Pathology Grand Rounds, February 22, 2024 - Ranjit Bindra, MD, PhD

February 29, 2024

Information

Ranjit Bindra, MD, PhD, Harvey and Kate Cushing Professor of Therapeutic Radiology and Professor of Pathology, presents, "Exploiting DDR Defects in Cancer: Stories from the Bench to Bedside and Beyond . . . "

ID11393

To CiteDCA Citation Guide

00:00All right, everyone, welcome,
00:04although he really needs no introduction.
00:06It is my honor, on behalf of the Pathology
00:10Department Grand Rounds Committee,
00:13to welcome our Grand Round speaker today,
00:15Doctor Ranjit Bindra.
00:16He is the Henry and Kate Cushing
00:19Professor of Therapeutic Radiology
00:21and of course professor of
00:24Pathology and also neurosurgery.
00:26He's also well known in our
00:29department as he's a graduate of
00:32our experimental pathology PhD
00:34program and Yale's MDPHD program.
00:37He's obviously everyone is aware that
00:41he's nationally and internationally
00:43recognized for his discoveries on onco
00:45metabolite induced dysregulation of
00:47DNA damage response and repair and
00:50how he's been able to exploit those
00:55defects to for therapeutic gain.
00:57And so maybe lesser known fact I have
01:02on good authority that Ranjit is
01:05actually an accomplished drummer as well.
01:07He also won the the the Bio CT
01:11award for Entrepreneur of the the
01:13Year a few years back.
01:14And so without further ado,
01:16I'd like to welcome
01:18Ranjit. Great, thank you.
01:23OK, great. Well, thanks so much.
01:25I'm really excited to be
01:26speaking in a live audience,
01:27not just a zoom meeting.
01:29So this is great to hear.
01:30My disclosures only be focusing
01:32on the last company that modified
01:34bio just at the very, very end.
01:36But we're going to make this
01:37very academic and obviously,
01:38so topics that we'll cover today,
01:41we'll start off sort of our quest to
01:43translate work from bench to the bedside.
01:45Some key concepts,
01:46progress and learnings for those who've
01:47seen these intro slides have changed them.
01:49So they're updated this year.
01:52And then we'll move on to our recent
01:54discovery of DNA modifiers and how
01:56we're trying to exploit loss of
01:58MGMT and glioma and then sort of
02:00deviate a little bit further into
02:01the horizon and what we're actually
02:03trying to do outside of glioma.
02:05And sort of in a bigger picture,
02:07what directions were going on?
02:08And then we'll sort of have an epilogue of,
02:10you know where do we go from here.
02:12So just to get started,
02:14So synthetic lethality therapeutic index,
02:16I think this audience doesn't
02:17need much of a background on that.
02:18You have two pathways A&amp;B that are
02:20somewhat relevant or parallel pathways.
02:22You knock one out and the other
02:24ones out in the tumor cell,
02:25then you have a selective targeting
02:27of the tumor over the normal tissue.
02:30And I've always been fascinated.
02:31I always give credit to David
02:32Stern because he and I,
02:33when I was a graduate student,
02:34asked to ask him the question said,
02:36well,
02:36let's let's talk about the history
02:38and maybe look up this old paper from
02:401945 where the original Drosophila
02:42work were synthetically valid,
02:43meant you could never see it in a
02:45screen because you both knocking
02:47out both pathways,
02:47you wouldn't actually see it pop up.
02:49You'd have to engineer it for it to happen.
02:52And it wasn't until 1997 that Lee Hartwell,
02:55Steve Friend and colleagues
02:56published the SEMOL paper 1997,
02:58which are in the late 90s rather
03:00where they actually talked about
03:01this concept of synthetic lethality.
03:03It was fascinating because at
03:04that time they didn't really know
03:06about BRACA mutations.
03:07They had not been fully discovered and
03:09they thought this could have a potential.
03:11But soon after it was in 2005 that we
03:14had the back-to-back Nature papers
03:16when I was a graduate student in the
03:18Xpath program showing that Olaparib
03:21was active against Braca mutant
03:24cancers and that really opened up a
03:26PARP inhibitor bracket synthetic without it,
03:28which became the first sort of case
03:31or poster child for this approach.
03:33I still remember this.
03:34This is back in 2005 or so when I
03:37actually got a vial of the drug.
03:39I'd met one of the lead authors
03:40on a plane to a Gordon conference
03:42and he actually mailed it to me
03:43without an MTA and actually tested
03:45it in VCA Braca wall type mutants
03:47in Peter Glaser's lab and you can
03:49see the effects
03:49quite striking. And now we have
03:514 FDA approved PARP inhibitors.
03:53So we've been making a lot
03:54of progress in this space.
03:56What I think is fascinating is,
03:57you know, monotherapy PARP inhibitors,
03:58obviously, FDA approved,
04:00It's a synthetic without a success story.
04:02But we see this both in the academic
04:04realm and in the biotech VC realm.
04:06This is really only successful example
04:08that we've seen of this approach.
04:11There's a lot being tested,
04:12but we haven't seen this fully bear fruit.
04:15And so just some thoughts on this.
04:16And so why,
04:17you know why is that we come back
04:19to our therapeutic index curves
04:20that know those who know me well,
04:21we present this a lot of, you know,
04:23it's very simple.
04:23We need to separate the red
04:25and the green curves.
04:26You need a therapy that's going to
04:28selectively target tumor over normal tissue.
04:29I always go back to my other
04:31mentors like Simon Powell,
04:32Chair of Radiation Oncology and Sloan.
04:33During my residency you said you
04:35know without tumor selectivity,
04:36if you're trying to develop
04:38a radio sensitizer,
04:39you might as well just give 10 more Gray,
04:40add that to your prescription and
04:42that's always sort of resonated with
04:43me when I first started my laboratory.
04:45And so a couple of reasons I think
04:47we're we're stumbling a little bit
04:48and we still have more work to do is
04:50I think even though this is obvious,
04:52we need a molecular tumor specific biomarker,
04:55something that's homogeneously expressed
04:56in the tumor cells and not in the
04:59normal tissues and something we can
05:00readily detect and that's that's real.
05:02This is something that Mark O'Connor
05:04who discovered elaporib a good friend
05:06of mine who noticed who notes that you
05:08need to have well separated curves.
05:10You need to have on a clonogenic
05:12survival assay log skill differences
05:15to see to see this translate into a
05:17heavily appreciated population in a
05:20phase one clinic clinical trial setting
05:222nd 1/3 which something we've been
05:23doing a lot is combination therapy.
05:25Often monotherapy is not enough and so
05:27you need to add another drug but you
05:29have to be very careful because you
05:31shift that red curve over but you're
05:32going to drive the green curve over
05:34and you need to make sure that you
05:35don't equally shift them because you're
05:37doing no better than the single agent.
05:39And then finally,
05:40I would also,
05:41we don't have time to go too much into this,
05:43but you need model systems
05:44that will accurately model the
05:46normal tissue toxicities.
05:48The best example is looking at things
05:50like PARP inhibitors combined with
05:52chemotherapy in that heme toxicity.
05:54You don't see that in a mouse
05:56because mice actually don't use HR,
05:58believe it or not,
05:59they use anonymous end jointing in
06:00their bone marrow and you have to
06:02actually do rat tolerability studies.
06:03So you actually get fooled that you
06:04can do a combination therapy that
06:06you can't actually get to the doses
06:08you need to in clinical trials.
06:11So those are just some of the the
06:12quick learnings. As a background,
06:13obviously some of our earlier work
06:15in 2017 was focused on a very unique
06:17discovery of synthetic lethality
06:19of Brachanus phenotype between
06:21onca metabolite producing tumors.
06:23Those are tumors that have IDH one and IDH
06:25mutations interaction with PARP inhibitors,
06:28lot like the Bracha story for breast
06:30and ovarian cancer in elaporim.
06:32We published that story a while back.
06:34We moved that on about a year later
06:36into other ONCA metabolites in
06:37the citric acid cycle like femuric
06:39hydratase and succinate dehydrogenase.
06:41And then we actually spent a lot of time,
06:43this is work with Peter Glazer's lab,
06:45which is great to sort of join
06:46forces with him.
06:47Again in 2020, we broke the mechanism.
06:49We were able to show exactly
06:51how that interaction occurred.
06:53And then in in the years following we,
06:55we tested a lot of interesting ideas
06:57of looking at combination therapies
06:58as I've alluded to earlier that may be
07:01the key for some of these interactions,
07:02PARP inhibitors and ATR inhibitors
07:04and then looking at a lot of different
07:06tumor types like AML for instance,
07:08which we reported on 20/22.
07:11But then we wanted to really
07:12translate this into the clinic.
07:13And before I get to our main story,
07:15kind of want to give you an update on
07:16where we are since we made that discovery
07:17because we've still been working hard at it,
07:19but it's been a long slog what
07:22we what we thought is,
07:23well,
07:23there's a number of different tumor
07:25types here that we go after with
07:27Alka metabolite producing phenotypes
07:28or or citric acid cycle mutations.
07:30We spent a fair amount of time.
07:32We wrote about 7 clinical trials
07:35from about 2018 to about 2022.
07:37I served as either Co I or Co
07:40Pi or collaborator on these.
07:42And we obviously won't go into all these
07:44studies but I want to tell you that
07:46we are learning a lot from these trials.
07:47Paul Leader and colleagues,
07:49we we published an initial case
07:51series looking at IDH mutant
07:53ZINCOMAL sarcomas where we actually
07:55saw about three or four patients
07:57that were exceptional responders.
07:58We amended an existing protocol that
08:00he had to use Olaprip for BRAC and
08:03his tumors without BRAC and mutations.
08:05And as you can see here a very
08:07nice we had a a 14 month sustained
08:10near PR for this patient.
08:11And actually when this patient
08:13progress we're able then amend,
08:15amend another protocol to
08:16put them on a PARP ATR,
08:18inhibit accommodation based on some
08:19of the work from our laboratory.
08:21And that patient actually had a pretty
08:22durable response for quite some time.
08:24We published that in 2021.
08:26We then continued to try to find
08:29off label opportunities to chest
08:30PARP inhibitors in these onco
08:32metabolite producing tumors.
08:33This is an example of an SDH deficient
08:36gist that we saw in Pete's tumor board
08:38and Charles Singh and Juan Vasquez
08:39and actually Forzano Prezankar should
08:41also mentions the senior author here.
08:43We actually had a pretty remarkable response
08:46looking at temozolomide combined with
08:48olaparib in this particular case here.
08:52More recently though,
08:53we've now really embraced this
08:55combination approach looking at
08:56alkylator combinations like TMZ
08:58combined with PARP inhibitors.
09:00And this is a trial we started a
09:02few years back called ABC 18 O1.
09:04And I just want to give you a little
09:06update on where we are with this as
09:07well because it kind of gives you a
09:09sense of what it's like when you try
09:10to translate work from the bench,
09:11the bed size.
09:12So many people are involved in
09:14this type of work on this trial is
09:16recurrent IDH mutant gliomas adult
09:18and we also have a pediatric version
09:20of this with Asher marks looking at
09:22the PARP trapping PARP inhibitor BGB
09:24290 and using a full dose of the BGB
09:27290 with a very low dose of Tamizola
09:29and almost as a sensitizer to BGB 290.
09:31And this is a was a phase one two
09:33trial that broke out into three
09:35cohorts of which should find as
09:38porous patients that are alkylated
09:40refractory alkylator naive patients,
09:42R&amp;B and the exploratory GBM.
09:44All these are IDH Newton gliomas.
09:47We built a lot of exciting sort of
09:50correlates correlatives and oratory studies.
09:52For instance,
09:52we built a phase zero trial here we're
09:54able to actually biopsy patients and
09:56and look for drug levels and enhancing
09:58and non enhancing disease and also
10:00show you some some cool stuff we did
10:02looking at MRI velocity studies to
10:03track tumor growth because as some
10:05of you know these IDH mean tumors
10:07grow rather slowly in in patients and
10:09can be difficult to track responses.
10:11So just share a little bit of an update
10:13just because of time but I want to
10:14kind of give you a flavor of where
10:15we are on on for this stage of the project.
10:18So this is the the phase zero trial
10:20we we we wrote it was really exciting
10:22to sort of get this off the ground.
10:23It's one of our first phase zeros
10:25we've done and as you can see
10:26we were able to give the drug at
10:28a full dose 60 milligram BBID,
10:30the PARP inhibitor without the
10:31tamizole mod for seven to 10 days.
10:33And then we're able to take the
10:35patients to the OR and essentially
10:36biopsy not not only enhancing
10:38disease but non enhancing disease
10:40to look at whether we truly had
10:42blood brain barrier penetration of
10:43our drug and then obviously blood
10:45for normalization or comparison.
10:47We developed a protocol to
10:50develop BGB to detect BGB 290.
10:52This was Jing Lee's group at Wayne State,
10:54the Carmanos Cancer Institute
10:56to to detect the BGB 290.
10:58And I'll just show you some of
10:59the report results that we just
11:01reported at Snow an updated version
11:03and published recently.
11:05We can see here it was quite nice.
11:06We could see that we could detect high
11:08levels of total pimiprid that's BGB 290.
11:10The green circles are in
11:12non enhancing disease
11:13and then the red is the enhancing
11:15tumor and then we can detect,
11:17detect substantial levels
11:19of unbound drug shown here.
11:21And when we calculate KPUU we
11:23get these numbers shown here.
11:25It's a little skewed because
11:26of the plasma numbers.
11:27So we sort of normalize in the far right,
11:30but this is actually the first study
11:31to show that both enhancing and on
11:33enhancing tumor we can detect the
11:35part where neighbor BGB 290 which
11:37is thought to be seen as penetrant
11:39but was not known as only project
11:41modeled in the animals to to
11:43penetrate the blood brain barrier.
11:45Another flavor of this,
11:46because we're still continuing
11:47to now look at responses,
11:48I can tell you today is we don't
11:50have you know overwhelming responses
11:52with this combination therapy,
11:54but we do have outliers that are
11:56actually seeming to have some response.
11:58And as I mentioned earlier,
12:00IDH main tumors are fascinating
12:01because they grow incredibly slow.
12:03And as if anyone knows from the
12:06audios trials that recently are now
12:08an NDA filed for the IDH inhibitors,
12:10these tumors actually have to be
12:11tracked over time because they
12:12often don't shrink.
12:13It's more disease stability
12:14that you're looking for.
12:15So we worked with Ben Ellenson
12:17who's developed an MRI velocity
12:18protocol working in the company
12:20called Neosoma where they can
12:22automatically segment the enhancing
12:23and non enhancing areas of disease.
12:25Then volumetrically average and you
12:27can essentially get three prior
12:30Mris and then three Mris after
12:32treatment and then get a a sense
12:34of disease of velocity again.
12:36We just reported these results at Snow
12:38at the Neuro Oncology Conference last year.
12:40And what you can see here,
12:41looking in arm A,
12:42this is the alculator refractory patients.
12:44I know it's a little bit small here,
12:46but you can see three patients,
12:48an example.
12:48It's pretty clear that you had progression
12:51and then either disease stabilization or
12:53arguably some amount of disease reduction.
12:55And we saw this in the Arm B,
12:57the the arm B as well.
13:00Again,
13:006 cases here were other disease stability
13:03or disease regression and also in the
13:06exploratory cohort of recurrent GBMS.
13:08Again, so this is A work in progress,
13:11but was real, real fun to work with.
13:12Ben and his group at UCLA just sort of
13:15track some of these tumor responses.
13:16So just wanted to kind of give you sort
13:18of a sense of where we are with that,
13:20with that work,
13:21as it's been a labor of love for us in
13:23terms of trying to translate that work.
13:25But I really want to focus on more of
13:27our recent discoveries in the laboratory,
13:29which is we've been really,
13:31really excited about South for this.
13:33We need a little bit of background.
13:34So I'm a GB M radiation oncologist
13:37by training.
13:37I went into CNS because there's only CNS,
13:40right.
13:40So only three trials to know
13:41for gliomyemics in clinic today.
13:43And it's very,
13:44it's always nice,
13:45not like pathology where you
13:46have like hundreds of papers
13:47you have to memorize.
13:48But you can see here there's 333 key trials.
13:51The main one here is the Walker trial 1978.
13:55The Walker, Walker and colleagues actually
13:58randomized patients to surgical resection
14:00for GBM followed by adjuvant chemotherapy,
14:03sorry followed by adjuvant radiation or
14:06adjuvant alkylator chemotherapy BCNU,
14:08the structure shown there which will be
14:10important or the combination of both.
14:11This trial put radiation oncology
14:13in the map in the CNS space.
14:14Essentially all patients now from then
14:17on received post OP radiation therapy
14:20for GBM and then we had about a million
14:23negative trials until about 2005.
14:24And then we had the pivotal trial.
14:26This was the STOOP trial which essentially
14:28tested a different alkylator, A,
14:30a, a more selective mono functional
14:33alkylator called temazolamide and
14:34looked at whether adding it concurrent
14:37with temazolamide with radiation
14:39rather and adjuvantly Temazoma would
14:42have a survival benefit.
14:43And they're even though this
14:45doesn't look very impressive,
14:46this is actually a was a pretty pretty
14:48big result for the glioma field because
14:51we've had so many failures in the past.
14:53A follow up study based on this
14:55trial though made a key distinction.
14:57It was the patients that had tumors
14:59with silenced methylone E methyl
15:01transferase and this is ADNA repair
15:02protein that we'll talk about in a moment.
15:04When you're stratified by that you get
15:07a huge difference in overall survival.
15:09So this,
15:10this real,
15:10this paper from Hagee ET all that that did
15:13the analysis from Roger Stoops data really
15:15established that MGMT was a biomarker.
15:17I would actually argue if you
15:18talk about synthetically valid,
15:19this is one of the first synthetic
15:21lethal interactions that had clinical
15:23significance where tamazolmide
15:24works in MGMT silence tumors.
15:26But we could argue that for for hours.
15:29So now delving a little bit
15:30deeper into the laboratory side.
15:32So why? Why is this the case?
15:35So I've been fascinated ever since I was a
15:37graduate student here in the early 2000s.
15:38Molecularly,
15:39what we understand is that temozolomide opens
15:42up into this methyl diazonium ion shown here,
15:45and it methylates the O six
15:47position of guanine.
15:48Now only about 5% of the damage
15:50from TMZ is actually at the O six.
15:52It's actually all over the bases,
15:53but this is the important one for
15:55the cytotoxicity that you see
15:57with MGMT SCIENCE TUMORS.
15:58So first,
15:59normal cells express high levels
16:00of this enzyme called methyl
16:02guanine methyl transferase.
16:03It's a suicide enzyme.
16:04It plucks that O 6 methyl off and
16:07other potential lesions as well.
16:09But if you don't have MGMT what
16:11you get is during replication
16:12with about 90% bypass efficiency,
16:14the polymerase will drive across
16:16that O 6 methyl guanine and it
16:18will miss pair with thyme.
16:19This activates mismatch repair
16:20which basically says you've made
16:21an error to the polymerase.
16:23Polymerase does the same thing over
16:24and over again and you get this
16:26thing called futile cycling which is
16:28very fascinating because essentially
16:29the the at at one point mismatch
16:31repair essentially realizes cannot
16:33actually, you know change
16:35polymerase's errors.
16:36I'm sure it doesn't actually do it
16:38that dramatically but it stimulates
16:40apoptosis and this actually accounts
16:41for the cytotoxicity even though it's
16:43a very small amount of the damage.
16:45And what we now understand is that
16:47over the last three or four years
16:49quantitatively we can say that
16:51large fractions of of tumors that
16:53respond to TMZ that have MGMT.
16:54Silence promoters.
16:56They simply just knockout
16:58mismatch repair again.
17:00We knew about this in the
17:0190s and the early 2000s,
17:02but it was really these two papers
17:04I'm presenting here in the last few
17:06years that have quantified this.
17:07And so just to show exactly
17:09what's happening is you're
17:10simply turning off the firewall.
17:11So now you're knocking out mismatch repair.
17:13You're actually creating mismatches,
17:14all of the genome genetic instability,
17:16but there's no fire alarm to to
17:19trigger apoptosis or futile cycling.
17:21Sort of like a tree following
17:22the woods if no one's around.
17:23Does it really make a sound?
17:25So if you're sort of just checking e-mail,
17:28just the the take home message here.
17:29TMZ works well on MGMC.
17:31Silence, tumors OK,
17:32but resistance emerges and I'd say in
17:35in half of all gliomas it's actually
17:38clonally driven mismatch repair mutations.
17:40So for us this was a sort of multi
17:42forks in the road where when we saw
17:44this in in the clinic this is actually
17:46an example we're seeing patients,
17:48we said well this is fascinating,
17:49these patients still have silence
17:52MGMT and they've acquired A mismatch
17:54permutation TMZ is not working.
17:55So how could we fix this?
17:57Can we come up with a new therapy that
17:59still exploited the MGMT biomarker
18:00and we first thought about creating
18:03alkylators that could be MMR independent.
18:05So just let's just make superpotent
18:07alkylators that would just crush anything
18:09that walked and well we've done that before.
18:12So the Walker trial,
18:13I showed you BCNU back in the day,
18:14non selective alkylator, Kintara,
18:16this is a drug called VALO A3
18:18reformulation of an old drug from the 70s,
18:21non selective alkylator.
18:22Again this actually just failed in the
18:24GBM agile phase three trial and a drug.
18:27We'll talk about Azelastone in a moment.
18:28These are effectively steamrolling the DNA.
18:30This is my 4 year old's steamroller
18:33to be dramatic there we also thought,
18:36OK,
18:36so we're not going to do that
18:37because that's going to not going
18:38to have that therapeutic index
18:40constraint that we really wanted.
18:41So what about reactivating MMR?
18:43That could be cool.
18:44We actually had a a pitch competition
18:46maybe Demetrius remembers the pitch
18:48competitions back in several years
18:50ago to to start a company around this
18:51and and this got shut down because
18:53there's too many mutations that you
18:55can't you know custom reactivate
18:56mismatch repair if there's mutations
18:58all across the open reading frame.
19:00But we landed on this idea of creating
19:03alkylators which are MMR independent
19:04but would retain the MGMT dependence.
19:07So harkening back to this
19:08therapeutic index curves that I
19:09talked about earlier.
19:10So the red and the green curves
19:12because we know these tumors still
19:13have silenced MGMT in the cases that
19:16acquire the mismatch repair mutation.
19:18So with that we turn to Seth Herzan.
19:19We had been working together in the
19:21past and we got our band back together.
19:24Seth is actually a really good guitarist
19:26and we haven't played a show together,
19:27but I'm still playing drums trying to quit
19:29my day job, which I because I'm here today.
19:31I haven't been able to do it yet.
19:33But so in 2018 came to Seth and I said,
19:36listen, you know, we've been working on
19:37some DNA repair inhibitors a while back.
19:39So let's let's figure out a way to
19:40come up with these new molecules.
19:42We called it Secret name with
19:43Crowbar into O 6, the CEO 6 project.
19:46You know,
19:46at the time I was doing the IDH parp story,
19:48he was working on all his crazy antibiotics
19:50that he makes and publishes in big journals.
19:53And it all centered around a simple question,
19:55which is, can we change the R group here?
19:59And again, I'm not a chemist,
20:00my dad's a chemist.
20:01I did not inherit his chemistry gene.
20:03But it certainly was very
20:04interesting in the idea of, like,
20:06there's got to be something that we could
20:08put on here that could be removed by MGMT,
20:11right?
20:11But.
20:12But independent of mismatch
20:13repair status for cell killing.
20:15OK.
20:16So at that point, I'm very,
20:18very grateful to have talented
20:19trainees come in the lab.
20:21Kingston Lynn who's here today.
20:22Susan Gable at Thinkston Clinic
20:24in a cycle of life because Susan
20:26Gable did her PhD in Peter Glazer's
20:27lab and then came to my labs.
20:29And so these two folks came in and
20:31we had a a mission for both of them,
20:34which is essentially, you know,
20:35put in our group.
20:36That's not a methyl,
20:37that's more complex that can
20:39be removed by MGMT.
20:40But kill cells independent our status
20:43and semi jokingly told them you know
20:46let's look at this plot and sort
20:47of walk them through it and said OK
20:49look at the green curve you get MGMT
20:51silence mismatch repair proficient.
20:52Very nice to sell killing with TMZ.
20:54OK yellow line is when you knock
20:56out mismatch repair you see the
20:57drug now looks invisible and then
20:59if you have MGMT doesn't matter
21:00because the methyl group gets ******
21:01** and we sort of semi joking and
21:03said make some molecules and do
21:05not call us back until the yellow
21:07line drops below the green.
21:08Do not bring the black or the blue line down.
21:11Obviously we Kingston knows we're
21:13always available as mentors but they
21:15made about 100 molecules and one of
21:17the molecules fulfilled the criteria.
21:19I still remember and I'm sure Kingston
21:21and Cesar remember the day that
21:23they emailed us about this molecule.
21:24Literally isogenic cell lines
21:27selectively killing cells based
21:29on MGMT status but independent
21:31of mismatch repair status.
21:33Really, really nice sort of
21:35potential therapeutic index here.
21:37And so just to dive into some of the data,
21:38we spent a lot of time where we had
21:40that phenotype and we didn't really
21:42understand exactly what the mechanism
21:44was and we're going to get to the
21:46chemistry a little bit towards the end.
21:48And so Susan came in and started doing
21:50a lot of DNA repair functional assays.
21:53And what you can see here she did
21:55the neutral comment simply asking,
21:56you know, are there an increase
21:58in DNA double strand breaks in the
22:00background of the double negative cells.
22:02And so first I'll just walk you through
22:04here's TMZ and then you know it's a
22:06little bit of a good figure minus plus,
22:08that's MGMT minus miss,
22:10mismatch appear proficient,
22:12you can see that.
22:12So you get you get double strim
22:14breaks and we had published on this
22:15a few years back that made sense.
22:17But with the KO50 molecule,
22:18the minus plus and the minus minus
22:20you don't really get a a huge
22:21difference in double strim breaks.
22:23And note that when you knockout MGMT
22:25and mismatch repair here with TMZ
22:27the drugs invisible in terms of the
22:30damage and that's a futile cycling
22:32inducing strand breaks and apoptosis.
22:34But here clearly not double
22:36strand break clear mechanism.
22:38She then actually pulled the paper
22:40from the mid 90s and this is what
22:42I love about trainees working with
22:43trainees has pulled this out herself
22:44in an old cross linking comet assay
22:48and this is how this assay works.
22:50So here's some of you don't know.
22:52You take nuclei,
22:53you put it in agros gel and you
22:55run it across electric field
22:56and if there's DNA breaks,
22:57they'll sort of fade behind it like a comet.
23:00And so in this assay drug
23:01induced DNA double strand breaks,
23:02single strand breaks will induce a comet.
23:04IR induced breaks will induce
23:06an even bigger comment.
23:08But if you have a drug that has
23:09a cross linking effect, OK,
23:11it and you, the way we add, you know,
23:14we add these drugs and then treat,
23:15it'll actually stitch the DNA up,
23:17it'll prevent that comment from forming.
23:19OK.
23:19And this was used in the Fanconi
23:21anaemia days.
23:22So very,
23:23very nice assay to consider.
23:24So I'm going to walk you through
23:25what she found because this basically
23:27gave us our first clue on exactly
23:29what was happening.
23:30So let's just walk through the
23:310 Gray shown here.
23:32So DMSO no, no Comet Tail Mitomycin.
23:35See, that's a cross linking agents, you know.
23:37So no, no effect there.
23:39TMZ induces single tram breaks
23:41and some double tram breaks.
23:42You can see a significant effect there.
23:45And then KL50 some damage but
23:47but not as significant.
23:48Now when you add the the radiation sort of
23:52probe for the potential crossing in activity,
23:54you can see here that when
23:56you add mitomycin C,
23:57when you go to high doses
23:58you're stitching the DNA out,
23:59OK. And so now you're unable to create
24:02that common tail and then TMZ as expected,
24:05no cross linking activity. But then KL50,
24:07we started to notice an appreciable
24:09sort of restriction of that comment.
24:12The pictures are there, shown on the right.
24:13So this is the first realization that this
24:16is probably a cross linking mechanism,
24:18but selected to the MGMT minus background.
24:22She then went on and did some DDR foci assays
24:26and I'll just sort of gloss over because the
24:28time won't be able to go through all this.
24:30But look at the red shaded box here again.
24:33What was great about doing this
24:35isogenically modeling each MGMT
24:36versus mismatch repair status.
24:38You know the red shaded box here you
24:40can see basically DMSO versus TMZ.
24:43Again, the drug is invisible,
24:45damage is being induced,
24:46but it's not being detected.
24:47We can see KL50,
24:48now we're getting induction DNA damage.
24:50You can follow that kinetically over time.
24:52You can see or by 96 hours TMZ versus KL50.
24:56The yellow line,
24:57that's the MGMT deficient,
24:58double negative mismatch for deficient.
25:00You can see that that in the
25:02KL50 get induction damage.
25:03And then what I think is really cool is we
25:05then segmented the nuclei by DNA content,
25:08just geometric averaging and segmentation.
25:11You can see here this is done actually
25:13with the CMD when you look at G1S and G2.
25:16What was really cool here is you actually
25:18get damage in every phase in the cell cycle.
25:20So this actually suggests that the the
25:22cross linking activity that we think is
25:24there based on the on the comet assay
25:26and then the cell cycle profile suggests
25:27that doesn't require replication.
25:29Even so this is not a futile cycling
25:31mechanism unlike TMZ for the real DNA
25:32repair efficient out this is actually
25:34phosphor RPA which is interesting
25:36because our patient not be loading in
25:38G1 or G0 just at a conference over
25:40the weekend in Cancun DNA repair
25:43conference and someone had to do it
25:45and went there and actually spoke
25:47to some people about RPA can form at
25:49single strand DNA in G0G1 cells and
25:52so mechanistically really cool stuff.
25:54So now what about the chemistry.
25:55So going to gloss over a fair amount
25:58of Kingston's thesis just because of
25:59time today and I know it presented
26:01elements of the story before,
26:03but Kingston and Seth Herzan really
26:06dove in the chemistry up on sign sale
26:09with Susan sort of back stopping on
26:11the DNA repair mechanistic studies.
26:13And this is basically what they
26:15figured out and they taught me to
26:16say when you talk about chemistry,
26:17say things like 4A4B and 4C,
26:20very chemistry savvy.
26:21SO4A is the molecule.
26:24So Kale 50 is a fluoro ethyl
26:26instead of a methyl and so Team
26:28Z would have the methyl here.
26:29Very simple but elegant change
26:32that fundamentally changes
26:33the way this molecule acts.
26:35You get a ring opening which then
26:37creates this fluoroethyl diazonium
26:39which is this reactive intermediate.
26:41This fluoroethyl diazonium then attaches
26:42to the O six position of guanine and
26:45what Kingston and and Seth and others
26:47have shown is that MGMT can remove it.
26:49So the old name for MGMT is
26:51alkyl guanine transfer AGT.
26:52So it's not just a methyl
26:54guanine methyl transferase,
26:55even though we think of it just as
26:58MGMTMGMT can remove this pretty readily.
27:00But what what's fascinating is
27:02Kingston pulled out some belief
27:03papers from the 80s or 90s.
27:05If I remember where he found some
27:06oligo studies where they looked at
27:08fluoro ethyl addicts of the O 6:00
27:10and had some literature precedents
27:11that it was actually forming
27:13an ethanol guanine intermediate
27:14which would be highly unstable,
27:16could possibly cross link.
27:18And that's exactly,
27:19you know you know blowing past a
27:20lot of the work that that he and the
27:22Horizon Laboratory did to prove this,
27:24but essentially show with a very
27:26slow T 1/2 this actually forms
27:28ethanoguanine intermediate molecule
27:296 and then cross links with this
27:32adjacent cytosine as you can imagine.
27:34Now this is MGMT dependence.
27:36So if you don't have MGMT you
27:37you have the slow,
27:38slow reactive process but not
27:43dependent on mismatch repair activity.
27:45So very, very nice mechanistic studies there.
27:48But now I sort of want to,
27:49you know,
27:50step back to the clinical data because
27:52this is something that had always
27:54vexed me even during residency and as
27:56an attending is as I showed you earlier,
27:59if you look at the Walker trial in 1978,
28:00well they, you know,
28:01they did an alculator, they used BCNU.
28:04And so why didn't a cross
28:06linking Alculator work back then?
28:07You know, could it be,
28:08you know, you know what what,
28:10what are the factors that the trial was
28:11negative because the Stoop trial in
28:13the far right was positive with Alculator.
28:14But I told you it's a mono functional
28:17alculator that doesn't cross link.
28:19And we actually spent a lot of time I
28:21think my first five years in attending,
28:22I was delving the literature.
28:24I was obsessed with the idea that
28:26Oh well it's because BCNU is given
28:28every six weeks and so it's not
28:30given during fractionated radiation.
28:31So they missed the opportunity
28:33for radio sensitization.
28:35We spend a lot of grant money on that,
28:36not true,
28:37doesn't really matter and
28:39so we'll we'll go into that.
28:41But if you actually look at the
28:42chemistry and I really you know
28:44thank the trainees for teaching Seth
28:45Roson for teaching me this stuff.
28:47But if you look at the these
28:48reactive chlorine, these are very,
28:51very efficient non selective cross
28:53linking sort of payloads as we'll call them.
28:56Compare that with this methyl here,
28:57which is essentially non cross
29:00linking MGMT dependent.
29:01Futile cycling is the mechanism.
29:03And actually as I we started to dive
29:05into this literature around the
29:07time that we started this project
29:08started going deeper
29:10and it turns out between
29:12and actually became friends with Rose Stoop
29:14who's now a mentor and A and a friend.
29:16And actually he talked to
29:18to Roger about this as well.
29:20There was a in between around the time
29:22of BC and U and before the Stoop trial
29:24there were actually other tamizolamide
29:26precursors that were tested in the
29:28clinic that kind of give a little bit
29:30of sense of what likely happened.
29:32So DTICI think many of you may know is
29:35the carbazine we use this in Melanoma,
29:37but there's actually a drug called azolastone
29:39that was developed before tamizolamide.
29:41And I'll just have you note there the,
29:44the chlorine,
29:45the similarity here very reactive
29:47drug and this was actually brought
29:49into the into phase one and two
29:51clinical trials before temozoline
29:53made its debut into the clinic.
29:55And actually just to sort of get,
29:56you know, going to the rabbit hole,
29:57please not TMI or too much information here,
30:01but this drug was made by actually a post
30:03grad named Robert Stone and Aston University,
30:05Aston University and in the UK,
30:08hence the name Azelastone.
30:09And it has this sort of chlorine
30:11reactivity that I mentioned earlier.
30:13And again, very interesting
30:14because the molecule at Kingston,
30:16it actually looks quite similar to this,
30:18right,
30:18But it only only differs by flooring.
30:20The chemist will say,
30:21well this is important because flooring
30:23is actually a very poor leaving group
30:25and some of us remember that from
30:28college level chemistry and when
30:30Robert Stone made this molecule they
30:31actually tested this in animals.
30:33They actually flatlined a number
30:35of different tumor models.
30:37And actually this book that that
30:39Kings and I read about this,
30:40they actually made a poster called
30:42Azolastone the movie because they were
30:44going to cure cancer with this molecule.
30:46This was going to be the alkylator
30:48of all alkylators.
30:49But alas it went into multiple
30:51phase one trials.
30:51The drug was also called Monozola
30:53line and actually failed in the mid
30:5580s because of dose living toxicity.
30:56They tried multiple scheduling
30:58regimens and and and then soon
31:00after Roger Stoop came on board,
31:01picked up TMZ and then ran
31:03that into the stoop trial.
31:04And so it's fascinating for me when we
31:06think about this and this competitors
31:08made this this really funny poster to
31:09make fun of them for failing I guess
31:11back in the day it was a lot more fun
31:14And if you think about it clinically,
31:15so in the clinic we use lomastine
31:17to salvage patients when they failed
31:19team azolamide or if they just you
31:21know recurrent glioma patients and
31:23again not a chemist but the red shaded
31:24box will show you the chlorine.
31:26Again similar warhead here,
31:29highly reactive and interestingly even
31:31though we salvage patients with Lumosity,
31:33it really has no survival
31:34benefit in recurrent glioma.
31:35This is something that we
31:37struggled with for a while.
31:38And so I would argue that or we
31:40would argue rather that this is
31:42again a therapeutic index play.
31:43The slow cross linking activity of
31:46KL50 with the MGMT dependency sort
31:49of possibly makes it the the best of
31:51both worlds in terms of having more
31:53DNA damage that's MMR independence.
31:55It doesn't fall prey to a mismatch
31:58permutation which I think is the key.
32:00So to get at this we looked at this,
32:02we went back from serve clinical
32:03observations and went back to the Herzon
32:05laboratory and also our laboratory
32:07to look at this little more deep
32:08because again this is all just sort of
32:11hearsay without some preclinical data.
32:13And so Eric Kuzman and and Kingston
32:15and folks in Herzon lab actually
32:17then measured the rate of ICO.
32:19Interesting cross link formation
32:22using a very nice elegant technique
32:23which I I won't go into because I
32:25couldn't do justice to us this was
32:27just is it hopefully about to be
32:29published and deposited in chem RVX.
32:30We can see here looking at the floor
32:32out the with this cross linking
32:33assay in vitro you can see indeed
32:35very slow cross linking activity.
32:37OK, so if you have MGMT,
32:40arguably if a normal solo has MGMT,
32:43there will be time for it to pluck that off.
32:45By contrast, if you look at the chloroethyl,
32:48that's the mitozole, my version, or the CCNU.
32:50Arguably you can see very rapid cross
32:53linking activities T 1/2 of 6.3 hours.
32:56So this likely is consistent with the idea
32:58that if you're cross linking too quickly,
33:00you're not going to have.
33:01Even if MGMT can get to
33:03that lesion and remove it,
33:04it's unlikely to have as much of
33:06A therapeutic index as something
33:07like a fluoroethyl that has a very,
33:09very slow T 1/2.
33:11And again,
33:11I'll I'll note that this is really
33:13the 1st for this KL50.
33:15It's the first time this molecule has really
33:17ever been described by Kingston and Susan.
33:20So then we brought that chemistry
33:22observation back to our laboratory and
33:24we just like to do the thing we do,
33:26which is cloning survival assays over
33:27and over again in Isagenix cell lines.
33:29Some people hate us when we do this,
33:30but we think it's important.
33:32So look at TMZ.
33:33I've walked you through that date again.
33:34The methyl group futile cycling, right?
33:36The green line becomes invisible
33:38with the yellow line.
33:40Then look at kale 50 again,
33:41very nice therapeutic index here.
33:42You're going to very high doses,
33:44200 micromolar kale 50 and you're not
33:47killing anything that has MGMT intact.
33:49Now let's compare CC and U and
33:51mitozolamide and indeed you can
33:53look at azolastone and you you do
33:56get MMR independent cell killing.
33:58But it's that therapeutic index.
34:00We would argue it's it's with both
34:02myzolamide and CC and U very potent alkalis.
34:04But that window is narrow and
34:06some people say,
34:06well then why don't you just dose
34:08the patients 150 micromolar 100.
34:10It's not easy.
34:11As many of you know in a clinical trial
34:13getting those doses right across a
34:15very heterogeneous group of patients,
34:17you're going to need this
34:18wider therapeutic index.
34:19So it's red in the green.
34:20Curves need to be far apart.
34:23So just to summarize for
34:24this part of the talk,
34:26you know what we believe is happening.
34:27This is actually a slide summary from
34:29Susan Gable who now has her own lab
34:31here at Yale looking at tamozolamide.
34:33Again, futile cycling,
34:35removed by A rapidly removed
34:37by MGMT expressing cells.
34:38Futile cycling then induces tumor
34:42cell death in the absence of MGMT,
34:45but requires MMR proficiency
34:47when you knockout MMR.
34:48The lesion,
34:50essentially invisible mitazolamide,
34:52the chloroethyl,
34:52So very fast acting forms at Athena Guanine.
34:55OK and is MMR independent but
34:59has an MGM is is less dependent
35:03on MGMT status and then KL50,
35:06which we would say is sort of the
35:08possibly the Goldilocks phenomenon
35:09but has has the best of both worlds.
35:12OK and I'll just show you one example.
35:15Because this date has been published,
35:16I'd like to move on to sort of some
35:17of our more recent unpublished work.
35:19This works incredibly well.
35:21We sent this date we sent these
35:23molecules to Jan Sarcoria the at the
35:25translational brain tumor Center at the
35:27Mayo Clinic and asked them to compare.
35:29Let's look at TMZ lomastine and
35:31KL50 and let's look at intracranial
35:33GBM xenographs that have acquired
35:35this this aggressive phenotype MGMT
35:38science mismatch period efficient
35:40and you can see here on the left as
35:42expected and this is aggressive model.
35:43All the animals 30 days the vehicle
35:45are are dead Temazolamide is invisible
35:47under these conditions lomusting we
35:49wouldn't expect it to work as that
35:51therapeutic index issue in this experiment,
35:53no efficacy here on the right though.
35:55You know we've been doing these
35:56types of experiments for about 12
35:58years in our own lab and pretty
36:00remarkable efficacy here.
36:01This is an 8 fold improvement in
36:04overall survival as a monotherapy
36:05or again TMZ has no effect.
36:07So really,
36:08really excited about the in vivo data that
36:10really is building the story of of this,
36:12this MGMT dependency and the mismatch
36:15repair independence could have
36:17some some potential therapeutic
36:19implications because that's sort of
36:20a summary of the initial discovery.
36:22So where are we going from here.
36:24So one of the things that we're
36:26interested in is it turns out
36:27that MGMT promoter methylation,
36:28we only talk about that in
36:30like CNS tumor board.
36:31We never like to think about this that
36:32MGMT could be silenced in other cancers.
36:34It turns out that subsets of all cancers AML,
36:38colon,
36:38sarcoma and lung,
36:39they all actually have silence for
36:41whatever reason have have subsets
36:43of cancers have silence MGMT.
36:45And so get at this.
36:46The team did a use the Prism
36:48screening platform up at the broad.
36:50I'd encourage anyone who's
36:51interested in using this platform.
36:52It's a really cool pooled bar coded
36:56drug screen $10,000 a molecule and
36:59you actually compared to some of their
37:01existing data and what not and it's
37:03basically 930 cell lines across 45
37:05lineages. You essentially send your drug
37:08up there and what they'd said is OK,
37:10let's ask, let's look at KL50 and
37:12look and they've got all the genomic
37:13data or a seat data that you can
37:15correlate with it and say are there.
37:17When we treat with KL50,
37:18are there any specific genomic
37:21biomarkers that correlate with
37:22sensitivity or resistance and MGMT
37:24was the true correlate was the
37:27reproducible correlate for KL50 activity.
37:29And when you break it out by different
37:31cell of origin types for the cell lines,
37:33you can see MGMT low is in the
37:36orange and MGMT high is in the green.
37:38You could see across the board all
37:40different tumor types when you have low
37:42MGMTKL 50 is significantly more active.
37:46So this prompted us to move on and this
37:48is work that as Susan finished up her MD,
37:50PhD in our laboratory or sorry her
37:52sorry her residency in our laboratory
37:53and then went on and started her own
37:55lab and we're now our two laboratories
37:57are collaborating.
37:58She went and started sampling PDX
38:02libraries across a number of CROs
38:05and academically.
38:05We were able to find a number
38:07of of models here.
38:08These are all different tumor types,
38:10some that have lost mismatch repair,
38:12some that have lost MGMT or both.
38:14Focusing in on 2 examples here shown
38:17here and you can see in this case
38:20this these two models both silence
38:22MGMT one loss MLH one and one loss
38:25MSH 2 and again pretty remarkable
38:27data for monotherapy efficacy.
38:29Getting back to the the idea of
38:31like if you've got data like this,
38:32this is the type of stuff you want to try
38:34to move in the clinic because there's
38:35a chance we could see an efficacy in a
38:37heavily pretreated phase one population.
38:38You can see here this is looking at
38:41Melanoma model and a lung model.
38:44Again TMZ versus KL50.
38:46TMZ as expected invisible under
38:48these in this tumor genotype,
38:50very nice tumor growth delay with
38:52KL50 and and I'll just know this is
38:533 doses times 3 cycles and then we
38:55stop dosing and then you go out to
38:57day 80 and we've got essentially
38:59sustained tumor regressions and we see
39:01this also for lung cancer as well.
39:03We've since been now moving on
39:04to different tumor types.
39:06I talked to you about our interest
39:08in AML and started a collaboration
39:10with Stephanie Helene and what you'll
39:12see here is in is a petite in our
39:14laboratory post doc started modeling
39:16this in a number of different AML cell lines.
39:18Pulled some of that data from the
39:20PRISM screen, the 930 cell line data.
39:23And you can see that when you look at
39:26TMZ and KL FIT MGMT low versus high
39:28you can see largely very nice correlation.
39:32And he's now what he's been doing
39:33is doing the same thing we did
39:35earlier with the other models is
39:37doing isogenic knockouts now asking
39:38the question of you know MGMT status
39:41versus mismatch repair status.
39:42And you can see here in these
39:44models now you get a very nice,
39:46this is KL50 in a in a molem 13 AML
39:48model where we knockout mismatch
39:50repair and then in MGMT deficient
39:52that's the red and the blue shown here
39:55and you see very nice activity here.
39:57We're now working with Stephanie,
39:58don't have the data to show today
40:01looking at her PDX models because
40:02we actually think there could be
40:04a potential to use some of these
40:05molecules in the AML setting,
40:06the subsets that have silenced MGMT.
40:11So really in the last you know five
40:14about 10-10 minutes or so sort of talk
40:17about where we're going from here.
40:19So the first thing which is really
40:21interesting is we had this molecule KL50
40:23never really been described before.
40:26We want to translate this in the clinic
40:27and we had you know gotten really lucky
40:28with the IDH PARP story because there
40:30was FDA approved PARP and intervals and
40:31we just needed to call those companies,
40:33write the trials and then and
40:35then run them here.
40:36There's really no source of KL50 and
40:39we've started a few companies before
40:41this so we sort of knew how to do this.
40:43But ultimately to to cut to the
40:45chase here we ended up just spinning
40:47out our own company and this was
40:49great to work with Kingston who as
40:51a as a MDPHD student and then Seth
40:53Herzan's Co Pi and then my long time
40:56business partner Kevin Ragan.
40:57We had a nice write up and end points
40:59about two years back and this is just a
41:01glimpse of the founding team and right
41:03around that time Kingston very proud.
41:04It's Forbes 30 under 30.
41:07I'm still waiting for 50 under 50,
41:08but I don't think they're going
41:10to have one it's too but so,
41:13so what's the company doing.
41:14So it's been really great.
41:16So,
41:16so really the company is now
41:18taking that tool compound KO 50
41:20and really now engineering it for
41:22ready for prime time so to speak.
41:24Turning that into what we
41:25call development candidate,
41:26some of you know what that means,
41:28but essentially suitable for
41:29Ind enabling studies.
41:30Didn't want to focus too much on
41:32this because of you know it's
41:33more company related stuff,
41:34but it has the original molecules,
41:36metabolic liabilities that preclude
41:37it from going to the clinic.
41:39We welcome anyone who reads
41:40the paper to try to do that.
41:41It's not possible.
41:42So we've been able to engineer
41:43that molecule and we have that.
41:45We have a new version of KL50 that
41:47has very good PKPD properties
41:49enhancing its penetration and whatnot.
41:51And the company hopes to over the
41:53next year perform the necessary
41:54ID enabling studies to drive this
41:56into the clinic.
41:57And we hope if all goes according to
41:58plan that we can actually bring this
42:00into patients about a year from April,
42:02which would be really, really exciting.
42:04But we have to close our Series A first,
42:05which is going to be still a bit of a path.
42:09And what's really exciting is
42:10we think that we can use these,
42:12this molecule KL50 for tumors
42:15outside of the brain.
42:17We really think there's a potential
42:18here to move this into things like
42:20colon cancer where 30 to 40% of tumors
42:23are MGMT silence and we've modeled this,
42:25this is just an example mod 16,
42:27this is sort of a a next generation
42:30kill 50 before our our development
42:32candidate called mod 246.
42:33But you can see here mod 16 in a colon
42:35cancer model, a flank model MGMT science Mr.
42:38prepared efficient very nice dose dependent
42:41activity 10 Meg per keg treatment regimen
42:45here inducing A tumor regressions.
42:47And we're excited about this because
42:49we've been working with folks that
42:51you know well like Kirk Schopper
42:52pathology and Mike Tuccini already
42:54looking at whether we can do alkylator
42:56DNA repair inhibitor combinations.
42:58For example in MGMT silence colon cancer.
43:01And this is this is Mike Tuccini
43:03study that he ran recently with Kirk
43:05Chopper developed some really cool
43:07assays detect MGMT expression on this
43:09case using temazolomide in a Labra.
43:11This is before we discovered KL50 and
43:13now Mike's actually moved on to looking
43:16at Temazolomide in an ATR inhibitor.
43:18And so we're really excited for this
43:20because what we believe is not only can we
43:22test KL50 as a monotherapy in these cancers,
43:24we could actually probably combine this with
43:26other agents like DNA repair inhibitors,
43:28PARP inhibitors and ATR
43:31inhibitors for example.
43:33And then in the last sort of few slides
43:35just kind of talk about sort of some
43:37of the Wilder stuff that we're doing.
43:39And so I showed you that mechanism earlier,
43:40right, with all the, you know, 4A4B4C.
43:42So when you focus on the blue box area,
43:46we've got this kind of crazy idea.
43:47I know our lab can sometimes be a
43:49little going off the beaten path,
43:51but we call this project breaking DDR
43:53if you guys ever see a Breaking Bad.
43:55So it's just sort of a little wild,
43:57but but I promise you there there's
43:59some sanity here.
44:00So if you think about it,
44:02we're creating cross links that are
44:04specifically active in MGMT silence,
44:06misreactor deficient cells and arguably MGMT.
44:09Science misreproficient we are now
44:11by making this simple fluoroethyl
44:14substitution for the for the methyl
44:17group here we're actually now making
44:19the futile cycling pathway that
44:22Tamizoli works totally irrelevant.
44:23So now the cell is actually being
44:26forced because of its genomic
44:27biomorg because it lacks MGMT.
44:29It's now being driven into a cross
44:32link repair pathway probably right.
44:34So the question is are we now able
44:36to at a bigger picture create DNA
44:38modifiers right create novel analogues
44:40that for instance here create a cross
44:42link or create a double strand break.
44:44So we're making new analog sets
44:46sets lab and we've got James,
44:47Ilia here and others that are grad student
44:49that are working on other analogues
44:51and and different DNA repair defects.
44:53But we could actually force a cell to
44:55switch from one repair pathway to another.
44:58OK.
44:58And so this could be an enormous
45:00opportunity for novel combinations
45:01of KL50 with DNA repair enhibbers
45:04that either you wouldn't think were
45:06possible or targeting DNA repair
45:08proteins that are not, you know,
45:10really thought to be relevant,
45:11but they become relevant.
45:12So for this,
45:13I'll just show you a little bit
45:15of some of our prelim data.
45:16And this is Colin a post doc
45:17in our laboratory.
45:18So he's been addressing this
45:21and also James and others in our laboratory,
45:23but he's been leading the efforts of
45:25creating a focused DNA repair gene library.
45:27And we're always excited about potential
45:29collaborations in this space because
45:30he's really spent a lot of time the
45:32last year building this platform
45:33working with Select and Agilent and the
45:35Agilent sequencing profile platform
45:38about 335 DNA repair and response
45:40genes targeting 6 guide RN as per gene
45:44standard sort of protocol shown here.
45:46And essentially looking at what
45:47are the nodes of sensitivity and
45:50resistance for your drug of interest
45:52that are related to DNA repair.
45:54Obviously you can do a whole
45:55genome crisper screen and we have
45:56aspirations of doing that eventually.
45:58But these are giant experiments by
45:59doing a focus screen you can do,
46:01you know, you know,
46:032015 centimeter dishes and and you'll
46:06be and it's relatively tractable.
46:08And so I'll just give you just a
46:11little smattering of some of the
46:13data that he's produced recently.
46:14So the first thing he did is started
46:16looking at KL50 and TMZ and ran it
46:18through his Christmas screening platform.
46:20And I should have been remiss of
46:21it and mentioned Sam Friedman,
46:22the bioinformaticist in our lab
46:24that built the platform for the
46:25analysis of this data.
46:26And what you can see here just as a glimpse,
46:28you can see very nice in terms of
46:31everything here is sensitive knockout of
46:33that gene induces sensitivity and then
46:35everything on the right induces resistance.
46:37TP 53 comes out,
46:38but it's at a pretty low magnitude
46:40of effects there.
46:41But you can actually see if you notice
46:43there's a lot of Fanconi genes and
46:45interesting genes that are are involved.
46:47When you overlay this with TMZ,
46:49it gets really interesting.
46:50And so I'll show you this data here.
46:53He's doing a lot of work here but I'm
46:54just summarizing because of time.
46:56You can see here now in these different
46:59quadrants you've got when you compare
47:01KL50 versus control and TMZ versus
47:03control what you can see are the
47:04genes that are whose knockout not
47:06when knocked out and do sensitivity.
47:08TMZ only versus resistance to TMZ
47:10only and then sensitive to KL50 only
47:13And you can see some interesting like
47:15one sort of knew that but you know
47:17arguably still pretty interesting
47:18then one on the right here this was
47:20actually great because you could
47:21see the mismatch repaired genes all
47:23come out when you knock them out
47:25and become resistant to TMZ.
47:26And you'll notice actually for the
47:27again the DNA repair official is
47:29the missing gene here is MSH 3.
47:30So the two three complex which repairs
47:33loops and not insert mismatches was
47:35not a determinant of of resistance.
47:37So then functionally validating the
47:39screen as a as a really great way
47:41to fingerprint molecules.
47:43But we can see one gene actually
47:45popped out that was really interesting
47:47and this is called B Rip One which
47:49is also known as Frank J.
47:51And so we've been interested in
47:53understanding this further and
47:55to to get at this we reached out to
47:57Sharon Kanter who's done a lot of work
47:59in the Fang Jay space and Colin reached
48:00out to her to see if we could validate
48:02this in some Fang Jay knockouts.
48:04All lines and here's 3 knockouts all
48:05lines that we got you can see her MGMT.
48:07Science. Mr. repair proficient it's a
48:10double negative and then MGMT proficient
48:14but knockout sorry knockout MGMT.
48:17Chemically those expensive guanine
48:19and you can see there's no real
48:21effects of knocking out, thank Jay
48:24in terms of tamzolamide sensitivity.
48:26And again we would not expect
48:27that from the crisper screen,
48:29but you can see some this
48:30is short term growth delay.
48:31So we still have some clonogenics to do here.
48:33But just you can see in the
48:34middle here that in this HEK,
48:36this hex align,
48:37MGC science mismatch very deficient,
48:39you can see a very,
48:40very nice effect of a Fank
48:43Jay inducing sensitivity.
48:44So this platform is exciting because
48:46just sort of going the previous slide,
48:49what we're trying to do,
48:50what we're now going to be looking
48:52at is actually using this platform to
48:54actually start fingerprinting different
48:56alkylators as we make different
48:57model modifiers and warheads to sort
48:59of see how the landscape shifts.
49:01And actually we're always open
49:02to collaborations,
49:03you know shoot calling or or mean
49:05e-mail if you're interested in
49:06testing a drug out in our platform.
49:07And depending on the interest in the fit,
49:10we could, we could certainly collaborate,
49:11we have this kind of running pretty well.
49:14So with that sort of just conclusions
49:16and future directions hopefully I've
49:18shown you that we've identified the
49:19first MGT dependent mismatch repair
49:21independent alculator which has a
49:23very favorable TI which I believe
49:25potentially meets the constraints
49:26of of what could be successful as a
49:29synthetic lethal targeting strategy
49:30that can make it in the clinic.
49:32We've spent a lot of time loosening
49:34the mechanism of activity then a
49:36lot of validation experiments that
49:37are presented here today.
49:38We'd argue this is a whole new
49:40way to exploit DNA repair defects
49:42and hopefully you'll see James
49:44Elias upcoming RIP talk.
49:45I don't know if it's coming up soon,
49:47but sometimes doing this for
49:49HRD or brach immune cancers,
49:51we can do this for IDH immune cancers
49:53with actually inhibit out BH.
49:54There's a lot of different pathways
49:56we can go here and I think we
49:58can actually do really novel
50:00combinations here looking at DNA
50:02preparing hip accommodations.
50:03What I haven't shown you Juan Vasquez,
50:05postdoc in our lab and now has his own
50:07lab here is looking at immunotherapy
50:09combinations again because of immunogenic
50:11cell death from cross linking.
50:13And of course very exciting to
50:14launch this into a company.
50:15We just shared a little bit.
50:16There's a lot going on with modify
50:18but didn't want to focus on on
50:19that today which we hope will be in
50:21the clinic in about a year or so.
50:23So of course as always I
50:24just make the slides here.
50:25So I got to thank the people
50:27actually do the the work and
50:28folks that I'll mention here run
50:30to me or long time lab manager
50:32Colin who did the crisper screen.
50:34Kingston Lynn and the PhD
50:36student graduating soon.
50:37Pratik who did the post doc
50:39doing the AML work.
50:40And I've eluded some of the other
50:42folks like James and and others
50:43and then Susan Gable who now just
50:45started her laboratory and Juan
50:47Vasquez and of course the Herzog
50:49Laboratory who's been really a joint
50:51project through and through and then
50:53finally thank all the folks that
50:54fund our work and got ended on time.
50:56This is great.
50:57All right.
51:04So we have time for questions or yeah, sorry
51:09the one you like your second or less of
51:11a second talk about immune interactions
51:14they've done about lung cancer,
51:15colon cancer and the DNA response problems.
51:18We know those are very sensitive in
51:21the but I wonder if the but how much
51:23are you pushing on the idea that you're
51:25going to be enhancing the effectiveness
51:27by also enhancing that pathway.
51:29We already know it was
51:33important that the so of of mismatch repair
51:36loss or yeah so it's interesting because
51:38you know there's been a lot of interest.
51:39I think this may diagonally answer
51:41your question and tell me if it doesn't
51:44is a lot of people are now trying to
51:46give tamazolamide to induce Microsoft
51:48instability and mismatch repair.
51:50And there's AGI trial of the air
51:52through the trial where they actually
51:54took MGMT silence colon cancer.
51:55And they found that when you
51:57give them TMZ the tumors respond,
51:58they become resistant,
51:5960% of them get mismatched pair mutations.
52:01And then they went on to get immunotherapy
52:03and they thought that they would respond
52:06but the responses were quite limited.
52:08And this is Keith Liggin up at up in
52:11Boston pathologist who would has shown
52:12that it's likely that there's just not
52:14enough time for NEO antigen formation.
52:16So I think acquired mismatch of pair
52:20mutations after TMZ for instance
52:21will not respond to immunotherapy
52:23like we think we did.
52:24We do think though the KL50 induced
52:26cross linking could be immunogenic
52:28cell death could sensitize that
52:29sort of gets at your question.
52:31But
52:33yeah it's beautiful work and I think
52:37you know your work illustrates to
52:38training is the power of having you
52:42know a broad perspective historically
52:44what what what didn't mean why and
52:47then the focus on individual residues,
52:50the obsession that you've shown in your work,
52:53you know the details of all of this work.
52:56So Congrats. Thank you.
52:58I I have a basic question.
53:02The MGMT dependency, this is related
53:06to the the tumor lacking that enzyme,
53:09but the wild type cells have
53:11that enzyme as you understand.
53:12Yeah. So that's in part a great
53:17tolman of specificity there. Yes. And
53:22I was a question. The question is,
53:29yeah, it's a tough crowd, tough crowd.
53:33Demetrius always asking questions. The
53:35question is, you know,
53:36I was thinking, you know,
53:38would it be useful to wash
53:39it out because you have,
53:40it's so sensitive. Right.
53:41So you could add back
53:43that enzyme essentially.
53:45So. Oh, yeah. Yeah.
53:46Oh, interesting. And I don't know
53:48if that's if that would be useful,
53:50but it seems like your wild
53:51type does have that enzyme.
53:53So you're really not washing out there.
53:54I was thinking in terms of toxicity,
53:56but it will give you even greater.
54:00Yeah even wider
54:02dose. Yeah. Yeah.
54:02No, no, it's it's actually we're we
54:04have a it's kind of a secret project.
54:07Some grads didn't get mad if I talk about it.
54:09But we're trying to go the other way
54:11which is trying to like a radiation
54:13activated version where where you knock
54:15out MGMT like because MGT unmethylated
54:17tumors is a huge unmet need there
54:19because then the therapeutic index.
54:21So again this is a diagonal
54:22answer to your question,
54:23but where we could actually
54:25have MGMT inhibition,
54:26so O 6 pencil guanine that's
54:28activated only in the radiation field.
54:30So that's kind of one way
54:32we're getting at that.
54:33But to your point of some people have tried
54:35to do like rescue experiments and whatnot,
54:37I think it's it's hard,
54:38I think it would come down to to
54:40timing and sequencing just to
54:41try to magnify the therapy index.
54:43Yeah.
54:43In the
54:44animal studies that fail, do they fail
54:46because of progression? They fail, yeah,
54:52yeah. Yeah. I think,
54:53I think the therapeutic index is the issue
54:55with the mouse studies we show is is,
54:57is and we didn't show this because the
54:59companies mainly working on this is the
55:01Heen tox is is severely dose limiting.
55:03And and actually even with KL50 the the,
55:06the, the therapeutic index that we get
55:08if you do a rat tolerability study
55:10which is a better which I've been
55:12taught is the is a better surrogate for
55:14human Heen tox and like an actual rat.
55:17Straight up you know, 30 day,
55:18five day on observation.
55:20The Heen tox is dose limiting.
55:22And so that is probably the biggest issue.
55:23That's interesting.
55:27Think about dosing the blood
55:29based enzyme like that. That's
55:30interesting. Oh yeah. Yeah.
55:31And they could be taken up in
55:33the Heen compartment. Yeah.
55:34We should talk actually.
55:35Yeah. Yeah. Yeah.
55:37It's something to consider.
55:38Yeah. Because if you're getting
55:39dose limiting tox. Yeah.
55:41You have this enzyme dependent,
55:43you know, to leave. Yeah. Yeah.
55:44There's a you might be able
55:45to take advantage of that,
55:46especially if it's Heen related.
55:48Yeah. We should talk later
55:49if there'd be an interesting
55:50way to selectively delivery.
55:51That's cool. Yeah. Yeah. Like,
55:56yeah, we'll see.
56:07Yeah.
56:14Yeah,
56:18100%, Yeah.
56:21Yeah, it's very, no,
56:24it's very stressful actually.
56:25We're thinking about this a lot
56:27both academically and then obviously
56:29the company but academically and
56:30actually been relying a lot of
56:32pathology colleagues here for
56:33for input because one thing I've
56:35been digging in the literature,
56:36it's like so MGMT promoter
56:37methylation as many people in this
56:39audience know better than me.
56:41You know it's sort of a cut off
56:42and sort of arbitrary and so but
56:44then on the same side we've been
56:45trying to do an MGMTIHC essay,
56:47we've been doing some TM as
56:48right now we're working on this
56:50working with Kurt and then others.
56:53And the issue with the MGMTIHC is it
56:55it seems to not be as the threshold,
56:58the dynamic range where even if
56:59it's out on IHC there's still
57:01low level expression.
57:02Craig Orbinski,
57:03I don't know if he's a neuropathologist
57:05at Northwestern talks about this
57:06all the time and the other elephant
57:08in his room is the IHC is negative.
57:10When you treat the TMZ there is
57:12some data that MGMT can be re
57:15expressed and depending on the
57:16promoter methylation sites and so.
57:18So what I think is going to be
57:20the answer is a combination.
57:22So in the colon cancer literature
57:24recently what they're doing is
57:26both IHC for MGMT and promoter
57:29methylation and they're actually
57:30doing this thing called the methyl
57:32beam assay which you guys probably
57:33know more than meets like a digital
57:35MGMT promoter methylation.
57:36So Long story short is I think
57:38trying to have a double selection
57:40for homogeneously silenced by IHC
57:42and meets the criteria for promoter
57:44methylation will be key because there
57:46are partially methylated cases that
57:48are going to totally screw this up.
57:49It's related
57:51but I
57:54don't know, I don't want
57:58that. Oh yeah, because they are, yeah,
58:01100% yeah, so so TET is some of the,
58:03you know as question was
58:05about TET expression.
58:06So mutations in the 10/10/11 trans low case,
58:10which is hydroxy methyl cytosine
58:12maintenance in AML.
58:13We're doing this with Stephanie Helene's
58:15lab where TET mutations are common.
58:17We're trying to see if TET will be
58:20a predictor for MGMT expression and
58:21IDH mutations as many people know,
58:23inhibit TET as well.
58:25That's another project that we're
58:26trying to get that which has been
58:27harder to do than we thought.
58:29But yeah, it's a great question.
58:30Yeah, I have a question.
58:31One of the sort of tumor you showed
58:34with MGMT insulation is breast cancer,
58:37breast cancer with the DRC one
58:38deficiency in general line,
58:39the bottom is really nice.
58:41And the other showroom MIC mutation.
58:43But when they used it in
58:45the semantic mutation,
58:46the trial team made it and they
58:49did the HR score and they should
58:52benefit from Barbara negative.
58:54They only have semantic.
58:55So what's your thought the HR revealed?
58:57Mm hmm. Mm
58:58hmm. Yeah. I mean for you know,
59:01for that, that's a great question.
59:02I mean, there's even questions
59:03about loss of the second allele.
59:05Susan, Don check at Pennis
59:07asked that question as well.
59:08It's confounded though because I think
59:10the new PARP one selective PARP numbers
59:12that you like like AZD 5 three O 5,
59:14I think they may be able to get
59:16enough PARP inhibition to hit even
59:18if it's a happenence efficiency.
59:20But you know, on a side note,
59:22we're also looking at MGMT loss and HRD,
59:24this is Susan Gables Labs doing that as well.
59:26I don't know if I have a
59:27full answer for you though.
59:28Yeah, good, great.
59:29So I just wanted
59:31to know how how much question.
59:42Yeah, it's it's it's we're
59:43alluding this earlier.
59:44I think the correlation between
59:47MGMT promoter silencing and
59:49and MMGMT protein expression,
59:50it's quite variable and that's it's
59:52a little bit in the glioma world if
59:55you're promoter methylated MGM TS out.
59:57But in other cancers they're finding
59:59that you can be promoter methylated but
01:00:01still express MGMT the protein level.
01:00:03So I think there's a lot of variability,
01:00:04which is I think.
01:00:05Going to be important especially for this
01:00:07because the MGMT dependency is so exquisite.
01:00:09So great. Cool.
01:00:13I think we're at the top of the hour.
01:00:14Thank you.