Intergenerational Epigenetic Inheritance of Cancer

October 08, 2021

Information

October 7, 2021
Yale Pathology Grand Rounds
Bluma Lesch, MD

ID6970

To CiteDCA Citation Guide

00:05So today I'm delighted to introduce Dr.
00:09Bluma Lesch as our pathology
00:12grand rounds speaker.
00:13Many of you may know Doctor Lash
00:15is an assistant professor here
00:17in the genetics department,
00:19but what you may not know is that LUMO is
00:23actually a graduate of Yale University.
00:26Before she went on to the MD PhD
00:30program at the Tri Institutional
00:33Program at Weill Cornell.
00:35Rockefeller MSKCC,
00:37which did her pH D with Cori
00:39Bargmann and then she went on to do
00:43postdoc at the Whitehead Institute.
00:45With David Page and Richard
00:47Young as Co advisors.
00:50Her research focuses on the mechanisms
00:53of transcriptional regulation in
00:55the context of both development and
00:57evolution and in the short time since
01:00she started her lab here in 2017.
01:02She's actually been awarded the
01:04prestigious steel Cyril Scholar
01:06Award and and just received the Pew
01:09Biomedical Scholar Award in 2021.
01:12And so, without further ado,
01:15I'd like to introduce Dr Lash
01:17and her grand rounds.
01:19Talk will be an intergenerational,
01:22epigenetic, epigenetic inheritance of cancer.
01:33Alright, thank you, let me
01:35just share my screen here.
02:06Can everyone see that? Yes, OK,
02:11so thanks to the lobby for inviting me.
02:13I'm I'm really excited to tell you
02:15about what we're working on and
02:16also looking forward to helping me,
02:18hopefully making some
02:20connections in your department.
02:22So I'm going to talk about
02:24Internet intergenerational
02:25epigenetic inheritance of cancer,
02:26which is a fairly dense topic
02:29and I want to start out actually
02:32with a fairly simple video.
02:35This is a video of, UM,
02:37the first cell division and an
02:39embryo single cell zygote with two.
02:43The two pronuclei coming together to
02:46meet and undergo the first cell division
02:49and generate the two cell embryo,
02:51and I'm showing this video both
02:53because it's just really cool.
02:55This is one of the most amazing
02:57points in the life cycle and
02:59I really like watching it,
03:01but also because I want to make a point.
03:04In this video that that sort of sets up
03:06the problem that we're interested in.
03:09And the point is so here you
03:11can see the two pronuclei.
03:13Hopefully you can see that one of them
03:15is much brighter than the other one.
03:17The brighter one is the paternal
03:20pronucleus coming from the sperm,
03:23and the reason it's brighter is because it's
03:27loaded up with histones tagged with GFP,
03:31and this GFP tagged histone
03:32was coming from the oocyte.
03:34The fact that this nucleus this nucleus
03:36is so much brighter means that all of its.
03:39The stones,
03:40all of it's upper genetic protein
03:42information has been removed and completely
03:45replaced with oocyte provided histones.
03:48So this is a you know,
03:49kind of the common conventional
03:51wisdom in the field that come,
03:54that all epigenetic information coming
03:56from sperm at least is completely
03:59replaced at fertilization and reset.
04:02And and so.
04:03With that in mind.
04:08This is a question one of the questions
04:11that my lab is especially interested
04:13in two life experiences of the parents
04:15affect the health of the progeny,
04:18and one would think since life experience
04:21tends to be encoded in epigenetics,
04:24but the fact that all of this epigenetic
04:26information is reset at fertilization would
04:28mean that the answer to this question is no.
04:31But what I'm going to try and get across in
04:34this talk is that the answer is actually yes,
04:36and we're going to try and
04:38uncover what some of the.
04:39Mechanisms for that might be.
04:41And we can break down this question
04:44into two more specific questions.
04:46One is gene regulatory information
04:49inherited across generations. So is this.
04:51This is a molecular question.
04:53What's happening at the biochemical level?
04:55And then the second question is,
04:57is this inherited regulatory
04:59information biologically meaningful?
05:01In other words,
05:02does it actually affect phenotype
05:04or is it just a biochemical
05:06signature that we can follow with
05:07our various molecular techniques?
05:09Uhm?
05:11Before I get too deeply into the problem,
05:14I'm going to give you a quick
05:16introduction to epigenetics.
05:17This is probably review for most people,
05:19but I just want to make sure
05:21everybody is on the same page and
05:23what I'm showing here is a consensus
05:25definition for what epigenetics
05:27is that came out of a cold Spring
05:29Harbor meeting about 10 years ago,
05:31and it has four key components stably
05:34heritable phenotype resulting from
05:36changes in a chromosome without
05:38alterations in the DNA sequence,
05:40and I want to make the point.
05:41That come off when we talk about epigenetics.
05:44We're talking about the second half of
05:46this sentence changes their chromosome
05:48without alterations in a DNA sequence,
05:50and those are the things shown here
05:52that are generally considered to be the
05:55carriers of opportunistic information.
05:58Covalent modifications to DNA,
05:59mostly in the form of DNA methylation,
06:02positioning of nucleosomes
06:05along the chromosome.
06:07Covalent modifications to histones,
06:09which can affect positioning along
06:11the chromosome transcription factor
06:14binding and large scale structural
06:16arrangements of the chromosomes.
06:19These are the kinds of things that
06:20you see a lot of data on that we
06:22tend to study and pay attention to,
06:24but I want to point out that a key
06:27aspect of this definition is that
06:29this effects phenotype and but it's
06:32heritable so so I and I want to make
06:35sure that that we're addressing.
06:36So those questions during this talk.
06:42My love is especially interested in in one
06:45particular aspect of epidemic information,
06:48we tend to focus on this level of
06:51organization at the nucleosome.
06:52The nucleosome is the basic
06:55structural unit of chromatin,
06:57and it's composed of an
07:00octamer of eight of eight,
07:03histone 8 histone proteins,
07:06so two tetramers.
07:09And 147 base pairs of DNA
07:12wrapped around its optimer.
07:16Each of these histones has a
07:18globular domain that makes up
07:20the core of the nucleosome,
07:22but importantly also has an end
07:24terminal tail that sticks out,
07:26and it's this floppy tail that gets modified.
07:29I'm currently and so we tend to.
07:31This is the structure we tend to
07:33portray this as a cartoon like
07:35this and this is what you'll see
07:36throughout the rest of the top,
07:38but at the biochemical level,
07:41covalent modifications to these
07:43histone tails can affect the
07:45stability of this nucleosome and
07:47in turn effects the extent of gene
07:50regulation of a particular locus.
07:53Uhm? So when we talk about resetting
07:57of epigenetic information,
07:59there are,
08:00and we're focusing on on his
08:04own modifications.
08:05There are actually two aspects of this.
08:07There's one that I showed already
08:09in the in the in the early embryo
08:11where his zones are kicked out of the
08:14eternal nucleus and some extent reset
08:16in the maternal practically as well.
08:19And there's also a resetting
08:21that happens in sperm before
08:23you even get to fertilization.
08:25So sperm actually removed
08:27most of their histones.
08:28Most of their nucleosomes as they
08:30mature and replaced them with
08:32another protein and their packaging
08:34protein called crudo means,
08:36and that's so that they can
08:38compact their their their genome.
08:40Very,
08:40very small and fitted into a sperm head.
08:43As we've all seen.
08:45And so.
08:46There's a loss of epigenetic
08:48information at this stage,
08:49and there's a loss of epigenetic
08:51information at this stage.
08:53But importantly,
08:53there is a little bit of
08:56retention of histones in this
08:58during this protamine transition,
09:00and we're actually not sure the
09:02extent to which those retained
09:04histones actually passed through
09:06into the into the early embryo.
09:12So in complete counterpoint to the extent
09:15of epigenetic resetting that happens at
09:18fertilization and before fertilization,
09:21there's quite a lot of evidence indicating
09:24that phenotype can be affected by paternal
09:27experience or by parental experience.
09:31And what I'm showing here
09:32are two classic examples.
09:34These are phenotypes that are actually
09:37conveyed generally by DNA methylation.
09:40But these are papers that that in
09:42some ways kicked off this field.
09:45This is a locus called agouti viable yellow
09:48which is an allele of the agouti gene that
09:51affects coat color among other things.
09:54And these mice are all genetically
09:56identical but carry different levels
09:59of DNA methylation at this locus
10:01which affects the extent to which
10:03they they express or don't express.
10:06This brown agouti coat color.
10:09And it turns out that the maternal
10:13epigenetic state can affect the chances of
10:16the offspring having the same epigenetic
10:18state and having a similar code color.
10:21These again,
10:22this has nothing to do with being
10:24encoded in the genome itself.
10:26So this is a maternal effect just to
10:28show you that it's not only maternal is.
10:31Similar effects can be seen and
10:33another locust called Accent,
10:34and this causes a kick tail and
10:36can be transmitted either from
10:38a father or from the mother.
10:40Uhm?
10:43A whole pile of other studies
10:45has looked at this kind of thing,
10:47not just in terms of morphology
10:50and coat color, but in terms of
10:52things like metabolism and behavior.
10:54So I'm showing here a couple of the
10:56most common types of phenotypes that
10:58have been studied in this context.
11:00A fairly well established
11:02one is paternal diet,
11:04so a high fat diet in the father can
11:08result in altered insulin response and
11:10altered glucose processing and offspring.
11:12This is an example from a paper in 2015.
11:16Uhm, a number of papers have looked
11:18at drug exposures in the father,
11:20so here's an example where a father
11:23exposed to nicotine has offspring that
11:26actually are better at processing nicotine.
11:29So this is actually survival.
11:31Or if you give them toxic doses.
11:34The offspring of exposed fathers are
11:37better able to process those doses
11:39and survive compared to offspring.
11:41If not you fathers.
11:43It also affects some sort of
11:45addictive behaviors and and and
11:48responses to to drugs at that level.
11:51And then finally there are a number
11:53of ways you can stress out a mouse.
11:54Essentially they all result in
11:57increased cortisol levels and offspring
11:59of stressed out fathers tend to
12:02have altered responses to stress.
12:04Specifically,
12:05a lower a smaller increase in cortisone
12:09following a stressful event and in some
12:12cases they have behavioral changes as well.
12:15So this is all in rodents.
12:17This has been looked at
12:19to some extent in humans,
12:20although obviously it's harder to do these
12:23kinds of experiments in human subjects.
12:25The best examples are in.
12:28One example is in a cohort from Sweden
12:31from the 19th century where they
12:34had very careful records of harvest
12:37yields and health outcomes overtime,
12:41and in that study their studies from the
12:44early 2000s they were able to show that.
12:46Good nutrition in the paternal granfather
12:51resulted in increased cardiovascular
12:53events and diabetes in grandchildren.
12:57This is a follow-up study that was done
13:00more recently in a cohort of of almost
13:0230,000 people over three generations,
13:04also in Sweden,
13:06and here they actually see an effect on
13:09cancer so well nourished grandfathers,
13:11well nourished paternal grandfathers
13:14have a a correlation with a significantly
13:18increase rate of cancer in grandchildren.
13:22Uhm?
13:24So.
13:25The answer to this question do life
13:28experiences of parents affect the
13:29health of progeny appears to be yes,
13:32but.
13:32The molecular mechanism by which these
13:35these experiences are transmitted
13:38remains a pretty significant black box,
13:42and so we're trying to get at these
13:44questions of what information is
13:46transmitted and whether it's meaningful,
13:48and we've decided to do this
13:50by simplifying the system.
13:51So rather than expose then making an
13:54environmental exposure change like
13:56diet or stress or drug treatments
13:58were going to make a genetic
13:59change where we have a little bit
14:02more control over the effects of.
14:04Add change on epigenetic state and
14:07then evaluate what those effects are
14:09in offspring and try and connect that
14:12molecular chain at each point along
14:15the way. So here is our strategy.
14:18Make the make a genetic change.
14:19In my father we examine the sperm and we
14:23examine the offspring within this firm.
14:25There are three main modalities of
14:28epigenetic information that will look at.
14:30One is small.
14:31RNA is so actually this is something that
14:34we in my lab had paid less attention to.
14:36Although there are plenty of
14:37other people in the field,
14:38but I've looked at from DNA metalation
14:40so it could be like modifications to
14:44the DNA and histone modifications. Uhm?
14:47Our general strategy is very simple,
14:50so we're taking advantage of the fact
14:53that males have only One X chromosome
14:55and that the X chromosome encodes
14:58a variety of epigenetic regulators.
15:01So if we delete and epigenetic
15:03regulator on the X chromosome in a male,
15:06we make a complete knockout and we
15:09knock that we crossed that mail
15:11to a genetically wildtype female.
15:14If you remember your Punnett squares
15:16from high school biology, all male.
15:18Offspring of that cross are going to
15:21get a normal X chromosome from their
15:23mother and a normal Y chromosome
15:25from their father.
15:26Even though the father was a
15:28complete knockout for this scene.
15:30Which means we can evaluate the
15:32phenotype of these F1 males and if
15:35we see a phenotype that implies that
15:37it's a result of epigenetic changes
15:40in the paternal germline rather than
15:42genetic changes that are inherited.
15:47As a genetic target,
15:48we picked the gene called JTX.
15:50It's also called KDM 6A,
15:52and you may see that name pop up later
15:55on in the slides and we picked this
15:58gene first because it's X linked,
15:59so it meets our first criterion.
16:02It's also a well established
16:04chromatin regulator,
16:05and we know a specific enzymatic
16:08activity that this gene has it.
16:11See methylates metalation
16:12at Lysine 27 of histone H3.
16:16This is a well studied histone modification
16:18and a well established enzymatic activity.
16:22For this protein.
16:23We also know that it has non non
16:26enzymatic effects on chromatin so
16:29it regulates enhancer assembly by
16:32by interacting with a complex.
16:34That place is a different
16:36chromatin modification.
16:37These are both relatively well
16:39defined functions for this protein
16:42and gives us some hope of being
16:44able to trace this molecular chain.
16:46From sperm to the next generation.
16:50We also know that UTX is
16:52biologically significant.
16:53It's a developmental regulator.
16:54It's been implicated in cardiac
16:57development and in about police this and
16:59we know that it's a tumor suppressor,
17:01which I'll come back through
17:03a little bit later. Uhm?
17:05So, uhm, here's how we set this up.
17:09For those of you who are interested
17:11in the details,
17:12we set this up as a conditional knockout.
17:14So the TX is deleted only in the germline,
17:17not in the somatic tissue of the parents,
17:19which reduces the possibility of
17:22having secondary or indirect effects.
17:24And we're using a creed that that is
17:27basically complete to the excision by birth,
17:30so U TX is lost throughout surmounted
17:34Genesis from starting from.
17:35Very early in the lifetime of the animal.
17:38Uhm?
17:40We can confirm that this knockout it's
17:43complete using Western blot for UT X.
17:45There's a tiny bit of a residual band here,
17:47and that's because this is done
17:49on whole testis,
17:49which has some somatic tissue which
17:52therefore still expresses UTX.
17:55Uh, the first thing to note is that,
17:58UM,
17:59that blossom PTX has essentially
18:01no effect on fertility,
18:04so I'm showing here a cross section of
18:07the seminiferous tubule in a control,
18:09and the UTX conditional knockout,
18:12which will be shown as Cao and
18:14the tubules are pretty identical.
18:17They contain all the different
18:19fault mental cell types in this
18:22developing seminiferous epithelium.
18:24Sperm look normal and the numbers of
18:26sperm are about the same and these
18:29minds can produce pops at the same levels.
18:31So this is.
18:32This is whatever we see is not due
18:35to some sort of secondary effect
18:37of just screwing up fertility.
18:41Uhm,
18:41so then we start to look at
18:43offspring and the first thing that
18:45we noticed was that offspring of
18:47these mice have a reduced lifespan.
18:49They still live a good long time,
18:52so they make it out to about a year.
18:54But then they start dying and then
18:56another cohort of them lives a
18:57little bit longer and then dies
18:59out in about a year and a half.
19:01Uhm? And we did this.
19:03Our initial cohort was in a
19:05mixed genetic background.
19:06Just to you know, make sure this was real.
19:09We also did this in an inbred
19:12background and can confirm same effect.
19:14Incidentally, and sort of fascinating to me.
19:17The shape of the curve is actually the same,
19:19so there's a drop off around the year and
19:22another set of deaths a little bit later.
19:24And then of course, what we want to
19:26do is look into what's going on.
19:28What's causing these deaths,
19:29and to do that we took each
19:31mouse that died as they died and
19:33did it complete necropsies,
19:35sampling each of about 12 tissues.
19:40And I'll tell you right away,
19:41we actually have no idea what's
19:43causing this first set of deaths.
19:45These mice drop dead overnight and
19:48don't appear to have any histologic
19:52differences relative to controls.
19:55But in the second set of dots,
19:57we found something interesting,
19:58so this is just raw data showing
20:01counts by age,
20:02and obviously the controls guide
20:04to the controls also have various
20:06issues because they're just old mice.
20:09But most strikingly,
20:10and there was a substantial increase in
20:13the number of tumors we found in offspring
20:16of JTX knockouts compared to controls.
20:19And this is tumors found
20:20across multiple tissues.
20:21The most common that we saw in
20:24histiocytic sarcoma, and then,
20:25which is common in mice.
20:27But we saw many more in EU TX
20:30offspring compared to controls.
20:31And then the next most common
20:34are cellular carcinoma and lung.
20:35Various lung tumors,
20:36which I will also come back to
20:38a bit at the end.
20:41Uhm? So that was very interesting
20:44and we also wanted to know what
20:46happens across multiple generations.
20:47So we did this follow-up experiments
20:50where I'm not showing details here,
20:52but basically we set it up so that we
20:55could look at F2 mice that were the
20:58product of two successive generations
21:00of loss of BTX in the germline.
21:02So in other words,
21:03the germ cells never have a chance
21:06to reset their epigenetic information
21:08because you taxes is continually
21:10lost across these two generations.
21:16When we look at these F2 mice,
21:18we see a similar reduction in lifespan.
21:22Uhm, and and the similar sort
21:25of shape overall of the curve.
21:27Although this first drop off is his absence.
21:31Interesting like.
21:33And again we saw a significant
21:35increase in the in the numbers
21:37of tumors that came up in mice.
21:40And interestingly,
21:41although the overall tumor rate was about the
21:45same between in the F1 and F2 mouse cohorts,
21:48there was a a noticeable difference
21:51in the severity of the tumors.
21:54So we see a lot more aggressive tumors
21:56and F and F2 minutes compared to F1 mice.
21:59They tend to be in the same tissue.
22:01They look a lot worse, and relatedly,
22:04each individual mouse has more tumors.
22:08And, and here we're counting
22:10independent tumors,
22:10either within the same teacher
22:12or in multiple tissues.
22:13It was not uncommon to see mice with with.
22:18323 independent tumors and
22:20occasionally four or five.
22:21And so something is.
22:23This implies that something
22:25is accumulating epigenetically
22:27across generations in the
22:29absence of this chromatin
22:31regulator in the male germline.
22:36Uhm, so here. So we have a phenotype.
22:38So this now we're at the level of the
22:40studies that I showed at the beginning.
22:42We see a phenomenon of a change in in
22:45phenotype in the offspring of males that
22:47have undergone a certain perturbation,
22:50and we can conclude that perturbing
22:52epigenetic state in the male germ cells
22:54reduces lifespan and increases tumor rate
22:56in the genetically wildtype offspring.
22:58And now we want to go back
22:59to the original purpose,
23:00which is to understand what
23:02genetic regulatory improvement
23:03with gene regulatory information.
23:05In other words, what epigenetic.
23:06Information is actually inherited.
23:11Uhm, and I'll just remind you that we
23:13know that UT X is a tumor suppressor,
23:15so when UTX is actually lost
23:18in the tissue itself,
23:19it promotes tumor formation and
23:22variety of different tumors.
23:25And just in case you're wondering
23:26how much people care about this,
23:28this is a paper that came out
23:30in nature a month or two ago,
23:32again showing that UT X axis tumor
23:36suppressor and attributing this
23:39to formation of a phase separated.
23:42Regions in the nucleus,
23:44based on an intrinsically disordered
23:46domain at the center of the TX protein.
23:52Alright, so now we wanna know what
23:54gene regulatory information is
23:56inherited from the knockout fathers.
23:57And can we link this to what's
24:00known about JTX molecular function
24:01as a as a tumor suppressor?
24:04To do that, we're going to look again,
24:06not so much at noncoding RNA's, but at.
24:10DNA methylations and histone modifications.
24:13And we started out by by thinking about
24:15histone modifications because as I mentioned,
24:17we know that UTX has enzymatic activity
24:21as a demethylase of this modification.
24:24H3K27 travel.
24:27So the first thing we did was look
24:29at this modification and we performed
24:31the chip seek experiments in sperm.
24:33And now we expect that use because
24:36you TX is a demethylase.
24:38It removes this modification when we
24:41delete JTX we should see an increase in
24:44K27 trimethyl signal and that is in fact
24:47what we see across two replicates in sperm.
24:51A moderate increase globally
24:53in K27 Trimethylation.
24:57So we thought create.
24:57This is exactly what we predict.
24:59Now we all have all we have to do
25:01is figure out exactly what genes are
25:04most affected by this change in K27
25:07TRIMETHYLATION and we'll have our answer.
25:10So we started looking at individual
25:12genes and it turned out as usually
25:14happens that the story was a little
25:16bit more complicated than I thought.
25:18So here is an example of our chip seek data.
25:21This is shown as signal tracks
25:23on the genome browser and so here
25:25gene models at the bottom.
25:27And if you look at the gene promoters
25:30and Gray is controlled data and in in
25:33oranges data from EU T X knockout sperm.
25:36And you look at the promoter regions.
25:38You actually see a drop in signal in
25:41the knockout relative to the control.
25:43So this is the opposite direction
25:45of what we had expected.
25:47And it turns out,
25:49if we look across the entire genome.
25:51What's happening is that in the knockout,
25:54the signal is dropping right at the
25:56promoter with where normally the
25:58signal is highest and it's increasing
26:00in these intergenic regions,
26:02so it's essentially redistributing
26:05across the genome and flattening out.
26:08We could interpret that as
26:11randomization or overall remodeling,
26:13but the punchline for us was that this
26:15doesn't actually allow us to identify
26:18specific genes that are being MIS
26:20regulated and contributing to the phenotype.
26:22It just tells us that something is
26:24wrong with K27 trimethylation sperm.
26:26So we decided to take an alternate
26:29tack and we know that.
26:33That K 27 Trimethylation is is
26:38antagonistic with DNA methylation.
26:41Uhm,
26:41so this is they actually have
26:43a very complex relationship,
26:44but in general we see that this
26:47modification and DNA match DNA
26:49metalation are anticorrelated genome.
26:52So we decided to look at what
26:54happens to DNA methylation in the
26:57sperm of these knockout mice.
27:00When we did that,
27:01we found actually a very striking effect.
27:04So it's only a relatively small
27:06subset of regions that exhibit
27:09changes in DNA methylation.
27:11If you look globally across the genome,
27:13there's no global change,
27:15but among these regions that exhibit
27:18changes. The vast majority of them
27:20have increases in DNA metalation,
27:22so there's some bias or it's
27:24game of DNA methylation.
27:25In the context of this disruption is
27:29overall disruption in K27 TRIMETHYLATION.
27:32Uhm? Interestingly, if we then look
27:35in somatic tissue of the offspring,
27:38we see the same effect.
27:40So again, there's an overall game
27:43at DNA metalation at a subset of
27:46LOCI and a clear bias towards gain,
27:48as opposed to lots of DNA metalation in this.
27:52In this case, bone marrow of the offspring.
27:56Alright, so now the question in your
27:59mind should be how many of these sites
28:02that game DNA metalation in offspring
28:04were also gaining DNA methylation
28:06in the sperm of the knockout.
28:09And the answer is that not all of them,
28:13but many of them are are hypermethylated
28:15in both the sperm and the bone
28:18marrow of offspring,
28:19which implies that this DNA metalation
28:22change may be persisting across
28:24generations and helping to alter
28:26gene expression in somatic tissue.
28:29This turns out to be about 200 low side.
28:32Uhm?
28:34And we can convince is an example of just
28:37showing the signal at one particular look.
28:40It's almost two,
28:41and then we can actually validate it
28:43using a different assay at the same locus.
28:48Uhm,
28:48so now we wanna know.
28:50OK this seems like a a important set
28:53of targets that may be contributing
28:56to our phenotype.
28:57We wanna know what's special
28:59about this set of 200 regions.
29:01Turns out there's not really much bias
29:03in terms of where they are in the genome.
29:05There's a bit of a cluster
29:06here on chromosome 5,
29:07but generally they're scattered
29:09across the genome,
29:10and I'll summarize a whole bunch
29:12of analysis by telling you that
29:14they're not enriched.
29:14It repeats,
29:15they're not enrich step imprinted regions.
29:18They're not particularly enriched
29:19near promoters.
29:20With jeans and they're not
29:22enriched at CPG islands,
29:23which is another dumb class that we might
29:27expect to see biases and DNA methylation.
29:30Uhm?
29:32What they are enriched for is
29:35changes in K27 Trimethylation,
29:37which is where we started.
29:40So the regions that have the biggest
29:42perturbation in K27 tend to also have
29:45the greatest gains in DNA methylation,
29:49so there may be a functional link there.
29:52They're also enriched and enhancer regions,
29:55and this is marked by a different
29:58histone modification that's
29:59generally associated with enhancers.
30:01We see that these persistence
30:04hypermethylated regions tend to occur
30:07in regions that have enhancer activity.
30:10And that was nice,
30:12because 'cause enhancers we
30:13can potentially linked to genes
30:15and linked to functions.
30:17So we did that using an established
30:19software for making those kinds of
30:22associations and what I'm showing
30:24here is actually all of the enriched
30:28phenotypes associated with with the
30:31the persistent hypermethylated region.
30:34So there's a clear link.
30:35There's a clear functional link here
30:38to tumorigenesis as sort of evidence.
30:41By past phenotypes that have been
30:44associated with these enhancer regions.
30:49To get a little bit more at how
30:51this might actually be occurring,
30:53we took a closer look at the sequences
30:56associated with these 200 or the
30:58sequences of these 200 regions,
30:59and we looked for enriched motifs
31:02and we were able to find 3 enriched
31:05DNA motifs in these sequences.
31:07All of which contain CPG sites,
31:09so these are the sites that
31:11game DNA metalation,
31:12so these are probably where the
31:15DNA replication is seeking place.
31:17And interestingly,
31:18two of these three have been
31:20tested to see if the presence or
31:23absence of DNA methylation effects
31:26binding of a transcription factor.
31:28So this is a site that is usually
31:31bound by transcription factor
31:32called elk one in the presence of
31:35DNA methylation elk when binding.
31:37Is significantly inhibited
31:39and same thing here.
31:41This is normally bound by GPA and the
31:44presence of DNA metalation there's a
31:46significant inhibition of binding,
31:48which implies that these regions
31:51that gained DNA metalation and sperm.
31:54And then retain that DNA
31:56metalation in offspring.
31:58There may be an effect on transcription
32:00factor activity and G and downstream
32:03gene expression that may in turn
32:05affect the phenotype and maybe
32:07tumorigenesis in those tissues.
32:09Interestingly,
32:09at around the same time that
32:11that we did this study,
32:13another paper came out looking at you.
32:15T X function in magnetic tissue.
32:18And and they found almost exactly
32:20the same enriched set of motifs among
32:23a set of genes where they found
32:25when they knockout EX they see a
32:27miss regulation of gene expression
32:29and they see they see that UTX
32:33binds directly to these promoters.
32:36So here you see out for a while
32:38and GPA coming right out of this
32:41analysis as well.
32:42So this implies that UTX may
32:45actually directly bind these sites
32:47in developing sperm altered DNA
32:50methylation at those sites, the DNA.
32:53The DNA metalation is retained in
32:56matter poetic tissue of offspring.
32:59And effects from binding and
33:01downstream expression based on these.
33:04These classes of transcription factors.
33:08One example of this effect
33:10is that the ranks 2 gene,
33:13so this is one of the hypermethylated
33:15DMR is that we identify it within
33:18that DMR we see an elk one
33:20binding site within that elk.
33:22When binding site is a CPG site and
33:24that site has more population in
33:28knockout sperm compared to controls and
33:31more methylation in healthy bone marrow.
33:35And knockout an offspring of
33:37knockouts compared to controls.
33:39Since is any absence of any disease,
33:41phenotype, and implies that they
33:43made that this may predispose these
33:46healthy offspring to generating.
33:48Blood tumors this is a analysis
33:51of of ranks to target.
33:53Genes are now looking downstream in
33:56the in the gene regulatory network.
33:59This set of genes sort of as significantly
34:02perturbed and gene expression
34:04space in the presence of disease.
34:06But even before we see disease,
34:08there's a shift in the expression
34:10of these downstream genes towards
34:12a more disease like state,
34:14again implying that may be predisposed
34:17by these epigenetic changes induced by
34:20by UTX last in the previous generation.
34:28Alright, so from that we can conclude
34:31that the sperm of beauty X knockout males
34:35carries changes in histone methylation
34:37K27 trimethylation and DNA methylation,
34:40some of which persists offspring.
34:43And it leaves open two questions that
34:46we're sort of in the process of pursuing.
34:49One is, how are these changes
34:50established in the developing germ line?
34:52So we we what we've added so
34:54far is mature sperm.
34:55But we know that UTX is lost at the
34:57beginning of spring out of Genesis.
34:59So what are those initial changes
35:01in epigenetic state?
35:02That's set up views?
35:04DNA methylation differences
35:06that appear to persist?
35:09A second question is how are they maintained
35:11in developing tissues of offspring?
35:13So again we have a time point of sperm.
35:16And then we have a time
35:17point of adult offspring.
35:18Actually pretty old adult offspring
35:20that are starting to develop tumors.
35:22What happens during all the time in between?
35:25How do we maintain these epigenetic marks?
35:28Even in the context of of rapid changes in
35:32developmental gene expression and so on?
35:35And finally,
35:36how do they actually predispose
35:38to tumorigenesis?
35:38I showed you one example in the
35:41in the alteration of ETS binding
35:44transcription factor binding,
35:46but we'd like to identify sort of
35:49some more global rules about how
35:51this effect may be taking place.
35:54So what I'm going to show you is
35:56just a little bit of preliminary data
35:57to try and address these questions.
35:59These are things that were very much in
36:02the process of trying to understand still.
36:05So First off we looked at where you
36:07at when you text is expressed during
36:10strategy MC developments and what I'm
36:13showing here is single cell RNA seek
36:15data that's been put onto a pseudo time axis.
36:19So these different colors are
36:22progressive stages of schematic
36:23demik development starting here
36:25in red or starting over here,
36:28and the sporadic janitors progressing
36:30through meiosis and through
36:32the final stages of sperm.
36:35Developments and intimate tours firm.
36:39DMC1 is a marker for a specific
36:41cell type and a specific stage.
36:43This is the very beginning of the
36:46entry into meiosis when chromosomes
36:48are actually inducing double strand
36:50breaks and starting recombination.
36:52So this is a critical time
36:54during spermatic Genesis,
36:55and it has to be very highly regulated,
36:57and you can see that U TX is most strongly
37:00expressed right exactly at this time.
37:02So it's a very specific time in Toronto,
37:04Genesis, and implies that it may
37:06have for very specific developments.
37:08Functions, and there's also
37:10a bit of expression,
37:11and the earlier strategem excels and
37:14a little bit in the somatic cells.
37:17And we can confirm this.
37:18Uhm, this is using single molecule RNA fish,
37:22so each of the green points here is a
37:26single molecule of RNA ATX RNA in blue.
37:30Is DAPI showing the chromatin and
37:33you can see that there's JTX RNA
37:36molecules in these cells that are at
37:38the edge of the seminiferous tubule.
37:41These are the progenitor and early meiotic
37:43cells that I referred to just a minute ago.
37:47And but expression persists
37:48actually a little bit into later
37:51stages from out of Genesis,
37:52but is definitely highest in
37:54these early progenitor cells,
37:56confirming what we saw from
37:58the single cell RNA seed.
38:01In addition, I told you at the very
38:04beginning of this talk that some lots of VTX
38:08doesn't affect fertility, which is true.
38:10These mice are completely fertile.
38:12There's no change in sperm count,
38:14but we do see a little bit of
38:15an effect on spermatogenesis.
38:17This is an extremely mild change,
38:20but we see a statistically significant
38:23increase in the number of spermatogonia,
38:26so these early stragetic genitor cells.
38:29As I said, either by market,
38:31by this solid four marker or
38:34by sort of counting by Ajani.
38:38And that implies that although
38:41nothing is affecting fertility,
38:43something is off in these cells.
38:45Something has been changed in the gene
38:48regulatory network that's causing them
38:50to not progress completely smoothly,
38:52so through their normal development.
38:57And we also looked at gene expression in in
39:02testis and whole testis and consistent with
39:05the fact that we see very that we see very
39:09little change in phenotype in the knockout.
39:11And in the fact that you TX is expressed
39:14only in a very small subset of cells,
39:17we don't see massive changes at
39:18the level of the whole test.
39:20So that's what we expect.
39:21But we do see a couple of interesting
39:24changes, and one of them is KLF 10.
39:26So Cliften is a transcription factor
39:28that's known to be a direct targeted VTX,
39:32significantly downregulated in JTX knockouts,
39:36and we know that it acts in a
39:38variety of developmental contacts,
39:40often along with TGF beta.
39:42And so this is a good candidate for, UM.
39:45For it, sort of for understanding how
39:48transcriptional networks may be perturbed,
39:51starting from the very beginning of
39:52loss of BTX turns for how to Genesis,
39:55and we're actually waiting
39:56for single cell RNA.
39:58Seek data to come back.
40:00That will hopefully allow us to look at
40:03gene expression changes specifically in
40:04the subset of cells that do express JTX,
40:07which hopefully will be.
40:09A little bit more expensive.
40:13And then finally, uhm,
40:14I getting into sort of to the other end of
40:18the spectrum trying to understand what's
40:20actually happening in these offspring
40:23tissues to promote tumorigenesis.
40:25We've actually shifted our
40:27attention recently.
40:28We have been focusing on Magic Poesis based
40:31on the histiocytic sarcoma phenotype,
40:34but have started focusing
40:35a little bit more on lung.
40:37And there are a couple of reasons for that.
40:38One is that slang is a little bit
40:41easier to collect than bone marrow,
40:43but also that we saw this really
40:47interesting effect in the F1 and F2's,
40:50which is just actually there
40:52are two parts of it.
40:53One is that.
40:56Long is the only tumor that we never
40:58saw spontaneously occur in controls,
41:00and we see it only in the
41:02offspring of JTX knockouts,
41:03so it seems to be fairly specific.
41:06And two is that it has this.
41:08It had a very strong,
41:09effective anticipation where the
41:11tumors in the F2's were significantly
41:14worse than the tumors in the F1F ones
41:17tended to be small self contained data
41:19nomas and the F2 is often they would
41:21take over the entire chest cavity.
41:23So it seems like there's something
41:25especially sensitive in the in
41:28the lung that may be picking up
41:30on these changes that do not get
41:32reset from generation to generation.
41:34In keeping with that,
41:36we see a substantial set of genes that
41:39are miss regulated in both F1 and F2 long.
41:42So this is these are completely
41:45genetically wildtype mice and
41:47completely histologically normal lung,
41:51so this is pre any disease and
41:53yet they share about 200 genes
41:55that are commonly MIS regulated,
41:57implying that there's some sort of
41:59common process that's going on there,
42:01and these genes are enriched for a
42:03variety of biological processes.
42:05And just as an example,
42:07the most enriched processes relate
42:10to protein targeting to the membrane,
42:12and these are some of the genes
42:15included in that category.
42:17And I'm showing this where each
42:19column is an individual mouse.
42:22To emphasize that these effects are
42:24very consistent from individual
42:26to individual.
42:27Again,
42:28despite the fact that they're
42:30genetically identical and there is
42:32no apparent disease in these tissues.
42:36And finally, in trying to understand again
42:39what the links are between these sets of
42:42miss regulated genes and tumorigenesis,
42:45we've also noticed that the sets of genes
42:48that are commonly miss regulated in F1 and
42:51F2 lungs are enriched for being Mick targets.
42:57To make, of course is a is a.
43:00Extremely common oncogene,
43:02and it's been implicated previously in
43:05regulating abrass dependent lung cancer,
43:08so this may be sort of an interesting
43:12connection to what's actually
43:13happening to initiate these tumors.
43:15To predispose these lung tissues to tumors.
43:20Uh. So so in conclusion,
43:23basically we've found that perturbing
43:25epigenetic state in male germ cells
43:28reduces the lifespan and increases
43:30tumor rate and wild type offspring,
43:32and that's perma VTX knockout.
43:34Males carries changes in histone
43:37modification K27 TRIMETHYLATION,
43:39and in DNA METALATION,
43:41and some of those changes
43:43persist into offspring.
43:45And to me,
43:45this implies that a common set of
43:47regulatory networks may be sensitive
43:49to epigenetic perturbation in the
43:51germline and in tumorigenesis.
43:53In other words, there are some jeans,
43:57some promoters,
43:58some regulatory regions that are
44:01especially sensitive to any kind of
44:05epigenetic change be at loss of BTX
44:08or potentially an environmental.
44:11Influence or or toxin and that.
44:15The germ line is particularly prone
44:18to misregulation of those regions,
44:21and those regions are especially.
44:26Have an especially strong tendency to
44:29contribute to initiation of pyrogenesis,
44:33so there's a correlation that may
44:35actually be causation across generations,
44:37but at least is an interesting
44:39set of networks to explore.
44:41And finally,
44:42this sort of fascinating idea that although
44:46you TX is not required for fertility,
44:49it may.
44:49It may have a role in the germline,
44:51and tuning heritable epigenetic information.
44:54So in other words,
44:55the Organism has an interest in
44:57controlling what epigenetic information
44:59is passed across generations,
45:02and that U TX may be one of the tools
45:05that it uses to control that information.
45:08So finally I just want to acknowledge a
45:11bunch of people who have contributed to this,
45:14uhm,
45:15Ben Walters is a postdoc who's done most
45:17of the work on this project in my lab,
45:20Shannon created the fertilization video
45:23that I showed and Allison is underground.
45:26He's been helping Ben out how Ming
45:28is a student who has done some
45:30really cool work on this project
45:32that I actually did not show here,
45:34and I want to make sure to acknowledge
45:36this is a project that I initiated.
45:37As a postdoc at Whitehead and David Page,
45:40my mentor,
45:41there was extremely generous and
45:43letting me start this very risky and
45:47very expensive project without much
45:49promise of of where it was going to go,
45:52and likewise hope funds for Cancer
45:55Research initially funded it.
45:57Similarly had a very risky stage.
46:00Rod Brunson is a veterinary Paul
46:03tough Ologist who did the the looked
46:06at actually every single necropsy
46:08slide for this project she's on and
46:11Ben are hematologists who helped
46:13with that phenotyping.
46:15Elizabeth Morgan as apologists at
46:17Brigham Hill helps validate some
46:19of our results and grant gave us the
46:22TX mice and Tyler provided advice for
46:25the project and thank you again for
46:28the invitation and for listening.
46:30And I'm looking forward to.
46:31Answering any questions.
46:37Thank you so much Bluma.
46:39Uhm that was it was really fantastic.
46:42I think a lot of us think about changes
46:45to methylation marks a lot in in
46:48terms of you know, different tumors,
46:51but we don't really get to think a lot about.
46:55The mechanisms underlying what sort
46:57of what those changes are sort of.
46:59What is the origin of those changes
47:01and and and sort of all the different
47:03contexts that you mentioned here,
47:04and the germ line.
47:06So thank you so much.
47:08There are actually a couple of questions
47:11already, so doctor Sklar asked.
47:13Where the tumor types heritable
47:16between the F1 and F2 generations.
47:21Uhm, meaning oh, so if you have a
47:23specific if you have an F1 with a
47:25particular tumor, do the offspring
47:27of that F1 get the same tumor so?
47:32Uh, yeah, I think that.
47:33I think that's the question.
47:35OK yeah, that's a great question and
47:37one that we do not have data to answer
47:40because the we never actually bred those.
47:43Come the F1 that we phenotypes we
47:45never bread and F ones that we bred.
47:47We never phenotype so we were never
47:50able to make that specific connection
47:52from the F1 to F2 generation.
47:54It's only at the population level but yeah,
47:56that's a great question.
47:59OK and then come from Katie
48:02Politi bloomer really nice talk.
48:04Have you looked at the lung tumors to
48:07see whether they have K Ras mutations?
48:09Mice can develop them spontaneously
48:11and as you mentioned K.
48:13Ruskin cooperate with MIC.
48:15Yeah also a great question. We have not.
48:18We haven't looked at any genetics
48:20yet in these mice so we haven't
48:24looked to see what kind of underlying
48:26spontaneous mutations there may be.
48:29We're in the process of setting up
48:31to try and do some XM sequencing
48:34from the slides that we have.
48:36It's sort of variable in quality,
48:38but but we definitely want to look
48:40at that, yeah?
48:44OK, and then there's also a question in
48:48the Q&A box from Chen and many of these
48:52top dogs in F1 and F2 lungs are RP.
48:56LRPS jeans. Do you see any changes
48:59of translation translational changes?
49:03We have not looked at it.
49:04Yeah, we we definitely want to,
49:06but we haven't collected that data yet.
49:08This is all pretty new.
49:09All that like data, yeah?
49:12OK, so there's lots of there's
49:14a lot of interest, so Krishna,
49:16who's one of our residents asked,
49:18does such findings have potential
49:20impact on epigenetic paternal
49:22screening and or genetic counseling
49:25in couples planning to conceive
49:27or with male infertility issues?
49:30Uhm, that's a great question I.
49:34So yes and no.
49:35I think on a practical level it
49:37would be very difficult to do any
49:40epigenetic screening at this point.
49:42I think it that's some future goal
49:44might be to identify a set in humans,
49:47a set of sort of high risk
49:49epigenetic loci that we could do
49:52some sort of array on or something.
49:54And then we could look at things like that.
49:56Or right now we just don't have any idea
49:58we can't like whole genome sequence,
50:00every infertile couple or whole genome
50:02bisulfite sequence every infertile.
50:04Couple uhm.
50:05But I actually think like looking
50:07for germline mutations of epigenetic
50:10regulators is a potentially really
50:13fruitful thing that people don't do
50:16right now because I think what one
50:18thing that our study has shown is that
50:21increase rates of mutation of these
50:24regulators can affect phenotype even if
50:27the mutation itself is not inherited.
50:30So in other words,
50:31if there are,
50:32if you're if the father is producing
50:34sperm that carry these mutations.
50:36Even if the sperm that creates the
50:38embryo doesn't carry the mutation,
50:40it may actually be relevant information
50:42for understanding the health of the embryo.
50:44So and that's something that we
50:46may be might be able to do at
50:47kind of a screening level.
50:51And finally, Doctor Prasad asked, can these
50:54heritable epigenetic changes be reverted?
50:58Uhm? So far, no, at least
51:01not a controllable way.
51:06They are in a very exploratory
51:09set of experiments.
51:11It would be cool to try using sort
51:14of epigenetic drugs and seeing
51:15what that would affect that has,
51:17although that's obviously not locus specific,
51:20so that may have a lot of secondary
51:23changes that we're not expecting.
51:25My lab has thought about using
51:26CRISPR type approaches to do
51:28this in a locus specific way,
51:29but that's obviously not.
51:32Sort of applicable then it
51:35potentially clinically applicable
51:36and is a experimentally challenging,
51:39so I haven't managed to actually do it yet,
51:42so that's a great question as well.
51:44We don't yet know the answer to
51:46bloom. I thought I really enjoyed your
51:49talk and I thought it's fascinating that
51:52you have mice with with different code
51:56colors and that was epic genetically
52:01determined. And and which brings me to.
52:05Is that possible that the skin color in
52:07humans is also epigenetically determined?
52:11Because clearly there is a correlation to
52:14environment and sun exposure and different
52:17regions in the world and the skin color.
52:22Yeah yeah. So. So on one level.
52:26Certainly there's an epigenetic
52:27component because it's responsive,
52:29as you said, responsive to environment.
52:31So if you if you spend a lot of time
52:34in the sun then generally your skin
52:37gets darker and that's epigenetic.
52:39It's there's there's.
52:41It's not necessarily gene regulatory based,
52:44but then the second question might
52:46be to wet expenses that heritable
52:49and that we don't know.
52:50I think in the in the case of mice.
52:52We had a specific locus that
52:54we knew is responsible for, UM,
52:57for coat color and and and then
52:59a specific regulatory change.
53:02In that case,
53:03there was a transposon that was inserted
53:05in the locus so we could try or the
53:07people who did the study could track,
53:09you know, the extent to which
53:11that transposon was methylated.
53:12Basically,
53:12in humans I don't know if we know
53:16of alleles that could do that.
53:19Sort of an analogous situation
53:21so we don't have.
53:22Any specific data sort of
53:24demonstrating that it could be true,
53:26although we don't have anything
53:27showing that it can't be true either.
53:36K I think I don't
53:39have any other questions.
53:41I'd really just like to thank you again.
53:44Bluma, it's really fantastic talk and
53:47I just like look forward to you know,
53:52helping you connect with anyone in our
53:55department and it'll be really great too.
53:59You know work work together
54:01in in whatever capacity?
54:04Yeah, no thank you again,
54:05it was great to to be here.
54:09Hey, uhm. I guess we can't come.
54:15Uh, I guess that that's sort of
54:17it so we can come log off there.
54:20I've there aren't any other questions.
54:22Thank you again.
54:24Thanks everyone for attending.