Epigenetic regulation of immune evasion and drug resistance in melanoma and beyond

December 07, 2022

Information

Yale Cancer Center Grand Rounds | December 6, 2022

Presentation by: Dr. Qin Yan

ID9236

To CiteDCA Citation Guide

00:00Got it. So it's my distinct pleasure
00:03to introduce Chen Yan to us today.
00:06He's one of the invited speakers for
00:08this year for the Melanoma program.
00:10So for those that don't know,
00:11Melanoma program is a fairly well
00:14established 1 going back to the 1980s
00:16when we started the first what wasn't me,
00:18other folks started the first
00:21interdisciplinary disease team,
00:22John Kirkwood and Steve Arians specifically.
00:26And then as the years went by,
00:27Ruth Taliban, who's sitting here,
00:29wrote the first version of the Yale.
00:30Or in skin cancer first funded in I
00:33think 2006 or 7 or something like that?
00:36We just submitted the 4th iteration.
00:40So one of the best things about working
00:42here is actually our colleagues and I
00:44think Chin actually exemplifies that.
00:46So you came to us from from Harvard where
00:50he worked in the lab of Bill Kalen,
00:52actually on epigenetics and renal
00:53cell carcinoma.
00:54But at some point it became clear
00:56that some of the things that he was
00:58studying were very applicable to
01:00Melanoma as well and he submitted a
01:03developmental research project to
01:04the sport in its previous iteration.
01:06And that subsequently blossomed
01:08to a full project.
01:09We are thrilled to have Chen working with us.
01:11We couldn't,
01:12we couldn't ask for a better collaborator,
01:15both in terms of his scientific depth
01:16and in terms of his personality.
01:18He's definitely one of us.
01:19And I actually don't care that he's the
01:21scientific Co director of the breast unit.
01:23As far as we're concerned, he's ours.
01:25So without further ado,
01:26chin, the floor is yours.
01:27Thank you.
01:31Well, thank you Harry for your kind
01:33introduction and and and I was also
01:35like to thank my normal program for
01:37nominating me here to present here.
01:39I would say Cancer Center ground is
01:42one of the event that actually led
01:44me to work on Melanoma and on my way
01:48back from Grandma's talks and I was
01:50working together with Marcus Bosenberg.
01:52I bought a decade ago and we were
01:55talking about Jerry 1B who might be
01:57which might be important in Melanoma.
01:59I was working on Jerry one.
02:01Because I generally knockout my and well,
02:04we just started the collaboration and
02:06it's a very fun collaboration and this is
02:08something I'm going to tell you today.
02:13So let me get this started.
02:17Fixed the pointer.
02:22So this is my disclosure.
02:25So what I'm going to do is first
02:27you give you a very quick overview
02:29of cancer epigenetics and then you
02:31tell you two stories related to
02:33this histone demethylase KDM 5B,
02:35how it recognizes drug resistance
02:38and immune evasion.
02:40So as many of you know,
02:43the epigenetics is study of
02:45health heroical traits that does
02:47not depend on the underlying DNA
02:49sequences and the major epigenetic
02:51mechanism include DNA methylation.
02:54Put his own structure,
02:56histone modifications and non coding on it.
02:58There's a number of regulators of
03:01IPG netting mechanism including the
03:03coronary modernness which are involved
03:06in moving the nuclear zones around
03:08and rider eraser and the readers
03:11of histone or DNA modifications.
03:15So what I'm going to tell you
03:17today mainly focus on KDM 5B which
03:19is an eraser which is involved in
03:22removing a certain modification
03:23and sandbag one which I'm touched
03:25upon which is the right approach.
03:30So many of you are quite familiar
03:32with this hallmarks of cancer
03:34and what I'm going to tell you a
03:37little bit about is the immune,
03:39the immune invasion that
03:41cancer cells have to achieve.
03:43And if you look at it on the right side,
03:46this is a new,
03:47those are new hallmarks that have
03:50been added to the hallmarks of
03:52cancer and two of which actually
03:54quite related to epigenetics,
03:56including unlocking phenotypic.
03:59Plasticity and epigenetic reprogramming.
04:03So that's what I'm going to tell you today.
04:07So as many of you know epigenetic can
04:11epigenetics can regulate many of the
04:14cell fate and also a lot of mechanisms
04:17are involved in anti tumor immunity
04:20and just on the tumor cells for example,
04:23it has been shown DNA machination,
04:26histone modifications have been
04:28involved in regulating tumor antigen
04:31expression and cytokine secretion,
04:33PDL one expression and
04:35also chromatin structure.
04:37Have been shown to be important
04:39to response to cytotoxic attack,
04:41and those modifications are also
04:43important on other immune cells,
04:46including cytotoxic T cells,
04:48dendritic cells and macrophages,
04:50which is not duplicated here.
04:53So just a brief introduction on my
04:55laboratory and we are interested
04:56in cancer epigenetics of course.
04:59And one of the area we are interested
05:01in is a cancer metastasis shown here.
05:04Just one of the example where we showed
05:07one of the target called CCR two is a
05:10driver of breast cancer metastasis.
05:13And you can look at here,
05:15if you knock down CCR two,
05:16you can suppress the ability of those
05:18breast cancer cells to metastasis to the
05:20lung and if you overexpress CCR two,
05:22you.
05:22And rescue this phenotype.
05:24And of course we are very interested
05:26in the immune invasion part of the
05:28talk I'm going to talk about then and
05:31this is something that I will mention later.
05:33And so I'm not going to go over this figure.
05:36And we are also interested in drug
05:38resistance and I will tell you about our
05:41work on the drug resistance in Melanoma,
05:43but this is a diagram actually found a.
05:47Had breast cancer walk where we
05:49showed that trastuzumab resistant
05:51cells have increased oxidative
05:54phosphorylation and if you block
05:57oxidative phosphorylation with only a,
06:00if you combine that with transfusion level,
06:02you can actually regress the tumor
06:05formation by the drug resistant cells.
06:09As a matter of because we
06:10are interested in the area,
06:12we are also interested in
06:14developing epigenetic drugs.
06:15And I will tell you some of the work
06:18on KDM 5 inhibitor development.
06:20And this is a some work that we have
06:23done a couple years ago where we
06:26characterized I potent bromodomain
06:28inhibitor where we show that this
06:31bromodomain inhibitor and HW 870 can
06:33not only inhibit the ability of the
06:36cell tumor cells to grow but you can.
06:38Also hit on the macrophages by
06:41suppressing the expression of CSF 1A,
06:44critical regulator of macrophage
06:46polarization and the macrophage
06:49proliferation and this drug actually
06:51have entered the phase one clinical
06:52trial in in China and moving
06:54into phase two very
06:55soon.
06:57So my laboratory had been focusing
07:00on a group of England called KDM 5
07:03histone demethylase and and as you can
07:06see here this group of vendors have
07:09four of them and they they are called
07:12KDM 5 ABC D or Jared 1A1B1C and 1D
07:15and all of those have this team JC
07:18domain which is the Jumanji C domain,
07:21it's hydroxylase domain and the by
07:24hydroxylation of the methanation.
07:26Group and the removal of formaldehyde.
07:28They actually can demate the histones.
07:32So this group of landline can demonstrate,
07:35try and demonstrate nice thing
07:37four on histone H3.
07:39And because those machination
07:41marks are critical marks for
07:43actually transcribed genes,
07:45so by doing so this group of
07:48online can silence transcription.
07:51But that's not the whole story.
07:52And all those protein actually have
07:55other domains including 80 rich
07:57interactive domain which is involved
08:00in DNA binding and some of the PhD
08:03fingers which are involved in binding
08:05specific histone modifications.
08:07In addition,
08:08they can interact with many other
08:11proteins involved in chromatin remodeling
08:13and transcription recognition.
08:15So they have.
08:16It has been documented that
08:18this group ENDLINE cannot only.
08:20The transcription repressor they
08:21can be transcription activated
08:23in some other settings.
08:26So today's talk we'll we'll be,
08:28I'll be focusing on on this protein
08:31called Kadian 5B or Jerry 1B.
08:33Also another known name
08:35is called the PLU One.
08:38Because there's a number of.
08:41Evidence showing that uh
08:43Kadian 5B has oncogenic role.
08:46It was shown to be overexpressed
08:47in many cancer types,
08:48including skin cancer.
08:50Initially was identified as a
08:53downstream gene downstream of her
08:562IN breast cancer because it was
08:59shown to be downregulated by anti to
09:02anybody in her two overexpression cells.
09:05And these have been shown by 90 points
09:07group that is amplified in luminal
09:10breast cancer and it's a potential
09:12luminal linearity driving oncogene
09:13and we have shown in any in mouse.
09:18Me, I'd be single cells that are
09:22Jerry 1B can recruit Gallant St to
09:25regulate Fox A1 expression and that
09:27contribute to estrogen receptor
09:29target gene expression and in
09:31fact if you look at the estrogen.
09:35Except the positive tumors
09:37in for breast cancer,
09:38higher activity of Jerry won't
09:41be OK and 5B is correlated with
09:46poor prognosis of those patients.
09:50And and then you point out group
09:52has also shown that Kadian 5B can
09:55promote transcriptomic heterogeneity
09:57and this actually contribute to the
09:59therapeutic resistance and this is
10:02just one of the mechanism that this
10:05could contribute to resistance.
10:07I will tell you more about our
10:10work on a different angle.
10:12In addition,
10:13when we deplete KDM 5B first
10:16initially in breast cancer cells
10:18in syngeneic mouse model,
10:20you can see down regulation of
10:22KADIAN 5B can decrease the ability
10:25of those tumor cells to grow.
10:27And it was shown by Mihan honing
10:31scope that if you suppress.
10:35Expression in normal cells initially,
10:37those tumor cells actually grow faster.
10:39However, after you serial transplantation,
10:42those cells still crash,
10:43so suggesting that it's required
10:46for Melanoma maintenance instead of
10:49putting refreshing initial proliferation.
10:51And they was shown in multiple groups
10:54including ours that KADIAN file be
10:56is involved in drug resistance and
10:58shown here just one of the example by
11:01actually by a company constellation
11:03where they showed in multiple cancer
11:05cell lines including Melanoma.
11:06Here if you compare the effect of
11:10Canadian five inhibitor on parental
11:13cells or drug tolerant persister
11:17cells if you actually in this
11:19case they did a pre treatment of.
11:21Both S and and they should have shown
11:25that the KADIAN 5 inhibitor cannot
11:28inhibit the growth of the parental cells,
11:31but they can prevent the emergence
11:35of the drug resistant tolerant.
11:38Would DP cells or drug tolerant
11:41persister cells or drug resistant cells?
11:44In prostate cancer if we cost the
11:49KADIAN file be knockout model to
11:52the P-10 knockout model in process
11:55specific deletion and where you
11:57can see P-10 knockout model can
12:00form a prostate cancer.
12:02But if you get relocation 5B you can
12:05normalize the those those prostate
12:08tumors basically you can see the
12:12the the size is much smaller.
12:15Now I want to move back to Melanoma
12:18because this is a focus on our talk today.
12:221st when we looked at the TCA data set,
12:25this was done by Goran, a tenant in
12:28the graduate student at that time.
12:30In exposing like who is final right now?
12:34Where he showed that high
12:37expression is associated with poor
12:40survival of Melanoma patients.
12:43So now we decided to look at the
12:47Melanoma when we when we followed some
12:50of the work from Marcus Bosenberg.
12:53I have about my normal propagating cells.
12:57Was published more than a decade ago that
13:00if you look at the mouse Melanoma cells,
13:03you can sort them to three
13:06different populations,
13:07the P75P-75 positive cells,
13:11CD 34 positive cells or the
13:14double negative cells.
13:15If you look at the ability
13:17of the cells to form tumors,
13:19the CD 34 positive cells can form
13:23tumors very efficiently and the
13:25double negative cells can do so.
13:28With less efficacy but still works
13:31and the PDP 75 positive cells
13:34do not actually form tumors if
13:36they put them into modern mice.
13:39So we decided to look at this more
13:43systematically and when this is
13:45just a diagram show a table showing
13:47and many of the Yale University
13:50mouseman normal cell lines generated
13:52by Marcus Bosenberg Snapstory and
13:54those cell lines are generated was
13:57fun back six animals and you can do
14:00use those and use those cells for
14:05syngeneic transplantation experiments.
14:07And two of the cell lines we.
14:09Used here uh Young 11.7 which will
14:13actually I will use it also later
14:16on on for e-mail invasion studies
14:19and also young ones 3.3 cells.
14:21The reason why we chose those cells
14:24because they only have two populations
14:26so these 34 positive and city 34
14:28negative both of them can form too much.
14:30So this provide a nice system to look at
14:34the the population changes and when we put.
14:38Drugs on onto them.
14:41So we used the because those
14:44are mutant tumors and we treat
14:48those cells with rough inhibitor.
14:50In this case we use actually
14:52use the PX4 or three, two over.
14:57Stephanie.
14:57Umm,
14:57as you can see here,
15:00if you compare the parental cells and
15:02you have more CD 34 positive cells.
15:04If you look at the resistance the
15:07drug resistant cells you have
15:10which we delicate as the Yom Young
15:131.73 R or resistance,
15:15they have more city 34 negative cells.
15:18When you look at the the effect
15:22of the Bureau of inhibitor on
15:24those soap sub populations,
15:26you can see 3034 negative.
15:28Those are more resistant to be
15:32off inhibitor treatment because
15:34there's less growth inhibition.
15:37And this phenomenon is also reversible.
15:41If we treat those,
15:42you can see that they shifted
15:45to the left side,
15:46meaning CD 34 negative cells.
15:49However,
15:49if you remove the drug after a couple
15:52passages and they will shift it back
15:54to the parental cell population.
15:56So one of the things that was
15:58actually who it was a graduate student
16:01once Marcus and I basically should
16:04notice that there's an increased
16:06expression of KDM 5B if we treat
16:09those cells with BRAF inhibitor.
16:11And this is shown in young
16:141.7 cells, 3.377 cells.
16:15But also when you compare the parental
16:18with the resistance cells you see the
16:21similar increase of KADIAN fab expression.
16:25And this is reversible if you take
16:28out out and be rough inhibitor and
16:31the expression drops down and it's
16:34showing 1.7 cells as well as 3.3 cells.
16:39So when we did the genetic experiment
16:43when we knocked down kidding 5
16:46expression by a as shown here.
16:49We can see in the one point 11.7 cells,
16:54there's a decrease of CD34 negative
16:57cells after we deplete eighteen 5B.
17:00When we look at the phenotype and
17:02it's consistent to what other people
17:05have seen in other Melanoma setting,
17:07if you knock down killing five,
17:10you actually increase the ability
17:12of them to grow in vitro.
17:17And then those cells are actually
17:21more sensitive to inhibitor treatment?
17:25So this is not only.
17:28Two in most cells but also in human cells,
17:32this is you Max cells.
17:35If you knock down Killian 5B
17:37and you can see induction HPK
17:394 trimethylation which is the
17:42substrate of the enzyme and you
17:45can see those cells grow faster.
17:47However, they are less sensitive,
17:51they're they're more sensitive
17:53to BF inhibitor treatment.
17:57And if you look at this in animal models,
18:00uh, similar things happens when we treat
18:03cells with borough inhibitor and you can
18:06see KADIAN file being level increase
18:09and if you take away the inhibitor,
18:12you can see the level drops down.
18:18So Umm, and then we look at the if you
18:20look at the population of the cells,
18:23you can see increased.
18:26City City for negative cells.
18:30When we treat the cells
18:33with Burrough inhibitor,
18:34when you take out the inhibit that way
18:37and then those would not normalize.
18:40So Umm, and this is all consistent
18:43with our data and others have shown,
18:46which you're not showing
18:47here on that KADIAN filing.
18:48Hebetor can suppress the emergence
18:51of drug resistance cells.
18:53So to summarize this part,
18:55we see we have showed that 634
18:57negative cells are more resistant
18:59to BRF inhibitor treatment and BF
19:02inhibitor can increase C30 four
19:04negative cells and you can induce
19:07KADIAN file be up recognition and
19:10this is reversible and kadian Fabian
19:13N can reduce this population cells and
19:17induce drug resistance sensitivity.
19:20So now I want to switch switch gear
19:23to talk about uh immune evasion
19:25and firstly I want to start with
19:27this cancer immunity cycle on which
19:29many of you know have seen before.
19:32Basically this is a diagram showing
19:35that the cancer cells interact with
19:38the immune system and and there are
19:41many ways that cancer cells have
19:44adopted to evade immune evasion
19:47to evade the immune response.
19:49So as a matter of fact,
19:50because of this mechanism and some drugs
19:54have been developed including the anti PD1,
19:57PDL one anti 4 antibodies as well as the
20:02ways to push the effect of on the T cells.
20:06However,
20:06there's not much he's really actually
20:09known about the trafficking of T
20:11cells to tumors and the infiltration
20:12of the T cells into the tumor
20:15at that time when we started.
20:18And what's known about epigenetics,
20:20uh, in this setting,
20:22many of you are quite familiar with
20:25this concept about code tumor and
20:28hot tumor code tumor are not really
20:31responsive to another treatment,
20:33but the hot tumor will enable them to
20:37be responsive to even checkpoint block.
20:41And a sub couple for epigenetic.
20:43Uh regulators have been shown
20:46to be critical for this.
20:50Code to how the transition and
20:54if we treat this the tumors with
20:57a couple of inhibitors against
21:00those targets like DMT inhibitors.
21:03Two inhibitors. You can ship them to
21:06be more hard to become more hot hot.
21:10Um, in some of the settings,
21:12not in all the settings.
21:15And this is kind of related to what
21:18we are trying to do and at that time
21:22actually a couple of years ago before
21:25that and we have looked at the cadian 5B.
21:29And how it's related to other genes when
21:33we look at the TCA Melanoma data set?
21:36And to our surprise,
21:38actually KADIAN 5 expression was
21:40shown to be negatively correlated with
21:43many of the immune related genes.
21:46And if you look at those top
21:48signaling pathway,
21:48those are all immune system related
21:50genes and those are negative
21:52coordinate with expression.
21:53If you look at the the identity
21:56of those genes,
21:57those shown here are the gene
21:59names and on the right side of
22:01the Spielman score and you can see
22:03many of the silo kinds,
22:05for example interferon gamma and
22:10TNF A6796 O 10 which are involved in
22:12T cell recruitment are all negative
22:14coordinate with cadian fair expression.
22:17And some of the targets for immune
22:19checkpoint blockade PDL one CPT
22:21for also negative coordinate
22:23with KADIAN fabric expression.
22:24This will be important when we
22:26are trying to look at the e-mail
22:28checkpoint blockade resistant tumors.
22:32So when we looked at the KDM 5 be
22:35expression protein expression in
22:37melanomas and if you compare nonresponse
22:41boundaries with and responders,
22:43we can see increased expression location
22:455B which is shown in red in those non
22:49responders compared to the responders
22:51and and the quantification is shown here.
22:55So this motivates us to look at the role
22:57of Canadian file be using animal models.
23:00So as I mentioned earlier.
23:02Um, Marcus Bosenberg snapped,
23:04had generated a series of
23:07why UM or young models.
23:09Uh, one of the models that we started
23:11to use is this Yamaha 1.7 models.
23:14Yeah, this stand for ER stands
23:17for exposed to radiation,
23:19meaning those cells will radiate
23:22so that they have more mutations,
23:25they can generate more antigens that
23:28can be recognized by the immune system.
23:31So when we knocked out Kadian 5B in
23:33those cells, as you can see here,
23:35those cells can initially can grow,
23:38then they got fully rejected after a while.
23:41And the more importantly when we challenge
23:45those animals with the cells control
23:48cells which normally can grow very well,
23:51they never grow up.
23:53So this is very important imagine that if
23:56we have to treat patient with a drug and
23:59case for example in this case KADIAN 5
24:02targeting drug and those patient will not,
24:06will not have recurrence because
24:08they they will should never grow up
24:12because the immune memory response.
24:14So this is actually translated
24:15to 100% survival,
24:17which is uh, this is great.
24:20And then we look at the UM uh T
24:23cell infiltration when we compare
24:25the control cells and KADIAN file B
24:29knockout tumors at the very early stage,
24:32you can see the T cell infiltration
24:34either by e-mail,
24:35histochemistry as well as fax analysis.
24:39And there's another way to say that
24:41this is immune system dependent on we
24:44compared the ability of cells to grow
24:47in wild type cell wild type mice or rats.
24:50Deficient mice and as you can see here
24:54the the the the B6 is the wild type.
24:58Those two curves are what I have
25:00showed you before and if you look at
25:02the the ability of cells you grow in
25:05rectification mice the control goes
25:07here and then can be deficient once grow
25:11kind of similarly although slightly slower.
25:14So this basically set up the
25:17stage that Kadian 5B which is
25:20critical for immune evasion.
25:23So the next question is what's the mechanism,
25:26right?
25:27So how to?
25:29To understand this,
25:30we are look dead on a sequencing
25:33comparing Yammer 1.7 cells,
25:35probably knockout versus wild type.
25:38And we can see there's an induction
25:40of a lot of signaling pathway
25:43involved in DNA on a sensing pathway
25:46and showing here that generation analysis,
25:49the instrument parts where you
25:52can see there's an enrichment
25:54regarding like research pathways
25:56at Sonic DNA sensing pathway,
25:58those are all induced after
26:00you get rid of Canning Vale B.
26:02So now how does this actually work?
26:05And are those sensing pathway
26:07critical for the function of KDM 5B?
26:10As as many of you know that the
26:12double strand DNA double strand on
26:15the Ascension sensed through those
26:17pathways and double strand DNA is
26:19sensed through cgas sting pathway
26:21to big TV K1F3F7 and the interferon
26:25response and this need to induction
26:28interferon stimulated genes.
26:30And the double stranded on a could
26:32be sensed through regard MDA 5
26:34maps Altos three and basically
26:36also signals through and activate
26:38interference and steam engines.
26:40So what we did is we knock cloud
26:43each single component through
26:45this pathway and see what happens
26:48when we knock out the Canadian 5B.
26:49As you see it does not grow in the
26:52wild type cells do grow if we combine
26:55that with knockout of the mouse or
26:58steam and the important mediator of.
27:00Christian Arnie or double Strand
27:02DNA sensing pathway,
27:03you can see partial rescue right here.
27:06If you get rid of both of them
27:09you see much better rescue.
27:11So we went on and when the upstream
27:14when we get rid of the sea gas or
27:17MDA 5 and you can also see partial
27:20rescue if you get rid of both of them.
27:24You can see pretty good rescue response
27:27in this case when in two independent.
27:31So how many established that?
27:34Now we want to understand why
27:36those pathways are activated.
27:38So why would the sense that we notice
27:41is that when we compare the control
27:44cells with the knockout cells,
27:46we can see the induction of double
27:50stranded on a in Kadian 5B knockout
27:53cells and then we have seen this
27:55also in two months as well.
27:57And this motivated us to go back and
28:01to realize our only sequencing data.
28:04For expressing those retro elements
28:08and those retirement are part of junk
28:11genome and then people are totally
28:13normally ignore and it turned out
28:16to be very important in this case.
28:18And what we have seen is that
28:20we knock out Kadian 5B,
28:21we can see induction of those
28:24retro elements and especially
28:26some of the endogenous retrovirus.
28:29Animals.
28:29And the one with which is called MOV 30 and
28:34you can see multiple of those showing up.
28:37And then we study is actually
28:40critical for the interferon response
28:42because if we knock down M30 with
28:45SRAM as you can see here,
28:47you can see the down recognition or
28:51interference imagines suggesting that
28:53this is at least partially contribute
28:56to the interferon induction and maybe
28:59the response to e-mail evasion.
29:02And the one thing that we were
29:04puzzled about is that since I
29:06showed you that both DNA and only
29:08sensing password are required,
29:10where are those DNA coming from?
29:12And we postulated that those
29:14DNA will be coming from
29:16reverse transcription of those only
29:19species that that would generate
29:23through after we get rid of Kadian 5B.
29:27And one experiment we did is use
29:31reverse transcriptase inhibitor.
29:32This is a cocktail of reverse transcriptase
29:35inhibitors used for HIV treatment and
29:37where we see if you treat the cells with
29:40those reverse transcriptase inhibitor.
29:42You can see suppression of the interference
29:46imaging expression suggesting that this
29:49DNA might be created through this pathway.
29:52So now with all those mechanisms,
29:54now the question is can we
29:57translate it to targeting this?
29:58The quick question is that can
30:01we induce under tumor immune
30:03response with KDM 5 inhibitors?
30:05So as I mentioned because there's a lot
30:09of evidence showing that KDM five are
30:13critical for cancer initiation progression.
30:17So we have started working on
30:19this on to by multiple methods to
30:23develop locating file inhibitors.
30:26So initially with that panel ground from
30:29Yale Small Molecule Screening Center
30:32now called Yale Center for Monica.
30:35Discovery,
30:35we have done some screening,
30:37biochemical screening for KADIAN
30:405D methods inhibitor.
30:41And initially we did 100,000 compounds
30:44with those as a preliminary data we
30:46were able to obtain support for NCI
30:49experimental security program where we
30:51were able to assemble a team about 30
30:55scientists to to develop those inhibitors.
30:58So we have done a high school screening
31:01about 200,000 compounds those are high
31:03quality compounds and have done extensive
31:06medicinal chemistry optimization of some
31:08of the compounds and we have solved.
31:1225 uh crystal structures,
31:14can you find a way with different inhibitors
31:16and shown here just the two of them,
31:19basically showing that they combined
31:22very tightly to the active site.
31:25One thing that I want to mention that those
31:27inhibitors are all pancaking from inhibitors.
31:29They hit both all Canadian five
31:32family members because the
31:34Catholic side is very similar,
31:37very similar for all those
31:40Canadian 5A family members.
31:44So even with with those and we decided to
31:48ask what the Canadian five inhibitor can do.
31:52And the one thing that we decided to do is
31:56we selected four KDM 5 and inhibitor here.
32:00Those are high quality specific
32:02calling from inhibitor.
32:03As you can see here they all induce
32:05HK for translation which is the
32:07substrate of the reaction and then
32:10did not do anything to the other
32:12of the histone modifications.
32:14And we did those actually in I'm 6-7
32:17and multiple human breast cancer cells
32:20and and when we looked at the gene
32:23expression changes to our surprise we
32:25see the top pathway that's upregulate
32:28are those interference signaling
32:30pathway at that time I was like
32:32interfering pathway is not something
32:34I want to work on not so much now.
32:38So, so anyway,
32:39so when we see there's an induction
32:41interfering pathway and we have
32:43seen this in multiple cell lines,
32:46multiple drugs.
32:47And we were able to.
32:50Understand how this actually worked.
32:52And at the end we were able to
32:55show that KADIAN 5 inhibitor can
32:58induce H3K4 termination at the
33:00steam promoter and by doing
33:02so, it actually induce Stein expression.
33:06And this need to the interferon
33:09stimulated gene expression and
33:11listening to the T cell infiltration.
33:15So this is a little bit different from
33:18what other people have been trying to.
33:22Uh, to activate this pathway
33:24through either using Steam agonist,
33:26which the limitation of those drugs and is.
33:31Many of the cancer cells you
33:33actually have Stein silence.
33:35So by inducing Stein and this
33:37provide another mechanism how we
33:39can activate this signaling pathway.
33:42So now and we actually tested KADIAN 5
33:45inhibitor in multiple human Melanoma
33:48cells and we can see induction of
33:51sting and in this case in Western
33:54border here and the induction
33:56of interference steam engines.
33:59And so we thought this is the shoe
34:01bat and the Canadian five inhibitor
34:03is going to work.
34:04And to our surprise, nothing happened.
34:07And when we took put this in
34:09the mouseman normal cells,
34:10the Yammer 1.7 cells,
34:12the model system that we have tested.
34:142 into 2 Canadian farm inhibitor and
34:19the retro element was were not induced,
34:23the interference images were not induced,
34:26nothing happened.
34:27So we did not want to give up
34:30because we thought maybe there's
34:32some limitation of the drugs and so
34:35we did those rescue experiment to
34:37understand whether the critical the
34:39community activity is required or not.
34:41So what we did is that for.
34:44I'll call it Yama cells.
34:45We reintroduced either wild type or
34:48mutant KADIAN 5B into those cells.
34:51Those mutant are dead Canadian 5B.
34:55And as you can see here,
34:57in both cases you can see wild type
34:59or mutant Canadian 5B can suppress
35:02the expression of retro elements and
35:04those interference stimulate genes.
35:06Moreover,
35:07both of those can induce the
35:10growth of those tumors.
35:12So now what?
35:13Now we are back to the starting point
35:16and kind of depressed right at time.
35:19So we went on and decided to look
35:22at all the repressive mechanisms
35:24and to see which one might work.
35:27And one of the things that
35:28we decided to look at is,
35:29is actually inhibitor for
35:30example and those are two higher
35:32quantities that true inhibitor,
35:34it did not do much either.
35:36Umm, and then uh,
35:39it.
35:39There's some clue that HK9
35:41message transfers would work,
35:43and we use a pretty dirty
35:45actually actually canine method.
35:46Transfers inhibit the code channel thing
35:48and it can inhibit actually K9 translation.
35:52You can see induction of MOV 30 and some
35:55of the interferon stimulated genes.
35:58So now there are multiple HK9
36:01methyltransferase and so we knocked out
36:03each single one of them to see which one.
36:06Is critical when we knock out the G9A or
36:10SO39H1 and it did not really do anything.
36:12But when we knockout set
36:14B1 which is shown here,
36:16you can see robust induction on mobile 30.
36:20So this is what was a great news.
36:22So at that time we're quite excited.
36:25And then when we did call e-mail
36:28precipitation experiment,
36:29we actually can see that KADIAN
36:31file B can interact with set DB1.
36:34When we did set DB1 IP,
36:36that's the pull down of Kadian 5B by Sade 1.
36:42And then uh, this is quite exciting.
36:46Then we decided to map the binding of
36:50KDM 5B and set DB1 and shown here just
36:54the the the heat map where we ranked
36:58those KADIAN file B target genes where
37:00you can see KADIAN file B combined
37:02them very well in wild type cells,
37:05not so much in knockout cells.
37:07When we look at set DB1 binding,
37:10you can see amazingly overlapping
37:12binding of the set DB1 and the HTK 9
37:17formation which is the product of set
37:22DB1H3K9 formation is a repressible mark
37:25that can suppress gene expression.
37:28And to our surprise, when we look at
37:31the HK4 translation and imagination,
37:33which are the substrate of the Kadian 5B,
37:37you can actually do not see much effect.
37:40Suppress a suggestion that KDM 5B
37:43function add message function is
37:45probably silenced in this setting.
37:47So now those are all.
37:53Important and now we want to look at this in.
37:58Drug resistance setting and
38:00in this case e-mail checkpoint
38:02blockade resistance setting.
38:04When you look at the KTM 5 be expression,
38:06it's actually lower in the patient
38:09with computer response to anti
38:11PD1 blockade compared to the ones
38:14with the progressive disease.
38:16So this is suggesting that if
38:18we can lower expression you can
38:20make the reason tumor sensitive.
38:23Indeed that's actually true and
38:25we use this young 1.7 model.
38:28Which is the parental model for the
38:32Yammer 1.7 I have showed you before.
38:34This model is resistant to all immune
38:39checkpoint blockade, PD1 blockade.
38:41If you look at this, nothing happens.
38:43If you throw CTO four anybody
38:46on then nothing happens.
38:47If you combine them still nothing happens.
38:50In this very refractory model,
38:53you can see if you get relocating 5
38:55you can already see some response.
38:57If you combine with PD1 blockade
39:00you see synergistic response.
39:02It can extend the survival of those animals.
39:06You can basically double the
39:08survival of those animals.
39:09And this is just one of the PD1
39:11resistant model and when we look at
39:13the another model which is the Yammer
39:15interfering gamma resistant model,
39:17you can see similar phenotype.
39:21So lastly, is this also true in humans?
39:26When we compare the the KADIAN 5
39:29expression with the indulgence
39:31retro elements part of the category
39:34of the retro elements,
39:35you can see the the ones with high
39:37Acadian 5 be expression was shown.
39:40On this you have no expression
39:41of some of the.
39:44You always showing here just one example
39:48RV 2637 and it's anti correlated
39:51with Kaden 5 expression and the
39:53expression is correlated with the
39:55better response to PD1 blockade is
39:58opposite to what we see with PKD and 5B.
40:00So to basically to summarize this part
40:04of my talk which I showed you that Kadian
40:075B can interact with set DB1 and and
40:10you can recruit set DB1 to the targets.
40:13To deposit actually K9 traumatization
40:16to silence retroelements,
40:17if you gather with locating 5B you
40:19can activate endogenous retroelements.
40:21You can activate double stranded
40:23on Ascension pathway and double
40:25strand DNA sensing pathways through
40:27the reverse transcription process.
40:29It I need to the better representation
40:31of the MHC one and the cytokine secretion
40:35lead to higher immunogenicity and better
40:38response to e-mail checkpoint blockade.
40:41So although with the first group
40:43that show that Kadian 5B is critical
40:47for immune evasion,
40:48we are not the first group to so
40:50shows that B1 has this function and
40:52multiple groups about the similar
40:54time show that said B1 is involved in.
40:59Suppressing tumor immunogenicity and
41:03and and this is just multiple papers
41:06basically by multiple groups and this
41:08add to basically add to the what.
41:12Uh, what?
41:13What do we know about epigenetic
41:15regulation of the viral mimicry pathway?
41:20Basically I've showed before that double
41:23DMT and SD one can do this and here we
41:26just showed up and said one can do this
41:29and all those inhibitors will be able
41:32to induce those biometric response and
41:35the firm response and better response
41:38to e-mail checkable and blockade.
41:40So now I would like to thank all
41:42the people involved in this and
41:45especially Marcus Bosenberg group and
41:47where we had the fun collaboration.
41:49A decade on collaboration and and the
41:53drug resistant work is led by Shawnee
41:56anew and Sami Zang and the immune
41:59evasion they walked the net by Samin
42:03Jan and Samin has actually started.
42:07Isn't Professor Ship at
42:09Shanghai Tech University?
42:11And on the some of the bad formatting works
42:13are done by Western East High and the glory,
42:16and also like to thank all the
42:19youthful members for the kind of help
42:23through the course of this project.
42:27And when I try to start on Melanoma,
42:30the SPORE members welcomed me
42:33with open arms and that's how
42:36I can get where we are here.
42:39And I'd also like to thank all the other.
42:41Funding agencies for their support
42:43as you can see in a couple of
42:46Melanoma Research Foundation,
42:47Melanoma research alliance have
42:48been very helpful in supporting our
42:51research in Melanoma and I would
42:53like to thank you all for your
42:55attention and I welcome any questions.
43:09Or maybe 1 back.
43:35Yeah. That's a great.
43:36So the question is whether we
43:39have tried to combine KDM 5
43:41inhibitor with sting agonist,
43:43that's great suggestion.
43:44And we have thought about this,
43:46but we have not had the
43:48time to do this experiment.
43:50Yeah, which we should have done, yeah.
43:54OK.
43:56That was a great job. Thank you so much.
43:58And went back and forth a little bit
44:01between 10:20 and five inhibition
44:03and 25 B specific inhibition.
44:06And I know that you think the KDM 5B
44:09is the most important one. What about?
44:13And so we have, we actually have
44:18been working on breast cancer and
44:20also some other cancer types where
44:23we have seen is that in actually
44:27maybe I'll show you one slide here.
44:29This was just published basically
44:32this is MC38 with the colorectal
44:35cancer where when we treated those.
44:39So those animals, uh tumor bearing
44:41animals with KDM 5 inhibitor,
44:43you can suppress the ability to grow.
44:44So it works incorrect cancer.
44:47Also when we look at the breast cancer
44:49you can see they have some new efficacy.
44:52You can also combine that with
44:54PD1 blockade and we can have
44:56I would say additive effect.
44:57So it works in multiple cancer types.
45:02It's just where we need to find
45:05the correct cancer types and
45:07subtypes even to so that we were
45:09able to use those inhibitors.
45:13Yeah, yeah. So, yeah,
45:15those are all planning.
45:17Can you invite me here with us?
45:18So, so one of the things that we
45:21are trying to do is to develop
45:24KADIAN file family members,
45:27specific degraders.
45:28And with the.
45:32Because with the protect or some
45:34other similar kind of mechanism or
45:36molecular glue type of mechanism
45:38you can develop a specific
45:40degraders against KDM 5 and we
45:42are actually working on that.
45:44We have some potential degraders that
45:49work specifically on Canadian 5B and
45:52some of them work on multiple all
45:54kidding 5A in different settings.
45:58Online question from City Chin.
46:03Yeah, I can. I can read. I can read it.
46:05So the question is, do I anticipate?
46:09Other epigenetic reader and writer
46:13to have similar effect, yes,
46:15because actually I have showed
46:18you one in one of the diagram.
46:21There are multiple other ones on this.
46:26OK. Yeah. Once which have been shown
46:30have similar effect and although I
46:34have to say in different cancer types
46:36and they have different effect and
46:38we just need to find the right one
46:41and that work in the in our setting.
46:47OK, that's.
46:51Question for you.
46:55So you showed that Kenny M5
46:57views anticorrelated with all sorts of
47:00new vectors, both positive and negative.
47:06What about the cell types?
47:11DC.
47:14Well, that's a great question.
47:15We have not looked. Yeah.
47:18So you just need to do something
47:21also analysis too or just analyze
47:23single cell data to see to see that.
47:25Yeah, it's great, great suggestion.
47:26Yeah, should do that.
47:30I don't know.
47:33Wonderful mechanistic. Right.
47:40My question is regarding Katie M5B.
47:43And it's a deck that seemed to be asymmetric.
47:49You showed us. I was wondering whether
47:51they would be animators that could
47:53maybe the scaffolding effect of paying
47:565D its interaction with 71 and whether
47:59those could be more appropriate for
48:01who gets PDL 1 increased response.
48:07Yeah, that so answer the question is
48:10whether we should inhibit the scaffold
48:14function location 5B, which is great.
48:17So that's something that we are thinking
48:19along the way because we have to first
48:22of all we need to identify the domains
48:24that are critical for those interaction.
48:26And then one of the things that we're trying
48:29to do is you to look for those domains
48:31that are involved in interacting with
48:34said said B1 and then those inhibitors.
48:37To have more specificity as you as
48:41you suggested to target this pathway
48:44and it's probably better than getting
48:475 inhibitor or 71 inhibitor which
48:49might have some other off target
48:51effect that we don't want to see.
48:55That's what interesting,
48:56because when you think of,
48:57for example, LC-1, there are requests
48:58that seem to be targeting the.
49:04But we know that they actually
49:06impact step folding effects,
49:08other proteins that play a role in one.
49:15I wonder if those types of
49:17indicators are out there.
49:19Yeah, you could be made, but uh,
49:21we we don't have those yet.
49:23Work in progress.
49:30OK. If no more questions. Thank you.