Pathology Grand Rounds May 12, 2025

Transcript

• 00:00So doctor Frank McCormick is
• 00:02the chair of tumor biology
• 00:04and cancer research at UCSF.
• 00:06He's had a long and
• 00:07storied career in biotech before
• 00:09coming back to academia, perhaps
• 00:11most notably,
• 00:13founder and CSO at Onyx
• 00:15Pharmaceuticals,
• 00:16which, develops sorafenib, the RAF
• 00:18inhibitor, and also palbasiclib,
• 00:20the CDK four six inhibitor.
• 00:24On top of that, he's
• 00:25been leading the RAS initiative
• 00:27at the NCI for over
• 00:28a decade now.
• 00:30And, I think, you know,
• 00:31most notably not most notably,
• 00:33but outside of his contributions
• 00:35to research,
• 00:37in my short time with
• 00:38him, I've learned that he's
• 00:39a a staunch proponent for,
• 00:41cancer research and for funding
• 00:43of scientists, which is more
• 00:45important now perhaps than ever.
• 00:46So please,
• 00:47join me in welcoming doctor
• 00:49Frank McCormack.
• 00:50Okay. That's awesome.
• 00:56That's a great introduction. Really
• 00:57brief. I like that. Thank
• 00:58you.
• 00:59And thanks a lot for
• 01:00inviting me. It's a real
• 01:01honor to be invited to
• 01:02come and talk
• 01:03back here again. But I
• 01:04came back here in, twenty
• 01:05nineteen was the last time
• 01:06I talked here. This is
• 01:07before the pandemic, which seems
• 01:09like a different world.
• 01:11But we survived the pandemic.
• 01:12We all survived Trump, I
• 01:13I hope. That's all.
• 01:17Anyway.
• 01:18Okay. So today, I'm gonna
• 01:19talk about,
• 01:20the the RAS pathway and
• 01:22specifically
• 01:23about a couple of drugs
• 01:24we developed that target KRAS
• 01:25cancers.
• 01:26But before I get into
• 01:27the drug, development and discovery
• 01:29part, I wanna give you
• 01:30a bit of background on
• 01:31the MAP kinase pathway in
• 01:33a few contexts, which are
• 01:34slightly outside of cancer and
• 01:35areas you may not be
• 01:36as familiar with because this
• 01:37pathway is involved in, in
• 01:39several different indications,
• 01:41beyond its well known role
• 01:43in in cancer.
• 01:45So that's, San Francisco as
• 01:46you can see, and,
• 01:48these are my disclosures. And
• 01:50to distract you from reading
• 01:51them, the picture below is
• 01:52the is the building,
• 01:54in Frederick, Maryland, where I've
• 01:56been spending a week a
• 01:57month for the last ten
• 01:58years, working on a project
• 02:00that Harold Varmus,
• 02:02initiated, really, to target,
• 02:04pancreas cancer primarily, another KRAS
• 02:06driven, cancers. Because at that
• 02:07time in twenty thirteen when
• 02:09we started, we had nothing
• 02:10for those patients at all.
• 02:12So he basically assigned
• 02:14fifty people from an from
• 02:16a a contract with the
• 02:17NCI
• 02:18and told them, okay. You
• 02:19guys are gonna work on
• 02:20RAS, and, he asked me
• 02:22to be the project leader.
• 02:23So I've been going there
• 02:24every every,
• 02:25every month or so for
• 02:26the last, ten years.
• 02:28And I'm very proud to
• 02:29say that we have actually
• 02:30developed three drugs from that
• 02:32initiative, which are now in
• 02:34clinical trials, and I'll tell
• 02:35you about two of them
• 02:36in the second part of
• 02:37my talk. So just as
• 02:39a preview, one is a
• 02:40a KRAS G12 c inhibitor,
• 02:42which is the on stage
• 02:43of of KRAS, which hasn't
• 02:45hasn't been done before. And
• 02:46the other one is a
• 02:47drug that prevents
• 02:48RAS proteins activating p I
• 02:50three kinase, and we think
• 02:51will be useful in a
• 02:52whole range of indications. The
• 02:53third one, which I won't
• 02:54talk about, is a pan
• 02:56KRAS drug, which is, also
• 02:57now in the clinic.
• 03:01K. So,
• 03:10Oh, I'm I'm sorry.
• 03:11Using the wrong advancing system.
• 03:13Okay.
• 03:16Okay.
• 03:17Now I'm set.
• 03:18So,
• 03:19again, introduction to the, the
• 03:21RAS pathway in normal cells.
• 03:23And I wanna stress that
• 03:24the only pathway that RAS
• 03:25proteins activates in normal cells
• 03:27is the RAS MAP kinase
• 03:28pathway as far as I
• 03:29know. No evidence for any
• 03:31other RAS effectors being important
• 03:33in normal cells. If anybody
• 03:34has information or,
• 03:36wants to contradict that, I'd
• 03:37be delighted to hear about
• 03:38it, but, I'm pretty sure
• 03:39that's the only pathway the
• 03:41RAS activates.
• 03:42So,
• 03:43you're all being familiar with
• 03:44the general concept here that
• 03:46RTKs activate,
• 03:47Sos one and other related
• 03:49proteins to put RAS from
• 03:51the off state,
• 03:52into the on state where
• 03:53RAS engages the map kinase
• 03:55pathway,
• 03:56and then gets turned off
• 03:57by gaps such as these
• 03:58three here, which I'll talk
• 04:00a bit more about in
• 04:01just in just a moment.
• 04:02But although its pathway is
• 04:04well trodden, it's in the
• 04:05textbooks, there's still lots of
• 04:07elements to this pathway we
• 04:07don't really understand. In fact,
• 04:09most of it we don't
• 04:10understand really in in-depth.
• 04:13For example, a critical part
• 04:14of this pathway is the
• 04:15tyrosine phosphatase SHIP two,
• 04:17which is thought to dephosphorylate
• 04:19a critical substrate and enable
• 04:21all this to happen.
• 04:22We don't think that's true.
• 04:23We don't think SHIP two
• 04:25phosphatase is actually,
• 04:26important as an activity, but
• 04:28the protein definitely is important,
• 04:30and we don't really understand,
• 04:31how it works.
• 04:32We don't really understand,
• 04:34which gaps are in play
• 04:35in different cell types. We
• 04:37don't really understand how this
• 04:39complex is regulated to regulate
• 04:40BAP kinase, and so on.
• 04:42And if you actually understand
• 04:43or or look at the
• 04:44details by which RAS activates
• 04:46RAF, for example, it's really
• 04:47complicated. There's multiple steps.
• 04:49This is a massive oversimplification.
• 04:52But still, the the hierarchy
• 04:53is definitely correct, but the
• 04:54details are yet to be
• 04:56filled in.
• 04:57So just to get you
• 04:58all on the same page,
• 04:59if you take a cell
• 04:59line, add EGF, for example,
• 05:01you activate RAS GDP
• 05:04levels within about about one
• 05:05minute. So this pathway gets
• 05:06kicked on, and then it
• 05:07gets turned off again in
• 05:09about the same time frame,
• 05:10by,
• 05:11in in the gaps such
• 05:13as neurofibromine and RAS, a
• 05:15two, turning off again.
• 05:17And the spike of RAS
• 05:18activity
• 05:18translates to a spike of
• 05:20phospho work,
• 05:21as a as a signal
• 05:22goes down the pathway, and
• 05:23the end result of this,
• 05:25pathway,
• 05:26can affect all kinds of
• 05:27different aspects of cell growth
• 05:28and metabolism
• 05:29differentiation
• 05:30and even cognition, as I'll
• 05:32mention later. But the output
• 05:33of the pathway is very
• 05:34much context dependent even though
• 05:35the pathway is, you know,
• 05:37highly conserved between different tissues.
• 05:39Using a phospho ERK sensor,
• 05:41you can see that when
• 05:41you add growth factors to
• 05:42cells, you get a spike
• 05:43of signaling down the pathway.
• 05:44This is phospho ERK coming
• 05:46on and off again in
• 05:47a very nice controlled spike.
• 05:49If you add more EGF,
• 05:50you get more spikes. So
• 05:51each of these spikes is
• 05:52the same size and duration,
• 05:54and is highly regulated. It's
• 05:56just you get more,
• 05:58spikes if you have more
• 05:59more growth factors. So in
• 06:00normal cells, this pathway is
• 06:02extremely precisely regulated, so on
• 06:04and off very quickly. And
• 06:05this,
• 06:06pattern will be different with
• 06:07different growth factors. The duration
• 06:09will be different and so
• 06:10on, but,
• 06:11a whole number of proteins
• 06:13jump in to regulate this
• 06:14pathway with great precision.
• 06:17We've studied this pathway in
• 06:19cancer, obviously, but also in,
• 06:20in Noonan syndrome,
• 06:22which is, a disease in
• 06:24which individuals,
• 06:25inherit a germline mutation in
• 06:27one of the genes in
• 06:28this pathway.
• 06:29And we like this studying
• 06:30this disease because these are
• 06:32slightly activated versions of a
• 06:33normal counterpart but in a
• 06:35normal cell background. So unlike
• 06:36cancer where everything has gone
• 06:38to hell, these are wild
• 06:39type cells with just one
• 06:41slight gain of function mutation.
• 06:43And these gain of functions
• 06:44can happen in pretty much
• 06:45every part of this pathway.
• 06:47And they're all weak alleles
• 06:48that don't cause cancer, and,
• 06:51individuals can survive these activating
• 06:53events, but they end up
• 06:54with a very similar kind
• 06:55of phenotype.
• 06:56These individuals have distinct facial
• 06:58features, as you can see
• 06:59here. They have some cardiac
• 07:01problems, and they have short
• 07:02stature.
• 07:03This is a very common
• 07:04disease. Actually, one in a
• 07:06thousand people or more have
• 07:07Noonan syndrome. It's mild, not
• 07:09life threatening, but it does
• 07:10have phenotypes and consequences.
• 07:13So based on this, you
• 07:14say this pathway is involved
• 07:15the RAS pathway is involved
• 07:16in differentiation and development because
• 07:18these are the effects you
• 07:19see when you hyperactivate
• 07:21this pathway.
• 07:23And the frequency of this
• 07:24disease is much more than
• 07:25you'd expect just from the
• 07:26mutation rate, in in, in
• 07:28any particular gene in the
• 07:30genome.
• 07:30And the the, the reason
• 07:32for that, or a potential
• 07:34reason was revealed recently by
• 07:35a paper that
• 07:37is, I think, the really
• 07:38game changing paper to me,
• 07:40and it's slightly gross. So
• 07:41hang with me.
• 07:43So this is a study
• 07:44in which,
• 07:45Anne Gereilly in Oxford,
• 07:47did a study in which
• 07:48she,
• 07:49looked at
• 07:50clonal expansion of cells in
• 07:52aging male testes.
• 07:54So what she did was
• 07:55took slice of testes from,
• 07:57from elder men. This is
• 07:58a seventy,
• 08:00eighty year old, ninety year
• 08:01old guy, and then did
• 08:02deep sequencing on little regions
• 08:03of the testis,
• 08:05sequencing for genes in the
• 08:06MAP kinase pathway, so, like,
• 08:08sixty different genes. So each
• 08:10of these circles here represents
• 08:11clonal expansion of a cell
• 08:13in the testis. Okay?
• 08:15And the colors are coded
• 08:16by which of the genes
• 08:18are named on this list
• 08:19here. So in this particular,
• 08:22slice to this test, as
• 08:23you can see, there are
• 08:24all kinds of focal amplifications
• 08:26of, of these cells,
• 08:29and the genes involved
• 08:30are genes in might may
• 08:32be a spec, FGF receptor,
• 08:33FGFR2,
• 08:34FGFR3,
• 08:36and most of the genes
• 08:37involved in syndrome, including,
• 08:39PTKN11,
• 08:40SHIP2, and and RAS.
• 08:43So the really interesting thing
• 08:44about this is that just
• 08:45like as in any other
• 08:47tissue, as you age, you
• 08:48get point mutations in RAS
• 08:49which and and RAS pathway,
• 08:51which cause clonal expansion of
• 08:52your cells in your body.
• 08:53These are the cells that
• 08:54make sperm.
• 08:56So these mutations, which are
• 08:57selected for for amplification in
• 08:59the testes, are transmitted to
• 09:00the offspring.
• 09:02And there, they they cause
• 09:03the syndrome, but the selection
• 09:04was actually in the testis
• 09:06of the, of the aging
• 09:07individual. So that's why they're
• 09:08much more frequent than you'd
• 09:09expect, and these are then
• 09:10disseminated throughout the population. So
• 09:12here we have human evolution,
• 09:14at work.
• 09:16So, again, Noonan syndrome is
• 09:18probably caused by these
• 09:20mutations in aging
• 09:21testes.
• 09:22Also, FGFR3
• 09:24mutations
• 09:24cause dwarfism, achondroplasia.
• 09:27And just as a sidebar,
• 09:29the company a company I
• 09:29cofounding for BridgeBio
• 09:31has an FTFR3 inhibitor,
• 09:33in clinical trials,
• 09:35for people with dwarfism, and
• 09:37amazingly, the drug accelerates growth
• 09:39of these individuals to normal
• 09:40growth rates without any toxicity
• 09:42because it's used at a
• 09:43very low dose. That'll be
• 09:44coming out soon, but that's
• 09:46just sidebar.
• 09:48Anyway, so,
• 09:51that that's the Nuna syndrome,
• 09:52and we've we've looked at
• 09:53all the genes that involved
• 09:54in the syndrome and have
• 09:55now have a mechanism for
• 09:56how each of these are
• 09:57activated except for SHIP two,
• 09:59which we're still working on.
• 10:02Now another another disease which
• 10:03is much more clinically serious
• 10:04is neurofibromasosis
• 10:06type one. So in this
• 10:07disease, individuals either inherit or
• 10:10acquire a,
• 10:11mutation in the neurofibromine
• 10:13gene.
• 10:14That's a gap that turns
• 10:15RAS
• 10:16off. So through loss of,
• 10:18neurofibromine,
• 10:19RAS protein starts to accumulate
• 10:20in the GTP bound state
• 10:22just as they as they
• 10:23do in cancer, but not
• 10:24as severe. Okay? So I've
• 10:25given a partial break here.
• 10:27So this is also pretty
• 10:28common. One in three thousand
• 10:30five hundred people. That's a
• 10:31hundred thousand people in the
• 10:32US have NF one disease
• 10:34of varying severity.
• 10:35The heterozygous state,
• 10:37the the phenotype is cognitive
• 10:39defects,
• 10:40vascular disease and osteoporosis.
• 10:43But,
• 10:44these, by far, the most
• 10:46problematic for the kids and
• 10:47the families are the cognitive
• 10:48defects because these kids have,
• 10:50autistic like phenotypes and that
• 10:52they have a disruptive behavior
• 10:54and are very difficult to
• 10:55deal with in a family
• 10:56situation. So this is a
• 10:57big problem.
• 10:59But worst yet for the
• 11:00kids, they lose a second
• 11:02allele,
• 11:03randomly.
• 11:04They get clonal growth of
• 11:05benign tumors, which can be
• 11:07either on the skin such
• 11:08as these dermal neurofibromas
• 11:10or plexiform neurofibromas
• 11:12that grow around nerves.
• 11:14Now these are benign, but
• 11:15as they grow over time,
• 11:16they can be extremely painful,
• 11:17and they can't often be
• 11:19surgically removed because they grow
• 11:20around nerves. So this is
• 11:21a really, awful disease, and
• 11:23it's a lifelong disease because,
• 11:25you know, this is a
• 11:25germline,
• 11:26mutation,
• 11:27and it just sort of
• 11:28get gets worse over time.
• 11:30These kids are also,
• 11:32at risk of some malignancies,
• 11:33including,
• 11:34those listed here. But the
• 11:36worst, which happens in about
• 11:37ten percent of the cases,
• 11:39they can lose genes in
• 11:40the polycomb repressive complex and,
• 11:43become, malignant. And these are
• 11:45malignant peripheral nerve seed tumors,
• 11:47and these are fatal.
• 11:49So, this is a disease
• 11:51which has been known for
• 11:52a very long time when
• 11:53gene was discovered and shown
• 11:54to be a protein that
• 11:55regulates RAS, as we showed
• 11:57in nineteen ninety.
• 11:59All the complex phenotypes here
• 12:00turn out to be probably
• 12:02just too much active RAS.
• 12:04A little too much in
• 12:05the in the, heterozygotes,
• 12:07and way too much in
• 12:08the clonal cells would lose
• 12:09a second allele.
• 12:12So we put a lot
• 12:12of work into understanding how
• 12:14the NF1 protein is regulated.
• 12:16It's,
• 12:17it's a really interesting protein.
• 12:19It's a gigantic protein. It's
• 12:20two thousand eight hundred amino
• 12:21acids long. For a long
• 12:22time, it was just drawn
• 12:23as a stick like this
• 12:25with the gap domain in
• 12:26the middle, a sec fourteen
• 12:27domain, which is involved in
• 12:28membrane localization, and then unknown
• 12:30territory all across the whole
• 12:32gene. But then we and
• 12:33others solve the cryo EM
• 12:34structure of the protein, and
• 12:36it turns out to be
• 12:37a really beautiful structure. It's
• 12:38actually a head to tail,
• 12:40dimer,
• 12:41in this sort of infinity
• 12:42shaped protein, which is mostly
• 12:44made up of coils of
• 12:45alpha helices. And then there's
• 12:47RAS binding that main from
• 12:48one protein here and then
• 12:49one here.
• 12:50And this gigantic protein is
• 12:52taken to RAS in the
• 12:53membrane through as adapt to
• 12:55protein called SPREAD1,
• 12:57and SPREAD2,
• 12:59and also through binding directly
• 13:00to activated receptors like c
• 13:02kit. And once this gets
• 13:04to the membrane, it turns
• 13:05RAS off. This whole process
• 13:06is extremely
• 13:08tightly regulated.
• 13:09Let's say a level of
• 13:10expression spreads and phosphorylation probably
• 13:12of NF one, and binding
• 13:14to the receptor and so
• 13:15on. Because we know from
• 13:16genetics, if you have half
• 13:18as much of its protein
• 13:19as normal, you have a
• 13:20major disease.
• 13:21So it's got to be
• 13:22very precisely regulated.
• 13:24So we're doing a lot
• 13:25of work on that.
• 13:27Now there's not much good
• 13:28news in the NF one
• 13:29world, but there was some,
• 13:31a few years ago when
• 13:32Brigitte Wideman of the NCI
• 13:34pediatric oncology group performed a
• 13:36clinical trial on a MEK
• 13:37inhibitor
• 13:38to turn down the MAP
• 13:39kinase pathway,
• 13:40in patients with,
• 13:42plexiform neurofibromas.
• 13:44So this is a brilliant
• 13:45clinical trial because this is
• 13:47not a disease which is
• 13:48life threatening. So we're not
• 13:50measuring clinical endpoints in terms
• 13:51of survival or progressive free
• 13:53survival. This these endpoints were
• 13:55a reduction of tumor burden
• 13:57and also reduction of pain
• 13:58and other quality of life
• 13:59issues. But she got the
• 14:01drug approved, and this is
• 14:02now, widely used with kids
• 14:03for NF one.
• 14:05So the effects can be
• 14:06dramatic, but,
• 14:08those of you who've been
• 14:09involved in clinical use of
• 14:10mecadipid does know these drugs
• 14:11are toxic. And most kids
• 14:13go off this drug after
• 14:14a few months because of
• 14:15the side effects,
• 14:16are really bad and and
• 14:18to the GI and other
• 14:19issues. So we need a
• 14:20better way of dealing with
• 14:21this disease.
• 14:22And
• 14:23hope is on the way,
• 14:25and it is just hope
• 14:26at this point.
• 14:27But we think that most
• 14:28of the, activity
• 14:30that's activated when you lose
• 14:31your neurofibromine protein is actually
• 14:33KRAS,
• 14:34and we think that because
• 14:36if we measure the ability
• 14:37of RAS family proteins to
• 14:38bind to neurofibromine, the NF
• 14:40one protein,
• 14:41KRAS is the major partner.
• 14:43And if we knock out
• 14:44NF one cells, KRAS is
• 14:46the one that comes to
• 14:47life, most actively as we
• 14:48and others have shown.
• 14:50So this all hangs together.
• 14:52This means the drugs that
• 14:53are now being developed for
• 14:54cancer,
• 14:55which have the ability to
• 14:57target wild type KRAS could
• 14:59be tested in NF one
• 15:00disease, and we we are
• 15:01planning on doing exactly that
• 15:03with a pan KRAS inhibitor.
• 15:05And the theory is that,
• 15:06it should be as potent
• 15:07as a MEK inhibitor but
• 15:08without all the side effects
• 15:09because we're only eating KRAS
• 15:11and not all the RAS
• 15:12proteins that MEK,
• 15:14would deal with. So we
• 15:15hope that would be, heading
• 15:16for the clinic, in the
• 15:17next, year or two, one
• 15:18of several KRAS drugs, which,
• 15:21hit wild type as well
• 15:22as mutant alleles.
• 15:26Now I mentioned that, that
• 15:28autism and behavioral problems are
• 15:29a big issue with NF1
• 15:30patients. These issues are even
• 15:32more severe in people who
• 15:34have lost one copy,
• 15:36of the, SYNGAP gene, which
• 15:38is another gap related protein
• 15:39shown down here. This one
• 15:40is brain specific,
• 15:42but people lose this by
• 15:43chance head in the in
• 15:45heterozygous state,
• 15:46and, they they then succumb
• 15:48to, very severe forms of,
• 15:50autism, including, epilepsy.
• 15:53So these are some of
• 15:54the mutations, which have been
• 15:55mapped to cause, autism.
• 15:57So this is not proven
• 15:58really formally yet, but it
• 16:00looks like too much map
• 16:01plan is pathway in specific
• 16:03parts of the brain,
• 16:04can lead to, these kind
• 16:06of behaviors.
• 16:07Again, it hasn't been formally
• 16:08proven, but now we have
• 16:09drugs actually to test that
• 16:10possibility,
• 16:11very specific KRAS and other
• 16:13RAS pathway,
• 16:14inhibitors.
• 16:15So
• 16:16I believe that the RAS
• 16:18field will will now start
• 16:19to pay more attention to
• 16:20the effects of RAS signaling
• 16:22in the brain. Ras proteins
• 16:23have always been known to
• 16:24be very abundant in the
• 16:25brain, but, since obviously,
• 16:28these
• 16:29until recently were thought to
• 16:30be non proliferative tissues,
• 16:32there's always been a question
• 16:33as why do we why
• 16:34do you have so much
• 16:34map size activity in the
• 16:36in the brain? So that
• 16:37needs to be revisited.
• 16:41Anyway,
• 16:43move towards cancer.
• 16:44So, in cancer, we have
• 16:46a different situation again. Now,
• 16:48mutations occur in KRAS, codons
• 16:50twelve, and sixty one as
• 16:51you definitely all know.
• 16:53These mutations,
• 16:54make RAS proteins resistant to
• 16:56all gaps, so they get
• 16:57turned off by any any
• 16:58gaps in the cell can
• 17:00no longer, turn off this
• 17:01protein.
• 17:02So these proteins accumulate in
• 17:04a GTP bound state as
• 17:06shown by this,
• 17:07t l c analysis of
• 17:08nucleotides bound to RAS for
• 17:10wild type RAS, mostly GDP
• 17:12bound, q g
• 17:14q sixty one no. G
• 17:15twelve c, mostly GDP bound,
• 17:18and then,
• 17:19g twelve c and v
• 17:20and q sixty one. So
• 17:21the mutant pure proteins accumulate
• 17:23in the GTP bound state,
• 17:24and that's,
• 17:26part of what causes cancer.
• 17:28But the consequence of this
• 17:29lack of regulation is we've
• 17:30lost that pulsatile spike of
• 17:32signaling that you see in
• 17:33response to EGF. And now
• 17:35we have a situation where
• 17:36you have tonic activation of
• 17:37phospho work.
• 17:38These spikes actually, obviously, this
• 17:40tonic activity is not necessarily
• 17:42as as high as a
• 17:43spike in in a normal
• 17:44cell. It's just on the
• 17:46whole time. And this persistent
• 17:48activation of the MAP kinase
• 17:49pathway,
• 17:50from as you can see
• 17:51from RAS GDP levels or
• 17:52from fossil
• 17:53work, this causes
• 17:55massive change in the transcriptional
• 17:56profile in cells and can
• 17:58lead to their, change of
• 18:00a normal cell into a
• 18:01tumor cell.
• 18:02So that is the basis
• 18:03of RAS driven cancer, persistent
• 18:05tonic activation of the MAP
• 18:07kinase pathway.
• 18:08Pathway is not on fire.
• 18:09It's just always on, and
• 18:10that's a very different signal
• 18:12for the cell to interpret.
• 18:13And, that, I think, is
• 18:15the basis of the whole
• 18:15thing.
• 18:16But RAS proteins do more
• 18:18than that in cancer as
• 18:19I'm sure you all appreciate.
• 18:21If you ask,
• 18:22Google
• 18:23what RAS proteins do in
• 18:24cancer cells, you'll probably find
• 18:26a picture like this, which
• 18:27is based on, you know,
• 18:28one of fifty diagrams for
• 18:29literature in which RAS proteins
• 18:31can activate multiple downstream effectors
• 18:33from various publications
• 18:35over the years.
• 18:37At the RAS initiative, we
• 18:38did a sort of crowdsourcing
• 18:39version of this and asked
• 18:40the community of RAS people,
• 18:42what do you think RAS
• 18:43proteins do? And, of course,
• 18:44they all voted for their
• 18:46favorite protein in the in
• 18:47the pathway,
• 18:48and you get a sort
• 18:49of conglomerate picture like this
• 18:50in which RAS is here.
• 18:51Here's the map kinase pathway.
• 18:53The RAS proteins also activate
• 18:55other downstream
• 18:56effect as shown here.
• 18:58Now all of these,
• 18:59pathways are based on data
• 19:01from cell lines, and the
• 19:03bottom line is that,
• 19:05the RAS always engages MAP
• 19:06kinase,
• 19:07and can activate other effectors
• 19:09in different cell lines, and
• 19:10that's the answer that you
• 19:10get from CHaT GPT if
• 19:12you ask the same question.
• 19:14MAP kinase pathway, yes, always.
• 19:16Most tumor cells also activate
• 19:18p I three kinase. Some
• 19:19activate RAL GDS and some
• 19:21additional effectors, which are kind
• 19:23of more rare, but they're
• 19:24still real. They're just less
• 19:25common.
• 19:27So that's the situation in
• 19:28in cancer.
• 19:29But then as as a
• 19:30biochemist, you can ask, well,
• 19:31how does a small protein
• 19:32like RAS, which has a
• 19:33very small effective binding region,
• 19:35so twenty to ten amino
• 19:36acids, how can it possibly
• 19:38engage so many different downstream
• 19:39proteins effectively and turn them
• 19:41on,
• 19:42to engage downstream pathways.
• 19:45So,
• 19:47this is what I think
• 19:48is happening. So in normal
• 19:49cells, RAS proteins only activate,
• 19:52pathways pathway I mean, RAS
• 19:54proteins.
• 19:55However, RAS proteins have cousins,
• 19:57such as rRAS2
• 19:58and MRAS and and RAL.
• 20:00And if you overexpress
• 20:02a mutant form of RAS
• 20:03at high levels, it can
• 20:04cross over,
• 20:06and then interact with PI3
• 20:07kinase
• 20:08or with SHOCK two or
• 20:10with other members of these
• 20:11proteins
• 20:12just because the proteins are
• 20:13so similar.
• 20:14So,
• 20:15as I'll show you in
• 20:16just a moment, the binding
• 20:17size of these proteins are
• 20:18are very
• 20:20similar.
• 20:21So this is like sort
• 20:22of, unwanted inappropriate,
• 20:25interactions of RAS with other
• 20:26effectors.
• 20:27It kind of reminds me
• 20:28of, Donald Trump at a
• 20:30beauty patent.
• 20:32If I can use that
• 20:32analogy.
• 20:34So,
• 20:36inappropriate rations, which shouldn't happen
• 20:38normally, but do because we
• 20:39have a high level of
• 20:40mutant RAS in the GTP
• 20:41bound state.
• 20:42And, actually, this sort of
• 20:43goes the other way as
• 20:44well. Mutations in our RAS
• 20:46and our RAS two proteins
• 20:47that cause,
• 20:49Luna syndrome can crossover and
• 20:51activate in that kinase pathway.
• 20:52And that's why they all
• 20:53can do that, whereas they
• 20:54don't do that in normal
• 20:55tissue. Okay? This is a
• 20:56gain of function as a
• 20:58result of over expression and
• 20:59mutation in in the target.
• 21:01So that's what we think
• 21:02how,
• 21:03RAS proteins can pick up
• 21:04all these different interactions by
• 21:06sort of barging in on
• 21:07their cousins,
• 21:08which look very similar. So
• 21:10just for those of you
• 21:11who like details, this is
• 21:12a
• 21:13the the RAS, g domain.
• 21:16This is the effect of
• 21:17binding region of the RAS
• 21:18protein. So h n and
• 21:19K RAS are identical, but
• 21:21so is, MRAS, RS, RS
• 21:23two, and RIT one. They
• 21:24all have the same,
• 21:26effect of binding region.
• 21:28So all of these cousins
• 21:29can interact with RAF,
• 21:31for example, but only canonical
• 21:33RAS proteins activate it,
• 21:35and that's because activation has
• 21:37a is a two step
• 21:38process, binding followed by a
• 21:40second engagement of a different
• 21:41part of the protein to
• 21:42turn on the actual activity,
• 21:44and that's unique to the
• 21:45canonical RAS proteins.
• 21:47So, in the case of,
• 21:49of of RAS and RAF,
• 21:51this is the, the minimal
• 21:53binding domain of,
• 21:54of RAS and RAF that
• 21:55we solved,
• 21:57many years ago.
• 21:58In addition to that, we
• 21:59solved a bigger piece of
• 22:00RAF, including the sister enriched
• 22:02domain shown here,
• 22:03which now
• 22:04extends the footprint on RAS
• 22:06to a much bigger footprint,
• 22:08and this CRD interaction with
• 22:10RAS is essential for the
• 22:11activation process.
• 22:13Okay? So if you don't
• 22:14have that, you can get
• 22:15binding, but nothing happens.
• 22:17And, so we think that's
• 22:18that's interaction here is actually
• 22:20really interesting, also interesting,
• 22:23interesting drug target because this
• 22:25direction is weaker than the
• 22:26RBD, and it actually has
• 22:27some specificity for different RAF
• 22:29isoforms.
• 22:30But that's the general principle.
• 22:31We have a binding piece
• 22:32and then a sort of
• 22:33activation piece, and those are
• 22:35unique to different members of
• 22:36the RAF family, but they
• 22:37can cross over. If you
• 22:38over express proteins,
• 22:40too much. They can they
• 22:41can pick up interactions which
• 22:42are not supposed to happen.
• 22:46Okay. So,
• 22:47the, the world of,
• 22:50of drug discovery in the
• 22:51RAS world really started in
• 22:53a in a serious way
• 22:54in twenty thirteen
• 22:56when Kevan Shokat at UCSF,
• 22:58had the brilliant idea of
• 22:59targeting the g twelve c
• 23:00allele of KRAS.
• 23:03So g twelve c is
• 23:04common in non small cell
• 23:05lung cancer because it's a
• 23:06hallmark of cigarette smoke.
• 23:09But cystain is a reactive,
• 23:11residue, and by sheer good
• 23:13fortune,
• 23:14Kevan's lab is close to
• 23:15the,
• 23:17a group at, at at
• 23:18UCSF that developed a cystain
• 23:20tethering library for finding small
• 23:21molecules applying to cystains and
• 23:23proteins.
• 23:24So So K band put
• 23:25two and two together under
• 23:26the screen with that library
• 23:28to find compounds that were
• 23:29bind to g twelve c,
• 23:30and he found compounds, as
• 23:31you all know, I'm sure,
• 23:32and found a compound,
• 23:34that binds in the, in
• 23:35the pocket here that hadn't
• 23:36been seen previously. It's an
• 23:37induced pocket, actually.
• 23:39It's called the switch two
• 23:40pocket. It tucks in under
• 23:42here through covalent interaction.
• 23:45So with this experiment, he
• 23:46killed two birds with one
• 23:47stone.
• 23:48First of all, he got
• 23:49a a compound of binds
• 23:50to RAS with high affinity.
• 23:51It's covalent.
• 23:52And secondly, this is specific
• 23:54for the mutant allele, so
• 23:55it should be a safe
• 23:56drug. So, obviously, everybody jumped
• 23:59on this immediately and then,
• 24:01followed this
• 24:02this path, and now we
• 24:03have some fifteen different D12C
• 24:05inhibitors,
• 24:06in the clinic, And, also,
• 24:08having bust open the sort
• 24:09of the knot of RAS,
• 24:11now we see this pocket
• 24:12can be used for actually
• 24:13non covalent interactions,
• 24:15as well.
• 24:18So these this summarizes RAS
• 24:19inhibitors in the pipeline.
• 24:21The
• 24:22pink ones here are approved.
• 24:24That's from Amgen and Marathi.
• 24:25All the blue ones are
• 24:26in clinical trials right now.
• 24:28This is probably
• 24:29out of date after those
• 24:30of you went to ACR.
• 24:32Can't have, you know, escaped
• 24:34the fact that half a
• 24:35torsion KRAS inhibitors.
• 24:37So these are all in
• 24:38the clinic, but the majority
• 24:40of these are knockoffs of
• 24:41KVAN's original idea, the,
• 24:44compounds.
• 24:45And, by chance, the pocket
• 24:47that KVAN discovered is only
• 24:49accessible,
• 24:50apparently, at that time in
• 24:52the, off state of the
• 24:53protein, the GDP bound state.
• 24:55Okay? So these are all
• 24:56off state inhibitors.
• 24:58Now for those of you
• 24:59who think about RAS at
• 25:00all, you'll think, well, hang
• 25:02on. What's the point of
• 25:03inhibiting the the off state
• 25:04of RAS when you wanna
• 25:05hit inhibit the on state?
• 25:06But turns out RAS protein
• 25:08cycle between the two states.
• 25:09So a compound that traps
• 25:10the protein in the off
• 25:11state prevents it becoming activated
• 25:14is a very good strategy.
• 25:15Right? Trapping it in the
• 25:16off in the off state,
• 25:17so it's seen.
• 25:20However, there are compounds in
• 25:21the clinic now, including when
• 25:22I'll mention, just coming up,
• 25:24BBOA five twenty, which does
• 25:26actually,
• 25:27bind covenylate to cysteine in
• 25:29a GTP bound state.
• 25:31This is really hard to
• 25:32do. We did this at
• 25:33Frederick,
• 25:34Merrill Anderson collaboration with
• 25:36a partner of Bio Oncology,
• 25:38because the the the GTP
• 25:39bound state is much tighter
• 25:41than the GDP bound state.
• 25:42It took years of medicinal
• 25:43chemistry to find a compound
• 25:45that would fit in into
• 25:46that that pocket,
• 25:47but we we were able
• 25:48to do that.
• 25:50I'll come back to that
• 25:51in just a minute and
• 25:51tell you how it works.
• 25:53These hit, t twelve d,
• 25:55which is the most common
• 25:56allele overall in human cancer,
• 25:58and, these in the clinic,
• 26:00and these are preclinical, and
• 26:01there are many more behind
• 26:02this. These are not covalent
• 26:03for the most part. There's
• 26:04one but most of them
• 26:05is not covalent.
• 26:06But some of these, like
• 26:07this Murati compound, which is
• 26:09published
• 26:11this has a binding constant,
• 26:12to RAS, to the
• 26:15to RAS, which is in
• 26:16almost unmeasurably tight. It's it's,
• 26:18it's an off rate, which
• 26:19is, in a matter of
• 26:20hours.
• 26:21The binding constant is sub
• 26:22way sub picomolar,
• 26:24and it's non covalent.
• 26:26And it just goes to
• 26:26show, you know, someone has
• 26:27said, you know, ten years
• 26:28ago, do you think we'll
• 26:29ever find a picomolar compound
• 26:31that binds to RAS in,
• 26:32you know, noncovalently?
• 26:35It it said, well, no
• 26:36way.
• 26:38But it that's that's the
• 26:39way it is. So, it
• 26:41just sort of got cracked
• 26:42piece by piece.
• 26:43So these compounds are, again,
• 26:45in the clinic. These are,
• 26:47other versions of different alleles.
• 26:49These are PAN KRAS that
• 26:50I mentioned earlier, might be
• 26:52used for NF1 disease. We
• 26:53have one at BridgeBio,
• 26:55this one here, which is
• 26:56in the clinic now. And
• 26:57we're developing a a version
• 26:58of that for the for
• 26:59treating, NF one.
• 27:02So from, you know, nothing,
• 27:04twenty thirteen, we now have
• 27:05a whole, you know, whole
• 27:06pipeline of RAS, inhibitors. Okay.
• 27:10Okay. So, again, these are
• 27:12the the two, that were
• 27:13first approved.
• 27:15I think these went into
• 27:16clinic pretty quickly, and and,
• 27:18Amgen and
• 27:19and and Marathi had these
• 27:20drugs approved,
• 27:22as a single agents or
• 27:24in this case, in combination,
• 27:25with cetuximab
• 27:26in twenty
• 27:27four. However,
• 27:29although there was somewhat somewhat
• 27:30of an advantage in of
• 27:32progression free survival early on,
• 27:34overall,
• 27:35there's no real survival benefit
• 27:37of using these drugs, compared
• 27:38with standard of care chemotherapy.
• 27:41So this was obviously disappointing.
• 27:44In some ways, not too
• 27:45surprising because we know that
• 27:46if you inhibit the RAS
• 27:47pathway, you activate upstream signaling
• 27:49pathway by de repressing EGF
• 27:51receptor, for example, and and
• 27:53that has to be dealt
• 27:53with. But I think most
• 27:54people were expecting a much
• 27:56more dramatic response than this.
• 27:58And, the problem is these,
• 28:00tumors acquire resistance,
• 28:02very quickly,
• 28:03and this has been studied
• 28:04quite carefully.
• 28:05About at least, I think,
• 28:07the current betting is about
• 28:09two thirds of the all
• 28:10the, resistance mechanisms
• 28:12relate to upstream signaling. Okay?
• 28:14And that's because of a
• 28:15of a fundamental flaw in
• 28:17the original concept. Although it's
• 28:18a brilliant concept, the floor
• 28:20is that the drug has
• 28:21to bind to the inactive
• 28:22state of the
• 28:30sorry.
• 28:31The drug has to bind
• 28:32to the inactive state of
• 28:32the of the RAS protein,
• 28:34which is fine.
• 28:35So the drug binds here
• 28:37and then, basically, locks it
• 28:38in this state and prevents
• 28:39it going back to the
• 28:40active state. The problem is
• 28:42the tumors can quite easily,
• 28:44apparently, find ways of increasing
• 28:46upstream signaling.
• 28:47So if SARS is activated
• 28:49and gets to the inactive
• 28:50form of RAS first, it'll
• 28:52kick the protein back to
• 28:53the active state, and off
• 28:54we go again down this
• 28:55pathway.
• 28:56So the drug has to
• 28:57compete with endogenous
• 28:59RTK signaling.
• 29:00And tumors have found ways
• 29:01of amplifying RTK signaling,
• 29:04or even making more KRAS
• 29:06so the drug can't keep
• 29:07up and can't it can't
• 29:08hold back the protein in
• 29:09and out of state because
• 29:10there's tremendous pressure on upstream
• 29:13to get the to get
• 29:14the protein back into the
• 29:15on state.
• 29:17So the initial
• 29:19once this was realized, the
• 29:20number of clinical trials were
• 29:21started with drugs that hit
• 29:22inhibit SHIFT two to try
• 29:24and prevent upstream signaling or
• 29:26SARS or RTK signaling,
• 29:28and for the most part,
• 29:29except in lung in the
• 29:30colorectal cancer,
• 29:32for different reasons maybe, this
• 29:33has not been very successful
• 29:35because these drugs are just
• 29:36toxic. They're just they're like
• 29:37growth factor signaling inhibitors that
• 29:39don't have any any, any
• 29:41specificity. So
• 29:43combining
• 29:44the off state with upstream
• 29:45inhibitors has not been a
• 29:46successful strategy, I would say.
• 29:48So a better strategy would
• 29:50be to find a compound
• 29:50that hits the GTP bound
• 29:52state directly, and then, you
• 29:54wouldn't have this problem
• 29:55in in theory.
• 29:59So that's what we did.
• 30:00As I said, this took
• 30:01a long time, for we
• 30:02did this in collaboration with
• 30:03the group at, at British
• 30:05British Bio Oncology.
• 30:07We were able to do
• 30:07this because right from the
• 30:08get go in the RAS
• 30:09initiative,
• 30:10our mandate was to attack
• 30:11pancreatic cancer,
• 30:13not g twelve c lung
• 30:14cancer. But pancreatic cancer is
• 30:16mostly g twelve d and
• 30:18v, which are much more
• 30:19GTP bound than g twelve
• 30:20c, and we wouldn't be
• 30:22able to use the covalent
• 30:23approach.
• 30:24Right from the start, we
• 30:25worked on
• 30:26drugs that hit the GTP
• 30:27form of mutant RAS proteins.
• 30:30And, eventually,
• 30:31as we're making drugs,
• 30:32for to target pancreatic cancer,
• 30:34we realized we could quickly
• 30:35quite easily make one of
• 30:36those into a g two
• 30:37l c compound because cystine
• 30:38was right close by to
• 30:40our active site. So we
• 30:41have some experience in targeting
• 30:43the on the on state,
• 30:44which we translated into this
• 30:46compound.
• 30:47So here's some specs. This
• 30:48is all published so, recently,
• 30:49so I won't go into
• 30:50any detail. It's extremely potent
• 30:52with,
• 30:53seventy, p q m l
• 30:54I c fifty and cell
• 30:55lines, very, very clean, good
• 30:57good drug like properties,
• 30:59and,
• 31:00off we go.
• 31:02Of course, there's twenty other
• 31:04Rasnovas in the clinics.
• 31:06So showing this is better
• 31:07than the other ones is
• 31:08is our challenge, but, we
• 31:09believe that the on state
• 31:10will be more effective, but
• 31:12we need to prove it.
• 31:13But you can ask, okay.
• 31:14Well, the the off state,
• 31:15we can now sort of
• 31:16imagine how it works. It
• 31:17traps the protein in the
• 31:18in the active state. How
• 31:19did the drug work if
• 31:20it binds to the on
• 31:21state? How does that prevent
• 31:22signaling?
• 31:24Well, this shows, first of
• 31:25all, it engages the target
• 31:26much more quickly
• 31:28than the approved drugs. So
• 31:29this is a RAS RAF
• 31:30engagement, target engagement assay. We
• 31:32shut down RAS RAF binding
• 31:34within minutes, whereas the two
• 31:35approved drugs that clinical doses
• 31:38takes, you know, hours. That's
• 31:39because it takes a long
• 31:41time to hydrolyze GTP to,
• 31:43to to the GDP form
• 31:45so the drug can actually
• 31:46interact with the with the
• 31:47with the target. This is
• 31:48well known.
• 31:49So faster and more complete
• 31:51target engagement, we hope, will,
• 31:53translate into better clinical outcome.
• 31:55We don't know yet.
• 31:56But the way that it
• 31:57works is to take advantage
• 31:58of something which we we,
• 32:00noticed when we first started
• 32:01working on structures of RAS
• 32:02proteins,
• 32:05many years ago.
• 32:06So Symantje and Frederick solved,
• 32:08structures of different mutant RAS
• 32:10alleles, and he noticed that
• 32:12in some structures,
• 32:13the switch one region, which
• 32:15is where RAF, interacts, that's
• 32:17the effect of region I
• 32:18showed you earlier,
• 32:19can be in one of
• 32:20two different, configurations,
• 32:21confusingly called state one and
• 32:24state two.
• 32:25So this is state,
• 32:27one in which the switch
• 32:28one region, the effective binding
• 32:29region is away from the
• 32:31main,
• 32:32body of the g domain.
• 32:33Whereas in in, state two,
• 32:35it's tucked down and makes
• 32:37contact with the magnesium,
• 32:38and directly to the, GTP
• 32:40in the GTP molecule in
• 32:42the, active site. So this
• 32:44is the productive state of
• 32:45of RAS that interacts with
• 32:46its effectors. Okay? We know
• 32:48that because we cocrystallize
• 32:50RAS with RAS RBD, and
• 32:52you can see that switch
• 32:54one is in,
• 32:56the state two configuration.
• 32:58Okay?
• 32:59So,
• 33:00this dynamic switch between or
• 33:03or equilibrium, I should say,
• 33:04between state one and state
• 33:05two has been known for
• 33:06many years from,
• 33:08NMR data from,
• 33:10Karl Bitzer and Whittinghofer and
• 33:11others from way back. Who
• 33:13looked at, NMR
• 33:15spectra from emitted from the
• 33:16phosphates in the RAS protein.
• 33:18So these are this is
• 33:19thirty one p NMR
• 33:20where you can see resonances
• 33:22from the alpha, beta, and
• 33:24gamma phosphates in in the
• 33:26g domain.
• 33:27So, the point here is
• 33:28we can tell state two
• 33:29from state one by the
• 33:30resonance of the the gamma
• 33:32phosphate, gamma one, gamma two.
• 33:34And we add our drug
• 33:35to these, g domains, the
• 33:37protein starts to accumulate in
• 33:39state one, which is the
• 33:40form that can't bind RAF.
• 33:42Okay?
• 33:43So we think the drug
• 33:45sort of forces its way
• 33:46into the, into the g
• 33:47domain, this tiny pocket, and
• 33:49then,
• 33:50pops out the switch one
• 33:52region into a state where
• 33:53it can no longer interact
• 33:54with RAF. And then the
• 33:55protein is GTP bound but
• 33:57dead.
• 34:00K? So that's,
• 34:01that's our theory.
• 34:03And papers published to support
• 34:04that. So that actually should
• 34:06that will interfere with all
• 34:07effective binding at, let's say,
• 34:09at,
• 34:10at,
• 34:11switch one.
• 34:13K?
• 34:14So as I said, this
• 34:15the drug has now been
• 34:16through a dose escalation in
• 34:17phase one and, it's now
• 34:18being tested in combination with
• 34:19checkpoint inhibitors,
• 34:21because,
• 34:22in lung cancer,
• 34:24the ideal position to be
• 34:25in to make a real
• 34:26impact on patients is to
• 34:28go to frontline
• 34:29or first line,
• 34:30say, of of, treatment in
• 34:32which,
• 34:33a a drug is combined
• 34:34with a checkpoint inhibitor. Right
• 34:36now, the first line would
• 34:37be checkpoint plus a chemo,
• 34:40but we'd like to replace
• 34:41chemo,
• 34:42with a a a RAS
• 34:43inhibitor. So that's the goal,
• 34:44and that's what every company
• 34:45that has a RAS inhibitor
• 34:47is shooting for. Is that
• 34:48true?
• 34:51Okay. But it hasn't been
• 34:52easy.
• 34:53Well, nothing's easy, but it's
• 34:54been particularly difficult because
• 34:57these g twelve c inhibitors
• 34:58are covalent,
• 35:00and,
• 35:01the problem is if you
• 35:02if you give a high
• 35:02dose of a covalent compound,
• 35:04the theory is it can
• 35:05haptenize proteins in the liver.
• 35:07It can react with proteins
• 35:08randomly in the liver, and
• 35:09they become antigens. So when
• 35:11you add that effect to
• 35:12a checkpoint inhibitor, now you
• 35:14start to get an immune
• 35:15problem, in the liver.
• 35:17Not sure that's true, but
• 35:18that's, I think, the theory.
• 35:19Yeah. K. So
• 35:21yeah.
• 35:22So,
• 35:23so far, the compound that
• 35:25I mentioned, the the, OnState
• 35:27BB, BBO a five twenty,
• 35:29is so potent. We can,
• 35:32get really good equivalent effects
• 35:33of target engagement at much
• 35:35lower concentration, and we've not
• 35:36seen any liver tox yet,
• 35:38even at the highest dose
• 35:39in combination with,
• 35:41with, checkpoint inhibitors. But that's
• 35:43all, you know, playing out
• 35:45over time.
• 35:46So we will we will
• 35:47see. That one is,
• 35:49in the hands of the
• 35:50of our clinical colleagues. See
• 35:51see what counts out.
• 35:53No. Anyway
• 35:55okay. So the second
• 35:56drug, which entered the clinic
• 35:57a bit little after the
• 35:58the first one, is called
• 36:00the breaker.
• 36:01K?
• 36:02So the breaker, another massive
• 36:03collaboration between the, rash initiative,
• 36:06at Frederick National Lab, headed
• 36:08up by,
• 36:09this manchu,
• 36:11and bridge by bridge by
• 36:13bio oncology therapeutics,
• 36:15headed up by,
• 36:17Pedro Beltran.
• 36:18And Eli Wallace is the
• 36:19CEO of this, company we
• 36:21set up really just to
• 36:22develop these three drugs out
• 36:23of the Federal National Lab.
• 36:25Eli is a drug discovery,
• 36:27expert. He developed,
• 36:29selamatinib when he was at
• 36:31Array and the, HIF two
• 36:32alpha inhibitor at at Peloton.
• 36:35So it's a great team,
• 36:37and they developed this compound
• 36:38that we call, the breaker.
• 36:41I should also say that
• 36:42this required a lot of
• 36:43computational,
• 36:45input from group at Lawrence
• 36:46Livermore National Labs that have
• 36:48a supercomputer
• 36:49that until recently was used
• 36:50to, model thermonuclear explosions.
• 36:54When they got fed up
• 36:55with doing that, they came
• 36:56to the NCI and say,
• 36:57so is there a really
• 36:58difficult project you can work
• 36:59on
• 37:00using our supercomputers?
• 37:01So we've been working with
• 37:02them on modeling RAS, RAF
• 37:04interactions,
• 37:05in silico and also on
• 37:07developing, drugs and UA was
• 37:08the head of this group,
• 37:10at Lawrence Livermore Lab.
• 37:13So this drug,
• 37:15prevents RAS activating PI three
• 37:17kinase.
• 37:19People in the RAS world,
• 37:21will know that the relationship
• 37:22between RAS and PI three
• 37:23kinase has been known for
• 37:25a long time in a
• 37:25kind of vague way,
• 37:27because it's not clear whether
• 37:28RAS proteins directly activate PI
• 37:30three kinase or they do
• 37:31it take advantage of PI
• 37:33three kinase from other sources,
• 37:34and it's really, really been
• 37:36very difficult to figure out.
• 37:38We know that,
• 37:39inhibiting MAP kinase pathway,
• 37:41with a RAS inhibitor
• 37:43plus PI three kinase inhibitor
• 37:44has synergistic effects.
• 37:46So,
• 37:47this is shown in this
• 37:48mouse model from my colleague,
• 37:50Martin McMahon, who showed that
• 37:51if you activate the MAP
• 37:52kinase pathway with a V
• 37:54six hundred e BRAF, which
• 37:55is just basically RAF is
• 37:56on fire. Okay? There's no
• 37:58BRAF kinase activation. It's just
• 38:00on fire.
• 38:01So in a lung cancer
• 38:02model, this, allele,
• 38:04causes tumors and mice die
• 38:06of the disease. If you
• 38:07add on top of that
• 38:08a PI3 kinase mutation, now,
• 38:10the mice die much more
• 38:11quickly and the tumors are
• 38:13much more aggressive.
• 38:14So that's the sort of
• 38:15gain of function combination, and,
• 38:17reciprocally, if you inhibit these
• 38:19two pathways with,
• 38:21particular inhibitors of these two
• 38:22pathways as shown by Jeff
• 38:23Engelman and colleagues, you can
• 38:24cure
• 38:25KRAS tumors in mice.
• 38:27That was a challenge that
• 38:28Talajax put out a long
• 38:29time ago. We had ten
• 38:30thousand bucks for anybody who
• 38:32could cure a KRAS tumor
• 38:33in mice, and this was,
• 38:34I think, the first time
• 38:35it was done using two
• 38:36different drugs.
• 38:39However,
• 38:40the combination of these drugs
• 38:41is really toxic.
• 38:43Mecanibers, as I mentioned for
• 38:45NF one, are toxic,
• 38:47especially if it does high
• 38:47enough to have an effect
• 38:48on a malignant tumor. And
• 38:50PI three kinase inhibitors are
• 38:51toxic for multiple reasons, including,
• 38:54generation of hypoglycemia
• 38:56because insulin signaling depends on
• 38:58p I three kinase.
• 39:00So in theory, this is
• 39:01a great idea, but it
• 39:01hasn't been possible to find
• 39:03a combination of drugs which
• 39:04is safe enough,
• 39:06to to test this in
• 39:07in people.
• 39:09So just to show you
• 39:10the complexity of this situation,
• 39:11this is an experiment done
• 39:12by my ex postdoc many
• 39:13years ago where he transfected
• 39:15into cells a whole bunch
• 39:16of different RAS alleles or
• 39:17different RAS family members
• 39:19and and then look for
• 39:20their effect on different isoforms
• 39:22of PI three kinase,
• 39:23alpha, beta, gamma, and delta.
• 39:25And you can see here
• 39:26that six different RAS proteins
• 39:28can activate alpha.
• 39:30The same,
• 39:31SACE can do,
• 39:33delta.
• 39:33Two do sorry, gamma. Two
• 39:35do delta, and none of
• 39:36them do beta.
• 39:37The g proteins can activate
• 39:39the p I three kinase
• 39:40beta.
• 39:41So in a typical cell,
• 39:42you can get inputs to
• 39:42p I three kinase from
• 39:44multiple sources going to different
• 39:45isoforms.
• 39:46That makes life, you know,
• 39:47really complicated to figure out
• 39:48what RAS is adding to
• 39:50that, that picnic.
• 39:52Well, this was, solved at
• 39:53least,
• 39:55in a sort of very
• 39:56specific way by a really
• 39:57brilliant experiment by,
• 39:59Julian Downwood,
• 40:01who showed, many years ago
• 40:02back in two thousand seven
• 40:04that if you, engineer mice
• 40:06so that, I I got
• 40:08goosebumps thinking about this experiment.
• 40:09It's, you know, really awesome.
• 40:12He engineered mice in which,
• 40:14he took a mutant allele
• 40:15of p p I kinase
• 40:16alpha, p one ten alpha,
• 40:18mutated the, the RAS binding
• 40:20domain with these two mutations
• 40:21so that this
• 40:23domain can't interact with any
• 40:25RAS protein. Okay. RAS, MRAS,
• 40:27whatever cannot bind. That site
• 40:29is dead.
• 40:30So he in several experiments,
• 40:32but the most important one,
• 40:33he'd knocked this allele into
• 40:35mice, which already had established
• 40:36tumors,
• 40:37okay, in adult mice. And
• 40:39the effect was that the,
• 40:41it this is systemic disruption.
• 40:43Systemic disruption was well tolerated.
• 40:45They didn't have any any
• 40:46serious problems. They went on
• 40:47to live, you know, full
• 40:48and productive lives. So that
• 40:50says that in normal adult
• 40:51mice, at least, this interaction
• 40:53is not essential.
• 40:55Okay?
• 40:57However,
• 40:57and and and,
• 40:59importantly,
• 41:00no you didn't see any
• 41:01effect on insulin homeostasis.
• 41:02So this shows that, insulin
• 41:05activation of PFK kinase doesn't
• 41:07need input from RAS.
• 41:09But he did see tumor
• 41:10stasis or regression in different,
• 41:12mass models, including each activated
• 41:14EGFR or activated RAS.
• 41:17So from a drug discovery
• 41:18point of view, this is
• 41:19a green light. You know,
• 41:19it's a target which is
• 41:21safe, apparently, in mice at
• 41:22least, doesn't cause the expected
• 41:24talks of an, PI3 kinase
• 41:26inhibitor,
• 41:27and actually has clinical benefit,
• 41:29particularly in combination as he
• 41:30showed.
• 41:31So, this was a, you
• 41:33know, a really good start.
• 41:34And then, in parallel,
• 41:37Sally Levers at,
• 41:38his colleague in in the
• 41:39UK,
• 41:41made the same mutations in
• 41:42flies.
• 41:43So fly and flies, the
• 41:44whole pathway, RAS pathway is
• 41:46identical really to humans.
• 41:48But in so in flies,
• 41:50she made my flies in
• 41:51which this interaction is also
• 41:52disrupted by the same mutations.
• 41:54And these flies that have
• 41:56no interaction between RAS and
• 41:57PI three kinase go on
• 41:58to leave normal, happy, and
• 42:00productive lives.
• 42:01So,
• 42:03except when it comes to
• 42:04tie time to lay eggs
• 42:06so that they are defective
• 42:07at egg laying.
• 42:09So in this paper, Sally
• 42:10concluded that the interaction between
• 42:12RAS and PI three kinase
• 42:13is only necessary in certain
• 42:15situations, which which in flies
• 42:17requires a ton of protein
• 42:18synthesis to make eggs,
• 42:20and in, mice or probably
• 42:21in humans,
• 42:22during development,
• 42:23during angiogenesis,
• 42:25and during malignancy.
• 42:26That's where you need to
• 42:27turbocharge
• 42:28the system to get a
• 42:29sustained activation of PI3 kinase
• 42:31or stronger activation in these,
• 42:34none, in these pathogenic situations.
• 42:38And this is kind of
• 42:39sort of a
• 42:41maybe a dumber way of
• 42:41doing the same experiment in
• 42:42which, we took, cells in
• 42:44which we knocked out all
• 42:45the RAS genes, h n
• 42:46and k and r r
• 42:47s, r r s two,
• 42:48and m r s by
• 42:49adding to cells
• 42:50in which k ras is
• 42:51under
• 42:52a a usable promoter. Get
• 42:53rid of k ras, MAP
• 42:55kinase signaling stops,
• 42:57but p h three kinase
• 42:58signaling does not in in
• 42:59response to multiple different growth
• 43:01factors.
• 43:02So EGF, for example, does
• 43:04not need any RAS protein
• 43:06to act as a AKT,
• 43:07not as IGF one.
• 43:09PDGF
• 43:10actually is better than the
• 43:11absence of RAS proteins because
• 43:12when you knock out RAS
• 43:13proteins, PDGF receptor levels go
• 43:15through the roof, by the
• 43:16way. But, certainly, you don't
• 43:18need any RAS proteins to
• 43:20activate PI three kinase in
• 43:21response to RTKs. Okay?
• 43:24Which is the reciprocal same
• 43:26conclusion that, Julian did.
• 43:29Okay. So, again,
• 43:30verifying this interaction is being
• 43:32really, potentially valuable drug target.
• 43:35So why didn't Julian,
• 43:37or anybody actually,
• 43:39find a drug when they
• 43:40this paper was published in
• 43:41two thousand seven?
• 43:43When we started the RAS
• 43:44list, I called up Julian
• 43:45and said, okay. Well, how
• 43:46are you doing on finding
• 43:47a breaker for this interaction?
• 43:49The phenocop is your mutant
• 43:50mice. And he said, he
• 43:52tried. I collaborated with a
• 43:53company in the UK, a
• 43:54screen for inhibitors.
• 43:56The problem is
• 43:57the binding constant between RAS
• 43:59and kinase is about twenty
• 44:00micromolar.
• 44:02That's really too weak to
• 44:03do a robust screen for
• 44:04interaction, so they barely bind
• 44:06really in the test tube,
• 44:07and there were no crystal
• 44:08structures available to model compounds.
• 44:10So the whole field, they
• 44:11got stuck.
• 44:13And then, quite honestly, the
• 44:14whole field got fixated on
• 44:15RAF and MAP kinase inhibitors
• 44:17and ignored the pathway
• 44:19for years,
• 44:20but it's back.
• 44:22Not just for the stuff.
• 44:24Everybody else is now realizing
• 44:25that PI preconews,
• 44:26is a resistance mechanism to
• 44:28RAF inhibition.
• 44:30Anyway, so that stole the
• 44:31whole field. K?
• 44:34However,
• 44:35by sheer luck,
• 44:37I was, in in Tokyo
• 44:38a while ago. I have
• 44:40a long standing collaboration with,
• 44:42Daichi Sankyo, now oncology group
• 44:44in in Shinagawa
• 44:45in the suburb of Tokyo,
• 44:47where I've been going every
• 44:48year for the last twenty
• 44:49five years. And I was
• 44:50there a few years ago,
• 44:52one of the the people
• 44:53in the in the group
• 44:53there, Kazu, said, hey. The
• 44:55diabetes group at, Daishankyo
• 44:58have found a compound
• 44:59that they were trying to
• 45:00find an
• 45:02orally available insulin memetic compound,
• 45:05and they found a compound
• 45:06that glues RAS to PI
• 45:07three kinase.
• 45:09So
• 45:10so, okay,
• 45:11we'll take it. So,
• 45:13they offered me the compound
• 45:14to, to to test, and
• 45:16then use that for for
• 45:17the reason I'll just show
• 45:18you.
• 45:19So here's the deal. So
• 45:20they would this is insulin
• 45:21signaling. Okay? Insulin activates, insulin
• 45:23receptor, obviously. IRS one recruits
• 45:25PI three kinase here, and
• 45:27then activates,
• 45:29AKT. This promotes GLUT4 translocation,
• 45:31which causes glucose uptake. This
• 45:33is what you need when
• 45:34you need to, take take
• 45:36up glucose. Right?
• 45:39So they wanted to find
• 45:40a drug which would replace
• 45:41insulin so that a diabetic
• 45:43could just take a pill
• 45:44and then
• 45:45take up glucose without having
• 45:46to have an injection.
• 45:48So we get a phenotypic
• 45:49screen for compounds that would
• 45:51do just that, and the
• 45:52screen bait was based on
• 45:53translocation of GLUT four, to
• 45:55the plasma membrane in response
• 45:57to a library of compounds.
• 45:59And amazingly enough, they found
• 46:00compounds which do exactly what
• 46:02they wanted.
• 46:03But these compounds
• 46:04promote, glucose uptake
• 46:06in the absence of insulin
• 46:08with a with a, sort
• 46:09of kinetic curve or those
• 46:10curve very similar to insulin
• 46:12itself.
• 46:13So that looked that looked
• 46:14pretty good.
• 46:15And then they spent years
• 46:16trying to figure out which
• 46:17part of this complicated pathway
• 46:19does the drug act on.
• 46:20And they narrowed it down
• 46:21to PI three kinase alpha,
• 46:23and then they did mass
• 46:24spec on PI three kinase
• 46:25alpha plus or minus compound,
• 46:27and they found,
• 46:28our old friend, ras two,
• 46:31found to, PI two kinase
• 46:33alpha in the IP.
• 46:35And then mass spec also
• 46:36shows some KRAS, but this
• 46:37was a major RAS isoform.
• 46:39So we now know that
• 46:40this compound, t nine two
• 46:42seven,
• 46:43is a molecular glue which
• 46:44sticks RAS proteins in general
• 46:46to PI three kinase alpha
• 46:48and then just jams on,
• 46:50GLUT four translocation and glucose
• 46:52uptake. Okay?
• 46:54So,
• 46:56that's, that's a good start.
• 46:57So we we've working on
• 46:58this compound for a while.
• 46:59And, Frederick, we got all
• 47:00of the compound to test
• 47:01which RAS isoforms that lice
• 47:03best. So we went back
• 47:04to our panel of single
• 47:05isoform mesh that only have
• 47:07one RAS protein,
• 47:09each of these, and then
• 47:10dump the glue onto this,
• 47:12this panel. And you see
• 47:13that each of them activate
• 47:14AKT in response to the
• 47:16glue, but ras two is
• 47:17by far the strongest,
• 47:19interaction.
• 47:20And that correlates with the
• 47:21fact that ras two, is
• 47:22the one of the better
• 47:23binders to the ad glucanase
• 47:24alpha, actually.
• 47:26And if you just dump
• 47:26the compound on the cells,
• 47:27you activate,
• 47:28AKT signaling
• 47:30without affecting
• 47:31phospho ERK, okay, as you'd
• 47:32expect probably.
• 47:34Okay. So it's a glue
• 47:36that activates map kinase p
• 47:37I three kinase signaling through,
• 47:39pretty much any any RAS
• 47:40protein.
• 47:42So we were able to
• 47:43use this compound to solve
• 47:44a structure of KRAS alpha
• 47:46for the first time, and
• 47:48this is published, just a
• 47:49few months ago. So this
• 47:51is KRAS. This is the
• 47:52RAS spani domain where Julian's
• 47:53mutants were made. This is
• 47:54the kind this is the
• 47:55kinase domain, which is distinct
• 47:56from the RAS binding domain,
• 47:58and these are other structural
• 47:59protein parts of the of
• 48:00the protein.
• 48:02So these are the RAS
• 48:03proteins that that bind to
• 48:04PI three kinase alpha. K
• 48:05RAS, as I said, has
• 48:06a IC fifty about twenty
• 48:07micromolar,
• 48:08but our glue is dropped
• 48:10to six nanomolar.
• 48:11So So just that compound
• 48:12alone adds three orders of
• 48:14magnitude in binding.
• 48:16RRAS two binds at four
• 48:17micromolar without glue as MRAS
• 48:19is similar. So Symantje was
• 48:21able to solve the structure
• 48:22of these, these proteins bound
• 48:24to PI two kinase alpha
• 48:25without any glue to help
• 48:26verify that the glue wasn't
• 48:27doing something really weird,
• 48:29and this is not published.
• 48:31And just to come back
• 48:32to the cancer context,
• 48:34the mutant alleles in in
• 48:35in the KRAS are the
• 48:36most common,
• 48:37g twelve d and d
• 48:38twelve v. These are the
• 48:40two RAS isoforms that have
• 48:41the highest affinity for p
• 48:42I three kinase alpha.
• 48:44D and v, the lowest
• 48:45I c lowest k m,
• 48:47k d,
• 48:49compared with, wild type or
• 48:51d twelve c and so
• 48:51on.
• 48:52So it could be coincidence,
• 48:54but it seems likely to
• 48:55me that the reason that
• 48:56d twelve d and v
• 48:57are so prominent in human
• 48:59cancer is because they can
• 49:00activate PI3 kinase as well
• 49:02as MAP kinase, and that
• 49:03combination together is, you know,
• 49:04a very bad combination for
• 49:06the tumor.
• 49:08Anyway, so now we've
• 49:09saw the structure of the
• 49:10of the complex for the
• 49:11first time, and, the moment
• 49:13we saw it, saw the
• 49:14structure, it's obvious that we
• 49:16could convert the glue to
• 49:18a breaker. And that's because
• 49:19the crystal structure of, PI
• 49:21three kinase bound to the
• 49:22blue so is that most
• 49:23of the contact is actually,
• 49:25with the blue compound is
• 49:26is with p I three
• 49:27kinase alpha.
• 49:29So this stabilizes p I
• 49:30three kinase alpha into a
• 49:31RAS binding structure, which takes
• 49:33out a ton of energy
• 49:34of binding and also makes
• 49:35contact with the RAS protein
• 49:36directly. It has more energy
• 49:38that gives you three orders
• 49:39of magnitude in in binding.
• 49:42So this is the blue
• 49:43compound,
• 49:43and so I naively thought
• 49:45it wouldn't take long to
• 49:46make this blue compound into
• 49:48a compound that binds to
• 49:49the same site but now
• 49:50repels
• 49:51KRAS. Okay? So now it's
• 49:52now it's a breaker.
• 49:55Well, the the chemistry group
• 49:56at BridgeBio Oncology with the
• 49:58group of Frederick succeeded in
• 50:00doing this, but it took
• 50:01about two or three years
• 50:02of very heavy med chem
• 50:03and structural biology
• 50:05because the glue does cause
• 50:06a conformational change in the
• 50:07protein, which is basically more
• 50:08complicated,
• 50:09And we had to find
• 50:10a compound that didn't just
• 50:11serve as a a breaker,
• 50:13was also gonna be a
• 50:14good drug that with great
• 50:15PK properties,
• 50:16good biodistribution,
• 50:18etcetera, all the things which
• 50:19academics don't like to think
• 50:20about that really make or
• 50:22break a drug in the
• 50:22clinic.
• 50:23So it took all that
• 50:25time to make this compound.
• 50:26And it turns out when
• 50:27we got the, first structures
• 50:29of our first candidates, we
• 50:31saw it was a cystine
• 50:32close by the compound, so
• 50:33we made it into a
• 50:34covalent compound just to add
• 50:35to binding,
• 50:36just a little linkage here.
• 50:39So this compound binds to
• 50:40p ad hoc kinase alpha
• 50:41covalently and prevents
• 50:43KRAS and other RAS proteins
• 50:45binding.
• 50:46And,
• 50:47this shows inhibition of binding
• 50:49of of g twelve d
• 50:51to p one ten alpha
• 50:52in a pull down assay.
• 50:53It binds tightly enough to
• 50:54do a pull down, and
• 50:55the IC fit is about
• 50:56six nanomolar.
• 50:58And
• 50:59very importantly,
• 51:00this compound, the, the breaker
• 51:02compound,
• 51:03doesn't affect the kinase activity
• 51:05of of PIAT three kinase
• 51:06alpha. Unlike alpelosib,
• 51:07which is the drug approved
• 51:08for treating,
• 51:10patients with p ad free
• 51:10kinase mutations.
• 51:12This directly inhibits the kinase,
• 51:14domain. This prevents RAS activation
• 51:17of the kinase, but doesn't
• 51:18affect the kinase directly. Okay?
• 51:20So
• 51:21the the compound breaks this
• 51:23interaction. It doesn't affect the
• 51:24kinase activity.
• 51:26That's that's really important because
• 51:28it allows, insulin homeostasis to
• 51:30continue as we'll see.
• 51:32So for the first time,
• 51:33we could then ask, well,
• 51:34how many RAS driven cancers
• 51:35or other cancers depend on
• 51:36the RAS PI3 kinase interaction?
• 51:38Previously, it was totally unknown.
• 51:41So,
• 51:42turns out to be a
• 51:43lot.
• 51:44Once they don't care are
• 51:45p ten null cells as
• 51:47you might expect,
• 51:48but tumor cells of mutant
• 51:49KRAS or mutations in PI3
• 51:51kinase alpha, helical, or
• 51:53kinase domain mutations,
• 51:55show pretty good effects,
• 51:57not down to zero.
• 51:58But by far, the most
• 52:00responsive subset
• 52:02of those tumors would amplify
• 52:03H02,
• 52:04H02 new.
• 52:05Okay?
• 52:06That was a shock to
• 52:07us because in the literature,
• 52:09H02 new is definitely known
• 52:10to activate PI3 kinase, like,
• 52:12you know, on fire, but
• 52:13it's been shown previously not
• 52:14to use canonical Ras proteins.
• 52:17Okay? So it's assumed that
• 52:18that interaction is straight binding
• 52:20of p eighty five p
• 52:22eighty five p eighty five
• 52:22you have to kind of
• 52:23straight to,
• 52:24HER2 new or to HER3,
• 52:25its partner, and activation without
• 52:27any need for any Ras
• 52:28proteins. But this shows that
• 52:30this interaction is completely dependent
• 52:32on a RAS protein interacting
• 52:34with PI3 kinase alpha.
• 52:37So that was a shock,
• 52:39and this shows basically a
• 52:40dispatch of that with multiple
• 52:41cell lines to show this
• 52:42is the sort of the
• 52:43breadth of the cell lines
• 52:45we studied.
• 52:46So in cells with looking
• 52:48KRAS,
• 52:49most of the activity does
• 52:51actually come from,
• 52:52from KRAS itself.
• 52:54So we believe that's true.
• 52:56So this is the g
• 52:57twelve c cell line, but
• 52:58we add either the
• 53:00breaker, which is, ten two
• 53:02zero three in red or
• 53:04a pan KRAS compound, the
• 53:05red MET g, six two
• 53:07thirty six, which hits HN
• 53:08and KRAS. So they look
• 53:09the same in response to,
• 53:11AKT. Alpelosib
• 53:13also slams it, but it's
• 53:14much less potent, same in
• 53:16this cell line.
• 53:17So here we see we
• 53:18don't not not all the
• 53:19PI three kinase activity is
• 53:20coming through,
• 53:21k through RAS. Some must
• 53:23be coming from other other
• 53:24sources directly from receptors or
• 53:26from g proteins, but, you
• 53:27know, most of it does.
• 53:29But in in HER2 new
• 53:31cells, most of it comes
• 53:32from something else.
• 53:33So here we do the
• 53:34same experiment. We add the
• 53:35RevMed compounds, and it does
• 53:37have it has no effect
• 53:37on PAKT in this, amplified
• 53:40new cell line, but the
• 53:42breaker
• 53:42really shuts down
• 53:44PAKT very effectively,
• 53:46as well as
• 53:47pretty much. And then b
• 53:48t four seven four,
• 53:50same. There's a bit of
• 53:51a bit of an effect
• 53:52here from wild type HN
• 53:53and KRAS proteins,
• 53:55from,
• 53:55this analysis, but most of
• 53:57it comes from something else.
• 53:59And before you all ask,
• 54:00what is it, RS two?
• 54:03We think it's not. So,
• 54:04we've knocked out ras two,
• 54:06best we can, and ras
• 54:07and mras,
• 54:09and none of those account
• 54:10for this effect.
• 54:11But now we're stuck with
• 54:13another another protein, presumably, not
• 54:15necessarily, but, presumably,
• 54:16binding at that RAS binding
• 54:18domain, which is essential for
• 54:20HER2 new signaling, but we
• 54:21just don't know what the
• 54:22hell it is.
• 54:24So, any suggestions?
• 54:26I'll be delighted.
• 54:27We have a mass spec
• 54:29analysis going on right now
• 54:30with, t I p kinase
• 54:31alpha with Julian's mutants and
• 54:33blue and breaker, etcetera. It
• 54:34might sort of reveal it
• 54:36for the, the last resort
• 54:37of the intellectually bankrupt just
• 54:40to just to a mass
• 54:40spec and hope it solves
• 54:41the problem for us, but
• 54:42we we haven't gotten any
• 54:43good ideas as to what
• 54:44this could be.
• 54:47Anyway,
• 54:48this drug is now in
• 54:49the clinic, so you might
• 54:50say, well, who cares? But,
• 54:51I do.
• 54:54Anyway,
• 54:56so important feature of this
• 54:57drug is that it shuts
• 54:58down signaling
• 55:00in these genotypes that doesn't
• 55:02affect,
• 55:03insulin signaling.
• 55:04So, in a mouse model,
• 55:06we can shut down AKT
• 55:07in tumors just like alpettosib,
• 55:09but in contrast to alpettosib,
• 55:11we don't provoke accumulation of
• 55:12glucose and we don't cause
• 55:14hypoglycemia.
• 55:16For clinical people here will
• 55:17know that patients being treated
• 55:19with alpolis for, eight
• 55:21of the mutant
• 55:23breast cancers,
• 55:25often,
• 55:25go off drug because of
• 55:26the side effects caused by
• 55:27hypoglycemia.
• 55:29So So this compound does
• 55:30not cause hypoglycemia,
• 55:32okay, in the mouse models,
• 55:33even though it can be
• 55:34as effective.
• 55:35So we hope it will
• 55:36be at a better version,
• 55:37a safer version of alpelacib
• 55:39with the same potency, hopefully,
• 55:41and
• 55:42and we have a tox.
• 55:45You also know probably that
• 55:47activation of p I three
• 55:48kinase makes many,
• 55:50drugs,
• 55:51fail to, work effectively in
• 55:53different contexts in in cancer
• 55:55cells. So here, looking at
• 55:56combination of the breaker with
• 55:58actually g twelve c inhibition.
• 56:00And this is I draw
• 56:00attention to this one. This
• 56:01is a model in which
• 56:02we have mutations in PIP
• 56:03one and STK eleven, which
• 56:05make lung cancer cells relatively
• 56:07resistant
• 56:08to G12C inhibitors.
• 56:10So, this is the p
• 56:12I three this is the
• 56:12G12C inhibitor. This is the
• 56:13breaker. Together, we see really
• 56:15nice responses.
• 56:16The same is true in
• 56:17these other models.
• 56:19So that kinda makes sense,
• 56:21but we see that kind
• 56:22of additive or super additive
• 56:24effects in all the models
• 56:25we've tested so far, which
• 56:26include the follistrant, palvaciclib,
• 56:30trastuzumab,
• 56:32and even chemotherapy
• 56:33where
• 56:34breaker alone, has an effect,
• 56:36but in combination with irinotecan,
• 56:38we see really substantial effects.
• 56:41So we don't really understand
• 56:43why this would be. We
• 56:44know, generally, the PIAT mechanics
• 56:46makes cells more difficult to
• 56:47kill.
• 56:48That's been known forever. We
• 56:49don't really know what the
• 56:50molecular mechanism is, and we
• 56:51don't know that this will
• 56:52translate in the clinic. Obviously,
• 56:53this is all,
• 56:55in vitro model. So,
• 56:57but we will see because
• 56:58this compound is now moving
• 56:59through dose escalation in phase
• 57:00one clinical trials. And once
• 57:02we reach a certain dose,
• 57:03we can then test it
• 57:04in combination,
• 57:05and we'll definitely test it
• 57:06in combination with t twelve
• 57:07c and other KRAS drugs
• 57:09plus,
• 57:10these other drugs, which I've
• 57:11just mentioned.
• 57:13So, you know, fingers crossed
• 57:14for us at least. We'll
• 57:15see on the patients, see
• 57:17what happens.
• 57:18So this kind of summarizes
• 57:20the very complicated story of
• 57:21PI three kinase. So insulin
• 57:23receptor activates,
• 57:25glucose uptake without any RAS
• 57:26involvement that we know of.
• 57:28Okay? So,
• 57:29breaker doesn't affect that.
• 57:31Most normal signaling doesn't need
• 57:33RAS either as I showed
• 57:34you, so it don't affect
• 57:35normal RTK signaling. But under
• 57:37these circumstances,
• 57:39we do need a RAS
• 57:40protein
• 57:41either to turbocharge
• 57:42the reaction, as I said,
• 57:44or some other mechanism yet
• 57:45to be determined.
• 57:46But,
• 57:48we're we are working on
• 57:49this mechanism, but this is
• 57:50where the the breaker compound
• 57:52could work or directly on
• 57:54tumors driven by KRAS where
• 57:55the mutant alleles
• 57:56enable them to bind to
• 57:57PI3 kinase alpha, which they
• 57:58don't normally do, and we
• 58:00can interrupt that with the
• 58:01breaker.
• 58:03And with that, I will
• 58:05thank the people of the
• 58:06team from National Lab, and
• 58:08Lawrence Livermore National Lab. And
• 58:09my lab, part of work
• 58:11here in California.
• 58:13So, and with that, I'll
• 58:14I'll stop and be happy
• 58:15to take questions. Thank you
• 58:16very much. Thank you.
• 58:26Yes.
• 58:54Who's that? Yeah. That's that's
• 58:55a great question. Yeah.
• 58:58Sorry?
• 58:59May not be the direct
• 59:01one.
• 59:03Yeah. Correct. Yeah. We we
• 59:04we don't really know how
• 59:05it's working. We don't see
• 59:06we we do see that
• 59:07the,
• 59:09ras two, you know, binds
• 59:11to,
• 59:12p one ten in those
• 59:13cells and are most likely
• 59:15still engaged with the receptor.
• 59:17The drug interrupts that.
• 59:20Drugs that inhibit HOTO new
• 59:21make that all fall apart,
• 59:22so it's RTK dependent
• 59:24binding. But beyond that, we
• 59:25haven't got a handle on
• 59:26the mechanism. But, you know,
• 59:27obviously,
• 59:28you might have much better
• 59:29ideas than I do about
• 59:30how to deal with that.
• 59:31So it's, it's very, very
• 59:33strange. And and we only
• 59:34see the really dramatic effect
• 59:36of dependence on cells that
• 59:37amplified,
• 59:38HOTA new and or her
• 59:39three.
• 59:40But don't think it happens
• 59:41during normal signaling. It's a
• 59:43gain of function, but what
• 59:44it is, we don't know.
• 59:46Yeah. And we haven't we
• 59:47haven't really gone through all
• 59:49RTKs either really, but, honestly,
• 59:50just in that panel of
• 59:51cancer cells that I showed
• 59:52you where Houton new jumps
• 59:54out. But, it simulates that
• 59:55the EGFR mutations also do
• 59:57respond, but not as well
• 59:58as the Houton.
• 01:00:10Yeah. Very good very good
• 01:00:11thought. Yeah. Yeah. We should
• 01:00:12do that.
• 01:00:14Yeah. Maybe we should what
• 01:00:15I'd like to have is
• 01:00:16a simple system with, you
• 01:00:17know, like a like a
• 01:00:18math where you put things
• 01:00:19back in again and you
• 01:00:19can take them away again.
• 01:00:20But this
• 01:00:22is now we're working in
• 01:00:22cancer cell lines. It just
• 01:00:23makes it more complicated. But
• 01:00:24yeah.
• 01:00:27Yeah.
• 01:00:28I will we'll talk about
• 01:00:30it later maybe.
• 01:00:31Thank you.
• 01:00:33Hi.
• 01:00:35Some cancers don't have obvious
• 01:00:37genetic drivers of PI three
• 01:00:38kinase, you know, in the
• 01:00:39pathway. I was wondering if
• 01:00:41this with the screening of
• 01:00:42cell lines, like, suffer or
• 01:00:43to do that,
• 01:00:44you probably find some surprises
• 01:00:46where the breakers have really
• 01:00:48substantial effects,
• 01:00:50in the lines
• 01:00:51that don't have an obvious
• 01:00:52driver.
• 01:00:55Second really is, other RAS
• 01:00:56and deals that's not on
• 01:00:58person. NRAS put on sixty
• 01:00:59one. Are you seeing any
• 01:01:01activity with this drug? There's
• 01:01:02needs in
• 01:01:05cancers that don't have.
• 01:01:07Right.
• 01:01:09Well, second part of the
• 01:01:10question, yeah, we do see
• 01:01:11effects on,
• 01:01:12cell lines with h mutant
• 01:01:14or mutant. Same it's the
• 01:01:16same binding site version. There
• 01:01:17are actually cell lines with
• 01:01:18mutant two also, which it
• 01:01:20also blocks. But here's here's
• 01:01:21all of those.
• 01:01:22Okay.
• 01:01:24First question,
• 01:01:25there's a lot of cell
• 01:01:26lines that,
• 01:01:27that we see responses that
• 01:01:28have wild type p I
• 01:01:29three kinase for sure. Right?
• 01:01:30That's we know that.
• 01:01:33How many of those don't
• 01:01:34have known drivers that could
• 01:01:36affect p I three kinase?
• 01:01:38I'm not so sure because
• 01:01:39RTKs definitely can activate p
• 01:01:41I three kinase directly. Right?
• 01:01:43So, Like,
• 01:01:44blocks and things like that.
• 01:01:45I mean, that wouldn't be
• 01:01:47expected, but Yeah.
• 01:01:49Surprises.
• 01:01:49Right.
• 01:01:51I think
• 01:01:52lines are completely resistant, but
• 01:01:53I'm not hundred percent sure
• 01:01:55about that. Very good, very
• 01:01:56good point. I need to
• 01:01:56come back and check.
• 01:02:06Okay.
• 01:02:08Any questions
• 01:02:10from an AI component?
• 01:02:12Does the greater change of
• 01:02:13the PI distribution
• 01:02:15on the memory
• 01:02:16ever yet created
• 01:02:18products?
• 01:02:20Their their localization
• 01:02:21is not their localization signal.
• 01:02:24Yes.
• 01:02:26That's one of the that's
• 01:02:27one of the possibilities on
• 01:02:28our list. We haven't been
• 01:02:29able to address that yet.
• 01:02:30In fact,
• 01:02:32don't really know how RAS
• 01:02:34cooperates with,
• 01:02:35PIP kinase anyway, but one
• 01:02:37model is that,
• 01:02:38KRAS particularly pluses PIP two
• 01:02:40around itself in the membrane.
• 01:02:42It's like a it's like
• 01:02:43a little puddle of PIP
• 01:02:44two induced by RAS itself.
• 01:02:45PIP two obviously is a
• 01:02:46substrate for p I three
• 01:02:47kinase.
• 01:02:48So it could actually be
• 01:02:49redistribution of lipids in the
• 01:02:51membrane to give the kinase
• 01:02:52more substrate.
• 01:02:53A change in signaling today.
• 01:02:56Yeah. Yeah. Yeah. All these
• 01:02:57signaling proteins have multiple c
• 01:02:59two domains and second and
• 01:03:00so on. Yeah. Yeah. Those
• 01:03:02are questions which might well
• 01:03:03be true, but it kinda
• 01:03:04makes me nauseous to think
• 01:03:05about. It's so difficult to
• 01:03:06do the experiment,
• 01:03:07but that might well be
• 01:03:08the case. Yeah. We don't
• 01:03:09really understand what the complex
• 01:03:11looks like, you know, before
• 01:03:12or after drug yet,
• 01:03:13but we hope to solve
• 01:03:14that out. Great great point.
• 01:03:18Yeah. Alright.
• 01:03:19No. We haven't used AI
• 01:03:20yet. Actually, we have to
• 01:03:21develop a drug.
• 01:03:22So,
• 01:03:24you said it took two
• 01:03:25or three years to figure
• 01:03:26out, structure for the breaker.
• 01:03:28Could you share,
• 01:03:29what were maybe some surprises
• 01:03:31in that process that may
• 01:03:33have led to it.
• 01:03:38Yeah. Well, actually, that's kinda
• 01:03:39short in a way. I
• 01:03:40mean
• 01:03:42so getting a hit is
• 01:03:43easy, but getting a drug
• 01:03:44that's that has a PK
• 01:03:45properties that are compatible with
• 01:03:46orally available
• 01:03:47drug binding and not, you
• 01:03:49know, getting good target exposure
• 01:03:50and so on. That just
• 01:03:51takes a lot of empirical
• 01:03:52testing in mice, essentially. A
• 01:03:54lot of those stuff you
• 01:03:55can't do you can't predict,
• 01:03:57you know, a lot of
• 01:03:58those factors. Yes. A lot
• 01:03:59of it is just plowing
• 01:04:00through mice looking for things
• 01:04:01that have the right absorption,
• 01:04:03serum binding, distribution in tissues,
• 01:04:05and so on. A lot
• 01:04:05of it's just grunt work
• 01:04:06really, but honestly.
• 01:04:08And we do have a
• 01:04:08few surprises which are more
• 01:04:09interesting, and that is there
• 01:04:10are some compounds
• 01:04:11that affect our RAS two
• 01:04:13binding, but don't affect,
• 01:04:15the and other drugs that
• 01:04:17don't. And even though they
• 01:04:17all bind, they all affect
• 01:04:19KRAS, but only some of
• 01:04:20them affect RAS two even
• 01:04:21though the sequence is almost
• 01:04:22identical.
• 01:04:23So So the compounds do
• 01:04:24have different spectra of activity
• 01:04:26against different RAS proteins despite
• 01:04:28being so similar, which I
• 01:04:29didn't really expect, but that's
• 01:04:31become a bit of an
• 01:04:31issue. But,
• 01:04:33yeah, mostly it was just
• 01:04:35getting a compound that have
• 01:04:36the right affinity, the right
• 01:04:37PK properties, binding,
• 01:04:39safe, clean, high potency.
• 01:04:41That's hard.
• 01:04:43That's where AI may will
• 01:04:44make a difference in the
• 01:04:45future. If we can predict
• 01:04:46how to make drugs where
• 01:04:47I wish,
• 01:04:48a better add new properties,
• 01:04:50that would be a huge
• 01:04:50step forward.
• 01:05:40Directly possible. Yes.
• 01:05:42Yeah. I'm quite I'm a
• 01:05:43bit worried about that.
• 01:05:46Yeah. We think in some
• 01:05:47context that the blue compound
• 01:05:49on, yeah, the bridge top
• 01:05:50out, for example,
• 01:05:51combines RS two, cause a
• 01:05:53conformational change in p one
• 01:05:55ten, which could recruit other
• 01:05:56proteins, even chaperone proteins, and
• 01:05:58that could get in the
• 01:05:59way of the complex. So
• 01:06:00there there could be other
• 01:06:01players involved that we haven't
• 01:06:02yet seen. That's definitely a
• 01:06:04concern.
• 01:06:05And glue compound, though, works
• 01:06:06in in the test tube
• 01:06:07with recombinant proteins. So the
• 01:06:08glue, I think, is clean,
• 01:06:09but the breaker is not
• 01:06:10quite so clear what's going
• 01:06:12on. You know?
• 01:06:15Thank you.
• 01:06:16Okay.
• 01:06:19Thank you for your attention.
• 01:06:20Thank you.