Somatic Mutations Reveal Personal History of Development and Aging

March 30, 2023

ID9776

To CiteDCA Citation Guide

00:00Doctor Alexei Abizov is
00:02a physicist by training.
00:04In 2002 he graduated from the Moscow
00:07Institute of Physics and Technology.
00:09And by the way, I think that we heard
00:12that before where he obtained same place
00:17where he obtained his BS&amp;MS degrees.
00:20He then conducted graduate studies
00:22in biology and in 2008 received
00:24his PhD in computational biology
00:27from Northeastern University.
00:29In Boston, after working as a scientist
00:31at Yale University for almost six years,
00:33he opened his laboratory at
00:35the Mayo Clinic in Rochester,
00:37and his laboratory is purely
00:39analytical and studies variations
00:41in mutation in human cells and how
00:44they can affect health and disease.
00:46He was a member of the 1000 Genomes
00:48Project and then Code Consortium
00:50and was one of the leaders of the
00:53brain Somatic Mosaicism Network,
00:55and he's currently a member of the 2nd.
00:59All right, I'm actually a second scientist
01:01talking here who is not like organic person,
01:04so I'm more like studying genomics,
01:07but I hope I could do some and you
01:10will see how we can contribute to
01:11general knowledge from different areas
01:13and from different angles of studies.
01:15So as was discussed today,
01:17Organo is a quite powerful system.
01:19So you can do them for many organs and
01:22they actually give a promise to not only to
01:25recreate the organs but also use it as a.
01:28As a model where you can study development
01:31and different phenomenons in human.
01:34So one of the thing in organoids which
01:40didn't was not highlighted that much
01:42today is that organoids allow you to
01:45actually do genetic manipulation in
01:47the in the system and then see which
01:49effect is going to give a downstream
01:52on development or organ growth.
01:54And typically how it's done if you
01:56have like some kind of cell car,
01:58so typically it's IPS line so you can
02:00CRISPR edit it and introduce edit.
02:02But since the CRISPR is not 100% efficient,
02:06not all of the cells will have the edit,
02:07so some of them will, some of them won't.
02:09So typically what people do is
02:11a clonal selection,
02:11select a pure edited colony and what
02:16what can happen during this process
02:17which is not very obvious all the time.
02:20Is that actually our cells have mutation.
02:23So you can have in the initial culture
02:24you can have cells with a mutation
02:26and then on top of what you have cells
02:28which can be edited to unedited.
02:29And when you do clonal selection
02:31you can get a colony which may have
02:33no edit but with mutation.
02:35Or it can have edit and mutation or
02:37it may have edit and no mutation.
02:39So it's actually creates heterogeneity
02:41during this process of clonal selection.
02:45Of course if it's only one single mutation,
02:46maybe you don't care,
02:47but then the question is like how
02:49much everything is different.
02:50And we collaborated with a few labs and
02:54there were a few labs at Mayo Clinic,
02:56at Flores Lab at Yale and in Oklahoma.
03:01So we collected data for those experiments.
03:05We conducted whole genome sequence
03:06and of the clones and we just simply
03:08compared them and ask a simple questions
03:10how much is the clones which are being
03:12conducted analysis how much is a similar,
03:14it turn out that pretty much
03:16they're all different.
03:17So there is nothing which is similar.
03:19And for example as it is shown here,
03:22so there are three clones which are derived
03:24after CRISPR experiments and they're
03:25different here in the copy number variation.
03:27So this is the coverage across whole genome,
03:30across genome.
03:30And then you can see that here the coverage
03:33drops which is means heterozygous deletion,
03:35one of it was deleted,
03:36so in this line it was deleted and
03:38this line is completely normal.
03:40So, and this is quite large,
03:41this is kilobases,
03:43multiple kilobases that's that
03:44has been deleted
03:45in other experiments it's
03:46pretty much just say.
03:48So here is extremely large deletion which
03:50is manifested only in one of the lines.
03:52Here there is a small deletion and then
03:55also manifested and and the same here.
03:58All of this variations which are shown here,
04:00they actually were present in the
04:02initial culture in the initial
04:04self population that you use.
04:07So and if you can do exactly.
04:09If you can do similar analysis for
04:11the point mutation and you discover
04:13them so you can see hundreds of them
04:15uniquely in each clone and IPS line.
04:17So this effect is manifested. Why?
04:19Because we actually do a single cell clone,
04:22so we now there is this bottleneck
04:24during analysis when you analyze
04:26single cell or isolate single cell.
04:28So a simple conclusion from this
04:31kind of various experiments that
04:34clone and conculturing actually.
04:36Results to in non isogenic lines that
04:38the promise of doing CRISPR that
04:40you can actually can get perfectly
04:42isogenic and only one mutation that
04:44you introduced will be affecting
04:46differences in your analysis.
04:48However the step of cloning results
04:50in differences and people have
04:52to take this into account.
04:56But I think more, maybe excitingly
04:59importantly for us to understand that
05:02actually all cells are different.
05:03So no two cells even
05:05from the same individual,
05:06from the same culture as the same.
05:08So this example I gave you in culture,
05:10but pretty much the same is true for
05:13cells in our body, in every person.
05:15So all cells are different.
05:17And this is called somatic mosaicism.
05:20Somatic mosaicism gained a lot
05:22of attention recent years.
05:24Because we had a boost in technology,
05:27everyone knows of you development
05:29of sequencing technologies and also
05:31there were approaches to study
05:33mutation to the single cell level.
05:35However, the ideas of the somatic
05:36mosaic and they actually quite old.
05:38The first mentioning was in 2004.
05:40The conceptual idea was expressed
05:43in 2040 that first time it was
05:46published phrase somatic mosaic
05:48in scientific literature.
05:49In in 1945 that was the first
05:52time we expressed ideas that this
05:54can happen in a human.
05:56So I want to share with you
05:59today talk about two projects,
06:01one of them related to development
06:04and the other one related to agent.
06:08So what does the semantic Moses
06:12mean development.
06:13We as a mammal,
06:14so we developed from a single cell and
06:17we follow very defined developmental path.
06:19So fertilize deck first produce a
06:22blastosis and blastosis differentiate
06:23inter in and outer cell mass.
06:25Then inner cell mass differentiate
06:27into 3 drum layers and outer cell
06:29mass mostly produces placenta.
06:31And then there is a cell migration
06:34intermixing and making organs.
06:36If we look at this process in a little
06:38bit more details this is how it looks like.
06:40So this is a fertilize deck and
06:43this is differentiation pass which
06:45happens to different tissues so.
06:47For example,
06:47there is a blood tissue eventually developed,
06:50so brain tissue and any other
06:53tissues and and and cell commitments.
06:56So this process is this picture
07:00reflects your development,
07:02how this happened and differentiation.
07:04However,
07:05together and in parallel with the
07:08differentiation another thing happens,
07:10cells acquire mutation,
07:11so they divide,
07:12they acquire mutation and they also
07:15differentiate.
07:15So these two processes are coupled.
07:18And in fact mutations can happen
07:21at a different stage of development.
07:23If they happened really late,
07:24they may may be present only in brain and
07:26not present in the blood or vice versa.
07:28So a lot of analysis that currently being
07:31conducted in in medicine and genetics,
07:34we simply take blood,
07:35analyze it and and asking a question
07:37can we see something there.
07:39But doesn't necessarily guarantee you
07:40that mutation whatever you see in
07:42the blood can be found in the brain.
07:43So this is 1 aspect why it's
07:45important to study mosaicism.
07:47And 2nd aspect is,
07:48since these two processes are
07:50actually coupled development
07:52and acquisition of mutations,
07:54you can ask a question can we use one
07:57process like let's say mutations to
08:00infer something about the development.
08:02So specifically you can think
08:04of the following idea.
08:05So in mice when we do lineage
08:07tracing and try
08:08to define cell ancestry,
08:09so people typically use
08:11genetically modified mice, so.
08:13What is it shown here?
08:14So this is the first cell that I got
08:17is start dividing and at some point
08:20you can activate expression of let's
08:22say green fluorescent protein and like
08:24cartoonishly you will get such a result.
08:26Like you can usually see that
08:28there is some area in the mouse
08:30which is have some abnormal color.
08:32So obviously we cannot do that in human.
08:35In human is basically not permissible
08:37to make a genetic modification.
08:40However, you can use this idea of mutation
08:42because once the mutation happen in a cell,
08:44all progenies of a cell
08:45will inherit that mutation.
08:46And we know that majority,
08:48absolute majority of the mutation
08:49not going to do anything.
08:50So they're just purely marks of development.
08:53So like as it's shown here,
08:54so let's say you have a First Division,
08:56you will have mutation A1A2,
08:58then you have next division,
08:59let's say you have beta mutation,
09:01you have gamma.
09:01And then let's say we will take this cell,
09:04it will have three mutations.
09:05So all of the Progenies will inherit
09:07this unique combination is going
09:08to be like a bar code for a cell,
09:09like 3 unique mutations.
09:11And then in the adult person given that
09:15if you can analyze multiple cells,
09:17you can actually infer and see it and
09:20reconstruct ancestor of the cells.
09:21So we try to capitalize on this idea
09:25and study what can we and try to see
09:28what we can say about development.
09:30So specifically we did the following.
09:31So we took a person.
09:34So we did multiple skin biopsies
09:37from a person from different areas we
09:40extracted fibroblast cells and from
09:43fibroblast cells we derived IPS lines.
09:45So here we leverage the property
09:48of IPS lines that they are clonal.
09:51So every IPS line that you derive it
09:53will be a perfect clone of one single cell.
09:56So you can think of IPS line
09:58as basically a single cell.
10:00Which we analyze and then we discovered
10:03mutation in this single cells and
10:06by sharing of the mutation we can
10:09reconstruct ancest is 3 of the cells.
10:11So if they if they share mutation
10:12means they had a common ancestor,
10:14if they don't they don't.
10:15And then like you do simple phylogeny,
10:18I mean in fact it's not was the
10:20simple but I'm not going to go
10:22into technical details.
10:23So what did we find at the end?
10:25So this is a patient with
10:27a Tourette syndrome.
10:28So here so each branch basically
10:30represents your single cell and these
10:32three represents you their ancestry
10:34that we reconstructed from the mutation.
10:37So this is in our model.
10:39This is as I got which divided
10:41into 2 branches and mutations are
10:43shown by letters so ABCD.
10:45And we used Latin letters for this
10:48branch and brick letters for this branch.
10:50So OK so we reconstructed this.
10:52Is there anything interesting?
10:54So this bars over here show
10:56you the frequency of this mutations.
10:58In cells from different tissues.
11:00So red is blood,
11:01blue is saliva and yellow is urine.
11:04So what we see right away is that this for
11:07mutations which marks this entire branch,
11:09they're present in 9180 and 69% of cells
11:13and cells from this accordingly they will
11:17be at much smaller fraction, 2 to 10%.
11:20So right away we notice that
11:22there is a huge asymmetry.
11:24SO2 blastomers, they already not
11:26contributing equally to the adult body.
11:29So that was really surprising to us.
11:32And this asymmetry to some extent
11:34projected to the following divisions,
11:36so divisions.
11:37So we thought maybe it's just something.
11:40So we call this dominant lineage and
11:42then we call this recessive lineage.
11:44So we thought maybe this is
11:46something related to a patient,
11:47maybe it's maybe overall not a big deal.
11:50So we actually recruited another just
11:52healthy individual and did pretty much the
11:55same analysis and found identical result.
11:57So we reconstructed this lineages
12:00and one branch was dominating and
12:02the other one was really recessive.
12:06There was some some slight differences
12:08that there was a lot of mutation here.
12:11But however,
12:11the overall effect was the same.
12:13So we published that study earlier in 2021.
12:18Yeah,
12:18it was early 2021 and this result was
12:21partially replicated by three follow
12:23up studies just few months later.
12:25Why I say partially is that as a
12:28studies using slightly different
12:30approach obtains the same result but
12:31not in all of the individuals so.
12:35Roughly half of the individuals
12:37where such linear construction was
12:39conducted have this strong asymmetry,
12:41and another half doesn't have
12:42the strong asymmetry,
12:43and we don't know exactly what it means.
12:45It could be that this is a
12:46finistic variation between people,
12:48or it could be that we simply
12:50in those individuals.
12:50We haven't found it yet.
12:51This is a symmetry.
12:52You just need to sample more cells.
12:54Obviously,
12:54the more cells you sample from a person,
12:56the more power you have to
12:58find such a symmetries.
12:59So.
13:00But what is the interpretation
13:02of this asymmetry?
13:04Technically speaking,
13:04you can think about two effect effect #1.
13:08So if the blastomer divides and creates 2
13:12lineages initially is I got this divide.
13:15So it could be that one lineage
13:19somehow has an advantage.
13:20Maybe it proliferates faster or maybe
13:22the other lineage proliferates slower.
13:24So this is what happens in reality.
13:26But when we construct.
13:28When we make a retrospective reconstruction,
13:30because we actually don't
13:32have information about time,
13:33we only have information about ancestry,
13:35it looks to us that there was
13:37only two divisions and we assume
13:38there was scenario like this.
13:39However,
13:40in fact may have been this case
13:42where for three divisions in this
13:43lineage there was only two of this.
13:45This is just cartoonish,
13:46but showing you that there is some asymmetry,
13:48like internal properties,
13:50that's possible explanation number one.
13:53Another one maybe that.
13:57This lineage which we call recessive
14:00cells of this lineage have some
14:02intrinsic property where which
14:03for example leads them to higher apoptosis.
14:06So so this cells.
14:08So in reality they this cell divided
14:10into 2 but that cell didn't survive
14:12and maybe this cell didn't survive.
14:15So then every time we divide
14:17it acquires mutation and start
14:19with some mutation and one
14:21retrospectively we reconstruct.
14:22Then it looks to us that it
14:24was one single mutation,
14:25was one single division with many mutation.
14:27But in fact it was not
14:32here that the cells which
14:34we put with a question mark,
14:36they don't necessarily may up up
14:38toes or maybe happen something else.
14:41Particularly they may go to placenta
14:42and since we are not something
14:44placenta from the living individuals,
14:46we are not able to observe it.
14:49And this possibility is quite
14:50exciting because it may suggest
14:52the following scenario.
14:53At this point we cannot prove it.
14:55But this is on the level of hypothesis
14:57that two cells from the very beginning
15:00could be already unequal in their fate,
15:02and one of them, one of the cell,
15:04one of the blastomer from the beginning,
15:06already knows that it will
15:07mostly make a placenta,
15:08and the other blastomer will know
15:10that it mostly will make a dull body.
15:11And then during this development,
15:13so this will organize,
15:15the development will go this way.
15:18And that's why when we are
15:21analyzing leaving individual,
15:25one of the lineage will be
15:27underrepresented and the other
15:28lineage will be overrepresented. So
15:33that's that's about development.
15:35And then let me slightly shift gears
15:37and go about somatic mutation.
15:39What can we infer about aging?
15:44This project was conducted in frame
15:46of brain somatic Moses network.
15:48This network was established by
15:51NIMH National Institute of Mental
15:53Health with the aim of understanding
15:56how much somatic mutations can
15:58contribute to neurological diseases.
16:01So there were six diseases targeted
16:03and the idea will discover all of the
16:06type of somatic mutations in the brain.
16:08But today I will mostly focus
16:10on the three diseases.
16:12Simply because for this for brains
16:13with these diseases we had whole genome
16:15sequencing data and for others we don't.
16:17And we actually like really whole
16:19genome data because we can analyze
16:21entire genome and it gives us a lot
16:24of information and also I will be
16:26focusing mostly on point mutations.
16:29So when we started this project in 2015,
16:33I believe at that time the question
16:36number one was can we discover
16:38some ASIC mutations or not.
16:40The approach in this project was
16:42different from that lineage tracing
16:44I just talked about because here we
16:46did not analyze individual cells.
16:48So here we took a brain and
16:50analyzed bulk of the brain.
16:51And last question,
16:52can we see some mutations there?
16:54So obviously we cannot see
16:56mutations which are extremely rare.
16:58We can only see relatively frequent
17:00mutations and the question was
17:02how do we actually discover them.
17:03So just briefly what we did,
17:06we simulated somatic mutations by mixing.
17:10Different samples with different genomes.
17:12So this way variations which I inherited
17:17will be present as a frequency
17:19which is not in 100% of cells,
17:21just a fraction of cells,
17:22and this thereby will simulate
17:24somatic mutations.
17:25The Long story short,
17:27basically nonexistent methods allow
17:28us to do comprehensive and accurate
17:31discovery of somatic mutations at that time,
17:34so we actually had to develop
17:36some new approaches.
17:37So we set up a big experiments,
17:39so we did that in the in the
17:42consortium and we generated tons
17:44of call predictions from many labs
17:46and we made elaborate efforts to
17:48validate as many sites as possible.
17:50So here basically showing you
17:51a validation for 400 sites.
17:53So every column corresponds to A1
17:55mutation and every row corresponds
17:57to a different type of data we
17:59collected for this mutation.
18:00So we for this call,
18:02so we did the multiple sequencing
18:04of the brain regions,
18:05we did multiple replicas,
18:07we did multiple validation and so on.
18:09So as you can see from this experiment,
18:12majority of the things we find we
18:14eventually deemed false positive
18:15and only small fractions were
18:17true real thematic mutations.
18:19And based on this experiment
18:21I'm not going to overload you
18:23with the technical details,
18:24but at the end we came up with a
18:27way how we can discover mutations.
18:29From just sequencing brain
18:31bulk at high coverage,
18:33so few of the kind of snippets of what does
18:37the what was the key in this discovery.
18:40So we had to adjust existing methods
18:43to become more sensitive to find
18:45mutations at the lower frequency.
18:48So we had to use accessibility marks.
18:50What does it mean? Our genome has a
18:52lot of repeats so we actually had
18:54to throw about 25% of the genome.
18:57Where reads we are using short reads I'm not
19:00able to reliably call somatic mutations.
19:03And then what was important we have
19:07to use panel of normal mask which
19:10is we see for reproducible errors
19:12across multiple samples.
19:14So this was a quick three key steps
19:16and eventually one of the lab in the
19:18brain somatic mosaicism developed.
19:23Artificial intelligence approach,
19:24a machine learning approach to
19:26make a final filtering step.
19:28So at the end, so this is performance
19:31assessment of our performance,
19:33how we see it for our developed approach.
19:37So now mutations between frequency of
19:40about 20% of frequency and about 2%
19:42of frequency can be discovered with
19:45relatively high sensitivity of about 70%.
19:48So we are missing a lot.
19:49And we're missing a lot because we
19:51have to throw out this repetitive
19:52region of the genome.
19:53But we can guarantee that what
19:55we find in in that 75% of the
19:59genome is actually quite accurate.
20:01Accuracy is about so false.
20:04Positive rate is around 5%.
20:07So what do we find?
20:09So we analyzed 131 brains.
20:12There were normal brains,
20:14brains with stress syndrome,
20:15ASD and schizophrenia.
20:16So those are shown here.
20:18And since we find only relatively
20:21high frequency mutations,
20:22so we don't see all like
20:24hundreds and thousands of them,
20:26we see relatively small number
20:28between 20 to 40 per brain.
20:30And this number is quite consistent
20:32across all of the cohorts across
20:34normal and the disease cohort.
20:36So that was kind of in some sense expected.
20:40So however what was unexpected
20:42that there are few brains.
20:44Where mutation burden is much much high.
20:46So please pay attention that
20:48this is log scale.
20:49So like here the count of mutation is roughly
20:522000 that we detected in A1 single brain.
20:55So we call this brain hypermutables.
20:57Initially we saw that maybe something
20:59wrong with the data and we did all
21:02the possible checks so here it just.
21:04Showing you that they are not outlined
21:06in terms of phasing it to the haplotypes.
21:08They are not outlined in
21:09terms of mutation spectrum.
21:10They are not outlined in terms of like
21:12as a more refined mutation spectrum,
21:14everything.
21:14What we could actually point out
21:16that these are real mutation.
21:18So these brains in fact have much
21:21higher count of detective mutation.
21:24So interestingly enough that
21:25if we now put age together with
21:28the count of this mutation,
21:30we see a positive correlation.
21:32So here what we saw,
21:33we show the age of the brain and
21:36here is a mutation count that we
21:38detected and we see the positive
21:41increase that aging brains have.
21:43All the brains have higher fraction of
21:47brains with increased mutation count,
21:50so obvious association.
21:52And then the question we was
21:54of course how do we explain it?
21:55What does it mean?
21:58So two brains here which are circled.
22:02They actually had a one missions
22:04mutation in the NRAS gene and then Ras
22:06is a quite well known cancer driver gene.
22:08So it it was damage and mutation over there.
22:12So this brain had a mutation in the anthorgy.
22:15So this actually also quite famous
22:18cancer driver gene and this one the
22:20brain with the highest mutation count
22:22actually if you look at the copy
22:24number profile across the genome.
22:26So this is cortex,
22:27this is hippocampus and this is chromosome.
22:30So what you can see that cortex
22:32is absolutely normal
22:33and then hippocampus.
22:35So there is a gain of chromosome 7.
22:37It's mosaic, so it's not
22:39present in all of the sand,
22:41all of the cells and loss of chromosome 10.
22:43So it's also mosaic.
22:46This is actually gain seven loss 10.
22:48It's a classical example of glioblastoma.
22:50So this is how it's often
22:53time being diagnosed.
22:54When we talked to clinicians in the Mayo
22:55Clinic, they said well if we see this,
22:58it's not enough to say it's,
23:00it's we diagnose it as a cancer.
23:02However, if you go to the third
23:04promoter and find mutations there,
23:05then we'll we'll just call it a cancer case.
23:08We actually went to the third promoter.
23:10So it was masked out as a repetitive regions.
23:13So that's why initially we
23:14didn't see this mutation.
23:16However, if we actually go to that
23:17position where this mutation happens
23:19to actually see it, so it's there.
23:21So this person has all the
23:23hallmarks of cancer.
23:24Gain of seven,
23:24loss of 10 and 3rd promote the mutation.
23:27So we think that it's probably
23:29undiagnosed case in the person or
23:31maybe misdiagnosed because that the
23:33official diagnosis is schizophrenia.
23:36So that's already we have multiple
23:39evidence pointing to the cancer driver
23:42genes and actually if you could be
23:44late all of the mutations we have.
23:47So these are all mutations
23:48in the cancer driver genes.
23:50Typically we see them in the
23:51brains which are hyper mutable.
23:53So there are few brains which
23:54are non hyper mutable.
23:55Sometimes we see mutations in
23:57the cancer driver genes but
23:58typically there and this is a
24:01statistically significant enrichment.
24:02So now we have association with age
24:04and then we have association with the
24:06mutations in the cancer driver genes.
24:08So we think that likely what we
24:10observe is the following scenario.
24:12In a normal brain we have
24:14a diversity of clones.
24:15And then if we randomly pick up one clone,
24:17this is just a two dish example.
24:18Mutation in that clone will be
24:20at relatively low frequency and
24:22we are not going to be able to
24:24see them by sequence and bulk.
24:26However,
24:26if you gain some proliferative advantage
24:29this clone start to expand and
24:32clonal diversity significantly drops.
24:34Then mutations present in one
24:36clone in this green line they
24:39will rise to the high frequency
24:41and then by sequence and bulk.
24:44We are able to detect it.
24:45So the question is what actually
24:47cell type expands because in brand
24:50we have multitude of cell types and
24:52the question is what happens there?
24:54So that's one case which I already
24:57went over it once with this game
25:00seven and delete loss of 10 so this.
25:06In hippocampus we see this unemployed is
25:08and all of these mutations that we detect,
25:11we also find it in hippocampus.
25:12So this plot is showing you.
25:14So every column corresponds to a mutation.
25:17There are two regions and color
25:20represents your frequency.
25:21So these mutations are shared
25:23between cortex and hippocampus and
25:25that probably developmental origin.
25:26So these are early one and these
25:28are specific to hippocampus,
25:30most likely reflecting this clonal expansion.
25:34So we think that in this case
25:36it's probably glial cells again.
25:38So this is our strong hypothesis
25:40which need to be proven.
25:42So hypothesis #1,
25:43if you expect inspect mutations which
25:47happened in cancer driver genes,
25:49then you will notice genes like DN,
25:51MT3A, tattoo ID H2 and these genes
25:55have been before implicated into
25:58their hematobological malignancy,
26:01basically blood cancers.
26:03So another possibility from this
26:05analysis we think that it could be
26:08hematopaetic cells so which are
26:10expanding in the blood but somehow
26:12they penetrate into the brain and we
26:15we are able to detect it by by this
26:18analysis particularly we know that
26:20aging brains actually they are more
26:23prone to blood brain leakage and as
26:25the last hypothesis that we have is.
26:30Comes from the brain NC7,
26:31where we already had mutation in the
26:34NRAS gene and so the plot is the same here.
26:37So columns represent mutation.
26:38But now we have different brain
26:40regions and different cell fractions.
26:42And what we saw that mutations present
26:45everywhere in the cortex bulk,
26:47they're present everywhere in stratum bulk,
26:49but also they were present in the
26:51stratum interneural fractions that we
26:53isolated from this region and actually
26:55the quite high frequency just to prove that.
26:58What we did, we isolated cells
27:00from this stratum interneurons,
27:02single cells, so isolated 16 of them.
27:05And we genotype mutations in the
27:07single cells, so coverage was not high.
27:10So our genotype inefficiency was roughly 50%.
27:13So half of the mutation,
27:14we can see half of them, we're not.
27:15But you quite clearly can see
27:17that in the 8 cells,
27:19they clearly have almost all of the
27:21mutations in this stratum in the neurons
27:24from from this that we discovered in bulk.
27:27And half of the cells you
27:28don't have anything.
27:29So it's it's an ultimate proof that first
27:32of all that this is quantal expansion
27:34in a single some all mutations present
27:36in a single cell in a single lineage.
27:39And then it strongly suggests that stratum
27:41interneurons may have expanded and and
27:43then lead to this quantal expansion.
27:45So to summarize, so,
27:47so this next hypothesis,
27:49this is interneurons just to summarize
27:51that what we think may have happened,
27:53this is 3 hypothesis.
27:54One possible origin of this quantum
27:57expansion is during neurogenesis,
27:59so where interneural generated
28:01in one specific area of the brain
28:03and then migrate and populate the
28:06brain during development.
28:07That's hypothesis #1 and then when we
28:09do sequencing so we are able to see it.
28:11Next hypothesis is that this is gluogenic
28:14origin probably happened late in life,
28:15so at some time during lifetime
28:18cell acquire proliferative advantage
28:21and then repopulate brain area.
28:23Is going to be localized in particular
28:25region and the other one that
28:27there is a hematopaetic origin that
28:29actually clonal expansion happens
28:31in the brain and then they penetrate
28:34sorry happens in the blood and then
28:37the cells somehow
28:38penetrate brain and that will
28:40be excited to see what they
28:42may do if this is the case.
28:44So the last thing what we did is
28:48there some functional relevance
28:50of the discovered mutations.
28:53Sorry, this is my own timer,
28:56I have one minute. So is the question was
29:02do they do do they have any
29:04relevance to diseases that we study?
29:05This neuro, psych,
29:07psychiatric when we did the standard
29:09analysis using overlap with exams and
29:12predicting possible functional variants.
29:13So we didn't really see anything specific.
29:15The only signal we can find is
29:17when we try to predict mutations,
29:19affected enhancers.
29:20So what we saw the over representation
29:23of mutations which create binding
29:25sites particularly for the
29:27transcription factor like Mace 1,
29:29Mace 2, Mace 3 and we saw that in the
29:33ASD this transcription factors they
29:35actually very important in development.
29:37So our hypothesis that semantic
29:39mutations can contribute to the
29:42ASD phenotype during development by
29:44affecting regulation or dysregulation
29:47of development and.
29:50By affecting transcription factor binding.
29:52All right. So my conclusions,
29:55conclusion slide about all this.
29:57So mutations actually very frequent.
30:00So no,
30:01no two cells in our basically in our
30:04body will ever have the same genome,
30:06probably from the very beginning.
30:10And we can use this to track cell lineages.
30:12So we already see the effects
30:14from the very beginning,
30:16this created lineages are not.
30:20Symmetrical and maybe I
30:21didn't highlight it enough,
30:23I kind of bolted it here.
30:24So this approach for lineage
30:26tracing that we have, it's,
30:28it was done for the living individual.
30:31So basically it's a way like every
30:33everyone sitting in the room we
30:35can conduct the same study and
30:37reconstruct your personalized history.
30:38What happened to you when you are really,
30:40really little like literally very small and.
30:47In brain we see between 20 to 60 mutations
30:50and this effect of hyper vitability
30:52is there are three hypothesis which
30:53we are in the process of trying to
30:55figure out which one of them is true.
30:57And actually maybe all three of
30:58them are true in different brains.
31:00And let me conclude by acknowledging
31:03my colleagues.
31:04Of course a lot of work is done with
31:06as I said with a Flora Walk arena and
31:09my lab is purely analytical and we
31:12rely on collaboration with other labs
31:15to generate data and analyze it and.
31:17Of course, big signs goes to BSMN
31:19because it was a big project with
31:21enrollment of more than a dozen lab.
31:23And thank you very much for your attention.