Yale Psychiatry Grand Rounds: March 19, 2021

March 19, 2021

"Multiscale Imaging: Revealing Links Between Neuronal Networks, Behavior, and Disease"

Michael Higley, MD, PhD, Associate Professor of Neuroscience, Yale School of Medicine

Information

ID: 6306
To Cite: DCA Citation Guide

Download Transcript

00:00Doctor Crystal couldn't be here today,
00:02so he's asked me to introduce
00:05our guest speaker today,
00:06Doctor Michael Higley Doctor Higley
00:08is currently an associate professor
00:10in the Department of Neuroscience.
00:12He got his MD and his PhD at
00:14the University of Pennsylvania,
00:16and then he went on to do his
00:19postdoctoral work at Harvard with
00:20Bernardo Sabatini where he worked on
00:22the basal ganglia and particularly
00:25synaptic mechanisms in the basal ganglia.
00:27He's gone on in his work here at Yale.
00:31To continue his efforts to do very
00:34basic neuroscience at the synaptic
00:36and cellular level and to extend
00:38that work to understanding circuits
00:40that are relevant for brain disease,
00:43whether it be psychiatric illness
00:45or neurological illness.
00:46And he's going to talk to us today
00:49about his work on multiscale imaging,
00:52and in this case it's not fMRI,
00:55its cellular and circuit level
00:57image Ng and he's going to lead us
01:00through what some of the current
01:02cutting edge techniques are.
01:04In preclinical science to reveal
01:07links between neuronal networks,
01:09behavior and disease.
01:10Mike, go for it.
01:14Marina, thank you and thanks to
01:15all of you for coming today.
01:18It's it's an incredible privilege and
01:20a pleasure to be here, especially
01:21giving talks to two local communities.
01:23One of the things that I would I would more
01:26than than welcome is both during the talk.
01:29If people have questions just
01:31just jump right in and ask them.
01:33And certainly afterwards if anybody wants
01:35to follow up with anything you know,
01:37an enormous part of these kinds of talks
01:39is to potentially ferment collaborations
01:41and an future interactions between.
01:44All of us here,
01:45at both the preclinical and clinical levels.
01:47So happy to keep talking afterwards as well.
01:50So today as Marina said,
01:52what I'm going to try to give you a
01:54sense of is some of the methodological
01:57approaches that my lab and others,
01:59both here at Yale and another locations
02:01have been using to study brain function,
02:03its relationship in Part 2 to
02:05North psychiatric disorders.
02:06So I'm not going to give you like an
02:09incredible deep dive in any one story,
02:12and in fact I'm going to.
02:14Tell you a few stories or parts
02:16of a few stories,
02:17especially in service to that introduction.
02:19Most of what I'm going to tell
02:21you is actually unpublished,
02:22and in fact a little bit of it at the end
02:25is going to be even quite preliminary.
02:27But again,
02:28hopefully it'll be an interesting
02:30and exciting opportunity to sort
02:32of learn what we've been doing.
02:34OK,
02:34so this is a bit of an outline
02:37about what I'd like to cover,
02:39so first I'm going to tell you a little
02:41bit about what behavioral state means to us.
02:44Then I'm going to go in for
02:46awhile about some of the methods,
02:47especially at the imaging the fluorescence
02:49based imaging methods that we use,
02:51and finally towards the end,
02:53I'll tell you what we've been
02:54working on in in the vein of neuro
02:56psychiatric disorders or models
02:57of neuro psychiatric disorders.
03:01OK, so first you know this is sort
03:04of like a really basic question.
03:06I mean, what is a behavioral
03:08state or what do we mean by it?
03:10And at the end I'm not going to
03:12give you any great conceptual answer
03:14to that question because I'm not
03:16totally sure there there is one,
03:18at least in the field and what I will
03:21stick to is largely a set of operational
03:23definitions and I'll go into that.
03:25But in a sort of vague ish way.
03:28What we in the field, both both
03:30preclinically and clinically often mean.
03:32By behavioral state could be
03:33levels of arousal or alertness.
03:35You know.
03:36Obviously,
03:36this sort of the mice and this
03:38this image here you know sleep wake
03:41transitions are a very obvious example
03:43of changes in behavioral state,
03:45but even just relative attention to
03:47what's going on in your environment
03:49versus sort of just passively hanging
03:52out and being in a very relaxed state.
03:55So attention concentration,
03:56something that comes up a
03:57lot in functional MRI field,
03:59sort of resting state versus
04:01maybe a task engaged state so.
04:03So these are just kind of words that we
04:06often used to describe behavioral state
04:09at a neural or at a nervous system level.
04:12Other words or phrases that often come up,
04:16you know,
04:16synchrony or correlational
04:17structure of brain activity.
04:19Default Mode network is something
04:21that we hear a lot about specific
04:23modulatory systems like acetylcholine
04:25or norepinephrine are thought to play
04:28important roles in behavioral state
04:30and and also sort of notions of top,
04:33down or.
04:34Higher order regulation of brain function or,
04:37say,
04:38perceptual ability versus bottom
04:40up or or sometimes a sending
04:43or sensory driven activity.
04:45And of course,
04:46all of these things are are profoundly
04:49linked to disruption in a number of nurse,
04:53psychiatric disorders.
04:53Depression, anxiety,
04:54attention deficit, you know,
04:56schizophrenia,
04:56autism so clear mechanistic links
04:58between these categories is really
05:00still a big topic of investigation
05:03in a number of labs and just sort of
05:06broadly in the field of neuroscience and.
05:09And as I said,
05:11I'm not going to provide any
05:13any strict links between them.
05:15Only to you,
05:16you know,
05:17sort of provide a sense of how we
05:19think about some of these things and
05:20and where we are as a field at the moment.
05:23OK.
05:23So how do we measure behavioral
05:26state in rodents?
05:27So my lab works entirely in rodents,
05:29really in mice,
05:30and this is an example of a head fixed
05:33mouse of the sort that we might use
05:35in any number of our experiments,
05:37and in fact many that I'll tell you about.
05:40And if you just watch this video,
05:42you'll see that the mouse
05:44is doing a few things,
05:45and I'll point out a couple.
05:47So one perhaps most obviously the mouse
05:50is running and I think he'll stop for
05:52a second and then start up again.
05:54But tracking that locomotion
05:56at head fixed but locomoting on
05:58this freely moving wheel is a
06:00very easy marker of
06:02something we can.
06:03We can sort of distinguish right?
06:05Not running, running or
06:07questions and locomotion.
06:08You can also notice his whiskers
06:10and some of his facial musculature
06:12is sort of constantly moving.
06:14It stops sometimes, then moves again,
06:16so these fine motor movements,
06:18especially in rodents.
06:19Whisking is a very obvious one,
06:22is another thing we can use to
06:24categorise behavioral states.
06:26And finally,
06:27if you if you sort of zoom in on his eye,
06:31here is a little bit subtle
06:33in this particular movie.
06:34There's not a huge dynamic range,
06:37but again,
06:37his pupil diameter also
06:39fluctuates throughout the movie,
06:40and that is another variable
06:42that we can use to sort of divide
06:45the animals behavioral, state.
06:47And so again,
06:48I'll be explicit that these are really
06:50operational definitions of behavioral state,
06:53and there are a number of ideas,
06:55some better.
06:56Tested and others about neural mechanisms
06:58underlying these different variables,
06:59but for the moment will really
07:01just stick to them as a sort
07:03of operational definitions of
07:04behavioral States and there's
07:06a long list of different labs,
07:08including ours.
07:08Also just Gardens Lab here at
07:10Yale that have worked on this,
07:12and we know that these behavioral
07:14states have a pretty big impact
07:16on behavior even in rodents,
07:17and so this is just an example
07:19of some data from our lab
07:21that we published last year.
07:23The details aren't terribly important,
07:24but on the X axis here I'm plotting.
07:27The intensity of a visual
07:29stimulus that the animal is being
07:31presented with visual contrast
07:32and on the Y axis is the animals.
07:35Correct performance on this visually
07:36guided task and what you'll see
07:39is a very characteristic sigmoid
07:40relationship between those variables.
07:42So as the visual contrast goes up,
07:44the animals performance gets better.
07:46And if we divide the animals
07:48behavior into times when it's,
07:50say running versus not running
07:52or in the bottom plot,
07:53you know when the pupil is large in
07:56diameter versus small in diameter.
07:58You'll see that those higher arousal states,
08:01either from locomotion or pupil,
08:03correspond with a left shift in
08:05that perceptual curve and also an
08:07increase in overall performance.
08:09So again,
08:10the mechanisms of this are
08:12not fully understood,
08:13but simply to point out that changes
08:16in behavioral state directly
08:18correspond to changes in at least
08:20perceptual ability in rodents.
08:23And so we get.
08:24Please jump in with any with any
08:25questions if anything comes up.
08:27OK,
08:27so how do we measure neural
08:29activity in our lab?
08:31So this is really going to be the meat
08:33of the meat of my lab does primarily,
08:36but also of the talk and so this
08:38is all going to be fluorescence
08:39imaging and we're going to in these
08:42studies use genetically encoded
08:43indicators and I'll talk a bit
08:45more about how we get those into
08:48the into the brain or into cells,
08:50but they can either,
08:51you know,
08:52be transgenic mice where the mice
08:53express the number of these indicators,
08:55sort of in their genome,
08:57or else we can use, for example,
08:59viral vectors to drive expression.
09:01I'll come to that just a minute,
09:03but probably the most
09:05common indicator in our lab,
09:06and honestly in the preclinical
09:08field of neuroscience in general,
09:10are indicators that report
09:11calcium free concentrations of
09:13cytosolic calcium within neurons.
09:14And G camp is probably the most
09:17ubiquitous and most well known.
09:18This is a green flora for it's it's
09:21molecularly a molecule of GFP green
09:23fluorescent protein fused account module,
09:25in which calcium binding protein and you
09:28can sort of see the structure of it here,
09:31and so you've got this this.
09:33GFP molecule circularly permuted
09:35variant of GFP bound to calmodulin and
09:37so when calcium binds this protein,
09:39changes its confirmation and
09:40changes the fluorescent properties.
09:42The molecule there are also red
09:44shifted variants of these Jr.
09:46Camp and Jay are gecko or two variants
09:48just that use two different fluorophores,
09:51an Ruby and an Apple.
09:52These are red shifted floors,
09:54but again bound it to calmodulin.
09:56Most of these have been developed
09:58by the groups that that.
10:00Nelia research campus outside of Washington,
10:03DC.
10:03The other molecule that's going
10:05to be important for the talk today
10:08is a sensor for acetylcholine.
10:10This something actually,
10:11the Marine's lab, is used as well.
10:13This is a tool that was developed by you
10:15Longleys lab at the University of Peking.
10:18They called it a Ch 3.0.
10:20This is their recent variant of it.
10:22It's it's maybe not the most distinct name,
10:25but that's what we're going to
10:27call it for today, and this is GFP,
10:29bound to a variant of a cholinergic
10:31muscarinic M2 receptor.
10:32But the same idea applies.
10:34Basically it's a it's a.
10:36Floor for bounded molecule and
10:37so when acetylcholine binds it
10:39undergoes a conformational shift that
10:41changes its fluorescent properties.
10:43So why is this useful?
10:45So here I'm showing you kind of a
10:47cartoon example for calcium, right?
10:49So here we've got a neuron.
10:51This blue ball expressing G camp.
10:53This is fluorescent Reporter and
10:54when we shine blue light in this
10:57case 480 animator light on this,
10:58normally nothing happens.
10:59You don't get any green fluorescence,
11:01it just sits there.
11:03However, when the cell is active,
11:05when it fires action potentials,
11:06for example, that depolarizes the membrane,
11:08it opens voltage gated calcium channels
11:10which allow calcium to rush into the cell.
11:13The calcium then binds.
11:14Thejy camp and now in its calcium bound form.
11:17When we shine blue light on that,
11:20this molecule will now emit green light,
11:23and so we can collect these green
11:25photons in various imaging modalities.
11:27But essentially this increase
11:29in green light means that there
11:31was an increase in intracellular
11:33calcium which we use as a proxy
11:35for increased neuronal activity,
11:37and so that will be really
11:39important for most of this talk.
11:41Similarly, this cholinergic Reporter AC H.
11:443.0.
11:44So in this case,
11:45this is a membrane bound protein,
11:47and so when acetylcholine is
11:49released from wherever it happens
11:51to be released from, if it binds this
11:53protein in the membrane of the cell.
11:56The same idea. Now suddenly you can get
11:58green fluorescence from this molecule,
12:00and so in this case the green
12:02fluorescence reports the extracellular
12:03presence of acetylcholine,
12:04which is bound to these membrane receptors.
12:08Alright. So the modality that I'm
12:11going to talk about most today for
12:15imaging is a one photon widefield.
12:17We usually call it mesoscopic imaging and
12:21what this is is a method for imaging the
12:25entire at least dorsal surface of the
12:28mouse cortex at one time has some advantages.
12:32We can do this through the intact skull,
12:35so the surgical invasiveness
12:38is relatively low.
12:39The the fluorescent reporters that we
12:42use are quite right and so you can get
12:45really nice signal even through the skull.
12:47You can see the entire cortex.
12:49As I said, sort of one time.
12:51So if you're interested in
12:53interactions of different regions,
12:54that's fairly straightforward.
12:55The perhaps disadvantage is that you cannot
12:57resolve individual neurons this way.
12:59These are from aerial signals,
13:01probably smeared out over at
13:02least 100 microns or more.
13:04So in some sense it's a little
13:06bit akin to do fMRI,
13:08or maybe even like a continuous.
13:10Local field potential or
13:11or EG kind of signal,
13:13but still with fairly high
13:16spatial and temporal resolution.
13:18So this is just one quick example.
13:20I want to get everybody oriented
13:22a little bit.
13:22So this is kind of a cartoon of
13:24a microscope that we would use.
13:26So we've got big objective that
13:28goes over the mouse is head.
13:29All of these mice are going to be head fixed,
13:32but awaken free to run and so
13:34forth is in that movie.
13:35I showed you a moment ago and this is
13:37a view sort of an unprocessed image of
13:39what you can see through the microscope.
13:41So this is the front end of the mouse.
13:44The mouse is head is pointing
13:46up so side side midline.
13:48Back end and to all these dark
13:50lines that you can see that's the
13:52vasculature of the brain of the
13:54cortex that shows up quite nicely.
13:56This is a mouse which is
13:58transgenically expressing G cap.
13:59So in this case every excitatory
14:01neuron in this mouse cortex is
14:03expressing G count an when we take
14:05a movie we process it a little bit
14:08just to look at changes in signals.
14:10This is what you get and this is
14:12set up three times a little bit
14:14faster than than real time,
14:16but hopefully you can see this.
14:18Pretty interesting and dynamic set
14:20of fluctuations that are happening
14:21across the entire cortex,
14:23and this is for a mouse that's just
14:25sitting there, not doing anything,
14:27so this is sort of a very quick
14:29introduction to how dynamic the
14:31cortex is even when we are supposedly
14:33not doing anything at all,
14:35just just sitting there.
14:36And so this work was described
14:38in a couple of papers that we
14:40published just last
14:41year. If you're interested in more details,
14:44but I'm going to use this method
14:46for the next several slides,
14:47so just as a reminder.
14:49This is the top down view of the mouse.
14:52Brain Front is up.
14:53Back is down side to side.
14:57So while that last mouse was transgenic,
15:00much of the data that I'm going to
15:02show you today uses also another
15:03new method that we've developed in
15:06collaboration with Mike Rares Lab.
15:08Also here at Yale.
15:09And this is to use viral vectors to
15:11drive the expression of transgenic
15:13proteins rather than having to make a
15:16mouse where you say knock in a gene.
15:18This is, it's not terribly complex,
15:20but it's it takes a lot of work and,
15:23and so this is in some respects a
15:25much easier way to get transgenic
15:27expression in mice.
15:28And so in this case we take a mouse pup.
15:32This really only works well in
15:33the first couple of neonatal days,
15:36and we take a viral vector
15:38adeno associated virus.
15:39In this case,
15:40this is pretty ubiquitous
15:41in neuroscience these days,
15:42and it turns out that the serotype
15:449 variant of AAV crosses the
15:46blood brain barrier really well,
15:48especially in the early post Natal period.
15:50So we take a virus in this case.
15:53Two different viruses,
15:54one driving our camp,
15:55which is this redshifted calcium
15:57indicator that I mentioned?
15:58And one driving this H 3.0 cholinergic
16:01signal so we mix those viruses
16:04together and we inject them into the
16:07transverse sinus is of this mouse
16:10is brain into P0 to P1 period and
16:13what you see here on the right is
16:15some Histology demonstrating the
16:17massive and robust and widespread
16:19expression of both the green
16:22fluorescing cholinergic indicator,
16:24the red fluorescent calcium indicator,
16:26pretty much throughout the entire brain.
16:29So this neonatal virus injection
16:31protocol is really incredibly powerful.
16:32Works for anything that you can
16:34express via AV,
16:35and in fact I'll show you some some
16:38additional data at the end of the talk.
16:40For some more nervous psychiatric models,
16:42but regardless that this isn't
16:44otherwise wild type mouse that you
16:46just sort of picked out of the cage.
16:49And now you can get really nice
16:51expression of these indicators.
16:54And so again,
16:54this was first described in a
16:56paper published last year.
16:58An much more methodological description
16:59of this is in this paper by Hamodia
17:02doll from last year as well,
17:03and it was only hamodia made by
17:05Chris Lab that they really sort
17:08of pioneered this tool.
17:09Alright, so we're going to use now.
17:11This dual expression approach of
17:13the red shifted calcium indicator
17:14in the green fluorescent cholinergic
17:16indicator for some studies.
17:17I'm going to show you about.
17:19Alright,
17:19this is probably going to be the
17:21busiest slide that I show you
17:23through the talk,
17:24so I'm going to walk you through
17:26it in some stages,
17:27but there's there are a few
17:29important things to point out,
17:31so this is work that was done
17:33by sweater Lohani,
17:34who is a postdoc in the carbon
17:36lab here and Andrew Moberly,
17:38whose adjoint postdoc between my lab.
17:40Not injustice Liberia.
17:41So first this is a similar cartoon to
17:44what I just showed you a few minutes ago.
17:47Only instead of there being one CMOS camera,
17:49we now have two CMOS cameras,
17:51one to detect the red signal which is
17:53calcium in this case and wanted to take
17:55the green signal which is acetylcholine,
17:57and in this case, but again you've got the
18:00mouse sitting on the wheel and so forth.
18:03So these are just a couple images similar
18:06to you know the movie I just showed you.
18:08Again we're looking down on the
18:10top of the mouse brain up, up.
18:12Is that the front of the brain down
18:15is the back and here we've got signals
18:17in the top row for the H 3.0 and in
18:20the bottom row our camp and we can
18:23divide this these cortical regions
18:25into different chunks by a number of
18:27different methods and I'll talk a bit
18:29more about that in a moment as well.
18:31But here you can sort of make out.
18:34I've drawn on an ROI or region of interest.
18:37It's sort of a little circle ish
18:39thing around a frontal region.
18:41This sort of motor cortex and also
18:43another here around the back region
18:45or visual areas and so in those
18:47areas we can now plot the fluorescent
18:50signal that's coming from those two
18:52areas as a function of time so we can
18:55see how these signals vary overtime
18:57for these two different regions.
18:59And so first I just show you the our camp
19:02data and so now you can see in purple above.
19:05This is the variation in that red
19:08fluorescence corresponding to cortical
19:09neuron activity via calcium in V1 in
19:11purple and then just below that in red.
19:14Is this this motor region and
19:15below all of that at the bottom?
19:18I've also drawn traces that
19:19correspond face map here.
19:21This is the facial musculature.
19:22It's mostly whisking,
19:23so this shows you sort of when the
19:26animal is whisking and then when it
19:29stops and when it was a bit more.
19:31Similarly,
19:31we have the pupil diameter plotted
19:34as a continuous function of time,
19:36and then we also have running
19:38speed on the wheel.
19:39You can see that these variables are
19:42roughly correlated with each other,
19:44and that's something that we
19:46see typically that there's some
19:47underlying state variable of arousal,
19:49whatever that means that is
19:51strongly correlated with a number
19:53of these behavioral variables,
19:55and so you'll note that the fluorescent
19:57signals the purple and red,
19:59actually agree roughly.
20:00With the fluctuations in these
20:03behavioral state variables as well.
20:05And so this is the calcium signal.
20:07Again, correspond to local cortical activity.
20:09We can do the same thing looking
20:11at the green fluorescence,
20:13which is the acetylcholine readout.
20:14And again,
20:15you see fluctuations in the
20:17acetylcholine that also track
20:18these behavioral state variables,
20:19and we can get that for both.
20:22In this case,
20:23this motor region is also the visual cortex,
20:25so there's two main points then that I'd
20:28like you to take from this whole figure.
20:31The first is that as you note,
20:34these signals are very dynamic in time,
20:37so they they fluctuate at both
20:39fast and slow timescales.
20:41The slower timescales correspond roughly
20:42to these behavioral state transitions,
20:44probably locomotion and pupil are the
20:47easiest to see as they go up and down.
20:50You see you sort of corresponding variation
20:53in the in both the cholinergic in that
20:56in the calcium signal so there's a lot
20:59of temporal dynamics and the other.
21:01Point is that there's a lot
21:03of spatial heterogeneity,
21:04and that perhaps you can appreciate most
21:06just looking at the images at the top so
21:09you can also see that in the traces as well.
21:12And for calcium, I mean maybe that's not
21:14surprising in the sense that different areas
21:16of the brain or or at least the cortex,
21:19are doing different things
21:20at different times.
21:21OK, it was perhaps a little surprising
21:23for the acetylcholine in that,
21:24at least in the early days of
21:26the field and early meaning,
21:28say, 50 years ago,
21:29there was a sense that some of these
21:31brainstem modulators might really be
21:32just sort of global arousal signals,
21:35and so you might have expected that.
21:37OK, you know when arousal goes
21:38up acetylcholine everywhere,
21:39it goes up,
21:40and then when arousal goes downhill,
21:42going everywhere goes down every sort of.
21:44A simple readout of behavioral
21:46state and what we find,
21:48and this isn't this isn't
21:50necessarily surprising,
21:50especially given recent anatomical
21:52findings about the diversity of cholinergic
21:54projections throughout the brain.
21:55But nevertheless,
21:56what you can see here is that
21:58even the acetylcholine is
22:00incredibly spatially heterogeneous.
22:01And so acetylcholine release in
22:03different parts of the cortex can be
22:06remarkably UN coupled from release
22:08in other locations in the cortex.
22:10So so, really.
22:11Acetylcholine is not just sort
22:12of a generic signal.
22:14Of overall state, but in fact as a as a,
22:19you know,
22:21primary component of dynamic
22:23cortical variables.
22:25And just to finally make that
22:27that last point a little bit,
22:29these black traces that you see here and
22:31here are the instantaneous correlations
22:32between the two cortical regions,
22:34and so you can just see there that.
22:37So,
22:37for example,
22:38the correlation of acetylcholine release
22:40between M2 and V1 starts out high,
22:42then it actually drops a little bit,
22:44then it goes up again.
22:46So just the correlational structure of
22:48these signals varies as a function of time,
22:50so this is really about religious
22:52descriptive view of the kinds
22:54of studies that we're doing.
22:55And so now let me go into a little bit
22:57more detail of some of our analysis.
23:02So the first thing that we do is
23:04spend a good amount of time trying
23:07to understand what behavioral
23:09state means or what some of these
23:12transitions mean quantitatively.
23:13An for the next little part of the talk.
23:17I'm really going to focus on the two
23:20major motor signals that we study,
23:22locomotion and facial movement here.
23:24It's called face map.
23:26This is named after the software that we
23:29use to extract these these movement signals.
23:32Primarily corresponds to whisking
23:34in detail if anyone is interested.
23:36It's essentially a principle
23:38component based decomposition of
23:39the video ography of the face,
23:41so you basically take a movie of of the
23:44animals entire face and you decompose it
23:47via principal component analysis to get
23:50a bunch of features that describe that.
23:52But nevertheless most of it
23:54really just agrees with whisking.
23:56That's that's the most dominant component.
23:58So looking at these traces,
24:00you'll know a few things so.
24:02Locomotion is perhaps easiest, right?
24:04So the animals just sitting there
24:06quiet for most of the time,
24:08and this sort of pinkish box sort
24:09of illustrates this sustained low
24:11level of locomotion.
24:12Really, no locomotion at all,
24:14and then later on,
24:15the animal starts running.
24:16It has these two little bursts of running
24:18and then a slightly longer period as well.
24:21And note the time scale here.
24:23This is 25 seconds for the bar,
24:25so so this this whole trace is
24:28actually a fairly long period of time.
24:30And then if you look at the at the facial
24:33movement in the in the below trace,
24:36you see that it fluctuates much,
24:38much more rapidly,
24:39but there's some agreement,
24:40especially if you look over the right side,
24:43you get these sort of sustained changes that
24:46correspond to to the balance of locomotion.
24:50So we just think about this
24:51a little bit conceptually.
24:52What this tells us is that
24:54when the animal is not running,
24:56sort this big pink bar of quiescence.
24:57You actually have some periods where
24:59where the animals whiskey or work where
25:01this facial motion energy is high,
25:03and then you also have some
25:04periods where it's slow,
25:05so you can get very big fluctuations
25:08in facial movement even when
25:10the animal is not running.
25:11You then look there with the far right for
25:14the periods where the animal is running.
25:16Every time that the animal is running
25:18for a sustained period of time,
25:20the facial motion energy,
25:22the facial movement is high,
25:23and so this suggests sort of a
25:25serial progression of arousal,
25:27which is that first you have like
25:29quiescence and no facial movement.
25:30Then the animal starts having
25:32some facial movement.
25:33They can go back and forth,
25:35and then at some point the
25:37animal might start to run.
25:38Whenever it runs,
25:39it's always whisking,
25:40so you've got this sort of three stage
25:42process of like total quiescence.
25:44Facial movement without running and
25:46then facial movement with running and
25:49we think that that represents a sort
25:51of serial progression of level of arousal.
25:56So what I'm showing you here then,
25:58is sort of the average fluorescence
26:01signals in the in the two indicators
26:03that I told you about divided
26:05or actually really in this case,
26:08subtracting the period of sustained high
26:10versus period of sustained load data,
26:12and so we just take the top row
26:15first here and So what we see
26:17for this cholinergic indicator,
26:19the AC H 3.0 when when you subtract
26:22high facial movement versus low facial
26:24movement you get really bright.
26:26Red signal across pretty much the
26:28whole cortex telling us that there's
26:30a lot more acetylcholine released
26:31during high facial movements than
26:33than low facial movements.
26:34So so there's a big difference there.
26:36It goes up when you look though,
26:39between transitions between no locomotion,
26:40locomotion, it's still all red.
26:42You know it,
26:43but it's a very it's much smaller,
26:45so there's a big increase in
26:46acetylcholine release when the animal
26:48starts having facial movements.
26:49And then there's a very small additional
26:52increase when the animal starts running.
26:54Then look at the bottom row for our campus.
26:57The calcium indicators for the
26:59readout of cortical activity.
27:01It's a bit the opposite,
27:02so when the animal whisks,
27:04there's a modest increase in
27:06the amount of neural activity,
27:08but there's a much bigger increase
27:09in the amount of cortical activity
27:11when the animal starts running,
27:14and so this is already starting to
27:15give us a sense that acetylcholine
27:18and cortical activity are not
27:20perfectly coupled with each other,
27:22and might be signaling different aspects of.
27:24Behavior.
27:27And actually so all of this data just
27:29to point out is is described in A
27:31in a manuscript by sweat and Andrew
27:32that's currently on by archive.
27:34If people are interested in more details.
27:38The one other little thing I'll put in here,
27:41I'm just going to mention this briefly.
27:43I'm not going too much detail,
27:45but at the same time we're
27:47doing all this image in.
27:48We have an electrode placed in cortex,
27:50which essentially gives us an Electro
27:52cortical gram or or E kog written here.
27:55This is again the sort of local
27:57field potential dynamics,
27:58the electrical signaling
27:59going on in the cortex,
28:00and what we see when you also
28:02look at low versus high arousal,
28:04whether it's facial movement or
28:06locomotion is something that's been
28:07described for decades and decades.
28:09Which is that the high frequency
28:12activity high frequency electrical
28:13activity goes up when the animal is
28:16higher aroused and the amplitude
28:17goes down and this has been
28:19interpreted for ages ascential E as
28:22local decorrelation of the network.
28:23So it basically means that neurons that
28:26are near each other in a local network
28:29or actually becoming less correlated
28:31with each other when arousal goes up.
28:33So just bear that fact in mind.
28:36It's not critical,
28:37but I mention it 'cause 'cause we've done it.
28:42So these data are really about the
28:44amplitudes of these fluorescent signals.
28:46You know how does the average signal change.
28:49So acetylcholine goes up.
28:50Calcium goes up when the
28:52animals arousal level goes up.
28:54But really, the major advantage of
28:57this imaging approach that we've
29:00got where we can see the entire
29:02cortex at one time is that we
29:05can look at the coordination of
29:07different areas so we can look at
29:09relative changes in these signals in
29:12different cortical areas overtime.
29:14And you might imagine that.
29:17Most of us think that behavior
29:18happens because of the coordinated
29:20activity of the nervous system,
29:22and so if it's coordinated,
29:23nervous system activity that's
29:25driving behavior really coming up
29:27with methods that allow us to see
29:29that coordination rather than just
29:31saying OK area a goes up or down.
29:33It's really the interactions
29:34that matter the most,
29:36and so the power in part of this mysa
29:38scopic imaging is that we can look at
29:41the coordination of different areas,
29:43so we're going to do that
29:45now for these two signals.
29:47Relative to changes in behavioral state.
29:51So the first thing that I'll mention
29:54is I alluded to it before when I
29:57said we could draw areas around
29:59safe motor or visual or whatever.
30:01How do we do that in a quantitative way?
30:04And how do we do it in a way that
30:07you know allows us to sort of say,
30:10OK, this area is the same area in
30:12mouse a mouse beam else whatever?
30:14It's not so different,
30:15for example then then then work
30:18in the fMRI community as well.
30:19We use an Atlas,
30:20and in this case we use an Atlas
30:23called the common coordinate
30:24framework version three,
30:25which was developed by the Allen
30:27Brain Institute in Seattle,
30:29and what they did was used a whole
30:31lot of anatomical labels and some.
30:34Fiber tracing labels to come up with an
30:36average parcellation of the mouse neocortex,
30:39and that's illustrated here.
30:40And so again,
30:41you know front is up back is is down.
30:44Since you got motor areas,
30:46you've got some at a sensory areas in
30:49the middle visual areas in the back.
30:51Auditory is quite lateral and so forth,
30:54and so the utility of such an
30:56Atlas is maybe a bit obvious
30:59given the points I just made,
31:01it allows a certain regularity
31:03across animals.
31:04Up its validity remains a little bit.
31:08Unclear,
31:09and I'm not going to talk about
31:10this in any detail other than
31:12just to mention it right now.
31:14There are certainly other ways of
31:16parcel eighting the cortex one could
31:18do it in an activity based way.
31:20You could in fact look at say
31:21OK Pixels in our movie that are
31:24most correlated with each other.
31:25Well,
31:26they belong to the same area
31:27and you could do that in a lot
31:30of different quantitative ways,
31:31but you could come up with some
31:33parcellation based on the activity.
31:34It turns out,
31:35for reasons that aren't totally clear,
31:37those kinds of functional parcellation's
31:39don't map onto these anatomical atlases.
31:40All that well and the reasons for that are,
31:43as I said,
31:44are unclear and is actually something that
31:46we're quite interested in pursuing a lot.
31:48But for the moment for convenience sake,
31:50and certainly for these data,
31:52we're just going to stick to this
31:53Allen Brain Atlas and we're going
31:55to say that area is circumscribed
31:56by these borders correspond
31:58to functionally related areas.
32:02So now we can make plots like this and
32:04you may recall a couple slides ago.
32:06I sort of showed you that running correlation
32:09between say motor and vision and so all
32:11these sort of complicated shapes are is.
32:13So if we just look at the one on
32:16the far left for the moment, right?
32:18We've got all of these cortical regions
32:20defined by the Allen Atlas on the
32:22bottom axis and also on the left axis.
32:24And so we're going to do a
32:27pairwise correlation.
32:27So every pair of areas we're just going to
32:29take the average correlation in signal.
32:32Between, you know,
32:33say motor and vision or motor and
32:35auditory or motor and sensory.
32:37So every little box in this matrix
32:40is the average correlation between
32:41a pair of cortical areas,
32:44and so you can see the entire
32:46pairwise matrix here,
32:47and we've done this now for
32:50the cholinergic signal.
32:51On the left is a stage 3.0 and our
32:53camp on the right the calcium signal
32:56and rather than show you the absolute
32:59correlations what I'm showing you is that.
33:02Difference in correlations
33:03across behavioral state.
33:05So in this case this is high facial
33:08movement minus low facial movement and
33:11the fact that pretty much all of this
33:15is red in both graphs tells you that
33:18the correlations the pairwise correlations,
33:20the sameness or the similarity in
33:22activity between two areas goes
33:25up for pretty much everything.
33:27So both acetylcholine and calcium
33:29signaling are becoming more correlated for
33:32every most pairs of areas in the cortex.
33:35When the animal starts whisking,
33:37so.
33:38Firstly,
33:38that's a little bit interesting
33:40because I just told you a moment ago
33:43that from these electrophysiological
33:45recordings that we make,
33:46that provides evidence that local circuits
33:48are actually becoming less correlated,
33:50and that's really sort of
33:52party line right for ages,
33:54everybody knows that arousal goes with
33:56reduced correlations of local circuits,
33:58and what I'm actually showing you here
34:00from these large scale imaging studies
34:02is that being circuits big networks
34:04across the whole cortex or actually
34:07becoming more correlated with arousal.
34:09And that we see that for both cholinergic
34:11signals and for for for calcium.
34:13So that's kind of cool and
34:16a little bit unexpected.
34:17But something that emerges from
34:19this from this image in modality.
34:21So this is for facial movements.
34:23Now we're going to do the same
34:26thing for locomotion.
34:28It looks different.
34:30So the calcium signal in our camp
34:32looks similar ish.
34:33I mean most things are red.
34:35It's a little less red.
34:37Sorry that the Gray boxes are those
34:39that were not significantly different,
34:41so if it has a color in it,
34:44it's significantly bigger or smaller.
34:46If it's just grey,
34:47it was non significantly different.
34:49So the calcium signal goes a little bit less,
34:52but someone but the acetylcholine goes
34:54robustly in the opposite direction.
34:56So what this is telling us is when
34:58the animal starts whisking the
35:00correlations in acetylcholine release
35:02across the cortex are going up.
35:04But suddenly when the animal starts running,
35:06the release of acetylcholine in different
35:08cortical areas completely decorrelate's.
35:10I will just say right now we
35:12really don't know what that needs.
35:14It is a purely observation.
35:16Ull point from these data were very
35:18interested in what it means and
35:20many follow up studies are being
35:22being carried out in our labs.
35:24And it's also in others to sort of
35:26think about mechanistic relationships,
35:28but at the moment I mean this
35:30is this is quite striking and
35:32and again emphasizes that.
35:34Acetylcholine is not sort of a
35:36really simplistic signal of arousal,
35:38as really interesting dynamics that
35:40seem to be differentially coupled to
35:42different kinds of behavioral states.
35:43It also tells us that calcium
35:46signaling acetylcholine signaling
35:47don't always go together,
35:48and so the relationship there is
35:50something that we're very interested in.
35:54Alright, so. This is a summary
35:57figure than of what we've sort of
36:00extracted from all of that data.
36:02So if I was to give you any sort of take
36:05home that was divorced from the from the
36:08details that it might be something like this,
36:11so we envision, right that arousal,
36:13whatever that means,
36:14operationally defined, again,
36:15not not really conceptually,
36:17goes up in some kind of monotonic fashion,
36:19an it operationally goes from the
36:21animal really just sitting there
36:23completely passively to the animal,
36:25beginning to whisk to the animal whisking,
36:27and also running.
36:28On a wheel,
36:29and that that transition seems to go
36:32with an overall increase in the amount
36:35of acetylcholine that's released
36:36and the overall activity of the of
36:39the cortical networks themselves.
36:41If we then though,
36:42look at the correlations between
36:44areas within the cortex,
36:46we see that local cortical circuit
36:48correlations tend to go down.
36:50That's from the electrophysiological data,
36:52and many,
36:53many other labs,
36:54but the large scale cortical network
36:56correlations go up.
36:57That's that's the Arkham correlations,
36:59but the acetylcholine follows this very
37:02interesting non monotonic relationship
37:04where correlations first go up and then
37:06go down again as the animal progresses
37:09through this this arousal continuum.
37:11So this is sort of the take home now.
37:14I mean,
37:14obviously this is.
37:15This is a little bit unsatisfying in the
37:18sense that we don't necessarily know,
37:20at a mechanistic level what this means.
37:22But really,
37:22I mean,
37:23I would emphasize that this is
37:25kind of the cutting edge of
37:27preclinical work at this point,
37:28and so This Is Us starting to lay the
37:31groundwork for understanding more
37:33about about how these signals interact.
37:36Alright,
37:36so now I'm going to switch briefly
37:39to one more methodological riff
37:42on this widefield imaging.
37:44And I've alluded to local circuits and
37:46individual neurons within a small region,
37:48and that they might be doing
37:50interesting things and then obviously
37:51I've shown you a lot of data for
37:53large scale circuit organization.
37:55Well,
37:55what if you really wanted to get
37:57a nice readout of both of those
38:00two things simultaneously?
38:01So what are single cells doing water,
38:03large networks doing and how
38:05do they interact?
38:06So Dan Barson is an MD PhD student
38:08who is in my lab and also Co.
38:11Advised by Mike Rare and Dan and
38:13Alijah Modi and in my prayers lab
38:15came up with this approach where we
38:17combine this widefield mysa scopic
38:19imaging that I've been showing you
38:21with what's a little bit my labs
38:23with bread and butter which is 2
38:26photon imaging and I'm not going to
38:28go into a lot of detail of what that means.
38:31If you're not familiar with it,
38:33but essentially two photon imaging
38:34is cellularly resolved imaging,
38:35although over a much smaller.
38:37Field of view and you'll see
38:39that in just a second.
38:41So in some sense this is mashing
38:43up two different microscopes,
38:44Amy's a scope and a two photon
38:46microscope that we sort of just
38:48slammed together on optical table,
38:50made some tweaks to it, and allows us
38:52to use both modalities at the same time.
38:55And it basically looks something like this.
38:57So on the left these are kinds of images
39:00that I've been showing you now for a
39:02few minutes and on the right these are
39:05two photon images, so this Gray box.
39:08This sort of field of view.
39:10Is the entire T of a tiny little box,
39:13shown here in the dotted lines,
39:14so this whole big box here is really just
39:17a tiny little bit of all of the cortex,
39:20and these faint white circles
39:21or blobs that you see here.
39:23Each one of those is a single
39:26neuron in the cortex.
39:27So maybe I'll play the movie.
39:29And So what?
39:30You can appreciate now is these data
39:32are being acquired simultaneously,
39:34so we are now watching.
39:35So every time one of these
39:37little blobs here lights up,
39:39that's that neuron firing action potentials
39:41and every time an area over here lights up,
39:44that's that cortical region being active.
39:46And so this approach now allows us to say,
39:49well, OK, we in this neuron is active.
39:51What cortical regions are active,
39:53and vice versa when a particular
39:55region is active,
39:56even if it's far away?
39:57From where we're imaging
39:59these cells doesn't have any.
40:00Impact on the output of this
40:02single neuron and so really,
40:04you know,
40:04we feel like this is a little bit
40:06of a groundbreaking approach to try
40:08to relate multiple scales of neural
40:10activity really from the single cell
40:12level up to large scale networks an.
40:15As one slide to sort of illustrate
40:17what we can do with that,
40:19I'll just walk you through this example.
40:21So this blue trace here.
40:22This is just the time series,
40:24the activity of a single neuron.
40:26So we just take the fluorescence
40:27from one neuron.
40:28We plotted overtime and you can see it.
40:30It goes up in these little
40:32spiky things and then is flat,
40:34and so everyone of these little transients
40:36is when that cell fires an action potential.
40:38Or maybe a few action potentials.
40:40We can then take that Nisa
40:42Scopic movie right,
40:43which is a movie.
40:44It's a whole bunch of frames,
40:46and functionally you can take every
40:48frame from that movie that corresponds
40:50in time to one where one of these spikes
40:52are and you average those together.
40:54You only average the ones that are
40:57present when this cell is spiking,
40:58and that gives you this image here and we
41:01call this a an event triggered average.
41:03We call it a cell centered network
41:06and what it is is the areas across the
41:08cortex again front and front is up.
41:11Back is down.
41:11These are the areas of the cortex
41:14that are strongly coactive or
41:15correlated with this one neuron,
41:17and so you can imagine that you
41:19could do it for every neuron in your
41:22field of view and what it turns out
41:25is that individual neurons have very
41:27different SCMS telling us that the
41:29long range connectivity of single
41:31cells varies a lot, even for two cells
41:33which are right next to each other,
41:36and so we're using this approach
41:37to learn something about
41:39connectivity between big and small,
41:41and just to bring in some of
41:43the behavioral state data.
41:44It also turns out that those networks are
41:47dynamic as a functional behavioral state,
41:49and here these are just two examples.
41:51Top is 1 example of autumn is
41:53another where we calculated this
41:55sort of average network for two
41:57different neurons divided into,
41:59say, whisking in quiescence.
42:00And what you'll see is the fact that
42:03the left image and the right image look
42:06different tells us that the coupling
42:08of each of these two neurons to the
42:11large scale cortical network differs
42:12whether the animal is whisking or not.
42:15And here on the right,
42:16these are just the correlations of these
42:19two Maps for a whole bunch of cells.
42:21So if the left and the right were
42:23exactly the same, you get a one,
42:25and if they were sort of
42:27totally unrelated to each other,
42:29you get a zero.
42:30And So what you see is that for
42:33all the neurons in cortex it
42:35kind of spans that range,
42:36and so is the animal transitions
42:38across behavioral states.
42:39Some cells really don't care their
42:41large scale connectivity doesn't change,
42:43and other neurons completely reorganized
42:44their large scale connectivity.
42:46As a function of the animals
42:48behavioral state,
42:49so again,
42:50really providing this this cool evidence
42:52that connectivity is not only anatomy.
42:54Connectivity is very much a functional
42:57property of what the brain happens
42:59to be doing it at any moment to time.
43:02And the last thing I will say
43:05about all of this.
43:07Is that because these indicators
43:09are genetically encoded?
43:10We can express them in pretty much
43:13whatever cell type you're interested in,
43:15so if you want to say OK,
43:18these are all for excitatory neurons.
43:20These are kind of the connectivity Maps
43:23for excitatory neurons in the cortex.
43:25Well,
43:26what if you're interested in interneurons?
43:28We can just as easily express these
43:30indicators selectively in Gabaergic cells,
43:32in subtypes of Gabaergic cells,
43:34and derive these kinds of metrics
43:37for all different.
43:38Cortical neurons to learn about how
43:40different cell types are connected.
43:42We can almost go one better than that.
43:45What if it turns out that you have
43:47a mosaic animal in which some of its
43:50neurons are mutant or transgenic?
43:52Especially for maybe some interesting
43:54disease gene.
43:55We can then compare mutant cells and
43:57say wild type control cells from the
43:59same animal and ask how does the
44:02cell autonomous deletion of that gene alter?
44:05Or perhaps deletion of that
44:07gene cell autonomously disrupt?
44:08The large scale connectivity of
44:10those neurons.
44:10So essentially this gives us a
44:13very powerful readout into the
44:15connectivity of single cells.
44:17And again,
44:17this is this has been described
44:19in a paper from last year.
44:21OK,
44:21so in the last part of the talk
44:24I'm going to shift to something
44:27that is perhaps maybe a
44:29little bit more interesting to
44:32this Community, which is which is.
44:34Our work is very preclinical work.
44:37Nevertheless into models of neuro
44:39psychiatric disorders and my lab
44:41in close collaboration with just
44:43Cardens lab primarily works on
44:45autism spectrum disorder models.
44:47I'll just say from my perspective
44:50I this the notion of what these
44:53genetic models represents.
44:55Is not clear.
44:56I generally don't think that a
44:58mouse can be autistic in in a
45:00way that is is meaningful, right?
45:02These are animals in which
45:04we've deleted genes.
45:05Those jeans have been linked
45:07to autism spectrum disorders,
45:08and so we hope that these model
45:10systems give us insight into how
45:12genes regulate brain activity,
45:14how that ultimately translates
45:15into the clinical phenotypes
45:17associated with this disorder
45:18with these disorders is actually,
45:20I think, a much harder question,
45:22and something that is very
45:23difficult to do in a rodent.
45:26So I just want to put that out there
45:28'cause I feel sort of strongly about that.
45:30In fact,
45:31I think it puts the preclinical study
45:33in actually in a better position to
45:35not try to claim any sort of face
45:37validity and simply say these are.
45:39These are mechanistic studies to understand
45:41links between genes and brain activity.
45:43OK,
45:43so I'm going to tell you about our
45:46work in a model of Rett syndrome,
45:49you know,
45:50with the caveat that I just said
45:52attached and specifically these are
45:55going to be mice with mutations in me.
45:58CP2,
45:58methyl CPG binding protein 2 which is the
46:01causal gene disrupted in Rett syndrome,
46:03and so as many you probably know,
46:06Rett syndrome in human patients goes
46:08with with cognitive impairment with
46:10intellectual disability seizures.
46:11This sort of very characteristic
46:13loss of learned skills.
46:15Early in life,
46:16language deficits and then some really
46:19interesting stereotype and movements
46:20that also go with with a taxi as well.
46:23For you know,
46:24just just for saying that me CP2 is
46:27a is a transcriptional regulator and
46:29so it's disruption causes an enormous
46:32host of changes at the genetic,
46:34molecular and cellular level.
46:36So really we're looking at that
46:38sort of functional consequences
46:40of disruption of a gene that has a
46:43number of pathways that can interact with.
46:47So the really cool thing that we've
46:50discovered works, maybe surprisingly,
46:51and so this is work that's
46:53been done by Antara Majumdar,
46:55who is a post grad in the card lab and
46:59repent who is a technician in the card
47:01lab is to take these viral vectors.
47:05And I previously showed you
47:06that they work really well for
47:08expressing G cap or cholinergic
47:10indicators everywhere in the brain.
47:12They also work really well
47:14for driving crisper cast 9.
47:16Related proteins and so for these mice,
47:19these are transgenic mice in which CAS
47:229 is expressed in all excitatory cells,
47:24so these mice are otherwise fine.
47:27It just turns out at the moment it's
47:30methodologically easier to start
47:32with the transgenic kastein mouse,
47:34so these might express cast 9
47:36in all their excitatory cells.
47:38We then inject two different viral vectors.
47:41These AAV 9,
47:42one is driving a GFP tagged
47:45guide RNA targeting me CP2.
47:47And for some experiments I'll show you
47:49in a second another is just driving
47:51this red fluorescent calcium indicator.
47:53Our camp one be so first here in
47:55the center is just a little bit
47:57of confirmation that this crisper
47:59cast 9 strategy works.
48:00So what you see on the left panel?
48:02That's the guide.
48:03RNA and GFP expression that that again
48:05you see pretty much throughout the brain.
48:08And here on the right the red
48:10is staining for me, CP2,
48:11the green is the GFP.
48:13The guide RNA expressing cells and you'll
48:15see that there's really no overlap there.
48:17So basically every cell.
48:18It has guide RNA.
48:20We have successfully deleted me CP2
48:22expression and that's shown here.
48:23You know for the population,
48:25so control mice have.
48:26You know, roughly 90% expression of me.
48:29CP2 in all cells.
48:30And that's not down to about 20%
48:32expression in the in the crisper model.
48:35So it's not perfect,
48:36but I'll also point out,
48:38right that cast 9 is only
48:40in the excitatory cells,
48:41which is only about 80% of
48:43all cortical neurons.
48:44So since this is about 20%,
48:46it's actually suggestive that were in fact.
48:49Probably deleting me CP2 from
48:51almost all the excitatory cells
48:53that are expressing kastein.
48:54You could do this in a transgenic
48:56mouse in which kastein was
48:58even more broadly expressed,
49:00and we would presumably have
49:02even even larger effects,
49:03and then this is just a Western blot
49:06data showing as a function of me
49:08CP2 protein and controls the crisper
49:10mutants show about a 75% reduction in MCP,
49:13two protein,
49:13and this is in comparison to a
49:16standard knockout for me, CP2,
49:17which shows an almost complete
49:19loss of of me CP2 protein.
49:21So somewhat remarkably,
49:22you can use viral vectors in an otherwise.
49:25Wild type mouse and get robust
49:27loss of me CP two.
49:29So does that mean anything
49:31functionally for these animals?
49:33So now this is work analytical
49:35work done by Hospice Tia postdoc
49:37in my lab and lavonna Mark,
49:39who is a Yale Singapore undergraduate
49:42who's been Co advised by Jess and
49:44I an lavonna was just admitted to
49:46the Yellae NP program.
49:48It's we're aggressively trying to
49:50convince her that she wants to to
49:53come to Yale for her PhD as well.
49:55So what had Austin lab did was the same
49:58kinds of correlational analysis that I?
50:00And you about so these are these
50:02are just matrices showing the
50:04pairwise correlations of activity
50:06in different cortical regions.
50:08And they're going to use graph
50:10theory based analysis, and again,
50:12I'm going to kind of just just blow
50:14through any of the details here and
50:17tell you that the variable I'm going
50:19to show you is called centrality.
50:22And it's really a mathematical
50:23representation of how a given parcel.
50:25In this case they have blue dots,
50:28is connected with the rest of the cortex,
50:31so a a blue dot that is directly or
50:33indirectly coupled to a small number of
50:36other regions would have a low centrality.
50:39Whereas a region of blue dot that's
50:41really connected to a whole lot of other
50:43regions would have high centrality.
50:45So it's it's just a measure of
50:47how functionally connected one
50:49area is with with another.
50:51So the first thing I'm going
50:53to show you is this,
50:54which is probably very
50:56difficult to make sense of,
50:57but I throw up here just really
50:59quickly and what we've plotted
51:00here now is the centrality for
51:02all of these different regions.
51:04Again,
51:04all the different cortical regions
51:06comparing control mice in black,
51:07the black bars and the ME CP2.
51:10CRISPR mutants in Gray.
51:11And then we divide that up
51:12into quiescence and locomotion.
51:14And so you can sort of squinted
51:16this from bed,
51:17and you'll see that there are
51:19some differences in the bars
51:20in different regions and.
51:22If you want us to let your eyes go to that,
51:26that's fine,
51:26but this is a little bit of an easier
51:29way to appreciate what we see here,
51:32and it's similar for quiescence
51:34in locomotion.
51:34So I'll just describe quiescence here.
51:36What I'm plotting here is
51:38the difference in centrality.
51:40Basically,
51:40the black bars minus the Gray bars up
51:42here for each different cortical region,
51:45so it's the it's the controls
51:47minus the mutants,
51:48so everywhere that it's red,
51:50it means the controls are.
51:52More connected and everywhere that it's blue,
51:54it means that the mutants show higher
51:57connectivity and so you can see this
52:00is very spatially heterogeneous.
52:01There's some really interesting
52:03networks whereby in some cases
52:05mutants show higher connectivity
52:07and in others the the controls do,
52:09and that's roughly similar,
52:11although not perfectly similar
52:13as a function of brain state.
52:16So this is really where I said at
52:18the very beginning that some of
52:20the data we're going to be quite
52:23preliminary and this is these are these.
52:26Are they?
52:27So I don't fully know what this means,
52:30but what it's telling us is that
52:32these this mutation strategy allows
52:34us to see very different patterns
52:36of connectivity across the cortex,
52:38suggesting that these might be
52:40classical disruptions seen in some of these.
52:43Some of these models of
52:45autism spectrum disorder.
52:46I'm not going to see any data,
52:48but I will say that we have worked
52:51on another gene called REI One,
52:53which is retinoic acid induced gene
52:55one which is the causal gene and Smith
52:57Magenis syndrome and it actually looks
52:59quite similar and an REI wanted me to
53:02have nothing to do with each other,
53:04so it's really quite intriguing
53:05and what we've sort of started
53:07this project to look at is to ask
53:09whether or not different models of
53:11autism spectrum disorders which
53:13made genetically or molecularly
53:14have nothing to do with each other,
53:16converge on similar network level phenotypes.
53:18And so this is sort of an example
53:20of the kinds of network phenotypes
53:22that we're actively exploring,
53:24and in some of these models.
53:26And you might imagine could easily
53:28be applied to whatever your favorite
53:30model of of neuro psychiatric
53:32disorders might be to the last slide
53:35that I'll show you is something
53:36that's perhaps equally cool,
53:38which is that we can do sort of some standard
53:41benchtop behavioral assays in these mice,
53:43and so in this case,
53:45this is the three Chamber sociability assay.
53:47You're simply asking whether or not
53:49a control or mutant mouse prefers
53:51to hang out with a conspecific,
53:53or prefers to hang out by itself in an empty.
53:57Cage and you'll see the
53:59control data shown here.
54:01This is sort of typical and I'll
54:03sort of be transparent and say this
54:06is the black bars are males and
54:09females lumped together that we don't
54:11see much of a difference between
54:13males and females of control mice.
54:16However,
54:16when we then look at the mutants and
54:19we divide them into males and females,
54:22we get very interesting and different
54:24phenotypes, whereas the males,
54:26the male mutants, much prefer to.
54:28To be alone,
54:29the female mutants show and even
54:31heightened preference for being
54:33with conspecifics,
54:34and so this is also preliminary.
54:36It's actually already significant,
54:37but this is sort of part of
54:40a much larger study,
54:41but I will sort of emphasize
54:44that these are these AAV mice.
54:46These were otherwise wild type C57
54:48mice that we used viral vectors to
54:50drive loss of functioning me CP two
54:52that produces both substantial network
54:54dysregulation and standard behavioral
54:56deficits seen in these mutants.
54:58So I think it illustrates.
55:00Through the power of these viral
55:03tools for expanding the OR for
55:05making more flexibel,
55:06the ability to study
55:08mutations of various genes.
55:12So OK, so to summarize everything
55:14that I've I've told you,
55:16so one behavioral state is something
55:18that we don't have a great handle on.
55:21We define it operationally,
55:23but it clearly corresponds to a
55:25very high dimensional combination of
55:27both motor and autonomic activity.
55:29Arousal seems to be associated with
55:31increases in both cholinergic signaling
55:33and just general cortical activity
55:35decreases in local circuit correlations.
55:38But really interesting dynamic
55:40changes in large scale correlations.
55:42Loss of me CP2 expression via this
55:45viral crisper strategy drives changes
55:47in both network activity as well
55:49as behavioral deficits and really
55:51sort of bring all that together.
55:53You know,
55:54simply to say that these viral vectors,
55:57music, scopic, imaging,
55:58genetic editing,
55:59all these strategies in combination
56:01provide incredibly powerful tools for
56:02dissecting relationships between cells,
56:04circuits, and behavior,
56:05and so let me just finish by acknowledging
56:09again everybody that did all the work.
56:12So Dan Barson did all of
56:14the dual to Pizzo imaging.
56:16Had Aspen, Insteon,
56:17lavonna.
56:18Mark Levan is also advised by
56:20the by Jess Card,
56:21and have been doing all of the graph
56:24based analysis on this topic data.
56:26Andrew Moberly,
56:27an sweater Lohani in the card lab,
56:29did all of the dual calcium an
56:32acetylcholine imaging and Antara Majumdar
56:34an remote Panton justice lab did.
56:36The initial virus injections and
56:38did all of the Histology of the
56:40Western blotting data for the.
56:42For the Christmas stuff,
56:43we've been really generously funded by
56:45both the NIH, the Simons Foundation,
56:47and the Smith McGinnis Engine
56:49Research Foundation.
56:49For some of this work,
56:51as well as getting important support
56:52from the Copley Foundation here,
56:54yeah,
56:54so thank you all so much for
56:56bearing with me and I'm happy
56:58to answer any questions.
57:03Thank you Mike. Maybe you can stop
57:05sharing your screen and we can have
57:06people chime in with questions.
57:08There is 1. Technical question in
57:10the chat from Lauren Zima Low and
57:12the question was whether the controls
57:145050 male and female or where they
57:16skewed in one way or the other.
57:19No, the
57:19yeah, so that's really quite preliminary.
57:21The controls there I I think it's I think
57:24those controls are six animals and I
57:26think it's two males and and for females.
57:29So in some sense you know you should.
57:31You should take those statistics as
57:33a with a bit of a grain of salt for
57:36sort of an internal presentation,
57:38but I would say though that.
57:40They're really for those six animals.
57:42The four females and two males.
57:45They're pretty similar in their distribution,
57:47and even if you just look at them separately,
57:51it's clear that the male and
57:53the female mutants really
57:55seemed to push well outside
57:57those distributions.
57:57So along those same lines,
57:59Huda Zoghbi's lab has extensively
58:01characterized the full knockouts of me.
58:04CP2 has she seen *** differences
58:06in social interaction?
58:10I don't believe so.
58:11I don't want to say no at all 'cause
58:13I don't know if she's ever shown it,
58:16but one of the I don't
58:17know if it's a weirdness,
58:19but it's one of the facts of
58:20the MCP 2 field in general.
58:22Is that most of the rodent
58:24work is done in the males.
58:26The males have much bigger phenotypes,
58:27so me Susan excellent gene and so most
58:29of the work has been done in the males,
58:32though as you probably are aware,
58:34in clinical cases it's much more often
58:36females that are considered now that
58:38maybe because even in the humans.
58:39Loss of any CP2 in males is so
58:42devastating that in many cases it
58:45may not be compatible with life.
58:47So that may be why the human
58:49studies are actually biased to the
58:51less severely affected subjects.
58:53So one of the nice parts about this also.
58:56OK, so first point as well, right?
58:58So our viral expression method is post Natal.
59:02You know,
59:03if you want to study prenatal changes,
59:05obviously that doesn't work for that.
59:06But if you want to say spare
59:08any really early disruption and
59:09look at gene function later,
59:11it has an advantage in that regard.
59:13So let's study some of these *** differences,
59:15perhaps a bit more easily than
59:17we might with the transgenics.
59:19Thank
59:19you, Zoran. Do you want to
59:21ask your other question?
59:25I'm not sure if I can be
59:26heard well and can click
59:28no logical challenges here so.
59:31Is it possible to Anna is really love this
59:34seminar and its high-end Tour de force
59:37fantastic in all these fluorescent tests?
59:40Sometimes in in a simpler models.
59:42It's good to put a some
59:44kind of a negative control.
59:46Would that be 2 technologically challenging
59:49or have you thought about that?
59:51Yeah, I don't like a third probe
59:54that doesn't react to anything yet.
59:56OK, said, that's a.
59:58That's a fantastic question.
01:00:00And so yeah, so I didn't go into
01:00:03a lot of the details, but so one.
01:00:06There's a lot to unpack there,
01:00:08so so yes, getting just sort of a
01:00:11basil readout of how much flora for
01:00:14is in in a given region is really
01:00:16sort of of critical and right,
01:00:19so you'd want some readout that
01:00:21wasn't activity dependent.
01:00:22More than that changes in hemodynamics,
01:00:24especially right because
01:00:25hemoglobin absorbs photons,
01:00:26especially in these visible light ranges,
01:00:29and that that absorption varies
01:00:30as a function of oxygenation.
01:00:32So there are hemodynamic artifacts
01:00:34in in a lot of this as well,
01:00:37so figuring out those controls
01:00:38is a huge part of what we do,
01:00:41and there are two ways that
01:00:43we do to get around that.
01:00:45So firstly, well, I didn't say this before.
01:00:47If you illuminate G camp with
01:00:49UV light instead of blue,
01:00:51it actually fluoresces independently
01:00:52of its calcium concentration and
01:00:54it basically behaves like GFP.
01:00:56So you can use UV illumination
01:00:57and the G camp as a as a sort
01:01:00of activity independent readout,
01:01:02and so that can be quite useful as
01:01:05a normalizing. Signal in addition.
01:01:09You can use fluctuations in that signal
01:01:11as a readout of hemodynamics because
01:01:13we know it's not calcium dependent,
01:01:15so any changes in that signal,
01:01:17or presumably from hemodynamics.
01:01:18We can also collect the backscattered
01:01:20green photons,
01:01:21so we're shining green light on that
01:01:23issue for the our camp fluorescence,
01:01:25so the our campus fluorescing in the red.
01:01:28But you can also measure the
01:01:30green photons that just bounce
01:01:32off the brain and come back up.
01:01:34Those are also very sensitive
01:01:36to hemodynamic changes,
01:01:37and so you can measure fluctuations in
01:01:39that backscattered green fluorescence.
01:01:40Is another readout for hemodynamics and
01:01:43then regress those hemodynamic signals
01:01:45out of the data to essentially correct
01:01:47for those those in accuracies does.
01:01:49That does that answer the question?
01:01:52Yes, it it.
01:01:54Even a ratio might be a possibility
01:01:57between those two and then you can
01:02:00actually standardize the output but.
01:02:04It must be too complicated
01:02:05to do in a real experiment.
01:02:07No, no. I mean, as I said,
01:02:10I mean we use a regression based approach to.
01:02:12I mean which is essentially a ratio, right?
01:02:15I mean the regression just gives
01:02:16you the beta coefficients for the
01:02:18basically the fractional contribution
01:02:19of 1 signal to the other.
01:02:21So thank you. Al
01:02:29I can't hear you. Aw, we can't hear you.
01:02:36Maybe put it in the chat and then
01:02:40and I'll answer. I'll ask it for you.
01:02:43OK, there you go OK is it working OK?
01:02:47Sorry bout that.
01:02:48So what I was wondering is a great talk.
01:02:52I was wondering about this discrepancy
01:02:54between decreased local correlations
01:02:56and increased global correlations
01:02:57during periods of high arousal.
01:02:59Yeah, and I was wondering if you steal
01:03:02that as relating to like the effects of.
01:03:06Is it a calling on Amex or this is
01:03:08sort of like spatial scales at which
01:03:11acetylcholine is being released?
01:03:13Like meaning maybe the the during
01:03:15high arousal it's be the fluctuations
01:03:17in acetylcholine are happening
01:03:18on more defined social spatial
01:03:20scale or or something like that?
01:03:22Yeah, I mean I think that's, uh,
01:03:25those are sort of the questions
01:03:27that we're asking as well,
01:03:29and I don't have any great answers,
01:03:32so it's been it's been shown recently by by
01:03:35a couple of labs that if you for example,
01:03:38I mean so.
01:03:39Firstly,
01:03:40individual basil forebrain neurons
01:03:41which are the main supplier of
01:03:44acetylcholine to the cortex,
01:03:45have pretty divergent axons.
01:03:46So some go to some places,
01:03:49some go to others,
01:03:50and so the coordination of acetylcholine
01:03:52release probably substantially involves
01:03:54the coordination or lack of coordination.
01:03:56Between individual neurons within
01:03:58the basal forebrain and so as we
01:04:00as we try to learn more about
01:04:02what regulates their activity,
01:04:04that's probably going to give us
01:04:06some insight into you know both the
01:04:08increases and then decreases the
01:04:10correlation of of acetylcholine
01:04:11release in the cortex.
01:04:12And there are also a small number of
01:04:15local interneurons within the cortex
01:04:17that can release acetylcholine.
01:04:18Quite honestly,
01:04:19I don't think we have an understanding
01:04:21of the relative contributions,
01:04:23right?
01:04:23Our assumption,
01:04:24although it's not based on a
01:04:26whole lot of quantitative data.
01:04:27Is that most of the acetylcholine
01:04:29that we're seeing comes from
01:04:31from basil forebrain projections?
01:04:32But but it's certainly possible
01:04:34that there's a.
01:04:35There's a substantial contribution
01:04:37from local release and that may play
01:04:40some role in whether you see you know
01:04:43increases or decreases correlation.
01:04:45Doctor data that was a great talk Mike.
01:04:48So just to follow up to ALS questions,
01:04:51I was wondering if it's possible
01:04:53that these local circuits are being
01:04:56entrained at a particular say,
01:04:58oscillatory frequency that permits
01:04:59greater coherence across these
01:05:01anatomically connected large scale circuits.
01:05:03Is that one sort of underlying sort of?
01:05:06Yeah, I mean very much,
01:05:08so right?
01:05:08I mean so gamma activity is certainly
01:05:11a proposed mechanism for a long time
01:05:14of coordinating long distance regions.
01:05:16One of the things that I didn't emphasize,
01:05:19mostly 'cause it's a problem that
01:05:21we can't fix,
01:05:23is that the calcium indicators
01:05:24are really slow,
01:05:26so the timescales over which we're
01:05:28seeing fluctuations are really
01:05:30slow on the order of you know,
01:05:32I could be generous and say
01:05:34hundreds of milliseconds,
01:05:35but probably even seconds you know,
01:05:38whereas you know gamma activity to
01:05:40to coordinate regions is obviously
01:05:42much faster than 40 Hertz or so,
01:05:44so it's possible that those
01:05:46oscillations could be.
01:05:48The envelope of those oscillations
01:05:50might be what we're seeing
01:05:52that you know, maybe the coordination
01:05:54is is driven by by gamma say,
01:05:56but that's all we really have access to.
01:05:59Might be the envelope of those correlations.
01:06:02It's really challenging with calcium
01:06:04imaging to get a more direct readout.
01:06:07We're optimistic, though we haven't
01:06:09really gotten it to work yet.
01:06:11That voltage imaging,
01:06:12especially with the music scale approach,
01:06:14may provide an ability to see
01:06:17fluctuations like like gamma
01:06:18at this sort of spatial scale,
01:06:21but that's that's.
01:06:22The bleeding edge.
01:06:23Sort of like just just past the
01:06:24bleeding edge at the moment and stuff.
01:06:29I just wanted to to make a note
01:06:32that using the same sensor,
01:06:34the Estel cooling sensor and vibra
01:06:37photometry of principle neurons
01:06:39in the basal lateral amygdala,
01:06:41we also see fairly clear dissociation
01:06:44between the activity of Astle colon
01:06:46and the structure of the release of
01:06:49Astle calling in the structure and
01:06:52coordination of firing of the entire
01:06:54excitatory network in the structure.
01:06:57So if it is inducing,
01:06:59let's say, local Theta, or like.
01:07:01Local gamma it's not sufficient
01:07:03to to coordinate the network.
01:07:05The excitatory network that doesn't
01:07:07mean that there isn't a selection
01:07:10of neurons within that network,
01:07:12but overall the rhythms are not sufficient
01:07:15to coordinate the whole network locally.
01:07:19So Doctor Cederbaum,
01:07:20you want to ask your question.
01:07:23Yeah, so I mean not to try to get out of
01:07:27the realm of what you've done too far.
01:07:30Risk getting.
01:07:30I don't know for an answer here,
01:07:32but wondering if there's an opportunity
01:07:34here. Also to study patterns of glial
01:07:37activity and whether there are receptors
01:07:39or channels that are specific to
01:07:40various glial populations that could be
01:07:42leveraged to do that with this technique?
01:07:45Yeah, it's funny. You should say that.
01:07:47So one of the things that I dearly dearly
01:07:50love about Yale and having been here,
01:07:52is how much amazing collaborations
01:07:54have sprung up here and so.
01:07:56So Carla Rothlin's answer of
01:07:57doshas groups have really been
01:07:58interested in neuroinflammation,
01:08:00particularly as regards astrocytes
01:08:01and glial cells in general.
01:08:02For awhile,
01:08:03and so we have a very nascent
01:08:05collaboration going on with them
01:08:07starting to look at at glial activity.
01:08:09So as I said, I mean all of these
01:08:11indicators are easily expressed,
01:08:13and I mean anything that
01:08:14you've got a pre line for.
01:08:16We can express it now that is a
01:08:19little bit of a rub right because
01:08:21it actually turns out and this is
01:08:23this is a bit new to me as well.
01:08:26Genetically targeting astrocytes is.
01:08:27Is not quite as easy as one might think,
01:08:30so it's been known for awhile that like
01:08:32GFP or at least the GFP cream mouse
01:08:35is not that specific for astrocytes.
01:08:37You do get neurons labeled and in fact
01:08:39even some of the more recent blinds that
01:08:42have been developed like like LDH Creek,
01:08:44may also label neurons,
01:08:46especially if if expressed
01:08:47early in development.
01:08:48So so yeah, we can do it.
01:08:50We're very interested in doing it,
01:08:52especially as a part of collaboration
01:08:54with some of the groups at Yale.
01:08:56There are some remaining
01:08:58methodological challenges.
01:08:59To making sure that the signals that we see
01:09:02say just come from the cells of interest,
01:09:05But yeah,
01:09:05yeah.
01:09:09Any other questions for Doctor Higley?
01:09:14OK, well thank you all for
01:09:16joining us this morning.
01:09:17Thank you for taking time out of your day.
01:09:19Thank you Mike for presenting these
01:09:21interesting data and for pointing
01:09:22out how the preclinical and the
01:09:24clinical can be interpreted together.
01:09:26I have 111 quick answer is I saw somebody
01:09:28asked, does this map on to resting
01:09:30state fMRI and so let me give one
01:09:33other plug that was that you know
01:09:35that's you one other plug right?
01:09:37So so in a collaboration with
01:09:38Todd Constables Lab here at Yale,
01:09:40we also just published a paper.
01:09:42In nature, methods doing simultaneous
01:09:44miscopy calcium imaging and
01:09:45fMRI and that's allowing us,
01:09:47at least in mice,
01:09:49to make direct relationships
01:09:50pretty bold and calcium,
01:09:51which we hope will at least
01:09:53serve as a bridge for relating
01:09:56the cellular level analysis to
01:09:58human bold data as well so.
01:10:01Good way to good way to end.
01:10:04Excellent way to end. Thank you
01:10:06everybody and have a good weekend.