"Altered RNA Splicing: a driver event in pancreatic cancer” and "Defining New Pathways in Colorectal Tumors with Mismatch Repair Deficiency"

April 28, 2022

Information

Yale Cancer Center Grand Rounds | April 26, 2022

Presentations by: Luisa Escobar-Hoyos, MSc, PhD and Rosa Munoz Xicola, PhD

ID7767

To CiteDCA Citation Guide

00:00Just a minute after noon, so I'd like to.
00:05A few people are still arriving,
00:06but I'd like to to welcome everybody today
00:10to the Cancer Center ground grand Rounds.
00:13And for those of you who who don't know me,
00:15I'm Mark Lemmon.
00:16I'm stepping in for Eric Weiner today
00:20because Eric is otherwise engaged.
00:23I'm mark them and I'm deputy
00:25director of of the Cancer Centre.
00:28And so I'm I'm I'm channeling.
00:33Actually, which is why I won't do it so.
00:36But the great,
00:37great to have you all here and great to have.
00:40Louise Escobar Hoyos and rose and
00:44we know Zikula to talk with us
00:48today and we so we begin with with
00:52Doctor Louisa has Escobar holes.
00:55Who is an assistant professor
00:57of therapeutic radiology.
00:59She received her Masters degree in
01:01Biomedical Sciences at the University
01:02at Del Valley in Cali in Colombia.
01:05And then as a Fulbright Scholarship,
01:06pursued a cache in in cancer,
01:09molecular and cellular pharmacology
01:10at Stony Brook University,
01:11where she was mentored by
01:13Doctor Kenneth Schroyer,
01:15then completed her postdoctoral training
01:16at Memorial Sloan Kettering Cancer Center.
01:18Commented by Doctor Stephen Stephen
01:21Leach and Omar Abdel Wahab,
01:23and the overarching goal of Doctor Escobar.
01:27Hoyos's lab is to develop new
01:30approaches to tackling pancreatic
01:32and lung cancers with lots of.
01:34Really very exciting work.
01:36Exciting new work going on and lots
01:39of innovation and specifically her
01:41team is currently trying to seeking
01:44to understand and target somatic
01:46mutations and importantly aberrant
01:48RNA processing in these tumors in
01:50order to to develop novel therapies.
01:53So it's a great pleasure to have
01:55you start us off.
01:57Louisa and I look forward very much
01:59to your talk, so please take it away.
02:03Wonderful thank you Mark. Let me just
02:06can everyone see my presenter mode.
02:09Sorry not my presenter. My full slide
02:12slide that's all very good perfect. So
02:14thank you Mark again for that Nice
02:16introduction and thank you everyone
02:18for participating today in in the
02:20Cancer Center grand rounds and I'm
02:22excited to share with you a little bit
02:24of the research that we've been doing
02:26in our lab in terms of finding how
02:28altered our release by splicing is a
02:31driver event for pancreatic cancer.
02:33So if your disclosures I'm part of the
02:36Scientific Advisory Board of QDX Diagnostics,
02:38I won't be presenting the work
02:40that I have with them today.
02:42I'll be talking about this compound.
02:43This small molecule inhibitor of
02:45splicing that has been provided to
02:47us by age 3 biomedicine and currently
02:49we're in discussions to launch a
02:52clinical trial using this compound.
02:53Based on the research that I'm
02:55going to show you today.
02:56So for many of you,
02:59it's not unheard of that pancreatic
03:01cancer is a very lethal malignancy,
03:04and in here is just plotting
03:07the survival over time.
03:09For the major cancers and what we
03:11can see in this evidence is that
03:13unfortunately we haven't been able to
03:15make that much improvement in the five
03:17year survival rate of pancreatic cancer,
03:19and this could be attributed to many reasons.
03:21It's it's a. It's a disease that
03:24is diagnosed once it has been.
03:26It's already systemic.
03:27The first line,
03:28chemotherapy and immunotherapies,
03:30are ineffective,
03:31and the available targeted therapies
03:33are only available to 1% of
03:35the cases that have actionable,
03:37actionable mutations.
03:38So there is a really.
03:40Strong clinical need to understand more of
03:42these tumors and develop new treatments.
03:45So just to introduce a little bit
03:46of the mutational landscape for
03:48these tumors and mainly today,
03:49I'm talking about pancreatic
03:51ductal at no carcinomas or petax,
03:53the most common form of pancreatic cancer,
03:56so we know that there are driven by a
03:58first mutation, first hit mutation,
04:00activating mutations in cameras,
04:02and also is very common to find
04:04T53 mutations,
04:05so we'll talk about a little
04:06bit more of these mutations.
04:08There is also sometimes it appears that
04:10other tumor suppressors are mutated,
04:12and then after these four top genes.
04:15There's really a sea of mutations
04:16that appear at a very low frequency,
04:19so using mouse models over the
04:22last 1520 years or so we have been
04:24able to kind of dissect a little
04:26bit the genetics of this disease.
04:28We know that the normal pancreas,
04:30if we engineer key rest mutations
04:31into the US and ourselves,
04:33these mice will start develop,
04:35panning or pancreatic intraepithelial
04:38lesions that will progress into
04:41pancreatic cancer if we add
04:43an additional mutation in P50.
04:45And so basically,
04:46this is this tumor follows kind of a
04:50two hit hypothesis notes and model
04:52and all with the with you know to
04:55enhance the activity of kieras over the time.
04:59Now for many years we've known about
05:01these two mutations driving this disease,
05:02but really we haven't made much effort
05:05to understand how these oncoproteins
05:07cooperate in the case of pancreatic cancer,
05:10and even other tumors.
05:11So a few years ago when
05:13I was gonna start as a postdoc at MSK,
05:15these studies came out in the molecular
05:18subtypes of pancreatic cancer with the
05:21squamous and basal subtype being the
05:24most aggressive molecular subtypes.
05:26And when you look into what genes are
05:28differentially expressed into the subtype,
05:31there is a small subset of genes that it's
05:34overexpressed in this molecular subtype.
05:36And when you look into their mutation status,
05:38they're highly associated with
05:39gain of function mutant, P. 53.
05:41And these genes that are being
05:44enriched are the majority are codeine
05:46for splicing regulatory proteins.
05:49So after I read these reports I kind
05:51of got interested on understanding
05:52a little bit more alternative RNA
05:55splicing and just to remind everyone
05:57during alternative splicing.
05:59Not only we remove the introns of genes
06:01but also there could be a selective
06:04retention or skipping of exons and this
06:06can lead to proteins that had opposite
06:09functions or no protein being formed.
06:12And we all have our favorite gene,
06:13and sometimes we don't study the
06:16alternative splicing of these gene
06:17and this pathway in general is a
06:19very potent and plastic that can
06:21actually explain a lot of the features
06:23that happen in cancer cells.
06:25So based on those reports I
06:29asked the question,
06:30is there a connection between mutations
06:31and P53 and alterations in RNA splicing
06:33and so first we took the RNA sequencing
06:36from many patients and we divided
06:38them into three different groups,
06:41either if they had truncating.
06:42Mutations in P53 meaning loss
06:44of function mutations of P53,
06:47gain of function mutations or
06:49mutations that make the protein
06:51going from a tumor suppressor to
06:53an uncle protein or wild type P53.
06:55And we compared this glycine.
06:57Differences between these tumors and
06:59in the case of pancreatic cancer,
07:01the most common hotspot missense
07:04mutations are these four listed here.
07:06So here are the 1st results.
07:08So in in the X axis you have
07:11a measurement of alternative.
07:13Slicing of axons and the different tumors,
07:15and here is each one of the mutations.
07:1853 compared to the wild type P53 tumors,
07:20and each one of the hotspot mutations
07:23compared to truncated P53 and what you
07:25are can appreciate is that all these
07:27hotspot gain of function mutations
07:30change alternative splicing and the
07:32R175 one of the most common ones,
07:33actually changes the most.
07:35So based on those correlation
07:36studies we started asking well.
07:38Is P53 changing splicing
07:40and pancreatic cancer?
07:42So we went ahead and developed.
07:43Three different model system patient
07:45derived organoids where we can
07:47actually shut off the expression of
07:49the mutant P 53 and after doing deep
07:52RNA sequencing and splicing analysis,
07:53we can see that there are these different
07:56exons that are either preferentially
07:58retained in red or preferentially
08:01spliced out in the context of mutant
08:03P53 in complex to complement this
08:06model we generated a murine cell line,
08:09also with the same capacity
08:11to shut down mutant P 53.
08:14And again,
08:14we were seeing these changes
08:16or swapping of axons.
08:17And lastly, we took a
08:20pancreatic precancer mouse,
08:22panning organoids where we knocking
08:25the mutation of our 175 and again we
08:28were seeing that even in early stages
08:30after early 19 of the mutation we were
08:32seeing this differential splicing of exons,
08:35so we wanted to ask,
08:36what are the specific features
08:38of these exons that are being
08:40either retained or skipped?
08:41And we found that there is this this that.
08:44The retention of these axles is not random.
08:47All the promoted axons after splicing
08:49those ones that are going to be
08:51retained in the mature M RNA are highly
08:53enriched for seas while they're repressed,
08:55exiles were highly enriched in a S&amp;G's,
08:58suggesting that this was pretty much
09:01a program established in these cells,
09:03so we wanted to focus on what were
09:06the MRE's that were being coded by
09:08these gain of of policy axons and so
09:11here we're summing into one of these.
09:14Barneys Gap 17.
09:16We're in mice and human.
09:18This Exxon 17,
09:19which is a policy Axon,
09:21is alternatively spliced,
09:23and here are the raw sequencing of
09:27the of the reeds of this Axon and
09:29what you can appreciate is that
09:30whenever the mutant P 53 is present,
09:32there is higher rates versus
09:34when you knock it down.
09:35There is a decrease on the
09:37retention of these Axon,
09:38but not the neighboring axons,
09:40and we saw this pattern across
09:42the marine cell line that we had
09:44engineer in the panning organized.
09:45That we had also crisped.
09:48So from here we actually went and said,
09:50well,
09:51let's go back to patient derived
09:53samples and let's see if the retention
09:55of these policy accounting gap 17 is
09:57it exclusive for the R175 mutation,
09:59or is it also found in other
10:01gain of function mutants of P53?
10:03And the answer was yes,
10:04it's actually retained and not only are 175,
10:07but other gain of function mutant P53.
10:10As you can appreciate here from this
10:13targeted PCR. So then the question what?
10:15It was well,
10:16what is the consequence of incorporating
10:19policy exons into M RNA's over time,
10:22and so when we started looking at all
10:24of the M RNA's that were incorporating
10:26policy axons in the presence of mutant P.
10:2853,
10:29we found that a family of proteins
10:31called the GPA,
10:33the GPA's activating proteins or gaps
10:36were actually gaining these policy Axon.
10:38In fact 25% of total gaps encoded by
10:41the human genome were gaining policy
10:43exons and just to remind everyone.
10:46The gaps do.
10:47They actually accelerate the GTP
10:49hydrolysis of Ras proteins so
10:51they bring it from the on state,
10:54which is bound to GTP to the
10:57off state bound to GDP.
10:59And just to give you a sense of what,
11:01how was this exon impacting the
11:04protein we were seeing that these
11:06policy actions were inframe and when
11:08they got translated they encoded.
11:11For prolines,
11:12highly rich proline tails in the
11:15sea terminus of the parties and
11:17here is just an example to show
11:20you that actually these are
11:22different molecular weights of the protein.
11:25Here is the promoted by
11:26P53 with the policy Exxon.
11:28And here's the repressed
11:30P53 isoform of gap 17.
11:32Without the policy Axon,
11:33so both can be produced and
11:36translated in in the self.
11:37So at this point we were
11:39faced with the question well.
11:41What happens with the CARAS state of either
11:46GTP bound state or GDP bound bound state?
11:49Whenever you have a plus policy
11:51gap 17 or a minus policy gap 17,
11:55and so we did this in cell experiments
11:58where we actually first overexpress
12:01either the policy gap 17 or the minus
12:05policy gap 17 and we actually did a
12:08pull down to capture GTP bound K rest.
12:11And then we did.
12:12We used an antibody that it's only
12:14recognizing the mutant form of kiras
12:16to determine the the the levels of
12:20active cares in these cells and
12:23what we found was interesting,
12:26which is in the presence of policy gap 17.
12:29The isoform promoted by mutant P.
12:3153.
12:31The levels of active care as were
12:35actually maintained.
12:36However, as soon as we overexpress
12:39the minus policy gap 17 that.
12:41Isoform that is repressed by mutant P 53.
12:44We saw that the levels of active
12:46care has decreased and also the
12:48active levels of Arc which is
12:50downstream of of cameras were
12:52also significantly decreased,
12:53and so it was interesting to see kind
12:56of like the different effects on the
12:59active form of Keras in the presence
13:02or absence of this gap 17 isoforms.
13:04So then we went and did a
13:07self free essay where we took
13:09while type carrots or mutant.
13:11The arrest,
13:12and we incubated it with either the
13:15policy gap 17 or the minus policy gap
13:1717 and what you can see is that there
13:20was no much difference in the cell.
13:23Free assays in terms of their capacity
13:26to hydrolyze GTP bound cameras,
13:28and this was very odd and surprising
13:30to us because actually in the cells
13:33they were maintaining different
13:35levels of Keras bound to GTP.
13:37So this made us go back to the drawing
13:39board and start thinking about.
13:41What happens in the context of
13:44cells in the activity of gaps?
13:46It turns out that gaps are
13:48usually cytoplasmic proteins that
13:50when calls to deactivate Keras,
13:52they go to the membrane and
13:54that's when they actually promote
13:56the hydrolysis of the GTP.
13:58What we were seeing was the following
14:00and the presence of mutant P53.
14:02When you have the policy gap
14:0517 being expressed,
14:06the gap mainly localizes into
14:08the title plasm of the cell.
14:10Even when we gave it signals
14:12to go to the membrane.
14:14However, when you knock down P.
14:1653 out of the cells,
14:18you can see that there is this the the
14:21gap 17 that is now not expressing policy.
14:25Exxon now can more likely reach
14:28the membrane and promote the
14:30hydrolysis of Keras.
14:32And so we were excited to find
14:34these because that led us to
14:36a model where we had for the
14:37first time kind of discover
14:39how these two owner proteins,
14:41mutant cares and mutant 53
14:43synergizes in the following.
14:44OK, our model suggests that in the presence
14:47of a wild type B 53 or the loss of P53,
14:50the cells actually lose policy
14:53axons across multiple M RNA's,
14:56mainly the gap in RNA's and after
14:59this M RNA gets translated.
15:01It encodes gaps that actually are
15:04efficient at reaching the membrane and
15:06being more efficient at hydrolyzing GTP
15:10bound cameras and promoting tumor growth,
15:12but not as much as.
15:14When you have the hotspot mutant,
15:16because now this time you are
15:18gaining a policy Axon and when that
15:21mRNA gets translated it has now
15:23these reach domain of prolines that
15:26prevented from reaching the membrane.
15:29Maintaining an active care
15:31estate and more tumor growth.
15:33So just to go back to our model and
15:35the genetics of pancreatic cancer.
15:38So I've told you before that you
15:40needed Karras and mutant P 53 and
15:42what our findings had suggested is
15:44that in the presence of just mutant.
15:46The rest you still have the active rest,
15:48but then when mutant P 53 comes specifically,
15:51the gain of function mutant of P53,
15:53you Now have an altered RNA splicing
15:57and and a feedback loop that now
15:59prevents the gaps from being active.
16:01And then in this way you can enhance the
16:04oncogenic signaling and activity of key
16:06areas and this is our model system currently.
16:09So that was great and we published
16:11this a couple of years ago.
16:13So then we came back into the
16:14lab and we started thinking,
16:16well,
16:16how can we target RNA splicing
16:19and pancreatic cancer cells?
16:21And so recently,
16:23this small molecule compound,
16:26H3 B 8800 it started being tested
16:29in phase one clinical trials,
16:31and they got interested in our research
16:35with pancreatic cancer and mutant 53.
16:37So basically what H3 B 8800 does.
16:40It binds to one of the.
16:41Course splicing proteins as F3V1 and
16:44prevents this whole machinery the
16:46spliceosome to bind and recognize
16:48fully the M RNA's, and so we were.
16:51Our hypothesis was well,
16:52if mutant P53 tumors really
16:54depend on ultra RNA splicing,
16:56they'd be more sensitive to toddler.
16:58They would be more sensitive
17:01to any perturbation into the
17:03splicing machinery with the AD 800,
17:06so we launched what we call a
17:09mouse trial where we took mice.
17:11That either had tumors that had
17:14mutant P53 in them, so that's red,
17:17or that lacked mutant 53 inone
17:20which are blue.
17:21And then we randomize these animals
17:23to either receive 8800 or vehicle.
17:26And what you can appreciate is
17:28that the solid lines,
17:30which are the animals that receive 8800,
17:32they all benefited from having
17:35from receiving the compound.
17:36However,
17:37the animals that survive and
17:39benefited the most were those
17:41ones that had mutant P53 in them,
17:43suggesting that these mutations
17:45sensitizes these tumors.
17:46To this lysine modulator.
17:48And when we did RNA splicing
17:52analysis and after we treated
17:54these tumors with the 8800,
17:55we can nicely see how.
17:58H3 B 8800 was repressing the
18:00retention of that policy Axon in the
18:03cells in a function as a function
18:05depending on the concentration.
18:07So you know other words.
18:09This compound was reversing the key
18:11splicing events that we had seen
18:13in the presence of mutant P. 53.
18:15Lastly, we have now established a human
18:18model where we have isogenix cells that
18:21express different forms of mutant P.
18:2353 and we know that when we
18:25treat them with this compound,
18:27the mutants and particularly are
18:29more sensitive to these compounds
18:31when you compare them to the
18:33counterparts when they don't have P53
18:35or when they have a wild type P53.
18:37So based on these results we
18:39are now in discussions with HB,
18:41biomedicine and row Invad sciences,
18:44who recently bought the 88.
18:46100 compound because we would
18:48like to start a phase.
18:49Two clinical trial where
18:51we combine Gemma Broxton,
18:52which is one of the first
18:54gamma standard of care,
18:55chemotherapeutic lines for pancreatic
18:57cancer patients and start escalating
18:59doses of 8800 for patients whose
19:02tumors have gained a function.
19:04Mutant of P 53 so hopefully
19:06we can launch this soon.
19:08So let's back.
19:09Let's go back to you know what are
19:12the mutations that Dr pancreatic cancer?
19:15What have we understood
19:16and how can we target?
19:17To drive personalized medicine,
19:20so,
19:20as I mentioned before,
19:22we now understand that KIERAS
19:24mutations are the most common
19:25mutations and they are required
19:27hit mutation to form tumors.
19:29We know that 10% of the cases have
19:32familiar pancreatic cancer and
19:33most of them have mutations in ATM
19:36and when they have these mutations
19:38they are giving PARP inhibitors
19:40and that's why we do molecular
19:43molecular profiling industry rumors
19:45to identify this cohort of patients.
19:47To have actionable mutations.
19:50We also know that,
19:52as I mentioned before,
19:53that 30% of the sporadic pancreatic tumors,
19:56which are the most common ones are
19:58driven by gain of function 53.
20:01And as I mentioned before,
20:03we're hoping to start a clinical
20:05trial using 8800.
20:06These glycine inhibitors to see if
20:09we can bring a targeted therapy
20:12for these sporadic tumors.
20:14Now we're still facing the challenge
20:16that we still don't understand.
20:1830. What is the mutation?
20:20That drives the other 30% of
20:23pancreatic tumors because we also,
20:26because we already know that the other
20:2830% is driven by loss of function P.
20:3053.
20:30So for these two last groups.
20:32Unfortunately,
20:32right now we don't have any targeted therapy
20:36or any trials that are going to launch here,
20:39so we were curious to know
20:40well what is sporadic,
20:42what other mutations, Dr,
20:44sporadic tumors of pancreatic cancer,
20:47and so to answer that question
20:49I mentioned before that.
20:50You know there is a such a a
20:53large number of mutations that
20:55appear in very low frequencies,
20:58but it's hard to study each one of
21:00these mutations to understand well.
21:01Are they driver mutations or
21:03are they passenger mutations?
21:05And so we took an inform
21:07approach where we went
21:08back to the basic contact concept.
21:10Sorry of mutual esclusiva,
21:12so just to remind everyone we know that
21:15mutations may be mutually exclusive,
21:17meaning that if P53 is present.
21:20The mutation in P53 is present,
21:22then it would turn into a viable tumor cell.
21:25If other mutation is present but not P53,
21:28it could be viable,
21:29but if both mutations are present,
21:31it could be synthetic, lethal,
21:33and the mutual exclusivity of these
21:35mutations is because sometimes these mutual
21:38exclusive mutations either have the same
21:40function or impact the same pathway and
21:42that's what makes them driver mutations.
21:44So we started conducting a mutual exclusivity
21:48analysis by taking advantage of C.
21:50Bioportal.
21:51Which has the.
21:53Mutation signatures of over 3000
21:56patient samples of pancreatic cancer,
21:59and so the 1st results that we
22:01derive from this analysis are are
22:03shown here in this volcano plot.
22:05So on the right hand side we have the
22:08mutations that Co occur with mutant P.
22:1053 and in the left hand side we
22:11have the mutations that are mutually
22:13exclusive for P53 and this was the
22:16side of the volcano that we were
22:18interested in because this potentially
22:19could tell us what mutations were
22:22driving this disease aside from mutant.
22:2453 so as a as a proof of concept,
22:27key results in the middle is
22:29the first mutation.
22:30It appears for all of the tumors,
22:32but here's where it got surprising to us.
22:34One of the most mutually exclusive
22:36mutations to P53 was mutation in SF3B1.
22:39It's a score splicing protein.
22:42Then it on this same side we
22:45have mutations in RBM 10,
22:47which is another splicing factor.
22:49But on the core occurring
22:51side we had a U2AF1,
22:53again another mutation in another.
22:54License factor,
22:55so if our hypothesis was true,
22:57it's potentially that
22:59mutually exclusive mutations,
23:01meaning S3B1 and RBM 10 could be
23:04drivers of pancreatic cancer and just
23:07assuming what type of mutations are
23:10present in S4B1S4B1 in pancreatic
23:13cancer has a driver mutation.
23:16Very hot spot mutation in case 700.
23:18E RBM 10 is mainly truncating mutation.
23:21So basically you're losing the
23:24function of RBM 10 and U2AF1 has
23:27a hotspot mutation in S34F.
23:30So we our question was,
23:32are any of these three mutations
23:34driving pancreatic cancer?
23:35And so we took advantage and we
23:38started generating genetically
23:39engineered mouse models.
23:41So here's the case.
23:42C model system where it's only
23:45driven by ACARAS mutation.
23:47And what we found and what we
23:49expected was that these animals
23:50should only form pannings or
23:52pancreatic and triphenyl neoplasias,
23:54those precancer states.
23:56So then we cross this KC animal with a
24:00U2AF1 mutant animal for the S34F mutation.
24:04And what we found is that
24:06actually there are pannings,
24:07but not as much as we expected.
24:09And most importantly,
24:10there was no Peacock in these animals.
24:13But surprisingly,
24:14the animals that had them,
24:16the Keras mutation and the SFRB 1 mutation,
24:19foreign pancreatic tumors,
24:21same as the animals that we cross to have K,
24:26res,
24:26and RBM ten loss.
24:28So here's just the quantification
24:30done by our pathologist,
24:32who you can see that there is only pdac
24:34and the animals that have the nutrition
24:37in S4B1 and the nutrition in our BM.
24:3910 There is more pannings also in these
24:41animals and they succumb to the disease.
24:43Very early on,
24:45so we are now in this hypothesis
24:48that we're trying to further test which is.
24:51We believe now that pancreatic cancer
24:54cells that have a mutant carras
24:57actually require a splicing switch
24:59in order to become tumor cells,
25:03and most likely the majority
25:05of these of these tumors will
25:07develop through a mutant P53,
25:09which I showed you before how it
25:11drives alternative RNA splicing,
25:13but we're now fathering.
25:16Starting how these SF 3B1 mutation,
25:20and RBM ten loss also drive the the
25:23the disease based on a splicing change,
25:26and these animals are now being
25:29characterized by a couple of
25:32postdoctoral fellows in my lab,
25:34and so I just want to quickly mention
25:37that they have obtained really
25:38interesting results in terms of
25:40what are the splicing defects that
25:42these proteins mutated proteins.
25:44Are leading to.
25:46They are very similar to the
25:48P53 splicing changes.
25:50We do tons of deep RNA sequencing
25:53into this model systems.
25:55We do several.
25:56We run several algorithms to
25:58determine the splicing changes into
26:01not only the marine model systems,
26:03but also patient derived samples.
26:07I just want to skip quickly through
26:09this just so I can get here to
26:11how are we going to target these
26:13mutant splicing factor proteins.
26:15So similarly,
26:16we use the 8800 compound and we are
26:19now finding that also these mutant
26:21cells are very sensitive to this compound.
26:24We are also finding that these
26:27mutations confer sensitivity to certain
26:30chemotherapeutic agents. In this case.
26:33In particular, the case 700 E.
26:35As of Feb,
26:36one is more sensitive to
26:37gemcitabine than it is to five FU.
26:39So this is important because these
26:42mutation profiling can also help
26:43to decide what would be the best.
26:45Chemotherapeutic agent who assigned
26:47to a particular patient and when
26:50we did combination studies on
26:52mixing gemcitabine with 8800 in
26:56mutant versus wildtype cells,
26:58we can see that the mutant cells are
27:01more sensitive to the combination
27:03of jam and 8800 more so than the
27:06wild type cells suggesting that this
27:08combination of therapy could be
27:10important to treating the patients
27:13that have these K 700 E mutation.
27:15That's up 31,
27:17so I just want to finalize by
27:20saying that I'm currently based
27:22on our findings on Mutant P.
27:2453 and mutant SFB.
27:25One and RBM 10 laws as the drivers,
27:28as pancreatic cancer.
27:30All of these mutations leading to
27:32changes in alternative splicing.
27:34We're hoping to also bring into the
27:37trial patients eligible patients
27:39that are case 100 mutant or have RBM
27:4210 lost to be eligible for this.
27:45Glycine anti silicene therapy that
27:47we wanna lounge in combination
27:49with Gemini and Gemini 8800 and
27:51so with that I wanna wrap up by
27:53saying thank you to everyone here
27:55for your attendance today to all the
27:58people in my lab who are leading
28:00this effort to all our collaborators
28:02and also to our funding sources.
28:05Thank you very much and I'll
28:07take any questions. Thank you.
28:09Thank you so much Teresa,
28:10that was really fascinating
28:13work at great, excellent stuff.
28:16If people have questions for Louisa,
28:20please put them in the chat and
28:23I can read them to her and she
28:26can go ahead and answer them.
28:28I had one quick question.
28:31While people are formulating their thoughts,
28:33which is it is intriguing that
28:37the mutations and the and effects,
28:38and indeed the the the.
28:438800 are all focusing on you two.
28:46Do you have some? He's out.
28:51It may be migraines,
28:52but what does that mean, mechanistically?
28:55Yeah, thank you Mark.
28:57So basically the compound targets
29:00mutant SF 3B1 and so it's the
29:03tumors have mutant SF 3B1.
29:06They are more sensitive to this compound,
29:08so that's the case for S3 one.
29:10But we also know that the if if
29:13tumors highly depend on splicing.
29:15There there are more sensitive to
29:18this compound because they cannot
29:20tolerate a double perturbation of the
29:23splicing changes and the splicing.
29:25Machinery and so that's how we are
29:29attributing the sensitivity of of
29:32mutant P 53 and mutant RBM 10 to 8800,
29:35and I think more dissection of the
29:39mechanism of of the drug within you.
29:42Know RBM 10 and P53 can further elucidate
29:45why are they so sensitive to this compound,
29:48at least for P53.
29:50We know that in certain cases
29:52it reverses the effects.
29:53The splicing changes that mutant P 53.
29:56Is promoting.
29:59Make makes sense and the other
30:01question I had was that we're going
30:03back to the gap 17 story. Do you see?
30:06In other circumstances the if you look
30:09through other cells and and indeed
30:12tumors that aren't don't have the
30:15the gain of function P53 mutations.
30:18Do you see the the gap 17 with
30:22the policy Exxon in other places?
30:25Yeah, so that's a good question.
30:27So for example.
30:28We've looked into other cancers
30:30that are not key, rest driven,
30:32but have this mutant form of P53,
30:34and we see that indeed the
30:37splicing changing gap 17 occurs.
30:39Now we have also seen some other tumors
30:42where mutant P 53 is not present and
30:45we still see the splicing change,
30:47and we think that this is attributed
30:50to the overexpression of a splicing
30:52factor called H&amp;R AMPK that today
30:54I didn't have time to go into,
30:57but we think that this splicing.
30:59Regulator also promotes the policy.
31:03The the policy acts on inclusion in M RNA,
31:06so we think that there is not
31:08a single pathway to to promote.
31:10The policy acts on retention in gaps.
31:15This is fascinating and I have a
31:18question in the chat from from
31:20Timothy Robinson who says great talk.
31:23I agree with the way you described
31:26using mutually exclusive analysis to
31:27find events within the same pathway.
31:30Did you look at Gap 17?
31:32Aberrant splicing based on mRNA
31:35directly to identify other drivers?
31:39So I'm not sure if I'm
31:40understanding the question.
31:41If I if we looked into into
31:44other pathways that are not
31:47linked to to the cares pathway.
31:49I think the question is
31:50actually did you look?
31:51Did you look at aberrant
31:53splicing based on M RNA to
31:54identify other drivers that
31:56might be other than gap 17?
31:57I think that's
31:59yeah. So all the splicing changes
32:01that we identified are based on mRNA
32:04sequencing and based on splicing
32:06analysis that we conduct. But if I.
32:09But I can also mention that the gaps are
32:12only 5% of the events that mutant be 50,
32:15three, 5% of the event is splicing
32:17events that mutant P 53 is triggering.
32:20So there are other M RNA's that affect
32:22other pathways that are being impacted by
32:25the aberrant splicing by mutant P. 53.
32:27So we just went with the gaps to start
32:30with because of course of the relevance
32:32and the path and the Keras pathway.
32:35But we are there is a student in the lab
32:37who's actually trying to understand.
32:39What?
32:40What is the role of the other splicing
32:42changes in other M RNA's that are not gaps?
32:47And presumably, in that context,
32:49I mean, it's kind of interesting
32:50that the gap 17 effect is so
32:53kind of singular in a sense,
32:55and you presumably in the other cases
32:57it's really going to be a combination
32:59that's going to be the constellation
33:00of those changes that are key,
33:01which is going to be interesting
33:03but tough to tease out
33:04exactly. So, as I mentioned before,
33:06we are seeing that 32 gaps encoded
33:10by the genome of of 120 dots that
33:13are encoded are being differentially
33:16spliced, we manipulated.
33:17One, but if you imagine manipulating
33:20several of them and forcing
33:22policy axons to be excluded,
33:24the the effect might be synergistic
33:26in terms of the the the cell
33:29proliferation and the tumor growth.
33:33Sure. Any other questions in the chat?
33:39So we don't seem to have at
33:40the moment of so that we could,
33:41and we we we should probably move on.
33:44So thank you very much, Lisa.
33:46That was a fascinating stuff with
33:48the enormous about everyone.
33:49And to think about it. Thank you.
33:52So, so let's move on for the second
33:57half to Doctor Rosa Vinod Zikula.
34:01So Doctor Zickler is an assistant professor
34:04of medicine and digestive diseases.
34:06She received her PhD from the
34:08university app Autonomo de Barcelona.
34:11I apologize for my pronunciation
34:13and oncology and her postdoctoral
34:14training at the Institute of Cancer
34:16Research at the University of
34:18Illinois and at Yale University.
34:20Dr Zickler's long term goal is to decipher.
34:22Known genetic alterations that
34:24predispose to colorectal cancer
34:26development and her research focus
34:29is on understanding molecular and the
34:31molecular characterization of sporadic
34:34and hereditary colorectal cancer
34:36with an interest in understanding
34:38the biological differences among
34:40racial groups to develop her
34:42translational research doctors.
34:43Zigler is a key player in several
34:47repositories and consortia that
34:49recruit cancer patients and
34:51then collecting biospecimens.
34:53And and clinical data and Doctor
34:55Nikola will will tell us about defining
34:59new pathways in colorectal tumors
35:01with mismatch repair deficiency.
35:04So thanks so much Rosa for doing this.
35:05I really look forward to your talk.
35:08Thank you, let me share.
35:15Can you see properly?
35:18OK, so thank you so much for giving
35:20me the priority to show you all our
35:22most recent data on the topic of
35:25mismatch repair, deficient tools.
35:27So the outline of the talk is going to.
35:30I'm going to explain you give you an
35:32overview of the mismatch repair and the
35:34phenomena of microsatellite instability.
35:36And then I will explain you the
35:38clinical phenotypes and challenges in
35:40the molecular that I diagnosis of the
35:43tumors that have mismatched efficient.
35:45Then I will explain you the association.
35:49That we are describing between
35:51deficiency of RAQUE and DNA helicases
35:54in Lynch like syndrome cases.
35:56And then I will show our most
35:59recent publication that describes
36:00the identification of tumors with
36:02a high likelihood development and
36:05immune response through mutational
36:07signature profiling.
36:10So here in the left you can see that
36:12it's a cartoon that shows the the mosque
36:15important for the main proteins that are
36:18involved in the mismatch repair system.
36:20The mismatch repair system is the inner
36:23repair system that identifies mismatches like
36:26single base base or like larger mismatches.
36:29And there's two main complexes,
36:31the mute test that it's formed by message 6
36:34and Message 2 and Message 3 and a message 2.
36:38So these proteins are the
36:39first ones to recognize them.
36:40As my tools and then the mute L complexes
36:44recruited to help fix the the mismatches
36:48and mutl is formed by PMS two and MLH 1.
36:51So in the genome there are these
36:54sequences that are called microsatellites
36:56that are prone to acquire alterations
36:58when any of the proteins of the
37:01mismatch repair not working.
37:03So here you can see here you can
37:07see sorry this is on the way.
37:09Here you can see a microsatellite
37:11microsatellites are short and repetitive
37:13sequences present in coding and
37:15non coding regions of the genome.
37:17And when the any of them is not
37:20working this Microsoft.
37:22That's accumulate deletions or insertions.
37:24So when the size of the microsatellite
37:27cannot be properly kept during
37:29replication of DNA in the cells,
37:32the phenomenon of microsatellite
37:34instability isn't identified in tumors.
37:40So I MSI can be identified
37:42in a variety of tumors,
37:44but as you can see here on the table
37:46and in the graph in the material tumors,
37:48colorectal and stomach are the
37:51tumors that have a higher incidence
37:54of microsatellite instability.
37:56So, in colorectal tumors,
37:58about 10% of a sporadic tumors have
38:01mismatch repair deficiency and these
38:03deficiencies due to CPG island promoter
38:05musculation of the gene mileage one,
38:08which I show you that it's a.
38:10It's one of the two proteins
38:12that form the metal complex,
38:14so when there's a promoter methylation,
38:17there's an addition of transcription
38:19of the gene and it and resulting in
38:22the loss of expression of the protein.
38:24So here you can see the difference.
38:26Between normal expression by
38:28immunohistochemistry and loss
38:30of expression and a significant
38:32number of these tumors,
38:35they also present this hot spot
38:37mutation in the Bureau of Gene.
38:39Here you have the mutation and these
38:42two are molecular events are used
38:44to differentiate and tumors that
38:47develop through a sporadic events.
38:50Then the tumors that develop MSI
38:52but they are developing in the
38:55setting of hereditary disease.
38:58So Vince Syndrome is the the tumor.
39:02It's cancer syndrome.
39:03There is due to germline mutations
39:06in this mismatch repair genes.
39:08It's actually the most common
39:10cancer syndrome of all it's present.
39:12It's estimated that one in 270 people
39:14in the US carry one of the mutation
39:17in one of these genes and these
39:20individuals have present this syndrome
39:23presents as penetrance about 70 to 80%,
39:26which means that.
39:27That in in 70 to 80% of the cases
39:30individuals that carry a mutation,
39:32they end up developing
39:33cancer and when they develop,
39:34cancer is usually associated
39:36with an early age of onset.
39:38So clinically,
39:39Lynch syndrome patients present with
39:42fewer polyps than other colorectal
39:44cancer inherited syndromes,
39:46and the tumors localized in
39:48the right side of the column.
39:50And this lynch patients.
39:51They have a high risk of
39:53developing multiple cancers.
39:55Colorectal cancers are
39:56diagnosis or over time.
39:58And another clinical feature
40:00that it's important to remember.
40:02Sorry,
40:02remember from these patients is
40:04that the Lynch syndrome is actually
40:07a multi cancer syndrome affecting
40:09different organs and here you can
40:11see the list and it's significantly
40:13important to remember that because
40:15actually female lynch patients they
40:17developed for example like in the material,
40:19they have a higher incidence
40:21of developing endometrial than
40:22colorectal cancer.
40:25So because I explained you that
40:27Link syndrome is the most common
40:29cancer syndrome and because of
40:31all this clinical features that
40:34these patients have nowadays,
40:35all in the midfield and Jay
40:38cancers are are supposed to be
40:40tested for the for Lynn syndrome.
40:43So how this works is all these cancers.
40:46They are tested with immunohistochemistry
40:49for the expression of the four main
40:52proteins of the mismatch repair if.
40:54Because of the expression in MSH
40:572 MSH 6 or PS2 is identified,
41:00then the patient should be referred
41:02to cancer genetics for testing and
41:05contrary if the loss of emulate
41:07one or PMS or the conduction
41:09of emulate one and PMS two is
41:11identified by immunohistochemistry,
41:13then there is the one.
41:15Methylation should be tested and
41:17if there is no methylation then
41:19the patient should be referred
41:21to cancer genetics and in any
41:23way if anyone in the identifies.
41:25MSI case, but there was no.
41:28I'm even Histochemistry tested,
41:30but the physicians have a clinical concern.
41:33Then these patients should be
41:35preferred to cancer genetics.
41:39So in general, in the cancer genetics
41:41clinic was we've been facing,
41:43is that about 50% of the suspected link
41:46syndrome patients that are referred.
41:48They actually test negative for Jim for
41:52having germline mutations in the genes.
41:54And this case is where name as Lynch
41:56like syndrome because they are similar
41:59to lynch like but there's no mutations.
42:02So as as a definition these lines
42:04like syndrome patient patients,
42:06they develop tumors at the MSI.
42:09They don't have resolution of image one.
42:11They don't have the hotspot be
42:13600 imitations and they don't
42:15have a germline mutations either.
42:17So what are these things like cases they
42:20actually could be Lynch syndrome cases,
42:23but that due to difficulty on
42:25identifying mutations or because they
42:27have like encrypting mitigations.
42:29Maybe we have not been able to
42:30then defy them,
42:31or they could actually be heritary
42:34cases that they might be due to general
42:37mutations in other genes and that.
42:39They end up developing MSI as a driver
42:44effect, not as a cancer driver effect,
42:49sorry.
42:51Sorry that they developed because
42:54other germline mutations but they
42:57actually the MSI was an effect of
43:00the development of cancer but they
43:02could just be sporadic cancers.
43:05So to address these these challenges,
43:08we have developed 2 main projects,
43:11one and the general level and
43:13another one at the semantic level.
43:15The general level with our aim was
43:17to identify the current deficient
43:19DNA repair genes and the cellular
43:22consequences that contribute to the
43:24development of colorectal cancer in
43:26lines like patients and at the somatic level,
43:29we aim to define molecular factors
43:31in the three types of mismatch.
43:34Deficient tumors the lynch lynch like
43:36and the viraf methylated ones which will
43:39contribute to diagnosis and treatment.
43:44So so our collaborations with the
43:47correct with the current concerns,
43:50we were able to describe the patients have
43:52a higher frequency of family history of
43:55colorectal cancer than sporadic cases,
43:58and you can see here how the standardized
44:01incidence ratio was 2.2 for the links in
44:04comparison to 0.48 for sporadic individuals,
44:08and and and these incidents
44:11of family history was.
44:13Actually lower than lead,
44:14so this kind of foods.
44:16The Linge like phenotype and
44:19in between between.
44:21Lynch and Sprite.
44:23We're also able to to show that the
44:27average age of diagnosis for Lynch like is
44:31significantly younger than sporadic cases.
44:33So these two features are suggest
44:36that a potential unidentified genetic
44:38predisposition induced in this in a group,
44:41at least in Group of Lynch
44:44like syndrome patients.
44:45So to address this,
44:47and because we believe that that is the case,
44:49we develop a a study including 654
44:55individuals from our Chicago Colorectal
44:58Cancer Center consortium cohort,
45:00and we performed that link screening
45:03testing that I mentioned before
45:06we identified 23 suspected links.
45:09Lynn syndrome.
45:12So from those we were able to have
45:16germline DNA from 15 of them and
45:18we perform XM sequencing and we
45:20identified that four of them were
45:23actually engaged and eleven were links
45:25like were classified as Lynch like
45:28because we didn't find limitations.
45:30So then we take it one step further
45:33and we wanted to identify if if
45:36any of these links, like patients,
45:39had mutations in other DNA repair genes.
45:42So we analyze 162 DNA repair genes and
45:46we were able to see that this links,
45:49like patients.
45:50They had the higher mutational burden
45:52and comparison to lynch to the TCG,
45:55a colorectal cancer cohort,
45:57and to control without cancer.
46:00So specifically,
46:01we identified four loss of function variants,
46:03one in body, one one in Werner,
46:06one in MCPH one and one in Rev 3.
46:11So then after this first study
46:13that we identified that links
46:15like were in bridge with mutations
46:16in the inner river jeans,
46:18we include decided to include two
46:21different independent series of lines,
46:23like patients to try to identify
46:25genes that maybe would be recurrently
46:28mutated in this in this phenotype.
46:31So when we did that in the first
46:34series with unified 6 genes that
46:36were mutated and had lots of
46:38function variants and interestingly.
46:40We found the same splicing variant
46:43in two different patients in the
46:45regular 5 gene and we actually
46:48perform a kinship analysis to show
46:50that and to prove that these two
46:53patients were not genetically.
46:55And they were not genetically.
47:02There were no this related.
47:05Sorry, because these patients were
47:07both coming from from Spain and we just
47:10wanted to make sure that there was no
47:12any family relation that we don't know.
47:16And then when we developed the analysis
47:18of the other series of patients,
47:20we again identified another loss
47:23of function variant in regular 5.
47:26So with that, if you've been able
47:28to follow my my talk and the the
47:31the notification of mutations
47:33in our original serious,
47:34we have identified 4 different
47:37mutations in genes that belong to
47:40the Dracula DNA helicase family.
47:43So here you can see the five the
47:47five proteins that are in this
47:49family regular one bloom Werner
47:51regular four and regular 5 and all
47:54of them share the same helicase.
47:57I mean.
47:58So these are the individuals
47:59that we have identified the two
48:01individuals with the same splicing,
48:03one with the insertion and
48:05from the original cohort.
48:06We also identified this individual
48:09with a mutation.
48:11So after that we were interested in
48:13knowing if maybe the mutations in
48:16this in this family of genes were also
48:19recurring in other cancer friendships.
48:21So to do that?
48:23First took individuals that were
48:25referred to the Smilo Cancer
48:27Genetics and Prevention program
48:28that when they were referred,
48:31they and they were tested
48:32and we in the clinic.
48:34They didn't find any mutations,
48:36say many known cancer predisposition genes.
48:39So we perform XM sequencing in 156
48:43breast cancer patients in 75 individuals
48:46that had different types of tumors
48:49that were not breast breast tumors.
48:52We'll sync clouded,
48:54MSH and PC.
48:55These are very rare type of familial
48:59colorectal cancer that affects individuals
49:02in different generations and that
49:04they develop cancer at the young age.
49:06But these individuals don't have MSI RMS.
49:10And lastly,
49:11we also identify mutations in the DC G.
49:14So with this analysis we were
49:16able to see that actually,
49:18like the higher a little frequency variants
49:20in DNA repair genes that are not the.
49:23Compare and then they are not know
49:26well established cancer predisposing
49:28genes and we all the identified
49:32mutations in the REQ DNA helicases
49:35in the lynch like phenotype.
49:38So then we went back to the
49:42families that we were able to.
49:44Contact again to in the defy
49:47if the mutations were shared.
49:49If these mutations with shared
49:50with other family members,
49:52so here these are the three families that
49:54will have with mutations in the right QL.
49:56So family A&amp;B are the the ones that
49:59share the same splicing variant,
50:01so here this is the program that
50:03developed for family aid that
50:05developed colorectal cancer.
50:06At 63.
50:07We were also able to sequence
50:09the tumor of this individual and
50:11we also found a missense variant
50:14in the in the tumor of this.
50:16Of this patient,
50:17and then the brother of this program
50:20had a small bowel cancer and he was
50:22also a carrier of the mutation.
50:24The family we we.
50:26This was the program that they
50:28are of collector cancer at
50:2964 very strong family history
50:32and then we tested the two sons.
50:35That one was a carrier and the
50:37other one was not a carrier and the
50:40rate of diagnosis was under 40s.
50:42And lastly this last one we.
50:44This was the program developed
50:47colorectal cancer at 66 and we
50:50tested this son that also had
50:52sorry also had the mutation.
50:55But however, these individuals
50:57in the second generation,
50:59because they are in their 40s,
51:01they might have not been able
51:03to develop cancer yet.
51:04So this this course aggregation
51:06study was not definitive.
51:10So then we wanted to to test what was
51:13the effect of having a heterozygous
51:15loss of function barrier in intestines.
51:19So to do that we went back to to our
51:22contacts in Spain and we extracted cells,
51:26extracted blood samples from the two of
51:29these songs that I show you in family.
51:32That one was a carrier and
51:33the other one was not.
51:35And we extracted that RNA.
51:37We did the red transcription and qPCR.
51:40To show that actually the the level of gene
51:43expression was significantly lower in the in,
51:46in the brother that had the mutation.
51:50And to test the effect in the Warner,
51:53we had to use a different approach because we
51:56didn't have access to that family anymore.
51:58But we were likely to acquire
52:01Lymphoblastoid cell line from family that
52:04had there were these one mutation and
52:08heterozygosity and from a control also.
52:11And this mutation is the one
52:12that the cell lines have.
52:14And it's just like a loss of function
52:16mutation just for an amino acids
52:18down the line from the actual.
52:20Colorectal cancer mutation that
52:21we found in one of the patients.
52:23So we extracted.
52:24We grow the cells.
52:25We extracted proteins and we show that
52:28again that there is an effect on the
52:31heterozygous and the protein expression.
52:34So with these we show that when
52:36there is a when these individuals
52:38have a heterozygous well as a
52:40function in these genes,
52:42they actually have a downregulation
52:44of the gene and protein.
52:47So then we were interested in knowing
52:49well if there is a downregulation,
52:51what's happening with the activity
52:53on the activity of the genes,
52:56and how is that?
52:57How are these sales managing DNA damage?
53:00Because again,
53:00remember that these are DNA repair genes.
53:03So to do that we grow the cells and
53:06we perform a flow cytometry analysis
53:08that was actually testing the
53:11quantity of forceful relation of the
53:14serene 139 residue of the history.
53:16Age to ax as an indicator of the
53:19damage and DNA double strand breaks.
53:23So when we did that,
53:24we determined the phosphorylation
53:26at different time points and the
53:29black are the wild type cells and
53:31the and the and Gray are the the
53:33ones with the headers I use,
53:35so it's true that that the first time
53:38the first time point might be a delay
53:41on the on on the phosphorylation we see
53:44that on the other time points there
53:47is a higher dose of the frustration and
53:50therefore an indicator that these cells have.
53:52The higher DNA damage
53:55and here as you can see,
53:57this is the difference
53:59between the heterozygote,
54:00the the heterozygous that has
54:02like a higher phosphorylation so.
54:07So right now we are also doing more
54:10analysis and we are testing for for example,
54:13for the effect of these variants
54:15in cell cycle because some of our
54:18preliminary data showing that maybe
54:20these cells are actually arrested in G1,
54:23but we have not had this data yet.
54:27So in conclusion from this aim
54:29I we believe that
54:30heterozygous loss of function
54:32variants in DNA repair genes
54:34such as Warner and regular five,
54:37could predispose to tumor development
54:39because they are enriched among
54:41the lines like cancer phenotype.
54:43They lead to gene down regulation
54:46and they increase DNA damage.
54:49So now turning it to the end
54:51two at the somatic level.
54:53Had to do develop these aim.
54:55We also included two different
54:57independent series of tumors that
54:59mismatch repair deficient tumors
55:00from the three different types,
55:03and we develop exam sequencing
55:05to identify somatic variants
55:08and loss of hydrazoic events.
55:11And with with this data,
55:12with this excellent data,
55:14we also were interested in defying
55:17the contribution of mutational
55:19signatures to these tumors.
55:21So mutational signatures are like
55:24a fingerprint of of the portrait
55:27of the mutations that the tumor has
55:30acquired over the development of the
55:33tumor and they some of them are well
55:36established and they are associated to,
55:37for example,
55:38exposure to carcinogens and other ones.
55:41Associated like in the case of the mismatch,
55:43repair to deficiency on DNA repair pathways.
55:47So they the these are the six
55:51current well established signatures
55:52that are associated with deficiency
55:53of the mismatch repair.
55:55So when these tumors have,
55:57when the tumors have deficiency and you
55:59analyze the the mutational signatures,
56:01you can see this one so so we
56:03were interested in knowing what
56:05was the contribution of these
56:07signatures to each of the tumors.
56:09So let me explain.
56:11So this colorful graph here.
56:13So we first read identified what
56:15were the mutational signatures that
56:17were contributing the most to each
56:19of the tumor and then we perform
56:21clustering to see whether the groups
56:23of whether the tumors that have
56:26a similar contribution of those.
56:29So here each each row is 1 tumor
56:33and each column is rotational.
56:35It's a contribution of to
56:37the mutational signatures,
56:39and here we are also having the phenotypes.
56:42So in here you can see the tumor
56:45is linch light lynch or the MSI?
56:48Isolated.
56:48And then in the last column here we
56:51are showing that which is the protein
56:53that each of these tumors have.
56:56Most of the expression.
56:57So when we perform this analysis,
57:00you can see with identified 2 of the
57:03that mutational signatures based on
57:05the the contribution of SBS 26 and 15,
57:10which are very well established.
57:12Mutational signatures associated with
57:13deficiency of the mismatch repair
57:15identified first the two clusters
57:17that are in breach with the Lynch.
57:19And the lynch,
57:20like and then we then defied this
57:23cluster that has a higher contribution
57:25of the tumors that are missing.
57:28MSI,
57:29MSI dated.
57:32So here I'm I'm showing you the
57:35different features associated
57:37with each of the the clusters,
57:39and as I mentioned,
57:40cluster two is enriched with MSI dated
57:43and also this cluster has specific
57:46clinical features that are well
57:48established with this type of tumors,
57:51which are that they develop preliminarily
57:54and female patients at an older
57:56age and that they are associated
57:58with the bright side location.
58:02So as a molecularly,
58:03as I as I explained you,
58:05this cluster is associated with
58:07thousand expression of mutl and mainly
58:10due to the manipulation of image
58:13one and and then there's tumors.
58:16They also have the higher
58:18number of frames if mutations,
58:19even though there is no difference in
58:22tumor purity that could be affecting this.
58:24But we didn't see that there
58:26was a significant difference,
58:27and they don't have a significant
58:28difference in their own TMB.
58:29So to like two more.
58:31Additional burden,
58:32so it's specifically to the friendships
58:34and what this suggests is that the
58:36the the tumors in this cluster.
58:38They actually have the higher
58:40level of Microsoft the instability.
58:44So one of the results of having
58:46a higher level of microsatellite
58:48instability could be that these tumors
58:50have a different new antigen load,
58:53so new antigens are these peptides
58:56that are generated after somatic
58:58mutations arise in the tumor.
59:00And as you can see here,
59:01you can see that the normal protein and
59:03this is a missense mutation in the tumor,
59:05so this is going to be 1 amino
59:07acid different from the self.
59:09The regular normal protein,
59:11but no antigens that are that are there.
59:15Develop from frame.
59:16Frameshift mutations are significantly
59:19different from the normal because
59:21they introduce a lot of well.
59:25Insertions and deletions.
59:27So these proteins.
59:29These peptides are significantly
59:31different from cells,
59:32and these new antigens which represented
59:35here by this dot are presented from
59:38through the HLA 1 receptor to the TCR.
59:43To the T cell receptors,
59:45and this is obviously a very simplified
59:47version of what's happening,
59:49but then when this is when when this
59:52is happening then the T cells identify
59:56the tumor cells as as non self,
60:00and then they're going to
01:00:02start the immune response.
01:00:05So we wanted to see how these new
01:00:09antigens and the direction of the HLA.
01:00:12Image of the patient were
01:00:13occurring based on the different
01:00:15clusters that we are defined.
01:00:17So to do that we use several
01:00:21bioinformatics pipelines.
01:00:22We use Poly solver to predict
01:00:25the HLA one alleles that we know
01:00:28that there's three of them using
01:00:31the germline XM sequence data.
01:00:34Then we use unaware tool to
01:00:36annotate all the mutations that
01:00:39we had identified in the in the.
01:00:42More excellent sequencing and
01:00:43then we use net,
01:00:45MCA,
01:00:46MHC pan that actually identifies
01:00:49what are what.
01:00:51What are the interactions between
01:00:53the HLA's and the new antigens?
01:00:57And then we took it one step
01:00:58further and we use narrow pred.
01:00:595 that actually this algorithm
01:01:03computes the recognition potential.
01:01:05So what it does is it provides a likelihood
01:01:09that this interaction is going to occur.
01:01:12And it's based on on the immune epitope.
01:01:15It's it's.
01:01:16It's this prediction is based
01:01:19on the TCR receptor rapporteur,
01:01:21that it's that it's.
01:01:25That he's present in the
01:01:27immune epitope database.
01:01:29So with that we took this likelihood
01:01:34and this recognition potential,
01:01:36and we score them,
01:01:38and we identified the ones that
01:01:40were at the highest 10% tile and
01:01:42the ones that were at the lower 10%.
01:01:45So we assume that if there is no selection,
01:01:49then the these interactions in
01:01:51the temple in the top percentile.
01:01:54Between the new antigens and the HLA,
01:01:57one should be the distribution of
01:01:59these alleles should be similar to
01:02:02the distribution of the patients
01:02:04and the little frequency in the
01:02:06patient population.
01:02:07So to test this hypothesis we we
01:02:11compare the actual frequency of the
01:02:14alleles in the patient population for
01:02:16each of the different clusters and
01:02:19the frequency and the distribution
01:02:21of the alleles in the ones that
01:02:23are selected as having the higher.
01:02:25Likely for the recognition and
01:02:27what we
01:02:28identified is that actually there was
01:02:31one specific allele B702 that were
01:02:34significantly in breach in this in
01:02:37the top 10% recognition potential,
01:02:40which that was not happening in
01:02:43the lower set of of interactions.
01:02:46So we think that the specific actually
01:02:50wanna leaves like the B702 could promote
01:02:53stronger immune immune response.
01:02:55And these tumors that are the ones with
01:02:58the higher microsatellite instability.
01:03:00And we believe that these down the line
01:03:02could be affecting the immune response
01:03:06of these tumors to and and how to
01:03:11immune immune checkpoint inhibitors.
01:03:13So obviously this is the the beginning
01:03:16of like expanding this work in
01:03:19the area of immune response by the
01:03:22immune checkpoint inhibitor response.
01:03:25So in conclusion,
01:03:26for him two molecular differences between
01:03:28the three different types of mismatch repair,
01:03:31deficient tumors could have a direct
01:03:34implication and immune response specific.
01:03:37One else could be driving the presentation
01:03:40of neoantigens among mismatch repair
01:03:42deficient tumors with the highest
01:03:44level of microsatellite instability.
01:03:46We probably specially this work,
01:03:48so overall the take home message is that
01:03:51our studies show that there's novel
01:03:53molecular heterogeneity among these.
01:03:55Under the efficient tumors and
01:03:57that understanding the clinical
01:03:59pathological features associated with
01:04:01this heterogeneous heterogeneity is
01:04:03essential to accurate diagnosis and
01:04:06prediction of treatment response in
01:04:08the setting of personalized medicine.
01:04:10And our future directions.
01:04:13It's to understand the molecular
01:04:15mechanism that associate trequel 5
01:04:17and Warner deficiency with this type
01:04:19of tumors identify immune regulators
01:04:21that determine response based on the
01:04:24type the specific type of mismatch
01:04:27repair deficiency,
01:04:28and investigate also the treatment
01:04:30response to immune checkpoint
01:04:32inhibitors based on this type of
01:04:34mismatch repair deficiency.
01:04:36So with that,
01:04:37just acknowledge our funding sources
01:04:39Martinek Albuch that is the the first
01:04:42dog in my lab that has one done most
01:04:45of the work and my collaborators
01:04:47in the US and also in Spain.
01:04:49And I'll be happy to take any questions.
01:04:53Thank you very much Rosa.
01:04:55A terrific work that's very interesting and
01:04:57we do have a couple of questions in the chat,
01:05:00so which hopefully I can read properly.
01:05:05So the first is from Jeffrey
01:05:08Townsend and Jeff asks,
01:05:10is the association of BRAF V600E
01:05:14with MSH mutation purely mutational?
01:05:17Or is there some more complex biology
01:05:20to the association and he asks because
01:05:22the trinucleotide signature in.
01:05:24Used by MSH is especially likely to
01:05:27make the B Rav 600 to E mutation. Yeah
01:05:30so so. The BRAF mutation in colon cancer
01:05:33is associated with the serrated pathway,
01:05:36so that's like the more biological.
01:05:38It's not this type of tumors,
01:05:40but for the for the Ms,
01:05:42I believe it's more like a
01:05:44motivational association,
01:05:45but the one that has been more
01:05:49described biologically is the one
01:05:51that the the the serrated pathway.
01:05:54Had tumors that they were writing.
01:05:56Passwords are developing,
01:05:58but this is like more like.
01:06:01Mutational that we used to
01:06:03mainly separate the sporadic
01:06:05from the hereditary ones.
01:06:09OK great thanks.
01:06:10And then the next question is
01:06:13from Ryan Jensen and Ryan asks.
01:06:16And one of the potential roles of
01:06:19of REC QL 5 is to prevent aberrant
01:06:22homologous recombination by displacing
01:06:24RAD 51 off single stranded DNA.
01:06:27And he wonders if in tumors from
01:06:29patients with loss of function,
01:06:31mutations in REC queue do you see
01:06:34increased chromosomal aberrations?
01:06:36Sister chromatid exchanges,
01:06:37or perhaps increases in microsatellite
01:06:40contraction or expansion.
01:06:43So all of these we have, we.
01:06:46There's so the the work done in Q L5
01:06:50and colorectal cancer is not very vast.
01:06:53So so right now what I can say is that
01:06:57we we just engineer a cell line that is,
01:07:00that has these mutations,
01:07:01which rupees per so we are going to
01:07:03have the cell lines that have like the
01:07:05heterozygous and homozygous and Val types.
01:07:07So we are going to be testing these
01:07:10kind of events that Brian is suggesting.
01:07:13So I don't have that information
01:07:15yet where I know that, for example,
01:07:17for Frank L5 is that there there's been one.
01:07:20There was one old paper that was
01:07:22showing that interestingly regular 5
01:07:26downregulation was identified in MSI tumors,
01:07:29and so I think that there is more
01:07:32than we can be learning about this
01:07:35and and I think that that's going
01:07:36to be one of our like next steps.
01:07:40Great thank you and I had one quick question.
01:07:43When you were going
01:07:45through and looking at the.
01:07:46The the red queue and other
01:07:49mutations in the Lynch like syndrome.
01:07:51I didn't have a sense for for whether
01:07:54it was clear whether that they're
01:07:55all loss of function mutations,
01:07:57like for example the T31K in that one
01:08:00family is that is that some is that a is
01:08:03that one so that one was mutation that
01:08:05we found in the tumor of that patient.
01:08:08We have not been able to
01:08:10test the other individual.
01:08:11So the germline variants that
01:08:13we are even defining in the
01:08:15germline are all loss of function.
01:08:17But we only have been able to test
01:08:201 tumor from these individuals.
01:08:23I can tell you, not for AQL.
01:08:25I know a lot of the data for
01:08:27one not a lot few data from
01:08:30Warner mutation somatically.
01:08:32There is the there identifying loss
01:08:35of function mutations and actually
01:08:37these tumors that have loss of
01:08:40function mutations in Werner they
01:08:43have a significantly higher number
01:08:45of them in comparison to the ones
01:08:48that don't have mutations there.
01:08:49MSI. So again another kind of.
01:08:53Another clue that there have there
01:08:55might be some some association between
01:08:58deficiency in these genes and MSI.
01:09:01However, association doesn't mean causality,
01:09:04so this is what I think that
01:09:06is what we actually need to do.
01:09:07More research to figure out
01:09:09these needs one or the other.
01:09:12Good, well I think it's been a great session.
01:09:18And lots of good questions
01:09:19and and two fantastic talks.
01:09:21So I'd like to just to finish
01:09:23by by thanking Luisa and Rosa.
01:09:25Very much for really giving very
01:09:27stimulating and exciting talks,
01:09:28great grand rounds and thank
01:09:30you very much everybody.
01:09:32Thank you. Bye bye bye.