Gene expression regulation by RNA modifications in cancer

February 14, 2023

Information

Yale Cancer Center Grand Rounds | February 14, 2023

Presented by: Dr. Sigrid Nachtergaele

ID9486

To CiteDCA Citation Guide

00:00First of all, I'd like to thank you all
00:02for attending our grand rounds today.
00:05It is my pleasure to present this special
00:09event, which is sponsored by the Chenevert
00:14family brain Tumor Center at Yale.
00:17And today we have the pleasure to
00:21welcome Doctor Sigrid Natural.
00:23And. She isn't an assistant
00:26professor of molecular,
00:27cellular and developmental biology.
00:30She received her PhD from Stanford
00:33University and during her graduate work
00:35in the laboratory of Rajat Rochetti,
00:38she developed novel chemical tools
00:40to study the Hedgehog signaling,
00:41uncovering new modes of steroid
00:43mediated regulation of this
00:45critical developmental pathway.
00:47She was a postdoc fellow which
00:48run hay at University of Chicago,
00:50where she set out to identify
00:53novel chemical modifications in M
00:55RNA and uncovered their functions.
00:57Her lifestyle is working on numerous
00:59aspects of RNA mediated regulation
01:01of cell signaling.
01:02And their current focus is on
01:04uncovering the regulation and function
01:06of chemical modifications on RNA.
01:08She was awarded a demo reunion postdoc
01:11Fellowship and subsequently a demo reunion.
01:14They fray award for breakthrough
01:16scientists for this work.
01:18And more recently,
01:19she was awarded the Distinguished Scientist
01:23Award from the Sontag Foundation in
01:28in to reward her work in brain tumors.
01:32So please welcome.
01:41Thank you so much for having me.
01:45Look, I need a phone book
01:46behind this podium. One second.
01:50Get rid of this.
01:53All right. Thank you so much for having me.
01:55And I'm really, really thrilled to
01:58be here particularly because of,
01:59I think part of the reason this came
02:02about is that because of the santag
02:04word application that I was submitting,
02:06I guess a couple of years ago now,
02:08Doctor Romero kindly not only
02:09answered my e-mail, but even agreed
02:11to chat over zoom for for a bit,
02:13which you know was incredibly generous
02:15as I prepared for for that application.
02:17And So what I'm excited to share
02:19with you today is a little bit about
02:20what our lab is doing with respect
02:22to gene expression regulation.
02:23By Arna modifications and I'm going to
02:26try to use the cursor on the screen so
02:29that the folks on zoom can also follow along.
02:33Now to start with,
02:34I don't really have any disclosures,
02:36but I will perhaps start with
02:37a bit of a disclaimer.
02:39You may have already realized that I'm a
02:41basic scientist, I am not a clinician.
02:44And you know, while I think,
02:47you know,
02:47our work is really rooted in sort of
02:49fundamental biological questions,
02:51I do consider myself sort of cancer
02:53biology adjacent.
02:54And I have been for quite a while.
02:56So my PhD work focused on the
02:58Hedgehog signaling pathway.
02:59So I sort of became quite familiar
03:01with the sort of hedgehog.
03:02Even medulloblastoma field and
03:03then during my postdoc I sort of
03:06collaborated with a few different
03:07cancer biologists sort of uncovered the
03:09roles of RNA modifications and cancer.
03:11So you know while we are very much
03:14a fundamental basic science lab,
03:16I,
03:16I really do not only enjoy but
03:18really sort of motivate the lab to
03:20think about sort of the applications
03:22and implications of our work for
03:24for cancer biology.
03:26But all of that being said,
03:27I rely a lot on collaborators and sort of
03:30dedicated dedicated clinicians to sort of.
03:32Would have you know helped motivate
03:33our and drive our work forward
03:35in in that respect.
03:36So today I'll give you a little bit of
03:38an introduction on our new modifications.
03:41It's not always sort of the the
03:43first thing you think about when
03:45it comes to cancer biology.
03:46So I will then sort of segue directly
03:49into some of the work that's already been
03:52done on RNA modifications and cancer.
03:54And then I'm going to describe
03:56actually the current limitations of
03:58a lot of that work including my own
04:00and and try to describe for you the
04:01sort of new approaches that the lab
04:03is taking to try to sort of improve
04:05our ability to really understand how
04:07these chemical marks might actually
04:09sort of regulate gene expression
04:11and how we might actually be able
04:13to translate that into the clinic.
04:16And so with that,
04:17I'll just dive right in.
04:18And I think,
04:19you know,
04:19when we think of the central dogma
04:21in terms of its sort of,
04:23you know,
04:24the fund,
04:24the foundations of gene expression
04:26regulation,
04:26this probably looks quite familiar.
04:28This is what I teach the
04:29undergraduates that I teach
04:30biochemistry over on the main campus.
04:32And so when we think about
04:33gene expression regulation,
04:34really what we're thinking about
04:35is the flow of information from
04:37the genome and DNA through RNA
04:39molecules and then on to proteins.
04:40But of course, you know,
04:42this is a rather simplified view and of
04:44course you know DNA needs to be replicated.
04:47We know that a lot of interesting
04:49reverse transcript cases exist to
04:51sort of move back from RNA to DNA
04:53and then of course protein based
04:54enzymes and factors of course
04:56regulate this at many steps.
04:58But on top of that, you know,
05:00even within each of these processes,
05:01of course there are many
05:03sort of intervening steps.
05:04And so as an RA biologist,
05:05I'm going to bias your view
05:07and say that the RNA,
05:08so to sort of protein transition
05:09is the most interesting.
05:10But of course some of my basic
05:12biology friends will disagree with me.
05:14But that being said, you know,
05:15even just to go from a functional
05:18MRI MRA molecule to a protein,
05:20we have to undergo many intermediate
05:22steps including things like capping,
05:24splicing, processing of other types,
05:27then translation into protein
05:28and then eventually.
05:30RNA decay.
05:31And as you know,
05:33you have probably already quite
05:35familiar at each of these steps DNA,
05:37RNA and protein.
05:38There are many different chemical marks
05:40that can regulate these processes.
05:41This is sort of a fundamental paradigm of of,
05:44you know,
05:45signaling and gene expression regulation.
05:48And so we know for instance that
05:49DNA can be methylated and that
05:51this is sort of a critical,
05:53you know,
05:54regulator of gene expression at
05:56the level of the genomic DNA.
05:58And then all the way on the other
06:00side we of course know that enzymes.
06:02Think of things like kinases,
06:04of course phosphorylated and modified,
06:06with many other post translational
06:08modifications that really dictate
06:09their activity and function,
06:11and in some cases localization.
06:13What some people are sometimes
06:15less familiar with is that in
06:17fact RNA molecules are also very
06:19heavily chemically modified.
06:20And this can sometimes be a little
06:22bit counterintuitive because when
06:24we think of our name molecules
06:26and particularly M RNA,
06:28we think of very transient
06:30sort of chemical molecules.
06:31So why would you sort of want to
06:33fine tune something that is such,
06:35you know,
06:35such a transient molecule in nature,
06:37but it turns out that RNA actually
06:40carries over 150 now known RNA.
06:43Qualifications,
06:44and I'm showing you just a
06:46very small subset of them here,
06:48but these range from everything from
06:50the addition of methylation marks to
06:53isomerization ins around different bonds.
06:56And you know.
06:57On the other end,
06:59we have sort of double methylations,
07:01but this is actually a very
07:02sort of limited view.
07:03And there's all kinds of incredible
07:05chemistry that can happen on
07:07RNA and particularly T RNA's
07:08to regulate gene expression.
07:11And so if we sort of overlay
07:13this on what
07:14we actually encounter in the cell,
07:16which is not a very linear pathway,
07:18we get a very complex picture like
07:21this where we have modifications that
07:23are in fact regulating literally every
07:26step of an M RNA during its life cycle.
07:28And so many of these chemical
07:30modifications are added,
07:31sort of as the RNA is being made.
07:33Not all of them are, but many of them are.
07:36And so the RNA modification machinery
07:39will in fact interact with.
07:41My cursor is frozen.
07:43Well, actually interact with
07:44the transcription machinery as
07:45the RNA is being made.
07:47The RNA.
07:48Most MRA's carry a 5 prime
07:50cap that is in fact derived of
07:53numerous modifications and then
07:55every step along the way from
07:58splicing to polyadenylation export,
08:00all of this sort of,
08:02you know,
08:02modifications have been implicated
08:03in many of these processes.
08:05They're exact functions.
08:06We don't always know,
08:07but we have sort of evidence to
08:10suggest that modifications can
08:11regulate every step of this process.
08:13Once an MRA gets exported,
08:16modifications can impact the
08:18association with translation machinery.
08:20So the ribosome itself can in fact
08:22detect sometimes modifications
08:23out or in your start codons,
08:25and then something that's going
08:26to come up a few times today.
08:29Our name modifications can
08:30also regulate M RNA decay.
08:32And now this picture is actually
08:34only focused really on M RNA.
08:35It turns out really all RNA's are
08:37modified really across all kingdoms of life,
08:39but for the purposes of today,
08:41we're mostly going to be focused on M RNA.
08:44And so when you zoom in very,
08:46very far down,
08:47really at the sort of chemical level,
08:49really what's going on here is that,
08:51you know,
08:51these small little chemical
08:52moieties can in fact dramatically
08:54change base pairing patterns.
08:56And so just as an example at the
08:57top here I'm showing you an AU
08:59base pair and a few examples of
09:01different modifications and how
09:02they would sort of impact base
09:04pairing at the chemical level.
09:06And so the addition of a methylation
09:09group here an N6 methyl adenosine,
09:12which is abbreviated M6A.
09:15Can slightly destabilize this base pair.
09:18If you actually shift that
09:20methylation group over just a bit,
09:22you actually get something that
09:24completely impedes base pairing
09:26altogether and then so on and so forth.
09:28You can imagine that as you
09:29increase the chemical diversity,
09:30you can really impact on RNA
09:32base pairing and structure and
09:34a really wide variety of ways.
09:36But at the sort of functional level,
09:38zooming back out a little bit,
09:39what does this really mean?
09:41And So what do these modifications
09:44actually do?
09:45So we're going to focus mostly
09:46today on one RNA methylation
09:48mark and six methyl adenosine,
09:50which I'm showing on the top left here.
09:53And it's a great example for
09:54a lot of reasons,
09:55but one of them is that we can
09:57actually ascribe some very specific
09:59M RNA regulatory mechanisms
10:00to this specific mark.
10:02So this methylation Mark is
10:04installed by a complex of
10:06a methyl transferase complex that
10:08contains metal three and metal
10:1014 proteins as well as some other
10:13accessory factors and it can be removed.
10:16By demethylase SES.
10:17Two of which I'm showing you here,
10:18called FTO&amp;L PH-5, essentially making
10:21this sort of a reversible process.
10:23You can add a methylation mark,
10:24you can remove it back and forth.
10:27But the consequences of of
10:29this particular methylation,
10:30particularly near the three prime end
10:32of M RNA's is is really interesting
10:34because in fact that mark alone is
10:37sufficient to sort of recruit the
10:39adenylation and decay machinery to M RNA's.
10:41And so essentially what happens is
10:43the this methyl mark is specifically
10:46recognized by multiple different proteins,
10:48but in many cases this protein
10:50called YTHD F2.
10:52And in that process,
10:54YTHD F2 then recruits the DND
10:56adenylation and decay complex,
10:58which effectively means that
10:59this M RNA is destabilized not
11:01necessarily at the chemical level,
11:03but because the actual decay machinery
11:06is getting actively recruited to this
11:08M RNA to decay it more quickly than
11:11it would if it was not methylated.
11:13And so this,
11:14it turns out,
11:15can then coordinate with other modifications.
11:18So what I'm showing you here is an
11:20example actually from development.
11:22This is for those of you that are
11:24interested in developmental biology,
11:25the maternal zygotic transition
11:27and zebrafish development.
11:29But really this is applicable to
11:31many other situations as well.
11:33And So what can happen is you can have for
11:35instance in this case a subset of genes,
11:37in this case the the maternal genes
11:40that are marked by this M6A mark.
11:42That decay machineries and recruited at
11:44the appropriate time and development,
11:46causing a drop in the maternal the
11:49levels of these maternal transcripts at
11:52the time at which the zygotic genes are
11:54the ones that are supposed to be activated.
11:57So to facilitate this transition,
11:59essentially maternal genes are being
12:00decayed by the presence of this mark.
12:03But there's a subset of transcripts
12:05in this context that actually
12:07need to be maintained and
12:08stabilized through this process,
12:10and it turns out that these the subset of.
12:13Transcripts is actually marked
12:14by a different modification,
12:16in this case 5 methyl cytidine.
12:18And in this case in fact the modification
12:21is recruiting a different set of
12:23machinery that is preventing these
12:24these transcripts from being decayed
12:26and allowing for sort of longer half
12:28lives and stability of those transcripts
12:31through this developmental transition.
12:33Now,
12:33it's important to note actually that
12:35we have arrived at this sort of
12:38very simplified picture by numerous
12:39studies in many different contexts,
12:41all studying different modifications
12:43on their own,
12:44separately.
12:44So we have sort of synthesized this
12:47information together to postulate
12:49a model whereby you could have
12:51different modifications coordinating
12:52these events all in one system.
12:54But I will point out that we actually
12:56don't have the power for the most
12:58part yet to actually detect all of
13:00this sort of in real time, you know,
13:02with multiple modifications.
13:03At once.
13:04But you can imagine that this kind of
13:06decay mechanism would not only be
13:08important for something like development,
13:10but also in cases of disease.
13:13And so as soon as this sort of
13:15decay mechanism was discovered,
13:17we had an explosion of work in many,
13:20many different areas, but particularly
13:22in various sort of forms of cancer.
13:26And so I'm showing you just a subset
13:27of papers here, most of them not mine,
13:29but one of them is and I will sort
13:31of use that as an example, but.
13:34Essentially by studying these
13:36different modifications in
13:37multiple different types of cancer,
13:39it was sort of realized that it might
13:41be that these modifications are also
13:43regulating transcript stability of
13:45critical oncogenic or tumor suppressor
13:47transcripts in the context of cancer.
13:49And So what I'm going to do today is sort
13:52of use one of these studies as an example,
13:54this one here,
13:55which is actually done in AML,
13:57but for reasons that hopefully
13:59will become clear are sort of
14:01relevant to some of the brain cancer
14:03work we're hoping to do as well.
14:04But really this is just sort of
14:06to give you a snapshot of sort
14:08of what we have learned,
14:09but also the significant limitations
14:11that we still encounter in this field.
14:15And so this story started,
14:17this is a collaborative effort.
14:19I was sort of, I was a postdoc in
14:22Juan he's lab and John Jin Chen was
14:24working on a big project essentially
14:26trying to characterize the roles
14:28of N6 methyl adenosine in AML.
14:31And they stumbled on sort of a really
14:34interesting observation with respect
14:36to R2 HYDROXYBUTYRATE or R2 HG,
14:39which at the time was sort of
14:41thought to be an uncle metabolite.
14:43And that was, there was sort of
14:45various lines of evidence for this.
14:46But essentially there was sort of a
14:49long history of of this metabolite
14:51being described as an Aqua metabolite.
14:53But there were some sort of weird
14:55results that I'll get into that sort
14:57of suggested that that might not be
14:59the case in every in every situation.
15:02And so this metabolite is actually
15:04derived from a mutant form
15:07of isocitrate dehydrogenase.
15:08And so I DH isn't the enzyme that's
15:11responsible for transforming isocitrate.
15:14To Alpha Ketoglutarate,
15:15which is a critical factor for
15:17many many different enzymes,
15:18including RNA modification
15:20enzymes in the cell.
15:22But it turns out that
15:24with specific mutations,
15:25as many of you might be aware,
15:27I DH can generate sort of a different
15:29a different form of this molecule card,
15:32excuse me, called R2 hydroxy glutarate.
15:35So it turns out that this
15:38particular form of this metabolite,
15:42R2 HD can in fact compete with
15:44alpha ketoglutarate at many of the
15:47sort of enzyme sites that require
15:49alpha ketoglutarate for activity.
15:51This includes cofactors for many
15:54different enzymes including histone
15:56DNA and RNA demethylase hes,
15:58which, you know,
15:59if you've encountered any sort of
16:02gene expression analysis of cancers,
16:03we know that many of these are dysregulated.
16:06And in many different cancers.
16:09And so sort of combined,
16:10we can sort of imagine a scenario
16:13where the presence of suddenly,
16:15you know,
16:15a much higher level of this metabolite
16:17R2 HG could significantly impact gene
16:19expression patterns based on the fact
16:22that it would sort of compete with
16:24these enzymes that are modulating
16:25all of these different groups.
16:28And so it was interesting about this
16:30particular study and when they sort
16:32of like roped myself and a few other
16:34people in my postdoc lab and on this
16:36project was the sort of interesting
16:38observation that when they just
16:40took a panel of AML cell lines.
16:42So these are grown in dishes
16:43and culture in the lab.
16:45There was a pretty striking difference
16:47in how sensitive they were to
16:50the presence of this metabolite.
16:52And so there was actually a
16:54panel of about maybe three times
16:55the size in the paper.
16:57And because this is published,
16:58I'm not going to go through
16:59sort of all the data,
17:00but I'm going to try to give
17:01you sort of the snapshots to
17:03sort of give you the idea.
17:04So in just this subset of cells,
17:08we can see that there is quite
17:10a few that are sensitive in the
17:13sense that the size of the circle
17:15means. How many cells are still viable
17:16and so at an early time point which is in
17:19Gray you can see all the cells are viable.
17:21But when you treat with R2 HG the
17:23the different colors are different
17:24time points and you can see that
17:26dramatically the circle gets really small
17:28really quickly and so this suggests
17:30a cell line that when you add this
17:33uncle metabolite a lot of cells die.
17:35But on the flip side,
17:36you can see in the bottom row here
17:38that there are actually quite a few
17:39that are also quite resistant to this.
17:41So despite using the same time course,
17:43the same dosing we have, you know,
17:45different cell lines that have
17:46ostensibly are the same thing,
17:47but of course we know that they're not.
17:49They're responding quite differently
17:51to the presence of this metabolite.
17:53And so this sort of started us
17:55off on a quest to sort of figure
17:57out why that might be true.
17:59And through this process,
18:01the Chen Lab noticed that one of
18:03the enzymes that was particularly
18:06disregulated across cell lines that
18:08were sort of sensitive and resistant
18:10to this metabolite was this enzyme FTO,
18:13which you might recall was one of the
18:15demethylase hes that I highlighted.
18:17So FTO removes the M6A mark from many
18:20different M RNA's and the sort of
18:23predicted outcome of this would be that
18:26you would have more M6A left on transcripts.
18:29Because you've sort of
18:31inhibited the demethylase.
18:32So that's a relatively simple thing to test.
18:34And so we can measure these
18:36modifications by various different means.
18:38The data I'll show you is a
18:40mass spec based measurement,
18:41but we did this by by other means as well.
18:44And so when you measure the
18:47levels of this M6A modification,
18:49we always normalize relative
18:51to the unmodified a.
18:53So that's sort of what this
18:55ratio here on the Y axis shows,
18:57can see that when you treat
19:00with drug for sensitive lines,
19:02you see this small but sort of reproducible
19:04increase in M6A levels like we predicted,
19:07whereas it was much more variable in
19:09the resistant lines and there was
19:11certainly no sort of consistent effect.
19:13So this is potentially intriguing,
19:15but it still didn't really
19:16explain everything.
19:17And so we sort of turned to like what,
19:20why would the presence of this
19:22metabolite actually have such an impact?
19:25And we got a big clue from this
19:27Western blot that was done in the
19:29Chen lab in which they showed that
19:32the levels of this demethylase FT O
19:34were vastly different in the sensitive
19:37lines versus the resistant lines.
19:39And conversely levels of Mick
19:42were dramatically lower.
19:44In the sensitive lines,
19:45then in the resistant lines.
19:47And that the levels of these two
19:50key factors actually change when
19:52you add this metabolite RHG.
19:55So this was intriguing because of
19:57course Mick is a critical factor
19:59that drives a lot of not just AML,
20:01but a lot of other cancers.
20:03And so we wondered whether we
20:04could sort of turn
20:06a sensitive line into a resistant
20:081 purely by increasing the
20:09levels of making the cell.
20:11And it turns out that you
20:12can actually do that.
20:13And so I'm just showing you.
20:14One snapshot here, but there's no more.
20:17One cell line was one sensitive
20:19cell line where essentially the
20:21red lines represent just no sort
20:24of additional mix just at baseline.
20:26And when you add the R2HG,
20:29you can see that there's a drop in
20:31the ability of the cells to proliferate.
20:33When you exogenously just add a
20:36bunch of mic expression to the cells,
20:38you can see that you can make
20:40them more resistant to it.
20:42So this is just one snapshot,
20:43but this is sort of.
20:45The model that we sort of,
20:47you know,
20:47took out from this and I will say that
20:49there was a whole bunch of sort of
20:50mouse work done as well that I was not
20:51involved in and that I'm not showing.
20:53If you're interested,
20:54you are welcome to check out the paper.
20:56But all together we sort of settled
20:59on this model where perhaps the
21:02relative levels of the demethylase
21:04and Mick were sort of essentially
21:07dictating where on a spectrum of R2
21:10HG sensitivity these cell lines were.
21:12So the idea would be if you have.
21:15A lot of mixed transcripts floating around.
21:18You add R2 HG that inhibits FTO.
21:22You have less demethylation.
21:25You get a mic target gene expression
21:28because essentially M6A levels are lower.
21:32Transcript decay is lower,
21:34meaning overall mic levels are
21:36high and so the mic itself and its
21:39target genes are highly expressed,
21:41however.
21:41In the opposite scenario where maybe
21:44you just have a lot of mic present or
21:47FTO is able to demethylase and remove a
21:50lot of that methyl that methylation mark,
21:53you can overcome this by essentially just
21:56boosting the amount of Mick transcript.
21:59So essentially the
22:01methylation facilitates decay,
22:03loss of methylation facilitates
22:04stability and essentially by toggling
22:06both of these variables you're able
22:08to get sort of the spectrum of R2
22:11HD sensitivity even in what is.
22:12Ostensibly cell lines from sort
22:14of the same cancer.
22:15Now there's a lot of limitations to this.
22:18I'm going to go into some
22:20of them in great detail.
22:21But of course a big one is that a lot of
22:23this is done in cell lines and mouse models.
22:25And we're sort of zooming in on
22:27one particular aspect of this.
22:29We know that this metabolite does many,
22:31many other things including regulation of
22:33like way more metabolites than you know.
22:36We could even particularly cover
22:37in this talk.
22:38But I hope this is sort of highlighted
22:40one way in which sort of the presence
22:42or absence of this methylation.
22:43They can sort of dictate where you are,
22:46particularly in the context of cancers
22:49in which there are ID H mutations.
22:52And you know,
22:53this sort of set me on a path of a
22:56little bit of a rabbit hole and.
22:59It turns out that I didn't know
23:00this at the
23:01time. I'm sure all of you already know this,
23:03that you know glioblastoma is another
23:05instance in which ID H mutation
23:07status can dramatically impact
23:09sort of prognosis for patients.
23:11And one of my students in
23:12the lab is working on this.
23:13Emily actually found this sort
23:16of lovely table that summarizes
23:18a lot of what we know about N 6
23:21methyl adenosine in glioblastoma,
23:23not even taking into account
23:25ID H mutation status.
23:27And if you just read a few lines,
23:29you can already start to see.
23:30There's a lot of conflicting data on this.
23:33So for instance depending on you
23:36know which paper we want to look
23:38at all sort of published within
23:40the same approximate time frame.
23:42We have many different M6A
23:44related factors we can look at.
23:47But even when we're looking at the same ones,
23:49we can find papers that say the presence
23:51of that or absence of and says oncogenic.
23:54Another one says it's tumor suppressive.
23:56We different,
23:57you know different target genes
23:59are being explored mix socks.
24:01To Fox M1, essentially there's very
24:03little concordance among any of these.
24:05And that's not to say that they're wrong,
24:07but I think it sort of reflects that.
24:09Depending on exactly,
24:10let's say what cell line or
24:12what system you're using,
24:13if you have something like a
24:15spectrum of sensitivity or just
24:16differences in gene expression,
24:18you can arrive at this very confusing
24:20pattern of data that I am not a clinician,
24:22like I said.
24:23But I would assume that this doesn't
24:24exactly scream confidence to you,
24:25that this is something that,
24:26you know,
24:27maybe would be interesting to
24:28pursue in the clinic.
24:30And so this sort of motivated me to
24:32think a little bit more about why we
24:34might be getting it not necessarily so wrong,
24:37but why are we so confused by this.
24:39And I think a lot of this comes
24:41down to the approaches that we're
24:42using to study these problems.
24:44So M6A in particular has been a very
24:47popular modification to study because
24:49it's actually relatively simple to
24:51sequence where it is in the transcriptome,
24:53you know,
24:54in any cell linear system you want.
24:56The reason for that is that
24:58there's an antibody based approach
24:59to to sequence M6A sites.
25:00And essentially what you do is
25:03you take an antibody and your
25:05favorite RNA sample of interest,
25:07you mix them together,
25:09you pull down your M6A containing
25:12RNA's on your antibody and then you
25:14use high throughput sequencing which
25:16is now so run-of-the-mill that you
25:17can really do this with even very,
25:19very small amounts of input RNA.
25:22So this has been fantastic in terms of
25:24driving the field forward and sort of.
25:26Giving us the ability to sequence and
25:28say in a lot of different contexts,
25:30but it's essentially driven people
25:33to to this sort of top down approach
25:37of trying to identify unifying
25:38sequence features and gene groups
25:40and use that to try to sort of
25:42generate functional themes.
25:43So let's identify the most
25:45dysregulated sort of genes when
25:47we, you know lose M6A and and see
25:49if that tells us anything about
25:51how it might be functioning in
25:53our favorite system of interest.
25:55But it turns out that this can be
25:57a little bit tricky to interpret.
25:59So when you do an experiment like this,
26:01you will get many, many genes and
26:02you will do all kinds of analysis.
26:04And that's sometimes very illuminating.
26:06But in some cases it's also worth thinking
26:09about what exactly do those data look like?
26:11And so it turns out,
26:12when you do an experiment like this,
26:14you get, let's say,
26:15a section of your favorite gene.
26:16It could be Mick,
26:17could be anything else.
26:18And you will get a whole bunch of
26:20sequencing reads at different places
26:22along your favorite transcripts.
26:24But what this does not tell you is
26:26whether you have a situation like
26:28this where you have multiple M RNA's
26:31where half of them have multiple M6A
26:33sites and half of them have none.
26:36Or you could have a situation
26:38like this where actually all of
26:40your transcripts have M6A on them,
26:42but in slightly different locations.
26:44You might think that seems like a bit
26:46of a detail, like, why do we care?
26:47But it turns out that that can
26:49actually have very profound
26:50functional consequences in terms
26:52of how you interpret the data.
26:54And so for instance,
26:55you know,
26:55this might actually represent functionally
26:58two different transcript pools.
27:00It could mean that one of these pools is
27:02getting localized to a specific place
27:04or getting decayed and one pool is not.
27:06And conversely,
27:07this could actually just mean that,
27:09you know,
27:10the exact site of the M6A doesn't matter,
27:12it just has to be somewhere in this vicinity
27:14and then it'll have the same outcome.
27:16But we can't sort of disentangle
27:19these two things with a simple
27:21IP sequencing experiment.
27:23And that's not to say we haven't made
27:24improvements to this these methods.
27:26We have made a lot of improvements to
27:28IP based sequencing that allow you to
27:30get more precise sort of location data,
27:32but a lot of those are not easily applicable
27:35necessarily to very low abundance.
27:37Examples which might sort
27:38of be clinically relevant.
27:40So.
27:42This sort of leads me to sort of
27:44posit to you that part of the
27:45reason we get so much confusing
27:47data is because we are actually sort
27:48of the data itself is is not as
27:51clarifying as maybe we would think.
27:53And so how do we change our
27:55approaches to sort of deconvolve
27:56some of these variables.
27:58And this is really sort of at the crux
28:00of essentially everything my lab studies,
28:02but we we sort of apply this in a
28:04lot of different ways.
28:08So just to sort of summarize the
28:10approach that I just gave you because
28:12I'm going to try to sort of convince
28:14you that our approach sort of
28:16provides an interesting alternative.
28:17The approach that I just described is
28:20what I would call a top down approach.
28:22So we do transcriptome wide sequencing
28:24usually for one specific modification.
28:27We, you know, sequence everything and
28:28then we try to pick out features that
28:31are common across all the transcripts
28:33that have our favorite modification.
28:35And this is in fact what led to our,
28:37you know, the identification of this decay.
28:39Mechanism that I described earlier.
28:40So it is definitely not a useless approach.
28:42It has been very productive,
28:44but I will argue that we sort of hit
28:46a wall in terms of really getting a
28:48mechanistic understanding of of how
28:50some of these modifications work,
28:51which has led to some confusing results.
28:55So what I have tried to do,
28:57you know in setting up the research
28:59program in our lab is to try to
29:01develop what I would call more of
29:02a bottom up approach.
29:03So we identify specific transcripts and
29:05then we do a deep dive to figure out,
29:08OK, what are all the different
29:11modifications on this transcript.
29:12Once we know that,
29:13can we identify the regulatory
29:15enzymes are there,
29:16you know variations across
29:18different diseases etcetera.
29:20My argument is sort of that if
29:22we can do this,
29:23you can then go and look in your next
29:25favorite transcript and say it does
29:26it also apply here and essentially
29:28build out the network from there.
29:30So the idea is to sort of not limit
29:32ourselves necessarily to these
29:34RNA that we're interested in,
29:35but to use them to identify rules
29:37and then try to figure out if
29:39those rules apply elsewhere.
29:41And so I'm going to drive describe
29:43sort of the two approaches we're
29:45using to sort of make maps like this.
29:47The first one is a mass spectrometry
29:50based approach,
29:51and this has really been pioneered in
29:53my lab by a student, Lauren Wilson,
29:54aided by many other people in the
29:57lab including Josh and undergraduate.
30:00But the essence essentially the idea is this.
30:02So what if we just took our favorite
30:04transcript and we designed a whole
30:06bunch of probes that are complementary
30:08to that transcript and we just
30:10purified it out of cells?
30:11This sounds crazy,
30:12but Lauren has actually gotten this to work,
30:16arguably for more abundant transcripts.
30:17But nevertheless, you know,
30:19it's actually working quite well,
30:20as I'll show you in a second.
30:22And the idea here is,
30:23OK,
30:23let's just hypothetically say that
30:24we can purify the transcripts
30:26that we're interested in.
30:28And then digest them into individual
30:31nucleosides down to sort of
30:33the individual module level and
30:35then analyzed by mass spec sort
30:37of everything that's in there.
30:39And so I'm happy to discuss the
30:41details of the mass spec with
30:43anyone who's interested.
30:44But the crux of it is that
30:46essentially between the
30:47the fragmentation patterns and the retention
30:50time on a specific column that we use,
30:53we can distinguish between even very
30:55closely related chemical species.
30:57So that we can distinguish even two
30:59different singly methylated species
31:01like and one methyl adenosine and and
31:03six methyl adenosine and we can go
31:05down the line and look at sort of any
31:07modifications we might be interested in.
31:10And so this has turned out to be
31:12relatively fruitful as I just alluded to.
31:14And so we're starting to do this
31:16admittedly with very abundant transcripts.
31:18And so the data I'll show you
31:19is for a very abundant but very,
31:22very big long non coding RNA called neat one.
31:25It's got some very interesting biology
31:27that I won't really go into at the moment.
31:31But essentially we can purify neat 1
31:34transcripts to really high enrichment
31:37relative to sort of baseline total RNA.
31:40So the exact numbers that you get for
31:42enrichment depend a little bit on
31:45what you're comparing it to, right?
31:47So are you comparing to an M RNA
31:49or a very abundant ribosomal RNA?
31:51But for instance,
31:52if we look at sort of a favorite
31:54housekeeping M RNA.
31:55We can enrich this long non coding
31:59RNA many thousand fold over what
32:01we would have in just sort of the
32:04baseline pool of RNA.
32:05And importantly,
32:06Lauren can now do this to sort of a
32:09level of abundance that we can actually,
32:11you know,
32:12get enough RNA out of this to
32:14digest and do mass spec analysis.
32:16And so the idea here is that we take this,
32:19you know,
32:20RNA of interest and we just look
32:22for our favorite modifications.
32:23I'm just showing you a few of them here.
32:26You can look at different cell lines,
32:27you know,
32:28whatever cell lines you might be
32:30interested in and then basically
32:32profile the relative abundance
32:34of different modifications.
32:35In this sample. In different cell lines.
32:39And so for instance here we have a 549
32:42and HeLa cells and you can see that
32:45the M1A levels are a little bit different,
32:48which maybe it's interesting,
32:49maybe it's not.
32:50We'll have to find out.
32:52The M6A levels are relatively
32:54stable and so on and so forth.
32:57And so this is just a snapshot,
32:59but I hope you can sort of appreciate that,
33:01you know,
33:01this is one way to get a much more
33:04unbiased picture of what might be
33:06present in your long and encoding.
33:08RNA of interest I will say in
33:10the interest of Full disclosure
33:11we cannot do this yet for MRA's.
33:13MRI's are much less abundant.
33:15Maybe someday we will get there,
33:17but at the moment we can at least
33:18take a stab at taking our favorite
33:20long non coding our names of interest
33:22and trying to see what modifications
33:24are sort of sprinkled throughout.
33:25And it will say another RNA we
33:27have applied this to is Merlot,
33:28one which many of you might be familiar with.
33:31It's another sort of abundant
33:33long non coding RNA that's
33:34been of interest in in a few different
33:36cancers and so we're taking a stab
33:38at looking at the modifications.
33:39From that arena as well.
33:42But again, I told you I
33:43would tell you limitations.
33:44So mass spec is a very powerful tool,
33:48particularly to get really specific
33:51chemical identity information.
33:52But you lose location information
33:54because we're digesting all of
33:56this up into tiny little pieces
33:57so that we can really identify the
33:59chemical species that are present.
34:01So that's a bit of a problem
34:03and as I already alluded to,
34:04it's also limited at the moment
34:07much more abundant transcript.
34:08So things that we can actually
34:09purify to a degree to where we can
34:11actually do Mass Effect on them.
34:13It is also limited in the sense that,
34:15you know,
34:16we certainly wouldn't be able
34:17to do this from something like
34:18a biopsy or a patient sample,
34:19which I'll get to later.
34:21But that's sort of, you know,
34:22the dream down the line.
34:24So what do we do to complement this approach?
34:27So this is where we turn to a
34:29sequencing based method and this
34:31is actually where I've had the
34:33tremendous good fortune to work with
34:35another member of the department.
34:36And Anna Marie Pyle,
34:38her lab is just down the hall
34:40and we've got a fantastic sort
34:41of RNA wing in the Yale Science
34:44Building and they discovered and
34:46characterized this interesting reverse
34:48transcriptase enzyme called marathon.
34:51And Marathon is sort of interesting
34:54from a few different aspects.
34:55So first, as the name suggests,
34:57it can reverse transcribe
34:59really long transcripts.
35:00I mean,
35:01I actually have never seen anything quite
35:03as processive as this particular enzyme.
35:06The other thing that it does,
35:07though that's particularly useful for us,
35:09is it tends to install mutations
35:11when it encounters a modified
35:12modification in the RNA.
35:14That's not to say every single modification,
35:16but many of them,
35:17and even some that are a bit more subtle.
35:20So you might already be starting to
35:21piece together that this would be
35:23relatively useful if we could just
35:25reverse transcribe an RNA and use
35:26mutations to sort of identify where
35:28possible modification sites might be.
35:31Now the one caveat here is that
35:33we can't always tell the specific
35:35modification just based on the
35:37fact that there's a mutation there.
35:39There's a lot of different ways
35:41you can get mutations,
35:42but the massive benefit of this
35:43type of approach is that you can
35:45do it even for M RNA's that are
35:47not very abundant because for a
35:49reverse transcription reaction.
35:50Need much less than something
35:52like from aspec.
35:53So Dorothy in the lab has been the
35:55one really pioneering this approach.
35:57And as you can see here,
35:59so this time I'm using BRCA 2 as sort
36:01of an example of an MRA of interest.
36:04And you can see that as you
36:06reverse transcribe with marathon,
36:07you see some specific peaks where
36:11you get mutation signatures.
36:13We thought we knew what these
36:14mutation sites represented were
36:15actually a little bit less short now,
36:17which is why I'm not telling you the
36:18identity, the identity of the modification,
36:20because we're still trying to work that out.
36:22But what's interesting is that you can see
36:25that there's actually different levels of
36:27mutation depending on the specific site.
36:30As I said though, an important
36:31caveat here is you can get mutations
36:33by a lot of different routes,
36:34including of course because the
36:35genomic DNA might be a little bit
36:37different than your reference.
36:38And so here you can see that we've
36:41actually just picked up a snip.
36:43You can tell that actually you can
36:45almost predict that based on the
36:47fact that it's so highly modified.
36:48But then these others you can see
36:50there aren't really sort of concordant
36:52DNA based mutations and so we're
36:54sort of trying to follow up what
36:56those modifications might be.
36:57And so you can see that there's sort
36:59of an iterative process where you
37:02could essentially either identify
37:03potentially interesting modifications
37:05by mass spec in the favorite RNA and
37:07then try to use sequencing based on that
37:10knowledge to identify where they are.
37:11Or you could start with sequencing,
37:14identify specific sites on
37:15a transcript of interest,
37:17and then try to work out what
37:18the modification is either by
37:19mass spec or sort of another,
37:21you know, orthogonal method.
37:23But by iterating this process,
37:25the idea is that basically
37:26you can get sort of a map of.
37:28And RNA of interests and really
37:29that RNA could be anything.
37:31I'll tell you a little bit
37:32about the ones that you know,
37:33we're working on at the moment.
37:34But the idea is that if you can
37:36get this sort of more complete
37:38picture of what's actually on there,
37:40you can then go in and look at,
37:41OK, if I do perturbation X or
37:43if I look in disease Y,
37:45you can actually start to look at specific
37:48changes in those specific locations.
37:50And So what are we actually trying?
37:53Whoops,
37:54trying to do with this so.
37:57There's a lot of different applications
37:58I envision and this is where I probably
38:01should have put disclosure at the
38:02beginning that says you know I would
38:05love always perspective from clinicians
38:06and more translational researchers
38:08sort of interesting targets to look at.
38:10But I'll tell you a little bit about
38:11the ones that that we are looking at.
38:13And so I've mentioned the long non
38:16coding RNA's neat one and maillot one.
38:19These are sort of our first model
38:22transcripts let's say for the mass spec
38:24approach because they are abundant,
38:25they are big and there's
38:27some interesting biology.
38:28Associated with them,
38:29which means that once we have
38:31modifications of interest,
38:32we can go in and try to perturb them.
38:33We can go look in hopefully either
38:37sort of more disease relevant samples
38:39than just cell culture lines and
38:41actually go and look at whether
38:44these changes actually translate to,
38:46you know,
38:47samples that are a bit more
38:49directly translationally
38:50relevant. We're also interested
38:51in sort of M RNA's like Braca 2,
38:54but many others that we can use to sort
38:56of highlight the sequencing approach
38:57and essentially again go in and find
39:00some interesting modifications and try
39:01to perturb them and monitor how they
39:04change under under different conditions.
39:07But I think thinking sort of bigger picture,
39:09what these approaches sort of allow us
39:10to think about is to actually monitor
39:12changes and modifications in real time.
39:14And so a problem that we're,
39:17we're sort of starting to think a little
39:18bit about this is driven really by Emily.
39:20In our lab is can we actually, you know,
39:23look at instances of things like drug
39:25resistance where we're sort of used
39:27to thinking about DNA based mutations
39:29that are making a protein resistant
39:30to the drug that's targeting it.
39:32But are there other changes beyond
39:34that that we're missing that we could
39:37actually sort of use either as biomarkers
39:39or diagnostics or something else that
39:41we could actually either you know,
39:44target or at least use to monitor
39:47the development of disease sort
39:48of at the RNA level?
39:50The consequences of IH1 status are
39:52really unknown here and this is
39:55really where I think where we've
39:56become sort of really interested in
39:58glioblastoma for a few different reasons.
40:00One of them is the sort of huge
40:02dichotomy with with DH one.
40:04But another is that, you know,
40:06it's a disease in which,
40:07at least based on what I've read
40:09and I'm happy to be corrected,
40:10there's actually a fair amount of
40:11evidence that a lot of drug resistance
40:13isn't really happening based on
40:15mutations at the DNA level suggesting
40:17that there might be some really
40:18interesting things we can look at at the.
40:20RNA level including things like modifications
40:23that aren't really driven by you know,
40:25DNA mutations specifically.
40:27And so hopefully I've convinced you that
40:29that you know what we're doing is sort
40:32of somewhat feasible and interesting.
40:34I think moving forward,
40:35you know,
40:36we've definitely have a really strong
40:38interest in glioblastoma and this is
40:40sort of what the Sontag Foundation has
40:43sort of funded us to look into Umm,
40:45thankfully with the brain tumor
40:47centers as support and guidance,
40:49but we're really interested in sort of.
40:51Applying this across all different
40:53types of disease and other cancers
40:55in particular because I don't think
40:57that our approaches are necessarily
40:58going to be limited to you know,
41:00glioblastoma specifically.
41:01But you know,
41:03I'm hoping that with a little bit of
41:05discussion or maybe I've peaked some
41:07interest in some of you here that maybe
41:09you can give us some great new ideas
41:10and spark some new collaborations to,
41:13to work on this moving forward.
41:16And so with that I would just like
41:18to acknowledge the people that are
41:19actually doing the work in our lab.
41:21And so I think this is a up to
41:24date picture for the most part.
41:26So we are mostly graduate student lab,
41:28but we have undergraduates and
41:30postgraduate researchers working with us
41:33as well. We're supported by a tremendous
41:36team of collaborators both in in a piles lab
41:40but also Brent gravely and Emmanuel Saliba.
41:42We have a joint NHR I grant to sort of study.
41:46Um, you know, develop methods to sort
41:48of sequence modifications more broadly.
41:50We're also very grateful to our mass
41:53spec support and the Yale Chemical
41:56Biology Instrumentation Center.
41:58And of course our funding sources through
42:01the NIH, the Hood Foundation and the Sontag
42:04Foundation that I already mentioned.
42:06We've got also great support
42:08from the RNA center at Yale,
42:10which is a a great community.
42:12But of course I would definitely
42:13like to thank the Cancer Center,
42:15not only for the invitation to actually
42:16come in and talk to you a little bit today,
42:18but I was truly,
42:19it was the most pleasant surprise
42:21when I emailed Doctor Amura about
42:22this sort of like random hey,
42:24I'm writing a grant about brain cancer.
42:26Like, would you be willing to chat and and
42:28maybe talk through some of these ideas?
42:30It was a fantastic interaction
42:32and great conversation.
42:33I hope this is sort of a productive
42:35collaboration in the future.
42:37And with that,
42:37I'll just leave it there and I'll
42:39take any questions you might have.
42:50Well, let's check.
42:53I don't think so.
42:54In the meantime, let's get some
42:55questions from the audience.
43:03I have two questions though kind of like.
43:08The first one being I was wondering
43:10if you if your studies especially
43:12when they looked at like locations
43:14which does obviously cause an
43:16increase in strain escalation,
43:18seeing whether using methylation can
43:20change body affects or potentially
43:23changes the amount of modifications
43:25tomorrow night, I'm not saying.
43:28It's a really great question.
43:29That's actually a PhD project in our lab
43:33essentially not exactly that but yes,
43:34so that's a fantastic question.
43:36So for the people on zoom that
43:37may not have caught the question,
43:38the question is essentially about the
43:40sort of effects of methylation status of
43:42DNA on sort of RNA modification status.
43:45So at the moment there's a lot of
43:48correlative evidence suggesting that
43:50either DNA methylation and or histone
43:53methylation could impact sort of
43:54the presence or absence of specific
43:56modifications and then that will of course.
43:58Relate with sort of gene expression,
44:00but we have a bit of a chicken
44:02and egg problem.
44:03I think intuitively you would assume that
44:05the DNA level stuff would impact the RNA
44:07level stuff and only in One Direction.
44:09But it turns out that that's actually
44:11maybe not true and that there's sort of
44:14interactions between sort of chromatin
44:17transcription machinery modification
44:18machinery that maybe actually going
44:20back and impacting chromatin.
44:22So the short answer is yes,
44:24it likely is impacting it.
44:25We don't exactly know how,
44:27but one way that we're trying to study.
44:30Just to look at the code transcriptional
44:31regulation of different modifications
44:32and if we can figure out kind of
44:33exactly where they're put on,
44:34then we can go and tinker with the
44:36DNA and then figure out if it's
44:38still happening or. Yeah, yeah.
44:41Question is.
44:44Retaliating.
44:47It's like I don't know.
44:56Yes. So we we always try to validate
44:58sort of in multiple different ways,
45:00not only sort of loss of the enzyme but
45:02also loss of the modification it turns out.
45:04So the shorter answer is yes, we do.
45:06The tricky part is that oftentimes we're
45:08dealing with very small changes and
45:09actually in the case of metal three,
45:11it's even worse because it's an essential
45:13gene and if you lose it for too long,
45:15cells are dead.
45:17So we do our best to sort of.
45:20Tune the perturbation so we can get
45:22a change in modification but not,
45:23you know, lose everything.
45:25But I will say that actually one
45:26thing that we've been trying to
45:28work out though it's been very,
45:30very difficult is to use these
45:32sort of crisper cast based systems
45:34to sort of target enzymes to
45:36specific modification sites.
45:37So the idea would be you take a dead
45:39cast 9 or a dead cast whatever.
45:41There's like 10 of them now I think
45:43that are essentially trying to tether
45:45enzymes using this this machinery to a
45:47specific place using a guide RNA and
45:48then the enzyme is around and should be.
45:50Essentially just be removing a
45:52modification at that specific site.
45:55We're trying some of these systems.
45:57We haven't gotten all of them
45:59to work very well yet,
46:00but that would be sort of a much more
46:02targeted and much better way to look at
46:04sort of very specific modifications.
46:07So,
46:07yeah, good questions. Yeah.
46:15Yes. So yes, so the question is have
46:18we looked at IMIDAZOLINE and LH3? Yes,
46:21I actually didn't talk about the M1A work.
46:25I have a long history with that
46:26modification from my postdoc. Actually,
46:29it's been a very tricky one to study.
46:31And if you're familiar with that literature,
46:33you might know there's been a lot of
46:34arguments about sort of prevalence,
46:35presence, location, etcetera.
46:37And so I think are a lot
46:39without the H3 as a postdoc.
46:41And and the sort of odd thing that
46:44I stumbled on and I think this is
46:46like buried in the supplementary
46:48figure whatever of that paper is.
46:50It was strange because I when I
46:53overexpressed out PH3IN cell lines,
46:54you can see an ink.
46:56So you see an increase in LH3,
46:57you would expect a decrease in M1A and
47:00we did see that but when we knocked down
47:02LH3 we did not see the sort of reverse.
47:06But interestingly other labs have
47:07and so I don't know if this is
47:09sort of a methods based thing or
47:11not I will say in an in vitro.
47:12Experiment,
47:13it absolutely will demethylated M1A.
47:16It will do that probably on RNA and DNA.
47:18The question is more just if in the
47:20cell itself they're encountering
47:22each other to the extent that you
47:24would need for it to regulate M RNA.
47:26So we're not sure we're we're sort
47:28of going to still try to work that
47:30out in a more targeted way if we can.
47:33We actually would love to target LPH
47:343 with a cast system and sort of
47:37specifically demethylated specific places,
47:39specific sites but that's been a tricky 1.
47:42For a lot of reasons, yeah.
47:46So
47:47I was fascinated by the use
47:49of antibodies for enrichment.
47:52Have those antibodies ever been used inside
47:54to try to identify location
47:56of the marks or any kind of?
47:59Yeah, that's a great question.
48:00So I don't know about Umm.
48:04So yes, a little bit in sort
48:08of cell culture based systems,
48:10it's often a little bit tricky unless you
48:12can get a very specific concentration of
48:14a modification in a very specific place.
48:17For something like M6A that's sort
48:19of diffuse on many different RNA,
48:21it can be a little tricky.
48:22That being said, for some things,
48:24you know particularly if you
48:27have very concentrated sort of.
48:29You know, a nuclear body or
48:30a cytoplasmic body.
48:31You might be able to detect them,
48:32I don't know.
48:33Various people have tried and
48:35to sort of varying success.
48:37I think it's something we need
48:38to look at a little bit more.
48:39The the other tricky thing with
48:41these antibodies is it turns out
48:42they're not all super specific and
48:44so then we can't interpret whether
48:46the signal we're getting is actually
48:48because of that modification or just
48:50it's binding to a lot of things.
48:53So something we still need to work out
48:55but I there's people there have been
48:57some interesting methods trying to
48:59sort of use proximity based you know,
49:01akin to a proximity ligation
49:03based strategy where you know
49:04if you can bridge two things.
49:05Together to kind of combine an antibody
49:08with something else to tell you that
49:10that Mark is there with some success.
49:12But the imaging based methods are
49:13a little bit behind I would say,
49:15yeah.
49:19At the beginning.
49:24I was wondering if that's drastic
49:26enough to affect the RNA confrontation,
49:30but you could essentially have proteins
49:31that couldn't or wouldn't be picked up.
49:35Yeah, that's a great question.
49:36So the question is for those of you on zoom,
49:38whether the impact of modifications on
49:40base pairing would be sufficient to
49:42maybe even impact essentially decoding I
49:44guess is what you're asking and and sort
49:46of introduce mutations into proteins.
49:48There is certainly evidence that the
49:51presence or absence of modifications
49:53can impact at least at the very least
49:56sort of start code on usage and to
49:58some extent a little bit of decoding.
50:01Not as familiar with the
50:02translation literature.
50:03There's certainly evidence for it in sort
50:04of in vitro translation systems in vivo.
50:07It's a bit harder to tease apart,
50:09but if you're interested
50:10in that literature at all,
50:11actually Shalini Oberdorfer's lab
50:13has done some really amazing work
50:15on a C4C and how it sort of dictates
50:18start code on usage through sort
50:21of base pairing interactions.
50:22So that's not our work,
50:24but it's beautiful and I would
50:25encourage you to take a look at
50:26it if you're interested,
50:27because yes,
50:27there there's certainly some evidence
50:29for that.
50:39So the question is the time scale
50:41of our modifications and the
50:42and how that sort of relates to
50:44chromatin and things like that.
50:46So this is something that we're
50:48still trying to work out so.
50:50We've been working on actually a totally
50:52different strategy I haven't talked about.
50:54We're trying to work out a dual
50:55labeling strategy in our labs.
50:57This is a different graduate student,
50:58Luke, who is working on this
51:00where we're essentially trying
51:01to combine nascent RNA labeling,
51:03something like an EU approach or
51:05a fourth European labeling with
51:07essentially trying to both Mark
51:09naison RNA and nascent methylation
51:12with a deuterated Sam analog.
51:13We haven't been able to pin
51:16down the kinetics.
51:17I have some very preliminary
51:18data from like way back in my
51:20postdoc that this can happen.
51:21Within minutes.
51:22But it's very hard to catch
51:24the dual sort of label.
51:26And so that's something that
51:27we're still working out.
51:29We're hoping that that will be working soon,
51:31but that's exactly the question we
51:33want to answer because we don't know
51:35if the timescales are relevant yet
51:36because we can't pin them down yet,
51:38if that makes any sense.
51:46Hmm.
51:51The.
51:57Yeah. So the question is whether the,
51:58the mutation rate sort of essentially
52:00correlates with what's actually
52:02happening at the modification level.
52:03So we're literally working with
52:05Anna Kyle's lab to try to generate
52:07the standards to figure that out.
52:08So what we essentially need to make
52:11is a calibration curve with known
52:13sort of modified oligos because
52:15based on some previous work with a
52:18different reverse transcriptase,
52:20we know that there is some correlation
52:21between the level of modification.
52:23And sort of the mutation rate in terms
52:26of it's somewhat accurate reflection
52:28because you know it's reverse
52:30transcribing not just the modified
52:31pool but also the unmodified pool.
52:32So there should be some concordance.
52:34It is however probably not perfect and
52:36so we kind of need to make a calibration
52:39curve essentially getting you know
52:40known modified oligos at zero percent,
52:4225 percent,
52:4350% and generating a curve to try to
52:45correlate where those where those
52:47are based on some really early work
52:50in our first M1A paper that for
52:52M1A we know that mutations.
52:54Um are relatively usually relatively strong.
52:57And in that case M1A tended to occur at
53:00about 20% of its given transcript pool.
53:02So like if you had a favorite
53:04transcript that was 1A methylated on
53:05average it was about 20% methylated.
53:07We never figured out what that means though.
53:09We don't know if that means most
53:11of it's being decayed or you know,
53:12anything like that,
53:13but we're still working that out.
53:20Thank you very much.
53:22Sacred support is amazing talk.
53:24It's really very rewarding to see
53:27this high end science coming to
53:29visit her here on the other side.
53:31Let's hope that this will be the
53:33beginning of many more collaborations.
53:36And on that note,
53:37I would like to remind you all
53:40that we do have an RFA out for
53:43funding laboratory research.
53:44There has more of a translational character.
53:47So you might want to check it out.
53:48Thank you very much.