Using Stem Cells to Explore Genetics Underlying Brain Disease

April 30, 2021

ID6553

To CiteDCA Citation Guide

00:00My name is Karen Marshall an I'm
00:02a PhD student in the genetics
00:04Department and Doctor Bloom malicious
00:06lab and it's my privilege to
00:08introduce our next speaker doctor,
00:10Kristin Byrne and Doctor Brown,
00:12and is a new faculty member in the
00:14Department of Psychiatry at the
00:16Yale School of Medicine and most
00:18recently was an associate professor
00:20in the Pamela Sklar division of
00:22Psychiatric Genomics at Mount Sinai.
00:24Her research integrates stem cell
00:26based approaches with CRISPR mediated
00:28genomic engineering strategies
00:28in order to study the impact of
00:31patient specific variance across.
00:32In between the cell types of the brain,
00:35the goal of her research is to uncover
00:37the convergence and synergy arising
00:39from the complex interplay of the many
00:41risk variants linked to brain disease,
00:43and her work is funded by the NIH,
00:46the New York Stem Cell Foundation,
00:48the Brain Research Foundation,
00:49and the brain and Behavior
00:50Research Foundation.
00:51Thank you doctor Brennan.
00:54Hey, thank you so much. Anne Anne.
00:57Firstly I want to thank Dean Brown uh,
01:00an IRA for organizing
01:01this really great event.
01:03I I really enjoyed myself so much.
01:06So to begin the only conflict
01:08I have to declare is that my
01:10husband works at our Venice.
01:12My talk today is about now validating
01:14all of the genetic findings that
01:17everyone's been talking about today.
01:19So using stem cells to explore the
01:22genetics underlying brain disease.
01:26And a lot of what I'm going to talk about
01:29today is in the context of schizophrenia,
01:32not because there's anything uniquely suited
01:34from of my tools to study this disease,
01:37but because it's a highly complex
01:40genetic disorder for which
01:42there still really are no good.
01:45Here's how it is.
01:46Extremely common and I'm showing you here.
01:48One of the most recent descriptions
01:50of the genetics of this disease
01:52and what I want you to see is just
01:54how there's almost 300 of them is
01:56about 250 common variants here,
01:58with really small effects of one
02:00or two percent increased risk,
02:01and then a few dozen rare variants
02:03that are either protein truncating
02:05variants or copy number variants.
02:07And most importantly,
02:09genetics here is not diagnostic.
02:11There are strong and significant group
02:13differences between patients and controls,
02:15but we can't yet harness the power
02:17of all this genic information to make
02:20meaningful insights for their patients.
02:22And so the goal then is to derive
02:25revised polygenic risk scores
02:27that improve diagnostics,
02:28predict clinical trajectories,
02:29and ultimately allow us to discover
02:32novel therapeutic targets.
02:34And again,
02:34here is the the common variant
02:36restructure for schizophrenia.
02:38The last time it was published in two
02:402018 is about 145 significant loci
02:43here over the genome wide significant line.
02:45But the question then here is which of
02:48these variants are causal disease variants?
02:51There's many sniffs and linkages.
02:52Equilibrium at each location.
02:54What are their target genes?
02:56Were the proximal target genes and
02:58there are often more than one.
03:00What are the distal target genes
03:03that are regulated?
03:04Far distances from these risk variants.
03:06Are there any context dependent
03:08effects and an altogether?
03:09Does this answer the question of
03:11variable penetrance in this order?
03:13And of course,
03:14because all of us carry dozens of risk areas.
03:17But the patients carry dozens
03:19and dozens more.
03:20As this really important question
03:22of how to experience interact in
03:24some because their phenotypic
03:26effects only occur in aggregate.
03:28Now and so these are the kind
03:30of questions my lab is asking.
03:32How do risk area risk variant
03:34effects vary with polygenic risk.
03:36Innocent project led by Christina
03:37Brandt across no development led by
03:40Liz La Marca between cell types.
03:41This is Michael Fernando and Sam
03:43Powell across sexes and open project
03:45that we haven't really started yet
03:47and Karina so and and Kayla towns.
03:49They're looking at how they are
03:51impacted by the environment,
03:53particularly stress.
03:55So again,
03:55turning back to this gys and
03:57asking how do we begin to validate
04:00both the variant effects and how
04:02their modulated in a dish?
04:03And we started here with this
04:06particular snip on chromosome 16,
04:07and I want to tell you why we picked
04:10it because something unique a curd at
04:12this location that was seen nowhere else,
04:15and there's 145 loci that are
04:17above genome wide significance.
04:19At this location,
04:20the same snip that was most significant
04:23for schizophrenia risk in the G
04:24was that's the Y axis was also the
04:27same snip that was most significant
04:29for regulating the expression of
04:31the nearby coding gene furin in
04:33a brain post mortem collection.
04:34And so here what we have is a single
04:37putative causal snip linked to disease
04:39risk and gene expression of a target gene.
04:42The next best examples.
04:43I look more like what we see here.
04:46It's nap 91 with a cluster of 20 or 30 jeans.
04:49That are associated with both disease
04:51risk ANAN target gene expression,
04:53but we can't really figure out which
04:54one or many of these top snips are
04:57responsible for the disease effect.
04:58It could be that there's one
05:00putative causal snip in that cluster
05:02that we haven't resolved yet,
05:03and it could be that each of
05:05these many snypes is confirming
05:07part of the risk at that locus,
05:09and so this is a project that was led by a
05:12former postdoc in the lab and then shroud,
05:15and I will head up forever say that
05:17she was the bravest postdoc to join
05:19the lab because the flip side of that.
05:22You know beautiful, clean data for fear.
05:24And is this analysis here.
05:25This is now the post mortem gene expression.
05:28The few and by individual brain sample.
05:30So there's about 600 brains in
05:32this collection.
05:32But what you'll see is these
05:34error bars are incredibly huge.
05:36There's a very,
05:37very significant effect of genotype at Rs.
05:394702, which is in the three prime UTR.
05:41The fear in gene.
05:42But there's a lot of variation
05:44between brains.
05:45That's because every brand came
05:47from a different person,
05:48and these people married at
05:49many many genotype genome wide,
05:51not just the Rs 4702.
05:53Some of these people had schizophrenia
05:55or some did not.
05:56They had different histories
05:57of antipsychotic treatment,
05:58drug and alcohol abuse.
05:59They were different sexes.
06:01They do have different causes in a
06:02different ages and different postmortem
06:04intervals before they were examined.
06:06And so our hypothesis was really
06:08that if we did the experiment in the
06:10dish in the same genetic background,
06:12we controlled everything and
06:14generated the neurons in parallel in
06:16neighboring wells or hypothesis was
06:17that we would see the same effect
06:19size but the variation between
06:21samples will be dramatically reduced.
06:23And so with that,
06:24Nadine took out to be set out to
06:26begin this editing of a single non
06:28coding snippet actually turned out
06:29to be much harder than we expected.
06:31Took about two years to achieve
06:33or what you can see here is that a
06:35single snip in the three prime UTR,
06:37the Fusion gene again have been
06:40edited successfully.
06:40And when she generated generated neurons
06:42from these isagenix mashta stem cell lines,
06:45just as we predicted,
06:46the team was able to see a significant
06:49decrease in fear and expression.
06:51Now,
06:51in the two years that it took to
06:54engineer these edits is actually
06:56discovered that this up three prime
06:58snip is in a Micron 8338 binding site.
07:01And so when you inhibit near 338,
07:03you can eliminate this EQ TL effect.
07:05Really,
07:06I think suggesting that you
07:07could see context specific
07:09effects, so here the ability
07:11of this noncoding snip.
07:12To regulate foreign expressions depended
07:14upon mere 338 expression in the cell.
07:17Moreover, changing nothing more in this
07:19noncoding snip was sufficient to decrease
07:22in the right length and alternate activity.
07:25Moving forward, this projects and picked
07:27up by Christina the last couple of years
07:29and over the course of this pandemic.
07:32Increasing evidence showed that fear and
07:34actually had a role in the entry of SARS,
07:36Co V2 the the fear and cleavage
07:38site is specific to SARS, Co.
07:40V2 not found in SARS, Co V1.
07:42It was hypothesize to be part of
07:44Wise are Scobie too is so much more
07:47infectious since we were sitting
07:48on these isagenix stem cell lines.
07:50Christina working together with
07:52Daisy Hoagland,
07:52invent Hanover's lab began to ask could
07:54we alter the susceptibility to infection?
07:56By changing just a single noncoding snip.
07:59So here she's pivoted and we're
08:01making lung alveolar cells in
08:02the lab and across two donors.
08:04She can show that those GG lung alveolar
08:07cells express less fear and and also
08:09that there are less susceptible to SARS,
08:11Co V2 infection.
08:12Think the visual here is really
08:14striking so you can see a protein
08:16in the SARS Co V2 genome here,
08:19labeled in red and those AA alveolar cells
08:21on the top are much more brightly infected,
08:24and those on the bottom we
08:26turn back to neurons.
08:27And again,
08:28as I told you previously,
08:29Gigi neurons express last year
08:30and but they're also less well
08:32infected by SARS Co V2.
08:33So I think this was a really fun direction
08:36to take the lab over the last year.
08:39I'll remind you,
08:40though,
08:40that many of these common variants
08:42don't only exert their effect
08:44on the closest neighbor gene.
08:46Our DNA is not packed into our
08:48nucleus in a straight line.
08:50Instead,
08:50it folds in a very organized
08:52way into the nucleus,
08:54in ourselves,
08:54and so a former MD PhD student lab for Sale
08:58Missouri and asked at the genome wide level,
09:01how were their cell type specific
09:03differences in chromatin folding,
09:04and how did this impact the target potential
09:07target genes of schizophrenia risk loci?
09:10So here he is,
09:11applying Heisey analysis in isagenix
09:13stem cell derived astrocytes and
09:15regenerate cells and neurons.
09:17So each of those schizophrenia
09:19risk loci have,
09:20you know,
09:20a number of target genes that are
09:23close by here from those 145 variants,
09:25he's calling 224 proxamol target genes,
09:27but an additional couple 100 target genes
09:30that occur in a cell type specific manner,
09:32and he was able to validate the
09:35impact of each loci at a distance.
09:37Here, for example,
09:38the product here in Alpha cluster.
09:40So here we have for schizophrenia risk
09:42NIPS that about 93 kilobases away from
09:45this product here in Alpha cluster,
09:47he applied crisper.
09:48To delete these nips and was able to
09:51show in the direction that deleting
09:53these risk nips impacted the expression
09:55of a very distal target gene.
09:58Here, PC DH, EA 10.
10:00Come. Now, of course, these risk
10:03variants don't occur in isolation.
10:05We all inherit them in combination,
10:08and so the questions that I think are
10:11most important to ask moving forward is
10:14how to risk variants some so Mail in in
10:17my group is looking at the convergence
10:19of schizophrenia and autism risk genes,
10:21particularly around synaptic biology,
10:23and epigenetics patrons been looking
10:25at networks of genes that change
10:27together in a coordinated fashion
10:29focused here on Alzheimer's disease
10:31and their microglia and neurons.
10:33Drivers of disease.
10:34Anne Michaels been looking
10:35at additive affect.
10:37So how does schizophrenia risk variance
10:39add within and across pathways?
10:42Here's some of the work that
10:44Adrian has a recently wrapped up.
10:46This was a collaboration with Mingwei,
10:48weighing in Bendzans lab.
10:49It began with a post mortem
10:51analysis of through all 350
10:53plus frames for brain regions,
10:55each looking for jeans that were
10:57coexpressed together and predicting
10:58the drivers of those networks.
11:00And so one network here I'm pointing
11:02out is M 64 for that predicted
11:04causal driver Gene was 80P61A.
11:06Atron knocked it down in EPS derived
11:09neurons and here you can show see that.
11:12In these two independent knockdowns,
11:14one and two,
11:14there is reduced neural activity and by
11:17electrophysiology that's actually reduced.
11:19Excitability of these knock down
11:21neurons Mingwei predicted a drug
11:23that was thought to increase
11:25expression of 80 P 61 eight drugs.
11:27NCH 51 and he was able to show in a
11:30dose dependent manner that it did
11:33just that and finally that treatment
11:35of neurons with NCH 51 was sufficient
11:38to ameliorate some of these deficits,
11:41partially restoring.
11:42Neural activity in 80P61A knock
11:45down Ipps neurons.
11:47Coming back to schizophrenia here
11:49this is work that was largely
11:51led by Sacramento when he was a
11:53PhD student in the lab.
11:55Here we're going after a handful
11:57of of common risk variants.
11:59Associated genes SNAP 91 and T snare.
12:01Here,
12:02some RNA seek showing that by
12:03CRISPR activation or inhibition.
12:05We can upregulate or down
12:07regulate those jeans.
12:08The other jeans that are changing
12:10with think our downstream network
12:11effects of these preparations that
12:13are enriched for brain pathology
12:15and specifically synaptic function
12:16genes and hereby electrophysiology.
12:18So I was able to show that reciprocal
12:21changes in step 91 expression that are
12:24reciprocal changes in synaptic activity.
12:26So increasing SNAP 91 increased excitatory
12:29postsynaptic currents and decreasing
12:31it decreased synaptic activity.
12:32Of course,
12:33these jeans again,
12:34we wanted to ask how they impacted
12:36neurons in combination and so
12:38here stuck started the project
12:40and then the Dean finished it.
12:42This is a RNA seek experiment
12:44that started with single RNA seek
12:46for our four top schizophrenia
12:47common variant risk genes.
12:49Nap nanny One T snare,
12:51CLC and three in fear and an Indian
12:53took those single gene perturbation.
12:55Arnie seeks and computationally
12:56added them together,
12:57yielding an expected additive
12:59model of what she thought would
13:01happen if we did comma tutorial.
13:03Perturbation but in parallel we
13:05actually did this comma toryal
13:06probation and somebody was able to ask how
13:09well the model performed and will hold.
13:11The mark was actually pretty good.
13:13About 82% of the genes changed as expected,
13:15but seven of the percent of the
13:17jeans were more down than affected.
13:19An 11% more up than expected,
13:21and so they ended further into the day to ask
13:24what types of genes were more down or up.
13:27And so there's more down.
13:29Genes were actually enriched for
13:31all of the major neurotransmitter
13:32released pathways in the brain.
13:34And the more up genes are enriched for
13:36both the rare and the common variants
13:39linked to schizophrenia and bipolar risk.
13:41And so I think what we're saying.
13:43So what we're seeing here is that
13:46studying these risk genes one at a
13:48time gives us a lot of the story,
13:50but not all of the story.
13:52And if you want to fully understand
13:54the biological impact of manipulating,
13:56risk variance is really important
13:57that we do it in combination.
13:59Along those lines.
14:00One last story here,
14:02pivoting to some other rare variants
14:04linked to schizophrenia risk.
14:05So this was.
14:06The most recent copy variant analysis of
14:08deletions linked to schizophrenia risk.
14:10We're focusing here on this one at 2 P.
14:1416.3, which encompasses a single gene,
14:16neurexin,
14:16one that is inherited,
14:18non recurrent Lee or with varying
14:20boundaries between donors,
14:21and that impacts one of the most
14:24highly Alternatively spliced
14:25genes in the human genome.
14:27Here's a look by long range sequencing
14:29at the number of neurexin one isoforms
14:31seen across different regions of the brain,
14:34and in fact these differences in
14:35direction one splice rapid horrors are
14:37sufficient to distinguish different
14:39types of neurons in the brain.
14:41So this study was Alco led by Aaron
14:43Player Tia former PhD student
14:45in my lab and Shoujo,
14:46a former postdoc in my
14:48collaborator ********* Lab.
14:48So here we have a cohort of
14:50for a rare direction, 1 cases,
14:52two of them with deletions in the
14:54five prime region of the gene,
14:56including the promoter in the
14:57first two exons,
14:58and two of them with deletions in
15:00the three prime region of the gene,
15:02including the second,
15:03third and last from their second,
15:05third, and 4th from last exons.
15:08This is long range sequencing
15:09analysis were considering how
15:11the direction one isoforms vary
15:12between the cases and the controls,
15:14and the first thing that I want to point
15:17out about 50% of the isoforms are decreased.
15:20Inpatient neurons relative to controls,
15:22and so you can really
15:24visualize how these neurons,
15:25how I support a differin you can see in
15:28purple the differences in abundance,
15:30another about a third of the isoforms
15:32were detected in the controls,
15:34but not in the patient neurons at all.
15:37These were some of the lower
15:39abundance control isoforms.
15:40And most surprisingly,
15:41I think we had to find 31 unique
15:44mutant isoforms that we were did
15:46that we detected in a patient.
15:48Neurons formed by splicing around
15:50that three prime deletion that
15:52we never saw in control neurons that we
15:54never saw in the postmortem human brain.
15:57Aaron cloned and overexpressed
15:58some of these are most abundant
16:00wild type and mutant isoforms.
16:02She was able to show in control neurons
16:05here starting in that left most bar.
16:07This wild type control neuron activity
16:09that by knocking it down with.
16:11Four different mutant rexon,
16:13one isoforms you could decrease
16:16neural activity in control neurons.
16:19In those five prime cases,
16:20the ones that are not thought to
16:22express Newton isoforms should be able
16:24to rescue decreased normal activity by
16:26overexpressing even just One Direction.
16:27When I spoke at a time and in the hub,
16:30three prime cases, the ones that
16:32did overexpressed mutant isoforms,
16:33she was never able to rescue activity by
16:36over expression of wild type isoforms.
16:38To really, we think the phenotypes
16:39are occurring through two mechanisms.
16:41First,
16:41a loss of neurexin one dose in all cases,
16:44but then in the subset of patients on the
16:46additive effect of mutant isoforms activity.
16:48And this is something we're
16:50continuing to explore in the lab.
16:52And so with that I want to stop
16:54and thank everybody in the lab.
16:56This has been an extraordinary
16:58difficult year.
16:58And if it wasn't for all of their hard work,
17:01we really wouldn't have been able
17:03to keep these experiments rolling
17:04in the data I talked about today
17:06was really led by Christina Atron,
17:07Nadine Erin for Sean and Soak it
17:09in collaboration with our really
17:11our key collaborators.
17:12So I will thank you and turn
17:13the floor back to our moderator.