Organoid Modeling of Neuropsychiatric Disorders

March 30, 2023

ID9773

To CiteDCA Citation Guide

00:00So where I transition into next speaker,
00:02which is Flora Vacarina,
00:03I'm going to introduce her.
00:05So Flora Flora received the MD
00:08from the University of Padua in
00:11Italy and then she spent a few
00:13years and europharma call as a
00:15neuro pharmacology fellow at NIH,
00:17where she completed internship
00:18and residency in psychiatry.
00:19Then she completed internship and residency,
00:23residency in psychiatry at Yale and.
00:28Then she did a research fellowship
00:30in developmental genetic of
00:31the Yale Child Study Center,
00:33where she subsequently became an assistant
00:35associate and then a full professor.
00:38And in 2010 she was appointed as a
00:42Harris Harris professor as a child Study
00:44Center for Development of Neuroscience
00:46at Yale University School of Medicine.
00:48And since 2009 she has been the
00:50director of the program in the
00:53neurodevelopment and regression.
00:54And I'm very happy.
00:56So he's also my great collaborator and
00:58I'm happy looking forward to talk.
01:01Thank you, Alexei.
01:08Okay. Well, thank you very much.
01:10Let's move on.
01:13So let me start by saying that
01:17humans are a mosaic of germline
01:20and somatic genomic variations.
01:23And we're all very different from
01:25each other for different reasons.
01:27And similarly, genetic risk for human
01:31disease rarely map to a single gene.
01:33I'm not going to say that this is unheard of.
01:36We just heard two very good examples of that,
01:39but it's a rare phenomenon.
01:40Most most disorders map to multiple genes,
01:44risk genes.
01:45So one effort of our lab is being to find
01:49convergence in biological mechanisms.
01:52In complex developmental disorders,
01:57of course, this present challenges,
01:59particularly when you're studying the brain.
02:01So let me explain to you what I mean
02:04by convergence. So here you see.
02:07The promoter with the gene downstream of it.
02:10And that could be thought of a convergence,
02:13right? Because the promoter eventually
02:16synthesize regulates product synthesis
02:17which regulates cellular function.
02:19But you can see up here, there could
02:22be a number of different mutations,
02:24right, that could actually lead to
02:26the exactly similar or same phenotype.
02:29And this could be a close by
02:31like it could be in a promoter,
02:32but it could be very far away like
02:34in an enhancer that could be hundreds
02:36of kilo base away sometimes or it
02:38could be in different chromosomes
02:39for the transcription factor that
02:41binds to that enhancer.
02:42And so this very disparate different
02:45mutation could actually lead to
02:47essentially the very same phenotype.
02:49So how are,
02:50how are we going to deal with that?
02:53And one way to study this is exactly
02:56what we've been talking about today
02:59is to study personal genomes and the
03:03way these personal genomes can develop
03:06in vitro in different ways and what's
03:08the influence on this development.
03:10So we've been using induced pretty buttons
03:13themselves to generate these organoids,
03:15which are 3D aggregates of neuro
03:18progenitors that can be interrogated using.
03:21Different assay at the genomic level,
03:23at the transcriptomic level and
03:26at the epgenomic level, right.
03:28And and hopefully we could also look
03:31at the 3D DNA confirmation to try to
03:34put all this together and come back and
03:37and derive a model for intersection
03:40between genes and phenotypes.
03:43So this sounds easy.
03:46It's not.
03:47Let me first go through what organoids
03:49are and why do we want to use them.
03:51So obviously they are a longitudinal
03:54model of brain development,
03:56so they respect individual
03:59genetic background largely,
04:01which is very important.
04:03And let's not forget they can be
04:05developed from living people,
04:07so we can still try to do
04:11correlations between phenotypes.
04:12And in vitro development of these systems,
04:17this is how they look like.
04:18We've been growing them for over
04:2010 years now and we found ways
04:23to grow lots of them in batches.
04:25If you section them,
04:27they have these complex layers
04:28of progenitors.
04:29This is, these are cortical organoids.
04:31So you see cortical progenitors,
04:33you see immature neurons piling
04:35up here on the external side.
04:37And then if you grow them for very long time,
04:41you eventually see something that looks
04:42more similar to the actual cortex.
04:44In Cortex,
04:45you see layer five and six developing
04:48before layer two and three.
04:49So it's an inverse development.
04:51You can see that at one month you
04:54see layer 6 neurons here on the
04:56outside of the progenitors in blue.
04:59And then at five months you
05:01start seeing not only layer 6,
05:03but also layer two and three in red.
05:06So really the development of this
05:09system seems to recapitulate,
05:10at least in great lines,
05:14what's happening in the real brain.
05:16And eventually you even have glial cells.
05:18You see here astrocytes of 5 1/2 months
05:21that develop in the organoids as well.
05:23So a long time ago, but.
05:25Five years ago we asked the crucial
05:27question right to what extent
05:29organoids look beautiful but do they
05:31are really similar to the real brain?
05:34So we did a paper where we took three
05:36fetal specimen and had a cortical
05:39specimen for those fetal specimen at
05:43about 1617 postconceptional weeks.
05:45And then we developed organo.
05:48We developed in just pretty bottom
05:50stem sets from skin fibroblast from
05:52those specimen generated organoid.
05:55Can analyze them over a time course of
05:57three time courses and compare them to the
06:01isogenic cortices and what we found here.
06:03You can see these are the
06:05actual samples at the bottom.
06:07These are the cortex themselves and these
06:09are the organoids over three time points.
06:12Compared to a large data sets of
06:15gene expression across human stages,
06:18you can see that while the brains the
06:21cortices of this specimen are a snapshot
06:25of development because they map exactly
06:28to a 1617 post conceptual weak cortex,
06:31the organoids are a range.
06:34Here they present a range of
06:36similarities that go back.
06:38Not only to the 16 postconceptional week,
06:41but back to April postconceptional week,
06:43and even possibly earlier for
06:45stages for which we don't have
06:48human brain to compare them to.
06:50So organoids are a way to look in back
06:53from stem cells to later developmental,
06:56fetal and late fetal developmental stages.
06:58So let me show you a few slides on an
07:01ongoing study on autism spectrum disorder.
07:04This is a data set of
07:1014 families where we take the problem
07:13with autism spectrum disorder and
07:15we compare to the unaffected Father.
07:18And we do this to avoid a spuriously
07:20negative background comparison between
07:22groups that may be very different.
07:24So we do intra family comparisons here.
07:28And another thing we're doing,
07:29we separated head circumference size.
07:32So we separated patients into
07:35macrocephalic which have a larger brain,
07:37larger brain size versus those that don't.
07:40And the reason for doing that is that
07:43about 20% of people with all these
07:47marmacrocephalic and they're often
07:49have higher severity of symptoms.
07:52So here you see a single cell data sets
07:55of this that we generated in this study.
07:58Each dot representing this U
08:00map represent a single cell,
08:02and they're grouped by
08:04transcriptome similarities,
08:05and this is one of the largest data sets,
08:08if not the largest of organoids by
08:10single cell sequencing represents about
08:14650,000 / 650,000 cells.
08:16And you can see that they're
08:18grouped into various cell types.
08:20And here you see at the
08:22bottom radial glial cells,
08:23there is a trajectory
08:24between radial glial cells,
08:26intermediate progenitors and
08:27eventually cortical excitatory
08:29neuron and inhibitory neuron.
08:31And they're annotated by canonical markers.
08:34And one thing I like to point out
08:37that over time you see that there is a
08:40trajectory where the progenitors decrease
08:43in quantity and neurons increase.
08:45Which is what's to be expected.
08:48And let me highlight an important
08:51distinction of two particular cell
08:54groups that in this organ or that
08:57reflect actual development and
08:58one is the pre plate during early
09:01neurogenesis and the cortical plate
09:02which will form the actual cerebral cortex.
09:05So the pre plate is a transient layer
09:08of cells that develop very early.
09:11And serves as has various developmental
09:14functions but then eventually disappears.
09:17And then soon after that you
09:19have the actual cortical plate,
09:21the six layer cortical plate developing
09:23from the same regular real cells and we
09:26have both type of cells in this organoids.
09:30And of course another thing that I
09:32want to highlight is the viability.
09:34This is given the large data
09:36set that we have,
09:37we could actually assess that.
09:39You can see by color each color
09:42represent a cell type and one.
09:45One source of variability of course is age.
09:48And you see light green are TD0 and in in
09:53darker green TD30 and TD60 various stages.
09:57And so of course that TD0 you have
09:59more progenitor regular cells in pink
10:01and later you have more neurons,
10:02but still there is a large
10:05variability between different preps.
10:07And so how do we deal with that?
10:09That's that's an important phenomenon in
10:12in this field that we need to understand
10:15and we need to possibly study, right.
10:18So I was talking about
10:20variability earlier on.
10:21So,
10:22so what's you this variability what,
10:24what is,
10:25what is the origin of this variability and
10:28so one thing obviously could be a number
10:31of factors like reprogramming like you know.
10:36And anything that has to do
10:37with batch effect of cultures.
10:39And of course some of this may
10:41have an effect,
10:41but we largely excluded them and
10:43it seems that one important source
10:46of viability is genetic background,
10:48because if we culture IPS line
10:51organoid from the same individual.
10:54Even if they're cultured in
10:56different batches or differentiation,
10:57they're still displaying more
10:59similarity than all the other ones.
11:01So we believe that any background is
11:04an important driver of these differences.
11:07And these differences,
11:08of course, they're not random.
11:09If we have different percentage, say,
11:12excitatory neuron or inhibitory neurons,
11:14it's not just a serendipitous phenomenon.
11:17And you can see here, for example,
11:19that is highly correlated.
11:21Back down here with expression of
11:24certain genes in progenital cells.
11:26So the reason why we have different
11:29percentage of a excitatory neuron
11:32and inhibitory neuron is because
11:34there is a different programming
11:35of this transcription factors in
11:37the progenital cells in those
11:39spreads. So they do reflect
11:42differences within each organoid,
11:44within each organ that's derived from
11:46a particular person and perhaps.
11:48Reflects intrinsic difference in in
11:51the development of each one of us.
11:53So what happens if we compare people
11:56with autism with their father?
12:06So when we compared gene expression,
12:09single cell gene expression
12:11between problems and their father,
12:13we got the first surprise and that was
12:16that when we analyzed macrocephalic
12:18and normal cephalic separately.
12:21We found that they don't intersect
12:23or they intersect very minimum.
12:25That means that the differential
12:28gene expression is largely
12:30specific to which AST subgroup,
12:32and you can see examples of that here.
12:34So for example,
12:35in red you see differential gene
12:37expressions that are increased,
12:39in blue that are decreased and the one
12:41that are increased in macrosympalic
12:43which reflect largely dorsal cortical
12:45plate neurons and their progenitors.
12:48And a decrease in transcript or inhibitory
12:51neuron are actually not the same that
12:54are in fact they're the opposite.
12:55So if these are increase in macrocephalic,
12:58those are decreased and the enormous
13:00ephalic also don't have any significant
13:03change in interneurons and that's reflected
13:06also by the relative abundance of cells.
13:09So in macrocephalic individuals
13:10you see an increase in these dorsal
13:13cortical plate neurons,
13:14a decrease in preplate.
13:16And in the normal cephalic,
13:18you if you have the opposite phenomenon.
13:20So this was quite puzzling,
13:22quite interesting and do we
13:24have an explanation of that.
13:27So here just to show you that even by
13:31immunocytochemistry we reproduce these
13:33differences that I just described.
13:36So what we think this reflects is
13:39actually a different difference
13:41in the actual pathogenesis.
13:43Because if I go back to the pre plate
13:45and cord and those are cortical plate
13:48enormous epalic people that we have
13:50an increase in pre plate neurons that
13:52basically say what does it say that
13:55this radio glia says prematurely exit
13:57the cell cycle generate more pre plate.
14:00These are transient population and
14:02there is less of course progenitor that
14:05generating the subsequent cortical plate.
14:07Whereas the microcephalic
14:09of the opposite phenomenon,
14:10this progenitor generates fewer pre
14:12plate and there is more progenitors,
14:14there is more later on to generate
14:17an exuberant cortical plate
14:19neuron generation and so why?
14:23Why do we think this is important?
14:24Why is this something that's of interest?
14:26Because obviously this could reflect
14:30differences in the actual pathogenesis of
14:33what we call homogeneously autism people.
14:36They may actually be not reflecting the
14:38same pathogenic phenomenon in development.
14:41So just to summarize this part,
14:43organoid.
14:43Reproduce the lineages and cell type,
14:47at least the major one that we
14:49see in protocol development.
14:50There is great variability that in
14:52genetic programs of differentiation
14:54across individual and there are two
14:57different formal ASD that perhaps
15:00are different in pathogenesis
15:02and potentially they could have
15:05potential implications of treatment.
15:07So the next question was why?
15:09Why do we have these differences?
15:11What the transcriptome is without?
15:13What's the origin of these
15:15transcriptomic differences?
15:16And this brought up,
15:17this is ongoing studies,
15:19still unpublished bring us to
15:22the next step which is the non
15:24coding element of the genome.
15:26So as you know those are
15:28the portion of the genome,
15:30the regular gene expression.
15:31So to analyze those what we did,
15:33we took the non holding genome
15:36segmented by using cheap seek data,
15:39chromatic immunoprecipitation
15:40in various regions,
15:42mainly enhancers, promoters,
15:44repressed regions and mixed regions.
15:48And then correlated them with gene
15:52derived the data sets of about
15:55173,000 gene linked enhancers.
15:56So took the enhancers,
15:58linked them to genes and then perform
16:01correlation analysis where we could
16:04actually correlate the enhancer
16:06activity to the transcription factor
16:07that was bound to that enhanced.
16:09So correlation between activity of an
16:12enhancer and and transcription factor RN,
16:16A/C levels for those.
16:18The transcription factor,
16:19the bound to it and correlation
16:22between the enhancers and
16:24the downstream link chain.
16:26And by doing that we built the
16:28regular and what we mean by regular
16:31is a map of this gene enhancer
16:33transcription factor interaction.
16:36And we could identify two
16:37different type of enhancers,
16:39ones we call activating enhancers
16:41because they're positively correlated
16:43with the downstream genes.
16:45And whereas the repressing enhancers are
16:47those that are negatively correlated
16:49with the downstream genes and you see
16:52an example of this phenomenon here.
16:54So this is the regulatory graph for emx one.
16:57This is one of the genes that was
16:59up regulated in microcephalic AST.
17:01And you can see that there
17:02is this enhancer here
17:06693906 which is the major
17:08enhancers that activates emx one.
17:10This is the correlation coefficients.
17:12There are other,
17:13but they're less less powerful at
17:16activating this gene transcription and
17:18this enhances upstream of this enhances.
17:21There are five transcription factors okay,
17:24and four are inhibiting this enhancer
17:26and one IOM which is another gene
17:28that was appregulated in ASD models,
17:30sophalic is actually activating
17:32that enhanced so.
17:34So that's one example of going
17:38upstream of gene expression and trying
17:41to find out what's happening above.
17:43And then another thing that we've
17:45been doing is actually looking
17:47at the transcription factor that
17:48drives the type specification.
17:50So we looked at the single cell rnac.
17:53Derived gene markers that are
17:56specific for certain cell types,
17:58say excitatory neuron for example,
18:01you see they're not here or
18:03inhibitory neuron or radial glia.
18:04And then found those transcription
18:06factors that actually can explain
18:08or are correlated with this cell
18:10type specific gene expression.
18:12And we derive sets of transcription factor
18:14for example that can activate all neurons.
18:16You see these are all neurons.
18:18And repressing already or perhaps
18:20the activator of a certain type of
18:23excitatory in human or a certain
18:25type of inhibitory in human.
18:27So this,
18:27this really improves our ability to go
18:30upstream of cell type differences and
18:32try to find out what are the upstream
18:35mechanism that regulate those cell types.
18:38And finally this is the
18:42regular of macrocephalic AST.
18:44So what we did here we displayed
18:47in the regular.
18:48The differential gene expression
18:49in macrocephalic ASD,
18:51and you see here a few genes that
18:53are upstream that are actually their
18:55transcription factor that regulate
18:56exactly for neuron development.
18:58You see emx one that I talked about before,
19:01and you see this analysis 693906
19:05that as you saw before,
19:07is regulated by IOMS,
19:08which is also regulated in ASD.
19:11And then these are connected to
19:13other transcription factor through
19:15enhancing that as you can see here
19:17the red means they're activated.
19:19So we're really going upstream and
19:21explaining this gene expression
19:23differences by the activity of
19:25those enhances.
19:26So we have enhancers that activate
19:28genes and we have enhancers like this
19:30one here that are downstream of genes.
19:32So for example this enhancers
19:34downstream activated by Neuro D2,
19:36by eoms and by Vhlh E22 which
19:40are all up regulated.
19:41So this is a self reinforcing network
19:44that is giving us some ideas of what's
19:48going on in the development of of
19:51this organized in these patients,
19:54but you might ask.
19:55Why do we care about this?
19:57Why do we want to show all these enhancers?
20:00What the reason is this again right?
20:02Because how do we know that one enhancer,
20:06say here,
20:06activate a certain genes if we
20:09don't make a regular?
20:10We need to make a regular in
20:12order to actually make these
20:14connections meaningful and possible.
20:15And not only that,
20:17we need to make this regular
20:18mean different individuals in
20:20as many individual as we can.
20:22In order to be able to figure
20:25out this relationship
20:26and to figure out how the non coding
20:29genome relates to the coding regions,
20:31we already found some intersection
20:33between our differential express
20:36genes and the spy genes and other
20:38data sets of ASD risk genes.
20:40But these are coding genes we need to
20:42do this work for the non coding part,
20:45so that's being our next step so.
20:50I don't have much time,
20:51but let me say that perhaps differential
20:55regular activity can explain one day ASD,
20:58differential gene expression and
20:59self state and can point to no coding
21:02element that are enacting these changes.
21:04And in the last few slides,
21:06let me go back to something that
21:10Andrea Chenko spoke to you just
21:13a few minutes ago. And in fact,
21:16we by chance have this very same figure
21:18here about Peter Lawrence French flag,
21:21which basically says that
21:22gradients are important, right?
21:24And these are the collab young
21:26collaborators in his lab and my lab
21:28that they've made possible this project.
21:30And we're all very grateful to them.
21:32But basically they built this Chamber
21:34which allows us to look at the
21:37orthogonal effect of two gradients,
21:38and we intagonist,
21:39which is posteriorizing the organoids
21:41and a Sonic a joke agonies, which is.
21:44Ventralizing them and these are the
21:46organoid cultures in this area and this
21:49is a slide that you already showed,
21:50so I'm not going too much in detail.
21:51We're reading the gradient by
21:53using gene expression.
21:55And just let me say that it's
21:58fantastic what we see because dorsal
22:01genes which are up here like TBR
22:04one and Fox G1 are expressing the
22:07dorsal portion of the chambers.
22:09And not in the ventral and
22:11cortical genes like Phase 2 E MX2,
22:13they're expressing the anterior chamber
22:15which is C5D is anterior and C1 is posterior.
22:19And then ventral gene like nkx
22:222.1 which you see here in mouse,
22:24it's a ventral gene in the basal
22:26ganglia instead is expressed in
22:28the basal ganglia in the ventral
22:31regions of the of the chamber.
22:32So this Chamber really seems
22:34to be able to make.
22:38Brain regions,
22:39specific brain regions
22:40different from one another,
22:42and we only need 5 days of
22:44exposure to this gradient.
22:45And after that we can remove
22:47the organo from the Chamber and
22:49culture them in the usual way.
22:51But let me say one thing.
22:53Individual variation.
22:54We've been talking about that, well,
22:56it turns out the out of seven IPS
22:59line that we tried in this gradient,
23:01they're all different.
23:02They don't respond in the same way.
23:04So these are the chambers,
23:06these are the gradients.
23:08Anterior, C5, posterior C1,
23:10dorsal and ventral.
23:11And you can see that the slope,
23:13the response is similar but
23:15the slope is different.
23:16What does that suggest?
23:18That at least in this type of
23:21assay we don't behave the same way?
23:23Organized from different people respond
23:25in slightly different way and perhaps
23:28that's due to genetic background and other.
23:31Epigenetic and other phenomenon
23:32that are peculiar to each one of us,
23:35so the our brains are not constructed
23:38according to this in exactly the same way.
23:40And that's my last slide,
23:43is EU map for this Chamber or
23:47derived organoid altogether?
23:49If we combine all the organic together
23:51make single cell RN A/C can make a U map.
23:53Well, obviously it doesn't look like
23:55EU map I showed you before, right?
23:57It's not just cortex.
23:59You have palamus, you have subparium,
24:01your midbrain with dopaminenergic
24:02neurons in there.
24:04You have floor plate,
24:05you have medium structure septum
24:07with corioplexus in there developing
24:09and cortex of course.
24:11And these regions come from different
24:14regions of the Chamber, right?
24:15So the pallium,
24:16which is the cortex,
24:17come from the anterior regions and
24:20if we project them to the mouse brain
24:23using a software called Box Hunt.
24:26Again, you see that the C5 anterior
24:29map mostly to anterior mouse
24:31regions and the posterior regions.
24:34Instead, C1 maps to posterior
24:36regions of the mouse brain and
24:38the same is for the ventral side.
24:40So in conclusion.
24:43This is a new system that we really want to
24:46exploit to make our organo more credible,
24:49to build organoids that are more similar,
24:52developed in a more in a way that is
24:54more similar towards the actual brain.
24:56Human brain actually develops
24:57and and and that's using organo,
25:00it is not using tissue culture
25:02dishes with with factors added.
25:04So let me finish by highlighting the
25:08contribution of all my colleagues in my lab.
25:11All of them have greatly contributed
25:13to this work.
25:15Jessica Mariani was the first one
25:17who developed an organo in my
25:19lab back more than 10 years ago.
25:21And Alex and so I have greatly
25:23contributed to the Chamber project.
25:25And then of course Alexei is
25:27collaborated with us for many years.
25:29You hear him soon.
25:31And Andrei Levchenko and his people
25:33have greatly collaborated with
25:36us for the Chamber project.
25:38And with that, thank you very much.