Yale Center for Immuno-Oncology Virtual Symposium

Name: Yale Center for Immuno-Oncology Virtual Symposium
Uploaded: 2020-10-19T19:17:50.07Z
Duration: 5 h 2 min 29 s
Description: October 16, 2020

October 19, 2020

October 16, 2020

Information

ID: 5796
To Cite: DCA Citation Guide

Download Transcript

00:00I think I'll actually get going with.
00:04I think I'm all set right?
00:06So why don't we just keep it this way?
00:09Not upset the Apple cart and then we'll
00:12go with that and I just wanted to
00:15welcome the already 161 participants
00:17that we have and I think it's really
00:20a reflection of what an outstanding
00:22schedule that we have planned for today.
00:25I'm really, really excited
00:26on Marcus Bosenberg, um,
00:28the interim director for the
00:29El Center for immunooncology.
00:31This is a center that is based on.
00:34Collaborative interaction between
00:35the Yale Cancer Center,
00:37led by Charlie Fuchs and the
00:39Immunobiology Department,
00:40led by David shots as well as other
00:43aspects of at Yale University
00:45and we're trying to coordinate
00:47an enhanced the work and I'm now
00:49on Koleji at Yale and you know,
00:52it's I guess one Boone of not being
00:55able to travel is that it's hard to
00:58imagine getting a slate of external
01:01speakers like this if everyone has to
01:03be moving around and traveling all the time,
01:07so.
01:07Will really enjoy the day I think for
01:10the trainees at Yale and the faculty,
01:12this is going to be really exciting.
01:14A couple of things in terms
01:16of procedural things.
01:17I think all of you have the agenda.
01:20That will be a couple of breaks throughout
01:23and a half an hour lunch session.
01:25We may need
01:26to adjust times
01:27a little bit. We hope that the
01:30speakers can roughly stay on time
01:32for a total of 1/2 an hour per
01:34slot, which will include some
01:36questions. Panelists including
01:37speakers can actually ask questions.
01:39For those of you as attendees,
01:41there's a chat function that
01:42you should use to ask questions,
01:45and the moderate yrs will
01:46help monitor that Anne will
01:48guide the question period.
01:50And so I think, you know,
01:52we're probably going to need
01:54all the time we can get today.
01:56So with all this exciting science,
01:58I want to thank all those speakers
02:00really very, very much for
02:01participating in this upfront 'cause.
02:03I'm sure some people will have to go
02:05in and out during the day and stuff,
02:08but you know, up front,
02:09at least for our first session,
02:11for those who are here.
02:13Thanks so much.
02:14And I'd like to introduce Jeff Ishizuka
02:16who is an assistant professor at
02:18Yale University and medical oncology,
02:19who has a background.
02:20In immunology as well,
02:22and we're very excited to have
02:24him as a faculty member here.
02:26Also, as a colleague of some
02:28of the folks who were speaking,
02:30I think,
02:30and even this first session and Jeff,
02:33please proceed with
02:34the first session.
02:35Great, just echo at
02:37Marcus said it's
02:38it's fantastic lineup today and it's an
02:40honor and pleasure to moderate the session.
02:42I'm really excited to be here and
02:44to introduce our fantastic speakers.
02:46Our first speaker is doctor
02:47rafi Ahmed Rafi along with some
02:49of our other
02:50speakers today is responsible
02:51for laying a great deal of the
02:53groundwork for there to be a Center for,
02:56you know, oncologix, Here or anywhere else.
02:58His investigations into
02:59T cell differentiation
03:00and function have guided
03:01the way to our current understandings
03:03of T cell memory and T cell exhaustion.
03:06In his foundational work in acute
03:08and chronic else, MV have been
03:10essential in defining the role of PD.
03:12One in limiting T cell function.
03:14Raffi is the
03:15Charles Howard Candler
03:16Professor of Microbiology
03:17and immunology at
03:18Emory University and
03:19director of the Emory Vaccine Center. The
03:22title of his
03:23talk today is
03:24T cell lifestyle and chronic
03:26viral infection and cancer.
03:27Be shows after the infection
03:29is cleared and in striking
03:31contrast of the virus persist,
03:33specially at very high
03:34levels. But you get T cell dysfunction.
03:37The conceptual breakthrough
03:38in understanding T cell dysfunction that
03:41came from the studies of Valenzia. So I'm
03:43excited to introduce our
03:45last speaker of the
03:46session. Doctor Amanda Lund. Amanda
03:48has made Seminole contributions
03:49to our understanding of the regulation
03:51of immune function by lymphatic
03:53vessels and organs
03:54and of their role in shaping the anti
03:57tumor immune response. Her
03:59elegant work has
04:00spanned from fundamental
04:01immunological discovery to the
04:02development of quantitative and bio
04:04engineering tools and human translation.
04:06Amanda is associate professor in the
04:08Ronald Oper Element Department
04:09of dermatology in
04:10the Department of pathology at the end,
04:12why you, Grossman, School
04:14of Medicine. The
04:15title over talked today is
04:16lymphatic vessels, immune surveillance
04:18and immune escape in Melanoma.
04:21Thank you very much.
04:22It's really a pleasure and an
04:24honor to be here today and able
04:27to present some of our work.
04:29So I think the organizers very
04:31much for this opportunity.
04:32Like really everyone else here today,
04:35my lab is interested in understanding how
04:37we can generate and mobilize effective
04:40immune responses that enable tumor control,
04:42and there's really three main ways
04:45that we think about doing that,
04:47and that is.
04:48First we need to activate the Antigen
04:51specific sort of adaptive immune
04:53responses during tumor development.
04:55Second, we mobilize that effector
04:57immunity into the tumor micro environment
05:00where it can engage its target and 3rd.
05:03We are designing strategies to combat
05:05the multiple mechanisms of immune
05:07suppression that we know are present
05:09within these tumors and suppress
05:11the ability of the effector immune
05:13response to mediate tumor killing.
05:15And it's been clear,
05:17of course,
05:18through the seminal work of the
05:20presenters before me and many others
05:22that we can target tumor immune
05:24surveillance in tumor immune responses
05:26at each of these different points.
05:28But it's clear that they
05:30are the challenges remain,
05:31and there remains subsets of
05:33patients and types of tumors where
05:35we have not been able to mobilize
05:38really significant responses
05:39and tumor control and patience.
05:41And one thing that I think we should
05:44remember is that the there's an
05:46Amazon anatomic framework over which
05:49we mobilize these effector responses
05:51that is critical for allowing both
05:54the initial sensing of a primary
05:56tumor as well as the infiltration
05:59and retention of that affect
06:01immunity within the tumor space.
06:04And while we appreciate that
06:06this framework is certainly in,
06:08in some ways designed to
06:10facilitate immune surveillance.
06:11In the setting of cancer where we get
06:14Ramada Ling and dysfunction within
06:16various different onomatopoetic
06:18structures within the entire
06:20body of the animal or patient,
06:22that these changes may present
06:25additional barriers to the ability
06:27of the effective T cell or more
06:30responses to really do their jobs.
06:32And when my lab has been interested
06:35now and for quite some time
06:37as the lymphatic vasculature,
06:39this is a part of our vascular system,
06:42which mediates the unidirectional
06:44transport of fluid and cells and
06:46lipids from peripheral tissues
06:48to secondary lymphoid organs.
06:50Were adaptive immune responses are initiated.
06:52We know that this is the necessary
06:54in requisite route for the movement
06:57of Antigen loaded antigen presenting
06:59cells that leave peripheral tissues
07:01migrate towards draining lymph nodes
07:03and thereby activate robust antitumor.
07:06Immune responses.
07:07And work that I'm not going to be
07:10able to show you today from my lab is
07:13really learning that the lymphatic
07:15vasculature can tune its transport
07:17function both at the level of the
07:20lymphatic capillary and from work by others.
07:22Also at the level of the collecting
07:24lymphatic to in some way change the
07:27quantitative amount and the type
07:29of information that is able to be
07:31delivered to the Sentinel Lymph node.
07:34So in this way really,
07:35I think we played,
07:36we think of at least the
07:38lymphatic vasculature is a
07:39really critical arm of the
07:41peripheral tissue immune response.
07:43That of course we are very
07:46interested in continuing to explore.
07:48In the context of cancer, however,
07:51lymphatic vasculature has been
07:52largely appreciated for its
07:54role in regional metastasis.
07:55We know that, particularly in Melanoma,
07:58the presence of tumor cells
08:00in the Sentinel Lymph node,
08:02which is the lymph node directly
08:04draining from the primary tumor bed,
08:06is A is a negative prognostic.
08:09This is true also in other types of tumors.
08:13We know that we can find these
08:15lymphatic vessels which were shown
08:17here in Green both within and around,
08:19developing tumor beds,
08:20and that the increased density of
08:22these lymphatic vessels increases
08:24with stage and is associated with
08:26lived in a metastasis, impatience.
08:28And importantly,
08:29we know that either in humans where
08:31we see overexpression of edge FC
08:33by tumor cells or by myeloid cells
08:36infiltrating the tumor bed or in my
08:38swear we engineer the overexpression
08:40of edge of see that this is both,
08:42this is associated with.
08:44Increased metastatic potential
08:45to the regional lymph node bed
08:47of those primary tumors.
08:49But when I started my post doc
08:52now sometime ago with melody
08:54sorts at the PFL in Lausanne,
08:56we were really interested in challenging.
08:58This is so the complete story and
09:01asking instead how the biology of
09:03lymphatic transport and the Sentinel
09:05Lymph node really impacts tumor immunity
09:08and thereby influences tumor progression.
09:10And we've learned a lot since then,
09:13and I think it's pretty clear that
09:15even in the context of tumors at these
09:18lymphatic vasculature plays are really
09:20critical role in determining the
09:22extent to which the immune responses,
09:24both activated against the developing tumor.
09:26But also how these responses are maintained.
09:29Are we first showed up back in really
09:32in 2012 that the overexpression of
09:34edge FC by tumor experimental tumors
09:37not only increases the metastatic
09:39potential for these tumor cells
09:42to migrate towards a limp mode,
09:44but also generated a much more
09:46inflamed tumor microenvironment
09:47that subsequently activated multiple
09:49mechanisms of immune suppression,
09:51both within the primary tumor bed as
09:53well as in the Sentinel draining lymph node?
09:57And this work really sort of.
10:00Established the paradigm by which
10:02we could think about perhaps the
10:04idea that olymp angiogenic tumor,
10:06one that is really engaged,
10:08the tumor associated news about
10:11across culture,
10:11might be more responsive to immunotherapy.
10:14And that hypothesis was tested
10:16by melodies group where we showed
10:18in Melanoma that over legacy
10:20overexpressing tumors are actually
10:22potently responsive to immunotherapy,
10:24including immune checkpoint blockade.
10:26And this was really beautifully shown
10:29also more recently by a group at Yale.
10:32He was lucky group who showed in
10:34glioblastoma that the induction of
10:36a lymph angiogenic response within
10:38what's more normally thought of as
10:41an immune compromised or privilege
10:43site could really drive potent
10:45immune responses and response to me.
10:47A checkpoint blockade.
10:48And so this altogether has really
10:51suggested the potential for using
10:53the lymphatic vasculature in
10:55lymphatic transport
10:56as a new target for immunotherapy
10:58to improve immune surveillance and
11:00really turned on they response
11:02against the developing tumor.
11:03But since starting my lab,
11:05we were really interested in exploring
11:08some additional parts of lymphatic biology,
11:10and that was really how to lymphatic
11:13vessels in the context of a
11:15tumor that's already inflamed.
11:17How do they interact with that
11:19inflamed tumor biology?
11:21And might they continue to regulate the
11:23effector phase of the immune response
11:25within that tumor microenvironment?
11:27An in work that I'm not going to be able
11:30to show you today for the sake of time,
11:33we've now learned that after
11:35mobilization of Antigen,
11:36specific immune responses and recruitment
11:38into tumor micro environments,
11:40that we in fact see that a subset
11:42of these antigen specific T
11:43cells actually leave those tumors
11:45through lymphatic vasculature,
11:47that their exit is regulated by
11:49lymphatic vessel, derived Chemo Kines.
11:51And and that these exiting T cells
11:54are actually quite functional and in
11:56so doing their exit actually limits
11:58the ability for these tumors to be
12:01controlled and for response to immunotherapy.
12:04What I want to talk to you about
12:08today was sort of our first insight
12:10into how the lymphatic vasculature
12:12might be contributing to the efficacy
12:14of tumor control within the tumor
12:17micro environment,
12:18and that work was the work of
12:21a graduate student in the lab,
12:23Ryan Lane,
12:23where we learned that lymphatic
12:25vessels are exquisitely sensitive to
12:27cytotoxic community that accumulates
12:29within the tumor micro environment.
12:31They adapt to that side of toxicity
12:33by expressing multiple different
12:35factors including PD, L1,
12:37which will talk more about.
12:39And by decoupling the ability of
12:41the lymphatic vessels,
12:43descent cytotoxic community,
12:44we could actually significantly
12:46improve antitumor immune control.
12:48And so all this together has
12:51really suggested,
12:51along with work from many other
12:54labs at the lymphatic vasculature,
12:56in addition to playing an important
12:58role in immune surveillance and
13:00immune and tumor recognition,
13:02can critically also regulate the
13:04ongoing immune responses in the
13:06tumor microenvironment and maybe
13:07plays an important role in sort of
13:10immune escape burning resolution.
13:12I think one of the experiments that
13:14really for me best illustrates
13:16this point is shown here.
13:18So what we did is we have.
13:20We took mice that lack dermal lymphatic
13:22vessels in the skin which you can
13:25see here in green compared to wild type mice.
13:28We immunize both of these mice
13:30with LC MB Armstrong to setup a
13:32potent protective immune response,
13:33and then we came in and challenge
13:36these animals with vaccinia virus
13:38expressing an LC MB antigen GP 33.
13:40And what you can see is that in
13:43wild type animals,
13:45upon this secondary challenge,
13:46you get an initial inflammatory
13:48response in the skin,
13:49which is shown here by ear swelling
13:52that resolves fairly quickly overtime.
13:54But we do this in my side completely
13:57lack lymphatic transport,
13:59despite seeing no difference in
14:01viral control.
14:01In this particular experimental setting,
14:03what we see is a progressive and
14:06dramatic and long lived pathology
14:08within this year tissue and you
14:10can see
14:11that even better histologically,
14:13where we get epidermal hyper polif eration,
14:16we see an expansion of the dermis
14:18and a dramatic accumulation
14:19of CD 45 positive leucocytes.
14:22Even at this time point well past
14:25when the wild type mice have.
14:27Resolved, and so this really tells
14:29us that in the absence of an exit
14:32route in this lymphatic transport,
14:34that that tissues have an impaired
14:37ability to return to homeostasis.
14:39Well, we also know from work
14:42from back I'm going hard as well
14:45as our group melody and others.
14:47Is that the lymphatic individual
14:49cell present within the unique
14:51structure of the lymph node has some
14:54really unique and interesting and
14:56immunological properties that allow
14:58it to scavenge and present anagen
15:00that allow these cells to interact
15:03with both CD8 and CD4T cells through.
15:06Most notably the inhibitory molecule PD,
15:08L1 to drive dysfunctional CD8T
15:10cell activation. And or deletion.
15:12And so this is really exciting,
15:14interesting paradigm where at
15:15least these endothelial cells,
15:16and at least in the lymph node.
15:19Played a really interesting and
15:21unappreciated role in maintaining
15:22peripheral tolerance,
15:23at least at steady state.
15:25And so when Ryan joined the lab,
15:28we were really interested in two questions.
15:31One is are these immunological
15:33properties really unique to the
15:35lymphatic endothelia cell found
15:36within the lymph node itself?
15:38Or might they be activated in
15:41the context of challenge and
15:43peripheral tissues and organ tumors?
15:45And if so,
15:46is there really any reason to believe
15:48that the lymphatic vasculature out
15:51in these peripheral tissues might
15:53be interacting with affecter,
15:55CDA positive T cells?
15:57So what Ryan first started to do?
16:00Is he he?
16:01He said, OK, well,
16:02let's look in the in our implantable tumors.
16:04And let's ask whether or not competitive
16:07deals with in this contest expressed
16:09PD L1 so big Englehart had already
16:11shown and we had seen ourselves that
16:13if you look at cutaneous lymphatic
16:15vasculature in the absence of any
16:18kind of challenge there fairly PD
16:20L one negative and that's what you
16:22can see here in this black line.
16:24But what was really exciting to
16:26see was that if we look at the
16:28lymphatic endothelia cells here
16:30in Read that have been extracted
16:32from a tumor micro environment.
16:34They expressed elevated levels of PD L1.
16:37And this was distinct from the
16:39blood vasculature which actually
16:40expresses consecutive PDL.
16:42One and remains more or less unchanged
16:44by the tumor micro environment
16:45and what was really clear to us
16:48as we looked at them botic in
16:50ethereal cell PDL 1 / a range of
16:52different tumor micro environments.
16:54We saw the the level of expression and
16:56the number of cells that were expressing
16:58it appeared to be tuned by context.
17:01And if you actually look across these
17:04models and you correlate the level
17:06of expression of PDL one with the
17:08numbers of infiltrating CD8T cells,
17:10you can see a nice correlation
17:12across these models which really
17:13suggested to us sort of in a sensing
17:16ability of the lymphatic endothelium
17:18to respond to the accumulation of
17:20Antigen, specific immune
17:21response and activate PDL 1.
17:24But the really important first question was,
17:27is there any reason to think that an
17:30onomatopoetic non tumor source of
17:32PDL one is relevant for antitumor
17:34immunity and so we decided to take
17:36a bone marrow chimera approach to.
17:39To answer that question,
17:40Ryan generated bone marrow
17:42chimeras using PDL.
17:43One knockout mice such that in
17:45red here PDL one was lost on non
17:47hematopoetic cells or in blue PDL
17:49one was lost in the hematopoietic
17:52compartment we implanted B16F-10
17:54urine melanomas which are PDL one
17:56expressing but known to be poor RE.
17:59Ponders to single agent I mean
18:01checkpoint blockade and we got equal
18:03tumor growth in all our cameras.
18:05But if we actually looked at the T
18:08cell compartment within these animals,
18:10we were really intrigued to see
18:11that within the tumor infiltrating
18:13lymphocyte compartment,
18:14whether we lost PDL one onomatopoetic
18:16or amount of poetic cells.
18:18We saw this boost in overall proportion
18:20of CD T cells present and that of
18:23those PDT cell presence we saw this
18:26elevation of PD one as we would
18:28expect to see in the full knockout.
18:32Interesting Lee,
18:33this seemed to be segregated
18:34by anatomic location,
18:35so if you actually looks in circulating
18:38populations in the hematopoietic knockout,
18:40this elevation and number was
18:41already present.
18:42We look in the spleen or the
18:44tumor draining lymph node,
18:46but not in an Onomatopoetic Control,
18:48which really suggested that the
18:50effect of non amounted poetic PDL.
18:52One seems to be limited to the
18:54tumor micro environment itself.
18:56I'm in order to sort of resolve more
18:59functional effect on the immune response.
19:02We performed adoptive T cell transfer
19:04experiments where we activated
19:06antigen specific OT one T cells,
19:08transfer those into B16F-10 ova,
19:10expressing tumor bearing mice,
19:11and then asked whether or
19:13not these effector T cells,
19:15which would directly home to
19:17the tumor micro environment.
19:18We're capable of enhanced tumor
19:20control in the absence of PDL one,
19:23and what we saw in fact was that
19:26of course had aquatic loss of PDL.
19:29One had a significant effect.
19:31On the ability of these T cells
19:33to mediate tumor control,
19:35but really interesting, of course,
19:36is not what aquatic compartment
19:38also seemed to be playing some
19:41role in limiting the effector
19:43T cell response in Vivo.
19:44And perhaps most strikingly,
19:46we then decided to take an
19:48immuno genic murine Melanoma.
19:49The younger 1.7,
19:50which we know is exquisitely sensitive to
19:53to checkpoint blockade as a single agent,
19:56and we're able to see now
19:58that in the absence of PDL,
20:00one on non amount of poetic cells that we
20:03were putting tumors essentially in stasis,
20:05leading to really significant changes
20:08in tumor control in these mice.
20:11And so this suggested that yes,
20:13they not.
20:13Amount of product compartment may
20:15contribute to T cell dysfunction within
20:17the tumor microenvironment environment,
20:18and I'll remind you that from our analysis,
20:21the main expressing cells
20:22were really the endothelium.
20:24Both blood and lymphatic.
20:26So we were
20:27curious as to whether or not
20:29this was specific to the tumor.
20:31We presumed it was not,
20:32and we went through an,
20:34looked at a variety of
20:36different inflammatory.
20:36Responses in specifically in skin in mice,
20:39we infected mice with vaccinia virus
20:41in the skin initiated delayed type
20:44hypersensitivity responses or induced
20:46in imiquimod based model of psoriasis
20:48and what you can appreciate is that
20:51in all three cases we see this really
20:54dramatic increase in PD L1 on the
20:56lymphatic endothelium within the
20:58inflamed skin and specifically not in
21:01Contra lateral distal compartments.
21:03Really suggesting that this was a
21:05lymphatic and a filial response.
21:07To the accumulation of potent inflammation
21:11within that within that cutaneous space.
21:15These models also provided the
21:16opportunity to actually look at
21:18the kinetics of this response,
21:20which I think is really valuable.
21:22So we took the vaccinia virus model
21:24and we looked at the expression of
21:26PDL one overtime on both the blood
21:28endothelial cells as well as the
21:30lymphatic endothelia cells and you can
21:32see that again as I mentioned before,
21:35blended filial cells have some
21:36constituent of expression which is
21:38enhanced overtime in this model,
21:40but with the lymphatic endothelia
21:41mu really see a cell population
21:43that goes from absolutely 0.
21:45It's almost 100% expressing PD L1 over
21:47the time course of this infection.
21:50And what we now, of course,
21:52is that this transition recruiting
21:54day three and a seven is exactly
21:56the time point at which we would
21:58be expecting the accumulation of
22:00a cytotoxic CD8T cell response.
22:02And so this suggested to us that
22:04perhaps the CD8T cell and its
22:06effector molecules where the QQ
22:08leading to increased expression
22:09as one might expect based on our
22:12understanding of PDL one biology.
22:15But we tested this specifically in two ways.
22:18First,
22:19we vaccinated are tumor bearing mice
22:21with an attenuated listeria vaccine,
22:23either expressing a matched tumor antigen,
22:26ovalbumin or not to generate non
22:28specific inflammation and boost
22:30CD8T cell infiltration into tumors.
22:32In the absence of antigen recognition.
22:36And you can see were successful in their
22:38vaccines strategy and when we look at
22:41PDL one in the lymphatic endothelia
22:43you can see that it's really only
22:45elevated in the context of potent and
22:48boosted antigen specific immune responses.
22:50We can do that using our adoptive T
22:53cell transfer therapy experimental
22:55therapy as well,
22:56where we activate OT one T cells,
22:59we transfer those into established.
23:01In this case,
23:02B16F-10 ova during melanomas,
23:04and then take those animals
23:06down seven days later,
23:07and what you see is that the lymphatic
23:10endothelia has really responded
23:12potently to the infiltration of these
23:15effector T cells and upregulated PD
23:17L1 as compared to two no transfer.
23:21And so I think you know the
23:23obvious molecule here mediating
23:25this was interferon gamma,
23:27and indeed that is the case
23:29whether we look in our vaccinia,
23:31viral infected skin or in
23:33our tumor bearing animals,
23:34we see a dramatic reduction in PDL,
23:37one on lymphatic endothelia cells
23:38when the animals are treated
23:40with Interferon Gamma neutralizing
23:42antibodies and of course interferon
23:44gamma itself is sufficient.
23:45Let cultures in vitro to induce
23:48expression of PD L1 and it does
23:51so in a stat one mediated manner.
23:53And so, in the absence of having
23:56the PDL one flox mouse is really
23:58gave us the opportunity to say,
24:00well we can now D couple lymphatic
24:03endothelia cell biology from cytotoxic
24:05community by removing the Interferon
24:07Gamma Receptor on epithelial cells to
24:09ask whether or not this allows for more
24:11persistent effective T cell immunity.
24:14And so we did that we generated lymphatic
24:16specific interferon gamma receptor
24:18knockouts first using the live one.
24:20Cree, we could show that lymphatic in
24:22Attilio cells extracted from those animals
24:24no longer could phosphorylate stat one
24:26in the presence of Interferon Gamma,
24:28which is what you see here.
24:30While we maintain that responsiveness and
24:32all other cell types that we checked,
24:35we could show further that when we
24:37treat these cells with Interferon Gamma,
24:39where we normally see this nice
24:41expression of PDL one,
24:42that that's really completely lost in.
24:44In vitro and so we now have a really
24:47interesting tool to understand
24:49the interplay between cytotoxic T
24:51cells in lymphatic vessels in Bebo.
24:53And so we did that.
24:55I'm going to show you some work
24:56with the viral infection first,
24:58'cause I think it's a really nice.
25:00A way to think about this model,
25:02and So what we did is we first looked
25:05at PDL one in vivo at these seven,
25:08which was our peak level of expression
25:10and you can see that in the absence
25:12of the Interferon gamma receptor
25:14that we dramatically reduce the
25:16ability of these lymphatic individual
25:18cells to produce PDL one during the
25:20course of this infection.
25:22It was really interesting is that this was
25:25associated with an enhanced pathology,
25:27so this is 10 days post infection.
25:29You can see that we have some
25:31thickening of the epidermis.
25:33We have keratinocyte necrosis,
25:35enhanced accumulation of leukocytes,
25:36and dermal thickening,
25:37and really in what was a
25:39lot of work done by Ryan.
25:41He eventually really narrowed it down
25:43to an elevation in the number of T cells
25:47that were present within these tissue.
25:49Really over most other cell
25:51types that we looked at.
25:52We saw accumulation of both
25:54the force and CD 8 San.
25:56If we looked at a vaccinia specific
25:59response to be using Betar tetramer
26:01we saw almost a two fold increase in
26:04CD 8 positive beat our specific T
26:06cells within these tissues and I'll
26:08remind you that these tissues are
26:10already incredibly inflamed to CDA T cells,
26:13and so this is really striking to me.
26:16What was perhaps more striking is that
26:18it did not help in any way to improve tumor.
26:22Sorry,
26:22antiviral control.
26:23And So what this really suggested to
26:26us is this idea that of course many
26:28others have think about all the time,
26:31which is that we must always
26:32maintain the production and rapid
26:34mobilization of protective immunity
26:36with our ability to return
26:37to tissue homeostasis.
26:39And perhaps what we're seeing in this data
26:41is an inability to maintain that balance,
26:44and that by enhancing protective immunity,
26:46it's really at a cost of
26:48tissue structure and function.
26:50But what this data would also suggest
26:53is that there's now an opportunity to
26:55think about tipping this balance in
26:58the context of tumor immunotherapy.
27:00And might these animals now see
27:02improved antitumor immune control?
27:04And so we turned back to the Yammer
27:071.7 immunogenic mirroring Melanoma
27:08line that again is very responsive
27:11to single agent checkpoint blockade.
27:13We could see in these tumors now that
27:15in the absence of lymphatic interferon
27:18gamma receptor that these added filial cells.
27:21In the tumor failed to upregulate PDL 1.
27:25And what was really exciting to see
27:27was that this was functionally very
27:30relevant that tumors implanted into
27:32these mice exhibited more tumor,
27:35more immune control then those implanted
27:37into the Cree negative controls,
27:39and then that sort of stabilization of
27:42tumor growth was dependent upon CD T cells,
27:46because when we depleted these cells,
27:48which is shown here in green,
27:51we completely rescued tumor growth kinetics.
27:54Ryan confirmed this in a second pre.
27:57This is now an inducible Creed driven
28:00by prox one where we see really very
28:03very similar results which is this
28:06delayed tumor growth and sort of stasis
28:09in aggregate that leads to improved
28:11control and so this is really the
28:14first data to my knowledge that in vivo
28:17disrupted lymphatic vessel intrinsic biology,
28:20independent of its proliferative growth
28:22and demonstrated Anna direct effect.
28:24On the accumulation and functionality
28:28of a factor immune responses.
28:32I just want to make this one last point,
28:35which is that we do not know if in
28:37fact PD L1 on the endothelium might be
28:40working through antigen presentation,
28:43but we do know that lymphatic
28:45endothelia cells can present Antigen
28:47and this is work that we did back with
28:49Melody Swartz where we showed that.
28:52Uh,
28:52not a bit too am chimeras where
28:55we've lost all him out of poetic
28:57presentation that we still see
29:00presentation and activation of T cells,
29:03particularly when we overexpress
29:04edge of see OneDrive,
29:06Lymphangiogenesis and that lymphatic
29:08endothelia cells in vitro,
29:09or really good,
29:11better than other sort of control
29:13lines at taking up an antigen that
29:16requires processing for presentation.
29:18And so there certainly is this
29:20idea out there that these cells
29:23could be cross presenting antigen.
29:25How all that works in the context of
29:28the Imperial system still remains
29:29to be really carefully tested,
29:31and so with that,
29:32I think what I've shown you today is
29:35all very consistent with the idea that
29:38inflammation itself is self limiting.
29:39What I think our work demonstrates,
29:42in addition to lots of other
29:44work by other people,
29:45is that interferon gamma helps the
29:47threshold a response by activating
29:48can pensa Tori immune resolution
29:50mechanisms in multiple both cell types,
29:52both local and systemic,
29:54and this altogether is in crucial for
29:56maintaining the Fidelity of immune responses.
29:58We should I showed you today that non
30:01about aquatic PDL one limits effector
30:04T cell function into specifically in
30:06tumor micro environments that dermal
30:09lymphatic vessels and tumor associated
30:12lymphatic vessels are really exquisite
30:14sensors of local interferon gamma
30:16production and that by interrupting
30:19the ability of the lymphatic desens
30:21accumulation of the cytotoxic immune
30:23responses we can actually drive more
30:26durable and persistent tumor control.
30:29In Marion Melanoma And with that I will stop.
30:33I want to thank you all very much for
30:36listening as I sort of resent this work.
30:38All this work was done at Oregon
30:40Health and Science University.
30:42We've since moved to NYU,
30:43but I'm grateful very much to all my
30:46colleagues at OHSU for all their support.
30:48Of course,
30:49Ryan Lane did all the work that I showed
30:51you today and that is project is being
30:54continued by Teddy Moody on to who is
30:56a new student in the lab and I'd be
30:58happy to take all of your questions.
31:00Thank you.
31:02Thank you so
31:03much, Amanda. That was wonderful.
31:04Maybe I can
31:05start off actually,
31:06so I'm really
31:08interested in this.
31:09Finding that interferon
31:10gamma can actually limit anti
31:11tumor immunity. And of course a
31:13couple other groups have suggested
31:15similar things, although not with this
31:17mechanism. To my knowledge, Danny
31:19Mens group or new medicines group.
31:22We also know
31:23that Interferon Gamma is
31:24required for some
31:25of the key functions and anti tumor immunity.
31:28How do you think about targeting interferon
31:30gamma to improve anti tumor immunity?
31:32Yeah, I mean I think it's
31:34an important question.
31:35I think what our work and as well as
31:38all of the other work that's been done,
31:41really points to is that this all of
31:43the cells in the tumor microenvironment
31:45are integrating this signal,
31:47potentially in different ways, right?
31:49And so as an aggregate,
31:50just neutralization of the cytokine,
31:52you know clearly may not have the kind
31:55of intended results that you want,
31:57so I think it is really challenging
31:59to think about how we use it for
32:02therapeutic purposes, but it.
32:03I think very clear from all this work
32:05that if you can start to tease out
32:08the different mechanisms that are
32:10activated in a cell specific way,
32:12we really understand all of these
32:14nuances and perhaps that will
32:16lead us eventually to
32:17a better understanding for how we
32:19use it. Thank you. We have one for
32:22Nick Joe Sheehan which
32:23is on cross presentation
32:24of antigens by
32:25Ellie Elks. Is it
32:27possible that this
32:28serves as a sensing mechanism for
32:30the presence of Antigen and tissues?
32:31Are ATRM cells located near the?
32:34Else sees, yeah, that's
32:35great, but I
32:36think that's exactly how we initially
32:38thought about this, so it's clear.
32:40Sort of two things that lymph
32:42node lymphatic endothelia cells
32:43actually promiscuously express a
32:45lot of peripheral tissue antigens,
32:47so it's steady state.
32:48They are presenting things
32:50like tyrosinase which,
32:51when we think and again these are
32:53all limited by the lack of really
32:56sell specific in vivo models.
32:58But we think that that expression of PD
33:00L1 on the lymphatic individual cells,
33:02an expression of something
33:04might draw says actually limits.
33:06The generation and an mobilization
33:08of Malana site.
33:09Specific T cells.
33:10For example,
33:11we have evidence that they are
33:14also presenting in that issue,
33:16and we know limp itself is full
33:19of processed antigens and that
33:21there's reason to believe that the
33:24lymphatic endothelia cells are in
33:26some ways bathed in sort of tissue
33:29derived factors and antigens.
33:30So you know how they might interact with
33:33local more quiescent T cell populations.
33:36It's really.
33:37A current area of study that
33:40we know not enough about.
33:44Great, another question from the from
33:46the chat is how important is this pathway
33:48for non antigen specific immunity?
33:50We saw the T cell immunity but you see
33:53other anti tumor immune cells kind of
33:55changing or biology here. Yeah that's
33:57a great question.
33:58We you know we haven't looked as
34:00carefully at that as we should still do.
34:03Really understand what else is changing
34:04in these tumor micro environments.
34:06That is either sort of dependent
34:08upon the changes in affect or
34:10immunity or completely independent
34:11and so that still remains an open
34:13question and something we need to
34:15look more carefully at for sure.
34:18But thank you so much.
34:20I think, no, Marcus, I'll
34:21just ask one quick question.
34:23It will break very soon after that.
34:25So Amanda, I think it's striking with
34:27the gamma induced changes that you're
34:29seeing in the endothelial cells.
34:30Do you think there are endothelial
34:32specific suppressive mechanisms that
34:34might be targetable 'cause we talked
34:35about gamma being difficult to work with?
34:37Bizarre things?
34:38Are there changes by single cell RNA seq,
34:40or other approaches that you think
34:42there's new suppressive pathways
34:44that are endothelial specific?
34:45That's
34:45really interesting.
34:46So yeah, so we have been thinking about that,
34:49but haven't really gotten.
34:50We don't answer for you, right?
34:52So what is really happening downstream
34:54in the lymphatic endothelia am?
34:55That might be more specific.
34:57It was just exactly what you're getting.
34:59I think I think that's something
35:01that you know we certainly should do.
35:03And we've been thinking about,
35:05I think the other way to think about
35:07your question from the perspective of the
35:10endothelium is engineering strategies to
35:11actually target in a more specific way.
35:13And there's certainly people have
35:15been thinking about vascular
35:16targeting therapies or.
35:17Or are ways in which to
35:19administer checkpoint on.
35:20There was some nice study by
35:22Susan Thomas recently where you
35:24could actually target lymphatic
35:25transport in the lymph node.
35:27And so I think it's probably going
35:29to be combination of these things.
35:31It's a better understanding of
35:32the biology as well as maybe some
35:35delivery strategies that will help
35:36us target this more specifically.
35:40Well, great, thanks.
35:41I think I'll take over Jeff
35:43Real quickly and will take care.
35:45Thanks so much, Jeff for moderating.
35:47But really thanks so much to
35:48the speakers to Raffi, Arlene,
35:50Ann, Amanda for really great
35:51talks with a lot of interest.
35:53A lot of stimulation,
35:55so we'll have a brief break.
35:56It's 1135
35:57now will be starting back
35:59up promptly at
35:5911:45 in about 10 minutes,
36:01so will be signed back on, you know,
36:04a couple minutes before that,
36:05and make sure that doctor Rosenberg stock
36:07is all good to go for the next session.
36:10But thanks again and we'll
36:12see you guys very shortly.
36:13Alright. See in a few.
36:28Alyssa.
36:31Yes Hello, it's Tristan Park.
36:33Am I supposed to log onto a separate
36:37link or just stay on this note
36:40Erie rate in advance cell Therapeutics
36:42Laboratory that's run by Diane Kraus
36:45with assistance from a variety of
36:48other individuals at yell that is
36:50doing till harvest and expansion
36:52in a GNP compatible way that is
36:55also now being really confused
36:58in patients were obviously very
37:00interested in developing these areas.
37:02Further and so you know,
37:04I think not just us,
37:06but a lot of the world owes a debt
37:10of gratitude to doctor Rosenberg
37:12and the longstanding approaches
37:14that he's developed in
37:16cellular therapy overtime. And
37:17I think it's really exciting
37:19and low mutation burden tumors.
37:22How these approaches still really are as
37:25promising as anything that's out there.
37:27And I'd also like to introduce
37:30doctor Kristen Park,
37:31who is an assistant professor of surgery.
37:34At Yale, Ann had I think previously
37:37at least cross paths with doctor
37:40Rosenberg at some point in her training,
37:43and she is both a practicing
37:46surgical oncologist,
37:47but also has interests in a variety
37:50of different aspects of anti cancer.
37:53Immune responses including cell therapies.
37:55Kristen, if you wanted to introduce doctor
37:57Rosenberg and get the session going,
37:58that
37:58would be great. Hello
38:00everyone, it is my great pleasure to
38:02moderate the session and introduce
38:04doctor Steven Rosenberg MD, pH, MD,
38:06PhD is currently the chief of surgery
38:09for the National Cancer Institute and
38:11the National Institutes of Health.
38:13Doctor Rosenberg is considered one of
38:15the Founding Fathers of the field of
38:17amino therapy and his groundbreaking
38:19work enabled the use of high dose
38:21interleukin two as one of the
38:23first immuno therapies for solid
38:25organ cancers and his studies on
38:27the cell. Transfer of immunotherapy
38:29using bulk Tills Mutation.
38:30Reactive tools party at TC are
38:33transduced. Lymphocytes have
38:34resulted in complete, durable
38:35remissions in patients
38:37with metastatic Melanoma,
38:38other organ tumors and lymphomas
38:40which I was able to personally
38:42witness and participate in as
38:44a immunotherapy and surgical
38:46oncology fellow back
38:47in the day, doctor Rosenberg has received
38:50many numerous prizes, including the 2013
38:52Keio Medical Science Prize,
38:54the Albany Medical Center,
38:55Price in 2018, and the and the
38:582015 Medal of Honor of the.
39:01American Cancer Society.
39:02We're extremely excited to hear about
39:05your latest work and your
39:07thoughts. The title of this
39:10talk is self transfer immunotherapy
39:12for patients with metastatic,
39:14metastatic solid cancer.
39:27good to good to see you again.
39:31Going to be talking
39:33next 25 minutes or so on the
39:36development of cellular immunotherapy's
39:39for patients with cancer.
39:44I work for the US government,
39:46so sadly I have no personal disclosures.
39:49The goal of the efforts that I'll
39:52be describing of the development
39:54of effective immunotherapy's for
39:56patients with metastatic cancer based
39:59on the adoptive transfer of immune
40:01cells with anti cancer activity
40:03were using lymphocytes as a living
40:06drug for the treatment of patients.
40:11There are two critical issues in the
40:14development of cancer immunotherapy's
40:16that I'd like to discuss this morning.
40:18First, the identification of
40:20antigenic targets on the cancer cell.
40:22Can potentially lead to the
40:24development of new treatments.
40:26And of course,
40:27the identification of the
40:28characteristics of the immune cells
40:30that can recognize and destroy cancer
40:32cells and will touch briefly on
40:34those efforts in this discussion.
40:38Or attempt to answer some of those questions
40:41using cell transfer immunotherapy.
40:44That has multiple advantages
40:46as an immunotherapy approach.
40:49For one, we can administer large
40:51numbers of highly selected cells with
40:53high avidity for tumor antigens.
40:56We can potentially identify the exact
40:58cell subpopulations and effector
41:00functions required for cancer regression
41:02in vivo because we have the cells that
41:06are being administered as a drug in
41:08a test tube and can therefore study
41:11them of provides advantages compared
41:13to other approaches like the use of
41:16cancer vaccines or immunomodulators.
41:19And perhaps most importantly,
41:20we can manipulate the host prior to sell,
41:23transfer to provide an altered
41:25environment for the transfer.
41:26It sells because the drug we're going to
41:29use is no longer no longer in the patient.
41:34And so as we begin this,
41:36it's important to understand and I
41:38want to emphasize this to the non
41:40immunologists that are listening.
41:42The conventional T cell receptor
41:44recognizes its target because
41:47it has Alpha and beta chains,
41:49where on the left that recognize
41:53a processed peptide.
41:55That's presented on the surface
41:57of that particular patients M.
41:59HC molecule, either class one or Class 2.
42:02That's very different from a chimeric
42:05antigen receptor Here on the right.
42:08In which we take a take an antibody
42:11and isolate the variable region
42:14of the heavy and light chains.
42:20Combine them into a single chain
42:22antibody and attach that single
42:24chain antibody to intracellular
42:26signaling chains in a lymphocyte.
42:29So when this chimeric antigen receptor
42:32is inserted into lymphocytes and
42:34now converts a lymphocyte from its
42:37recognition from a conventional T
42:39cell receptor to that of an antibody,
42:42and the recent success of the use
42:45of chimeric antigen receptors,
42:47in fact, the first FDA approved.
42:50Cell and gene therapy has come from its
42:52success in the treatment of patients
42:55with advanced lymphomas and leukemias,
42:57we learned a great deal from even the
43:00very first patient that was ever treated.
43:03A patient here in the surgery branch
43:05who had an advanced lymphoma who
43:08taught us about some of the problems
43:11involved in using car T cells.
43:13This was a patient who had
43:15multiple prior treatments.
43:17As you can see,
43:18starting with a an aggressive combination.
43:21Chemotherapy than a vaccine
43:22that checkpoint modulator.
43:23Then,
43:24in another aggressive chemotherapy,
43:26progress through all of these
43:28until he came to the surgery branch
43:31NCI in May of 2009 for treatment
43:33with his own autologous cells.
43:36Into which an anti CD 19 Chimeric
43:39Antigen Receptor was transduced and
43:41despite all of these prior recurrences,
43:44he underwent a complete regression and
43:46is an on going complete progression
43:49free survival over 10 years later and
43:52this showed in fact that very large
43:55amounts of tumor can be can be progressed.
43:59You see here in these slides on the left
44:02the media steinem had a large tumor mass,
44:06the.
44:06Second down on the right,
44:09another large mediastinal mass you
44:11see in the third slide down large.
44:15Again,
44:15shown by these yellow arrows,
44:18large lymph lymph nodes that are
44:20surrounding the aorta,
44:22vena cava and finally a very large
44:24spleen and large iliac lymph nodes,
44:27all of which regressed completely.
44:31Bone marrow is full of tumor cells
44:33that you see on the left and they
44:36disappeared completely as well.
44:37But there was a price to pay for this
44:40because these car T cells recognized CD
44:4219 that's present than normal B cells,
44:45and it's just as easy to eliminate
44:47a normal cell as a tumor cell
44:50when cells are reacting,
44:51and as you can see,
44:53these B cells disappeared completely
44:55at a time when the normal T
44:57cells as you see in the bottom,
44:59recovered completely after about 8 days.
45:02And the natural killer cells were covered.
45:04It took about 7 or 8 months for
45:08the B cells to recover as well.
45:12We went on to treat a large number
45:15of patients in the surgery branch.
45:18You see a about a 50% complete response rate.
45:21Kite Pharma then performed a multi
45:24institutional study received.
45:25Gotta almost identical results.
45:27We then transferred our technology
45:30in a cooperative research agreement
45:32to Kite Pharma in 2012 and 2017 they
45:35received FDA approval for this,
45:37and this then received a fair
45:39amount of publicity because in
45:422017 kite was sold to ghillie add
45:44Sciences for 11.9 million dollars,
45:46just five years after this
45:49technology was transferred to do
45:52that. So this treatment,
45:53which can reduce the number of
45:56normal cells expressing CD 19,
45:58is now available for treatment
46:00around the United States face
46:02and beginning in Europe as well.
46:04But here we see the cancer deaths
46:06in the United States last year.
46:09As you can see, 600,000 deaths in total.
46:12About 10% or 58 thousand
46:15were of the humid logic.
46:17The blood cancers,
46:18but the solid epithelial cancers accounted
46:21for 90% of all of all cancer deaths.
46:26And so the major challenge that's
46:29confronting oncologix as a whole.
46:30But certainly cancer immunotherapy,
46:32is the development of effective immune
46:35or therapies for patients with metastatic
46:37epithelial solid cancers that cannot
46:39be cured by any available treatment
46:41and result in 90% of cancer deaths.
46:43Again,
46:44these epithelial cancers have ducts.
46:46The lining of the Ducks are
46:48constantly turning over as they do.
46:50Mistakes are made in the DNA called
46:52mutations and its accumulation of
46:54mutations that actually results in the.
46:57And the cancer.
47:00So will car T cells be useful for the
47:03treatment of solid epithelial cancers?
47:05Certainly not based on what we know now.
47:08Car T cells require the use of
47:11monoclonal antibodies that recognize
47:12molecules on the cell surface.
47:14These monoclonal antibodies were
47:16first described by Kohler and Milstein
47:1945 years ago in a paper in nature.
47:21But there are two major problems as we
47:24look to apply this to solid tumors.
47:26There's been no monoclonal antibody
47:28ever identified that recognizes
47:30cell surface molecules that are
47:32unique to epithelial cancers.
47:33And as I mentioned,
47:35it's just as easy to kill these normal
47:38cells as there are tumor cells,
47:40and so the destruction of essential
47:42organs when using our T cells as
47:45a major major potential problem.
47:49And so it's mainly the use of
47:52conventional T cells with T cell
47:54receptors that I think have the highest
47:57likelihood of being effective in
48:00the treatment of these solid tumors.
48:03I'm going to concentrate remarks on
48:05the on the solid epithelial cancers,
48:07but we've learned so much from the
48:09treatment of Melanoma that we're now
48:11applying that I'd like to just very
48:14briefly review our results of 190 four
48:16consecutive patients that were treated
48:18with the patients own autologous tumor,
48:21infiltrating lymphocytes,
48:21and you can see a 55% recist objective
48:24response rate with about 1/4 of the patients
48:26undergoing complete durable responses,
48:28and this provided a valuable resource
48:30because we had about half of patients that
48:33were responding and half not responding,
48:35that enabled us.
48:37To do experiments to try to identify the
48:40reasons for response or non response,
48:43I should mention that of these
48:4546 complete responders,
48:4744 of them received just a single treatment.
48:50Again,
48:50it's a living treatment of cells
48:53can expand 10,000 fold over the
48:56first 2 weeks and patrol the body
48:59looking for their target antigens.
49:02But without the cell therapy with
49:05till therefore appears able to
49:07eliminate the last Melanoma so.
49:10And that then leads to the question of
49:12what till recognize that enables the in
49:15vivo destruction of this last Melanoma cell.
49:17And it was the clue that I mentioned
49:20here on the bottom of this slide,
49:22that specific cancer regression,
49:23in the absence of off tumor on
49:25target toxicities in patients.
49:27Let us to explore the role of specific
49:30cancer mutations as the targets of till.
49:32Because it's these mutations that actually
49:35causes the cancer and of course will
49:38not be expressed will be president or
49:40expressed in normal and normal cells.
49:43Adjust a point very important to
49:45understand as we mine the cancer
49:47exomes that is all expressed cells
49:50to identify immunogenic cancer
49:52mutations and that is for a mutation
49:54product to be a cancer antigen.
49:57It has to be processed intracellularly
50:00into a 9 or 11 amino acid peptide.
50:02And that pep side ask to fit and be presented
50:06in the groove of one of the patients.
50:09Surface MFC molecules and therefore only
50:12rare mutations are going to be antigenic
50:15because they have to fulfill these two.
50:17These two properties.
50:18And so we developed an assay to
50:21identify mutation reactive cells in
50:23common epithelial cancers that would
50:26enable us to identify not only to
50:28sell but the antigen being recognized.
50:31And if you follow this cartoon
50:34gastrointestinal tumor deposit or an
50:36epithelial cancer deposit is for sected,
50:39we isolate genomic DNA and RNA.
50:42We identify the whole exome and
50:45transcriptome sequencing to identify
50:47every mutation in the cell by
50:49comparing it to the normal genome.
50:52That can be accomplished in a
50:55laboratory in about 2 weeks.
50:58We then Spencer synthesize all of the
51:02mini genes that encode these mutations.
51:06Or identify peptides that encode the
51:10mutations that represent 25 Mer peptides.
51:14With the mutated.
51:15Amino acid in the middle of that,
51:1825 more. We then express those tandem.
51:22Any jeans or peptides representing
51:23every mutation in the patient's own
51:26autologous antigen presenting cell,
51:28and so any mutations that can be presented
51:31will be presented on the surface
51:33of that patients on MFC molecules.
51:36We then expand till and
51:38used in these melanin.
51:40These experiments till that mediated
51:42complete durable regressions as I've
51:44shown before and when they are code
51:47incubated with the antigen presenting cell.
51:49If any antigens are recognized we
51:52can detect them by using interferon
51:54gamma elispot assays or upregulation
51:57of Owenby Bierocks 40 activation
52:00molecules by flow cytometry.
52:02Again, it's the 25 more.
52:04That's the key to this.
52:05With the mutation in the middle because
52:07it has to encompass every peptide
52:10possibly presented by that by that mutation.
52:12This mutated amino acid can be
52:15the first of the 9 or 10 more.
52:17It can be the last of the 9 or 10 more,
52:21but there's no need to predict
52:23peptide binding.
52:24Every candidate peptide and all them HC
52:27of the patients are included in the screen,
52:30and there's no tumor cell line
52:32necessary to do these assays.
52:34Well,
52:35we perform this in 20 two consecutive
52:37patients with metastatic Melanoma.
52:39They had a total of those patients
52:41had a total of 13,000 mutations,
52:44but we screened about 4000 that
52:47were expressed.
52:48In the cell,
52:49based on the RNA seq data that led to
52:53the identification of 54 immunogenic
52:55neo epitopes at 82% of Melanoma,
52:58pets of patients have these
53:01mute these antigenic mutations.
53:03About 1.4% of all of the mutations
53:06were recognized and almost exclusively
53:08by CD8 cells and quite surprisingly,
53:11all of these 54 neoantigens were unique.
53:14None were shared among any Melanoma patient.
53:18Among any Melanoma patients.
53:22We then extended this 230 patients
53:25consecutive patients with
53:26gastrointestinal cancers.
53:27About 80% of these patients had
53:30immunogenic mutations that were
53:32recognized by the patients own cells.
53:35We screamed about 15,000 of
53:38these mutations to identify 210
53:40immunogenic epitopes.
53:41Again,
53:42only one point 3% of all of the
53:46mutations were actually immunogenic.
53:49And interesting Lee in
53:51the epithelial cancers?
53:52Almost half of these antigens are recognized
53:56by CD4 cells and half by CDA cells.
53:59In contrast to the CD 8 recognition
54:02in melanomas and once again
54:04of these 210 Neoantigens,
54:06all were completely unique to that patient.
54:09Except for two patients who shared the
54:12same cave rats immuno genic mutation
54:14that was restricted by the class.
54:17One antigen CW 0802.
54:20Back then,
54:21Let us do not only look at the GI cancers,
54:25but is UK and see in the left breast cancers,
54:29colon cancers, ovarian cancers,
54:31prostate cancer.
54:32In 195 consecutive patients that we studied.
54:3677%,
54:37regardless of the Histology
54:39appeared to contain T cells that
54:42recognize antigens on that cancer,
54:45that was a total of 360.
54:48Three immunogenic neoantigens and once again.
54:52Repeated story,
54:53all of these were unique except
54:55for 2K grass antigens expressed on
54:58a particular restriction element.
55:00We're the only ones shared except
55:03for those two.
55:04All of the remaining neoantigens were
55:06completely unique to the individual patient.
55:11Now, an advantage of targeting
55:14mutations is its applicability
55:16to target multiple cancer types.
55:18Now, most patients that were attempting
55:20to treat with these selected cells
55:23that recognize the mutated antigens.
55:25Most patients do not respond,
55:27then I'll go to the numbers in a minute,
55:30but I want to emphasize the advantage of
55:33this approach because it's not specific
55:36to any individual cancer diagnosis
55:38and let me show you some examples.
55:40The first patient that ever responded to
55:43the use of cells selected for reactivity.
55:46To these immunogenic mutations and
55:48we're still learning on how to best
55:51select those cells with a patient
55:53with a metastatic cholangio carcinoma
55:54should have detected me Jed resection
55:57of lung and liver metastases.
55:59She had received multiple
56:00chemotherapy regiments.
56:01We treated our first with unselected
56:04till from a resected lung lesion.
56:06She did not respond.
56:08We then used our tandem minigene
56:11and peptide approach to select
56:13cells to treat her with.
56:15She had a unique population of cells at
56:18targeted the nerby two IP cancer mutation,
56:21one of 26 mutations that we screen
56:23and this patient underwent a complete,
56:26durable regression.
56:27Now I'm going over seven years later you
56:30see the lung metastases on the left,
56:33which disappeared completely.
56:34She also had three liver metastases at
56:37disappeared completely on the right.
56:39That's just a whole left in the
56:41liver following the elimination of
56:43that meta static.
56:45Composite.
56:49This is a patient with a breast
56:51cancer who showed a dramatic response.
56:54She had a as an estrogen and progesterone
56:57positive invasive breast cancer.
56:59As you can see, received seven different
57:02treatment regiments before she came to us.
57:04Chemotherapy hormonal treatments,
57:05targeted therapies,
57:06until she came to the NCI in 2015,
57:09she had 62 mutations.
57:10We targeted four of them and she
57:13underwent a complete tumor regression.
57:15You can see on the left that
57:18large mass that was beginning to
57:20protrude through the chest wall.
57:22You can that disappeared.
57:24You can see in the lower portion
57:26on the right with the yellow
57:29arrows multiple liver metastases.
57:31It disappeared completely as well,
57:33and she remains disease free
57:35almost almost five years later.
57:38We target in that patient for different
57:41what appeared to be random somatic
57:44mutations of no particular interest,
57:46as oncogenic mutations that
57:49lead to her tumor regression.
57:52Another patient, now in the cervical cancer,
57:55was treated with this approach.
57:58She presented with the fan gating
58:00cervical mass into the vagina, lung,
58:03and intraperitoneal metastases.
58:04Received radiation therapy and
58:06cisplatin chemotherapy progressed,
58:08underwent hysterectomy and excision
58:10of ovaries but develop multiple liver
58:13lymph node intrabdominal metastases
58:14came to the NCI in May excuse me.
58:18In March of 2013 with three to 1775
58:21billion of our own tumor infiltrating.
58:24Lymphocytes selected for antitumor
58:26reactivity and experienced a
58:28complete regression of all disease,
58:29including a relief of our urinary
58:32obstruction and remains disease
58:34free over seven years later again,
58:36you see on the left yellow arrows pointing
58:39to each of these cancers deposits on the top,
58:43the lymph node,
58:44the next abdominal wall mass,
58:46another intraperitoneal lymph
58:47node and on the lower left,
58:50the lymph node that was obstructing
58:52arview order that the white area is the.
58:56Stand in her urine are you can see on
58:58the right all of these it disappeared.
59:01We remove the sense that she
59:03remains completely disease free.
59:04Now seven years later.
59:07And finally, a patient with colon
59:10cancer who had a meta static and
59:12aggressive metastatic colon cancer
59:14that was invading into her bladder.
59:17She had a partial cystectomy
59:19along with their sigmoid colectomy
59:21developed multiple lung metastases,
59:23radiation therapy to the bladder,
59:25suture wall combination chemotherapy.
59:27We respected too long metastases for to
59:30obtain tumor infiltrating lymphocytes.
59:32We selected the populations that could target
59:35her unique cancer mutations based on our.
59:38RSA, she had 61 somatic mutations.
59:42She had seven lung metastases,
59:44six of them disappeared completely,
59:46but one the 3rd row down
59:49on the on the right grew.
59:51We then respected that lesion,
59:53leaving her disease free and in the
59:56analysis of that respected lesion we found
59:59that she had lost part of chromosome six,
01:00:03which encodes her major history
01:00:05of Pat ability antigens.
01:00:06Therefore,
01:00:07no mutation could be presented,
01:00:09and that probably explains why
01:00:11that lesion progressed.
01:00:12But with respecting that she has remained
01:00:16now disease free over far five years,
01:00:19five years later.
01:00:22And so our results are summarized here.
01:00:25This is very much a work in progress.
01:00:27If you give unselected till in these
01:00:30chemorefractory metastatic solid cancers.
01:00:31What works in Melanoma does not work in them.
01:00:35We had no responses when we
01:00:37selected the till we began.
01:00:39We're beginning to see responses of 12% rate.
01:00:42We studied the shells and showed that
01:00:44they would express PD one after they
01:00:47were administered at very high levels.
01:00:50When we edit checkpoint modulators,
01:00:52our response rate in
01:00:53these epithelial cancers,
01:00:54it's now 23% and we're continuing
01:00:57to work on improving this treatment.
01:01:00So I would raise then two hypothesis,
01:01:03one that it's the recognition
01:01:05of random somatic mutations.
01:01:06That is the final common pathway
01:01:09explaining cancer regression for most
01:01:12immuno therapies for solid cancers and it's.
01:01:15Evidenced in part by the studies
01:01:17I've just shown you,
01:01:18but also by showing that by data
01:01:20showing that checkpoint modulators
01:01:22tend to be more effective when
01:01:24there are more mutations and you've
01:01:26seen the data with respect to tumor
01:01:28infiltrating lymphocyte therapy.
01:01:31Second hypothesis.
01:01:33Revolves around the identification
01:01:34of cancer antigens.
01:01:36We finally,
01:01:36I believe,
01:01:37no at a cancer antigen is any
01:01:40intracellular protein can potentially
01:01:41be a cancer antigen if it's mutated
01:01:45and processed intracellularly
01:01:46to a peptide that combined to
01:01:49the autologous M HC mols.
01:01:53Good news and bad news.
01:01:55The good news is that since
01:01:57all cancers have mutations,
01:02:00virtually all cancer histologies
01:02:02are potentially eligible for
01:02:04this kind of approach.
01:02:06The bad news, however,
01:02:08is that the treatment will be.
01:02:11It will have to be highly individualized
01:02:14for patients because each express
01:02:17different immunogenic mutations and
01:02:20therefore this treatment is going
01:02:22to be very complex to administer.
01:02:27Well, we're looking at a whole variety
01:02:30of improvements to simplify this.
01:02:31To make it more effective with the idea
01:02:34that if we can develop highly effective
01:02:36treatments for these 90% of patients,
01:02:39the genius of American industry
01:02:40will figure out ways to deliver it,
01:02:43as was the case for car T cells.
01:02:46There are a variety of
01:02:48areas that we're looking at.
01:02:50Each one could be the subject
01:02:52of a one hour talk,
01:02:54but I just like today to discuss
01:02:56the one that I've listed here
01:02:58in red and that is to transduced
01:03:01mutation reactive TCT cell receptors.
01:03:04That recognizes immunogenicity stations
01:03:05into naive or central memory cells.
01:03:08So we now have FDA approval to
01:03:10use transient vectors with minimal
01:03:12with minimal testing.
01:03:14That makes this practical.
01:03:15The idea behind this is that
01:03:17conventional tool sales have under
01:03:20God Replication for years in the
01:03:22patient in vivo and therefore
01:03:24developed an exhaustive phenotype.
01:03:26PD one expression lag 310.
01:03:28Three if we can take these T cell
01:03:31receptors and put them into naive
01:03:34or central memory cells,
01:03:35we might have a cell with an explosive.
01:03:39Proliferative potential to
01:03:41administer to the to the patient.
01:03:44In the first patient,
01:03:46we actually have done that with factors only.
01:03:50Uh huh.
01:03:51Been done with one in one patient
01:03:55receiving these P53 mutations.
01:03:57We've identified a whole series of.
01:04:01T cell receptors.
01:04:02As you can see here in this second and
01:04:06this from a variety of patients that
01:04:09recognize a variety of P53 mutations
01:04:12on a variety of class one and Class
01:04:152 immuno genic class 2M HC molecules.
01:04:19We saw a patient that had a
01:04:22meta static breast cancer.
01:04:24That had a P53 mutation.
01:04:28That was in fact.
01:04:31R175 H 175th amino acid.
01:04:34A mutation that's present in about
01:04:372% of all patients with with cancer.
01:04:42She was oh 201 as well.
01:04:46She was a patient who had a metastatic
01:04:49breast cancer again on who had
01:04:52undergone multiple treatments
01:04:53and progressed through them.
01:04:55Came to us quite ill.
01:04:57September of 2019 she had tumor
01:05:00surrounding her heart.
01:05:01We had to perform a pericardial window
01:05:05three days before her treatment,
01:05:07with 550 billion of our own cells
01:05:11that were reactive against P53
01:05:13that would taken from this Library
01:05:16of of T cell receptors that the
01:05:19doctor Peter Kim had developed along
01:05:22with others in the surgery branch.
01:05:26You underwent a 90% response,
01:05:28but then recovered it about six months.
01:05:31You can see on our X Ray here.
01:05:34She had on the left surrounding our heart,
01:05:36extensive disease in the pericardium.
01:05:38It was biopsy proven to be tumor
01:05:41that disappeared completely.
01:05:42As you can see on the right.
01:05:46She also had a.
01:05:48Extensive disease in her breast
01:05:50extending super subcutaneously and
01:05:52intracutaneous that was extending
01:05:54on to the left breast,
01:05:56all of which disappeared completely,
01:05:59but again about six months later she did.
01:06:03She did recur.
01:06:04And so we're continuing to pursue
01:06:06that approach,
01:06:08but as well are asking the second
01:06:10question that I mentioned at the beginning,
01:06:13and that is what is what are the
01:06:16phenotypic characteristics of the cells
01:06:18that mediate cancer regression in vivo?
01:06:20Because we have the cells in a test tube,
01:06:23we can apply a new approach of
01:06:25high dimensional singles del
01:06:27transcriptome analysis that analyzes
01:06:29the transcript tone of up to it.
01:06:3110,000 cells all-in-one assay
01:06:32and we use a T Sne analysis.
01:06:35A way to reduce the multiple
01:06:38dimensions of this of this data.
01:06:42Into clusters.
01:06:43After all, we're dealing with
01:06:45a high number of patients.
01:06:47Each one of Mitch maybe expressing
01:06:49up to 20,000 antigens.
01:06:51It's an immense amount of data,
01:06:53but by doing a stochastic near neighbor
01:06:56embedding analysis we can divide all
01:06:59of that data and reduce it to a 2
01:07:02dimensional plot called AT sne plot.
01:07:04And here is one of our first experiments.
01:07:07Doing that,
01:07:08we could take Melanoma samples that
01:07:10were analyzed by mass spectrometry,
01:07:12but you can of course.
01:07:14Do the entire transcriptome
01:07:17utilizing other techniques.
01:07:20It divided those cells
01:07:24into multiple clusters.
01:07:27And you can see here up to
01:07:2920 two different clusters,
01:07:31but 'cause we knew the exact
01:07:34transcriptomes the exact antigens that
01:07:36were being expressed by every cell and
01:07:38the T cell receptor in those cells.
01:07:41From this analysis we then examined.
01:07:44Each one of these 22 clusters.
01:07:47Looking at the transcriptomes to see
01:07:49if any group of these clusters would
01:07:52separate responders and Nonresponders.
01:07:54And as you can see here in this first
01:07:56and in this analysis of cluster
01:07:58one it was the only cluster that
01:08:01statistically significantly separated
01:08:03responders from Nonresponders and
01:08:05you can see that on the on the left.
01:08:10Cluster one, it turns out,
01:08:12was highly enriched for CD39 CD,
01:08:1569 stemlike lymphocytes that appeared
01:08:17to be the effector cells responsible
01:08:20for her Melanoma treatment.
01:08:22And as you can see here in the middle,
01:08:26if you look at the total number of
01:08:29cells that were administered that did
01:08:32not separate responders from Nonresponders.
01:08:36Respond as being in red.
01:08:38But when we took those same populations
01:08:42and analyze the double negative.
01:08:44Cells that administration of high
01:08:47an low numbers highly statistically
01:08:49significantly separated.
01:08:50The responder's for the high
01:08:53survival from the nonresponders,
01:08:55with a low survival,
01:08:57and so we're vigorously approaching
01:09:00approaching now the use of these
01:09:03high dimensional techniques to try
01:09:05to identify the transcriptomes of
01:09:07the exact cells that are responsible
01:09:11for a tumor regression.
01:09:14Well,
01:09:14I've attempted to describe our
01:09:16efforts to develop a blueprint for
01:09:18cancer immunotherapy that we can
01:09:20direct against epithelial cancers
01:09:22by targeting the immunogenic somatic
01:09:24mutations unique to that patients
01:09:26cancer and by raising libraries of
01:09:29T cell receptors against shared
01:09:31cancer antigens,
01:09:32each with a different restriction
01:09:34element for K brass and P53 mutations.
01:09:37I might conclude.
01:09:40With the with the following and
01:09:42that is at cell transfer therapy
01:09:44can mediate durable regressions in
01:09:46patients with metastatic cancer
01:09:48refractory to other treatments
01:09:50that T cells that recognize unique
01:09:52somatic mutations can be found
01:09:54until and also in PBL in
01:09:56patients with common epithelial
01:09:58cancers and that the identification
01:10:00and targeting of the mutations unique
01:10:02to each cancer or shared mutations
01:10:05such as KRS or P53 has the potential
01:10:08to extend cell therapy to patients.
01:10:10With Comrent Epithelial cancers?
01:10:13Thank you for your
01:10:15very kind attention.
01:10:18Thank you so much doctor Rosenberg
01:10:20for that marvelous talk.
01:10:22It was wonderful to hear all of these.
01:10:26All of this data, so I'm going to
01:10:29start reading the questions from the
01:10:31Q&A so this is from each young Kim
01:10:35success of tumor mutation reactive
01:10:37lymphocytes over bulk till suggest
01:10:39number of cells infused matters as
01:10:42both till probably has the specific
01:10:44T cells recognizing the antigens.
01:10:46What are the T cell numbers?
01:10:49That are needed to give the effect in bulk
01:10:52till versus tumor mutation reactive tools.
01:10:56So we give all the till that
01:10:59we can grow an virtually.
01:11:02All patients have received
01:11:05somewhere between 2:00 and 10:00.
01:11:08Times 10 of the 10 cells
01:11:11up to about 100 billion.
01:11:15Sells well 10 billion cells but
01:11:16we don't know the number that are
01:11:19actually required when it used.
01:11:21When one uses car T cells.
01:11:23Carl June has shown that even a single clone,
01:11:26a Type 1 cell,
01:11:27can expand to large numbers and
01:11:29mediate the regression of lymphomas.
01:11:31So if you have the right cell
01:11:33that can grow and have a
01:11:35high proliferative potential,
01:11:37you may not need many cells.
01:11:39OK, the
01:11:40next question is, are the CD 39 CD,
01:11:4469 negative resource like stem memory cells.
01:11:47Doctor Raffi discussed and if so,
01:11:50would T cells drip from draining lymph
01:11:53nodes or metastatic lymph nodes?
01:11:55Be more effective as a till source then.
01:11:59Tumor infiltrating lymphocytes
01:12:00from the primary tumor. So
01:12:04mouse studies have identified the phenotype
01:12:06of cells that can that are involved.
01:12:09CD 39 has been one. I don't think C.
01:12:1369 in combination with it has been recognized
01:12:16'cause there are very few of those cells.
01:12:20CD 103 is been identified.
01:12:22So there's some similarity with
01:12:24what's been seen in other models,
01:12:27but it is very unique and rare.
01:12:30CD39 CD, 69 cell that appears to
01:12:34mediate the regression. Again,
01:12:35most effector cells are CD 39 positive.
01:12:39It's only the stem like cell that
01:12:42gives rise to CD 39 positive cells that
01:12:45appears to be appears to be critical.
01:12:49Strictly a tumor infiltrating lymphocytes,
01:12:51but haven't looked at draining
01:12:52lymph node cells.
01:12:53They may well be a better source,
01:12:55but are, of course,
01:12:57a little more difficult to
01:12:59obtain from all patients.
01:13:00OK, and
01:13:01a question from one of our
01:13:03surgical Oncologist. Spectrally,
01:13:04no have you used a bit genetic
01:13:06drugs to attempt to increase M HC
01:13:09expression on tumor cells prior
01:13:10to utilizing till treatments?
01:13:12About 10 years ago we did a study giving.
01:13:15Escalating doses of Interferon Gamma,
01:13:18which is a very potent molecule
01:13:20that can up regulate class one and
01:13:24Class 2 antigens on tumor cells.
01:13:26But we reached very toxic levels
01:13:30of Interferon Gamma administration
01:13:32and did not see any significant
01:13:34increase in MA C molecules.
01:13:37If there are ways to up
01:13:40regulate them in Vivo,
01:13:41I think one could improve the effectiveness
01:13:45of these treatments.
01:13:46Can I ask a question?
01:13:48Sure, there are a couple of.
01:13:50Maybe it's a really great talk.
01:13:52Really, really enjoyed that.
01:13:54Also, I think I want to.
01:13:56I'm sure everyone or many people
01:13:57realize how much of a Tour de
01:14:00force identifying all those antigen
01:14:01specific T cells and epitopes are,
01:14:04but it's not trivial answer reason why not.
01:14:06Many people have done that,
01:14:08but identifying tumor reactive
01:14:09T cells up front in a high
01:14:12throughput way I'm sure is something
01:14:14that you thought a lot about.
01:14:16You mentioned for one BB.
01:14:17Now the CD 39 C 69 double negatives.
01:14:20What are your thoughts moving forward
01:14:22about identifying tumor reactive till?
01:14:24Prior to expansion and focusing
01:14:26on that subset.
01:14:28So
01:14:28I haven't talked about that much,
01:14:31but one of We are pursuing these high
01:14:34dimensional analysis approach is to try
01:14:36to answer that question starting with
01:14:39cells that are isolated from individual
01:14:41single cell suspensions from a respective
01:14:44lesion and he studies being done by
01:14:47Frank Lowery in the surgery branch,
01:14:50and in fact we can in fact do this
01:14:53analysis on these individual and
01:14:55individual cells and attempt to identify.
01:14:59Markers that might be significant and there
01:15:02are about 19 markers that he's identified.
01:15:05Perhaps the most prominent one is CX CL 13.
01:15:10As a as a marker,
01:15:12but there are gene signatures that can.
01:15:15We believe very early work.
01:15:18Identify the cells from the original
01:15:20tumor before their cultured.
01:15:22That might be a means of identifying
01:15:26those those cells and expanding them.
01:15:29So that's that's a challenge,
01:15:31but I think one that is.
01:15:33That will likely be possible to accomplish.
01:15:39Right Crystal maybe will. Thanks so much.
01:15:44Doctor Rosenberg. Yeah. It's
01:15:47my great the my great pleasure and
01:15:49it was a great pleasure to have
01:15:51Tristan be one of our fellows.
01:15:54I should mention that virtually all
01:15:56the work that I've presented over has
01:15:58accumulated over the last 20 years
01:16:00was done by Clinical Fellows who come
01:16:03to the surgery branch to do research.
01:16:05So Tristan, thank you
01:16:07for your contributions. Thank you
01:16:09so much. There's a picture
01:16:10of me with with the other
01:16:12fellows behind their
01:16:13strategically placed it
01:16:15over there for. They are
01:16:17good zoom work OK,
01:16:20so I guess I will introduce the next speaker.
01:16:27How do I do this? Alesso do we? So if
01:16:33crystal just goes ahead
01:16:34and shares her screen.
01:16:37Myself Crystal yeah yeah,
01:16:39so can you go to the two?
01:16:42Can you go to the green share screen
01:16:45button at the bottom of the window?
01:16:48You're my screen. Yep,
01:16:49great, I assume OK, it is my
01:16:52great pleasure also to introduce
01:16:54Doctor Crystal Mackall.
01:16:55She's currently the earnest and Amelia
01:16:58Gallo family professor of Medicine and
01:17:00Pediatrics at Stanford University.
01:17:02She's also the founding director of
01:17:04the Stanford Center for cell therapy.
01:17:07Prior to this, she was at the
01:17:10National Cancer Cancer Institute,
01:17:11initially as a fellow in pediatric
01:17:14oncology and remained there until 2016,
01:17:16serving as the chief.
01:17:18Of Pediatric Oncology Branch.
01:17:19Doctor Michael is a leading pioneer in
01:17:22the development of cancer immunotherapy's
01:17:24in the pediatric population.
01:17:26Her group was among the first to
01:17:29impressive the impressive activity of
01:17:31the CD 19 Chimeric Antigen Receptor
01:17:33in childhood leukemia and also
01:17:35developed a car product targeting
01:17:37CD 22 and is active in this disease.
01:17:41The title of her talk is next generation
01:17:44car T cells to overcome resistance.
01:17:47Well
01:17:48thank you Tristan and it's
01:17:50a real pleasure to be here.
01:17:53Many valued friends and colleagues at
01:17:55at Yale and clearly a center that is
01:17:59contributed greatly in always to our
01:18:01understanding of human immunology.
01:18:04So it's great to speak with you all today,
01:18:08and even even better than I have the honor
01:18:12and pleasure or following Steve Rosenberg,
01:18:15who is clearly a giant in this field and.
01:18:19You know, in fact,
01:18:21the reason I got involved in cancer
01:18:24immunotherapy back in the 80s was due
01:18:27to inspiration from Steve's work.
01:18:29So it's always great to hear his updates.
01:18:33So here are some of my disclosures
01:18:36and I guess I'm going to focus
01:18:40on car T cells today,
01:18:42but hopefully can provide some stories
01:18:45that will give value to understanding
01:18:48of General T cell Biology that might
01:18:51be relevant even beyond car T cells.
01:18:54And, as Steve alluded to following the
01:18:58first demonstration of the effectiveness
01:19:00of the CD 19 car in large cell lymphoma.
01:19:04At the surgery branch.
01:19:07The decade of remarkable progress.
01:19:10A Kurd,
01:19:11culminating in FDA approval of the
01:19:13first cell therapies and the first gene
01:19:16therapies approved in the United States.
01:19:19Kymriah forbis OLL and yes Carta
01:19:21for large B cell lymphoma.
01:19:23These were the first out of the
01:19:26gate and and you know this is
01:19:29an immense success story that is
01:19:32just catalyzed the whole field.
01:19:34As an academic in the space,
01:19:37what what we see as our job at
01:19:40Stanford is to understand the failures
01:19:42that remain in this space because
01:19:45despite all of the progress for
01:19:48both the children with beef lol.
01:19:50Elan or adult patients who get treated
01:19:53with these agents were still under
01:19:5650% for long-term Disease Control.
01:19:58And it is our belief that we can use
01:20:02this model system to really dive deep into.
01:20:06The factors that are required for the
01:20:09success, particularly in car T cells,
01:20:11and that these insights will fuel
01:20:14our work in going tord solid tumors.
01:20:17I do disagree with Steve that Cartis
01:20:20Osar Fatalii flawed for solid tumors,
01:20:23and I hope to prove him wrong.
01:20:26In my in the coming years.
01:20:30So as far as yescarta for B cell lymphoma,
01:20:33as I mentioned there,
01:20:34about 60% of the patients who go on to
01:20:38progress after receiving other therapy.
01:20:40And the interesting thing is that
01:20:42most of the action nearly all of the
01:20:45action happens in the first six months.
01:20:48If you are still in remission
01:20:50after six months,
01:20:51we have this really amazing long
01:20:53tail that looks to be curatives
01:20:55and so it helps in terms of
01:20:58understanding the those who are.
01:21:00Successes versus failures.
01:21:01By focusing on that six month time point,
01:21:04Anne and there are clearly a lot of
01:21:06ways that car T cells can fail here.
01:21:09From a recent review were sort
01:21:11of highlighting those that we
01:21:13think are the most important tumor
01:21:16heterogeneity and antigen loss
01:21:17is an area of great focus for us.
01:21:19I'm not going to talk about it today,
01:21:22but Suffice it to say that clearly
01:21:241/3 of the yescarta failures or
01:21:26associated with complete loss of CD
01:21:2819 and another third are associated
01:21:31with significant down regulation.
01:21:32But the other piece of this is really
01:21:35the functionality of the T cells,
01:21:37and we've spent a lot of time in that area,
01:21:41so I want to start.
01:21:42The first story is to talk about
01:21:44a cohort of patients treated with
01:21:46yescarta at Stanford on the commercial
01:21:49receiving the commercial product,
01:21:50and you know, the sad truth is,
01:21:53you can't get your hands on
01:21:55the product for those patients
01:21:56because of the contract you sign.
01:21:59You really aren't permitted to go
01:22:01into the bag, but what we can do?
01:22:04As we can study these patients
01:22:05after they receive their commercial
01:22:07product and try to understand
01:22:09what distinguishes those patients,
01:22:11who, for whom.
01:22:12The treatment is a success versus for
01:22:14those whom they treatment fails them.
01:22:17And this is our outcomes for
01:22:19the Yescarta Cohort.
01:22:20This is 32 patients that
01:22:22we are working with here,
01:22:23and the purpose of this story
01:22:26and you can see it.
01:22:27It exactly mirrors the Zuma one data
01:22:30and if you ask the question first,
01:22:33well maybe it just relates to
01:22:35how well the car T cells expand.
01:22:37In our hands we do not see a correlation
01:22:40between the area under the curve.
01:22:43In other words,
01:22:44the height and the duration of
01:22:46expansion and overall outcome that
01:22:48does correlate with neurotoxicity,
01:22:49but we do not see it correlating with
01:22:52overall outcome and so we asked the question,
01:22:55might we be able to dive deeper
01:22:58into these populations that expand
01:23:00after infusion and at an early time
01:23:02point be able to predict what's
01:23:04going to happen at six months,
01:23:06and so to do this.
01:23:08We've leveraged the power of mass cytometry,
01:23:11a technique that has been developed
01:23:14at Stanford and optimized,
01:23:15and there's lots of expertise
01:23:17in our community here,
01:23:19and I have the pleasure on working
01:23:21on this project with a very
01:23:24talented computational urologist,
01:23:25azina good and So what Xena asked,
01:23:28was whether she could identify
01:23:30early after infusion and now
01:23:32she is talking about day seven.
01:23:35So seven days after infusion she
01:23:37wants to understand whether there
01:23:39might be an early predictive
01:23:41biomarker that can be identified.
01:23:43That might predict what happens
01:23:45to these patients six months down
01:23:47the line and using flow cytometry.
01:23:50With these targets,
01:23:51she identified a total of 10
01:23:53meta clusters on day seven,
01:23:55and each of these meta clusters can be
01:23:57shown here visually as encompassing.
01:24:00You know there are subsets
01:24:02of meta clusters as well.
01:24:04You can see meta cluster 3 here
01:24:06has two subsets, for instance,
01:24:08and each of these subsets is defined
01:24:11by its own special combination.
01:24:13Of the cell surface and
01:24:15intracellular markers.
01:24:16Now of course,
01:24:17you can't do this with the
01:24:20without some computation,
01:24:21computational assistance,
01:24:22and so there are variety
01:24:24of these models out there.
01:24:26The lasso analysis was
01:24:27developed at Stanford to try to
01:24:30identify those clusters which are
01:24:32associated with a particular outcome,
01:24:34and again, here were asking whether
01:24:36we can identify an early biomarker
01:24:39associated with complete response
01:24:41at six months and the Lasso decided
01:24:44that three meta clusters were.
01:24:46The sweet spot at which you could
01:24:48get your highest level of accuracy,
01:24:51and so indeed what we've identified
01:24:53is there are three meta clusters
01:24:55associated with a positive outcome.
01:24:57The 1st two are Co expression of the
01:25:00marker CD57A marker that has been
01:25:02associated with senescence in T cells,
01:25:05and this is true whether it
01:25:07is a neither car T cells.
01:25:10Now we're not talking about non car T cells,
01:25:13these are only the car T cells on Day 7.
01:25:17And if you have more CD4 or CD8 cells
01:25:20that express CD 57 on day seven,
01:25:22you're much more likely with
01:25:24these very strong P values to be
01:25:26in remission at six months.
01:25:28There was a third meta cluster
01:25:29which had the inverse relationship,
01:25:31wherein if you had expansion of these cells
01:25:34you were less likely to be in remission,
01:25:37and that is a meta cluster that
01:25:38is CD 4 positive pelayos positive.
01:25:41So now the power of the
01:25:43of the mass cytometry is.
01:25:44You can learn a lot about
01:25:46these meta clusters.
01:25:47By looking at whatever your favorite
01:25:50antigen of interest is and Steve
01:25:52mentioned CD39A marker of T cell
01:25:54exhaustion and not surprisingly,
01:25:56these cells are CD 39 negative,
01:25:59but surprisingly they are 57 positive,
01:26:01which is typically been associated
01:26:03with senescence in the spent cell.
01:26:06And so I think this was really
01:26:09a surprise to us.
01:26:10They tend to express some PD one
01:26:13you can see here the Meta cluster
01:26:16that is the Helios positive is.
01:26:19The one that is likely to be a T Reg.
01:26:22So now you know it's nice if you
01:26:24can do this by mass cytometry,
01:26:27but the truth is mass cytometry is not
01:26:29a particularly iaccessible technology,
01:26:31and So what Xena wanted to do is
01:26:33now armed with these insights.
01:26:35Should could she go back to simple
01:26:37flow cytometry and create a simple
01:26:39flow cytometry assay that might be
01:26:41able to be used widely to identify
01:26:44those patients who are likely to
01:26:46have good versus negative outcomes,
01:26:47and indeed a relatively simple
01:26:49flow cytometry panel was.
01:26:50Able to identify these patients
01:26:52as shown here.
01:26:53Here's an example patient with a
01:26:56complete response and you can see
01:26:58how how many car T cells all these
01:27:01are car T cells now expressed.
01:27:03CD 57.
01:27:04They're also Tibet positive and
01:27:06the same for the CD 8 subset.
01:27:08Here's an example of a patient
01:27:11with progressive disease at six
01:27:13months and you can see this patient
01:27:15had very little of the CD.
01:27:1757 positive cells,
01:27:18but in contrast had a high number.
01:27:21Of cells that Express Helios and these were
01:27:23all clearly statistically significant.
01:27:26Now,
01:27:26what are the functionality of these cells?
01:27:29So now we go back and we start
01:27:31to stimulate the specific cells
01:27:34with either PMA and ionomycin,
01:27:36or through the car itself,
01:27:38and perhaps not surprisingly,
01:27:39given what we know about CD 57 we see that
01:27:44these cells express very high levels of
01:27:46grandson be they don't make very much.
01:27:49I'll two and it's true whether you.
01:27:52Simulate with PMA,
01:27:53ionophore or through the car.
01:27:55The Helios positive cells,
01:27:56not surprisingly, however,
01:27:58are not granzyme B positive and
01:28:00look for all the world like T regs.
01:28:03Now we can use an even deeper
01:28:06dive using that NX platform.
01:28:08Now we're doing single cell RNA
01:28:10seq with site seek to really try
01:28:13to confirm more about these cells
01:28:15and it really all panned out the
01:28:18way we expected because here now
01:28:21using site seek you can define.
01:28:23The cells that are CD 57 positive
01:28:26CD 4 positive and now you can look
01:28:29at the RNA seq for each of these
01:28:32populations shown here on you map plot.
01:28:35So here are the T Reg sitting up here.
01:28:39Here are the CD 4 positive in the blue
01:28:4257 positive in the CD 857 shown in Green.
01:28:46You can see that the whereas the T
01:28:49regs are a very narrow population.
01:28:52The 57 is really distributed throughout.
01:28:55A more broad based population,
01:28:57so here are your CD four cells.
01:29:00You've got the T regs here that are
01:29:03Fox P3 positive helio cells are not
01:29:06expressing TCF one nor expressing CD 39,
01:29:09so this is exactly what you
01:29:12would expect from a T Reg.
01:29:14Here are your CD 8 cells that are
01:29:17expressing CD 57 also expressing Tibet,
01:29:20the enzyme that gives rise to
01:29:23the CD 57 epitope.
01:29:25Granzyme B and also talks,
01:29:27although talks really was not
01:29:28a good discriminator here,
01:29:30and so these cells do not appear
01:29:32to show a features of exhaustion.
01:29:35And then if you dive even deeper into
01:29:38these cell populations now we can
01:29:40start looking at the degree to which
01:29:43these cells are cycling and the CD.
01:29:4557 positive cells is shown here on
01:29:48a new map showing the cell cycling
01:29:50are really showing lower levels
01:29:53of cell cycling than than on 57.
01:29:55Positive self so they really are showing
01:29:58features that would be associated
01:30:00with senescence already at day seven
01:30:02in the blood of these patients.
01:30:04Not surprisingly,
01:30:05there are clonally expanded as
01:30:07shown in the orange and blue,
01:30:09looking at clones that are expanded
01:30:12greater or less than 10% compared to the
01:30:15non expanded in the grey and the blue.
01:30:18So this is been a surprise.
01:30:21It's not what we expected,
01:30:23but when you take these high dimensional
01:30:26single cell immune profiling,
01:30:28sometimes I think this is the
01:30:30power of it that without the.
01:30:33Burden of a hypothesis.
01:30:35Sometimes you find things you don't expect,
01:30:37and in fact what we found was that
01:30:40higher numbers of CD 57 positive
01:30:42car T cells that look like they
01:30:45have a senescent phenotype on day
01:30:47seven predict a favorable outcome.
01:30:49Now what do we think is happening here?
01:30:52We think that when you have these cells,
01:30:55what this is telling you is
01:30:58that the product that
01:30:59was infused. Contained potent T cells
01:31:02that were able to become activated
01:31:04that the tumor expressed adequate
01:31:07antigen expression to be able to drive
01:31:09expansion of your car T and that the
01:31:12micro environment of the lymphoma
01:31:14was also receptive and so in some
01:31:17ways this is a biomarker we believe
01:31:20for all of the factors that line
01:31:22up to allow a productive response.
01:31:24The car T regs are another interesting
01:31:27twist we had not seen any evidence of car.
01:31:30T regs this.
01:31:31Bite lots of looking for them
01:31:34in our manufactured products,
01:31:36and so it suggests to us that maybe
01:31:39these are being induced in vivo and I
01:31:42think this is an area where we need to
01:31:46understand more why some patients do
01:31:49have this predisposition to generating
01:31:51T regs from their CD 19 car product.
01:31:54It's possible that with this in
01:31:56hand we can intervene earlier for
01:31:59these patients to improve outcomes.
01:32:02Alright,
01:32:02so now let's move onto another
01:32:05story and this is really trying to
01:32:07get to this issue that Steve was
01:32:10alluding to that you know car T
01:32:13cells so far in solid tumors have
01:32:15not demonstrated reliable activity
01:32:17and there are a lot of reasons for that.
01:32:20As I've alluded to,
01:32:22the tumor heterogeneity problem,
01:32:23difficulties with trafficking
01:32:24and also the fact that the solid
01:32:27tumors really induce substantial
01:32:28amount of T cell dysfunction,
01:32:30much of which is characterized
01:32:33by T cell exhaustion.
01:32:34Now,
01:32:35we also believe that this is compounded
01:32:37with car T cells because car T cells
01:32:39themselves are predisposed to exhaustion.
01:32:41And why do we believe that?
01:32:43Well,
01:32:44this is work that we did while I was
01:32:46still at the MCI when we were trying
01:32:49to induce car T cells that could
01:32:52induce regression of an osteosarcoma,
01:32:54and we wanted to,
01:32:55you know.
01:32:56Only look to see if we had a car
01:32:58that we knew was highly efficacious
01:33:00and that really is the CD.
01:33:0219 car up to now it remains kind of
01:33:04the gold standard in the field and
01:33:06we wanted to understand whether you
01:33:08could regress a solid tumor with the
01:33:11hostel micro environment with the car.
01:33:12If you had a good target and so we
01:33:15express CD 19 and Osteosarcoma,
01:33:17and indeed we saw a regression.
01:33:18But when we tried to target a second
01:33:21antigen GD 2 which is also expressed
01:33:23on osteosarcoma or car T cells,
01:33:25had no effect,
01:33:26and.
01:33:26And what we ended up learning was that the
01:33:29problem was not at the level of the inogen,
01:33:32but rather at the level of the car,
01:33:34and in this case the GD2 car,
01:33:37like many cars,
01:33:38tended to aggregate on the surface
01:33:39of the cells due to hydrophobic
01:33:42regions in the single chain FV,
01:33:43which led to signaling even in
01:33:45the absence of Antigen and when
01:33:48T cells received too much signal
01:33:49for too long of a time they become
01:33:52exhausted and So what we learned
01:33:54from this work was it this is an
01:33:56Achilles heel of car T cells that.
01:33:59Unlike the CD 19 car,
01:34:01which really does not show a
01:34:03propensity for tonic signaling,
01:34:04the vast majority of other car T cells
01:34:07do this to some extent, and the GD
01:34:10two was a particularly egregious example,
01:34:12and so we we tried to fix it.
01:34:15We've had trouble doing it to be honest and
01:34:18retaining the antigen binding properties,
01:34:20but but we thought we might be able
01:34:23to turn lemons into lemonade by simply
01:34:26using these highly tonically signaling
01:34:28cars to begin to study the biology.
01:34:30Of human T cell exhaustion.
01:34:32Because the truth is we didn't have a
01:34:35good model of human T cell exhaustion.
01:34:38We were really relying mostly
01:34:40on LC MGMV models in mice,
01:34:42which are of course very valuable
01:34:45but may not reflect what happens
01:34:47in human cells and and using a
01:34:50mutated version of the GD2 car we
01:34:53were able to see that within 12
01:34:55days we could take perfectly normal,
01:34:57naive healthy human T cells and convert them.
01:35:01Two full blown exhausted cells
01:35:02and using the CD.
01:35:0419 cars are controlled group.
01:35:05We had really good controls and so
01:35:08this was work led by Rachel Lynn in
01:35:10the lab and it shows you some of the
01:35:14evidence for this exhaustion in a dish model.
01:35:16The exhausting cells don't grow as well.
01:35:19They don't make much in the way
01:35:21of I'll two and they also have
01:35:23diminished interferon production.
01:35:25They terminally differentiate and
01:35:26they express all of the hallmark cell
01:35:28surface markers you would express.
01:35:30Expect of an exhausted cells.
01:35:32If you look at the transcriptome they
01:35:36really nearer the transcriptome of
01:35:38the T cells that are exhausted in the
01:35:41LC MV model or in let's say humans
01:35:44with hepatitis or other chronic viral
01:35:47infections and so now after validating
01:35:50that this is a valid exhaustion model,
01:35:52we set out to understand the biology
01:35:55a bit better and using a taxi quicker
01:35:58the genome wide approach to look
01:36:01at enhancer availability,
01:36:02we were able to identify that the
01:36:05greatest difference between exhausted.
01:36:07And non exhausted T cells.
01:36:09In terms of the epigenome,
01:36:12was the increased availability of
01:36:14enhancers that turn on the AP one
01:36:17transcription factor family and we
01:36:19saw dramatic overexpression of the
01:36:22AP one transcription factors here,
01:36:24and this was the puzzling result
01:36:26because of course we all think of
01:36:29AP one at least faucet in June as
01:36:32Canonical transcription factors
01:36:34that drive tso activation.
01:36:37So why would this be implicated
01:36:39in exhaustion? Well?
01:36:41It turns out there are a lot of
01:36:43other members of this family,
01:36:46and many of them actually induces
01:36:48suppressive transcriptional program
01:36:50and so we it LED us to the hypothesis
01:36:52that perhaps there was an overexpression
01:36:54of the exhausting or suppressive AP
01:36:57one family members in a relative
01:36:59deficiency of the activating AP,
01:37:01one of Foss in June and,
01:37:03and so we tested this by overexpressing
01:37:06Foss in June.
01:37:07Foss had no effect,
01:37:08but overexpression of Thi June in
01:37:10these exhausting T cells.
01:37:12Had a significant impact on the
01:37:14ability of these cells
01:37:15to perform most notably here with I'll 2.
01:37:18Now we also did the opposite experiment
01:37:20because if it was really an imbalance.
01:37:23Then, perhaps we could also restore the
01:37:25balance by using crisper to knockout.
01:37:27These inhibitory transcription factors
01:37:29and we found indeed that to be the case,
01:37:32especially June be bad.
01:37:33If 3 and IRF 4.
01:37:36And so now what we showed,
01:37:38and this is detailed in this nature paper.
01:37:41I know lots of data in there
01:37:44if you're interested.
01:37:45But basically in many models where
01:37:47we look if we simply overexpress
01:37:49see June and car T cells,
01:37:52we're able to dramatically change
01:37:54the potency of car T cells,
01:37:56especially against solid tumors.
01:37:58Here's an example using, uh,
01:38:00her two car against in Osteosarcoma,
01:38:02and you can see the dramatic regression when,
01:38:05see June is incorporated
01:38:07into the car construct.
01:38:08We see increased numbers of
01:38:11tumor infiltrating cells.
01:38:12We see decreased exhaustion expression.
01:38:14We see decreased terminal differentiation
01:38:16an we see that the say cells retain
01:38:20Poly functionality here increased
01:38:22out to production and also increased
01:38:25TF and Interferon gamma production.
01:38:27So what is the C Jun doing?
01:38:31Well,
01:38:31it's inducing really widespread
01:38:33transcriptional reprogramming in these cells.
01:38:35You can see that here are the June
01:38:39overexpressing cells compared
01:38:40to the HA control cells.
01:38:43You can see all of these cells
01:38:46associated with Stemness.
01:38:47All of these genes associated stemness,
01:38:49are overexpressed,
01:38:50and all of these genes associated
01:38:53with exhaustion or under expressed
01:38:55but interesting Lee.
01:38:56The exhaustion associated
01:38:58footprint did not change.
01:38:59We saw no difference in the
01:39:02exhaustion associated footprint,
01:39:03and so it was interesting because
01:39:06this demonstrated you could
01:39:07divorce the transcriptome from
01:39:09the epigenome of these cells,
01:39:11but it also.
01:39:12Raised the question,
01:39:14can you reverse the the epigenetic footprint?
01:39:17And so we've been working on
01:39:19another approach and this has been
01:39:20led by Evan Weber in the lab who
01:39:22has been looking to regulate car
01:39:24expression and chords a lot of folks
01:39:27want to regulate car expression
01:39:28mostly because they are thinking
01:39:30that this will help car T cells
01:39:32be more safe that we can tune the
01:39:34car T cell response and diminish
01:39:36the cytokine release syndrome
01:39:37and some of the other toxicities
01:39:39that have been observed in there,
01:39:40of course,
01:39:41numerous ways to do this one that we
01:39:43developed in our lab with Evan is
01:39:45this degron based system where you.
01:39:47Attach a peptide that tags the car
01:39:50for degradation in the proteasome,
01:39:52but the peptide is also druggable and
01:39:54can be inhibited with a small molecule
01:39:57and this works really quite well.
01:39:59You can see that with this drug on system.
01:40:03If you give the drug,
01:40:04and in this case,
01:40:06some of these are very antibiotics
01:40:08that are readily available.
01:40:10You can see a high levels of surface
01:40:13car expression versus when you take
01:40:15the drug away it degrades and it.
01:40:18And it works through the biologically
01:40:20relevant level of expression.
01:40:22You can see here.
01:40:23The difference in aisle 2.
01:40:25Another way to do this.
01:40:27However, rather than just engineering
01:40:29a specific decron to simply to use
01:40:32small molecules that inhibit cortisol
01:40:34signaling and we discovered that dissent
01:40:36nib is one of these small molecules,
01:40:39which can really quite potently as you
01:40:41can see here at modest concentrations can
01:40:44quite potently inhibit car activation
01:40:46and we think that this happens through.
01:40:49Inhibition of LCK now the beauty
01:40:52of this at neighbors.
01:40:54It's very reversible.
01:40:55So here you can have this at nib you know.
01:40:59In in you know,
01:41:01inhibiting you're killing but then,
01:41:02when you wash it out.
01:41:04You get the reverse so
01:41:06the disat nib inhibits,
01:41:07killing but when you wash it out.
01:41:10You can get the reverse so it's
01:41:12really quite a nimble system so
01:41:14whatever has been trying to do now
01:41:17is to look in the exhausting model
01:41:19and ask whether you could take a
01:41:22cell that is now already exhausted.
01:41:24It's already gone down the path
01:41:26remember what June was doing it
01:41:28with preventing exhaustion were now
01:41:29allowing cells to become exhausted.
01:41:32And then we're giving them a period
01:41:34of rest and we want to know can
01:41:37you reverse the exhaustion and
01:41:39the data with this at nib and with
01:41:42the degron model is really very
01:41:44impressive here you can see sales,
01:41:46now in a day 25 assay where you have
01:41:48high levels of exhaustion markers,
01:41:51but when you incorporate the fat nearby
01:41:54there from Day 4 or you can see from
01:41:57Day 7 Day 11 Day 14 or even Day 18.
01:42:00You start to see.
01:42:01Now reversal of these checkpoint
01:42:03molecules an acquisition now.
01:42:05Of your stem like molecules and
01:42:06even a change in the transcriptome
01:42:08and this is associated now with
01:42:11improved functionality and are
01:42:12feeling from these data were that
01:42:14there wasn't a point of no return.
01:42:16We simply got a more potent effect if
01:42:19the rest was allowed to go on for a
01:42:22longer period, but but we didn't know.
01:42:24Maybe there was a point of no return.
01:42:26So now, what Evan is done.
01:42:28His take this all the way out.
01:42:31Today,
01:42:3153 and you can see that applying
01:42:34the Saturn IB as late as Day 46.
01:42:36Is still enough to retrieve the
01:42:38cytolytic capacity of these cells
01:42:40and to greatly improve their
01:42:42ability to produce cytokinin OK,
01:42:44so now let's go back to the transcriptome.
01:42:48What's happening here?
01:42:49It's quite similar to what
01:42:51I showed you with June.
01:42:53You get massive transcriptional rewiring.
01:42:55Now.
01:42:55Another interesting piece of these
01:42:58experiments is Evan also used to
01:43:00control where he used anti PD one
01:43:03and just to make it clear anti PD
01:43:06one didn't do anything in terms
01:43:08of functional re invigoration,
01:43:09immodest invigoration.
01:43:10Of cytolytic capacity,
01:43:11but no ability to improve
01:43:14proliferation or or cytokine,
01:43:15and not surprisingly,
01:43:16therefore,
01:43:17PD one did not reverse the transcriptome,
01:43:20but the the rested period did
01:43:22reverse the transcriptome and we
01:43:24saw an owl acquisition of left
01:43:26one TCF 7 and diminished I RF 4
01:43:30so that's perhaps not surprising,
01:43:32but what was really notable
01:43:34in surprising was the degree
01:43:36to which rest is now reversing the epigenome.
01:43:39So here now is the A Taxi Cavan.
01:43:42Always off car you can see very
01:43:45different from the always signaling car
01:43:47and again PD 1 does not change this.
01:43:50But when you give these cells periods of
01:43:52rest you see epigenetic reprogramming
01:43:54and this isn't simply selection of a
01:43:57rare population that wasn't exhausted.
01:43:59We can show that by looking at
01:44:01the TC are receptors and it shows
01:44:04no evidence for clonal expansion.
01:44:06But we also see this by using test meta
01:44:09stat because what we see now is that
01:44:12this reprogramming is dependent on.
01:44:14EV H2 Mediated Trimethylation because
01:44:16it can be inhibited by an inhibitor.
01:44:20OK, so if you've been following
01:44:22the car field,
01:44:23you've seen that there are a lot
01:44:25of regulatable cars out there,
01:44:27and there are many ways to do this,
01:44:29and I think you know a lead candidate
01:44:31in terms of optimal platform
01:44:33has not yet been identified.
01:44:35Louis lemania is a very talented bio
01:44:38engineering graduate student in the lab,
01:44:39and he's been working on a regulatable
01:44:42system and he wants to use an FDA
01:44:44approved small molecule to be
01:44:46able to regulate these car T cells
01:44:48and so at Looe is done is use a.
01:44:51Protease that is a derived from hepatitis
01:44:54C that at baseline essentially cleaves
01:44:56the signaling domain off of the car.
01:45:00But when you apply now the small molecule,
01:45:03you prevent cleavage and this
01:45:05is shown on a western.
01:45:07You can see the full length car with the
01:45:10drug present becausw the proteases inhibited,
01:45:14but the absence of the full length
01:45:17car when the drug is absent,
01:45:19and we know that in this car.
01:45:22There is tonic signaling,
01:45:24and so it tends to express leg
01:45:26three and again with drug on.
01:45:28We get the lag three expression with drug
01:45:31off, we don't, so it's a drug on system.
01:45:34You're using a hepatitis C based antiviral,
01:45:36and when the antiviral drug falls below,
01:45:38then the car is degraded.
01:45:40Anne,
01:45:40I'm not going to show you today
01:45:42in the interest of time,
01:45:44but this model does show Efficacy in
01:45:46reversing toxicity in animal models,
01:45:48but perhaps more interesting,
01:45:49Lee is the question is,
01:45:51does this enhance efficacy and
01:45:53what Luis is found?
01:45:54Is that in multiple systems?
01:45:56That compared to a constitu tive on
01:45:58car which we just show here in black,
01:46:02the on off system using the antiviral,
01:46:04and he gives the antiviral daily to the mice.
01:46:08He gets better antitumor effect.
01:46:10We see this with a GD2 car with a
01:46:12B7H3 car against modulo blastoma with
01:46:15uh her two car against Osteosarcoma.
01:46:18The only exception is the CD 19 car
01:46:21and we believe that this is a car that
01:46:24doesn't tonically signal and doesn't.
01:46:27Rapidly acquire hallmark features
01:46:28of exhaustion,
01:46:29and so when you look at the fells in
01:46:32these animals that are receiving the
01:46:35Snip Trans car and the hepatitis C antiviral.
01:46:39Compared to the constitu tive again,
01:46:41what you see is a retention of more of a
01:46:45stemness profile, whereas the
01:46:47constitu tive cars rapidly terminally
01:46:50differentiate and again you can
01:46:52use this single cell RNA seq data
01:46:55to show a more in depth view.
01:46:57But there are no surprises here.
01:47:00Again, the Constitu tive cars
01:47:02are in pink and they are here,
01:47:05they lack I'll 7 receptor.
01:47:07They lacked 7 but the Snip based car
01:47:10retains a population that is stemlike
01:47:13but also is able to transition
01:47:16to the Granzyme B expressing.
01:47:19Factor which is so critical because stem.
01:47:21This is important,
01:47:22but the cells also need to be able
01:47:25to traffic through the process
01:47:27of differentiation to become
01:47:28full blown effectors.
01:47:30So conclusion from this T cell exhaustion
01:47:32occurs commonly in car T cells,
01:47:34and we believe it's a major
01:47:37factor limiting excess success,
01:47:38especially in solid tumors.
01:47:39See June overexpression
01:47:41endows exhaustion resistance,
01:47:42both in tonically signaling card
01:47:43details and also a non tonically ones.
01:47:46I didn't show you that today,
01:47:48but it's in the manuscript.
01:47:50And this occurs by potent transcriptional
01:47:53reprogramming that does prevent
01:47:55many of the hallmark features,
01:47:57but it doesn't alter the
01:48:00epigenetic footprint.
01:48:01In contrast,
01:48:02transient rest induces
01:48:03transcriptional reprogram but also
01:48:05the epigenetic reprogramming through
01:48:07a process that requires the Z,
01:48:09H2 and we find therefore,
01:48:11that regulated car T cells which
01:48:14were initially developed to
01:48:16prevent toxicity also show enhanced
01:48:19potency and we believe.
01:48:20That this result from the transient Reston
01:48:24tuning that's occurring due to the.
01:48:26The various system,
01:48:27so I want to.
01:48:28I think I've given credit to the
01:48:31people who did the work along the way.
01:48:33We've got a fantastic team at Stanford,
01:48:35a great clinical team that is
01:48:37conducting clinical trials alongside
01:48:39this fundamental work and also want to
01:48:41highlight the important collaboration
01:48:42with Howard Chang on this work.
01:48:44And these are our funders,
01:48:46and many of the folks so I can stop there.
01:48:49Thank you.
01:48:53Great, thank you so
01:48:54much doctor Mackle
01:48:55for that marvelous talk,
01:48:56I'm going to start asking the questions.
01:48:59That's all the Q&A.
01:49:00This is by Adam Rubin is the phenotypic
01:49:03diversity of Carti at day seven of
01:49:06function of the cells taken from the
01:49:08patient as as I can you predict the
01:49:11effectiveness of the transfusion
01:49:12product early on in the process.
01:49:15Yeah, I think as I as I alluded to,
01:49:18we would love to be able to go back
01:49:22to the product and the A Pheresis and
01:49:24we have work ongoing to do that with
01:49:27our investigator initiated trials.
01:49:29This is one of the problems
01:49:31with the commercial product.
01:49:33We are not permitted to analyze the
01:49:35A Pheresis or the product itself.
01:49:38It really negates our contract.
01:49:39So much for science there,
01:49:41so I can't answer the question.
01:49:45OK, this next question is from doctor Nick
01:49:48Josie is also one of our speakers later on,
01:49:51and faculty member here at Yale
01:49:53High Crystal CD 57 in CMD is
01:49:55associated with senescence,
01:49:57but is also representative of a functional
01:49:59effector response mediated by CD 50.
01:50:01Seven negative T cells because
01:50:03the immune system control CMP,
01:50:05do you think the presidents,
01:50:07the presence of CD 57 positive car T
01:50:09cells is telling us more that you're
01:50:12getting a good effector T cell pool?
01:50:15Which is the precursor cell for
01:50:17the city of positive self. This
01:50:19is exactly what I think is happening.
01:50:21I'm surprised it's already
01:50:22there at day seven, day seven.
01:50:24You've already taken a cell all away too.
01:50:26That's in essence, into me.
01:50:28It means they've gotten to the tumor.
01:50:30They found an antigen,
01:50:31and they've done what they need to do.
01:50:34And now they're back in the blood.
01:50:36I mean, this is the only
01:50:38way we can explain it.
01:50:39It is not what we expected.
01:50:41We expected to find a stem like phenotype.
01:50:44You know, all of these beautiful.
01:50:46You know, stem cell memory markers,
01:50:48and it's not what we found.
01:50:51OK, this next question is by Sue Keck.
01:50:54the CD 57 CD, 8 positive T
01:50:57cells also express CX3 CR 1.
01:50:59Ann are mostly found in blood.
01:51:02Do you think this migration pattern
01:51:04is one of the reasons why this
01:51:07populations associated with CR?
01:51:09Sue, I don't know. I I.
01:51:12My guess is you have a better
01:51:14understanding of that than I do.
01:51:16Maybe I will look into
01:51:18it. OK, and
01:51:20I think this will be our last question.
01:51:23Have you thought of a role for
01:51:25surface receptor signaling expense?
01:51:26Specifically CD 28 that may be further
01:51:29enhancing the tonics signaling driven by
01:51:31the car. Yeah, oh boy, so we,
01:51:33you know I'm not sharing that today,
01:51:36but we've spent so much time thinking
01:51:38about these proximal signaling features
01:51:40because one of the I mean in this gets
01:51:43to Steve's point that his concern that
01:51:45you won't ever find a truly tumor
01:51:48specific service cell surface molecule.
01:51:49I don't think you will,
01:51:51but I think what we have our therapeutic
01:51:54windows with car T cells and our
01:51:56work is shown very clearly that you
01:51:58need high levels of Antigen for car
01:52:00T cell to become fully activated.
01:52:02There are numerous examples.
01:52:03The GD two is a perfect one where
01:52:06we have that.
01:52:07In the clinic we get good expansion
01:52:09and we don't see any toxicity even
01:52:11though there is low levels on the on.
01:52:14The neural tissues peripherally
01:52:15and Centrally so,
01:52:16so why is that?
01:52:17And one of the things we've learned
01:52:19is that the transmembrane domain of
01:52:21the car is a really important feature.
01:52:24Determining the degree to which
01:52:26the proximal signaling that the
01:52:27strength of signal needed to activate
01:52:29the proximal signaling apparatus
01:52:31and a CD 28 transmembrane domain
01:52:33lowers your antigen threshold.
01:52:34And it appears that this may be related
01:52:36to interactions with the native CD 28.
01:52:38So yes,
01:52:39I think all of this could be very important.
01:52:41And then when you add tonic signaling to it,
01:52:44clearly the 28 transmembrane cards are
01:52:46more likely to Tonic Tonic Lee signal
01:52:48and this may all be part of that,
01:52:50and whether that's a good thing
01:52:51or a bad thing,
01:52:53you know it's just now
01:52:54that you understand it.
01:52:55You can use it as part of your engineering.
01:52:59Thank you so much, doctor, Michael.
01:53:01I think you've gotta break for our
01:53:04lunch period now and then will
01:53:07see you will see everyone again
01:53:09and for the afternoon session we
01:53:11look forward to seeing everybody.
01:53:13Thanks so much trust and also for moderating.
01:53:17I think there's a great session.
01:53:19Thanks doctor Rosenberg
01:53:20Crystal as well doctor McCall.
01:53:2311 oh 1:15 so just 15 minutes eat quickly.
01:53:25I know you just have to go to
01:53:28your refrigerator for at home,
01:53:29so we'll see you back.
01:53:31And we have a great session
01:53:33where halfway through Pam Sharma
01:53:34will lead us off at 1:15.
01:53:36So please check back in then thanks so much.
01:53:39See you soon.
01:58:12Good markets how are you Jim
01:58:14says hello as well thanks a bunch
01:58:17often in his office over there.
01:58:19I'm sure things
01:58:20were crazy down in Houston as well, you
01:58:23know, so you know it gets crazier
01:58:25by the minute just because it's
01:58:27Texas and it's it's just not
01:58:29complying with the way we think, but.
01:58:32Hopefully we're trying to
01:58:33survive as we all are,
01:58:35I think, yeah, so I think we'll
01:58:38try to get ourselves going here.
01:58:40So so for all of you who are back,
01:58:44which is actually the majority,
01:58:46you can always check attendance
01:58:48pretty quickly and zoom.
01:58:49At least by computer,
01:58:51but the next section would be
01:58:53session 3 and we have three speakers,
01:58:56a couple from Yale an we have
01:58:58Pam Sharma who our moderate
01:59:00are Cellino will introduce.
01:59:01Kelly is an assistant professor in surgery
01:59:04who I worked
01:59:05with very closely on
01:59:06the Melanoma unit has been a
01:59:08great addition to Yale a couple
01:59:11years back. Has experience in
01:59:12designing and evaluating immune therapies
01:59:14and it's very actively involved in both
01:59:17our clinical Melanoma unit as well as
01:59:19trying to advance anti cancer
01:59:21immune therapies. At Yale,
01:59:23so Kelly feel free to start.
01:59:26At its my absolute pleasure to
01:59:28introduce doctor Pam Sharma,
01:59:30who is a model for what all physician
01:59:33scientists really aspire to be.
01:59:35She's been highlighting
01:59:36major discoveries beginning
01:59:38in 2003 while she was still
01:59:40a fellow with
01:59:41looking at the New York new NYSE one.
01:59:44T cell epitope while she was still a
01:59:48fellow excellent Kettering and then
01:59:50some of her seminal work,
01:59:52and identifying Icos is
01:59:53expressing T cells as being part
01:59:56of the mechanism. Actions
01:59:57he till 8
01:59:58four. As you can see here,
02:00:00she's a scientific director
02:00:02of the immunotherapy platform
02:00:03at MD Anderson.
02:00:04She's a professor of Gu medical oncology
02:00:07there, and also the Co director
02:00:09of the Parker Institute and the
02:00:11Pi of many, many trials.
02:00:12And you know, really was a
02:00:14visionary in doing what we now
02:00:16call window of opportunity trials and
02:00:19trying to study while people are on
02:00:21therapy to figure out why things work,
02:00:23why they don't work. So again,
02:00:25we're all delighted to have you here.
02:00:28And welcome from another Queens
02:00:29girl who was in Texas and then
02:00:32came back home to the East Coast.
02:00:34So you're speaking about but what my
02:00:36whole family wants me to do is come back,
02:00:39but I'm having a lot of fun in Texas
02:00:41too and I'm really grateful for this
02:00:44opportunity to present some of our work.
02:00:46Thank you so much for that kind
02:00:48introduction an I am going to be
02:00:50talking a lot about the clinical
02:00:52trials today and how we interrogate
02:00:54the clinical trials in the laboratory.
02:00:56Let me see advancing here we go.
02:00:58So as many of you are aware,
02:01:01anti see teleport clearly open an entire
02:01:03new field called immune checkpoint
02:01:04therapy that goes without saying that
02:01:07the paradigm shifted as a curd in the
02:01:09last decade has been monumental in
02:01:11terms of the advances that we've seen
02:01:13not only for anti see till 8 four,
02:01:15but other immunotherapy agents.
02:01:17And of course the clinical benefit
02:01:19for all of our patients.
02:01:20So this cartoon is a bit outdated,
02:01:22but I will highlight sitali for
02:01:24here and PD one and PDL one which.
02:01:27Is the other agent actually anti
02:01:29PD one and anti PD L1 antibodies?
02:01:31Both agents that are now in an
02:01:33FDA approval Arsenal for patients
02:01:35with cancer as well.
02:01:36Obviously there are a lot of other
02:01:38targets that are being explored
02:01:40and we'll talk about some of them,
02:01:42but it's impossible to talk about
02:01:44all of them clearly,
02:01:46but these are other targets now
02:01:47that have come to the forefront
02:01:49of regulating T cell responses
02:01:51within the tumor micro environment.
02:01:53As I mentioned,
02:01:54the last decade has seen many
02:01:56many FDA approvals.
02:01:57Obviously a pillow Melbourne Sisi
02:01:59telephone was in 2011 and was the
02:02:01first but I just wanted to give
02:02:03everybody a broad overview of all
02:02:05of the approvals that have been in
02:02:08Kering across multiple tumor types,
02:02:09because really these agents are
02:02:11targeting the immune response and
02:02:13they don't really target anything
02:02:14related to the cancer cell itself and
02:02:17that makes it applicable then to many
02:02:19different tumor types and possibly
02:02:21for many different combinations.
02:02:22Strategies as well talk about.
02:02:25So the research questions though,
02:02:26that come to mind as we take
02:02:28care of these patients in clinic,
02:02:30is that clearly some patients are
02:02:32responding and others are not.
02:02:33So why is this happening?
02:02:35Are there biomarkers to predict
02:02:36a response and immune related
02:02:38toxicities that are associated with
02:02:40these agents and I'm sorry I won't
02:02:42be able to talk about the artwork
02:02:44on immune related toxicities today,
02:02:45but happy to take questions if
02:02:47someone has them.
02:02:48Are there biomarkers to enable
02:02:49patient selection for treatment with
02:02:50monotherapy versus combination?
02:02:52Can we increase the number of
02:02:53patients who respond and are there
02:02:55other pathways that can be targeted
02:02:57to improve clinical outcomes?
02:02:58So in order for us to try and answer
02:03:00some of these questions really we
02:03:02have to integrate the laboratory
02:03:04in the clinical research in a very
02:03:06efficient and rapid manner.
02:03:08At this point we have about actually 3000,
02:03:10so this slide is also a little bit outdated.
02:03:13About 3000 clinical trials,
02:03:14ongoing with either anti sitella.
02:03:16For anti PD one or anti PD L1 and
02:03:17so so clearly the clinical aspect
02:03:20has outstripped the basic science.
02:03:22I mean we know that the basic signs
02:03:24led to the development of this field
02:03:26as many of you know that the mice in
02:03:29the lab or in bread and the disease is
02:03:31homogeneous and that leads to great.
02:03:33Hypothesis testing that we can then take
02:03:36the data from the laboratory to the clinic.
02:03:38But as I mentioned,
02:03:39the clinic is now outstripping
02:03:41our scientific understanding,
02:03:42so we're really dealing with a
02:03:44lot of clinical data.
02:03:45And how do we get that back to the lab?
02:03:49So really, we're dealing with
02:03:50patients who are polymorphic.
02:03:52The disease is heterogeneous
02:03:53and this is called hypothesis
02:03:55generating data from the clinic,
02:03:56so we have to take this hypothesis,
02:03:58generating data back to the
02:04:00laboratory design the appropriate
02:04:01models in the laboratory to then
02:04:03test the hypothesis rigorously.
02:04:05So how do we do that in the clinic?
02:04:07Well,
02:04:08we have to rethink clinical trial design
02:04:10for one and many of you are aware that
02:04:12clinical trial design release phase
02:04:14one safety dose escalation phase,
02:04:16two efficacy in phase three
02:04:17comparison to standard of care.
02:04:19What we proposed was conducting
02:04:20these pre surgical or tissue based
02:04:22clinical trials which we call
02:04:24phase one or phase two a studies.
02:04:26And really these trials will
02:04:27allow us to not only have clinical
02:04:29signals but also to have biomarker
02:04:31analysis and mechanistic insights.
02:04:33And we started these studies in 2004
02:04:35and reported on our first trial.
02:04:37I'm around 2006 and the paper
02:04:39was published in 2008,
02:04:40but since then it's really taken
02:04:42off as a way for us to interrogate
02:04:44the immune responses in humans.
02:04:46Because these responses are
02:04:47slightly different.
02:04:48As you can imagine,
02:04:49then mice and also gives us a chance
02:04:52to study the lanja tude inal responses
02:04:54because in humans this is a dynamic
02:04:56process occurring multiple multiple
02:04:58stages of the disease as well as
02:05:01after multiple different treatment
02:05:02regimen that a patient can have.
02:05:05So the first trial we conducted
02:05:06in this way is shown here,
02:05:08and this again,
02:05:09this protocol is written in 2006,
02:05:11and as you can see,
02:05:12it's a small clinical trial.
02:05:14These do not need to be large
02:05:15trials as we think about phase
02:05:17two and Phase Three Studies.
02:05:19These are small trials.
02:05:20This was a 12 patient trial in patients
02:05:22with localized bladder cancer who
02:05:24were already scheduled for surgery,
02:05:25and we administered two doses of anti
02:05:28sitella for antibody prior to surgery
02:05:29so that we can have access to all of
02:05:32the tumor material at the time of
02:05:34surgery for the laboratory studies.
02:05:35But the trial gave us a lot of information.
02:05:38For one,
02:05:39it gave us a clinical signal for safety.
02:05:41So if you think about it,
02:05:43this was the first neoadjuvant
02:05:45clinical trial with immune
02:05:46checkpoint therapy conducted in 2006.
02:05:48Prior to any FDA approvals.
02:05:50And so it told us that we
02:05:51can actually give
02:05:52immune checkpoint therapy
02:05:53prior to surgery and now,
02:05:55of course there are multiple
02:05:57neoadjuvant clinical trials ongoing
02:05:58with immune checkpoint agents.
02:06:00It also gave us a clinical
02:06:02signal for efficacy,
02:06:03because this was the first clinical
02:06:04trial in patients with bladder cancer.
02:06:07And so clinical trials were on going
02:06:09in Melanoma but not in bladder cancer.
02:06:11And so three patients in this bladder
02:06:13cancer study actually had pathologic
02:06:14complete response is where all
02:06:16disease went away and the pathologist
02:06:18could not find any remaining tumors.
02:06:19So this was an indication that bladder
02:06:21cancer was going to be responsive
02:06:23to immune checkpoint therapy.
02:06:24And of course we design those studies
02:06:26later in the meta static setting
02:06:28for FDA approval in bladder cancer
02:06:30or from the laboratory standpoint,
02:06:32we really had a lot of access.
02:06:34Now to these large tumor samples that
02:06:36were taken in large amounts of cells.
02:06:38For all of the assays that we were proposing,
02:06:41so without looked at the pre
02:06:43treatment and I should point out
02:06:44here that we had some pre treatment
02:06:47but for those of you who take care
02:06:49of bladder cancer patients you know
02:06:51the pretreatment is very small.
02:06:52Biopsies,
02:06:53sewer pretreatment cohort actually was
02:06:54an untreated cohort that was staged,
02:06:56matched and went directly to
02:06:58radical cystectomy.
02:06:58So they did not receive the anti.
02:07:00See Tilly for drug and we were able
02:07:02to use that released comparison for
02:07:04posttreatment samples and you can
02:07:06see in the posttreatment samples we
02:07:08had lots of infiltrating lymphocytes.
02:07:10And these staying with CD3CD4 CD
02:07:128 Granzyme indicating T cells on
02:07:14activated T cells with Granzyme,
02:07:16and we also found B cells which
02:07:19are CD 20 positive is shown here.
02:07:22And when we compare the pre an
02:07:24post treatment or to untreated
02:07:26and posttreatment samples for
02:07:27differentially expressed genes,
02:07:28obviously with lots of lymphocytes
02:07:30infiltrating into tumor,
02:07:31we had lots of signaling pathways that
02:07:33were related to the immune response.
02:07:35What was surprising to us was that I
02:07:37cost was the top pathway listed here.
02:07:40So I cast had not been studied in
02:07:42human immune responses before and so
02:07:44we were puzzled by this and wanted
02:07:47to look at it a little bit deeper.
02:07:49So I cross again is inducible
02:07:51costimulator it belongs to the
02:07:53CD 20 agency tally for family is
02:07:54shown in this phylogenetic tree.
02:07:56Its expression.
02:07:57It has been known to maybe
02:07:59increase on activated T cells,
02:08:00however it has a very diverse
02:08:02role that's been reported,
02:08:03including a roll on regulatory T cells.
02:08:05Follicular helper T cells,
02:08:06and no role really an antitumor responses.
02:08:09So because we had access to
02:08:10all of these tumor tissues,
02:08:12now we can ask the question about
02:08:14what was happening with this icos
02:08:16positive subset in the setting of anti.
02:08:18See telling for treatment and
02:08:19you can see in nonmalignant
02:08:21tissues which we had access to.
02:08:23There were very few or about 13% of the CD.
02:08:26Four cells expressed I costs.
02:08:27Any untreated tumor tissues were
02:08:29somewhat similar in about 16%
02:08:30of the CD. Four cells expressing high costs,
02:08:32but after treatment with anti CD like
02:08:34for all of our patients, had this
02:08:37increase in the eye cast positive CD,
02:08:39four subset and some of our patients also
02:08:41had an increase in the eye cause positive
02:08:44CDs upset that I'm not showing here.
02:08:46At the same time, Jadwal Chuck was
02:08:48conducting a phase three clinical trial
02:08:50in patients with metastatic Melanoma,
02:08:51and so we had access to some of the blood.
02:08:54Samples of Judd had collected
02:08:56from those patients,
02:08:57and we could look to see whether or
02:08:58not I cast correlated with outcome,
02:09:01and you can see here to patients
02:09:02who had increased the levels
02:09:04and sustained levels of icons.
02:09:05Positive CD 4T cells had much better
02:09:07survival compared to patients who did not,
02:09:09so we also did another clinical trial.
02:09:12I want to point out here.
02:09:13This is an anti sitali 4 plus anti PDL
02:09:15one since of course combination therapy
02:09:17has been the way to move forward.
02:09:20This paper was just published
02:09:21about a week ago,
02:09:23but this is the combination neoadjuvant
02:09:24trial now and this is the first
02:09:27combination neoadjuvant trial
02:09:28in patients with bladder cancer.
02:09:29And again we were able to show that
02:09:31the trial led to patients having these
02:09:34pathologic complete response is not
02:09:36only in all patients who completely surgery,
02:09:38which are 24 patients.
02:09:39We had a 37.5% pathologic complete response,
02:09:42but we also found that in patients
02:09:45with 3D masses or T for a disease
02:09:47were very these patients do very
02:09:49poorly and I should point out that.
02:09:51All of these patients were also
02:09:53cisplatin ineligible so they
02:09:55could not receive chemotherapy.
02:09:56We also found a pathologic complete
02:09:59response for these patients of 42%
02:10:01and what we found that correlated
02:10:03with these outcomes you can see
02:10:06here the biomarkers of response
02:10:08were really related to T&B cells,
02:10:10and this was known as
02:10:12tertiary lymphoid structures.
02:10:13Again,
02:10:13in the Association between the
02:10:15T&B cells so the patients who had
02:10:17pretreatment samples with higher
02:10:19density of tertiary lymphoid structures
02:10:21did better in terms of responders.
02:10:24And again when we compare pre and
02:10:26post treatment samples is shown here.
02:10:27Also the patients who had an increase
02:10:29in the icons positive CD 4T cells
02:10:31within the tumor micro environment.
02:10:33Those patients did better in
02:10:35terms of responses as well.
02:10:37So these clinical data of course let
02:10:39us generate the hypothesis shown here,
02:10:41one of them being that the icon cycles
02:10:43legal pathways necessary for effective
02:10:45antitumor immune responses in the
02:10:47setting of anti seating for therapy.
02:10:49And to test this hypothesis we
02:10:51actually went back to the laboratory
02:10:53and we could end look at wild type
02:10:56mice as well as I cast knockout mice
02:10:58and I costly good knockout mice and
02:11:01wildtype mice injected with Melanoma
02:11:03cells could reject these tumors
02:11:04very well with anti see telling
02:11:06for therapy with 80 to 90% of the
02:11:09mice having long term survival.
02:11:10Where is the icons,
02:11:11knockout mice and Icos ligand knockout
02:11:13mice had impaired antitumor
02:11:14responses and only about 40% of
02:11:16these mice could reject or tumors.
02:11:19The second hypothesis that we had
02:11:21was that the icon Psychostick in
02:11:22pathway can be targeted and developed
02:11:24as a combination therapy strategy.
02:11:26And again we went back to doing this
02:11:28in wild type mice where we could
02:11:31target both icos ansi tally for
02:11:33at the same time and in the blue
02:11:35dotted line you can see combination
02:11:37therapy had significantly improved
02:11:38the responses in this model compared
02:11:40to any of the monotherapy ardion
02:11:42treated mice and the same experiments
02:11:44conducted in icos knockout mice.
02:11:45Of course,
02:11:46we lost the ability to have this response
02:11:48and so these data were published and
02:11:50now multiple companies have anti icos.
02:11:52Antibodies that they have in clinical
02:11:55trials and GlaxoSmithKline just reported
02:11:56on their data as as more this year,
02:11:59showing clinical responses
02:11:59with their antibody.
02:12:00So now I just want to switch gears
02:12:03a little bit because all of that
02:12:05data we did a while ago and I'm
02:12:07looking forward to clinical data
02:12:09with the anti icos antibodies but
02:12:11we had other questions we wanted to
02:12:14ask as well because as you can see
02:12:16from the list that I showed with
02:12:18the FDA approvals immune checkpoint
02:12:20therapy has been approved in lung
02:12:22cancer and Melanoma bladder cancer,
02:12:23head and neck cancers.
02:12:25And from this figure that I'm
02:12:27showing on this slide,
02:12:28these tumor types are known to be hot tumors,
02:12:31meaning they have lots of mutations
02:12:32and a result of lots of mutations.
02:12:35They have neo antigens that
02:12:36can be recognized by T cells,
02:12:38and so they have lots of infiltrating
02:12:40T cells within the tumor micro
02:12:42environment and therefore can respond
02:12:43to immune checkpoint therapy,
02:12:45which is, you know, puzzling tests,
02:12:47because, again,
02:12:47all tumor should have antigens,
02:12:49because all tumors are made up
02:12:51with some mutations,
02:12:52and so prostate cancer,
02:12:53which is shown here is considered
02:12:55to be a cold tumor.
02:12:56Because it has very few mutations,
02:12:58it does not have many infiltrating
02:13:00T cells and there have been 2 failed
02:13:02phase three clinical trials with
02:13:04antisec selling for in prostate cancer.
02:13:06So as a clinician I agree that
02:13:08immune checkpoint therapy is not
02:13:10working on these cold tumors,
02:13:12but as an immunologist.
02:13:13It puzzles me because the cold tumor
02:13:15should still also have antigens and
02:13:17T cells only really need one antigen
02:13:19before they can proliferate and expand.
02:13:21So we wanted to ask the question
02:13:23whether or not prostate cancer the
02:13:25antigens on prostate cancer really
02:13:27not being recognized by the T cells.
02:13:29Is there an issue with?
02:13:31Antigen recognition
02:13:31and in order to
02:13:32do that, we conducted this small
02:13:34clinical trial in 30 patients where
02:13:36we had metastatic castration resistant
02:13:38prostate cancer patients where we
02:13:40can respect one of the metastatic
02:13:41lesions or to primary prostate tumor.
02:13:43We can perform X omen RNA sequencing
02:13:45to identify to tumor mutations and
02:13:46then in the setting of giving these
02:13:49patients anti sitali for therapy.
02:13:50We can then take T cells and ask
02:13:52whether or not these T cells are
02:13:55capable of recognizing the antigens.
02:13:57Again, this paper was just recently
02:13:59published a couple of months ago so
02:14:01I won't go into all of the details,
02:14:03but I do want to show you this one.
02:14:06Figure when patients 7 where
02:14:08this patient had two mutations,
02:14:09one in Rogue wanting nucleotide exchange
02:14:12factor 37 and one in Dihydropyrimidine
02:14:14is shown here and these were single
02:14:17amino acid changes for these mutations
02:14:19and you can see that this patient
02:14:21did not have any detectable T cell
02:14:24responses in the Elispot assay
02:14:26at the pretreatment timepoint,
02:14:27which is which is noted.
02:14:29Noted here is pre AP but after giving
02:14:32anti see teleforce at the post AP
02:14:34one posted between posted before.
02:14:36Now you can see that these T
02:14:38cells are quite clearly capable of
02:14:40recognizing the mutated antigen,
02:14:42but not the Wild Type Antigen,
02:14:44and this is shown in relationship to
02:14:47negative into positive control here.
02:14:49So clearly the T cells are capable
02:14:51of recognizing these mutations
02:14:52and prostate cancer.
02:14:54So then the question becomes,
02:14:55is it possible T cells are not
02:14:57infiltrating into the prostate tumors?
02:14:59Is that the reason that prostate cancers
02:15:01are not responding to immune checkpoint
02:15:03therapy and in another clinical
02:15:05trial that we'd previously conducted,
02:15:07we were able to look at the
02:15:09data from these 20 patients,
02:15:11who then underwent antisec delay
02:15:12for therapy prior to surgery,
02:15:14and we had access to all of
02:15:16the tumor tissues.
02:15:17So then look to see what was happening
02:15:20in terms of tumor infiltrating T cells.
02:15:22And you can see here in the pretreatment
02:15:25samples there are very few T cells,
02:15:27of course,
02:15:27so this is really a cold tumor
02:15:29and I don't have the slide to
02:15:32show comparison to Melanoma.
02:15:33But if you were to compare prostate Melanoma,
02:15:35you would note that this is a very
02:15:37cold tumor with very few infiltrating
02:15:39T cells prior to treatment.
02:15:41However, after treatment,
02:15:42you can see now lots of infiltrating T cells,
02:15:45as shown here by the ihcc study.
02:15:47What we should have been prepared for,
02:15:49but we were in is that the immune
02:15:51response is tightly controlled.
02:15:53So if you drive it in One Direction,
02:15:56there will be compensable Tori pathways
02:15:57that are up regulated in order to
02:16:00control the response, and so on.
02:16:02Our gene expression studies with this
02:16:04clinical trial we found that PD L1
02:16:06in Vista was highly expressed as a
02:16:08result of treatment with anti sitali
02:16:10for so not only were We driving the
02:16:12T cells in but now we were having
02:16:15compensable Tori inhibitory pathways and
02:16:17just to show that we were able to
02:16:19confirm the gene expression data by HC
02:16:22in the protein expression is shown here.
02:16:24Prior to treatment, there really is
02:16:26no PD one PD L1 or Vista Expression.
02:16:28However, after treatment,
02:16:29I see telephone now that we
02:16:31have infiltrating immune cells,
02:16:32you can see PD one is highly
02:16:34expressed on the immune cells PDL.
02:16:36One is on the immune cells
02:16:38an on tumor cells as well.
02:16:40And now Vista is also on the immune
02:16:42cells and I should point out if
02:16:44this is a novel immune checkpoint.
02:16:46There was identified by Randy Noel's group.
02:16:48I'm just showing you hear the
02:16:50pre and post treatment samples.
02:16:52Pretreatment is in the black triangle
02:16:54and post treatment interent circles.
02:16:56You can see the PDL one is highly
02:16:59expressed now after true or the anti
02:17:01setelah foreign CD8T cells on CD.
02:17:0368 myeloid cells and on tumor cells
02:17:05and similarly for Vista you can city
02:17:08to CD4 and CD8T cells do have some
02:17:10Vista expression but Vista is really
02:17:13predominantly expressed on my lawd
02:17:14cells here so we wanted to compare
02:17:17to Melanoma and prostate samples.
02:17:19Since we all know when we give
02:17:21anti see teleported Melanoma.
02:17:23These patients have nice responses.
02:17:24However,
02:17:25when we give antisec teleporter
02:17:26prostate tumors.
02:17:27As I mentioned,
02:17:28there were two failed phase
02:17:29three clinical trials,
02:17:30so in Melanoma samples,
02:17:32what we found was we also had
02:17:34myeloid cells that were infiltrating
02:17:35as a result of treatment.
02:17:37But these myeloid cells tend
02:17:39to have an M1 Gene signature,
02:17:41meaning that they are participating
02:17:43in the antitumor response and
02:17:44leading to tumor rejection.
02:17:46However,
02:17:46in prostate tumors we found an M2
02:17:48gene signature meaning these myeloid
02:17:50cells are more immunosuppressive and
02:17:52we're still trying to figure out the
02:17:54signaling mechanisms that would allow.
02:17:56In one tumor micro environment from
02:17:57my lawd cells to be M1 and it had a
02:18:01different tumor micro environment for
02:18:02Milo excels to BM2 service and PDL.
02:18:04One are definitely potent inhibitors
02:18:06of human T cell responses.
02:18:07We took the T cells from these
02:18:09patients and we placed them in
02:18:11vitro with anti CD 3 and you can
02:18:13see that they can produce interferon
02:18:15gamma and Tina Fafa very well.
02:18:17However if the plate is coated with PD,
02:18:20L1 IG or Vista idea two together
02:18:21the T cells are no longer capable
02:18:24of producing interferon,
02:18:25gamma and TNF Alpha as well.
02:18:27So really,
02:18:28they're being suppressed by
02:18:29the Vista in the PDL 1.
02:18:32With that data in mind,
02:18:33we convinced Bristol Myers script
02:18:35conduct a clinical trial with
02:18:37anti see Taylor for an anti PD.
02:18:38One is combination therapy in patients
02:18:41with metastatic castration resistant
02:18:43prostate cancer and for the first
02:18:44time now we can see as you can see in
02:18:47this PSA graph below that patients
02:18:48who had high PSA in metastatic
02:18:50disease as a result of treatment.
02:18:52Now the PSA became undetectable
02:18:54and CT scans then showed
02:18:55resolution of metastatic disease.
02:18:57Again this paper was just
02:18:58published about a week ago
02:19:00and so the data is all there.
02:19:02I will point out that the combination
02:19:04therapy did lead to higher toxicities
02:19:06as a result of combination treatment
02:19:08and so to study was recently expanded.
02:19:11Now with four different treatment arms
02:19:13so that we can look at different doses
02:19:15and schedules of the therapy to hopefully
02:19:18minimize the toxicity but maintain efficacy.
02:19:21One of the things that we noted in
02:19:23this clinical trial that we were
02:19:25conducting is that the patients who
02:19:27were responding to treatment with
02:19:29combination therapy for patients
02:19:30who had soft tissue metastases,
02:19:32whereas as many of you may know,
02:19:34prostate cancer predominantly goes
02:19:36to the bones and patients with
02:19:38bone metastases were having less
02:19:39responses from our observations.
02:19:41So we looked at the soft tissue
02:19:43metastases from patients and the bone
02:19:45metastases from patients and what we
02:19:47found is that after treatment with
02:19:49anti sitali for soft tissue metastases
02:19:51were found to have a TH one signature.
02:19:54Meaning lots of Affecter T cells
02:19:56and interferon gamma production was
02:19:58occurring within the soft tissue.
02:19:59However,
02:20:00in bone metastases we found that there
02:20:02was no increase in the TH one response.
02:20:05Instead,
02:20:05there was an increase in TH 17 cells,
02:20:08and so that was puzzling to us.
02:20:10Why the bone metastases had an
02:20:12increase in TH 17 but not each one.
02:20:15So we took the clinical data and we
02:20:17went back to the laboratory and we
02:20:19model this in my swear we could inject
02:20:22castration resistant prostate cancer cells.
02:20:25Into the subcutaneous lesions of
02:20:26enter the subcutaneous space of mice
02:20:28to represent the soft tissue lesions.
02:20:30Or we can inject the cells into bones
02:20:33to represent the bone metastatic lesions,
02:20:35and then we treated the mice would
02:20:37anti C telephone,
02:20:38anti PD one is shown here in the red
02:20:41and you can see that again when the
02:20:44bone lesions TH one responses did
02:20:46not increase but the TH 17 did as
02:20:49compared to subcutaneous lesions where
02:20:51you had significant increase in that
02:20:53each one response and no increase in
02:20:55actually decrease in TH 17 response.
02:20:57So again,
02:20:57this was mimicking what we were
02:20:59seeing in the humans,
02:21:00and for many of you,
02:21:02you may be aware to TH 17 cells really
02:21:05their skewed into development as a
02:21:06result of Isle 6 and TGF beta signaling.
02:21:09So in this model we were able to look
02:21:11at all of the different cytokines
02:21:14in the bone micro environment,
02:21:15and again this paper was published last year,
02:21:18so the details are there.
02:21:19But focusing in on the Isle 6 and TGF
02:21:22beta pathways we can see that I'll six
02:21:24is very highly increased in the bone
02:21:27micro environment as compared to Serum.
02:21:29And it's in the bones regardless
02:21:31with if there's a tumor or not.
02:21:33So BMT is with a tumor and
02:21:35BM is without a tumor.
02:21:37However,
02:21:37TGF Beta is only high as a
02:21:39result of the tumor being within
02:21:41the bone micro environment,
02:21:42and we found the TGF beta
02:21:44was really being made
02:21:45as a result of osteoclast activity
02:21:47within the bone micro environment,
02:21:49and so with I'll 6 and TGF beta
02:21:51is a two cytokines that are
02:21:53highly elevated in bowling.
02:21:54You can imagine now that this would
02:21:56skew the T cells towards a teach
02:21:5917 phenotype rather than TH one.
02:22:01And so we propose blocking the TGF beta
02:22:03with an antibody is shown here in the
02:22:06green and when we blocked the TGF beta.
02:22:08Now for the first time we had
02:22:10decrease in tumor volume and
02:22:12increased survival of these vice.
02:22:13More importantly we were able to
02:22:15show that by blocking TGF beta in
02:22:18combination with immune checkpoint
02:22:19therapy for the first time that we
02:22:21had expansion of the TH one CD4T
02:22:23cells within the bones as well as
02:22:25expansion of CD8T cell clones within
02:22:27the bones and so as a result of this
02:22:30now we have a new clinical trial
02:22:32is currently undergoing review to
02:22:34block TGF beta in combination with.
02:22:36With immune checkpoint therapy for
02:22:38patients with bone metastases.
02:22:39But the studies with bone metastases
02:22:41actually had us thinking that there
02:22:43were unique subsets developing
02:22:45in different niche is so,
02:22:47although we may think of patients
02:22:49with metastatic disease is oh,
02:22:51they all have metastatic disease.
02:22:52The site of metastasis becomes
02:22:54very important for how the immune
02:22:56response will develop,
02:22:58and so this was also something
02:23:00we wanted to look at for other
02:23:02unique niches such as the brain
02:23:04were glioblastoma can develop.
02:23:06And as you know,
02:23:07immune checkpoint therapy does
02:23:08not work for glioblastoma,
02:23:10so we wanted to understand if there
02:23:12was also another unique immune cell
02:23:14subset that that is a part of the
02:23:17GB M micro environment and here we
02:23:19compare GB M to non small cell lung cancer,
02:23:22renal cell carcinoma,
02:23:23colorectal cancer and prostate
02:23:24cancer from patients.
02:23:25And we also compared untreated
02:23:27glioblastoma samples to anti PD one
02:23:29treated glioblastoma samples from
02:23:31our patients and what we were able
02:23:33to identify was a unique subset of
02:23:35CD 73 positive myeloid cells as
02:23:37shown here which is the L8 subset
02:23:40in our site off analysis.
02:23:41And then again,
02:23:42this paper was published recently
02:23:44in this year, so this L.
02:23:458 so upset with the CD 73 myeloid
02:23:47cells were able to conduct a single
02:23:50cell RNA sequencing from the patient
02:23:52samples and we were able to show that
02:23:55this is a macrophage signature that
02:23:56is very immunosuppressive in terms
02:23:58of the genes being expressed as well
02:24:00as hypoxia associated based on that data.
02:24:03We then conducted trial,
02:24:04then conducted studies in a model
02:24:06of CD 73 knockout mice compared
02:24:08to wild type mice and in this
02:24:10glioblastoma model you can see
02:24:11that in the city 73 knockout mice.
02:24:14Now when we give anti PD one and.
02:24:16Stacy telling for we can have improved
02:24:19survival as compared to the wild type mice,
02:24:22and this led to discussions now with
02:24:24other companies were developing a
02:24:26clinical trial based on targeting
02:24:28CD 73 in combination with immune
02:24:31checkpoint therapy.
02:24:32So I just want to finish up on other
02:24:34targets that we're thinking about that may
02:24:36have importance in the immune response,
02:24:38such as epigenetic pathways.
02:24:39Many of you are aware that as we
02:24:42target CD 28 and T cell receptor,
02:24:44that's the way to turn on T cells.
02:24:46You need both those signals
02:24:47for T cells to be turned on.
02:24:49However, as I mentioned before,
02:24:51when you turn on T cells,
02:24:52they are tightly regulated and then
02:24:54they find different pathways in
02:24:56which to regulate their responses.
02:24:57And in this setting easy H2 is
02:24:59increased as well so that easy H
02:25:01Tuukanen stabilize the Fox P3 Gene.
02:25:03Stabilization of the Fox P3 Gene then
02:25:05allows for regulatory T cell function
02:25:07instead of effector T cell function,
02:25:09and so in our patients we saw that
02:25:11when we gave Anti Sitella for therapy
02:25:13in patients who are not responding
02:25:15as well dance I see telephone.
02:25:17We found that these patients
02:25:19had increased levels of easy H2
02:25:21expression in their CD 4T cells.
02:25:23So to understand whether or
02:25:24not targeting easy H2 would be
02:25:26worthwhile as a combination strategy,
02:25:28we went back to the laboratory and
02:25:30looked at the Fox P 3G FP Mice an in
02:25:33these mice you can insert the Fox P3.
02:25:35Sales based on GF P expression
02:25:37and now we can ask whether or not
02:25:40easy H2 inhibition in this setting
02:25:42would then change the phenotype
02:25:44and the function of these cells.
02:25:46And So what we found with easy age 2
02:25:48inhibition was that we decreased Fox
02:25:50P3 expression and we increase other
02:25:52genes such as SL10 and Interferon gamma.
02:25:55Shown here.
02:25:56Importantly, not not only to
02:25:58phenotype change without the easy H2,
02:25:59the function of these cells were
02:26:01really is regulatory T cells,
02:26:03so they could suppress effector T
02:26:05cells are shown in this CFC delusion.
02:26:07However, after easy,
02:26:09it's two inhibition.
02:26:10These T cells were no longer capable
02:26:12of suppressing effector T cells
02:26:14because they now had more of an
02:26:16effector T cell function themselves,
02:26:18and so based on that,
02:26:20in a preclinical model,
02:26:22we could show that with combining easy
02:26:24H2 inhibition with Anti Sitali for we
02:26:27now improved survival of the mice as
02:26:29shown here and decreased tumor volume.
02:26:31And this let us then to negotiate
02:26:34with two different companies.
02:26:35Daichi, sancho, for the easy H12 inhibitor.
02:26:38And Bristol Myers Squibb for
02:26:39their anti sitali for antibody.
02:26:41And we designed a new clinical trial.
02:26:43So this clinical trial is now enrolling
02:26:45patients with metastatic prostate cancer,
02:26:46bladder cancer,
02:26:47renal cell cancer.
02:26:48We've treated treated three patients to
02:26:49date and so far the therapies well tolerated.
02:26:52I hope to have more data in
02:26:54terms of Biomarkers and Efficacy
02:26:55as we continue this study,
02:26:57but I think this is a good way
02:26:59to show that we not only go
02:27:01from the clinic to the lab,
02:27:03but we also go back to the clinic
02:27:05as we use that data to design
02:27:07and next set of clinical trials.
02:27:10So just to finish up,
02:27:11that obviously there are many,
02:27:13many targets that need to be considered.
02:27:15A lot of these are in
02:27:17preclinical or clinical studies.
02:27:18I think Anti Sitali for an anti
02:27:20PD one or PDL one or definitely
02:27:23the backbone for immune checkpoint
02:27:25therapies that are ongoing right now.
02:27:27But we do have to really move quickly
02:27:29between the clinical trials and
02:27:31laboratory interrogation so that we
02:27:32can have data to generate rational
02:27:34combination strategy so that the
02:27:363000 ongoing clinical trials are
02:27:38not just being thrown together.
02:27:40Without any thought behind him
02:27:41from a scientific perspective,
02:27:42so we set up the immunotherapy
02:27:44platform at MD Anderson.
02:27:46It's guided by an umbrella protocol
02:27:48that I wrote that allows us to
02:27:50collect samples from any and
02:27:51all patients at MD Anderson's,
02:27:53regardless of whether they are on a
02:27:55BMX trial or a regenerx on trial or
02:27:58GlaxoSmithKline trial or Genentech.
02:27:59Trial doesn't matter.
02:28:00The patient samples can be collected
02:28:02on this lab protocol and we have about
02:28:04100 ongoing clinical trials across
02:28:0618 Department where we're collecting
02:28:08the samples in these patients,
02:28:09samples can be interrogated.
02:28:11So that we can learn something
02:28:12from the patients in order to then
02:28:15design the next rational study.
02:28:16So I just want to conclude with
02:28:19immune checkpoint therapy is clearly
02:28:20joined the ranks of surgery,
02:28:22radiation and chemotherapy is a pillar
02:28:24of cancer treatment and combination
02:28:25strategies are going to be the future,
02:28:28including combination
02:28:28strategies with surgery,
02:28:29radiation and chemotherapy.
02:28:31Also,
02:28:31multiple immune checkpoints exists
02:28:33and these are very dynamic in their
02:28:35expression and they have to be
02:28:37evaluate in both pre and on treatment.
02:28:39Human tumor samples in order
02:28:41to guide therapeutic decisions.
02:28:42The organ specific micro environment
02:28:44will need to be considered in
02:28:46order to understand me.
02:28:47Logic,
02:28:48subsets and subsequent immune
02:28:49responses against cancer cells
02:28:50in these organs and pre surgical
02:28:52and tissue based clinical trials,
02:28:54really do provide a feasable platform
02:28:56to study biological effects in
02:28:58patients which then provide insights
02:29:00into mechanisms that can be targeted
02:29:02for rational combination therapies.
02:29:03So I have many,
02:29:04many people to thank and my
02:29:06funding sources are listed here.
02:29:08I definitely do want to thank
02:29:09the patients though,
02:29:10because as you can imagine,
02:29:12patients with localized disease do not
02:29:13need to participate in clinical trials.
02:29:15They can go just directly to
02:29:17surgery and we've had lots of
02:29:18success in having these patients
02:29:20participate in our studies.
02:29:21And we're very grateful.
02:29:23So thank you and I'm happy to take questions.
02:29:26So
02:29:26I'll start you off with the first
02:29:29question you based upon the
02:29:30work and that you very elegantly
02:29:32described from the 2019 cell paper
02:29:35and that you saw in humans. But when
02:29:37you're doing these
02:29:38window of opportunity
02:29:39trials, and if it does
02:29:41turn out that all of these
02:29:43different meta static nitches
02:29:45which are actually
02:29:46in locations that are actually
02:29:48very difficult to get access to
02:29:50while people are on therapy at all,
02:29:52how do you? How do you think
02:29:55that you can circumvent that?
02:29:56Or what strategies do
02:29:58you think that you'll apply to?
02:30:00Towards that end. Yeah, so this.
02:30:02You know it takes a great
02:30:04deal of investments.
02:30:05I want to give our institution
02:30:07credit for putting millions and
02:30:08millions and millions of dollars
02:30:10into the immunotherapy platform.
02:30:11And all of the associated networks
02:30:13that need to support it, right?
02:30:15Because then we have dedicated
02:30:16interventional radiologists,
02:30:17dedicated surgeons and dedicated pathologists
02:30:19that work with us to try and Design.
02:30:21The best way to do this.
02:30:23So we do have access to
02:30:24them getting a soft tissue.
02:30:26Anna bone metastatic lesion from
02:30:28the same patient at the same time
02:30:30of the biopsy and those samples.
02:30:32To come directly to the laboratory in
02:30:34the you know the way we designed a tubes
02:30:36to be sent directly to the laboratory.
02:30:39For those kinds of analysis.
02:30:40So it does really take this team
02:30:42effort with a lots of support
02:30:44behind it so that the studies are
02:30:46being prioritized and scheduled.
02:30:48You know we have schedulers to
02:30:49our dedicated for the scheduling.
02:30:51These.
02:30:51We have coordinators who take care of these.
02:30:54We have, you know,
02:30:55lots of people who spend time
02:30:57with the patients.
02:30:58Getting the informed consents.
02:30:59Imagine instead of spending an hour to see.
02:31:02For patients in clinic,
02:31:03I get to spend an hour with one
02:31:05patient to explain all of these
02:31:06steps so you know the institution
02:31:08is losing clinical billing
02:31:10dollars 'cause they don't get to
02:31:11bill all my clinical time,
02:31:13but they're gaining on the research
02:31:15side and not really takes foresight
02:31:16and insight and leadership for the
02:31:18institution to have made that a priority.
02:31:20So I really think that's what makes
02:31:22it work from our standpoint.
02:31:25I have a question Kelly if that's OK.
02:31:28So yeah, so you know.
02:31:30I actually also have to applaud your
02:31:32institution for the support they've
02:31:34given for a window out because
02:31:36it's pretty unusual as you know
02:31:38to have that level of support an.
02:31:40I think this also Harkins back to the
02:31:43kinetic nature of response
02:31:44of hot versus
02:31:45cold where this not as pathologist do.
02:31:48I look at plenty of these responses.
02:31:51It's not you.
02:31:52It's not uniform overtime,
02:31:53so when you look is important
02:31:55and with that in mind,
02:31:56with these window opportunity trials
02:31:58you had like week seven after two cycles,
02:32:00an initial trials with you have paths
02:32:02ers which are useful in a certain sense.
02:32:05But if you want to actually check out
02:32:07the immune response in earlier time,
02:32:09point probably would be nice
02:32:11where you actually see some tumor
02:32:12cells left that are
02:32:14being killed. So
02:32:15do you have some insight as to what you like
02:32:18right now in terms of optimal time points?
02:32:20'cause it's variable? I get that.
02:32:22But what sort of the window arranged for you
02:32:25now? You asked the question that our
02:32:27students always ask if we're looking
02:32:29at the tumors just when we actually
02:32:31have tumors or we don't have to more,
02:32:33then we're really defining just a
02:32:35response that leads to tumor growing.
02:32:37Or Jim are going away, right?
02:32:38You're not getting that real lanja tude inal.
02:32:41So we have been designing
02:32:42experiments where we're getting now
02:32:44weekly samples from our patients.
02:32:45Of course, that got interrupted
02:32:47a bit by Kovid, and so you know,
02:32:49we're trying to get everything
02:32:51back online for having more,
02:32:52but we do have small numbers
02:32:54now in our cohorts.
02:32:55Where we have weekly samples and
02:32:57we are able to see that the other
02:32:59thing is because we're able to now
02:33:01prioritize single cell RNA seq.
02:33:02Because the assays have gotten so much
02:33:04better at doing that on the smaller
02:33:06sample size so that we don't have to
02:33:08ask our pathologists to give us everything.
02:33:11'cause That's also becoming an
02:33:12issue because we have to leave some
02:33:14for diagnostic path and figure
02:33:15out when we can get the sample.
02:33:17So I think single cell RNA seq is
02:33:19also enabling us to do those analysis
02:33:21with an assay that we can follow
02:33:23overtime on smaller amount of cells.
02:33:25So that's. Hopefully data I can show.
02:33:27I want to say the other thing I
02:33:29didn't show here is that we've
02:33:31also been using bone marrow samples
02:33:32to represent the immune response.
02:33:34Instead of trying to do a tumor biopsy
02:33:36trying to get it from the bone marrow
02:33:38because we think that obviously the
02:33:40immune response evolving and also
02:33:42the bone marrow microenvironment
02:33:43that some cells also infiltrate,
02:33:44not just bone metastases but just normal
02:33:46patients taking a bone marrow sample.
02:33:48So we've been trying to do that also
02:33:50in looking at that by sight off to
02:33:52see if we can then see those lanja
02:33:54tude onal because that does not
02:33:56require a scheduling an appointment
02:33:58with Interventional radiology.
02:33:59Bone marrow can be done directly in clinic.
02:34:01When we see the patient.
02:34:02So those are things that we're
02:34:04trying to come up with.
02:34:05Not perfect by any means.
02:34:07About one more additional question
02:34:08from Sally Church.
02:34:09So easy H2
02:34:10Inhibitors as you know,
02:34:11have a lot of off target effects
02:34:14of the question really is.
02:34:15What are some of the anticipated
02:34:17toxicities that you may see
02:34:19overlapping with combinations
02:34:19and you could say epigenetics and in general
02:34:22when you combining that with
02:34:23a drug like
02:34:24Apple in math, yeah so
02:34:26we thought we were going
02:34:27to see a lot. Actually the trial
02:34:29was written with three milligrams
02:34:31per kilogram of the Apolima,
02:34:33which as you know is the dose
02:34:35that was approved in Melanoma.
02:34:36But then in renal cell.
02:34:38Sorry, that's my new puppy in renal cell.
02:34:40The dose is 1 milligram per kilogram
02:34:43because of toxicity issues and in
02:34:45bladder in lung cancer had to be
02:34:47Q Six week dosing because the Q3
02:34:49week dosing so we actually expected
02:34:50to have a lot of toxicities and
02:34:52again just running the gamut of all
02:34:54of the toxicities associated with
02:34:56immune checkpoint that we can be
02:34:58exacerbated within easy H2 Inhibitors.
02:34:59So everything from colitis all the way
02:35:02to dermatitis we were expecting all of it.
02:35:04We have to say the 1st three patients we
02:35:06treated at three milligrams per kilogram.
02:35:09And they're doing well, you know,
02:35:10crossing my fingers that continues.
02:35:12We're only giving two doses of the flu map,
02:35:14not the four doses as approved.
02:35:16So maybe that's also helpful
02:35:18because we found a two doses works
02:35:20in our pre surgical trial,
02:35:21so we're only giving two doses.
02:35:23And the other thing is the protocol is
02:35:25written to have a dose deescalation
02:35:26from 3 milligrams to 1 milligram
02:35:28per kilogram of diplomatically.
02:35:30Do run into the toxicity issue.
02:35:32Like I said,
02:35:32we haven't yet,
02:35:33and I hope we don't because the
02:35:35three milligrams per kilogram
02:35:36really does give the best response
02:35:38in terms of CD4 and CD8 responses.
02:35:40So we're hoping to be able to keep the
02:35:42three milligram per kilogram dose Ng.
02:35:44But yes,
02:35:44toxicity is something we always
02:35:46have to be careful with when we
02:35:48designed a combination studies.
02:35:49Great, thank you.
02:35:51Thank you.
02:35:55OK.
02:36:00Doctor Joshi
02:36:01you Are you ready to share?
02:36:05My screen thanks Pam.
02:36:07So I have the
02:36:09pleasure of announcing one of my colleagues,
02:36:11doctor Nick Joshi, who is an assistant
02:36:14professor of immunobiology here at Yale.
02:36:16He had initially done his PhD and they
02:36:19were smart enough to re recruit him
02:36:22back after finishing his postdoc at MIT,
02:36:24he's been honored as a
02:36:26Damon Runyon cancer fellow,
02:36:28also by the lung Cancer Research Foundation,
02:36:30earning awards as well as some of young
02:36:33investigator awards and Immuno Oncology.
02:36:35Here I will say that Nick is
02:36:38a wonderful collaborator.
02:36:39He's praised for his teaching,
02:36:41his mentor ship.
02:36:42And his inside in a sought after
02:36:44to work on a number of different
02:36:47projects throughout Yale.
02:36:48And he's also done a lot of really
02:36:50wonderful work on creating some novel
02:36:52animal models that have the best name in
02:36:55the world, ninja which he just recently
02:36:58published in nature communications.
02:36:59And today he's going to talk to
02:37:01us about investigating T cell
02:37:03responses and engineered cancer models.
02:37:06Alright, can you guys see see the screen?
02:37:10You're all set now.
02:37:11OK, thanks so thanks for the introduction.
02:37:14As you mentioned,
02:37:14I'm going to be talking about our
02:37:16results looking at T cell responses
02:37:18in these engineered cancer models,
02:37:20so there's been a lot of introduction
02:37:22in terms of how T cells function
02:37:24in the context of tumors,
02:37:26and I'm going to not go through
02:37:28that again in any detail,
02:37:29but just highlight two basic points
02:37:31that I think are really important to
02:37:33understand what we are interested in.
02:37:35The first is that I think there's been
02:37:38a lot of work by many, many groups, and.
02:37:41Including clinical work that's really
02:37:42highlighted the important role
02:37:43that immune checkpoint receptors
02:37:45play in terms of driving an
02:37:47immunosuppressive microenvironment.
02:37:48That's present tumors and this has taught us,
02:37:51the importance of these receptors
02:37:53and actually suppressing ongoing
02:37:55and actively suppressing ongoing
02:37:56T cell responses and we give him
02:37:59you know therapy now.
02:38:00Now we can drive our these responses
02:38:02back into a more optimal zone
02:38:04and get more antitumor effects.
02:38:06However,
02:38:06the there's been a real rise as
02:38:08as more patients are being treated
02:38:10with these drugs in the development
02:38:12of immune related adverse events,
02:38:14and this is being seen now very frequently,
02:38:17especially as I'll show you in.
02:38:19Same with combination therapies,
02:38:20and so this is telling us, of course,
02:38:22that this isn't all good,
02:38:24that there is a negative consequences
02:38:26blocking these checkpoint receptors
02:38:28and that these checkpoint receptors
02:38:29really play a very active role.
02:38:31In in actually maintaining tolerance
02:38:33towards self and maintaining
02:38:34peripheral tolerance.
02:38:35And as I mentioned now
02:38:37with combination therapy,
02:38:38this is really increasing the severity
02:38:40and frequency of the adverse events
02:38:42that are that are observed in.
02:38:45This tells us that these inhibitory
02:38:47receptors play non overlapping
02:38:49functions in terms of how they
02:38:51regulate the response and so in my
02:38:53lab we're really interested in trying
02:38:55to understand this balance between
02:38:57peripheral tolerance and effector
02:38:59T cell responses in the context of.
02:39:01Cancer,
02:39:01and so we've been developing
02:39:03animal models and I'll tell you.
02:39:06To try and understand this balance,
02:39:08I'm not going to talk about today.
02:39:10We have a whole program and trying to
02:39:12understand how peripheral tolerance
02:39:14is set up and how breaks down in
02:39:16the context of immunotherapy with
02:39:18this idea that if we can identify
02:39:20those mechanisms,
02:39:21maybe we can increase this window
02:39:23in which you can make an optimal
02:39:26response by suppressing the role
02:39:27of lumen system in terms of driving
02:39:29autoimmune responses.
02:39:30On the flip side,
02:39:32we're very interested in how T
02:39:34cells function in the context of
02:39:35developing tumors and how they
02:39:37really are impacted by the immuno
02:39:38suppression that's present within
02:39:40the micro environment here I think
02:39:41like a lot of groups were really
02:39:43interested in trying to identify
02:39:45means that we can use to try and
02:39:47make T cells more resistance in the
02:39:49micro environment or understand
02:39:51the process by which cells undergo
02:39:53differentiation so that we can understand
02:39:54what are the signals that are driving
02:39:56these processes and and how to best
02:39:58manipulate those signals to try and
02:40:00get better therapeutic responses.
02:40:01So of course. To understand this problem,
02:40:03we really need a good animal model
02:40:05where we can study T cell responses
02:40:07and compare them between what's going
02:40:09on an anti cancer response and maybe a
02:40:11peripheral tolerance response and trying
02:40:13to understand how that balance is achieved.
02:40:15And it turns out that some of the models
02:40:17that we use in general have a very
02:40:19difficult time with the second part,
02:40:21the peripheral image in the peripheral
02:40:23tolerance mechanisms and so to
02:40:25illustrate that I put together kind
02:40:26of this very basic slide where I'm
02:40:28going to describe the immunologist
02:40:30favorite technique of taking model
02:40:31antigens where we understand what the.
02:40:33With the T cells,
02:40:34are there recognized those antigens and
02:40:36what we've been doing for decades really
02:40:38has been putting these antigens within
02:40:39the context of viruses or cancer cells,
02:40:41or even expressing them in self tissues.
02:40:43And when we do that,
02:40:45we can now in program them into the
02:40:47mouse and we can study T cell responses
02:40:49against these different conditions
02:40:51and this tells us a lot about how
02:40:53the same T cells are influenced by
02:40:55these different microenvironments,
02:40:55which is very powerful tool.
02:40:58However,
02:40:58the caveat has been really trying to
02:41:00understand these responses against self
02:41:02self antigens and really peripheral
02:41:03self tolerance and the reason for
02:41:05this is actually a beneficial thing.
02:41:07It turns out that all the antigens
02:41:10that your body or they are encoded in
02:41:12the genome or most of those images are
02:41:15also expressed in the famous by text,
02:41:17and then they're presented by dendritic
02:41:19cells within within the Medal of the
02:41:21Thymus and autoreactive thymocytes can
02:41:23be exposed to this dendritic cell and
02:41:25then undergo a tolerance mechanism
02:41:27where they're either deleted or there
02:41:29turned into regulatory T cells.
02:41:31And so without this intolerance mechanism,
02:41:32we would have these autoreactive
02:41:34thymocytes that would be present in the
02:41:36in the peripheral T cell repertoire.
02:41:38And we would actually be able to
02:41:40study how they undergo torrents
02:41:41in the context of self tissues.
02:41:44But because of this deletion process,
02:41:45we actually don't have this peripheral
02:41:47T cell pool,
02:41:48and this is great for blocking an immunity.
02:41:51But in terms of experimental models,
02:41:52it's actually quite a difficult thing
02:41:54because the antigen expression mechanism.
02:41:56So like I say lock style locks.
02:41:58Imagine or a 10 Dusable Antigen.
02:42:00These systems really just aren't
02:42:01tight enough.
02:42:02To keep antigen off in the context
02:42:04of the Thymus and so you do undergo
02:42:06these peripheral,
02:42:07you go under those central tolerance
02:42:09and that confounds our ability
02:42:10to understand peripheral talents.
02:42:11So when I was a postdoc in Tyler Jacks lab,
02:42:14we started to try to tackle this problem,
02:42:16trying to figure out how we could
02:42:18make a neoantigen inducible model that
02:42:20is really a little bit different in
02:42:22the sense that we're not suppressing
02:42:24the neoantigen then turn
02:42:25it on. We're actually creating a
02:42:27neoantigen through the process of
02:42:29induction and the way we came up with was.
02:42:31I have referred to earlier.
02:42:33It's called ninja.
02:42:34So we took a this this constructor
02:42:36I'm going to describe the details
02:42:39here get very complicated.
02:42:40Very quick, but the key thing is
02:42:42that it has an inducible Neo Antigen
02:42:44that is created through an inversion
02:42:46mechanism and so therefore it
02:42:48doesn't physically exist in the in
02:42:50the genome until you turn it on now.
02:42:53The genetics of this are involved
02:42:55to recombinases one called Flippo,
02:42:56which turns on the Neoantigen
02:42:58through this recombination event.
02:42:59And then there's another one
02:43:01called called pre that turns on
02:43:03the ability of this other part
02:43:04we call the regulatory module.
02:43:06To make the flippo that will
02:43:08then act on the neoantigen.
02:43:10So how does this differ from
02:43:12what people have done before?
02:43:13So as I mentioned,
02:43:15we create the neoantigen in
02:43:16the way that we do,
02:43:18that is by taking an antigen
02:43:20substrate which is recognized
02:43:22by T cells is a very common set
02:43:24of T cells that we studied both
02:43:26in chronic viral infection and
02:43:28often in tumor models as well.
02:43:30These are tells cells recognizing
02:43:31the GP 33 and GP 66 episodes
02:43:34with LC MV and so we put these
02:43:36epitopes into a DNA substrate.
02:43:38And then we used to splicing sites
02:43:40to actually create a central
02:43:41Exxon within the DNA encoding.
02:43:43This,
02:43:43this this antigen and the Central
02:43:45X on this pricing event allows us
02:43:47to actually invert the in central
02:43:49Axon so that in this off state
02:43:51the sequences that make up the
02:43:53image and are not continuous,
02:43:55so therefore they are skipped and
02:43:57really can't make a full image.
02:43:58And because of this skipping process
02:44:00so we want to turn on the antigen,
02:44:03we just need to flip this exon around
02:44:05and the way we do that is we use.
02:44:08Non compatible Fritz sites which
02:44:09are responsive to that recombinase,
02:44:11called flippo that I mentioned earlier.
02:44:13Flippo Axon,
02:44:13these recombination sites and makes
02:44:15a permanent version that lines up
02:44:17these splice sites and now you get
02:44:19the production of the new engine.
02:44:20We call this the inversion induced.
02:44:22Join neoantigen or ninja and ninja
02:44:24has one other really helpful feature
02:44:26in the fact that this this module
02:44:28is actually encoded within a GF
02:44:29molecule such that when you turn
02:44:31on the Antigen now you go from a
02:44:33GF negative state to a GF positive
02:44:35state and the antigens within
02:44:37the Gino GF so we can actually.
02:44:39Read out things like image and
02:44:41silencing directly just by looking
02:44:43at GSP fluorescence.
02:44:45So just to make things a little
02:44:47bit more complicated,
02:44:48also describe how this regulatory
02:44:50module works and this is in the
02:44:52allele just on the on the three prime
02:44:55end of the oleo and what it allows
02:44:57us to do is to really spatially and
02:44:59temporally controlled the antigen
02:45:00expression very carefully within
02:45:02genetic models and also within within
02:45:04self tissues into your models.
02:45:05So the idea here is we have a flip
02:45:08oh that's actually broken in half,
02:45:10much like the neoantigen I showed you
02:45:12was where you need Creamer comma.
02:45:14Nice to actually.
02:45:15Invert that the DNA to actually
02:45:17allow you to express the Flippo and
02:45:19then you need to give doxycycline
02:45:20and tamoxifen in order to allow
02:45:22this foot boat and then act on
02:45:24the neoantigen module itself.
02:45:26And when you do that you get
02:45:28that permanent recombination.
02:45:29Now this becomes a neoantigen express
02:45:30himself and so as Kelly alluded to,
02:45:32we recently published this so I won't
02:45:34go through the details of how the
02:45:37mouse model works or the paper really
02:45:39other than to say we spent quite a
02:45:41bit of effort in that model to show
02:45:43that you could get inducible de Novo
02:45:45Neoantigen expression that paper site.
02:45:47To show that you get Dinovo in
02:45:49these neoantigen expression,
02:45:50and that because of this,
02:45:52this system is being set up the way it is,
02:45:56there's really no tolerance prior to
02:45:58induction within the peripheral T cell pool,
02:46:00and so you get naive T cells that
02:46:03can make very robust responses
02:46:05against viral infection,
02:46:06just like a normal mouse.
02:46:08You really have an intact pre
02:46:10activation immune repertoire
02:46:12that you can study how those T
02:46:14cells in undergo these processes.
02:46:16Additionally we showed.
02:46:17Just through and fishing with
02:46:18viruses and that in code.
02:46:20Comments or flippo are also turning
02:46:21on an engine using genetic means that
02:46:23you could really nicely get robust
02:46:25T cell responses when you turn on
02:46:27antigens and that the T cells would
02:46:29home to specifically the site where
02:46:31you were where you were turning on the
02:46:33antigen so you have very good again.
02:46:35The spatial temporal control over over
02:46:37antigens in this model is very is
02:46:39one of the highlights really I think
02:46:41will enable a lot of applications,
02:46:43so there's a number of different
02:46:45applications and for people who are
02:46:47interested we put this mouse in Jackson so.
02:46:49There's a lot of access that it wouldn't
02:46:51when it becomes available for distribution.
02:46:53Everybody will have an
02:46:54opportunity to use this.
02:46:56If you think it's something
02:46:57that we use for your studies,
02:46:59we are very interested in two questions.
02:47:01The immune microenvironment associated
02:47:02with tumors in healthy cells function
02:47:04within different tumor types of
02:47:06different tumor micro environments.
02:47:07And then we're also very interested
02:47:09in how this process breaks down in
02:47:11the context of peripheral tolerance.
02:47:13So for the rest of the time I'm going to,
02:47:16I'm going to focus on the story
02:47:18that we've been developing.
02:47:19It's an unpublished story that's
02:47:21been spearheaded by Kelly Connelly,
02:47:22who's a postdoc in my in my lab,
02:47:24and she's she's been working on
02:47:26this for a couple of years.
02:47:28Also,
02:47:28collaborating with two very
02:47:29talented graduate students and
02:47:30smooth Krishnaswamy's group,
02:47:31these are folks doing really
02:47:32amazing bioinformatics work,
02:47:33and they've done all the bioinformatics
02:47:35that I'm going to tell you about.
02:47:37And when the questions that
02:47:38we were really interested in
02:47:40is, there's been a real Sergeant.
02:47:41Our understanding of T cell
02:47:42differentiation processes and how.
02:47:44Chronic Inogen and signals within
02:47:46tumors drive script drives that
02:47:48process and we're really interested
02:47:50in how this might work in the
02:47:52context of developing tumors and so.
02:47:54Rocky and others have really harped on this.
02:47:57This model here,
02:47:58so I'm not going to go into any
02:48:00real detail on it outside to say
02:48:03that the real the past three or
02:48:05four years have been a flurry of
02:48:08activity just describing how this
02:48:09process that was once thought of as
02:48:12a monolithic exhaustion process.
02:48:13Now we're starting to understand
02:48:15that there are multiple subsets
02:48:17of cells within the T cell pool,
02:48:19and then each of these T cells differ
02:48:21in terms of their potential to Mount
02:48:23responses and participate in the response.
02:48:26The exhausted cell,
02:48:27which is sort of a terminally
02:48:29differentiated so really has
02:48:30restricted its functional capacity,
02:48:32and I think the other thing that
02:48:34goes along with that is a restriction
02:48:36of proliferative capacity that make
02:48:38this cell not very good at terms
02:48:40in terms of fighting, fighting,
02:48:42infections, chronic infections,
02:48:43or in terms of fighting fighting tumors.
02:48:45It's associated, of course,
02:48:47with the up regulation of these.
02:48:49These checkpoint receptors,
02:48:50and also this down regulation
02:48:52of TCF one and so there there is
02:48:55also a positive population within
02:48:56the within the tumor.
02:48:58Or within the within,
02:48:59the chronic infection that is a stem
02:49:01like population were going to let
02:49:03this this TSL sometimes in this cell
02:49:06is marked by its expression of PD.
02:49:08One expression of slanted sticks
02:49:10and also its expression of TCF and
02:49:12so one of the elements about this
02:49:14process that I found particularly
02:49:16fascinating is that the signals
02:49:18here that are potentially going to
02:49:19drive this process of exhaustion
02:49:21of terminal differentiation or
02:49:22chronic antigen and TGF beta.
02:49:24And these are signals are very highly
02:49:26expressed within the tumor microenvironment.
02:49:28And there's been work from several groups,
02:49:32including including this Snyder
02:49:34paper here and also.
02:49:36The sticky paper from from warehouse
02:49:38group that really highlighted the idea
02:49:41that T cells these stem like T cells
02:49:43really need to be inside tumors in
02:49:45order to mediate therapeutic effects,
02:49:46and this makes sense because we
02:49:48know that the micro environment
02:49:50has expression of PDL,
02:49:51one on tumor in this correlates
02:49:53with outcomes expression on immune
02:49:54cells that correlated outcomes.
02:49:56So the thought is that the the
02:49:58antitumor effect is really mediated
02:50:00from within the tumor.
02:50:01That suggests that the stem like
02:50:03cells are in the tumor.
02:50:05But again, this chronic antigen should be.
02:50:07It should be deleterious.
02:50:08Because the big question is,
02:50:10this process should be driving terminal
02:50:12exhaustion of these cells and of
02:50:14course in a short term this is fine,
02:50:16but because tumors develop over
02:50:18the course of months or years,
02:50:20the big question we had with how do
02:50:22you maintain this dim light population
02:50:24in the face of all these signals,
02:50:26that should drive its exhaustion
02:50:28and so to try and
02:50:29understand this we used another model.
02:50:31We cross our ninja mice to another
02:50:34model that we was familiar with from
02:50:36working in Tyler Jackson Lab and
02:50:38this is called the care SP3 model.
02:50:40It's a great model for studying lung
02:50:43adenocarcinoma in other tumor types
02:50:45that are driven by oncogenic kras.
02:50:47So the idea here is we give a Cree
02:50:49expressing adeno or lentivirus and this
02:50:52acts on genetic elements within lung
02:50:54cells and so we have a oncogenic form
02:50:56of care SG-12 deform but also have two
02:50:59flocks copies of the tumor suppressor P53.
02:51:01When you give Cree you activate chaos
02:51:03and droopy 53 in this single transform
02:51:05cell will now develop over the course
02:51:08of several months through a variety of
02:51:10different stages that mirror the stages
02:51:12that happen in patients who develop
02:51:14long adenocarcinoma and these stages.
02:51:16Now we can understand.
02:51:17How the immune system interacts with
02:51:19these tumors at different stages,
02:51:21and so we've we've introduced into this
02:51:23system using using the ninja leal the
02:51:25ability now to turn on you antigens
02:51:27in this tumor and the way we do that,
02:51:30again,
02:51:30is that when we give creepy poise
02:51:32this and then we treat them isa
02:51:35few weeks later with doxycycline,
02:51:36tamoxifen and this delay in treatment
02:51:38is really just do just to allow us to
02:51:41clear all these effects of the infection
02:51:43associated with turning on the tumor genes.
02:51:45And when we do this process we
02:51:47actually see a pretty big change
02:51:49within the tumors themselves.
02:51:51So this is what we call KP tumor,
02:51:53so it doesn't have neoantigens and this
02:51:55is sort of your classic cold tumor.
02:51:57There's really it's an immune desert
02:51:59in terms of T cells and B cells.
02:52:02Almost none of them in this tumor.
02:52:04In.
02:52:04In contrast,
02:52:04if you look at a KP ninja tumor
02:52:07and this is just showing you
02:52:08an 8 week keeping into tumor,
02:52:11you see that this tumor is very
02:52:13heavily infiltrated by by T cells.
02:52:14This is a very robust and
02:52:16high penetrance phenotype.
02:52:17Almost every every tumor at
02:52:18this time point is very heavily
02:52:20infiltrated by T cells in this manner.
02:52:22So just by turning on antigens
02:52:24within the tumor,
02:52:25we can actually elicit a very
02:52:27robust response that turns the
02:52:29macro environment from cold to hot.
02:52:31That's very exciting.
02:52:32'cause it allows us to study
02:52:34how this process happens.
02:52:35And of course,
02:52:36as I mentioned,
02:52:37we're putting in own neoantigen,
02:52:39since this system so that allows us to
02:52:40now look at the tetramer specific cells
02:52:42within the tumor and try and understand
02:52:44their differentiation at different stages,
02:52:46because the tumors are extremely
02:52:48small at these stages,
02:52:49we can isolate them and then
02:52:51and then purify out the cells.
02:52:52So we need a technique to identify
02:52:54which cells are in that issue,
02:52:56and so we've been using this intravascular
02:52:58technique that a number of groups
02:53:00have used where we inject labeled
02:53:01antibodies into the circulation.
02:53:03This labels the cells in the
02:53:04circulation and the cells
02:53:06that are in that issue, or protected.
02:53:07And what you can see from this graph
02:53:10here is that when we don't have tumors,
02:53:12so this is a mouse that's
02:53:13induced in the same way,
02:53:14but doesn't have crashed,
02:53:15so it can't form a tumor.
02:53:17You don't really get any sells.
02:53:18It looks like a V6 months.
02:53:20You don't really get any cells that go into
02:53:22the into the lung tissue in CDA T cells.
02:53:25In contrast, when you have a tumor now
02:53:26you see that there are a number of CD8T
02:53:29cells within the micro environment,
02:53:30so that between the two lung tumor tissue,
02:53:32which was a nice way of just confirming
02:53:34that these T cells are going to
02:53:36the long because of the Tuners.
02:53:38From these we can get on touch more
02:53:40specific cells which would be 333
02:53:41loaded and a C class one tetramers
02:53:43and we can look at the phenotypes of
02:53:46the cells that are within the tumor.
02:53:47And here we're just we're getting on
02:53:49PD one ansi and TCF one and looking
02:53:51at looking for these stem like cells
02:53:53which are marked by their dual
02:53:55expression about TCF and PD one and
02:53:57what you can see is both at early
02:53:59time points and at late time points
02:54:01there is a robust population of these.
02:54:04These TCF one PD one positive cells
02:54:05and that they are maintained even even
02:54:07several months after we've initiated tumors.
02:54:09These tumors still contain these cells
02:54:11and what we were particularly intrigued
02:54:13was by was when we started looking
02:54:15at the tumor draining lymph node.
02:54:16When we saw that there was a very large
02:54:19population of these stem like cells.
02:54:21In fact,
02:54:21almost all the cells within the lymph node,
02:54:24both at the early time point an at
02:54:26the late time point are expressing
02:54:28this TCF one and PD one and again I
02:54:30just want to remind you everything I'm
02:54:32going to show you from this point on
02:54:34is gated on Antigen specific T cells.
02:54:37So these are all specific to the
02:54:39to the tumors,
02:54:39yet they're located in the draining
02:54:41lymph nodes in terms of numerically,
02:54:43if we quantitate the number
02:54:44of cells you can see,
02:54:46there's about the same number of total
02:54:48GP 33 cells at both times in in the,
02:54:50in the lymph node in the in the tumor.
02:54:53But there's a big increase until
02:54:54in the total number of these stem,
02:54:56like cells within the draining lymph nodes,
02:54:58suggesting maybe these cells are could
02:55:00be could be important in this location.
02:55:02The other point that we noted from
02:55:04this very early analysis was that
02:55:05the T cells in the.
02:55:07The tumor really seems to be the
02:55:09only ones these TCF one low sells
02:55:10really seemed to be the only ones
02:55:12upregulating Tim 3 which is that
02:55:14marker of terminal exhaustion?
02:55:15the T cells in the lymph node
02:55:17aren't doing this.
02:55:18Many of the populations,
02:55:19and also when we look at the
02:55:21T cells within the lymph node,
02:55:23they seem to have a very robust
02:55:25stem like phenotype,
02:55:26so they're all expressing slam up six.
02:55:28Another marker of stem like T cells,
02:55:29and we look for function. We actually see.
02:55:32These cells are are very functional.
02:55:33We can. We can send them with peptide and
02:55:36see the majority of the cells that are in
02:55:38the in the in the tumor in the lymph node
02:55:41are capable of producing T Interferon,
02:55:43whereas only a fraction of the T cells
02:55:46in the tumor are capable of this.
02:55:48So try and get a better handle on
02:55:50what's going on with these populations.
02:55:52We perform single cell RNA seq
02:55:54and we also perform PCR seeks,
02:55:56so we're looking at the endogenous GP.
02:55:5833 specific styles identified
02:55:59with me to class one tetramers,
02:56:01and then we've used a technique with smooth
02:56:04at least meet his lab to try and Co.
02:56:06Embed the secret.
02:56:07Are they seek data and try and compare
02:56:10things using a technique called fate,
02:56:12but which allows us to understand
02:56:14the trajectory of differentiation
02:56:16very nicely and visualize that.
02:56:17I'm going to show you some
02:56:19some fake plots here.
02:56:20The 1st I'm going to show you is 1
02:56:22where we're comparing 17 weak lungs
02:56:24and 17 week draining lymph nodes.
02:56:26And when you put these together into
02:56:28the same plot you can actually very
02:56:30nicely detect different populations
02:56:31and I've just highlighted why we're
02:56:33going to call naive and stem like
02:56:35an exhaustion based on based on a
02:56:37handful of markers and what's really
02:56:38interesting about this is when
02:56:40we look at the lymph node cells,
02:56:42they all fall more towards this
02:56:44side of being stemlike,
02:56:45whereas the cells within the tumor fall more
02:56:47towards this exhaustion process and this.
02:56:49Eternally exhausted state,
02:56:50and this suggests that the signals
02:56:52that are the T cells are that are
02:56:55driving exhaustion really are being
02:56:56received and driving this process
02:56:58within the tumor consistent with
02:56:59what we saw with the facts analysis
02:57:01using an unbiased technique called
02:57:03pseudo time which allows us to
02:57:05draw developmental trajectory.
02:57:06You can actually see that these stem,
02:57:09like cells are the ones that are
02:57:11turning into creatively these
02:57:13exhausted cells and this is shown
02:57:15both by the plot here and then.
02:57:17Also very nicely visualized by the histogram.
02:57:19The love the long sales are are
02:57:21really much more developmentally
02:57:22differentiate more mentally advanced
02:57:24compared to the cells in the lymph node.
02:57:26And these are the same cells 'cause
02:57:28we did PCR sequencing.
02:57:30We've actually looked to show that
02:57:32the cells within the tumor within
02:57:34the lymph node are developmentally
02:57:35the precursors of the cells within
02:57:37the within the lung.
02:57:38And we know this because these are.
02:57:41They have the same Alpha beta chains,
02:57:43so we can actually track those
02:57:45cells between these locations.
02:57:47So one of the other elements that
02:57:49I'm going to show you is about
02:57:50comparing the T cells in the tumor's
02:57:52and then in the lymph nodes to show
02:57:54you how different the developmental
02:57:56trajectory of this house is.
02:57:57So if we compare T cells within
02:57:59the context of
02:57:59the of the draining within the tumor,
02:58:01what you can see is that those T cells
02:58:03that are from an early time point in
02:58:05the tumor are developmentally less
02:58:06differentiated than the ones when
02:58:08you look later and you can see this
02:58:10both in terms of where they are,
02:58:12where they are sitting in terms
02:58:14of populations.
02:58:14You can also see this by pseudo time.
02:58:16The majority of the.
02:58:17Late to ourselves or are actually
02:58:19within this within this dance
02:58:20cluster and this makes sense.
02:58:22We actually know in this model
02:58:23something from a previous version that
02:58:25we've confirmed in our model that
02:58:27essentially the micro environment
02:58:28goes from a very hot to a very cold
02:58:30environment and you can see an early tumor.
02:58:32There's a lot of these CD 3 positive
02:58:34cells within the tumor parenchyma.
02:58:36If you look at a later tumor,
02:58:38you actually don't see very many
02:58:39T cells at all,
02:58:40and so we know that there's this
02:58:42big shift in terms of the micro
02:58:44environment and this correlate's with
02:58:46the idea that the T cells in that.
02:58:48In that micro environment,
02:58:49may be undergoing this progressive
02:58:51exhaustive process.
02:58:52Or Sing to us about this.
02:58:56Was that in the context of the lymph node,
02:58:58there really isn't much of a change.
02:59:00So despite this fact that we've now gone
02:59:02from 8 weeks to 17 weeks in this business,
02:59:05really dramatic change within the
02:59:06tumor in terms of both the phenotypes
02:59:08of the cells and also their location,
02:59:10we really don't see much of a much of a
02:59:12difference in terms of the populations.
02:59:14There may be some more,
02:59:16a few more of these exhausted,
02:59:17more terminally differentiated
02:59:18cells within the lymph node.
02:59:20But the preponderance of the
02:59:21population is now still overlapping
02:59:23with what we saw at 8 weeks.
02:59:24That suggest that this is a very stable
02:59:26population within the lymph nodes.
02:59:28So we became very interested in
02:59:29trying to figure out this relationship
02:59:31between these two locations.
02:59:32You could imagine again that it's
02:59:34still possible that the cells
02:59:35that are in the lymph node,
02:59:36even though their canal cleark only
02:59:38related to the ones in the tumor.
02:59:40You could imagine that the ones
02:59:41that are in the tumor or actually
02:59:43locally maintaining themselves.
02:59:45And that that somehow this is allowing
02:59:46the population of Perpich perpetuate.
02:59:48Another possibility is that there's
02:59:49a migration process where the
02:59:51where the cells in the lymph node
02:59:53or having an important role in
02:59:54actually resupplying the tumor over
02:59:56the course of two native element
02:59:57to try and get a handle on this,
03:00:00we've treated the mice with FT by 7:20.
03:00:02These are talking this nice to
03:00:03these experiments are fairly long.
03:00:05We initiate the treatment arounds
03:00:068 about six weeks post infection.
03:00:08So these are when they have
03:00:09really really tiny tumors,
03:00:11but there still infiltrated by T cells.
03:00:12And then now we let them go for three
03:00:15weeks and try and look at what impact?
03:00:17Treatment had on the development of
03:00:19the two of the of the T cell response,
03:00:21and So what you can see is that the
03:00:24treatment with FTY 720 actually has a
03:00:26significant effect in terms of decreasing
03:00:27the number of stem like cells that
03:00:29are present within the tumor tissue.
03:00:31Just shown here, both in terms of
03:00:33their frequency and their number,
03:00:34and this effect doesn't really seem
03:00:36to occur within the lymph nodes.
03:00:38So the lymph node seems to be
03:00:40maintained over this period of time.
03:00:41Additionally, we saw it impact
03:00:43on the T cell function,
03:00:44so now what residual function was
03:00:46present within within the tumor T cells?
03:00:48It was now lossed in the context of
03:00:51treatments suggesting that this migration
03:00:52process of T cells from the lymph
03:00:54node to the tumor is very important,
03:00:57both maintaining the stem like
03:00:58population and also maintaining T
03:01:00cell function within the tumor.
03:01:02So the last point I want to make
03:01:04we we started looking into whether
03:01:06or not we could consider this may
03:01:08be a reservoir of T cells within
03:01:09the lymph node and maybe one that
03:01:11was more of a protected niche.
03:01:13And so we've tried to try to understand
03:01:15whether they are what's driving
03:01:16differentiation and whether the T cells in
03:01:18the lymph node are protected from that.
03:01:20The first thing I want to point
03:01:22out is that we look at PCR signals.
03:01:24We can see that the TR signals are
03:01:26very heavily enriched within the tumor,
03:01:27and as I mentioned, very early on TC,
03:01:29our signals are thought to be one
03:01:31of the major major drivers of.
03:01:33Of T cell exhaustion.
03:01:34And so the fact that very few of the
03:01:36cells that are in the lymph node are
03:01:38high for a number of different markers
03:01:40downstream of the CR suggests that
03:01:42those cells were not seeing nearly
03:01:44as much energy as they would if they
03:01:45are within the tumor micro environment.
03:01:47The other question the other,
03:01:49the other element that I wanted to
03:01:50highlight is this idea that the clonal
03:01:52dominance between the populations
03:01:53is very closely maintained and Rafi
03:01:55Ahmed showed a similar graph in his
03:01:57in his data about T cells with lymph
03:01:59nodes and tumors.
03:02:00So I won't belabor this point other
03:02:02than to say that this is very nice.
03:02:04They're very good correlation between
03:02:05the two locations in terms of the
03:02:07types of T cells,
03:02:08but the other thing that this is
03:02:10allowed us to do is to try and look
03:02:12at PCR motifs,
03:02:13and this is a little bit of an
03:02:15aficionados point,
03:02:15so I apologize if I lose people on this,
03:02:17but because we sack mice at 8 weeks
03:02:19and stack mice at 17 weeks,
03:02:21we can't really compare the T cells directly,
03:02:23but what we can do is we can try
03:02:25and identify motifs that are used
03:02:26by those T cells,
03:02:28T CR motifs and try and identify
03:02:29as similar motifs are used by the
03:02:31T cells in both time points and
03:02:33what you can see from this group.
03:02:35Ask is that if we compare T cells
03:02:37at early and late time points that
03:02:39there are a number of motifs that are
03:02:41actually conserved between the early
03:02:43time point in the late time point.
03:02:46This just suggests that the T cells
03:02:48that are present early are in some way
03:02:50conserved and you might imagine that
03:02:52if there was chronic antigen exposure
03:02:54and there wasn't reservoir that you
03:02:56might lose these very high affinity.
03:02:58The higher affinity clones,
03:02:59and ultimately that would cause
03:03:01clonal Kona loss.
03:03:02Here we're not seeing that we're actually,
03:03:04we think. Seeing that there maintained
03:03:06based on this motif analysis.
03:03:08So, just to summarize,
03:03:09I would have told you today we're thinking
03:03:11that that within the tumor microenvironment,
03:03:14the signals that are there,
03:03:15like a lot of groups have shown in
03:03:17like I think we're seeing in our
03:03:19data that those signals that are
03:03:21in the tumor microenvironment are
03:03:23very important for promoting this
03:03:24progressive T cell exhaustion,
03:03:26and we think absent a mechanism to
03:03:27maintain them that these T cells would
03:03:29ultimately undergo terminal differentiation,
03:03:31and you would lose the
03:03:32response against the tumor.
03:03:34So in order to maintain that response
03:03:36over the course of several months,
03:03:38we think that the immune system.
03:03:39That's up this lymph node reservoir
03:03:41which allows us allows it to protect
03:03:44antitumor T cells over long periods
03:03:45of time and really perpetuate
03:03:47the response through migration as
03:03:49opposed to setting unnecessary
03:03:51local maintenance that doesn't rule
03:03:52out the role of local maintenance.
03:03:54And it could be there are tumors
03:03:56don't have some critical elements
03:03:58like tertiary lymphoid structures.
03:04:00They definitely early time.
03:04:01Points don't have tertiary lymphoid
03:04:03structures,
03:04:03but it does suggest that that at
03:04:05least the lymph node could serve
03:04:07this role within within patients
03:04:09or within within tumors.
03:04:11I better that are present in our animal
03:04:13models and potential in patients,
03:04:15and we also think that this is
03:04:17important for potentially protecting
03:04:18those T cells from chronic antigen
03:04:20exposure and internal differentiation.
03:04:22One of the elements that I was
03:04:24particularly intrigued by was that the stem,
03:04:26like reservoir was really very similar,
03:04:28or when the tumors were hot or
03:04:30when they're cold,
03:04:31and it's possible that this is
03:04:32telling us that a cold tumor could
03:04:35really be associated with reservoir.
03:04:36That's actually functional.
03:04:37Definitely,
03:04:38it suggests that the tumor itself is
03:04:40the place at which determination of
03:04:41hot and cold is probably occurring,
03:04:43and that that if that is true,
03:04:45it could suggest that we have
03:04:47an opportunity here,
03:04:48even in patients with cold tumors
03:04:49to try and understand what's going
03:04:51on in the lymph node and then.
03:04:53Maybe to try and target Lindo these
03:04:55lymph node reservoir T cells in
03:04:57terms of therapies and there's
03:04:59been a number of papers.
03:05:00I think someone earlier referred to this.
03:05:02There's been a couple papers in
03:05:04the past couple weeks that have
03:05:06highlighted the idea that therapy.
03:05:07Maybe if you target to the lymph node
03:05:09you might get therapeutic effects.
03:05:11So with that I'd like to thank
03:05:13the people who are on this slide.
03:05:15I mentioned Kelly and smooth group.
03:05:17Wego is a collabora longtime collaborator
03:05:19who's been who helped us out with the
03:05:22motif analysis and people in his lab.
03:05:23Britt and Martina who worked on the.
03:05:26Paper that was published recently
03:05:27and then a few other people have
03:05:29assisted Kelly through the process
03:05:30and I'm happy to take any questions.
03:05:34OK, we have time. I think
03:05:36for a couple of quick
03:05:37questions. So the first one
03:05:39is based upon the data.
03:05:40I think you alluded to this a
03:05:42little bit on your last slide
03:05:44is do you think that there would
03:05:46be a role in? Promoting T cell trafficking
03:05:49to otherwise immune excluded term.
03:05:52Rise immune excluded tumors by
03:05:55targeting lymphangiogenesis.
03:05:57Yeah it's possible.
03:05:58I mean, we don't know.
03:05:59Excuse me, we don't know
03:06:01for sure what the what the.
03:06:04Whether the T cells aren't getting to
03:06:06the to the tumors through migration,
03:06:08we think they're probably getting
03:06:09there and just something about
03:06:11the micro environment is changed.
03:06:12Increasing lymphangiogenesis
03:06:13could do a few things.
03:06:15One is to increase the rate of
03:06:17flow of antigens, and there may be
03:06:19some deficiencies in that process,
03:06:20and it does suggest our data
03:06:22we think are showing US data.
03:06:24I didn't get a chance to show you.
03:06:26We think that this idea that
03:06:28maybe the antigen from the tumor
03:06:30is having an important role in
03:06:32terms of driving the migration so
03:06:33lymphangiogenesis could help in that.
03:06:35Aspect of the process,
03:06:37but those are experiments that we
03:06:39I think will need to follow up on.
03:06:41Next question is,
03:06:42do you think that this reservoir
03:06:45and humans would be
03:06:46combined to the draining
03:06:47lymph nodes? Or do
03:06:49you think that also
03:06:50the peripheral circulation
03:06:51may have a role? Yeah,
03:06:54I think it's been shown by a number
03:06:56of different groups that there are T
03:06:58cells in the peripheral circulation is
03:07:00not clear to me if those are resident
03:07:02in the circulation or if their T cells
03:07:04that are migrating from the lymph
03:07:06node to the to the to the tumor.
03:07:08In an you just capture him as a
03:07:10snapshot during when you look at them.
03:07:12But this idea of clonal replacement is
03:07:15actually much better in terms of there
03:07:17are a couple of papers on it trying to
03:07:19compare T cells from the lymph node or
03:07:21sorry from the circulation to the tumor.
03:07:23It's actually been quite hard.
03:07:24Even for us to try and identify
03:07:26ways to get to lymph node tissue
03:07:28to try and look for these cells.
03:07:31So that's just a problem that the
03:07:33field is going to have to wrestle
03:07:35with so we don't have a good sense
03:07:37for how their circulation in patients
03:07:39compared to the lymph node.
03:07:42And there's another
03:07:43question regarding tertiary
03:07:44lymphoid structures. If
03:07:45those could serve as
03:07:46a reservoir over a longer term
03:07:48compared to the draining lymph
03:07:50nodes. That's what we were thinking.
03:07:52I will say that at later time points
03:07:55that Week 20 in our models we do
03:07:58see tertiary lymphoid structures
03:07:59and it's probably it's likely that
03:08:01the majority of the stem cells
03:08:03that are present in the tumor or
03:08:05pressing within those structures.
03:08:06I think one thing that I've always
03:08:08thought that was interesting about
03:08:10the tertiary lymphoid structures,
03:08:11given its proximity to the tumor,
03:08:13it's likely a place where there's
03:08:15a lot of images,
03:08:17presentation and so it may be hard in
03:08:19some ways to maintain a population.
03:08:21In in a in a stem like State,
03:08:24because you're constantly exposing
03:08:25advantage and the distance of the
03:08:27lymph node may decrease that and they
03:08:29ultimately helped to allow a better
03:08:31maintenance of this population.
03:08:32But it's an open question still
03:08:34that will have to be addressed.
03:08:36Doctor Sharma?
03:08:37Yes, thank you. That
03:08:38was such a great presentation
03:08:40that thank you so much for that.
03:08:42Do you know in your model if your
03:08:45change now in the Neo Antigen epitope,
03:08:47you're able to have expression?
03:08:48Are you just changing
03:08:50T cell responses? Have you seen B
03:08:52cell responses change as well?
03:08:54Because, you know, we were seeing data
03:08:55now the B cell responses are also
03:08:57very important for these antigens,
03:08:59so just curious as to whether or not
03:09:01that's a model you can use there. In
03:09:04our model, the one I showed there
03:09:06not be selling engines and actually
03:09:08we've used that in another story that
03:09:10we're still working up that we've
03:09:12actually put Decelean mentions in,
03:09:14and it has a very big effect.
03:09:16Maybe next year I can talk about that,
03:09:19but it's very exciting when you
03:09:21put the cell antigens into tumors,
03:09:23it completely changes
03:09:24things. Yes, very great model.
03:09:26Thank you, will have that the last
03:09:28question from doctor keck. Linking some
03:09:30of your work with IRA Mehlman's
03:09:32work on PDL one. Expressing DCS.
03:09:34Do you see any differences in
03:09:36the expression of PDL one on D?
03:09:39Scenes between the tumor or in
03:09:41the draining lymph nodes.
03:09:42That's a great question. So thank you.
03:09:45We have not looked at something
03:09:47that we're actively now pursuing.
03:09:48We think that it's likely that that PDL
03:09:51one expressing DC's are going to be
03:09:54important in terms of interacting with
03:09:56these cells and an Irish work is cancer.
03:09:59Is nature. Cancer paper really was
03:10:01a very exciting window into those
03:10:03into those types of interactions.
03:10:05So is something that we're
03:10:07very interested in.
03:10:08I think we're probably not.
03:10:09Not the only ones who are interested in this,
03:10:12but I think as a field,
03:10:14that's something that we're going
03:10:16to find out relatively soon in
03:10:18terms of what those cells are doing
03:10:20and how they are participating
03:10:22in these types of responses.
03:10:23OK, great,
03:10:24thank you, so we're going to
03:10:26move on to the next speaker,
03:10:28so again, I have the privilege
03:10:30of getting to introduce another
03:10:31one of our colleagues here.
03:10:33Ehrenring errands and assistant
03:10:34professor of Immunobiology at Yale,
03:10:36who was recruited here after
03:10:38he got his MD PhD at Stanford.
03:10:40Aaron's been honored as a
03:10:4230 under 30 and healthcare,
03:10:44and he's also accused Stewart scholar,
03:10:46which she was word in 2018.
03:10:48Erin is a passionate scientist
03:10:50and you if anyone spent more
03:10:52than 3 minutes with Aaron,
03:10:54you cannot help
03:10:56but be excited and and have a great
03:10:59amount of enthusiasm for the work that
03:11:01he's performing in the work
03:11:03that you can perform with him.
03:11:05And Aaron is really focused on taking
03:11:08some very unique approaches towards.
03:11:10Engineered cytokines
03:11:11and also is
03:11:12doing a ton
03:11:13of other work
03:11:14and recently is
03:11:15published some of his Seminole work
03:11:17actually in nature earlier this year.
03:11:19So in the interest of
03:11:21time air and take it away.
03:11:24Yeah, thank you so much Kelly for
03:11:26that super kind introduction.
03:11:27I just want to point out to everyone
03:11:30that I'm way over 30 now and you
03:11:32know it's funny that Suquet just
03:11:34ask the question because you know
03:11:36she saw me just last year and she
03:11:38said you've been a couple years.
03:11:40Wow. You look a lot older so.
03:11:42Yes, I'm over 30 anyway,
03:11:44delighted to talk today in in this
03:11:47really exciting symposium in my
03:11:49lab we use structure based protein
03:11:51engineering to create pharmacological
03:11:53tools that we can use to probe
03:11:56complicated immuno regulatory pathways.
03:11:58And although the goal is biology,
03:12:00sometimes you know we do make things
03:12:02that have therapeutic potential happen
03:12:04involved in the commercialization
03:12:06efforts to bring those into the clinic.
03:12:09So these are my disclosures.
03:12:12So that a longstanding interest in
03:12:15developing cytokine therapies for cancer.
03:12:16Ever since I was an MD PhD student with
03:12:19Chris Garcia about 10 years ago in one
03:12:21of the major reasons beyond the fact
03:12:24that cytokines are incredibly interesting.
03:12:26If you have any interest in immunology,
03:12:28these are the central defining molecules that
03:12:31instruct all sorts of immune activities.
03:12:33Is that the cytokines were the first
03:12:35agent to unambiguously prove the paradigm
03:12:37that the immune system could be an
03:12:39effective target for cancer therapy,
03:12:41and that was most evident in the experience
03:12:44of high dose interleukin two therapy.
03:12:46In Melanoma and renal cell cancer,
03:12:49where a small fraction of patients about 16%,
03:12:52I'm responded to,
03:12:53patient responded to L2 in of those patients,
03:12:56a total of 6% had long lasting,
03:12:59durable remissions that could
03:13:01essentially be called a cure.
03:13:03I mean,
03:13:04this is really the first example
03:13:06of a survival tale on the capital
03:13:09matter survival curve that has
03:13:11made immunotherapy so compelling.
03:13:14So cytokines themselves are really
03:13:16compelling as potential agents
03:13:17for immunotherapy.
03:13:18That's like I said,
03:13:20because they do so many diverse
03:13:22activities on immune cells.
03:13:24So I'm like a checkpoint inhibitor
03:13:26that they do more than just tune
03:13:29an existing response where they
03:13:31can tap into hardwired programs.
03:13:33That instructing mean survival,
03:13:35proliferation,
03:13:35differentiation into different
03:13:36phenotypes in ultimately control
03:13:38effector function of immune cells,
03:13:40and they can do that Locali in
03:13:42an auto repair confession,
03:13:44or even have endocrine like.
03:13:46Effects systemically,
03:13:47the problem with cytokines is that
03:13:50they did not evolve to be drugs,
03:13:52but evolved to be signaling molecules.
03:13:55The immune system and so there are
03:13:58biological limitations inherent to
03:13:59how they have evolved to play a role
03:14:01in regulating immune responses that
03:14:04have curtailed their therapeutic use.
03:14:06So I'll choose a really instructive example.
03:14:09On one hand,
03:14:10it can stimulate potent antitumor immunity
03:14:13through cytotoxic T cells as well,
03:14:15stimulation of natural killer cells.
03:14:17But on the other hand,
03:14:19it also can stimulate the proliferation
03:14:22of immunosuppressive T regulatory
03:14:24cells that paradoxically inhibit
03:14:25the antitumor functions of I'll too.
03:14:28Similarly,
03:14:28because cytokines have such potent
03:14:30effects on our Physiology,
03:14:32we've of course evolved very strong
03:14:34negative feedback mechanisms to
03:14:36prevent runaway inflammation,
03:14:38and this is great to protect us
03:14:40from autoinflammatory disease.
03:14:42But in the setting of administering
03:14:44recombinant cytokine therapies,
03:14:45these same mechanisms can curtail the
03:14:48maximal efficacy of cytokine drugs.
03:14:50So really, our conviction and
03:14:52it's not just our own conviction,
03:14:55but that basically everyone
03:14:56who works with sciatica.
03:14:58Therapies is that we can't accept
03:15:01nature solution to cite a kind,
03:15:03so we need to engineer them for
03:15:07deliberate therapeutic purpose and
03:15:09tailor their activities to maximize
03:15:11their effect and desired cell
03:15:13populations and avoid those that have
03:15:16either safety or efficacy impediments.
03:15:19So when I started my lab a few years
03:15:22back and we wondered how their
03:15:24statically pathways that have been
03:15:26overlooked in tumor immunotherapy.
03:15:28In particular,
03:15:28we wondered if there were cytokines
03:15:31that had more selectivity to the
03:15:33very T cells that were doing
03:15:35the heavy lifting in the tumor.
03:15:37That is to say,
03:15:38tumor reactive antigen specific
03:15:40tumor infiltrating lymphocytes.
03:15:41It turns out that most cytokines
03:15:43that we give I'll to I'll 15 at 12,
03:15:46they don't really have selectivity
03:15:48toward these antigen experience
03:15:50management specific T cells,
03:15:51and so it was about this time at
03:15:53the single cell RNA sequencing
03:15:55revolution was coming to the
03:15:57foreign in Anderson's group.
03:15:59Publish this phenomenal paper where
03:16:00they perform single cell RNA sequencing
03:16:02on tumor infiltrating lymphocytes,
03:16:04and they were able to bioinformatic
03:16:06Lee extract AT cell activation score
03:16:08for every gene they've detected
03:16:09in the data set versus AT cell
03:16:12dysfunction score and what they found
03:16:14was that the best checkpoint targets
03:16:15the ones that you know of course,
03:16:18had so much success in the clinic PD,
03:16:20one seat away four,
03:16:22and a bunch of emerging targets
03:16:24that were on this upper right
03:16:26hand quadrant of the plot,
03:16:27meaning they were expressed in both
03:16:29activated and is Functional T cells.
03:16:31And that makes a lot of sense.
03:16:34These T cells will be activated
03:16:35because they're seeing tumor antigen
03:16:37and their dysfunctional because
03:16:38they're in the tumor microenvironment.
03:16:40Thought was a brilliant analysis.
03:16:42We wondered what about cytokines?
03:16:44And so we took their data and
03:16:46employed at every cytokine,
03:16:47receptor and pathway component that
03:16:49we could detect in the data set
03:16:52and to make a Long story short,
03:16:54would immediately jumped out at us.
03:16:56Was that the aisle 18 pathway
03:16:58was fairly unique,
03:16:59and then its receptor subunits,
03:17:00and even the cited kind itself.
03:17:03Or in that upper right hand
03:17:05quadrant of the plot,
03:17:06which made us think that that I lay
03:17:09team could be an open port on these T
03:17:12cells were trying to hack into them
03:17:15where we could deliver a selective message.
03:17:17More specifically to these antigen
03:17:19specific till as opposed to broad
03:17:22stimulation of lymphocytes that's
03:17:23known to cause cytokine release
03:17:25syndrome and unacceptable toxicity.
03:17:27So we actually first sought to confirm
03:17:29that the single scientist sequencing
03:17:31data with good old fashion flow
03:17:34cytometry and looking at mouse tumors,
03:17:36and what we found was that sure
03:17:39enough that the 18 Receptor
03:17:40was not very prevalent
03:17:42on T cells that were found in
03:17:45the periphery and the spleen,
03:17:47but the T cells in the tumor CD S in
03:17:49city force had abundantly upregulated
03:17:52I'll 18 Receptor and we also saw,
03:17:55as seen many times before, that.
03:17:57Natural killer cells,
03:17:58highly expressed al 18 receptor at baseline,
03:18:01so this is really a feature of
03:18:03innate capacity that T cells
03:18:05become antigen experienced. Again,
03:18:07the ability to respond to an 8 stimuli,
03:18:10in this case interleukin 18 and
03:18:12inconsistent with that idea.
03:18:13You can see that when you look
03:18:15within tumors that the cells
03:18:17that expressed the Allied Team
03:18:19Receptor it is exclusively found
03:18:21on the CD 44 very high cells,
03:18:24meaning that it really is marking
03:18:26those antigen experienced T cells.
03:18:28Within the tumor.
03:18:31So like I said, I only team it's
03:18:33an innate cytokinin just zooming
03:18:35out a little bit to give a little
03:18:38background on the biology.
03:18:39It's a member of the aisle.
03:18:41One family of static kinds which
03:18:43are essentially like alarm ends
03:18:45there made inside the cell in
03:18:47an inactive form with a with an
03:18:49inhibitory end terminal peptide that
03:18:51gets cleaved in removed by caspases
03:18:53downstream of the Inflammasome.
03:18:54Once these set of kinds of producing
03:18:56their mature form they exit
03:18:58the cell through a noncanonical
03:19:00secretion pathway that involves
03:19:01the formation of guests German.
03:19:03Wars in the cell membrane,
03:19:05then I'll 18 specifically signals by
03:19:07Hetero Dimerizing its receptor subunits.
03:19:09Eyelid team are Alpha and Alex nor
03:19:11beta to drive my TI88 signaling,
03:19:13which ultimately results in the
03:19:15activation of NF Kappa B and I got is
03:19:19pretty excited because the mighty 88.
03:19:21If you think about it is fairly orthogonal
03:19:23to most other immunotherapeutic agents
03:19:25that are currently in the clinic.
03:19:27Other side of clients or Jack stat.
03:19:30Other checkpoint pathways or
03:19:31item item TF super family.
03:19:33Traft rad so so my TI88 it's a powerful
03:19:37pathway in most 9088 coupled receptor
03:19:39agonists are not really competitive,
03:19:42compatible with systemic administration.
03:19:44You think about your TL are agonists,
03:19:47I'll one, etc.
03:19:48Like I said,
03:19:50You know it drives very
03:19:52strong signaling message,
03:19:53but the receptor is also expressed
03:19:55really on the right cells.
03:19:57That is to say a natural killer cells.
03:20:00Of course another innate lymphoid cells,
03:20:02but on an engine experience CD 8IN
03:20:04TH one cells and not importantly
03:20:06naive T cells but have not seen
03:20:09antigen or central memory T cells
03:20:11that have not seen antigen recently.
03:20:13There's also several reports that
03:20:15highlight team can inhibit the
03:20:17immunosuppressive function of
03:20:18T Reg or at least.
03:20:20Alter their programs away from
03:20:22immunosuppressive function and
03:20:23Tord a tissue repair phenotype.
03:20:25So all of these pieces of data
03:20:28together suggests that highlighting
03:20:29could be a really compelling agent
03:20:32as an immunotherapeutic both and
03:20:34immunogenic tumors through stimulating
03:20:36tumor reactive T cells as well as in
03:20:40in cold refractory tumors through the
03:20:43stimulation of natural killer cells.
03:20:46So we really floored to learn that
03:20:48that I liked Ben in the clinic
03:20:51before it was taken through phase two
03:20:53trials by GlaxoSmithKline and what
03:20:55they found was that for cytokines
03:20:57it was remarkably well tolerated.
03:20:59It could be given up to 2 milligrams
03:21:02per kilogram per day without treatment.
03:21:04Limiting toxicities.
03:21:05Very really astonishing for a cytokine.
03:21:07The problem was that it absolutely
03:21:09had no real efficacy to speak ofw
03:21:12on in the largest phase,
03:21:14two trial before 60 Melanoma patients,
03:21:16there was only one partial response.
03:21:18And I want to point out that these
03:21:20were not not just immunotherapy
03:21:22naive Melanoma patients,
03:21:23but actually treatment naive
03:21:24Melanoma patients.
03:21:25This study was done in the mid 2000s,
03:21:27so if there ever was a population we
03:21:30would expect to respond to therapy.
03:21:32This would be it.
03:21:33In this of course was a
03:21:34really disappointing result,
03:21:36and so the further development
03:21:38team has been largely curtailed.
03:21:40So this was a really striking paradox to us.
03:21:43How could this powerful cytokine
03:21:45hitting the right cells with the right
03:21:48message be so ineffective in the clinic?
03:21:50And so we dug into the data from
03:21:52these clinical studies that we found
03:21:54was that with repeated dosing of
03:21:57Islay team in these patients there
03:21:59was a waning pharmacodynamic effect.
03:22:01In this case, looking at Interferon
03:22:03Gamma released into the blood,
03:22:05I should say parenthetically the original
03:22:07name for all 18 was interferon gamma.
03:22:10Inducing factor which augurs well for
03:22:13immunotherapeutic but you can see
03:22:15that after what weekly dosing that
03:22:17many patients by 4 five 812 weeks
03:22:19had no increase in Interferon Gamma
03:22:21in the blood despite the fact that
03:22:24receiving the maximum dose of drug
03:22:26that decreased activity corresponds
03:22:27to a massive upregulation of a protein
03:22:30called Interleukin 18 binding protein.
03:22:32I'll 18 VP,
03:22:33which goes from single digit nanograms
03:22:35per mill in the blood to 10s to
03:22:38hundreds of nanograms per mill.
03:22:40In the blood.
03:22:41So this is a Ultra High Affinity
03:22:44soluble decoy receptor of I'll 18
03:22:46where it binds highlighting in
03:22:48sterically occludes its ability
03:22:49to engage the aisle.
03:22:5118 or Alpha Receptor highly
03:22:53overlapping interface,
03:22:54and I should also point out
03:22:56the importance of this.
03:22:57Gene is highlighted by the fact that
03:23:00the entire poxvirus family has stolen
03:23:02all ATP through horizontal gene transfer,
03:23:04and it's required for its virulence
03:23:07through escape from natural killer cells.
03:23:09So clearly the viruses have decided
03:23:11that this pathway is very important.
03:23:13This gene. Is a very effective one.
03:23:17You know,
03:23:18in our own biology I like team is is binding.
03:23:22Protein is upregulated downstream
03:23:23of Interferon Gamma where it
03:23:25completes a negative feedback loop.
03:23:27That's highly reminiscient of other side
03:23:29akinde sorry other immune checkpoints
03:23:31and concordant with that idea.
03:23:32If you look in that ECG data,
03:23:35there's a very strong correlation
03:23:37between 18 binding protein expression
03:23:39in other checkpoints that PD one
03:23:41ticket Tim three name your favorite.
03:23:43You'll see a high concordance.
03:23:46We looked ourselves, but you know,
03:23:48within human tumors,
03:23:49by immunohistochemical staining
03:23:50and what we found was that most
03:23:52tumors had at least punctate levels
03:23:54of staining on myeloid cells,
03:23:56typically macrophages in the tumor.
03:23:58But many types of cancer,
03:23:59including those that are not
03:24:01particularly Magenic,
03:24:02had extensive two to three plus levels
03:24:04of staining throughout the tumor.
03:24:06We also look systemically in the
03:24:08blood of cancer patients and what we
03:24:11found was that non small cell lung
03:24:13cancer patients have elevated levels
03:24:15of the binding protein at baseline.
03:24:17And these same patients,
03:24:18after treatment with anti PD one
03:24:20have further increases in the rally
03:24:22team binding protein levels.
03:24:24So all of this data together suggested
03:24:26that that maybe I'll a team would
03:24:28have worked well as immunotherapeutic
03:24:30but for this binding protein that's
03:24:32acting as a barrier to recombinantly
03:24:3418 amino therapy.
03:24:35And so it turns out my next door
03:24:38neighbor in the lab,
03:24:39Richard full Val their group had
03:24:41been working with our team binding
03:24:43protein knockout mouse for many years.
03:24:45So we obtain that that mouse from them.
03:24:48And graphs that with MC 38 tumors
03:24:51and compared it to the wild type
03:24:54litter mates and treated them with
03:24:56either sailing or eyelid team.
03:24:58And what we found was that
03:25:01just like impatience,
03:25:02wildtype Valley team has basically
03:25:04no activity to reduce tumor growth,
03:25:06inhibition or survival, whereas the mice
03:25:08that lacked alateen binding protein.
03:25:11Even though the tumor could still express it,
03:25:14did have a substantial increase in
03:25:16its sensitivity to wildstylez team.
03:25:18So this really.
03:25:19A confirmed to us that the binding protein
03:25:22was limiting the activity of violate team.
03:25:25So we wondered, what if we could take Kylie
03:25:28team binding protein of the equation?
03:25:30What if we can make a version of
03:25:32I'll 18 that was fully capable of
03:25:34engaging its receptor until but
03:25:36completely impervious to the binding
03:25:37protein that that's such a barrier
03:25:40to wild type violentine?
03:25:42So I should say this is a really
03:25:44tough engineering challenge because
03:25:45I'll 18 binds the binding protein
03:25:47and the receptor Alpha at.
03:25:49Like I said, a highly overlapping interface.
03:25:51But making matters worse is that the
03:25:53binding protein binds 10,000 times tighter,
03:25:55so we couldn't do what others
03:25:57have done with I'll two or L15.
03:25:59Or we can make selective mutations
03:26:01or regulation to a blade,
03:26:03one receptor interface and not the other.
03:26:05Very difficult to call your shot here.
03:26:07So our solution to this problem
03:26:09was was evolution.
03:26:10We use directed evolution with yeast display.
03:26:12To screen about 300 million
03:26:14variants of violate team mutated
03:26:15at that shared binding interface,
03:26:17and we selected for those clones
03:26:19that down the aisle 18 receptor.
03:26:22Whereas we counter selected against those
03:26:24that bound the binding protein and of course,
03:26:26repeated this process iteratively and
03:26:28what we saw as you monitor the selection
03:26:31process over round selection is that we see,
03:26:34you know that we're able to completely
03:26:36restore binding to the aisle.
03:26:3818 Alpha after Mutagen Ising this
03:26:40interface and selecting the variance.
03:26:42But we're able to completely prevent
03:26:44reactivity to the binding protein.
03:26:46Ultimately,
03:26:46you can see by the round five here where
03:26:49there's very strong binding to the R Alpha,
03:26:51none to the binding.
03:26:53Protein is a big contrast to the wild type.
03:26:56I'll 18 and we put these decoy resistant D.
03:26:58R18 variants onto NK cells in the dish.
03:27:01What we can see is they potently stimulate
03:27:04interferon gamma with equal or greater
03:27:05greater potency than wild type by lighting.
03:27:08But unlike wild set Valley team,
03:27:10they're entirely impervious
03:27:11to the decode receptor.
03:27:12Not at all.
03:27:13Inhibited by addition of the binding protein.
03:27:16And ultimately,
03:27:17we care about is what happens
03:27:19when we put this type of molecule
03:27:22into a preclinical tumor model.
03:27:24And so we started off with the tumor model.
03:27:27That's already been discussed
03:27:29quite a bit today.
03:27:30For for pretty obvious reasons,
03:27:32that's the Yammer Melanoma model
03:27:34that Marcus Lab created and
03:27:36what we found was completely
03:27:38consistent with our previous results.
03:27:40Completely consistent with Glaxo Smith,
03:27:41Kline saw in the clinic,
03:27:43wildstylez team does basically nothing
03:27:45in terms of tumor growth or survival.
03:27:48Where is the decoy resistant?
03:27:49I'll 18 had dramatic single agent
03:27:51activity that was able to produce
03:27:53very market tumor growth inhibition.
03:27:55In fact, it could clear established
03:27:57tumors in substantial fraction.
03:27:58The mice by itself to degree
03:28:00that was commensurate effect
03:28:01a bit better than NYPD one.
03:28:03In this model,
03:28:04and the two agents together produced
03:28:05a pretty brisk synergism, and,
03:28:07you know, not to belabor the point,
03:28:10but I just want to say briefly
03:28:12that we didn't just use one model.
03:28:14We use several,
03:28:15and we found that in multiple immuno genic
03:28:17models of various degrees of humanity.
03:28:19We confirm that the DRA team
03:28:22has that single agent activity,
03:28:24an synergism with anti PD one.
03:28:28So the timing of the mechanism a little bit.
03:28:31We first wanted to see what cells are
03:28:33required and it should come as no surprise
03:28:36that CD 8 cells were absolutely required.
03:28:38In these models there was a variable
03:28:40requirement for CD4 and NK cells,
03:28:42but clearly you know looking
03:28:44in the Rag deficient mouse,
03:28:45there was no activity of D,
03:28:47R18 or actually I should say,
03:28:49very,
03:28:50very slight tumor growth inhibition,
03:28:51but that we could we could restore
03:28:53that activity by giving T cells,
03:28:55and I think it's really interesting here.
03:28:57Is this particular study
03:28:59where we took a rag mouse.
03:29:01We adoptively transferred T cells.
03:29:02I'm from a wild type mouse.
03:29:05Is that that in itself is not enough
03:29:07to drive tumor clearance in these mice,
03:29:10but addition to the D R18 was
03:29:12able to enable those adoptively
03:29:14transferred cells to clear the tumors,
03:29:17and so I'm really hopeful actually
03:29:19that this data could provide a
03:29:21strong rationale to work with.
03:29:23People with my colleague Tristan
03:29:25parking and combine agents like
03:29:26I'll 18 Indy right inside the
03:29:28kinds with adoptive cell therapy.
03:29:30In any case,
03:29:31we're really curious to know what was
03:29:33going on mechanistically on the T
03:29:35cells and other immune cell populations
03:29:37after treatment with the R18.
03:29:39And so we perform single cell RNA
03:29:41sequencing of D, R18 treated tumors,
03:29:43and what we found was that compared
03:29:46to wild type or Saline treatment,
03:29:48that dear I team had elicited dramatic
03:29:50remodeling of the entire tumor.
03:29:52Immune macro environment,
03:29:53not just in T cells here,
03:29:55but also in the myeloid populations.
03:29:57We see the new appearance of a
03:29:59grainless site population even changes.
03:30:01In fibroblasts,
03:30:02but really we think that these
03:30:04changes are really being centrally
03:30:05driven by effects on the T cells.
03:30:08And so when you zoom in and the
03:30:10T cell clusters,
03:30:11what you see is that in wild
03:30:13type or Saline treated animal,
03:30:15the vast majority of the T cells
03:30:17are in this cluster tube,
03:30:19whereas in the D R18 treated animals
03:30:21the vast majority are in this cluster,
03:30:23one which is uniquely populated
03:30:25by the D R18 treated
03:30:26mice. I also want to point out that
03:30:29the writing also elicits a reasonable
03:30:31number of cells in this cluster. 5 here.
03:30:33And we actually analyze that the
03:30:36transcripts that define these clusters.
03:30:38What you see is that cluster two is
03:30:41defined by very high expression of talks,
03:30:44expression of exhaustion markers
03:30:45like CD244 and CD101.
03:30:47Very high expression of PD
03:30:49one and lag three and absent.
03:30:51Expression of costimulatory markers
03:30:52like CD 28 and I costs and concordantly
03:30:55decreased levels of effector cytokines
03:30:58and increased levels of immunosuppressive
03:31:00cytokines of TGF beta male.
03:31:0210 so really.
03:31:03We can say definitively that cluster 2.
03:31:06Is is the exhausted Lenny edge here.
03:31:08By contrast,
03:31:09cluster one is the exact opposite, right?
03:31:11And this is the DRT intercluster.
03:31:13Basically,
03:31:14no talks doesn't have those exhaustion.
03:31:16Markers instead has markers of
03:31:17T cell maturation.
03:31:18Cal RG1 I'll 18 receptors.
03:31:20The highest levels of Interferon Gamma and
03:31:22very high levels of costimulatory markers.
03:31:25So we would say that these are really
03:31:27a good example of that effect.
03:31:29Or Lenny Edge,
03:31:31and then that cluster five that I
03:31:33think is particularly interesting,
03:31:34has a little bit of PD one.
03:31:37As a little bit of a factor function,
03:31:39but importantly has intermediate
03:31:41levels of TCF seven,
03:31:42which encodes the gene product TCF.
03:31:44One hard to see in this heat map,
03:31:47but take my word for it,
03:31:49that there's there's a bit of yellow here in.
03:31:52This suggests that this population
03:31:53cluster five could represent
03:31:55that resource population that
03:31:56stem like precursor populations
03:31:57been talked about so much today,
03:31:59and so I'm not going to give
03:32:01too much background about those
03:32:03cells because Nick get raffia.
03:32:05Many others have already
03:32:06really discussed them.
03:32:07You know at much greater length than I could,
03:32:10but we really wanted to to look into
03:32:13the idea that that DRA team could be
03:32:15acting on this stem like population
03:32:17in Skewing the differentiation of
03:32:19these cells away from the exhausted
03:32:21Lenny Edge and Tord highly effective,
03:32:24highly effective effector Lenny edge.
03:32:25So we look by flow cytometry at the
03:32:28expression of talks in these cells,
03:32:30and what you can see is that Saline and
03:32:32wild type treated animals have high
03:32:34levels of talks on their initial experience.
03:32:37CD 44 high cells.
03:32:39Whereas with deer 18,
03:32:41that level is greatly decreased
03:32:43in concordance with that finding,
03:32:46we see that deer 18.
03:32:48Lymphocytes have the highest
03:32:50level of Poly functionality,
03:32:51meaning they don't just make 1 sided kind.
03:32:54That makes several interferon gamma,
03:32:56Tina Falfa, and grandson,
03:32:58so they're really quite active
03:33:01effector T cells.
03:33:02But you know,
03:33:03in addition to skewing the differentiation
03:33:05of these stem like precursor cells,
03:33:07we also wondered what was the effect
03:33:10on the frequency of these cells.
03:33:12Now we know that these cells
03:33:14are absolutely critical for
03:33:15immunotherapeutic responses,
03:33:17but the problem is that the checkpoint
03:33:19inhibition by itself and I just
03:33:21want to make a point that studies
03:33:23by Sadiki and others usually looked
03:33:26at combination of checkpoint
03:33:28with vaccination or CTA for blockade,
03:33:30which which will have
03:33:31other signaling effects.
03:33:33Particularly in in delivering a
03:33:34costimulatory signal but PD one blockade
03:33:36alone tends to actually consume.
03:33:38These cells were at least promote
03:33:40their terminal differentiation.
03:33:42This is what Nick hanging hanging script
03:33:44found, and also what we confirmed.
03:33:46Actually we found when we looked
03:33:48in tumors that were treated with
03:33:50anti PD one or Saline that there
03:33:53really wasn't a difference in the
03:33:55absolute numbers or frequency of
03:33:57these stem like precursor cells.
03:33:59Contrast at Dri team dramatically
03:34:01expanded the numbers and frequency of
03:34:03these of these precursor cells in the
03:34:05tumor by greater than an order of magnitude.
03:34:07And really,
03:34:08I think the MFI of the flow cytometry
03:34:10paints quite a picture that there's
03:34:12really more than a decade of staining
03:34:15of TCF one on these on the PD,
03:34:17one positive cells,
03:34:18and there are in the tumor,
03:34:20but I think Nick made a great point.
03:34:22Which is,
03:34:23you know,
03:34:23are these acting on the stem like
03:34:25cells in the tumor and maintaining
03:34:27their stemness or promoting them?
03:34:29Or is it acting upstream in the lymph
03:34:31node and we can't really pinpoint that
03:34:34particular distinction in our tumor,
03:34:35but I do want to point out,
03:34:38and I was inspired by Nick to put this
03:34:40in my talk at the last second that
03:34:42DRA team doesn't dramatically expand
03:34:44the frequency of these of these stem
03:34:47like cells within the draining lymph notes.
03:34:49I think this provides good evidence for Nicks
03:34:52hypothesis that the draining lymph node,
03:34:54maybe that the prime target for
03:34:57checkpoint another immuno therapies.
03:34:59I'm going to switch gears
03:35:00and just say that you know,
03:35:02I told you earlier that that alley team
03:35:04receptors is actually most prevalent.
03:35:06Lee found in on NK cells and if you
03:35:08look in the blood and you stimulate
03:35:11peripheral blood PB MC with ally team.
03:35:13But by far and away the vast
03:35:15majority of cells that respond,
03:35:17ex vivo are natural killer cells and
03:35:19so that made us wonder if there were
03:35:22there were settings where we could
03:35:23leverage that activity and NK cells
03:35:26therapeutically in an obvious candidate
03:35:27for that type of approach would be.
03:35:30In that immune checkpoint
03:35:31resistance setting where loss of
03:35:33antigen presentation is one of the
03:35:35dominant mechanisms of resistance,
03:35:37and in fact it's really a common finding
03:35:40where reduced or absent image C Class one
03:35:43is found in a large fraction of tumors,
03:35:46even prior to any therapy
03:35:48just in the primary setting.
03:35:49Depending on the tumor type,
03:35:51can reach as high as 50% or so.
03:35:56Up now know the dog with me nology would
03:35:58be that if you have a large mass of class,
03:36:02one deficient cells that those
03:36:04cells would be resistant,
03:36:05those cells will be quickly
03:36:06cleared by NK cells.
03:36:08But in reality we know from from.
03:36:10From David relays work that these cells
03:36:13become rapidly exhausted in the tumor in,
03:36:15we essentially have found
03:36:16the same effect that NK cells
03:36:19within tumors have very little activity,
03:36:21and to make a Long story short,
03:36:23we found that we could reverse
03:36:25that phenotype and have promote NK
03:36:27cell responses against numerous and
03:36:29Macy class one deficient tumors.
03:36:31Different models where we deleted
03:36:33beta 2M or where the cells naturally
03:36:36lacked absent in terms of looking
03:36:38at the NK phenotypes that we see.
03:36:40Is that the I-18 treatment
03:36:43drives and queso Mac?
03:36:44Maturation again drives that NK
03:36:47cell polyfunctional phenotype.
03:36:49So just to wrap up,
03:36:50by hopefully I've shown you today
03:36:52that what makes the alley team pathway
03:36:54compelling is that the receptors
03:36:56really selective toward the right T
03:36:58cells as well as natural killer cells
03:37:00that it could be a great target.
03:37:02But for this binding protein,
03:37:04and if you can find ways to avoid
03:37:06the binding protein that that Alley
03:37:07Team receptor agonism is remarkably
03:37:09effective in particularly expanding
03:37:11the numbers of TCF one positive cells
03:37:13and Skewing their differentiation
03:37:14away from the exhaustive Lenny Edge.
03:37:16I just wanted to point out that you know,
03:37:19we're really.
03:37:20You know,
03:37:20motivated to briskly move this
03:37:22into clinical trials,
03:37:23particularly here at the Yale Cancer Center,
03:37:26where our phase one team is so accomplished.
03:37:29And so I thought I'd put this picture,
03:37:31which represents about half a kilogram
03:37:33of our lead DRA teen compound.
03:37:36It's now been produced under GNP conditions,
03:37:38and we should be able to start our
03:37:41first trial in the first half of 2021.
03:37:44So I just want to briefly acknowledge
03:37:47the people who were instrumental
03:37:49in driving this program.
03:37:50My Postdoc Ting Joe,
03:37:51who is just accepted a job at
03:37:54Westlake University in China,
03:37:56graduate student or L Weitzman,
03:37:57who's really actually in the Kiko slab,
03:38:00built Dembski who's a dermatology
03:38:02assistant professor, and of course,
03:38:04our stalwart collaborator Marcus
03:38:05bosenberg in numerous other collaborators
03:38:07here at Yale and elsewhere.
03:38:09So of course,
03:38:10acknowledge our funding sources and
03:38:12thank you so much for your attention.
03:38:17Great, thank you so much air and it's been
03:38:21a pleasure to actually see that story evolved
03:38:24since the first time that we've heard that
03:38:28there's a question from the audience. So
03:38:31do you think D R18 would
03:38:34be useful in lymphomas? Wear? My T 88 is
03:38:38mutated in such things as
03:38:40ABC diffuse, large B cell lymphoma, or in
03:38:43diseases like Walden store macro anemia.
03:38:47Right, yeah, you thank you for that
03:38:50question so you know I'm not really sure
03:38:53what the expression Valley Team receptor
03:38:56would be on the tumor cells in that case,
03:38:59and if it would be desirable actually
03:39:02to further stimulate those pathways.
03:39:04But in terms of thinking about the potential
03:39:08clinical indications for D R18 therapy,
03:39:10I think lymphoma is a good one.
03:39:13Maybe not in monotherapy, but in
03:39:15combination with tumor optimizing auto.
03:39:17Sing antibodies CD20 CD 19,
03:39:19where we know that I'll 18 can
03:39:22robustly stimulated antibody directed
03:39:24cell mediated cytotoxicity ADC.
03:39:26Also, we haven't looked at this yet,
03:39:29but I really want to would be to look at
03:39:33you know 18 agonism in combination with.
03:39:37T cell, engager, therapies,
03:39:39bites, etc,
03:39:39as well as in engineered T cell therapies
03:39:42that that potentially could be used in
03:39:44these types of hematologic malignancy's.
03:39:46So yeah, I don't know if I if I can really
03:39:49say much about the exact mutation here,
03:39:52but I do think that the tile 18
03:39:55could be useful in those pathway
03:39:57in in those malignancies.
03:40:03Ask questions. Yeah,
03:40:06so I was intrigued by your your engineering
03:40:09graph and this idea that it looked
03:40:11like maybe I'm just misreading it,
03:40:14but it looked like gaining that the.
03:40:17The initial contract may have
03:40:19had lower no affinity for I'll
03:40:2118 and the decoy receptor.
03:40:23What really changed was the gain of
03:40:25Receptor expressed receptor affinity.
03:40:27Is that very common when you
03:40:29do these kinds of.
03:40:31Engineering projects are usually go the other
03:40:33way. No, it's yeah, interesting question.
03:40:35Yeah, that's basically universal
03:40:36feature of directed evolution.
03:40:37'cause when you create these
03:40:39libraries were mutagen ising the
03:40:40heck out of the interface right?
03:40:42The average number of mutations
03:40:43here will be like 20 plus mutations.
03:40:45Now of course we widdle that down
03:40:48in our process, where there are
03:40:49only a handful of really good ones.
03:40:52But you can imagine of that 300
03:40:54million mutants that we start with.
03:40:55The vast majority are complete garbage.
03:40:57In fact, they're ruining the molecule.
03:40:59It may not even fold.
03:41:01If it does hold, it certainly doesn't bind,
03:41:03so that's actually what you don't
03:41:05see binding in that first round is
03:41:08because 99.99999% of the students
03:41:10in the library complete garbage,
03:41:11so that's why you know what we're really
03:41:14doing is we're selecting back the
03:41:16mutations that were were busy honing
03:41:18in the needles in the haystack that
03:41:20were present from that large source of.
03:41:23Straw,
03:41:24I guess that's not useful.
03:41:27Can
03:41:28I ask a follow up on that and
03:41:30that's whether or not it's fast?
03:41:32If the mutation if there are mutations,
03:41:35now that you can identify,
03:41:37that may be in LA Team.
03:41:39Are there any patients that might
03:41:42have these kinds of mutations that
03:41:44would give them a phenotype like this?
03:41:47Yes, so in terms of you know
03:41:49mutations that patients might have.
03:41:51Where I like team would be
03:41:53naturally decour resistant.
03:41:54I highly doubt it just because it requires
03:41:56at least five mutations to get there.
03:41:58However, there are patients
03:42:00that have hypomorphic activity
03:42:01in El 18 binding protein.
03:42:02In fact, no Casanova just had a
03:42:05really interesting report of an
03:42:0611 year old girl sadly who passed
03:42:08away she had by allelic loss
03:42:10of violating binding protein.
03:42:11She was fine for most of her
03:42:13life until she got infected with
03:42:15Hepatitis A and she had to run away.
03:42:18Inflammation really highlights
03:42:19the important of these.
03:42:20Of these checkpoints,
03:42:21decor receptors negative feedback
03:42:23at protecting us in situations
03:42:25of extreme inflammation.
03:42:28Did you have question Pam?
03:42:30Go ahead. Yes,
03:42:31great talk really things phenomenal.
03:42:32Do you know if in your model if
03:42:34you use your I'll 18 agent with
03:42:37Anti Sitali 4 is that different
03:42:38than when you use it with anti
03:42:41PD one because I would expect
03:42:42it to be based on some of the
03:42:45mechanism mechanism to show but
03:42:46I don't know if you actually did
03:42:48those experiments.
03:42:49No we never did it and we should.
03:42:51I think the problem is that CTA forces two
03:42:54amazing in mouse models.
03:42:55Listening, he's happy to hear that,
03:42:57but I don't think that amazing 'cause
03:42:58we don't have all patients responding,
03:43:00so I can imagine that your design
03:43:02of your clinical trial would be
03:43:03anti patients who failed anti PD.
03:43:05One can now go on to get your anti PD
03:43:07one plus your L 18 targeting agents.
03:43:09So that would be the combination but
03:43:11it would be interesting to see an
03:43:13anti see Tilly for in the model if you
03:43:15could also do that with patients who
03:43:17failed Anti C telephone anti PD one
03:43:18because that would be spectacular.
03:43:20Yeah I think we think
03:43:21we need to pick the right syngeneic
03:43:23model to really test that but I'll just
03:43:25say briefly know Marcus move on but.
03:43:26We are working with your
03:43:28your colleague on Ming.
03:43:29At the phase one you did MBS MD
03:43:31Anderson to setup a clinical trial.
03:43:34One of the sites over there and
03:43:36so it would be really cool to
03:43:38think about how we can
03:43:39combine it with GTA4 as well.
03:43:41Actually Kelly has some models that
03:43:43will be useful for this where there's
03:43:45metastasis models for failure for
03:43:47both PD one plus CTA four blockade,
03:43:49which is what we commonly get to in Melanoma.
03:43:52So those will be useful models
03:43:54so I'd really like to thank Pam,
03:43:56Nick, Aaron. Great talks.
03:43:57Super exciting,
03:43:58this is when we were planning
03:44:00to start the next session. 5
03:44:02minutes 305 will go for the next session
03:44:05then thanks so much. Will see you
03:44:08guys in a couple of minutes and
03:44:10we've got two more great talks
03:44:12John where Emacs Krummel really
03:44:14really exciting stuff coming so
03:44:16come back and five minutes 305.
03:47:38IMAX and John John, if you can
03:47:40share your screen, I'm sure you're
03:47:42starting to do that anyway.
03:47:44That would be great. Thanks Pam.
03:47:50That should be good, right?
03:47:52It's perfect, just full screen it,
03:47:55and then we're good to go.
03:47:59It's like presenter mode for us,
03:48:02so there we go. How's that perfect?
03:48:06So I think it's a Friday afternoon
03:48:09and we're all looking to get,
03:48:11you know it's been a great day.
03:48:14A lot of lot of exciting stuff.
03:48:17I saw John at your listing
03:48:19in a lot of the talks,
03:48:21which is great sort of look a
03:48:24little bit like an old Ahmed
03:48:26lab meeting probably with Sue.
03:48:28Asking questions from a distance,
03:48:30you know, and
03:48:31a lot of
03:48:32lineages there from
03:48:34that point of view.
03:48:35But but anyway. Nick Joe,
03:48:37she will be moderating this session.
03:48:40Nick needs no introduction after
03:48:41his talk in the last session,
03:48:44so Nick fire away and let's
03:48:46get this going. Alright,
03:48:49thanks so much. So that's real
03:48:50honor to be able to induce the two
03:48:52people in this session to both my 2
03:48:55two heroes of line scientifically.
03:48:56And so I'm going to keep the introduction
03:48:59is very short so we can stay on time. John,
03:49:01we re I think most people in the audience.
03:49:04He needs no introduction has been a
03:49:06real thought leader in the field of
03:49:08trying to understand the mechanisms
03:49:09of T cell exhaustion.
03:49:10I'm just going to mention one thing
03:49:12that I noticed on his bio somewhere
03:49:14that he was actually appointed.
03:49:16He was actually named as one of
03:49:18the 37 under 36 at some point.
03:49:20I don't know how they get that
03:49:22very specific distinction,
03:49:23but that that is 1 claim to fame that he has.
03:49:27Of course many others.
03:49:28Many, many others.
03:49:29He's currently the chair of the
03:49:31Department of systems pharmacology
03:49:33and Translational Therapeutics
03:49:34at Penn and also the director
03:49:36of the Institute for immunology.
03:49:38So thanks for joining us to.
03:49:41And snacking and I guess like Aaron said,
03:49:43I'm definitely much older than 36
03:49:45at this point, but that's a whole
03:49:47different story about that one.
03:49:49Alright, So what I thought I would do
03:49:51is talk a little bit about some of
03:49:53our newer work trying to continue to
03:49:55understand the underlying molecular
03:49:56mechanisms of T cell exhaustion,
03:49:58so couple disclosures few things,
03:50:00but I don't think any of this
03:50:02is too relevant.
03:50:02What I'm going to talk about,
03:50:04so I'm not going to spend a lot
03:50:07of time in the introduction.
03:50:08There are a few simple points I want to
03:50:11make about the topic of T cell exhaustion.
03:50:13Over the years,
03:50:14we and many others studying T
03:50:16cell exhaustion in settings of
03:50:18chronic infection or cancer,
03:50:19or now autoimmunity,
03:50:20have, I think,
03:50:22come to the conclusion that exhausted
03:50:24T cells are distinct lineages that
03:50:26are as distinct from effector cells
03:50:28as memory cells or tissue resident
03:50:30memory or our other things and.
03:50:32And if that's true,
03:50:34then we can think about Therapeutics,
03:50:36the way an effector cell works.
03:50:38We have to think about Therapeutics,
03:50:40the way that they work on exhaust T cells.
03:50:43And so recently we've uncovered some
03:50:45of the transcriptional players in this,
03:50:47including talks,
03:50:48which lays down essentially the
03:50:49epigenetic tracks or the epigenetic
03:50:51program to divert exhausted T
03:50:53cells down that Lenny Edge,
03:50:54and actually a big role.
03:50:56What talks does is repressed the
03:50:58formation of terminal effector cells.
03:50:59It works in concert with TCF one,
03:51:01which of course has a role both in
03:51:04exhausted cells and memory cells.
03:51:06Although you can see a little bit of
03:51:08talks expression in some other cell
03:51:10populations and those other cell populations,
03:51:12when you get rid of talks.
03:51:14There's no effect,
03:51:15but if you get rid of talks,
03:51:17you simply don't get the
03:51:18lineages exhausted T cells.
03:51:19So we know a little bit about how this works.
03:51:23So first of all, why do we care?
03:51:25Of course, why do we care about this?
03:51:27Well,
03:51:28Raffi touching this little bit
03:51:29this morning and we care about this
03:51:31because exhausted T cells are at
03:51:33least one of the major cell populations
03:51:35responding to checkpoint blockade in humans,
03:51:37and probably more accurately, there.
03:51:39One of the major populations
03:51:40responding to P1 blockade.
03:51:42They may also respond to CTA for blockade,
03:51:44but there may be other cells
03:51:46involved there as well.
03:51:47So over the years,
03:51:48we and many others have examined the
03:51:50cells responding to PD one blockade using.
03:51:53Various forms of mean profiling
03:51:54and if you block the PD 1 pathway
03:51:57and then take the responding cells,
03:51:59look at them to try to understand
03:52:01what they are.
03:52:02The answer you get is that their
03:52:04share many of the same features as
03:52:06exhausted T cells from mouse models.
03:52:09So if you sort out the essentially
03:52:10the Ki 67 positive or the responding
03:52:12cells to PD one blockade in human
03:52:15Melanoma patients perform
03:52:16transcriptional profiling.
03:52:17What you find is that the underlying
03:52:20transcriptional program of
03:52:21these cells is most similar
03:52:22to an exhausted T cell.
03:52:24Yes, they re engage affecter biology
03:52:25when you block the PD 1 pathway.
03:52:27Yes, they turn effector genes back on.
03:52:30Yes they start Cliff rating but their
03:52:32broader wiring of what their cell identity
03:52:34is is actually much more related to
03:52:36exhausted T cells and effector T cells.
03:52:38We can see this in a variety of ways,
03:52:40and it's actually those exhausted T
03:52:42cells that correlate the most with
03:52:44the anti tumor response and we can
03:52:46find even in the peripheral blood
03:52:48some of the same T cell clones that
03:52:50you find in the tumors to Justin.
03:52:52What we see in the blood is actually.
03:52:54Playing a role in in the clinical response
03:52:57that we see and so from work from our group,
03:53:01but also many other groups,
03:53:02we know that exhausted T cells are the major
03:53:05responding cell type to PD one blockade.
03:53:08In humans we can identify exhausted T
03:53:10cells based on in humans Co expression
03:53:12patterns of inhibitory receptors is
03:53:14the easiest thing by flow cytometry,
03:53:16but we can confirm that transcriptionally
03:53:18and we now have good epigenetic
03:53:20data on human FINA,
03:53:22typically defined exhausted T
03:53:23cells than they actually are very,
03:53:25very distinctly.
03:53:26Exhausted at the epigenetic level as well,
03:53:28we can find some of these cells
03:53:30in the blood in some settings,
03:53:32rocky showed you this morning and head
03:53:34neck cancers are very hard to find.
03:53:36A blood in Melanoma.
03:53:37It's a little bit easier and allowing
03:53:38identifying which cell types are
03:53:40responding allows us to identify
03:53:41potential resistance mechanisms and
03:53:43their variety that I won't go into here,
03:53:45some of which are published in
03:53:47these papers and others here.
03:53:48Probably more expert in some
03:53:50of those things than I am.
03:53:53So,
03:53:53but to give a little bit more context
03:53:55on exhausted T cells and some of this,
03:53:57Raffi touched on this morning,
03:53:59and others have said along the way as well.
03:54:01I did want to touch in just a
03:54:03couple of key topics.
03:54:04So first,
03:54:05these are the cells responding
03:54:06to checkpoint blockade in humans,
03:54:07but they are epigenetically very
03:54:09distinct from other cell types.
03:54:10So this is just an example of
03:54:11studies that we published together
03:54:13with a study from McCain isgroup
03:54:14identifying the sort of epigenetic
03:54:16landscape of exhausted T cells,
03:54:17and there are a couple of very
03:54:19simple points from these studies,
03:54:21and since there are a couple years old,
03:54:23I just hit the high points.
03:54:24One,
03:54:25there are clear epigenetic distinctions
03:54:26and exhausted T cells that are markedly
03:54:28different from affecting memory T cells,
03:54:30including something like this
03:54:32and emblematic open chromatin
03:54:33region in the PD one locus PDC,
03:54:35one Locus Encoding Key,
03:54:37one that you only find and exhausted T cells.
03:54:40Or we now know regulatory T cells or TFH,
03:54:42all of which expresses very
03:54:44high levels of PD one.
03:54:46Overall,
03:54:46there are 18,000 open chromatin
03:54:48regions that differ from naive cells.
03:54:50A third of those are 6000 of those are unique
03:54:53to exhausted T cells.
03:54:55That's the same order of magnitude
03:54:57of difference that you have between
03:54:58a my Lloyd cell and a B cell,
03:55:01or between a B cell in a T cell,
03:55:03really indicating how distinct
03:55:04exhausted T cells are among the CDA,
03:55:06T cells, and so you can lay that
03:55:08out in principle component spacing
03:55:09the distance between the cells just
03:55:12illustrates that point I just made,
03:55:13but importantly,
03:55:14when you block the PD 1 pathway
03:55:15that epigenetic landscape
03:55:16doesn't change very much.
03:55:18In other words,
03:55:19despite the fact that you're
03:55:20reengaging transcriptional
03:55:21circuits of effector activity.
03:55:22Your epigenetic identity remains exhausted,
03:55:24so when that effect the PD one
03:55:26blockade wears off, which it does.
03:55:28Even if you continue to treat,
03:55:30the cell reverts to its
03:55:32essentially its identity,
03:55:33and its lifestyle is exhausted T cell
03:55:36rather than being converted into an
03:55:37effector or a memory cell or some other
03:55:40kind of more durable stuff population,
03:55:42and then I mentioned this already,
03:55:44but it's actually this transcription factor.
03:55:46This H MG Group transcription factor
03:55:49talks that actually programs the vast
03:55:51majority of the epigenetic changes
03:55:53associated with exhausted T cells.
03:55:55So I want to talk a little bit
03:55:57more about sort of the lineages
03:55:59development of exhausted T cells and
03:56:00this is a map kind of illustrating.
03:56:02Summarizing really,
03:56:03a lot of data in the field,
03:56:05but it kind of articulates this
03:56:06idea that early on after activation
03:56:08there are populations of.
03:56:09I'm sorry the terminology.
03:56:10This feels a little bit confusing.
03:56:12We call these the precursors
03:56:13because they are the precursors
03:56:15that can give rise to the memory.
03:56:16Lineages are the precursors that
03:56:18can give rise to the effectors
03:56:19and then the precursors that can
03:56:21give rise to exhausted T cells.
03:56:23We know that talks and TCF one play
03:56:25an important role here in determining
03:56:27which of these lineages the self goes down.
03:56:30And then once there's a commitment
03:56:32to the exhausted linkage,
03:56:34there are these exhausted.
03:56:35We call them progenitors.
03:56:36We identified them well over a decade ago.
03:56:39These cells that actually contain
03:56:41all of the ability to respond to PD
03:56:43one blockade those cells overtime
03:56:45in the response to persisting
03:56:47antigen can give rise to downstream
03:56:49populations of more term or
03:56:51terminally differentiated exhausted
03:56:52T cells including an intermediate.
03:56:53Population that expresses Tibetan
03:56:55circulates and then finally a
03:56:57terminal population that becomes
03:56:58resident in tumors or Ant issues.
03:57:00So the question for us is we wanted
03:57:02to understand more about this initial
03:57:04programming and we know that talks and
03:57:06TCF one do not explain the whole story,
03:57:09so we wanted to design a system to ask
03:57:12what else might be playing a key role
03:57:14in these fate decisions that determine
03:57:16the Lenny edge of exhausted T cells.
03:57:19So how do we discover novel
03:57:21regulators of this exhausted
03:57:22versus effector differentiation?
03:57:23Each choice at the beginning.
03:57:25And so,
03:57:26like Arlene and like others,
03:57:28city,
03:57:28Chen and others,
03:57:30Alex Marson,
03:57:30we've developed CRISPR screening
03:57:32system to try to identify some
03:57:34of those key regulators.
03:57:36Our system is a little bit
03:57:38different than Arlene's,
03:57:39but basically tries to achieve the
03:57:42same goals of doing in vivo screening.
03:57:44What we've done is use a cast 9
03:57:47transgenic mouse on LC MB specific 14
03:57:50background and into those P14 cells,
03:57:53we actually transducing retroviruses
03:57:54expressing the guide RNAs.
03:57:56And then put those retro virally transduced
03:57:58T cells back into infection match.
03:58:01My so we're doing the transduction
03:58:03and deletion using this in vitro
03:58:05incubation system and then we save an
03:58:08aliquot of those cells and then put
03:58:10the rest of them back into mice that
03:58:13are infected with the LCD Armstrong.
03:58:15Or else you clone 13 to get
03:58:18chronic infection,
03:58:18allow selection to occur in Vivo,
03:58:21which in our hands works several
03:58:23orders of magnitude more efficiently
03:58:25than in vitro selection.
03:58:27Isolate cells at various time points
03:58:29after that selection compared to the
03:58:30input to get a efficiency score.
03:58:32So we've done a few things here.
03:58:34This is all the work of a graduate
03:58:36student zeegen who actually
03:58:37just defended his thesis and our
03:58:39collaborators in ways she has really
03:58:41helped us optimize the system and
03:58:42what we wanted to do first is actually
03:58:45go in with a focused library,
03:58:46so we can't do genome wide,
03:58:48and this to be transferred too
03:58:50many T cells in the system.
03:58:51You're turning your chronic infection,
03:58:53acute infection,
03:58:54so we're putting in numbers
03:58:55of T cells that allow us to.
03:58:57These are the normal pathogenesis
03:58:59of whatever the model is for using.
03:59:01In our case that allows us to do
03:59:04targets about 1:50 targets at
03:59:06once with a 5 guides for Target.
03:59:08In this case,
03:59:09we're targeting only transcription
03:59:11factors as sort of the key regulators
03:59:13of those fate choices,
03:59:14so we went with the Library of
03:59:17150 transcription factors.
03:59:18And so, like Arlene, we use PD one.
03:59:21Targeting is one of the positive controls.
03:59:24When you do this with the Library
03:59:26of transcription factors,
03:59:27a few negative controls and PD
03:59:29one is positive control.
03:59:30PD one comes out as the winner.
03:59:32Not surprisingly,
03:59:32it's when you get rid of P1,
03:59:34you massively, essentially Dearie,
03:59:36Press the response and get this very,
03:59:38very robust expansion of those cells,
03:59:39and then in fact we no longer use PD
03:59:42one because the number of sequencing
03:59:44reads devoted to PD one or library
03:59:46kind of swamps out other things.
03:59:48But we also find a lot of
03:59:50the expected players,
03:59:51so this is sort of the opposite of
03:59:53bill came and called the up screen.
03:59:55This is the loss of function.
03:59:57Of the 80 FIR 4C Myc,
03:59:59without which you simply can't get it,
04:00:01T cell response.
04:00:02These were nice positive controls to another.
04:00:04If we target these transcription
04:00:06factors that are important in the
04:00:08very earliest stages of a factor
04:00:09or activated T cell programming,
04:00:11essentially nothing happens when
04:00:12we lose them from the library.
04:00:14We also pulled out Tibet mid like 2
04:00:17and several others that have a key
04:00:19role in initiating the activated
04:00:21T cell response, but we were
04:00:23more interested in what happens
04:00:24on the up side of the screen.
04:00:27So what are the?
04:00:28Other things that appear to
04:00:29repress the response of T cells
04:00:31during chronic infection or the
04:00:33initial development of exhaustion,
04:00:34and there were clearly some
04:00:36other interesting transcription
04:00:37factors here smads left one we
04:00:39heard about earlier Rocky Mount.
04:00:40You can't tell from this normalized score,
04:00:43but this in vivo system allows us to
04:00:45obtain somewhere between 20 and 100
04:00:47fold enrichment so many fold better
04:00:50than what you would get in in vitro
04:00:52screen that has to do just with the
04:00:54in vivo T cell expansion that you
04:00:56get and giving you a little bit more.
04:00:59Resolution of this.
04:01:00You can see PD one smad TCF 7 talks all
04:01:04coming up on the upside of this screen.
04:01:07IRA Forebet Fmic and several
04:01:09others in the downside.
04:01:10So looking at this a little bit
04:01:13more again looking for things that
04:01:14seem to repress the response.
04:01:16There are some interesting
04:01:17transcription factors here,
04:01:18and the Icarus family in the end.
04:01:20Fat family.
04:01:21But actually this transcription
04:01:22factor fly one really caught our
04:01:24attention one because we just
04:01:26simply didn't know much about it,
04:01:28but too because it was actually
04:01:29the second most prominent.
04:01:31Hit in the screen after PD one.
04:01:34So what do we know about fly one?
04:01:36So fly one is Annette family
04:01:38transcription factor regulates matter
04:01:40poetic and Progenitor Cell Biology.
04:01:41In other systems it has a key
04:01:43role in AML tumor progression.
04:01:45Probably playing a role in the
04:01:47AML stem cell tumor stem cell.
04:01:49And there are genetic associations of
04:01:51fly one deficiency with lupus and psorosis.
04:01:53I'm not sure I fully understand
04:01:55how that works,
04:01:56but it seems to play a role in
04:01:58some of the inflammation that
04:02:00accompanies both of these diseases.
04:02:02The place where this has been
04:02:04best studied is actually in this.
04:02:06Oncogene Fusion protein DWS fly.
04:02:08One mutation that drives Ewing
04:02:10sarcoma here what's interesting
04:02:11is fly one mediates chromatin
04:02:13remodeling in the tumor cells.
04:02:15Because the pryon domain of DWS
04:02:17essentially drags fly 12 regions of
04:02:19the genome where there shouldn't
04:02:21be chromatin rearrangement,
04:02:23and again ways that I think are
04:02:25still poorly understood,
04:02:27that actually plays a role in driving
04:02:29the oncogenesis of Ewing sarcoma,
04:02:31so that was interesting to us
04:02:34and suggested that a fly one was.
04:02:36Playing a role in chromatin remodeling,
04:02:38perhaps this was interesting to
04:02:40take a look at in CD8T cells.
04:02:42So we want to understand if we target
04:02:45only fly one, what would happen.
04:02:47So we chose two of the better guides,
04:02:50two 19360 and just wanted to confirm
04:02:52that they actually reduce protein expression.
04:02:54You can see that here and either
04:02:57Armstrong or clone 13 you get
04:02:59about an 80% reduction in protein
04:03:01and when you do that,
04:03:03you massively Dearie Press the initial
04:03:05response to acute LCD Armstrong infection.
04:03:07So you don't miss tenfold increase in the
04:03:10number of cells that they ate and a 15.
04:03:13When you look at what those cells look like,
04:03:16what you find is you've actually
04:03:18substantially increased portion of the
04:03:19response that expresses Cal, RG, one.
04:03:21So these now look like the more affecter
04:03:23like cells in Armstrong infection
04:03:25compared to the memory precursor.
04:03:27This is Day 15, so this is a pretty
04:03:30substantial increase in that killer
04:03:32T1 population at this time point.
04:03:34You can do the opposite,
04:03:36where if you overexpress fly one you
04:03:38actually repress the response just for
04:03:40the retroviral overexpression strategy
04:03:41can see that here at the eight word 815.
04:03:44Now what's interesting about this is that
04:03:47when you knockout fly one and you get
04:03:49this massive increase in cell numbers,
04:03:51you get this huge increasing
04:03:53the effective response.
04:03:54It looks like the memory precursor population
04:03:56decreases and it does proportionally,
04:03:58but the absolute number of these
04:04:00memory precursors does not change
04:04:02the numerical increase overall is
04:04:03essentially this affecter over sheet,
04:04:05so we wanted to ask what happened during
04:04:08chronic infection where that could be
04:04:10very valuable to have more effector cells,
04:04:12and indeed using different markers.
04:04:14Here CD 39 and Y 108 as a server for TCF one.
04:04:18You essentially see the same biology.
04:04:20That is when you target fly one,
04:04:22you push the response towards
04:04:24predominantly filling in this.
04:04:25Population of TCF negative or like
04:04:27108 negative CD 39 positive more
04:04:29affecter like cells in the early
04:04:31part of chronic infection.
04:04:33And that's just summarized
04:04:35over here numerically.
04:04:36On the right hand side again that's
04:04:39true proportionally for the increase
04:04:41in these affecter like cells,
04:04:43this decrease in the TCF rely 108 positive
04:04:45population is a proportional decrease.
04:04:47The absolute numbers actually
04:04:50remained constantly.
04:04:51So I wanted to understand what the underlying
04:04:53biology this was and if you do
04:04:55transcriptional profiling of
04:04:56these fly one deleted cells.
04:04:58What you find is a massive loss in
04:05:01the transcriptional signature of
04:05:02T cell exhaustion and a game in
04:05:04the signature of effector T cells,
04:05:06and you can sort of see this over
04:05:08here in the heat map where the jeans
04:05:11that are increasing expression on
04:05:13the bottom half is basically a who's
04:05:15who list of effector T cell biology
04:05:17including many of the Cal are molecules,
04:05:19CX3 CR 1, Tim 3.
04:05:21Taylor G1, Granzyme B.
04:05:22Whereas the jeans that are actually
04:05:24proportionally lower are some of the
04:05:26genes involved in progenitor biology,
04:05:28including TCF 7 ID 3 and others.
04:05:30Because supply one had a role
04:05:34in this EWS fly one.
04:05:36AKA Gene in regulating epigenetics
04:05:38wanted to look at the open chromatin
04:05:41landscape in the absence of fly one.
04:05:43And indeed if you do a taxi here the fly one.
04:05:47Deleted cells are very different
04:05:49from the control.
04:05:50If you map those changes to nearest gene.
04:05:53What you find is that a massive
04:05:55amount of this change actually occurs
04:05:57at genes related to the effector
04:05:59response you see increased chromatin
04:06:01Accessibility near genes and pathways
04:06:03involved in effector biology and
04:06:05a decrease relatively speaking.
04:06:07In chromatin Accessibility around genes
04:06:08that may be involved in progenitor biology,
04:06:10again,
04:06:11very similar lists of the
04:06:12jeans that are changed by RNA.
04:06:14If you look at the correlation
04:06:16between the two,
04:06:17there's a clear relationship you
04:06:18can look underneath this open
04:06:20chromatin to see what transcription
04:06:21factor binding sites may become.
04:06:23Less or more iaccessible.
04:06:24You lose Accessibility around sites.
04:06:26That combined, I RF molecules,
04:06:28particularly RF one and two,
04:06:30which may be interesting in terms
04:06:32of regulation of this exhausted
04:06:34T cell response by interference,
04:06:36but by far the biggest changes
04:06:39and increasing Accessibility at
04:06:40these sort of hybrid.
04:06:42Let's run sites massively increased in
04:06:44Accessibility when you remove fly one,
04:06:46so we wanted to understand the relationship
04:06:49of these changes to actually wear fly,
04:06:52one binds and performed,
04:06:53cut and run.
04:06:54Just to identify workflow is binding
04:06:56and then ask at those sites a water
04:06:59the sequences underneath those Ann B.
04:07:01How does that relate to the
04:07:03changes in chromatin Accessibility?
04:07:04And indeed we can find fly one
04:07:06binding at sites where chromatin
04:07:08becomes differentially accessible.
04:07:09The number one transcription
04:07:11factor binding sites under sites
04:07:13were fly one binds the fly,
04:07:14one binding site,
04:07:15although there are others here,
04:07:17including a runs binding site.
04:07:19What's interesting about this is
04:07:20that 78% of the places reply,
04:07:22one binds you see an increase in chromatin
04:07:25Accessibility when fly one is deleted.
04:07:27Suggesting that fly one is
04:07:30antagonizing chromatin Accessibility?
04:07:31Under those sites you find lots
04:07:34of effector genes close by.
04:07:35You do not find fly one binding at
04:07:38genes involved progenitor biology,
04:07:40So what this suggested to us was
04:07:42that fly one is playing a role
04:07:45in essentially safeguarding the
04:07:46effector response.
04:07:47The effector biology,
04:07:48so you don't have an overaggressive
04:07:50affective response,
04:07:51but actually has very
04:07:52little role in the transcriptional
04:07:54coordination of the genes
04:07:56involved in progenitor biology,
04:07:57which would be consistent with an increase
04:07:59in effector cells or Affecter lineage
04:08:02cells in acute and chronic infections.
04:08:04Without a loss of those progenitor,
04:08:06like cells for the cells will give
04:08:08rise to memory and exhausted T cells,
04:08:11so it's great fly,
04:08:12one repressive chromatin Accessibility
04:08:13binding effector like jeans,
04:08:15particularly those that
04:08:16will fly one runs motifs.
04:08:17Is there a connection between
04:08:19fly one and runx proteins?
04:08:21So we did an experiment to try to
04:08:23test this in the basic idea of this
04:08:26experiment was to delete fly one and
04:08:28enforce expression of the ranks.
04:08:29One or runs three.
04:08:31When you do this,
04:08:32we're using code transduction
04:08:33with two different retroviruses,
04:08:34one containing the guide to delete fly
04:08:36one and the other overexpressing runs three.
04:08:38You can just look at sort
04:08:40of the upper right quadrant.
04:08:41You do nothing about 10% of the response
04:08:44is double transduced with the empty vectors.
04:08:46If you overexpress runs three,
04:08:47you push a little bit better
04:08:49diesel expansion,
04:08:49and they're actually more killer.
04:08:51Do you want high cells?
04:08:52If you just delete fly one
04:08:54about the same effect,
04:08:55both in terms of numbers and
04:08:57also differentiation state.
04:08:58But if you do both things at the same time,
04:09:01you actually get at least an
04:09:02additive affect suggesting that.
04:09:04Removing fly one allows runx three
04:09:07to function more effectively.
04:09:09So fly one last cooperates with
04:09:11runs three overexpression to drive
04:09:13more of an effector expansion and
04:09:15effector response fly one seems to
04:09:16antagonize that should be runs three
04:09:18Accessibility and limit the effector program.
04:09:20There's some biology of Runx,
04:09:22one that I'm not coming into here,
04:09:24so this is the idea that you have only one,
04:09:27essentially safeguarding parts
04:09:28of the genome where runs three
04:09:31could bind when you remove fly,
04:09:32one runs three, has more Accessibility,
04:09:34can drive those,
04:09:35runs three dependent effector genes
04:09:37more effectively. So just
04:09:38in the last
04:09:39couple slides, of course.
04:09:40Want to address whether any of
04:09:42this matters at all? Of course.
04:09:44All goal of this is to see whether
04:09:46we can improve the efficiency
04:09:48of T cell responses during
04:09:50chronic infections and cancer.
04:09:51So let me just show you a couple
04:09:54of examples of how we tested
04:09:56this in models of infection,
04:09:57and so the simple experimental
04:09:59design here is we took mice,
04:10:01infected them with the pathogen,
04:10:03left them as you're just B6 mice.
04:10:05We either did not transfer any T cells
04:10:07we adoptively transferred in low
04:10:09numbers of P14 cells transduced with
04:10:11a control guide user cast 9014 cells.
04:10:13Or cast 9014 cells that actually had
04:10:16fly one targeted in all cases were
04:10:18using pathogens that express the
04:10:20GP 33 epitope matched to the P14,
04:10:22so we're putting in.
04:10:24So in chronic calcium,
04:10:25be putting in in grey just non
04:10:27targeted T cells gives us a little
04:10:29bit of a benefit because we're
04:10:31putting in diesel specific for
04:10:32the virus you see viral load go
04:10:34down and the Serum and liver and
04:10:36perhaps a little bit in the kidney.
04:10:38But when you actually delete fly one
04:10:40you see a substantially greater benefit.
04:10:42A controlling chronic infection
04:10:43because of loss.
04:10:44Apply one you do the same experiment in
04:10:46respiratory infection with influenza virus,
04:10:47where if you don't put in any
04:10:49T cells that might get sick in
04:10:51viral load is fairly high.
04:10:53You put in just non targeted T cells.
04:10:55Actually you do have a little bit of a
04:10:57benefit on pathogenesis and maybe a.
04:10:59Into viral control.
04:11:00Although it's not insignificant,
04:11:01you delete fly one.
04:11:02The mice are much healthier and you see
04:11:05substantial control of our replication.
04:11:07Interesting Lee in my still
04:11:08have high viral load.
04:11:09You see,
04:11:10actually much much greater
04:11:11expansion of the fly.
04:11:13One deleted T cells compared to control
04:11:15mice that had the same viral load.
04:11:18And then in systemic listeria
04:11:19monocytogenes infection,
04:11:20the stories the same,
04:11:21the mice do better when you
04:11:23actually delete fly one pathogen
04:11:26burden is substantially lower.
04:11:28So finally,
04:11:28what happens in a tumor setting where
04:11:30you might want to get tumor control.
04:11:32So in this case we did this.
04:11:34Actually rag two mice,
04:11:35so we could actually look
04:11:37at longer term affects,
04:11:38but you can do this also in B6 Mice
04:11:40and I'm happy to explain that a
04:11:42little bit more people are interested,
04:11:44but essentially you get the same answer.
04:11:46We established tumors and five days
04:11:48after the tumors have started growing,
04:11:49we do the same experiment putting
04:11:51in non control T cells or T cells
04:11:53that have fly one deleted and you
04:11:55can see that pretty much black
04:11:57and white answer that.
04:11:58Compared to T2,
04:11:59putting in just wild type T cells to
04:12:01control the tumor a little bit to
04:12:03fly one deleted tumors actually are
04:12:06substantially better control the fly one.
04:12:08Deleted T cells provide
04:12:10substantially better control of
04:12:11the tumor in these settings.
04:12:13So what I've shown you is this system,
04:12:16which we've turned optics,
04:12:17'cause everybody needs an acronym for
04:12:19their Christmas screening approach,
04:12:21is optimized for really focused,
04:12:23high resolution screens,
04:12:24gives us 20 to 100 fold resolution to
04:12:26understand the biology of T cell exhaustion.
04:12:29T cell differentiation in Vivo.
04:12:30We identified apply one as a key target and
04:12:33started to understand some of the biology.
04:12:36But there are also some other very
04:12:38interesting targets identified here,
04:12:40especially on the up side of the
04:12:42screen including Smad 2 EGRIR F2.
04:12:44One of the near 70 set or the newer
04:12:47are for a family members and ATF.
04:12:50Six loss of fly one enhances
04:12:52effector differentiation.
04:12:52But importantly,
04:12:53this occurs without compromising the
04:12:55memory or progenitor populations.
04:12:56So I think it's interesting to note that
04:12:58most of the other transcription factors
04:13:00that play this sort of toggling role
04:13:03between effector differentiation and memory,
04:13:05or effector and exhaustion when you delete
04:13:07them and promote effector differentiation,
04:13:10you actually lose the
04:13:11other lineages sells it.
04:13:12Sort of you can't have your
04:13:15cake and eat it too.
04:13:17This case will fly one deletion,
04:13:19since 51 seems to act as an
04:13:21effector lineages safeguard,
04:13:22rather than something that promotes
04:13:24the progenitor populations of
04:13:25exhausted or memory cells.
04:13:26We don't seem to have a loss of
04:13:28those key populations important
04:13:30for long-term immunity.
04:13:32The mechanism for fly one appears to be
04:13:35interacting with the access of Runx 3.
04:13:37Two genes involved in the effector
04:13:39biology and so removing fly one
04:13:41allows runs three to work more
04:13:44efficiently because it binds
04:13:45directly to fly one monks motifs.
04:13:47So we're working hard on using this
04:13:49information and the identification role
04:13:51for fly one and protective immunity
04:13:54to see whether we may be able to
04:13:56improve things like cellular therapies
04:13:58and car T cell models and other
04:14:00settings of adoptive cellular therapy.
04:14:02So stop here that you can.
04:14:04Is the student in the lab now.
04:14:06Postdoc is actually leaving to go
04:14:08to Brad Bernstein's lab very soon,
04:14:10who did all of this work in
04:14:12collaboration with you.
04:14:13My she's lab.
04:14:13Lots of collaborations with
04:14:15Shelly Burger on the epigenetics,
04:14:16Josephine Giles.
04:14:17Alex Wong played a role in some of
04:14:19the Melanoma work that I mentioned
04:14:21in the beginning.
04:14:22Omar, Coniston,
04:14:23all of the previous talks work
04:14:24with great collaborators in human
04:14:26Melanoma Melanoma side that I
04:14:28mentioned just in passing at the
04:14:30beginning and lots of thanks to
04:14:31a lot of the funding behind this.
04:14:33So stop there and happy to
04:14:35take questions at this time.
04:14:38Alright, Thanks John.
04:14:40So we have waiting for questions.
04:14:43I guess I wanted to ask
04:14:47about this question but I
04:14:50think this is an interesting idea
04:14:53and it's one of the areas that.
04:14:58Conceptually, it's very hard to put together,
04:15:01which is this idea of with the
04:15:03idea of stem like cells and
04:15:05trying to drive effector T cells.
04:15:08Always this trade off,
04:15:09and so you sort of have identified something
04:15:11here which may not have that trade off.
04:15:14And I like the mechanism potentially
04:15:16of being able to take the brakes off,
04:15:19but it does.
04:15:20It does suggest that these
04:15:22are different processes,
04:15:23especially as opposed to one process right?
04:15:25Like the that we've always.
04:15:27I guess I've always thought about
04:15:29the trade off as being you have
04:15:32to get one to get the other.
04:15:34That's that's what's going on.
04:15:36Or is it just a?
04:15:37Is there something more there
04:15:39than than I'm
04:15:39missing? I mean, that's
04:15:41what we think is happening and there
04:15:43are a couple different potential
04:15:44ways that that might play out and
04:15:47we don't have answers to this yet,
04:15:48but the simplest one is that fly
04:15:50one only plays a role after you've
04:15:52made that switch from whatever that
04:15:54progenitor precursor population
04:15:56is to the effector Lenny edge.
04:15:57Then in the effector Lenny edge fly,
04:15:59one is playing a role,
04:16:01probably to limit immunopathology.
04:16:02It just sort of restraining runs three.
04:16:04So we can start to test that.
04:16:06I mean, I think we really need to do to
04:16:08test that properly is do conditional
04:16:10deletion of fly one at later time points,
04:16:13so you know once you firmly established the
04:16:14exhausted lininger firmly established memory,
04:16:16then delete fly.
04:16:17Wanna make sure that there's not an impact.
04:16:19We also need to carry on some of these
04:16:21experiments a little bit longer and
04:16:23try to understand whether there's
04:16:24any long-term defect in memory or
04:16:26do re challenge experiments.
04:16:27We've not been able to do for
04:16:29some technical reasons.
04:16:30Right now,
04:16:31we're comfortable with sort of similar
04:16:33interpretation of that based on the
04:16:34data that we have that is just acting.
04:16:36Out of state,
04:16:37once you've already made the
04:16:38commitment to the effector Lenny
04:16:39edge and you could actually push the
04:16:41effector Lenny Edge even farther.
04:16:44So there's a couple of questions
04:16:45in the chat when I'm gonna
04:16:47ask is about the downsides to
04:16:49removing fly one. Yeah,
04:16:50so? I mean, there's always a downside
04:16:52to getting more effector cells,
04:16:53and that's immunopathology,
04:16:54and so this really depends on your
04:16:56setting and how much antigen there is.
04:16:58And things like that.
04:16:59And all the folks on the line who
04:17:01worked with LCD kind of understand
04:17:03how this might work and in the
04:17:05chronic infection experiments
04:17:06that I showed in one experiment,
04:17:07we actually had our cell numbers maybe 20
04:17:10or 30% higher than they should have been.
04:17:12And sure enough,
04:17:13we had lethal immunopathology.
04:17:14My cell had to be sacked because we lost,
04:17:17you know 3040% body weight.
04:17:18So yes,
04:17:19you do need to temper immunopathology
04:17:20and that does actually play role
04:17:22in humans in some settings and
04:17:24probably a subset of our kovid
04:17:26patients are experiencing some form
04:17:27of T cell mediated immune pathology
04:17:29where exactly the T cells fit in.
04:17:31There is a whole different conversation,
04:17:33so there is a downside if you're not careful.
04:17:37So sorry
04:17:38so the clarify John.
04:17:40So you're saying that there's
04:17:42a dose response relationship
04:17:43in terms of autoimmunity with
04:17:45with the fly one deleted cells.
04:17:47So if you give more adoptively
04:17:50transferred you see pathology. I would
04:17:52say it's a mythology, not
04:17:54autoimmunity. In this case and the way
04:17:57it manifests here is we can clearly see
04:18:00it in else MV because LCD is systemic and
04:18:03because there's an engine everywhere.
04:18:05You may see some of it.
04:18:08Influ if you're just on the
04:18:09wrong side of the Bell curve.
04:18:11But you're not going to see any of it really
04:18:13in this sort of Implantable tumor model,
04:18:15because the only place you have
04:18:16pathology would be in the tumor itself.
04:18:18That's one place the Antigen is,
04:18:20so the immunopathology risk actually can
04:18:21only really be evaluated when you also
04:18:23know where and how much energy you have.
04:18:25If it's systemic,
04:18:25the risk will be much higher than
04:18:27if it's local. OK, sorry,
04:18:30my zoom keeps cutting out
04:18:31and restarting,
04:18:32so I apologize if I missed something.
04:18:34There was one other question so
04:18:36I have to re open my chat that
04:18:38was there was about does removal
04:18:40of fly one Excel right?
04:18:42Resident memory development issue?
04:18:43Yeah yeah so great.
04:18:44I was actually just
04:18:45editing that paragraph of
04:18:46the paper this morning.
04:18:48So yeah, we actually think that
04:18:50this may be playing a role as well.
04:18:52Obviously runs three is a key role in
04:18:54resident memory and in some of the
04:18:56models that I showed you resident
04:18:58memory will play an important.
04:19:00Have an important role in protecting unity,
04:19:02especially that flew model that I showed.
04:19:04We actually don't know much about this yet,
04:19:06it's something that we're working on.
04:19:08We suspect that there will be an
04:19:09impact on resident memory as well,
04:19:11but we don't get it.
04:19:13Awesome, well thank you
04:19:15so much. John is real pleasure listening
04:19:17to you for the interest of time.
04:19:19I'm going to move on and it's a real
04:19:21pleasure also to introduce Max Promo.
04:19:23I'm not going to say much
04:19:25because of the interest of time,
04:19:27but there's a ton you could say about
04:19:29Max in terms of him having been sort
04:19:31of at the at the right places to be
04:19:34have a huge impact in terms of therapy.
04:19:36Was one of the first people to study.
04:19:39See till four in mice and really just
04:19:41has had a huge impact on the field.
04:19:43In his own lab,
04:19:45they've done a lot of just beautiful work,
04:19:47and I think that's the one thing
04:19:49that really stands out about.
04:19:51Max is work is always just
04:19:53amazingly beautiful.
04:19:53I can see in the back of his picture
04:19:56he's got one of his his, his.
04:19:58Images of a tumor or something
04:20:01in the back there.
04:20:02It's always beautiful work.
04:20:03So Max is the chair of sorry.
04:20:06Is the professor in the Department of
04:20:08pathology and he's a inaugural chair
04:20:10of the amino X initiative and it's real
04:20:13pleasure to have him to talk to them.
04:20:17Thanks Nick. In a room Mary yes great.
04:20:20Well I want to thank you Yale
04:20:22Cancer Center for setting this up.
04:20:24I'm hopeful that when Kovid is
04:20:25over that we can continue to learn
04:20:27how to do these kinds of meetings
04:20:29without flying all over the world.
04:20:31That's really useful for interactivity.
04:20:33I think for all of us,
04:20:35and we need to do figure out how
04:20:37to do this a little bit more,
04:20:39and obviously get the trainees involved too.
04:20:41But this has been great today in
04:20:44terms of moderating and the talks.
04:20:46So I'm going to tell us sort of
04:20:48a series of stories that feed on
04:20:50something that pointed out that
04:20:51we're really interested in how
04:20:53emerging behavior emergent behavior
04:20:55takes place in an immune system.
04:20:57It starts often with imaging
04:20:59to just ask what is,
04:21:00what do things look like an we do at
04:21:03the molecular level to look at how
04:21:05T solar separate act, for example,
04:21:07to understand how signaling gets going?
04:21:09We also look at, for example,
04:21:11how cells interact with one another in
04:21:13order to recognize and initiated response.
04:21:15And I just want to.
04:21:17Player quick movie.
04:21:18From that we're going to see in
04:21:20this T cell here that's in Green.
04:21:22It's probably not the way that you
04:21:24normally think of adi sailboat.
04:21:25If you do this kind of high resolution
04:21:27imaging called lattice light sheet,
04:21:28you see that what T cells do is
04:21:30they use these little finger like
04:21:32projections to probe the world to try
04:21:34to find peptide image, see complexes.
04:21:36And as this movie plays,
04:21:37the red is a dendritic cell,
04:21:39and you may think of a dendritic
04:21:41cells having these long dendrites.
04:21:42They're actually more like veils,
04:21:43and you can see them sweeping around
04:21:45the surface where the cell is probing
04:21:47and looking for package season.
04:21:49If you cut away that surface,
04:21:50one of the things you will appreciate
04:21:52is that the dendritic cell is
04:21:54doing as much of the work to font
04:21:56to be found to essentially wrap
04:21:58its membrane around the T cell.
04:21:59As the T cell is doing to probe the surface.
04:22:02So you see this incredible complimentarity,
04:22:04and I think this is a nice movie
04:22:06to demonstrate the idea that
04:22:08immune systems and we think about
04:22:09them in cancer or partnerships,
04:22:11and the idea that one T cell is
04:22:13the source of all cure is lovely,
04:22:15and it certainly is the basis of
04:22:17our favourites Italy for work.
04:22:19NP1 and all these things that are followed,
04:22:21but really we do need to think
04:22:23about the partnerships that take
04:22:25place and that's you know, visual.
04:22:26You can visualize that.
04:22:28And I won't actually show any movies
04:22:30in the interest of time by thinking
04:22:32about how cells work in a multi
04:22:34cellular array and you can look for
04:22:36example tumors in black here and
04:22:38these are various different mileage
04:22:39populations and you can think about
04:22:41how they arrange themselves to
04:22:43create the biology want and then
04:22:44the final emergent behavior.
04:22:46The connection is clinical outcome.
04:22:47So in cancer we have. Feet, feet down.
04:22:50Answer of whether whether a patient
04:22:51lives or dies and it's really a
04:22:53dependent on the emergent behaviors
04:22:55of how the multi cellular systems
04:22:57are setting up and those in term.
04:22:59Obviously at the lovable.
04:23:00So before I go into the data from our lab,
04:23:03I want to do something
04:23:04that's a little different.
04:23:06Maybe from what people have done
04:23:07in this in this seminar so far,
04:23:10and that's to remind everybody that
04:23:11while we've been super focused on the
04:23:13benefits of checkpoint blockade and
04:23:14all the various different ideas that
04:23:16it's brought forward about immunology,
04:23:18the rest of the immune system.
04:23:20Has been tapped in a huge huge explosion.
04:23:23I would say that many of us sort
04:23:25of choose to ignore I suppose,
04:23:28and also give you a couple examples
04:23:30because the immune system is no
04:23:32longer our textbook immune system
04:23:33where we're thinking about the
04:23:35difference between vaccination.
04:23:37Where you get a huge sort of clearing
04:23:39response versus a tolerogenic
04:23:41response where you ignore things.
04:23:43So just to give one example,
04:23:45in the last 10 years we've come to
04:23:47realize that the immune system will be
04:23:50necessary for everyone in this zoom today.
04:23:53To remember this talk,
04:23:54namely the macrophages of the brain,
04:23:56the microglia are now known
04:23:57to be pruning your synapses,
04:23:59and that pruning is essential.
04:24:01If you don't have the complement
04:24:03proteins in the microglia.
04:24:04To do that you will not remember this today.
04:24:07That is not the immune system that we're
04:24:09thinking about in general in cancer,
04:24:11but we should.
04:24:12Another example of that would be
04:24:14obviously the rise of the microbiota.
04:24:16What you realize about that is
04:24:18that the immune system is not
04:24:20ignorant of the microbiota.
04:24:21It is really carefully curated at.
04:24:23To the extent that you have T cells
04:24:25against most of your microbiota,
04:24:27and then it chooses how to deal with that,
04:24:30and that's that's a theme,
04:24:31and I think we can also see with some
04:24:33of the things that John talked about
04:24:36and we think about exhausted T cells,
04:24:38will exhaustion one can argue is a
04:24:40way that you can quarantine things.
04:24:42You can have a response to
04:24:44something that is dialed back.
04:24:45Perhaps if you want to leave latent virus
04:24:48alone and I won't go through all of these,
04:24:51but obviously our tourist chlorosis
04:24:52now as a macrophage based.
04:24:54Disease,
04:24:54it used to be just about lipids,
04:24:56so the immune system is showing up
04:24:58everywhere and I use this as a cartoon.
04:25:00This is from a perspective that underground
04:25:02Omaha and I wrote for science last year,
04:25:04where we basically point out that in the
04:25:06past we might have thought about the
04:25:08new system that going from tolerance to
04:25:10immunity or tolerance to destruction.
04:25:12Those were the things that we
04:25:14thought about dialing so we're
04:25:15in Immuno Immuno Oncology.
04:25:16We're trying to get it to destroy the tumor.
04:25:19But what's emerging from all
04:25:21the things that I just
04:25:22told you about are these roles of the immune
04:25:25system that you could call accommodation.
04:25:27Immune system is made to accommodate
04:25:29all the needs of your organs,
04:25:31that many of them are not about
04:25:34defense to microorganisms.
04:25:35So Michigan management issue metabolism
04:25:37is one that many in this room in
04:25:39this zoom will know quite well.
04:25:41Assisting tissue development.
04:25:42T cells are important, for example,
04:25:45for memory, duct formation to take place.
04:25:48And all of these things again,
04:25:50create this backdrop of the immune system,
04:25:52by which we've said, Well, you start,
04:25:54you start to need to understand how T
04:25:57cell receptor specificity paired with
04:25:58particular cell types can create this.
04:26:00What we're going in archetype,
04:26:02and it's here defining that as a
04:26:04collection of cells and linked states,
04:26:06possibly across tissue types,
04:26:07almost certainly following.
04:26:08It'll evolutionary design that do all the
04:26:10things that I show in these hexagons.
04:26:12And there are many more,
04:26:14I'm sure.
04:26:14And so the idea of this is that this
04:26:17is a higher level of abstraction
04:26:19than a single cell,
04:26:21and it's a little lower than disease,
04:26:23so it's recurrent motif in immune system.
04:26:25And if we think about the immune system,
04:26:27then also in the term of even the
04:26:29elimination of host cells and cancer,
04:26:31there are almost certainly,
04:26:32and I will show you the evidence
04:26:34that there are certainly collections
04:26:35of cell types that work together as
04:26:37archetypes that creates the kinds of
04:26:39immunity we need for antitumor responses.
04:26:41So for example, CDA T cells,
04:26:43and I'll tell you bout CC ones
04:26:45in the first story.
04:26:47Alright,
04:26:47so this doesn't mention two too much today,
04:26:50but I'm going to introduce this idea
04:26:53that is really how we came into this.
04:26:55Came back into cancer immunology
04:26:57was that we started to look in the
04:27:00sort of early teens at what cells
04:27:02in the tumor micro environment
04:27:04are really the antigen presenting
04:27:06cells on which T cell responses can
04:27:08be built and Long story short was
04:27:11that when are Miranda Bros took
04:27:13apart the tumor micro environment
04:27:15from an oven expressing tumor?
04:27:17She found it really only see 103 CC,
04:27:20one Denver Dick cells were
04:27:21able to induce T cells.
04:27:23This is XVX vivo,
04:27:24so in vitro to express nurse
04:27:2677 to express the 69 and these
04:27:28are the same cells she showed.
04:27:31They were,
04:27:31for example the predominant producers vial.
04:27:3312 they have PD one PD L1 is
04:27:37very consistent with what?
04:27:38IRA Millman has recently shown about
04:27:40these cells being again the primary
04:27:42antigen presenting cell that you want
04:27:44to D repress with checkpoint blockade,
04:27:47'cause they're the ones that present
04:27:49an antigen effectively and this is
04:27:51led to a really a cottage industry.
04:27:53In Miranda's paper she showed the
04:27:56essentially the frequency of genius
04:27:57that define these cell types,
04:27:59help you understand who's going to
04:28:01respond or actually this is even just
04:28:04who is going to live in general.
04:28:06Psycho number,
04:28:07but also who responds to check
04:28:09one blockade and a lot of other
04:28:11people have communists have done
04:28:12in create incredible work and
04:28:13sort of filling in the details and
04:28:15and carrying this story further.
04:28:17So if you understand that you
04:28:18understand then why we got interested
04:28:20in the question of what are the,
04:28:22what are the allies that are
04:28:24of the allies of the allies?
04:28:26So in other words, what is this?
04:28:28What is the archetype of the CC one?
04:28:30Why are they there in some peoples tumors?
04:28:32What makes them and have my
04:28:34we get more of them?
04:28:35So we set up.
04:28:36A system about 5 six years ago we
04:28:39called UCSF in your profile or insist
04:28:41big initiative to get tumors from every
04:28:44possible cancer indication we can.
04:28:46And to do very uniform sampling of
04:28:48them and the basic sampling strategy
04:28:50is not so different from maybe
04:28:52the pipeline that John mentioned.
04:28:54Our own one involves and has
04:28:56involved for a very long time,
04:28:58flow cytometry at high dimensionality.
04:29:00And then we sort individual applications.
04:29:02Nowadays we're doing single cell sequencing,
04:29:04but in fact we get a lot more better
04:29:06data out of deep sequencing of
04:29:08populations just because single cell
04:29:10sequencing tends to be very shallow pursell.
04:29:13So with that data set we could
04:29:15answer to the question I just posed,
04:29:17which is what is the archetype
04:29:18of the CDC one?
04:29:19What are the other cells that go with it?
04:29:22The way we did that is we took on.
04:29:24This is just some Melanoma patients
04:29:26and we each dot is a patient and we
04:29:28ask their frequency of CC ones and
04:29:30then we said what genes expressed in
04:29:32the tumor microenvironment have a
04:29:33strong correlation with the CC 1 numbers.
04:29:35And the one that was of interest
04:29:37is the cytokine.
04:29:38It's actually a League,
04:29:39and on a cell surface I'm often not
04:29:41secrete is called flip through Ligon
04:29:42and the reason it's interesting for
04:29:44those that know is it's the it's the
04:29:46dominant thing that drives CC one production.
04:29:48And the reason that's interesting
04:29:50is because this is from the tumor
04:29:52and So what you're realizing is that
04:29:54in the tumor there is a cell type
04:29:56that is your friend that is making
04:29:58flips religion that is making these
04:30:00cells and you want these cells.
04:30:01So it just so happened we started that
04:30:04just started making flips religion Reporter.
04:30:07And this is follows of sort of
04:30:09model that Rich Loxley is made
04:30:11a cottage industry as well,
04:30:13and for all kinds of cool discoveries.
04:30:15And here, just.
04:30:16Essentially,
04:30:17that's really look like in Locust Dr.
04:30:19Zeti FP and what you find is that
04:30:21Indiana the tumor micro environment.
04:30:23There's really not a lot of
04:30:25expression by the Celtics.
04:30:26You might have thought would
04:30:28expressed literally the predominant
04:30:29one though is NK cells.
04:30:30And here's the full change of the
04:30:32Reporter in NK cells versus controls.
04:30:35You also find some expressions
04:30:36in CD4 and CD8.
04:30:37Long story short,
04:30:38when you look in tumors that
04:30:40are responsive to checkpoint,
04:30:42so these individual patients are in rows,
04:30:45columns here and cell populations
04:30:47are in rows here, you essentially
04:30:49code people that are responders.
04:30:51Awana nonresponders is 0 and you ask what
04:30:54cell population predicts responsiveness.
04:30:56The number one cell population with a
04:30:59really high P value are the CDC ones.
04:31:02So you want to have those and the number
04:31:042 population that correlate's really
04:31:07strongly that those are NK cells.
04:31:10And many, many lines of evidence suggests
04:31:12now including work from Kitano racist,
04:31:15who's also that NK cells and CC
04:31:17ones existing kind of harmony
04:31:20in the tumor micro environment.
04:31:22But I bring this up also because if
04:31:25you're a sharp eye, you'll notice
04:31:27that some of the patients over here,
04:31:29even in this small data set,
04:31:31don't match this thing.
04:31:32I just told you.
04:31:33So there's 345 patients here that don't
04:31:35have that that archetype quite strongly,
04:31:37and what they have instead is they
04:31:39have a large number of CD4 cells
04:31:42and their concurrent CDC 2 here
04:31:43expressing BCA one dendritic cells.
04:31:45So it turns out that if you look at
04:31:48all responders to checkpoint blockade,
04:31:50you basically find that there are
04:31:52either of those that are high.
04:31:54PCA three in the cities he wants,
04:31:56or those that are high in the CZ
04:31:58one in the CC twos.
04:31:59So you either have a class of
04:32:01class one or Class 2
04:32:02immune response. That's the sign of my bread.
04:32:05Dough is ready just in
04:32:07case anybody is wondering.
04:32:08So this brings up this idea of Archetypes
04:32:11and I and I'm really just referencing.
04:32:13If you want to read more about these
04:32:15nature medicine paper where we describe
04:32:17what we call the type one archetype,
04:32:19that is NK cells that deliver
04:32:21flip through leg into CC ones
04:32:23in the tumor micro environment.
04:32:24Those the CDC ones then can activate
04:32:26and get to the draining lymph
04:32:28node where they activate CD8 CD S
04:32:31come back in an again CDC ones,
04:32:33and this is the This is the
04:32:35immune system that we want.
04:32:37However, there are also patients
04:32:39on which we can build an antitumor
04:32:41immune response that have a
04:32:42different kind of Dentrix.
04:32:44All this EC two that I just told you about,
04:32:47they can migrate to the lymph node activity
04:32:49fours seafort's go back to the tumor,
04:32:51but the key feature there is that
04:32:53the licensing of the CDC 2 is really
04:32:56regulated by regulatory T cells,
04:32:57and again I invite you to read that
04:32:59paper if you're interested and it brings
04:33:02up this idea that in the world now
04:33:04we have this antiviral sort of class.
04:33:06Two response C 4T cell stimulatory CDC twos.
04:33:09Macrophage phenotype,
04:33:10it's also interesting there,
04:33:11and these are CD4 based
04:33:12responders to checkpoint blockade.
04:33:13He said overall survival even without
04:33:15checkpoint blockade and you also have
04:33:17the ones that are more classically
04:33:18what you would have imagined.
04:33:19CDA T cells and then now the CDC ones in
04:33:22the NK cells as part of the partnership.
04:33:25The variant of this one that we defined.
04:33:27It's the regular regulated Class 2,
04:33:28and then there's kind of the rest of
04:33:31the world immune systems at large.
04:33:33What's more about this is that you can
04:33:35actually figure out the frequencies
04:33:37of these in different diseases,
04:33:38and it turns out that about 50%
04:33:41of Melanoma people respond to PD.
04:33:42One 2/3 of those have a class,
04:33:45one CDA response in less than 1/3 of the C4.
04:33:49In had net for example,
04:33:50though you really don't have a
04:33:52lot of class CD8T cells which you
04:33:54can build this response.
04:33:55OK,
04:33:55so this brings up this idea of what I'm
04:33:57going to call the reactive archetype,
04:33:59and this is from an image from a
04:34:01movie and in interest of time I'm
04:34:03just going to tell you what you,
04:34:05what you could see is that in blue
04:34:07there are T cells that here this
04:34:09is overexpressing tumor in black,
04:34:11blue or T cells that are swarming some
04:34:13dendritic cells that say reactive archetype.
04:34:15That's what we want but the
04:34:17dominant biology and tumor is tends
04:34:19to be CD 4T cells in CD 838,
04:34:20CDA T cells in this model.
04:34:22Swarming tumor associated macrophages.
04:34:24And if you find a lot of that,
04:34:28that's the majority of the
04:34:30mean system is doing OK.
04:34:31So why am I telling you that?
04:34:33So I'm going to tell you a very,
04:34:36very brief story.
04:34:36That's a very long story,
04:34:38and it's if I told its entire T of how
04:34:40we've been taking the data that I told
04:34:43you about it from Immuno Profiler.
04:34:45From all these different tumor types
04:34:47and doing the exact same thing I just
04:34:49mentioned you know flow cytometry sorting
04:34:51of populations and then just asking
04:34:53what is the immune landscape of tumors
04:34:55across all these different medications.
04:34:56So in one in one case you can just do
04:34:59this by the numbers and you can say
04:35:01if I do flow cytometry and I look at
04:35:03frequency of T cells or mileage cells
04:35:05or stromal cells in tumors and I have
04:35:07my different tumor types down here,
04:35:09you can say for example like happen
04:35:11HC and kidney are very T cell rich.
04:35:13Kidneys also very mild, rich for example.
04:35:17But you know,
04:35:18for example that happens is quite low,
04:35:21relatively speaking. And you know in that.
04:35:24So there's there's a sensually you
04:35:26can do a waterfall plot to say what?
04:35:29What's the immune composition of various
04:35:30are consumers and what you'll notice is.
04:35:32Even though I said that in I'm referring
04:35:34to the means being stacked here,
04:35:36there's a tremendous amount of variety,
04:35:38even within a tumor type single tumor
04:35:40type in terms of whether they are tend to
04:35:43be high or low for different operations.
04:35:45So we started to take all of this
04:35:47data and do live and clustering.
04:35:49So this is a dimension reduction version.
04:35:51And if you just take the frequency
04:35:53of T cells.
04:35:54Myeloid cells overall not
04:35:55distinguishing any of them are stroma.
04:35:57You may already find 6 populations that
04:35:59are going to be the fundamental basis
04:36:01for since essentially what I'm going
04:36:03over 12 populations of dominant immune
04:36:05systems across all kinds of different tumors,
04:36:08and this is done in here.
04:36:10In our data set,
04:36:11we can take the markers of those and
04:36:13find the same populations in TGA ascentia
04:36:16you're seeing ones that are high for T cells,
04:36:19high for myeloid cells,
04:36:20loafer stroma and various variations of this.
04:36:22Here there's two immune deserts.
04:36:24Those ultimately split into.
04:36:25And one further split will take
04:36:27place as we add some parameters,
04:36:29so that here's adding parameters.
04:36:30So we go to the point where we have,
04:36:33for example are dominant.
04:36:35This is using 6 features,
04:36:36T cells, myeloid cells,
04:36:37stromal cells, tregs, C4 CA ratio,
04:36:39and one that I'm forgetting.
04:36:41You find yourself sort of being
04:36:42able to divide the world up into
04:36:45eight archetypes of immune systems,
04:36:46and you can see the thing that I want to
04:36:50show you here is that when you go to 10,
04:36:53this is where you start to add in
04:36:55myeloid subsets and T cell subsets.
04:36:57That a lot of these basically
04:36:59keep their same track.
04:37:01So what.
04:37:01This is called a alluvial plot,
04:37:03and it basically like a River.
04:37:05It tracks where a patient was when you
04:37:08had it class by 6 features versus 10.
04:37:10So you see the majority of these
04:37:13perpetuate out to one just one
04:37:15future pen feature and you do create
04:37:17some sub sub phone.
04:37:18You always create subclasses
04:37:20when you add more features.
04:37:22But we like this 10 feature one at
04:37:24the moment and I'll tell you the
04:37:26reason why is that we get these.
04:37:27We give these things you can name immune
04:37:29rich CD 8 macrophage biased right?
04:37:31And that's based on all the
04:37:32different parameters told you about.
04:37:34But what's cool about that is that
04:37:35if you take other features like NK
04:37:38frequencies or plasma cell frequencies,
04:37:39or piece of frequencies using
04:37:41genes that represent those,
04:37:42you can see how well just using
04:37:44myeloid cells T regs and T cells
04:37:46already identify the ones.
04:37:47For example, that are NK rich,
04:37:49B cell, plasma rich,
04:37:50or just plain himself rich.
04:37:52So this is looking like the classes
04:37:54that you can find by just using a
04:37:56fairly small number of features.
04:37:58Define the dominant structure of
04:37:59immune system even in this populations
04:38:01that you didn't even count.
04:38:03Another way of looking at that
04:38:05is to just say OK.
04:38:06If I take the 12 different kinds of tumors,
04:38:09the Archetypes and I just ask
04:38:11about chemo kine expression.
04:38:12You can find that there is a
04:38:14cluster without us asking them to.
04:38:16They essentially cluster based
04:38:18on classes of chemo kines too.
04:38:19So again,
04:38:20the features of the tumor micro
04:38:22environment that come out of this
04:38:24suggests that we're really honing
04:38:25in on some fundamental chords of
04:38:27distinction and immune system,
04:38:28and this is the kind of ones
04:38:31that's coming up cool too.
04:38:32You take those same 12 archetypes.
04:38:34And you ask about gene expression
04:38:36in the tumor compartment,
04:38:37and you use some of the gene sort
04:38:40of expression modules like the ISG.
04:38:42This is by the way quite important in ours.
04:38:45Another covid studies butts in
04:38:46essence to associated genes,
04:38:48assassins,
04:38:48etc.
04:38:48And you can also see how the same populations
04:38:51tend to basically parse out based on
04:38:54what kind of jeans are in the tumor.
04:38:56And so this I think it brings us
04:38:58back to the idea that the immune
04:39:00system is just harnessing one of
04:39:03these programs that responds to
04:39:05various different gene expression.
04:39:07OK,
04:39:07so I'm going to move quickly into a
04:39:09technology story and then I'll be done.
04:39:11So.
04:39:12Many of us are in this mode right now
04:39:14of using the emerging technology of
04:39:17single cell sequencing for various
04:39:19different reasons and and one of
04:39:21those things that you get out of
04:39:23that we can just stay upfront is
04:39:25the ability to cluster populations
04:39:26of cells or patients or what have
04:39:28you based on dominant features that
04:39:30are maybe the computationally the
04:39:31biggest differences amongst them.
04:39:32And we do that and we get names of various.
04:39:35This is for myeloid cells,
04:39:36for example,
04:39:37from the six tumor and we get various
04:39:39different names of populations and
04:39:40with those we can do pretty cool
04:39:42analysis to look at how in one tumor to next.
04:39:45The lineages,
04:39:45for example going from the monocytes
04:39:47here in pink to the University
04:39:49of macrophages here in
04:39:50purple, are laid out on a
04:39:52trajectory and that's that's
04:39:53a very nice work of Trapnell,
04:39:55and many of us use these tools.
04:39:57So this is a nice way to
04:39:59find out the composition,
04:40:01but what are the partnerships?
04:40:02And this is kind of our interests,
04:40:05like how do we understand
04:40:06this in space? So
04:40:07we've been developing
04:40:08tools so that we can image tumor slice.
04:40:11Or any piece of tissue and watch
04:40:13how things behave and then ask what
04:40:16was each cell thinking. And I mean,
04:40:18thinking by the aspect of what RNA,
04:40:20where they expressing sort.
04:40:21It also could be a taxi,
04:40:23so I want to single cell sequencing
04:40:26analysis but we want to know
04:40:27going back to me to this slide,
04:40:30we want to know is this cell next to that
04:40:32sell and how are these cells spatially rate?
04:40:35So the trick that we figured out
04:40:37to do to do that is to essentially
04:40:40print barcodes onto onto cells.
04:40:41And the way we do that is we have an
04:40:44anchor oligonucleotide that can either
04:40:45be added to all your cells with an antibody,
04:40:49or for example all your cells with a link.
04:40:52And the key feature about this.
04:40:53Everybody from your DNA base
04:40:55pairing is that it has an overhang
04:40:57that is blocked by a photo cage.
04:40:58And that means that anything you might
04:41:00want to pair base pair with that.
04:41:02For example,
04:41:03if you brought in what is equivalently Inc,
04:41:05which is analogous nucleotide here,
04:41:06that has the 01 prime overhang
04:41:08to match the 01.
04:41:09It's not going to bind in less,
04:41:10we shine light on that.
04:41:12So if we shine light on this,
04:41:14these things are going to pop off
04:41:15and then we can print to that.
04:41:17And if we take advantage of the fact
04:41:19that in a microscope microscope is a
04:41:21spatial a system for spatial light.
04:41:23Direction right normally you would
04:41:24illuminate with a wide field light
04:41:26source in with blue here and here
04:41:28is the illumination source.
04:41:30If it's a mirror goes in and hits the
04:41:32sample and then you get fluorescence
04:41:34in that comes back to a camera.
04:41:36In this case we can add another light
04:41:39source that is actually itself printing
04:41:41light and therefore can I open up
04:41:43all the nucleotides for binding.
04:41:45And the way that works is you
04:41:47essentially use the same LCD projector
04:41:49chip that's in an LCD projector to
04:41:51turn on and off pixels that then turn
04:41:53them on and off in the samples space.
04:41:56And what we can do then is,
04:41:58for example,
04:41:58if we take a rectangular region
04:42:00as shown down here at the bottom,
04:42:02where every cell has the nucleotides
04:42:04added to it,
04:42:05we can selectively Uncage first here
04:42:07and add a die that's attached to a zip code.
04:42:10One of the zip code sequences here,
04:42:12that's blue, and then maybe use
04:42:13the middle on cage there at.
04:42:15Red uncaged them right now to green,
04:42:17and so we get something like this.
04:42:19This is by the way one of T cells
04:42:21in a couple and float around,
04:42:23which is why you see these
04:42:25cells that are off space,
04:42:26but in general you've now specially
04:42:28given a barcode to each of these
04:42:30regions and is the really cool
04:42:32thing here is you can do this
04:42:33in a chain reaction like the
04:42:35cooling of a PCR where if you if
04:42:37your oligonucleotide that you
04:42:38bring in is itself a caged.
04:42:40So instead of justice this piece that
04:42:42I just showed you in the previous
04:42:44the old one primes of code palie.
04:42:45You bring in something
04:42:47that has a new photo cage.
04:42:48Then you can do something like the following.
04:42:50Below you have a field of
04:42:52cells or a piece of tissue.
04:42:54You illuminate the left hand
04:42:55side only and you bring in
04:42:57unallocated get type for example,
04:42:58that's blue and so the right
04:43:00left hand side is now blue.
04:43:02Now you do 2 stripes like a bumblebee
04:43:04and you end up with four regions,
04:43:06so there's never got all of the
04:43:08nucleotide got read only got blue,
04:43:09only got blue red and you can do
04:43:12the same thing 50% and print only
04:43:14green and you end up with eight.
04:43:16So the formula for this in
04:43:17your math is 2 to the N,
04:43:19so you end up with like 256 different
04:43:21spatial regions that are highlighted
04:43:22by just eight print cycles.
04:43:24So that's a really fast way to do this,
04:43:26and this just shows you can do this
04:43:28pretty close to single cell resolution.
04:43:29Here we've got a region,
04:43:31for example is only ever printed
04:43:32in the bottom left with blue,
04:43:34so it's blue only and you can see that
04:43:36blue bar code is much higher in that region,
04:43:38and for example this one
04:43:40didn't ever get printed too,
04:43:41and all the variations in between.
04:43:43So these are basically identifiers
04:43:44of which cells came from where.
04:43:46And so the reason this There's
04:43:48a number of different ways
04:43:49that we've used this so far,
04:43:51and we're using it a lot more,
04:43:53but I'm going to just reference
04:43:55tumor volumes for a moment with
04:43:56reference to the idea of Archetypes,
04:43:58and also spatial temporal aspects of that.
04:44:00So this is a model that we've used a
04:44:02lot that is essentially putting tumors
04:44:04into mice that are cherry ova derived
04:44:06from our model that we made is PMT.
04:44:09It's a breast tumor model and
04:44:10goes into fat 14 days later.
04:44:12We can put in GOP level T1 cells that
04:44:15are against the over as we know.
04:44:17Four days later from that,
04:44:19we harvest and we the image
04:44:20of the tumor looks like this.
04:44:22So the border has lots and
04:44:23lots of details on it.
04:44:25The red tumor on the center is
04:44:26relatively sparse for Diesels,
04:44:27but still has T cells in there.
04:44:30And if we now barcode this,
04:44:31so the region that's highlighted
04:44:33with blue on the outside,
04:44:34we barcode that with one set of
04:44:36bar codes for a different set
04:44:37of bar codes on the inside.
04:44:39And we just do Disney and you sort
04:44:41of see all the different populations
04:44:42that you're used to and you start
04:44:44to see the beginnings of red,
04:44:46blue distinctions.
04:44:47You can really start to see those
04:44:49when you focus on the cell population.
04:44:51So here we pulled out the object
04:44:53or the collection of cells that
04:44:55represent T cells from this
04:44:57environment and you can
04:44:58see that the red cells,
04:44:59the ones on the inside or segregating
04:45:01pretty highly from the blue cells,
04:45:03and the difference between red cells
04:45:04and blue cells is exhaustion and
04:45:06terminal differentiation scores.
04:45:07And as best shown here where you
04:45:10basically compare blue cells to red
04:45:11cells in a volcano plot and some of the
04:45:14genes we just heard about from John C.
04:45:16F7 for example MIB and Slime family 6 R.
04:45:19Breast really in the outside region
04:45:21as cells for getting into the tumor.
04:45:23And as you go down differentiation
04:45:25pathways you see for example ID 2:05.
04:45:27If you other PD one being
04:45:29expressed highly as you go inward.
04:45:31So there's essentially starting
04:45:33to look like a gradient of
04:45:35exhaustion that permeates the tumor.
04:45:37And you can now because you have a
04:45:39whole data set you can rather gate
04:45:41on the monocytes and macrophages.
04:45:43And again you see now if you done
04:45:45that your object reduces to the
04:45:47monocytes and macrophages where
04:45:48now you can see that the red,
04:45:50red right hand side of this
04:45:52computationally generated Disney
04:45:53or Hue map represents all the cells
04:45:55came from the route from the center
04:45:57that left hand side has a lot more
04:45:59blue meaning from the margin.
04:46:01If you ask what this cell populations are,
04:46:03you see that C1,
04:46:04QA real high marker of Tams is
04:46:06on the right hand side.
04:46:08And Licensee, for example,
04:46:09big marker of early monocytes in
04:46:11the left hand side it allows you
04:46:14basically to say as you're going
04:46:16the trajectory from monocytes to
04:46:18tumor associated macrophages.
04:46:19This is how the spatial dimension goes.
04:46:21You go from being very much on the
04:46:24margin to being very much in the inside,
04:46:27so we're seeing kind of like a
04:46:30a coordinate differentiation of
04:46:31monocytes to Tams as we're seeing
04:46:33coordinate regulation of early entry.
04:46:35I think stem cells into the exhausted phase.
04:46:38The cool thing about this,
04:46:40just in just one hour.
04:46:44You can take multiple regions here.
04:46:46We've made four different concentric
04:46:47rings in a tumor micro environment,
04:46:49and you can find your favorite gene
04:46:51KLF two is as one of our favourites
04:46:53because it's almost always high on
04:46:55the inside of organs and low on the
04:46:57outside of organs and you can find
04:46:58other jeans that look like that and
04:47:00so that allows you to essentially
04:47:02to say what are the coordinate
04:47:04expression patterns in this cell type.
04:47:06But you can also now say what I
04:47:07wanna look for patterns that look
04:47:09like this in every other cell type.
04:47:11You can obviously look for the
04:47:13ones that are opposite.
04:47:14You can do that for other jeans like here,
04:47:17CR7 jeans.
04:47:17Again they tend to fall off as
04:47:19you go from inside outside,
04:47:20so this is really a discovery tool
04:47:22that starts to add to allow you to ask
04:47:25questions about how gene expression is
04:47:27coordinate Lee regulated over space.
04:47:29And I think the cool things that we're
04:47:31going to see about this is to ask
04:47:33what's special about these various
04:47:34different regions where different
04:47:35archetypes or different biology is pleasant.
04:47:37And in fact there's also to discover
04:47:39the Blacks, the famous black space.
04:47:40Whenever we image,
04:47:41we've chosen what we want to label,
04:47:43we don't label it, but we put Barcodes on it.
04:47:45We can find out what was there.
04:47:48So I need to think of Brazilian people
04:47:51for this, and this is a short list.
04:47:54John started the work that led to the CDC,
04:47:57one work of Miranda back in 2016,
04:48:002014 McHale showed the CDC two
04:48:02archetype audriana Mahal was involved
04:48:04in a lot of the archetype work.
04:48:06In general.
04:48:07Kevin did the NK cell work is now
04:48:10at the hutch candid zips.
04:48:12Lexie did a lot of the dominant
04:48:15archetype work and it will see
04:48:17come out fairly soon and.
04:48:19Kyles are imaging at Expert in
04:48:22Vincent really coordinates all
04:48:23of the profiler and UX work,
04:48:25so thank you all.
04:48:26Thank you everybody for coming and
04:48:28general sticking with us to the
04:48:30end here and again,
04:48:31the moderators for really
04:48:32a great great news today.
04:48:35Thanks Max, I was really excellent talk.
04:48:38So we have one question.
04:48:40In the in the Q&A right now and
04:48:42I'll ask that and then I may
04:48:44have some questions as well.
04:48:46So the question comes from Adam Rubin,
04:48:48the question is.
04:48:51It is the CD8 CD.
04:48:52One interaction in the TI,
04:48:54me also engine specific are those
04:48:57CDC ones activated and migrating.
04:49:00Yeah, so it is antigen specific so
04:49:02if we take for example P14 cell and
04:49:06mix them together with the with
04:49:08the CDC ones from ANOVA tumor,
04:49:10there's there's no priming.
04:49:12There's no, there's no formation of
04:49:14a synapse that's above background,
04:49:16so I think that's there are key
04:49:19McCain gradients that bring in cells.
04:49:21This definitely CR5 Comic Con gradient
04:49:23that could bring cells towards that.
04:49:26I don't think the interactions
04:49:28that would be antigen nonspecific.
04:49:30And I think that Miriam basically
04:49:32has really opened up something that
04:49:34we showed and didn't spend a lot
04:49:36of detail to that the CDC ones,
04:49:38about 20% of them end up expressing CR7,
04:49:41and that's a maturation signal for them
04:49:43to go to the lymph node we focused on,
04:49:46that she's very much focused on
04:49:48the idea that when they do that,
04:49:50there are sort of being matured
04:49:52and she's focused on the fact that
04:49:55that also corresponds to them.
04:49:56I would say fairly modestly,
04:49:58but noticeably upregulating PD L1.
04:50:00And that the importance of that
04:50:02obviously comes to play with iris
04:50:04demonstration that the main cell
04:50:05that matters is the CDC one for PD.
04:50:07L1 and I think the reason for
04:50:09that is that the other sales just
04:50:11aren't expressing a lot of energy.
04:50:14So if you D repress PDL one on a macrophage,
04:50:17it wasn't really doing a lot for you
04:50:19anyway in terms of management presentation.
04:50:21So it's DL1 is most important on the C one,
04:50:24so that's kind of a long answer.
04:50:26The question there are definitely
04:50:28subpopulations of these we see the
04:50:30CR7 high population is likely.
04:50:31The ones that are just about
04:50:32to transit to the lymph node.
04:50:33But they may do something before they go to.
04:50:37Great so John has a question on the panel. A
04:50:42Max that was really cool.
04:50:44Great stuff, the question relates to these
04:50:47recent papers on interferon autoantibodies,
04:50:49interference genetics snips and things is,
04:50:51I think, really interesting and sets the
04:50:53stage for the fact that people might
04:50:56have auto antibodies against cytokines,
04:50:58including interferons.
04:50:58I think creates a whole new
04:51:00layer of possibilities,
04:51:02so do you see any things in the archetypes
04:51:05that it looked like you have segregation
04:51:08based on my SGS to some extent,
04:51:10but how much of that do you
04:51:12think relates to production or?
04:51:15Impacts the Archetypes. Yeah,
04:51:16well just briefly going to the
04:51:18Casanova paper we just submitted
04:51:20ours our longstanding kovid paper,
04:51:22and it turns out that I think, well,
04:51:26we show that every Sevier patient
04:51:28is generating autoantibodies that
04:51:30are against the IST phenotype.
04:51:32Not just the type one interferon itself,
04:51:35but many of them are against some sort of
04:51:37cell surface epitopes that are on my SGS,
04:51:40so there's a bigger story there
04:51:42that's kovid related that's probably
04:51:44beyond the scope of today's meeting,
04:51:46but the one of the things that
04:51:48we're looking at is in the ISG.
04:51:50The tumors that show the ISG signature.
04:51:52Question is,
04:51:53is the source of that an ancient
04:51:56underlying viral infection and
04:51:57you can do that because when you
04:51:59sequence you get you can look and
04:52:02you align against all the known
04:52:03viruses you can find whether tumors
04:52:05have predominantly large amounts or
04:52:07particular subtypes of viruses in them.
04:52:09And you probably know that there was
04:52:12some nice papers this last year that
04:52:14show tumors with bacteria in them,
04:52:16and you know that that such that
04:52:18there's a much higher rate of ours
04:52:21are being residual E infected.
04:52:22But I think certainly was taught
04:52:24to us in textbooks we were taught
04:52:26about sterilizing immunity that
04:52:28when you finished an infection,
04:52:30you were back to being like a baby.
04:52:32You know bugs in you at all.
04:52:34And it's it's just not true, right?
04:52:37I mean,
04:52:37the fact that we can't find
04:52:39viruses in the blood,
04:52:41and you know,
04:52:42by PCR doesn't mean that there's not
04:52:44some transfer hanging around all over.
04:52:46So I think the SG might be interesting.
04:52:48Interesting one that it might be driven by
04:52:51residual tumor viruses that are in there.
04:52:53I don't know.
04:52:54Thanks for all your questions,
04:52:56'cause there's a lot of threads in there.
04:52:58The question of autoantibodies.
04:52:59And you did see it.
04:53:00Hopefully that one of the archetypes
04:53:02has a lot of plasma cells in it and
04:53:04you you showed that code with patients
04:53:06have a huge number of plasma glass,
04:53:08so that would be an interesting one
04:53:09to see whether those represent sort of
04:53:11continually production of autoantibodies.
04:53:12But we haven't done that.
04:53:14But we have the data set and
04:53:16we've had to share it, right?
04:53:17So Pam. I
04:53:19hate Max those great looking at
04:53:20your arch types based on the
04:53:22different tumors that you looked
04:53:23at where those primary tumors.
04:53:25Or did you also have access to
04:53:27metastatic lesions because you know
04:53:29the meta static micro environment
04:53:30may make a difference and I'm
04:53:32just wondering if the archtypes
04:53:33are really determined more by the
04:53:35primary colors themselves or to
04:53:37meta static micro environment.
04:53:39So the so the data I showed
04:53:40you is largely primaries.
04:53:42That does include some mid meta statics,
04:53:44but to your point there
04:53:45are things that seem to be.
04:53:47There's something when we take and
04:53:49you've done this answer to that.
04:53:50You take primaries and Mets from some people.
04:53:53They can look remarkably similar,
04:53:54but it's not a rule.
04:53:56It's clear that you can have some
04:53:58primaries in some Mets where the Met
04:54:00really diverges from the primary,
04:54:01and So what we did here is we basically
04:54:04treated the world as a garden's is
04:54:06like all garden variety tumors,
04:54:07whether their primaries or mats.
04:54:09Them into this analysis, right?
04:54:10So there's a separate question.
04:54:12One can answer, you know,
04:54:13like do primaries always look like
04:54:15their Mets or Mets liquor primary?
04:54:17And I think the answer is not always.
04:54:20I think it's probably dominantly true,
04:54:21but tissue tissue factors have a
04:54:23lot to do with things to write.
04:54:25If you go into the liver, you guys know.
04:54:28In Melanoma there's a big difference in
04:54:30going in along that what you have to
04:54:32do and what you have to immune system.
04:54:35You guys have in the first placement, OK.
04:54:38Yeah, I think I wanted to add to.
04:54:40That is whenever you have a tumor
04:54:43in any of those locations,
04:54:44it's mainly a failed immune response, right?
04:54:46So if there's still a tumor there an I
04:54:49think that's also sort of the beauty
04:54:51of some stuff than was talking about
04:54:53before about window of opportunity
04:54:55trials for you're actually giving
04:54:57therapies where you have responses
04:54:58that you can measure elsewhere,
04:55:00which we really have a hard time
04:55:02capturing in humans because those
04:55:03biopsies are not medically indicated
04:55:05and you have to pay for them,
04:55:07which I think MD Anderson's done
04:55:09a good job of trying to do.
04:55:11And we're trying to do this
04:55:13elsewhere as well,
04:55:14so I think keeping that in mind that if
04:55:17it's clinically indicated to remove a mass,
04:55:19it's usually not because the patient
04:55:21was doing so great that that you're
04:55:24trying to save the patient as result
04:55:26of that an it's 1 version of a
04:55:28failed response in a different issue,
04:55:30and I think that it's just useful
04:55:33to always annotate whatever the
04:55:34samples that you're getting,
04:55:36what they really represent is
04:55:37this in the middle of a response.
04:55:39Is this in the middle of a failed response?
04:55:42Is this on?
04:55:43Therapies does not run therapy,
04:55:45and all of those things matter.
04:55:47So I
04:55:48had a question about the archetypes
04:55:50and specifically this concept of
04:55:51whether or not you're defining there
04:55:53is different modules of function
04:55:55that the immune cells bring.
04:55:56So like the NK cell is not that
04:55:59different from a CD8T cell,
04:56:00and they kind of could bring
04:56:02the same thing is that is that
04:56:04really the concept there that the
04:56:06different cells are bringing in
04:56:08different modules and they're kind
04:56:10of interchangeable in a way?
04:56:13Well, it's a good question weather
04:56:14weather like a particular cell type
04:56:16can be substituted for another one
04:56:18and that then of course that comes
04:56:20down to how you define cell type.
04:56:21At some point lymphocytes
04:56:23would include NK cells at CDs,
04:56:24but then you can dive down and say
04:56:26only CD S in NK cells have cytolytic
04:56:28activity and so is cytolytic activity.
04:56:30The component of the archetype or is in
04:56:33lymphocytes or is it in case specific?
04:56:35You know the way that we've
04:56:37organized these has been based
04:56:39on very high level descriptors,
04:56:41and if it turns out you take enough of
04:56:43those and you end up with what look
04:56:46like pretty good class distinctions.
04:56:49That you know,
04:56:49predict other cell populations,
04:56:51whether they're interchangeable enough.
04:56:52I don't know that we have the
04:56:54statistical power of ask that right now.
04:56:56Like Are there like some NK cells
04:56:58that have caused one tumor in our 350.
04:57:00The tumors to get miss assigned and
04:57:02it really belongs in another one,
04:57:04because what the NK cell is doing
04:57:06in one tumor is different than
04:57:08what it was doing another tumor.
04:57:10But you know,
04:57:11I think those those things will
04:57:13have to be determined,
04:57:15and all we're trying to do here
04:57:17is to start a class distinction.
04:57:19Describer classes.
04:57:20I'm sure that if you started
04:57:22a new anchored on other ones,
04:57:24you might come to the same
04:57:26population somewhat differently,
04:57:27and sometimes you'll end up
04:57:28with an additional branch.
04:57:30The question about class distinctions,
04:57:31when does it matter?
04:57:33Like windows matter what the tumor hasn't?
04:57:35It was. Responsiveness is obviously key.
04:57:37One outcome.
04:57:38And we need to start it be coming
04:57:40at that from both sides as in OK?
04:57:42Who responds and then what are the
04:57:43sort of possible class structure out
04:57:45there and they start to match up.
04:57:46It will be when the wires hitting each
04:57:48other that we know that we got it right.
04:57:51Can I ask it? Maybe this is
04:57:53still to be determined as well,
04:57:55but the cells within a class?
04:57:57Are they all being targeted through
04:57:59targeting one? Or you know?
04:58:00So in terms of like therapy, right?
04:58:02So CTA for Axon cell a that then.
04:58:05Kicks off this whole class in that, so
04:58:08it's interesting question because I think
04:58:10the way I presented the way we
04:58:12think about this is that a tumor
04:58:14wouldn't be grown if it didn't
04:58:15have a dominant immune system
04:58:17that was supportive of the tumor.
04:58:19And yet, what we're trying to do when
04:58:21we're trying to get it to go is we're
04:58:24trying to harness the subdominant archetype.
04:58:25The population of cells
04:58:27that could work with us but
04:58:28aren't right, so that seems
04:58:30to be what
04:58:31we're doing with checkpoint blockade,
04:58:32is harnessing these populations
04:58:33of cells that could be doing good.
04:58:36But basically, the rest of the
04:58:37system is doing something bad.
04:58:39And so, So what are describing that dominant
04:58:41system is just saying what is the bulk mass
04:58:44action of the immune system look like?
04:58:46But that's really a different
04:58:48question than like, what else?
04:58:49Is there an Alpha often point to
04:58:51people this thing that everybody knows,
04:58:53and it isn't textbooks that if you take
04:58:56TH one and TH two is a class right?
04:58:58You can have gamma production or
04:59:00oil for production.
04:59:01If you look in any real lesion,
04:59:03it's never one or the other.
04:59:06You know, even when you're getting
04:59:07clearance and you've got lots of TH one,
04:59:09you will find TH two in there.
04:59:11And that's presumably because biology has at
04:59:13its heart the seeds of its own destruction.
04:59:15So we can.
04:59:15Basically then there might be a
04:59:17time when you want to be TH two,
04:59:19and so you keep the other guy around.
04:59:21And I think that's probably
04:59:22what we're trying to harness.
04:59:23And tumors were trying to harness
04:59:25the subdominant immune system.
04:59:26The tumor is managed to get mostly
04:59:27the mean system on its side,
04:59:29and we're trying to do is to
04:59:30grow that little tiny flower.
04:59:34Great, well this is an excellent talk.
04:59:36An excellent session with both John
04:59:37in and Max for speaking with that.
04:59:40I will conclude this session
04:59:41turned over to Marcus
04:59:42for concluding remarks.
04:59:44Thanks, I mean, what today really,
04:59:46really great talks all around and I
04:59:48really want to thank all of the speakers.
04:59:51So much for inspiring. We've had
04:59:53great attendance throughout the day,
04:59:55so it's made a huge impact primarily at Yale.
04:59:58But there are people from around the
05:00:00country who have been tuning in.
05:00:02Who had, you know,
05:00:03it's a good thing about finding
05:00:05things on the Internet. They
05:00:07can they can,
05:00:08you know, not
05:00:09zoom bomb, but sort of zoom
05:00:11bomb in a
05:00:12completely supported way.
05:00:13And I think that's great because.
05:00:15The goal of this is A to
05:00:17see where things are at,
05:00:19which is really spectacular.
05:00:20You think about 10 years ago?
05:00:22Well, Jim, you know,
05:00:2410 years ago was probably here,
05:00:26but you know, for and many
05:00:28others you know Max as well.
05:00:30Other folks were here as well,
05:00:32but the point is,
05:00:33is that where we were from
05:00:35a clinical point of view,
05:00:37an how household IO has
05:00:39become for everyone
05:00:40else. But I think
05:00:41if you really think
05:00:43about all the things you've heard today.
05:00:45There are so many unanswered
05:00:46questions that are still need to be
05:00:49addressed that we really don't know.
05:00:51These are really, really
05:00:52complicated questions. Cancers,
05:00:53complicated immunology is complicated.
05:00:54All of the subsets, things
05:00:56that we've been talking about.
05:00:57There's so much for all of the trainees to
05:01:00do out there to increase the
05:01:02number of patients that survive
05:01:03with immune based therapies,
05:01:05and I think that's at the end of the day.
05:01:08It's fun to understand all this,
05:01:10but what we're trying to do,
05:01:11I think in Part 2 is to save
05:01:14some lives along those lines.
05:01:16So I want to thank all of
05:01:18the speakers I want to thank
05:01:20especially alisa Matthews who
05:01:21was for the speakers. You obviously
05:01:23have met her in organizing,
05:01:24but the primary organizer from
05:01:26the El senor from you on Koleji.
05:01:28I want to thank the yell center,
05:01:30uh Cancer Center and Charlie Fuchs
05:01:32and let me know biology Department.
05:01:34With David Schatz also for supporting things.
05:01:36And it's a Friday afternoon.
05:01:38I hope all of you have a wonderful weekend.
05:01:41Thanks so much for participating.
05:01:42Thanks all of the folks
05:01:44again for being here all day.
05:01:46Thanks Max, I hope your air is better in
05:01:48San Francisco now than
05:01:49it was last time we
05:01:51talked. Seems like it is
05:01:52anyway. All thanks so much. Take
05:01:54care and have a great weekend. I buy thanks.