Pathology Grand Rounds, February 1, 2024 - William G. Kaelin, Jr., MD

February 02, 2024

Information

Yale Pathology Grand Rounds, February 1, 2024: William G. Kaelin, Jr., MD, presents on: "von Hippel-Lindau Disease: Insights into Oxygen-Sensing, Cancer, and Drugging the Undruggable."

ID11248

To CiteDCA Citation Guide

00:02All right, so it's just about 12:30.
00:04We'll get things started.
00:05For those who don't know,
00:06me, I'm James Elia.
00:07I'm a graduate student in
00:08the pathology department,
00:10and Doctor Caitlin was our choice
00:12for graduate student grand rounds.
00:15So each year, the graduate students
00:16select one Grand round speaker,
00:18and we're very happy to have him here.
00:20So I just wanted to give
00:21a brief introduction.
00:22So Doctor William G Caitlin
00:24Junior received his Bachelor of
00:25Science and Chemistry Mathematics
00:27from Duke University OR.
00:28He also began his research career when
00:31Doctor Kalin questioned his research
00:32mentors assumptions on a project.
00:34The mentor wrote that Mister Kalin
00:36appears to be a bright young man whose
00:39future lies outside of the laboratory.
00:42Dr.
00:42Kalin remained at Duke
00:43for his medical degree,
00:44where he first read about tumor
00:46angiogenesis and the highly
00:47vascular tumors of von Hippel,
00:49Lindau, or VHL disease,
00:50and learn more about the rapidly
00:53developing field of molecular oncology.
00:56Doctor Caitlin completed his
00:57internship and residency in
00:58internal medicine at Johns Hopkins,
01:00where he served as an Assistant
01:01Chief Chief of Service,
01:03learning more about VHL disease,
01:05in part so that you could grill
01:06trainees who questions authority,
01:08and in part because of the
01:09budding hypothesis on the VHL
01:11gene's role in oxygen sensing.
01:13Dr.
01:13Caitlin then joined Dana Farber
01:15as a Medical Oncology Fellow,
01:16where he determined the T and E1A
01:18binding site of the RV protein
01:20and helped identify E2F as a
01:22binding partner of the RV protein.
01:24He began his faculty position
01:26shortly thereafter.
01:27Just down the hall,
01:28his research focus turned to VHL,
01:30the topic of today's grand rounds.
01:32So on behalf of the pathology
01:33graduate students,
01:34our department and Yale School of Medicine,
01:36please welcome Doctor Bill Kalin.
01:45Well, thank you.
01:45Thank you, James for the very nice
01:47introduction and thank you to all the
01:49graduate students who invited me.
01:51I was recently I gave a lecture at
01:54the Cancer Center in South Dakota and
01:56at dinner the faculty confessed that
01:59they said we knew if the faculty invited you,
02:02you weren't coming to South Dakota.
02:03But if the graduate students invited you,
02:06you are on the next plane when there's
02:08some some element of truth that but I
02:10would have come whether the faculty or
02:12the graduate students at Yale invited me.
02:14So it's it's, it's quite fashionable
02:15now to have these disclosure slides,
02:17but often I find that people leave
02:19them up for literally 2 seconds.
02:21So it's like subliminal.
02:22So that's why I sort of leave it up.
02:23So you all digested this.
02:25So here are the people who've
02:27worked on Von HIPAA Lindau disease
02:30over the years of my laboratory,
02:31including your very own Chin Yan.
02:34And here's a list of some of our
02:36collaborators who have helped us with some
02:38of the work I'm going to describe today,
02:40including your very own David Brown.
02:42I might mention we're also
02:43collaborating with Joe Contessa,
02:45but in the interest of time,
02:45I won't be discussing that work today.
02:51So I went to high school at Rodger
02:53Ludlow High School at Fairfield, CT.
02:55I was rejected from Yale, but I I I did.
03:00I did get even because both my children
03:02actually attended Yale as undergraduates,
03:04including my son Pierre Tripp,
03:06who was on your national
03:07championship squash team.
03:09So obviously he got his athletic
03:11ability from his mother.
03:12So he was class of 2018, and I'll
03:15show you a picture of my daughter later on.
03:20Moreover, at the time I won the Nobel Prize,
03:23I was dating a very lovely woman,
03:25shown here named Karen Cantor,
03:28entering the hall at the
03:30arm of the Prime Minister.
03:32And there may be a handful of
03:34you in the room.
03:34Actually, no.
03:35She's the daughter of Fred Cantor,
03:37who was a beloved member of
03:39your faculty for many years.
03:41So all sorts of Yale connections.
03:43So when I was younger,
03:46as you heard,
03:46I went to medical school.
03:47I was so convinced I was
03:48going to be a clinical Dr.
03:49I spent an extra year at Johns
03:51Hopkins as the chief medical resident.
03:54But the person who changed my
03:56life was David Livingston.
03:57So after finishing my residency,
03:59I came to the Dana Farber in 1987
04:01to be a clinical oncologist.
04:04And then I had the good fortune
04:05of being a postdoc in his lab at a
04:07very good time to be in David's lab.
04:09Although frankly any opportunity to
04:10train with David would have been a good time.
04:12But the field was moving very quickly.
04:15The RB gene had just been isolated
04:17and I was given the opportunity to
04:20work on the RB tumor Spicer gene.
04:22So it was really David who sculpted me,
04:24crafted me, whatever you want to say,
04:26into a scientist.
04:27So I owe my scientific career to David.
04:29Sadly, we lost him about two years
04:32ago Now I thought what I would do
04:34is for the benefit of the young
04:36people and because you're supposed to
04:37dance with the one who brought you.
04:39I do have about 10 or 15
04:41minutes of historical review,
04:42so I won't be offended if some
04:43of my colleagues are checking
04:44their emails on their phones,
04:46and then we will move on to
04:48three unpublished stories.
04:49So of course,
04:49one of the most important decisions
04:51you have to make when you're starting
04:53your laboratory is what to work on.
04:55And fortunately for me,
04:56shortly after I started my own laboratory,
05:00this paper crossed my desk,
05:01which was the cloning of the gene that,
05:03when mutated,
05:04gives rise to the hereditary cancer syndrome.
05:06Von Hippel, Lindau disease.
05:07By the way,
05:08I left off one more Yale connection.
05:10So one of the few places that
05:12recruited me after my postdoc
05:14with David Livingstone was Yale.
05:16And I actually saw where my lab was
05:18going to be in the Boyer Center,
05:19which I walked past today.
05:21But the then chairman of medicine at
05:23Cadman and I didn't exactly have a
05:25meeting of the minds in terms of how
05:27you nurture physician scientists.
05:29So spices say that was a door
05:31I didn't walk through.
05:33So I joined the faculty at the Dana Farber.
05:35This paper crosses my desk and I
05:36knew from my clinical training this
05:38would be really be an interesting
05:40gene to study.
05:40Moreover,
05:41I thought from my work on
05:42the retinoblastoma gene,
05:43I was well positioned to study
05:45another tumor suppressor gene,
05:46namely the VHL tumor suppressor gene.
05:49So this is a disease that was
05:51described about 100 years ago.
05:52It affects about one in 35,000
05:54people around the world.
05:56And as you can see,
05:57these are some kindreds that
05:58have been followed at the NCI
06:00with von Hippelindau disease.
06:01And of course the filled circles and
06:04squares are affected individuals
06:06and the classical tumors seen in the
06:07syndrome are clear cell renal cell carcinoma,
06:10which is by far the most common
06:11form of kidney cancer.
06:13Blood vessel tumors called
06:14hemangioblastomas of the eye,
06:15brain and spinal cord and neuro
06:18Crest tumors called paragen gliomas,
06:20which when they arise in the adrenal
06:22gland are referred to as theocromocytomas.
06:24Now you can see that clinically this
06:26looks like an autosomal dominant disorder,
06:28but actually at the molecular
06:29level this is actually an autosomal
06:31it's actually caused by a loss
06:33of function mutation.
06:34So again,
06:34just to make sure everyone saw the same page,
06:36people with about HIPAA Lindow disease
06:38have have inherited A defective version
06:40of the VHL gene from mom or dad.
06:42In this schematic,
06:43it's the maternal copy.
06:44But they're initially OK because
06:46they have the remaining wild type
06:47allele and there's no evidence for
06:49HAPLO insufficiency for this gene.
06:51But unfortunately, if you're born like this,
06:52you have a 90% chance that one
06:54or more susceptible
06:55cells in your body will spontaneously lose
06:58or mutate the remaining wild type copy.
06:59And that's the cell that
07:00will go on to form a tumor.
07:02And as you would predict from
07:04the knowledge that germline VHL
07:05mutations predispose to for example
07:07clear cell renal cell carcinoma.
07:08If you now look at non hereditary
07:09clear cell renal cell carcinomas,
07:11you again see that both the maternal
07:13and the paternal copies of the VHL
07:15gene are frequently mutated or lost.
07:16But here both mutational events or hits
07:19if you will have occurred somatically
07:21in contrast to VHL disease where the
07:23first hit has occurred in the germline.
07:27So I knew from my clinical training that
07:29the tumors seen in von Hipolindo disease
07:30are notoriously rich in blood vessels,
07:32and that's because they frequently over
07:34produce BEGF and then they also occasionally
07:36cause excess red blood cell production,
07:38and that's because they sometimes
07:40ectopically produce erythropoietin.
07:42And what VEGF and erythropoietin have
07:44in common is that they're normally
07:46induced when cells or tissues
07:47are not getting enough oxygen.
07:48So that was the clue that by
07:50studying the VHL gene we might learn
07:52something about oxygen sensing.
07:54Now VEGF and EPO are under the control
07:57of a heterodemeric transcription
07:59factor called hypoxia decibal
08:00factor or HIP for short.
08:02And we knew from the work
08:04of multiple laboratories,
08:04including the work of my fellow
08:07laureates factors Semenza and Radcliffe
08:10that the alpha subunit is normally
08:12degraded when oxygen is available,
08:15hence hypoxia inducible factor.
08:18Whereas HIP beta,
08:18which is also known as RNT,
08:20is constitutively stable and
08:21over the course of time,
08:24our lab and others showed that the
08:25VHL protein is part of a so-called
08:27GABIC one and ligase complex that
08:29binds directly to hip alpha and
08:30targets it for proteasomal degradation
08:32provided oxidant is present.
08:34Whereas when oxygen levels are low
08:36or the VHL protein is defective,
08:38such as in kidney cancer,
08:40now HIP alpha can accumulate,
08:41dimerize with its partner protein,
08:43Orient and activate various and
08:45sundry genes such as VEGF and
08:47on occasion erythropoietin.
08:48And also point out that there are
08:51two mutational hotspots if you look
08:52at BHL families around the world,
08:54one is the alpha domain which
08:56recruits the rest of the ubiquin
08:57and ligase and the other is the beta
08:59domain which we showed is the actual
09:01docking site for the substrate.
09:02Now this of course begs the question,
09:04how does the VHL protein know,
09:06if you will whether oxygen is or
09:07is not available and hence whether
09:09it should or should not destroy
09:11HIP or mark HIP for destruction.
09:13And work That we did,
09:14and Sir Peter Ratcliff did working
09:16independently in parallel,
09:17showed that in the presence of
09:19oxygen one of two conserve prolial
09:21residues gets hydroxylated in hip
09:24alpha and this then generates a
09:26high affinity VHL binding site.
09:28The the enzymes that do the work
09:31here are variably called the Egolan
09:34or PhD Prol hydroxylases.
09:36They split molecular oxygen and
09:38use one of
09:39the oxidant atoms to hydroxylate the hip.
09:43They also require reduced iron,
09:45which explains an old observation
09:47that iron chelators when given to
09:48cells and culture will mimic or
09:50stimulate A hypoxic like response.
09:52They also require a cofactor which
09:55is variably called 2 oxygoodrate or
09:57alpha ketoglydrate if you prefer,
09:59which gets decarboxylated to succinate.
10:02Not shown here, however,
10:03is one other critical piece of the puzzle,
10:05which is These enzymes have very
10:07low oxygen affinities and hence
10:09they're poised to sense oxygen in
10:11a physiologically relevant range.
10:13That's in contrast to, for example,
10:14the collagen pro hydroxylases you
10:16might have studied in college
10:18and college biochemistry.
10:20The collagen pro hydroxylases have
10:22extremely high oxygen affinities.
10:23You'd have to be virtually anoxic
10:25before the collagen pro hydroxylases
10:27would become inactive.
10:29So these enzymes I just introduced to you,
10:32which I'll stick with the original and
10:33the meclids, are the EGELEN enzymes.
10:35So these are the pro hydroxylases
10:38that modify hip.
10:39But you can see these enzymes are
10:41part of a much larger superfamily,
10:43so-called 2 oxyglate dependent de
10:45oxygenases which includes many of the
10:49so-called Jumanji C histone demethylases
10:52as well as the Ted enzymes that are
10:56implicated in DNA demethylation.
10:58So let's go back to the Jumanji C proteins.
11:01So the Jumanji C histone methylases
11:04as first shown by Yi Zhang use
11:08a fairly similar chemistry,
11:09but what they do is they hydroxylate
11:11the histone methyl group which is then
11:14unstable and given off as formaldehyde.
11:16Now this raised the question in our
11:19minds and other people's minds,
11:21are these enzymes more like
11:23the enzymes that modify HIP?
11:25Meaning are they going to be very oxygen
11:27sensitive or are they going to be more
11:29like the collagen pro hydroxy ACES
11:30and be relatively oxygen insensitive?
11:32And it turns out you can find
11:34many examples of both.
11:35Some of the histamine methylases
11:37actually turn out to be exquisitely
11:40sensitive to oxygen availability.
11:41One example is the one shown here,
11:44KDM 6A, otherwise known as UTX.
11:47So this enzyme is every bit as oxygen
11:49sensitive as the enzymes that modify HIP,
11:51and so this provides A surprisingly direct
11:54linkage between oxygen availability
11:55and certain epigenetic marks.
11:57And we think this has relevance to
11:59the role of hypoxia, for example,
12:02during embryologic development in
12:04certain stem cell niches as well
12:07as within tumors.
12:08Now the other reason I showed you
12:10this superfamily of enzymes is,
12:12I must say I'm a bit of what we
12:14say on the wards.
12:15I'm a lumper rather than a splitter.
12:17I'm always looking for sort of
12:19common themes and commonalities.
12:20So I'm always looking for sort
12:22of unifying hypothesis.
12:23So now I want to introduce Otto Warburg,
12:25who you may know in the middle of
12:28the last century argued that altered
12:30metabolism was a cause of cancer.
12:32But then over the decades the debate was,
12:34well,
12:34is ultimate metabolism A cause of
12:37cancer a consequence of cancer,
12:39both or neither?
12:40And what what Warburg lacked was
12:42sort of a genetic
12:43smoking gun that alter
12:45metabolism could cause cancer.
12:46But we now know that some cancers have
12:49inactivating mutations and succinate
12:51to hydrogenase or fumarate hydratase,
12:53which lead to the accumulation of
12:55succinate and fumarate respectively.
12:57And it's clear these chemicals can inhibit
12:59these various two OG dependent enzymes.
13:01In fact you might have noticed that
13:03succinate was the product of the reaction.
13:05And then we learned from Burt Vogelstein
13:08and Haiyan and others that certain
13:10tumors such as certain brain tumors and
13:12leukemias have neomorphic mutations
13:14and isositrate dehydrogenase one or
13:16two which allow these enzymes to make
13:19milli molar amounts of a so-called
13:21onco metabolite 2 hydroxychlorate,
13:23which also can inhibit these various enzymes.
13:26So we could actually do it.
13:27Now an entire research seminar about,
13:29well, which of the enzymes are
13:31critical in which setting with
13:32which mutation and which cancer.
13:33But I'm not going to do that to you again
13:35because I'm a lumper rather than a splitter.
13:36So this is my lumpers view of how we get
13:39cancer from these various mutations.
13:41But I will say that early on the
13:44naysayers and for the students
13:45always beware of naysayers.
13:46You can always think of a reason
13:47why something's going to fail.
13:48It's not going to work.
13:50So the naysayers argue that even
13:51if you could make a drug that would
13:54prevent the production of two HG,
13:57it would not be helpful because many
13:58of these enzymes, as I just told you,
14:00are involved in epigenetic reprogramming.
14:03And the argument was those epigenetic
14:05marks would be relatively stable and
14:07would not reverse in a clinically
14:08useful time scale.
14:09So to sort of tackle that,
14:11Julie Lostman, when she was in my lab,
14:13decided to test whether there was
14:16an ongoing requirement for two HG
14:18and IDH mutant leukemias.
14:20And so she created a model based
14:22on a leukemic cell line called TF1
14:24where she created isogenic cells
14:26that were wild type for IDH one or
14:29had this canonical IDH 1 mutants.
14:30She then measured 2 HG levels and as
14:32expected when she put in the IDH one mutant,
14:34now you had milli molar amounts of two HG.
14:37And then she treated these cells
14:39with a tool compound developed by
14:41Agios that blocks two HG production
14:43and you see the two HG go down.
14:46More importantly, perhaps,
14:48she'd also shown that the cells
14:50with the wild type IDH One,
14:53despite being leukemic,
14:55can't proliferate in the absence
14:57of cytokines.
14:57So that's what's shown in red.
14:59However,
15:00she found that if the cells had
15:02this canonical IDH One mutant now,
15:03these cells could proliferate well,
15:06even without cytokines.
15:07It's sort of a hallmark of transformation.
15:09And now she's ready to treat the
15:11cells with the inhibitor and the
15:13cells stop growing.
15:15So for the students,
15:15this is one of the most dangerous
15:16results you're ever going to get.
15:18Because this is Exactly what
15:19you were hoping for.
15:21And that's when you get really lazy, right?
15:22So you want to call the editor.
15:24This is figure six. Look at me
15:26high fives up and down the hallway.
15:28But this is a dangerous experiment.
15:31A because I just said it's
15:32one we were looking for.
15:34But it's also dangerous because these are
15:35what some people would call down assays.
15:37So you add a chemical,
15:39two HG goes down, You add a chemical.
15:41The cells stop proliferating down and down.
15:44So if you were a cynic you would
15:46say this is just another poison that
15:48they sent you after that MTA that you
15:50negotiated for a year and it's just kill.
15:52The cells are just really sick.
15:53And since and this you could
15:55have measured any metabolite,
15:56but you just measure 2 HG and Gee you
15:58know this is true, true and unrelated.
16:00These cells are just dying, who cares?
16:02So you need a better experiment.
16:04So you got to push yourself
16:05to do a better experiment.
16:06So the better experiment was rather
16:08than give these cells the IDH mutant,
16:10she gave the cells a cell membrane
16:12permeable version of R2 HG that
16:14enters the cells and then gets trapped
16:17at tumor relevant concentrations.
16:18So now if you give the cells two HG,
16:21they're completely insensitive to a
16:23drug that prevents 2 HG production.
16:25And that's shown here is if you
16:26give this Ester to these cells,
16:28they start growing again.
16:29So now you really know you're on target,
16:30which is really, really important.
16:33So thankfully,
16:33because for this and other reasons,
16:35these drugs move forward and I'm happy
16:37to report we now have an inhibitor of
16:39both mutinite H1 and mutinite H2 for
16:42the treatment of IDH mutant leukemias.
16:44I wouldn't say these are home runs,
16:46although some patients certainly
16:47get a lot of benefit.
16:48But clearly there's heterogeneity
16:50and we have to understand
16:51that heterogeneity further.
16:53OK.
16:53So now returning to this slide,
16:55I already showed you again,
16:57it was certainly plausible that the
16:59tumors seen in VHL disease were all about
17:01hip and were all about an excessive hip.
17:04But another take home for the students
17:07is correlation plus plausibility
17:09is not proof of causation.
17:11We often get kind of lazy at that last step.
17:13OK, I can imagine it's all about hip,
17:15hip does a lot of things.
17:16Maybe that promotes tumor growth,
17:17but you have to do sort of the killer
17:20experiments, if you, if you can,
17:21just try to establish what you
17:23know what's really going on here.
17:24And again,
17:25that often requires sort of
17:27genetic paradigms.
17:28So we had shown some time ago that if
17:29you restore the function of VHL and a
17:32VHL defective wieno cell carcinoma,
17:33they lose the ability to form
17:36tumors and nude mice.
17:37So Kichi Kondo took these cells and
17:40introduced into these cells a version
17:43of HIP 2A that can't be recognized
17:45by BHL because the Prolil sites have
17:47been converted to alanine and these
17:49now once again form tumors and and
17:51a reciprocal set of experiments he
17:53took these cells and eliminated HIP
17:552A then with hairpin technology we
17:57now do this with CRISPR and these
17:59cells lose the ability to form tumors.
18:01And I will tell you,
18:02when we did the same experiments with
18:04the canonical member of the family,
18:05HIP One, we got the opposite result.
18:08So much to our surprise.
18:09On one level it is hip,
18:10but it's really hip to the less
18:14well studied member of the family.
18:16Now, not shown here are the kind
18:19of controls I learned about at
18:22the knee of David Livingston,
18:23who taught me how to be a scientist.
18:25So for the students,
18:26this is your required reading.
18:28I'm going to come back to that one more time.
18:29But this is the required reading
18:31for the kinds of controls you
18:32might want to do if you're going
18:33to do Target validation studies.
18:35Because I can tell you,
18:36our colleagues in industry have
18:38reported that 50 to 90% of the time
18:41when they go to replicate results from
18:43academic labs related to target validation,
18:45either the results are not
18:47reproducible or they're not robust,
18:49meaning they're only true under
18:51very narrow set of conditions.
18:52OK. So I think we have to all
18:54do a little bit better.
18:54I'll come back to that at the end.
18:57OK. So what can we do about this?
18:58Well, we certainly knew by the 90s
19:01that VEGF was under hip control.
19:03And thankfully,
19:03a number of companies were
19:05making VEGF inhibitors.
19:06And we argued if these inhibitors were
19:07going to work in any solid tumor,
19:09they should work in VHL,
19:10defective kidney tumors, for example.
19:13And that turned out to be true.
19:14So we're now up to 8 or 9, I think,
19:16VEGF inhibitors approved for
19:18the treatment of kidney cancer.
19:19So that's the good news.
19:20But not every patient with kidney
19:22cancer will respond to these drugs,
19:23and even those that do will
19:25eventually progress,
19:26although thankfully sometimes
19:27it takes many years.
19:28So how can we do better?
19:30Well, I hope some of the students
19:31are saying to themselves,
19:32kale and you dummy, rather than
19:34target anyone hip responsive gene,
19:36you should target HIP.
19:37And if you believe your own data,
19:39you should target hip too.
19:40Wouldn't that be nice?
19:41But again,
19:42the naysayers come out of the woodwork
19:43and they start saying you can't do
19:45that because HIP too is adna binding
19:48sequence specific transcription factor.
19:50It's not a member of the steroid
19:51hormone receptor family.
19:52So it doesn't have a natural pocket
19:53for a drug like small molecule.
19:55So forget about it.
19:57But fortunately,
19:592 very talented scientists,
20:00Rick Bruick and Kevin Gardner,
20:02who were then at UT Southwestern,
20:04identified a potentially druggable pocket
20:06in the so-called Pass B domain of HIP 2A.
20:10And they did high throughput screens
20:11at UT Southwestern and identified
20:13chemicals that could bind to this pocket.
20:16And in so doing and do some allosteric
20:18change in HIF to alpha such that
20:19it could no longer bind to aren't
20:21and hits no longer bind to DNA.
20:25Now these chemicals are what
20:27some of my drug hunter
20:29friends called publication grade
20:31but not pharmaceutical grade.
20:32They had too many blemishes and
20:34warts to really be drugs for people.
20:37But fortunately these chemicals
20:38were out licensed to a biotech
20:41company in Dallas or that was in
20:44Dallas called Peloton Therapeutics.
20:45And the chemist at Peloton improved
20:47these chemicals substantially in
20:48terms of their drug like properties.
20:50And they were kind enough to share with
20:52us a tool compound called PT2399 which
20:56at the time was only one or two atoms
20:58removed from the then lead compound.
21:01And so we had early access to this
21:03compound and we could show that this
21:05compound did everything you would want.
21:07And preclinical models of VHL mutant
21:09kidney cancer and similar findings
21:11were made by my former postdoc,
21:13Jim Brogarolis in his own laboratory.
21:16And so now it looks like maybe
21:17we're on to something,
21:19but I'll point out the assays
21:20in this paper were things like
21:22decreased HIF target gene expression
21:24decreased soft dog growth decrease,
21:27proliferation decrease,
21:28orthotopic tumor formation decrease,
21:30decrease.
21:30So once again,
21:31down SA after down SA after down SA.
21:33So just to give you an example,
21:34here's a lovely soft dagger essay
21:37that was done by Chincho in my lab.
21:38So you had 0.2 or two micromolar of the
21:41HIP 2 inhibitor and these renal carcinoma
21:43cells stop forming soft dagger colonies.
21:46So you get OK pretty good.
21:48But not shown here is, first of all,
21:50at 10 to 20 micromolar,
21:52This compound will kill every
21:53cancer cell we've ever looked at,
21:54whether it expresses if to or not.
21:56So by definition,
21:57that's off target.
21:59How would you know this is on target?
22:01Again,
22:01I've said this is the kind of experiment
22:04that gives cancer pharmacology a bad name.
22:06I used to hear people who did these
22:08experiments disparagingly referred
22:09to as sprinklers because they simply
22:12sprinkled noxious chemicals on
22:13cancer cells and watch them die.
22:15So,
22:15you know,
22:16I could have done this with Clorox bleach.
22:18I could have done this with formalin.
22:20You know what's what's really?
22:21Is this just another poison?
22:23But CHIN used CRISPR to make isagenic
22:27cells that had wild type HIP 2A or had
22:30hip 2A with a single amino acid change
22:32in that pocket that I showed you.
22:34That prevents the drug from binding to the
22:36pocket but otherwise leaves hip 2A intact.
22:38And now you can see we rescue the phenotype.
22:41Now eventually an approved version of
22:43this HIP 2 inhibitor which is called
22:45Bel Sudafan and I should declare I
22:47have a financial conflict of interest
22:48with Balsedifan went into testing
22:50for patients with advanced kidney
22:52cancer who had failed VEGF inhibitors,
22:54failed immune checkpoint inhibitors,
22:56etcetera etcetera.
22:56These are so-called swimmers plots.
22:59So you may know that each of these
23:00horizontal bars is a patient on
23:02the trial and how long they were
23:03on therapy at the time of this
23:05analysis of orientation.
23:05Here's one year on therapy.
23:07The black arrows were patients
23:09who were still doing well
23:11on therapy at the time of this analysis
23:13and the yellow star patients were patients
23:15who achieved a resist partial response.
23:17And I'll come back to the development
23:21of belsitovan at the end of the talk.
23:23So now I want to move to some newer data.
23:25So this is the first story is work of
23:27Nathan Chiroli, a postdoc in the lab.
23:30And he wanted to know, OK,
23:31we have this if two inhibitor,
23:33it's inhibiting the growth of
23:35these kidney cancer cells.
23:36But are all hip to target
23:38genes equally important or not.
23:40So the simple idea is you have
23:42hip two driving expression.
23:44We know it's hundreds of genes,
23:45but for this illustration let's pick 4.
23:47You come in with your inhibitor and now
23:50you down regulate these four genes.
23:52And so Nathan's idea was to now use CRISPR,
23:55a technology to activate HIP
23:57two target genes artificially,
23:59even in the face of the inhibitor,
24:02And to do this systematically go through
24:04all of the HIP two target genes.
24:06And in fact,
24:07as I'll show you in a moment
24:08for good measure,
24:09he just did this on a genome wide
24:11scale and then in silico looked
24:13for HIP two target genes.
24:15Now we're using a technology that
24:17was developed by John Dench and
24:20colleagues at the Broad Institute
24:21and this was the the preferred
24:23technology about two years ago.
24:24I can tell you they've improved upon it
24:26further, but this worked quite well.
24:28So the idea here is you have a nucleus state,
24:30CAS nine,
24:31that recruits A transitional
24:33activation domain BP 64.
24:35And then you you have some additional
24:37engineering of the CRISPR guides to
24:39bring in some other trans activation domain.
24:42So I'm sorry,
24:43this is kind of complicated and baroque,
24:44but it actually turns out to
24:46be incredibly robust.
24:47So now Nathan's ready to do his screen.
24:49He takes a HIF 2 dependent
24:51renal carcinoma cell line.
24:53He introduces the CRISPR,
24:54a guide library in cells that
24:57express that specialized CAS 9.
24:59And then he treats the cells with DMSO or
25:01the tool compound that inhibits hip two.
25:04And then he monitors guide
25:05abundance over time by next Gen.
25:08sequencing.
25:08So here's what the data looks like.
25:10Now I have to put up my glasses.
25:11So these this is the data from the
25:13cells treated with PT2399 and it turns
25:15out one of the top scoring genes.
25:18Let's see if I have an animation
25:19here is Cyclin D1.
25:21Now this immediately caught our
25:24attention because Cyclin D1 was
25:26already known to be a hip two
25:28target in kidney cancer,
25:29but interestingly and every other
25:31cancer that's been examined to date if
25:34if anything down regulates Cyclone D1.
25:36So this is probably part of the
25:38puzzle of why VHL loss causes
25:40kidney cancer but not for example
25:42lung cancer or colon cancer.
25:44So just to show you how
25:45beautifully this technology works,
25:47what I have here is a control
25:49guide or two different CRISPR
25:50activation guides from Cyclone D1.
25:52So now we either don't or do
25:54treat with the Hip 2 inhibitor.
25:56So now let's look at Cyclone D1.
25:58So if you down regulate hip two activity,
26:01you down regulate second D1.
26:03But now with the CRISPR A guides we can
26:05maintain the expression of Cyclone D1.
26:07This is specific because NDRG one
26:09is another hip two target gene.
26:11So you can see the CRISPR A guides
26:12are really specific or second D1 and
26:15now we can do validation assays.
26:17And I apologize that by my
26:19counts this is a busy slide,
26:22although what counts for a busy slide
26:24these days is certainly changing.
26:26So here I have cell number on the
26:28Y axis and days on the X axis.
26:31So now for orientation,
26:32here are the cells grown in DMSO.
26:35Here are the cells now treated
26:36with the Hip 2 inhibitor.
26:38But now if you artificially
26:39maintain second D1 expression,
26:41the cells are completely resistant,
26:43at least in this preclinical model.
26:46So we think there's an analogy
26:47to be made here potentially with
26:49hormone responsive breast cancer
26:50because the standard of care for
26:53many women with hormone responsive
26:54breast cancer is to be treated with
26:56an ER antagonist such as Tamoxifen
26:58together with a C DK46 inhibitor,
27:01CDK 46 being the catalytic
27:03partner for cyclin D1.
27:04And so you can imagine you get some
27:06synergy there because ER drives
27:08the expression of cyclin D1 in
27:09hormone response to breast cancer.
27:11And so now you're down regulating
27:13second D1 and you're hitting the kinase.
27:15So we think based on this analogy
27:16and we have some preclinical data
27:18to support this.
27:19It would be a good idea to combine
27:20a HEF 2 inhibitor with the CDK 46
27:23inhibitor and kidney cancer and
27:24such clinical trials have now begun
27:28now as often happens in experiments
27:30and another reason to do controls
27:32is it's I've been amazed over the
27:34years how often there's gold in
27:36the quote UN quote control arm or
27:38the control set of experiments.
27:39So now let's look at the cells
27:41that got the DMSO,
27:42not the Hip 2 inhibitor subjected to CRISPR.
27:45A Well,
27:45these data caught our attention
27:47as well because here's MEC,
27:49which had already been implicated
27:50in kidney cancer growth by others.
27:53Here's a gene that's less famous
27:55called SQST M1,
27:56which among other things activates NRF 2,
27:59which the gene name is NFE 2L2.
28:02So the reason we were excited about
28:04this is I have to say the most
28:06common implicon in kidney cancer
28:07or at least clear cell,
28:09I mean a carcinoma is amplification
28:11OF5Q And using horse and buggy
28:14technology about 10 years ago we
28:16deduced that the most likely target of
28:18the five Q applicon was probably SQST M1.
28:21So this provides some now additional,
28:22maybe even somewhat orthogonal evidence
28:24that maybe we were even correct.
28:27The other reason we like this
28:29is now let's go over here.
28:31So these are genes that when activated,
28:35confer a fitness disadvantage
28:38to the kidney cancer cells.
28:40So why might we care about that?
28:43Well, you may know that there are
28:45a lot of examples in cancer where
28:47chromosomal arms or sometimes entire
28:50chromosomes are missing. Excuse me.
28:54And the thought is that in many
28:55cases you're dealing with HAPLO
28:57insufficient tumor suppressors,
28:58where by reducing the copy number,
29:00you've lowered the expression.
29:01But since these are HAPLO
29:03insufficient tumor suppressors,
29:04we don't have the smoking gun of a
29:06mutation in the remaining allele as you
29:08would with a Knutson 2 hit tumor suppressor.
29:10But now we think we have a very
29:12powerful technology for reactivating
29:14these putative PAPLO insufficient tumor
29:17suppressors that are lost in cancer.
29:19And so we've now just as an example,
29:21we've made a custom CRISPR,
29:23a library for chromosome 1P which is
29:25frequently deleted in a variety of cancers,
29:27including famously neuroblastomas
29:29and are using CRISPR,
29:31a technology with focused chromosome
29:33or chromosome arm CRISPR guide
29:35libraries to look for the relevant
29:38HAPLO insufficient tumor express.
29:40OK,
29:40so now I'm going to completely switch gears.
29:42So if you didn't like that story,
29:43thank you for bearing with me.
29:44We have a completely different type of story.
29:46So it's been known for decades
29:47that clear cell Weiner cell
29:48carcinomas are highly immunogenic.
29:50They occasionally undergo
29:51spontaneous progressions.
29:53They have a high level
29:54of T cell infiltration.
29:55In the old days,
29:56they were occasionally cured with
29:58treatments like high dose interleukin 2,
29:59but some of those patients
30:00wound up in the ICU.
30:01The treatment was so toxic and
30:04many of these patients will respond
30:06to immune checkpoint blockade.
30:08So why do I tell you that?
30:09Well,
30:10you may know that in the case of some
30:13highly immunogenic tumors like melanomas,
30:15we think they're immunogenic because
30:17they have a high mutational burden,
30:18which is what I'm showing you here
30:21on the Y axis compared to the names
30:23of the tumor types on the X axis.
30:25But clear cell renal cell
30:26carcinomas smack dab in the middle,
30:28there's nothing conspicuous about
30:29clear cell renal cell, carcinoma cell.
30:31Why should it be immunogenic?
30:33So I think some widely underappreciated
30:36work was the work of Richard Childs
30:39at the National Cancer Institute
30:42dating back more than 15 years.
30:44So out of sheer desperation and
30:45I discussed this with him and
30:47it was sheer desperation,
30:48he took 74 patients with metastatic
30:50clear cell Reno cell carcinoma and
30:52treated them with allogeneic stem
30:54cell transplants as a source of
30:56potentially immunoreactive T cells.
30:58And remarkably,
30:58about half of the patients responded,
31:00including some who were durable responders.
31:02So about 10 to 12% of the patients
31:05were durable complete responders.
31:06And in one of the CR patients,
31:08one of the complete responders,
31:09he found donor derived T cells that
31:12were recognizing on the surface of
31:14the tumor cells a tenmor peptide
31:16derived from an endogenous retrovirus.
31:18And they were able to show that
31:20this retrovirus,
31:20which at the time was called Erbe,
31:22its expression was restricted to
31:24clear cell renal cell carcinoma,
31:26and it was not detectable in
31:28normal tissues or other cancers.
31:30Moreover, some but not all groups
31:33have have reported that the expression
31:35of endogenous retroviruses at least
31:37correlates with the probability that
31:39a kidney cancer patient will respond
31:42to immune checkpoint blockade.
31:43Now in the case of the Richard Child's ERB,
31:47they were able to show that this
31:49ERB was directly regulated by HIF
31:512 at the transcriptional level,
31:53which would very satisfyingly explain why
31:55it would be up regulated in kidney cancer.
31:57And so at this point I can introduce
32:00another postdoc, Chin Chin Chiang.
32:02She wanted to know,
32:03was this Richard child's Erba
32:06one off like a Unicorn?
32:08I mean great for this patient
32:10never to be seen again?
32:11Or was it the tip of the iceberg and
32:13was it trying to tell us something?
32:14So that's what she set out to do.
32:17So her hypothesis were that hip drives
32:20the expression of multiple Ervs,
32:22with erve,
32:23which has now been renamed ERV
32:244 simply being one example.
32:26So this being the Richard Childs ERV.
32:28And then she hypothesized that maybe
32:30some of these other Ervs like the
32:32Richard Childs ERV are transcribed and
32:34translated into MHC bound peptides.
32:37And I should point out we're
32:38not looking at every ERV,
32:39we're not looking at every remnant of an ERV.
32:41Excuse me, I'm sorry.
32:43We're looking at about 3000 Ervs
32:45that have been annotated to be
32:47relatively intact and where you
32:49can imagine them being transcribed
32:50and translated to some degree.
32:52And I should also point out that this
32:54is a collaboration with David Brown,
32:55who's sitting here,
32:56as well as with Kathy Wu,
32:58and we could not have done this
33:00work without them.
33:01So the first thing Chin Chin did was
33:03she took a rhino carcinoma cell line
33:05that lacks VHL and then she generated
33:073 isagenic pairs because again,
33:09for the students, you know,
33:09corroboration is your friend.
33:11So corroborating across multiple systems,
33:12multiple technologies, that's a good thing.
33:15So she either infected these cells
33:17with an empty expression vector or
33:19a vector encoding VHL she treated
33:21with vehicle or that HIP 2 inhibitor,
33:23or she used CRISPR to eliminate HIP
33:262A or treated with a control guide.
33:28I can point out,
33:29I should point out that at least
33:30in the short term these cells
33:32will tolerate loss of hip two if
33:33you keep them in high serum.
33:35And then with the help of David
33:37and his colleagues,
33:38we looked for Ervs and RNA seek data.
33:41So here are the Venn diagrams
33:43with with those three conditions.
33:45But again I'm a lumper rather
33:47than a splitter.
33:47So let's be maybe even overly stringent
33:49here with fairly stringent cut offs.
33:51There were 15 Ervs that scored
33:54in all three comparisons.
33:56And I'm,
33:56I apologize for the kind
33:58of crazy nomenclature,
33:59but that's where we are in the world of ER,
34:01BS.
34:01But it's always nice to have
34:03an internal control,
34:04even if you didn't know you
34:05had an internal control.
34:05SO1 internal control was we did rediscover
34:08the Richard Childs ERB, that's good.
34:10And we also rediscovered in the ERB
34:12which is sometimes called 3.2 which was
34:14in one of those JCI papers as being
34:16a potential predictive biomarker for
34:19response to immune checkpoint blockade.
34:21So then the question was well are
34:23are these things really under
34:25direct hip control or not.
34:26So we did chip seek multiple ways.
34:30One way we did it was to knock
34:33in using CRISPR combined with
34:35homologous recombination.
34:36We knocked an an endogenous flag
34:40tag into the endogenous HIP 2 locus
34:42and a renal carcinoma cell line
34:43that that allowed us to do a chip
34:45seek with a flag antibody.
34:47Or we did chip seek with an anti
34:49HIP 2A antibody in cells where we
34:52did or did not eliminate hip 2A.
34:56And again getting to you know corroboration
34:57and multiple lines of evidence.
34:59If I only had these two tracks and I
35:01squinted I guess I could convince myself
35:03there's a hip 2 binding site there.
35:06But now if I bring in the flag
35:07chip seek and the knock in cells,
35:09I think you can convince yourself
35:10there's a there's a hip to binding
35:12site parenthetically for for
35:14splitters rather than lumpers.
35:16Richard Childs had identified up with Peter
35:18to hip binding site on the shoulder here.
35:21I think he probably missed it,
35:22not that it really mattered.
35:24So there clearly is a hip
35:25to binding site here.
35:26But now let's look at some of our new Ervs.
35:28I picked 2 examples I like
35:29because for these two Ervs,
35:31for whatever reason,
35:32they are actually precisely bookended
35:34by hip 2 binding sites.
35:35So here's one called 5875 and you
35:37can see the hip binding sites here.
35:39And here's another one called 4818.
35:41Nice Hip 2 binding sites shown here.
35:43And all these and all these validated
35:46secondary experiments I'm not
35:47going to share with you today.
35:49So now could it really be that
35:51some of these other Ervs like the
35:53Richard Child's ERV are actually
35:55transcribed and translated?
35:56So we collaborated with Sterling
35:59Churchmen who helped doing Polysome
36:01Seek and again using very stringent
36:03criteria where now we'll overlap
36:05the RNA seek data with the chip seek
36:07data and the Polysome seek data,
36:08again probably using overly
36:10stringent cut offs.
36:11But again we rediscovered
36:12the Richard Childs ERB.
36:13We also discovered those two Ervs I
36:16just showed you a moment ago that are
36:19bookended with hip 2 binding sites.
36:21So now we're ready to ask,
36:22are these some of these Ervs also
36:25transcribed and translated into
36:27peptides that are displayed.
36:29And here we were helped immeasurably
36:31not only by David and Kathy,
36:33but also by Carl Klauser and Steve
36:35Carr at the Broad Institute.
36:37Suffice it to say,
36:38we can find these things in cell lines,
36:40but more importantly,
36:40we find some of these peptides
36:42in real kidney tumors.
36:44So here we have the data from
36:46the 1st 11 patients for whom we
36:48had kidney tumors
36:49samples. For six of these 11 patients,
36:52we had normal controlled tissue
36:55and we've identified again this is
36:57primarily the work of called Kauser,
36:59about 30 Ervs peptides,
37:02ERV derived peptides and in every
37:05case where we had normal tissue,
37:06they were not detected in the normal tissue,
37:08they were exclusively present
37:10in the in the tumor tissue.
37:13So we're excited that by continuing
37:15these sorts of studies we can start
37:17to learn more and more about which
37:18Ervs can be transcribed and translated
37:21and which maybe potentially could
37:22be the basis for various types of
37:25passive or active immunotherapy.
37:26I will also tell you although not shown
37:29here is if you now take a clinical
37:32grade hip stabilizer and treat other
37:33types of cancers such as melanomas,
37:35brain tumors, colon cancers,
37:37you dramatically up regulate
37:38the expression of various Erbs.
37:42OK, Story 3 undruggable cancer targets.
37:46Now, I think you probably know there's
37:48no shortage of genetically validated
37:51but undruggable cancer targets,
37:52although there's been enough progress
37:54on Ras in the past year that maybe
37:56RASP will leave this list shortly.
37:58But certainly there's no shortage
37:59of undruggables that we might
38:01want to think about accessing.
38:02I think generically there are a number
38:04of ways to go after these undruggables.
38:05In some cases,
38:06you can go downstream of the mutation.
38:09Effectively that's what we
38:10did with loss of VHL.
38:11We went downstream and targeted a hip,
38:14effectively exploiting an
38:16epistatic relationship.
38:17There's renewed interest in
38:19developing allosteric inhibitors.
38:20That's effectively what
38:21the HIP 2 inhibitor was.
38:22I've had a long standing
38:24interest in exploiting synthetic
38:26lethal interactions in cancer,
38:27but I'm not going to talk
38:29about the about that today.
38:30But I do want to talk about
38:32small molecule degraders,
38:33which I I deal with some trepidation
38:34with Craig Cruz in the in the front row.
38:36So he he's going to keep me honest
38:38through this section of the talk.
38:40So our group and Ben Ebers group working
38:42in parallel showed about five years
38:44ago that the thalidomide like drugs,
38:46the so-called image,
38:48act as molecular glues that
38:50recruit A ubiquitin ligase,
38:53not the VHL Ubiquin ligase,
38:54but in fact the cereblond,
38:56the ubiquitin ligase to now target for
38:59degradation to transcription factors
39:01called IKZF one and IKZF 3 which
39:03turned out to be near and dear to
39:05the hearts of multiple myeloma cells.
39:07Now in fairness this really
39:09kind of Harkins back.
39:11Oh,
39:11I should also say almost immediately
39:14people like Craig Cruz and Jay
39:16Brenner showed that you could
39:18chemically modify an image so that
39:20it would now go after another target.
39:23But I was about to say,
39:24in fairness,
39:24this general idea of having a
39:26chemical that acts as a matchmaker
39:28between an E3 ligase and a target
39:30goes back to the Seminole work of
39:32Craig Cruz as well as Raid Deshays.
39:34And just to introduce some nomenclature here,
39:37these molecules on the left are sometimes
39:39of course referred to as protax.
39:41I realize this is bringing
39:42Coles to Newcastle.
39:43These are sometimes referred
39:45to as molecular glues,
39:46and I and I think either you or Jay
39:48tried to call these degronomists.
39:49I think you need a publicist.
39:50I'm not sure that one caught on
39:52as well as the molecular glue,
39:53molecular glues or the protax.
39:56So when you think about it,
39:59however, there are lots of other
40:01ways a small molecule could
40:03degrade directly or indirectly,
40:04your favorite undruggable protein.
40:06Because protein stability
40:07is highly regulated.
40:09There are about 500 Ubiquitoligases.
40:11They're opposed by about 100 Dubs.
40:13Whether these ligases and dubs will
40:16recognize their targets are often
40:18influenced by post translation modifications,
40:20protein folding,
40:21protein protein interaction,
40:23sub cellular localization,
40:24et cetera, et cetera, et cetera.
40:26So we wanted to develop an
40:27assay for degraders that was
40:29actually mechanism agnostic.
40:30Now,
40:31we've talked a lot today about
40:32up assays versus down assays.
40:34And that's because 20 plus years
40:36ago some of my colleagues in pharma
40:37tried to beat into my head why?
40:39If we're doing a chemical
40:40screen or a genetic screen,
40:41you'd almost always rather do
40:43an up assay than a down essay
40:45for for at least two reasons.
40:46First of all,
40:47up assays tend to have better
40:49signal to noise characteristics.
40:51It's usually easier to see a
40:53positive and a sea of negative than
40:54a negative and a sea of positives.
40:56If you think about it,
40:56that's what makes astronomy work.
40:59But maybe more important for
41:00us as biologists,
41:01they're just lots of trivial
41:03ways to make a complex system,
41:04including a complex biological system,
41:06work worse rather than work better.
41:09So my thought experiment for the students,
41:11if you don't believe me,
41:12I hope it's just a thought experiment,
41:13is you go out to your car one
41:15evening and start randomly removing
41:17parts from your internal combustion
41:18engine and tell me how often it goes
41:20faster and how often it goes slower.
41:22So that's your that's your.
41:23Again I emphasize this probably
41:25should be a thought experiment.
41:26So we wanted to develop an up assay.
41:28So Sagar Kaduri when he was in
41:30my lab a very talented physician
41:32scientist developed this assay
41:34where your protein of interest is
41:36fused to a modified cited in kinase
41:38that was developed by Jeff Medin
41:40that will accept the non natural
41:41nucleoside and converted into a toxin.
41:43There's a little spacer,
41:44there's AV5 tag and then importantly off
41:46the same transcript you have a a GFP.
41:48So,
41:49so the idea is you can add a
41:51chemical or genetic perturbance,
41:53add BBDU and look for green survivors.
41:56And as this is an up assay this can be
41:58done in both arrayed format with chemicals.
42:00It can also be done in pooled
42:01format with CRISPR libraries.
42:02Just to show you a proof
42:04of concept experiment,
42:05here's a 384 well plate assay where
42:07the cells have the Imid target IKZ F1.
42:09I hope you can see there are
42:11three positive wells here.
42:12It turns out these two wells
42:14had an Imid pomalidomide,
42:16whereas this well had a chemical that we now
42:20know is an assay positive that interferes
42:23with BVD uptake by themselves,
42:26but that's easily detected.
42:28Then again, for proof of concept,
42:31Sagar did the following experiment.
42:32So again, the top plate
42:34is the IKZ of 1 fusion.
42:36The bottom plate is the counter screen
42:39counter screen with unfused CK.
42:41There are various controls in the outer
42:43columns which you can ignore for now.
42:46Otherwise each row has two different
42:48chemicals at 10 different concentrations.
42:50So here's an assay positive,
42:52so we don't care about that we're
42:53at the highest concentration
42:55somehow promotes the survival.
42:56But here's a true positive which
42:58was originally named Spout,
43:00and one it's been reported to
43:02be a modifier of autophagy,
43:03but that turns out to be a red herring here.
43:06But this turns out to be true,
43:07positive and and we know it's
43:09not just another image,
43:10first of all by looking at the structure,
43:12but secondly in contrast to the image.
43:14This doesn't require Cerebon
43:17to degrade IKZ F1.
43:20Now we thought our UP assay might
43:23allow us to revisit screening
43:25natural product collections.
43:27Now for those of you who are
43:30who haven't lived through this,
43:31there was a time back in the
43:33days of dinosaurs where screening
43:35natural product collections was
43:36a preferred way of finding drugs
43:38and the and the and the pros are
43:41enhanced chemical diversity.
43:43Nature's produced chemicals that
43:45I don't think any human chemist
43:47has ever been able to make.
43:49And natural product hits often have
43:51drug like product properties right
43:52out of the gate because in many cases
43:54that's what they were designed to do.
43:55They were designed, for example,
43:56to cross cell membranes.
43:58The cons are that contaminating toxins
44:01cause false positives and down assays
44:04and false negatives and up assays.
44:06And again we're doing up assays.
44:08It's historically difficult to isolate
44:10and identify the active chemicals
44:12in the mixtures of scores hits.
44:14This can take anywhere from years,
44:16decades to Infinity.
44:17And then sometimes it's not very frequently.
44:20It's difficult to identify the direct
44:22protein target for the chemical
44:25coming out of the phenotypic screens.
44:26Although I have to say there's some
44:28very clever and powerful genetic and
44:30biochemical techniques for doing this now.
44:32So we became,
44:34oh,
44:34that's how they should say so this is
44:36the unpublished work of Matt Boudreaux.
44:38So how can we leverage natural products
44:40and do screens with our UP assay?
44:42So we became aware of the work of Barry
44:46O'Keefe at the National Cancer Institute,
44:47who overseas the natural product collection
44:49that all of us paid for with our tax money.
44:51So you might as well ask for this
44:53collection at some point because you paid
44:55for it and his very simple but powerful idea.
44:58And again,
44:58for the students,
44:59simple is not bad.
45:00Simple is often like, good.
45:02OK, so I think we're drawn to
45:05the complicated fancy ideas.
45:06It's often the simple things
45:07that moves forth.
45:08So his embarrassingly simple but very
45:11powerful idea was what if I took this
45:14compound collection and there are over
45:15200,000 of these very complicated broths,
45:17extracts, mixtures housed at the NCI.
45:20And suppose we run them over a single column.
45:23So now each mixture is now
45:25represented by 7 fractions.
45:27And I apologize for all the verbage here,
45:29but suffice it to say let's pick the column.
45:31So it removes a lot of the nuisance compounds
45:34that plagues many phenotypic screens and in
45:38particular the so-called pain compounds,
45:40Pan interference compounds.
45:41And because we now have
45:43separated into 7 fractions,
45:44we've simplified the mixtures and
45:46that increases the probability that
45:48you'll get to the active chemical.
45:49They tell me their success rate is about 80%.
45:52If you get a hit from one of
45:54these single fraction comes,
45:55it helps with reproducibility,
45:57yada, yada yada.
45:58So for those of you are interested,
45:59this nice summary of this platform.
46:02So we moved this platform to Harvard
46:05and we decided this would marry
46:08well with some overall up assays.
46:10So Matt Boudreaux,
46:11whose picture I just showed you,
46:13he decided to take on mutated beta catenin,
46:16another classically undruggable.
46:17This could be updated now,
46:19but he's effectively gone
46:20through the entire collection.
46:22For those of you who do screens,
46:23the screen had very good statistics.
46:25But I want to kind of give you a sense
46:27of the power of this technology.
46:28So for all 7 fractions we get the
46:32unfractionated master mix if you will,
46:35as F0 and Z score here is a
46:37measure of the number of viable
46:40*** positive cells in the well.
46:42So in hindsight you could see that,
46:45yeah,
46:45maybe there's a little bit of a
46:47blip here with the master mix,
46:48but this doesn't come close to our
46:51threshold for calling a positive
46:52in the screen.
46:53If we had,
46:54if we moved our threshold down here,
46:56we just have thousands and thousands
46:58and thousands and thousands of hits.
46:59But now if you look at the seven
47:02sub fractions,
47:02you can see that F5 and F6 are
47:04a winner and we like this.
47:06We take great comfort if two adjacent
47:09fractions score because again this
47:10is a single column so it's unlikely
47:13that anyone chemical is going to
47:15be exclusively in in one fraction.
47:17So some of these do validate
47:19in secondary screens.
47:20This is a work in progress and it's
47:22unpublished, but I'll just show you.
47:23Here is an example of a fraction that
47:26where the Master was called 15 O 9
47:28and here are two adjacent fractions I think,
47:31well,
47:31it's called dash 5 and here's fraction 7.
47:34Here's an adjacent.
47:35But anyway,
47:36you can see we're degrading not only or
47:39down regulating the beta catena fusion,
47:41but more importantly we're down
47:44regulating endogenous beta catena
47:46and we're turning off beta catena
47:48target genes such as Axon 2.
47:50And for those of you who care,
47:52this extract is from a plant
47:55that's living somewhere in Belize,
47:58which brings up a whole other issue.
48:00If some smart chemists can't
48:01make this chemical,
48:01we may wind up going down to
48:04Belize and harvesting this plant.
48:06But another truism is,
48:07I've learned over the years,
48:08when you set up a screen,
48:09even a good screen,
48:10you get the things you were hoping to get,
48:13and then you get the things you
48:14didn't know you were screening for,
48:16but in fact you were
48:17screening for them all along.
48:18So here's another robust positive that
48:20scored as a pro survival chemical in
48:24that reporter system I described to you.
48:27So somehow we're inactivating Beta Catena.
48:30But now to our surprise when we
48:31did the western blots and here
48:33multiple fractions of this extract
48:34which is from a plant in Tanzania,
48:37you can see that actually the
48:39reporter and endogenous beta
48:40Catena abundance goes up not down.
48:43But it is apparently inactive because
48:45here is X and two which again
48:47reads out beta catena in activity.
48:49And when you look under the microscope,
48:50you can see that this chemicals inducing
48:53beta catena into go into aggregates.
48:55So maybe that's not such a bad thing.
48:57Maybe that's another way
48:58to inactivate beta catena,
48:59we'll just drive it into aggregates
49:02as opposed to destabilizing.
49:03Now I should point out for this
49:05chemical and I can't even say
49:07chemical singular now because
49:08these are only partially purified.
49:10I don't know whether this extract or the
49:13previous extract are going to hit a dead end.
49:16I can't tell you anything
49:17yet about true specificity.
49:18What I can tell you is
49:19because it's an up assay,
49:20they're presumably not just rat poison,
49:23but maybe they are affecting dozens
49:25if not hundreds of other proteins
49:27and so that has to be determined.
49:29So in closing,
49:30I showed you this a moment ago.
49:33I I wrote this paper in 2017 and
49:38it was my attempt to summarize in
49:41about 8 pages every common mistake
49:45pitfall artifact I had seen in
49:48the literature related to target
49:49identification and validation,
49:51including by the way some mistakes
49:52we made along the way.
49:53We all make mistakes.
49:54So this may be partially A mea culpa,
49:56but this was my attempt to
49:57say we have to do better.
49:58We can't have our friends in industry
50:00saying that 50 to 90% of the time
50:02they can't replicate our findings.
50:04And so these are some of the controls
50:05you might want to think about if you're
50:07doing these types of experiments.
50:08So the reason I wrote this is my wife
50:10was a breast cancer surgeon at the Brigham.
50:13She developed breast cancer.
50:15She actually self diagnosed
50:16herself with breast cancer in 2003.
50:18In fact, Dara Weiner, who's now here,
50:20was her medical oncologist and
50:22she survived her breast cancer,
50:23but she developed an unrelated
50:25glioblastoma in 2010,
50:27which took her life in 2015.
50:29In fact,
50:30here she is with Catherine
50:33graduating from Saybrook,
50:34if you're wondering.
50:35And Carolyn died about a month or two later.
50:39In fact, anyone here is a mother
50:41or has a you all have mothers.
50:43You know,
50:43I was.
50:44I was sure if there was
50:46any way Carolyn could live to see
50:48her eldest child graduate from Yale,
50:50she was going to figure out a
50:52way to do it, and she didn't.
50:54So I wrote this because if you think
50:56it's all about how many papers you
50:58publish in Cell, Science or Nature,
50:59you think it's all about whether
51:01you fooled Reviewer 3 one more time
51:03and you slipped another one in,
51:05and it's all about what societies
51:06you've been elected to, etcetera,
51:08please stop, because you're just
51:10becoming a dominant negative.
51:11The only thing you should care about is
51:14whether what I've discovered is so true,
51:16so reproducible and so robust,
51:18the next group of people can
51:20come and build upon that.
51:21So if this is all about ego gratification,
51:23please, please,
51:23please just stop.
51:24Because I can tell you everything we had
51:27available to treat my wife was based on
51:29science that was done 10 or 20 years ago.
51:32If God forbid one of our children
51:34or grandchildren gets this disease
51:35and statistically unfortunately
51:36it's likely to happen,
51:38they're going to be counting on
51:39the science we're doing now.
51:41So we have to do much better.
51:43So to end on a happy note,
51:44because I don't want to end on that note,
51:45I showed you these data,
51:46these were the data from the
51:48phase one two trial.
51:49The phase three data were positive
51:51and the FDA approved Bel Sudafan
51:54for the treatment of sporadic kidney
51:56cancer just about a month ago in
51:59patients who have failed VEGF inhibitor
52:01or immune checkpoint inhibitor.
52:03But another truism in medical oncology
52:05is most drugs and medical oncology
52:08work better in front line settings
52:11than in very late lines of settings
52:12where you've beaten the patients up
52:13and treated them with all sorts of
52:15noxious things and you've selected for
52:16all sorts of resistance mechanisms.
52:17So let's go back to those Von
52:19Hippel Lindau disease patients.
52:20So these patients developed so many tumors,
52:22they're often in careful surveillance
52:24programs where they'll get Mris
52:26every three to four months in an
52:27attempt to delay or prevent the
52:29need for repeated surgeries,
52:30such as repeated partial nephrectomies,
52:32which in the case of partial
52:34nephrectomies would leave them
52:35eventually functionally anephic.
52:36So we were able to convince Peloton
52:39and then Merck to treat 61 patients
52:41with von Hippelinda disease that
52:43had measurable kidney tumors that
52:45had never been treated before.
52:46They were just in these surveillance
52:48programs.
52:49So now you can see the the
52:50swimmers plots look even better.
52:52So once again the dotted line is one year,
52:55but now I think you can see that
52:57most of the patients are doing
52:58well and many of these patients
53:00went on to have a partial response.
53:02If you prefer waterfall plots
53:05then swimmers plots.
53:07Here are the changes in the size of
53:09the indicator kidney tumors that
53:11were the basis of these patients
53:13entering the trial.
53:14So again, I'm as you now know,
53:16I'm a lumper rather than a splitter.
53:17So I would say these are all going down.
53:19Some officially meet the criteria
53:21for resist response, some don't,
53:23but you know frankly if you're
53:25the patient as long as the tumor
53:26is getting smaller, that's good.
53:27We had also done some preclinical
53:29modeling that suggested that hip two
53:31was important in the blood vessel tumors
53:34and here are the hemangioblastoma.
53:35So again to be on the trial you
53:37had to have a kidney tumor,
53:38but you could have other tumors
53:39associated with VHL disease as well.
53:41So here are the hemangioblastoma shrinking.
53:43These patients also develop an
53:45unusual neuroendocrine tumor of
53:47the pancreas called peanuts.
53:48They responded very nicely and
53:52perhaps as importantly,
53:53I I mentioned these patients
53:55were in surveillance programs.
53:56So in Gray are the four years
53:59of surveillance before going on
54:00the HIP 2 inhibitor in green and
54:03everywhere you see a circle,
54:04a square, a diamond, whatever,
54:06that patient's going back to the operating
54:08room again to have a kidney tumor removed,
54:10an eye tumor removed,
54:11the spinal cord removed, tumor removed.
54:13And you can see that the,
54:15the frequency of the surgery goes way down,
54:18doesn't go to 0,
54:19but it does go way down.
54:21And so based on that,
54:22several years ago,
54:23the FDA just on the phase two data
54:25approved this drug for the treatment of
54:28Von Hippel Lindau disease just about 100
54:32years after Lindau's initial report.
54:34But just because statistics can
54:36sometimes be a little bit dry,
54:38to put a bit of a human face on this,
54:40we actually knew the trial was going
54:42to be positive long before any public
54:44presentation of the data because
54:46the VHL patients were posting on
54:48their social media pages They were doing.
54:50And you can imagine these patients
54:51have lived with the Sword of
54:52Damocles over their neck or hey,
54:54whatever the expression is,
54:55because they've watched this disease
54:56ravage their families generation
54:58after generation after generation.
54:59And they just assume they're going
55:01to probably die sometime in midlife.
55:02So here's a VHL patient saying,
55:04I never thought I'd see this day.
55:05And they're describing their tumors
55:07getting smaller, being stable,
55:09or in some cases disappearing entirely.
55:11And since everyone likes a movie,
55:13I was sent a vlog.
55:14I didn't know what a vlog was
55:15until this was sent to me,
55:17but I do have permission to
55:19show this patient's face.
55:20But this is another patient
55:21who was on that trial.
55:23Hey everybody,
55:23it's Justin,
55:24And I just wanted to give you a quick update.
55:26I am in a gondola right now in Taiwan.
55:29Over there is Taipei One O 1.
55:33The gondola is actually right
55:36by the Taipei Zoo,
55:37but I just wanted to give you a
55:39quick update and say I'm doing well.
55:41I'm enjoying my trip.
55:42If it wasn't for the PT2977 drug trial,
55:46I would have never been able to come out
55:48here and do what I'm doing right now.
55:50So I just wanted to thank Peloton
55:53and I hope Mark will fast track
55:55this drug for a VHL treatment.
55:58So if you guys are listed in,
56:00hopefully you guys will
56:01put on market to help VHL.
56:03But yeah, keep watching these videos.
56:07I'll be making more and
56:08I'll get better at it.
56:09And I have to get the angles right
56:10because they kind of look fat, you know?
56:13Anyway, so, so I.
56:14So I love that because that tells me
56:16he's back through his premorbid personality.
56:17Because at that age,
56:19your biggest concern should be whether
56:21you look fat on your blog and not
56:22how bad your MRI might look next
56:24time around and your doctor telling
56:26you to get your affairs in order.
56:28So with that,
56:28I thank you very much for your attention.
56:30I'm happy to take questions.
56:43Yes.
56:54Yeah, that's a great question.
56:55So we talk about this a lot.
56:57So the question relates to,
56:58in principle then,
56:59if you had a Hip 2 inhibitor,
57:01you would down regulate the expression
57:03of the ZR VS and that might make immune
57:05checkpoint inhibitor therapy work worse.
57:07And so maybe the combination
57:09would actually be antagonistic.
57:11I think we just have to do the clinical
57:12trial that that is a prediction.
57:14And so I won't be completely
57:15surprised if that's what we see.
57:16On the other hand,
57:17as I'm sure you well appreciate,
57:19there's so many benefits from
57:21combining 2 drugs that have
57:23distinct mechanisms of action in
57:25terms of treating or preventing
57:27resistance that I still wonder
57:28whether the benefits of having two
57:30drugs with very distinct mechanisms
57:32of action would outweigh this
57:33theoretical concern that in some
57:35cases they would be antagonistic.
57:36I can also imagine there might be
57:39some games you could do in terms of
57:40timing that might also be helpful.
57:42There
57:46is. There is that a student?
57:47Well, students should always
57:48go first if that. I can't,
57:49It's so dark back there I can't tell.
57:51This could be the Dean for all I know,
57:52but I this no, it's not the Dean.
57:54I'm quite sure now that I see the beard,
57:55it's not the Dean, but anyway
58:10yeah so. So let me do a medical student
58:12trick and answer a related question
58:14before answering your question.
58:16So one is I I I do think hip and
58:19and in particular hip too is part
58:20of the reason these Ervs are up
58:22regulated in the kidney tumors,
58:24but I don't think it's the only reason.
58:25So I also think there are widespread
58:29abnormalities and DNA and histone
58:31methylation and the kidney tumors
58:33that I think is creating a permissive
58:36environment for the expression.
58:37And we know this is true because
58:40for some of the Ervs HIP only
58:42regulates them if we add a DNA
58:44methyl transferase inhibitor and
58:45then you could see the creation of
58:47a hip binding site next to the ERB.
58:49So I think it's an interplay of HIP and
58:51a permissive epigenetic environment.
58:53Now I think you're asking,
58:53OK, that's fine,
58:54but if these things aren't regulated,
58:55why doesn't the endogenous immune system,
58:58oh and it's one other thing so far
59:00every time we've looked with these
59:01do seem to be tumor restricted
59:03and not seen in the normal.
59:04But you still,
59:04I think you're asking the question,
59:05well that's fine,
59:06but why doesn't the immune system
59:07recognize these kidney tumors
59:09that already have the Erbs?
59:10So that's a question for people like David.
59:13I walk around thinking that the
59:15T cells that we needed or wanted
59:18eventually lost that they became
59:20exhausted and that the tumor use
59:22various molecular signals and tricks
59:24to either obey the immune system or
59:26or to cripple the immune system.
59:28So I'm hoping that,
59:29you know,
59:30the immune checkpoint inhibitors
59:31are just the foot in the door
59:33to having better and better and
59:34better agents to allow the immune
59:36system to once again recognize the
59:37kind of things that they probably
59:39should have been able to recognize.
59:41Is that, is that OK,
59:42David, That is all right.
59:45Yeah, Yeah, yeah. Acceptable answer.
59:47He said good
59:49question about the other, the secondary.
59:53Yeah. So how about the
59:56occupation in bad competition?
59:59How does that practice?
01:00:00Yeah, so Chin's asking and I didn't
01:00:02make this point that VHL loss,
01:00:04although it's a critical
01:00:05first step in kidney cancer,
01:00:06is not sufficient for kidney cancer.
01:00:08You need other cooperating.
01:00:09Mutations such as in genes like PBRM
01:00:11One and BAP One and other other genes,
01:00:14many of which are involved
01:00:15in epigenetic regulation.
01:00:16In the case of PBRM One loss,
01:00:18we and others have shown that PBRM One loss,
01:00:21if anything,
01:00:21amplifies the hip activity even further.
01:00:24I'm not convinced that's true
01:00:25for BAP One and so we are trying
01:00:28to understand the biochemistry
01:00:29of the other gene products that
01:00:31are altered in kidney cancer.
01:00:32But I think in the case of PBRM One,
01:00:34it is partly about amping up
01:00:35even further the hip response.
01:00:39Yes,
01:00:43yes. So
01:00:54I'm wondering,
01:01:06I, I, I do in fact another thing we
01:01:08occasionally see and this is another
01:01:10old idea that's coming back in vogue.
01:01:13You know there are a lot of examples
01:01:14where we know in the cancer a
01:01:16certain oncogene signal is high.
01:01:18You know, we'll call it MIC,
01:01:19we'll call it E2F.
01:01:20And so the knee jerk response usually
01:01:22when we see an oncogenic signal very high
01:01:24is to try to inhibit it with a drug.
01:01:26But there's their data that go back 20
01:01:28or 30 years ago that show in some cases
01:01:31these cancers are just at the brink
01:01:33of apoptosis that if you could drive
01:01:35the oncogenic signal up even a little higher,
01:01:38the cells would die.
01:01:39So paradoxically in some cases,
01:01:41the answer I think is you want this oncogene,
01:01:43I'm going to give you so
01:01:44much of this oncogene,
01:01:45you're going to choke on it.
01:01:46So we have seen in certain settings
01:01:48using CRISPR A where further
01:01:50activation of a professional oncogene
01:01:52in that cancer causes cell death
01:01:54and you delete the CRISPR guide.
01:01:56So I think that's And so now
01:01:58we get into semantics.
01:01:59I think that is a form
01:02:00of synthetic lethality.
01:02:01If you knew there was a target
01:02:02that when inhibited further
01:02:04activated the Aqua gene.
01:02:13People have work to do,
01:02:15We should let them.
01:02:15We should let them go.
01:02:16All right. Thank you very much.