Yale Psychiatry Grand Rounds: Heninger Lecture: "Translational Strategies for Understanding Neural Circuitry Dysfunction in Schizophrenia"

Transcript

• 00:01Well, Matt, thank you very
• 00:02much for that very kind
• 00:03and generous introduction. And, Marina,
• 00:05my thanks to you and
• 00:06the, leadership of the department
• 00:08for inviting me here today.
• 00:10I always
• 00:11I just have to get
• 00:12rid of that message. Oh,
• 00:13okay. Enjoy,
• 00:15coming to Yale, which is,
• 00:17my second favorite department of
• 00:19psychiatry,
• 00:21but very close.
• 00:23The first.
• 00:24Not only because of the
• 00:25number of friends I have
• 00:26here, but because there are
• 00:27so many people past and
• 00:29present,
• 00:30in Yale psychiatry that have
• 00:31had a major influence
• 00:33just not only in our
• 00:34field, but on my own
• 00:35particular career path. And one
• 00:36of those individuals is is
• 00:38doctor Henninger. I never had
• 00:40the privilege of working directly
• 00:42with doctor Henninger, but I
• 00:44certainly read many of his
• 00:45papers which influenced me. And
• 00:46I also had opportunity, I
• 00:47think, primarily at ACMP,
• 00:50to he will not remember
• 00:51this, to have some conversations
• 00:53with him and to hear
• 00:55his presentations. And and one
• 00:57of the things that always
• 00:58struck me about it that
• 00:59I was thinking about I
• 01:00captured in the idea of
• 01:01a of a sober optimism
• 01:04about our field, that he
• 01:05was very
• 01:07that we could make progress.
• 01:08And I think
• 01:09but in his words, that
• 01:10dated back to his experience
• 01:11as
• 01:12a child seeing the impact
• 01:14of the polio vaccine and
• 01:16imagining that that could be
• 01:17true for psychiatry. But he
• 01:18was also very sober about
• 01:20knowing
• 01:21what we know, how well
• 01:22we know it, and and
• 01:24what we don't know. And
• 01:25I think that attitude is
• 01:26captured in this quote from
• 01:28him that says, you don't
• 01:28go to the moon in
• 01:29a wooden rocket ship. I
• 01:31mean, you have to wait
• 01:31until the pieces are ready.
• 01:33So the reason we haven't
• 01:34cured mental illness in fifty
• 01:36years is that we don't
• 01:36have our hands strongly on
• 01:38the causal factors. And I
• 01:40really agree and inspired by
• 01:42this statement, and what I
• 01:43hope to do today is
• 01:44to give you some ideas
• 01:45about how I think we
• 01:46are,
• 01:47and I hope with sober
• 01:49optimism,
• 01:50making advances in our understanding
• 01:52of one of these disorders,
• 01:53schizophrenia.
• 01:55So here are my,
• 01:57disclosures, research findings, a few
• 01:59consultations, none of which have
• 02:00any conflict with what I'll
• 02:01present today.
• 02:03But as I mentioned, I
• 02:04want to talk about the
• 02:05burden of schizophrenia,
• 02:07disorder which is common,
• 02:09typically chronic,
• 02:11associated with substantial comorbidities,
• 02:15and as a result,
• 02:16extremely costly,
• 02:18amounting to one of the
• 02:19leading causes of years of
• 02:20life lost to disability
• 02:22and premature
• 02:23mortality.
• 02:24Now I first became,
• 02:26developed a sense of compassion
• 02:27for people with schizophrenia and
• 02:29a curiosity about the illness
• 02:30when I was a third
• 02:31year medical student.
• 02:32But it wasn't until later
• 02:34in my training that I
• 02:35began to develop a strategy
• 02:38for how to approach the
• 02:39illness that was in part
• 02:40influenced by reading,
• 02:42this translation from the German
• 02:44by of Ewald Hecker's classic
• 02:46monograph
• 02:47on hebephrenia, which Kraepelin later
• 02:49used in his formulation of
• 02:51dementia precox.
• 02:52And towards the end of
• 02:53this descriptive monograph, Hecker said,
• 02:55the final excuse me. The
• 02:57final proof that hebephrenia is
• 02:58a unified form of mental
• 03:00illness
• 03:00in its own right
• 03:02can, of course, be established
• 03:04only
• 03:04on anatomical pathological grounds.
• 03:07But because of the still
• 03:08provisional character of our knowledge
• 03:10of brain anatomy, we must
• 03:11renounce this proof possibly for
• 03:13a long time to come.
• 03:15Three things struck me about
• 03:16this statement.
• 03:17First is obviously he was
• 03:18right. Here we are a
• 03:19hundred fifty years later still
• 03:21in pursuit
• 03:22of the final proof.
• 03:24But perhaps that was true
• 03:26because
• 03:27we had in, you know,
• 03:28in his quaint way of
• 03:29saying that we're ignorant.
• 03:31We had a we lacked
• 03:32the knowledge of neuroscience. And
• 03:34of course in the last
• 03:35half century, doctor Henninger's work
• 03:37and so many other people
• 03:38here, we've made major advances
• 03:39there. But I think the
• 03:40third thing that really resonated
• 03:42with me
• 03:43is that we needed the
• 03:44anatomical pathological
• 03:45grounds
• 03:46for this. And why this
• 03:48resonated is because before I
• 03:49went into psychiatry, I did
• 03:51a residency in internal medicine,
• 03:52and I was grilled in
• 03:54this model of the disease
• 03:55process.
• 03:58And, ultimately, applying this disease
• 04:00process model of schizophrenia, what
• 04:01we wanna understand
• 04:03is how is it that
• 04:04the etiology and apparent complex
• 04:06interplay between a large number
• 04:08of genetic liabilities and environmental
• 04:10risk factors,
• 04:11how do those intersect to
• 04:12produce
• 04:13a pathogenic
• 04:15process
• 04:16that results in a pathological
• 04:18entity?
• 04:19A set of disturbances in
• 04:20the brain that so alter
• 04:21the function of the brain
• 04:22that the resulting pathophysiology
• 04:25gives rise to the emergent
• 04:26properties that we recognize as
• 04:27the clinical syndrome.
• 04:29Now the dotted arrows here
• 04:30represent an extremely interesting and
• 04:32challenging scientific issue.
• 04:34But from a clinical perspective,
• 04:36I think imagining or thinking
• 04:38about the disease process is
• 04:40important because what we ideally
• 04:42wanna do with treatments
• 04:44is to normalize the pathophysiology
• 04:47so that even in the
• 04:48face of the persistent of
• 04:49the upstream factors,
• 04:51the clinical syndrome will remit.
• 04:53And what we seek to
• 04:54do with prevention, because primary
• 04:56prevention may not be possible,
• 04:57is to identify individuals who
• 04:59are at risk and intervene
• 05:01before the pathogenic process
• 05:03results in the pathology.
• 05:06But what struck me about
• 05:07this is that the pathology
• 05:09sits right in the center.
• 05:10And as I read about
• 05:11this, I, you know, Hecker
• 05:13and and things, I reflected
• 05:14back on
• 05:16an experience I had as
• 05:18a first year resident in
• 05:19medicine where my senior resident
• 05:21was always saying, when the
• 05:22disease process is an issue,
• 05:24the answer is in the
• 05:25tissue.
• 05:26Do a biopsy. He always
• 05:28wanted to get the tissue.
• 05:29Right? And, of course,
• 05:32resonates after you you know,
• 05:33you think about every other
• 05:35organ with illness. We've learned
• 05:36a lot by actually directly
• 05:38studying the tissue. But, of
• 05:39course, the human brain is
• 05:40a privileged organ. We don't
• 05:42biopsy
• 05:43biopsy it in in, psychiatric
• 05:45illness.
• 05:47So
• 05:48we have to rely
• 05:49on another approach.
• 05:51And for me, that approach
• 05:52has been
• 05:53using postmortem
• 05:55human brain
• 05:56as a proxy to what's
• 05:57actually happening
• 05:59in the living brain. And
• 06:00I think we need to
• 06:01do this because
• 06:03study of the human brain
• 06:04is the only way to
• 06:05capture
• 06:06the substantial
• 06:07and seemingly
• 06:08weekly discovery of features that
• 06:11are distinctive to the human
• 06:12brain.
• 06:14Secondly, actually examining the tissue
• 06:16permits an evaluation
• 06:18of molecular
• 06:20and structural alterations in the
• 06:21context of specific cell types
• 06:23and local synaptic circuits, things
• 06:25which we still can't get
• 06:26to yet with
• 06:28imaging techniques, and also we
• 06:30can still also look at
• 06:31distributed neural networks in the
• 06:32human brain.
• 06:33It serves as a key
• 06:34intermediary, I believe, between studies
• 06:36and model systems and in
• 06:37vivo studies demonstrating the importance
• 06:40of both of those
• 06:41areas of of research, but
• 06:43looking at them in the
• 06:44context of what we can
• 06:45learn from postmortem human studies.
• 06:47And finally, I think it
• 06:48provides
• 06:50an essential basis
• 06:52for disease related alterations in
• 06:54the brain that can tell
• 06:55us about target identification,
• 06:57what's wrong in the brain
• 06:58that we might develop drugs
• 07:00or devices against,
• 07:01target validation for targets that
• 07:03are discovered through preclinical or
• 07:04animal model systems, and then
• 07:06biomarker development so that we're
• 07:08better at stratifying patients and
• 07:09determining who should receive what
• 07:12treatment.
• 07:13So what I wanna do
• 07:14today and and consistent with
• 07:15what,
• 07:16Matt said in his introduction
• 07:17is to try to illustrate
• 07:18for you a strategy
• 07:20for dissecting
• 07:21the disease process of schizophrenia
• 07:24using studies of the postmortem
• 07:25human brain
• 07:27and
• 07:28the macaque monkey as an
• 07:30informative model system to help
• 07:31us interpret
• 07:33findings that we observe in
• 07:34the human.
• 07:36So I wanna start with
• 07:38dissecting this disease process model.
• 07:40I said the pathology
• 07:41is essential,
• 07:42but
• 07:44we weren't given the pathology
• 07:46when Kraepelin and Breuer defined
• 07:48the illness. What we were
• 07:49given was the clinical syndrome.
• 07:52So the strategy that we've
• 07:53used is, well, how can
• 07:54we start with the clinical
• 07:55syndrome or an aspect of
• 07:57it and then work backwards
• 07:59through this disease process?
• 08:00And so for a variety
• 08:02of reasons, including a lot
• 08:03of work that was done
• 08:04here at Yale, we focused
• 08:05in on cognitive dysfunction
• 08:08as a core and clinically
• 08:10critical feature of the illness.
• 08:12Why? Well, because cognitive dysfunction
• 08:15is highly prevalent
• 08:16in schizophrenia.
• 08:17Virtually everyone who carries the
• 08:19diagnosis
• 08:20regardless of their IQ
• 08:22underperforms
• 08:23cognitively
• 08:24compared to where they should
• 08:25be based upon
• 08:27their inherited cognitive aptitude.
• 08:30Secondly, these cognitive impairments are
• 08:31present and progressive
• 08:33well before the onset of
• 08:34psychosis as early as the
• 08:35first grade of school. Kids
• 08:37who fifteen or twenty years
• 08:38later will be diagnosed with
• 08:39schizophrenia tend to wind up
• 08:40in the bottom half of
• 08:41their class relative to peers.
• 08:43It's persistent across the course
• 08:44of the illness, doesn't wax
• 08:45and wane like psychosis. And
• 08:45Michael Green's, seminal studies replicated
• 08:47by others show
• 08:48that it's not the severity
• 08:50of psychosis, but it's the
• 08:52degree of cognitive impairment
• 08:57that predicts long term functional
• 08:59outcome for people with the
• 09:01diagnosis.
• 09:02And finally,
• 09:03and, fortuitously keeping with the
• 09:05alliteration of p's, we think
• 09:07of these cognitive impairments as
• 09:09a product of impaired cortical
• 09:11network oscillations.
• 09:13So to give you an
• 09:13example of where that,
• 09:15thought emanated from, here's a
• 09:17study that Ray Cho conducted
• 09:18in, our center a number
• 09:20of years ago
• 09:22in which he asked people
• 09:23to perform a cognitive control
• 09:25task that has a working
• 09:27memory component. That is during
• 09:28the delay period of the
• 09:29task, which is illustrated on
• 09:31the x axis here, people
• 09:32have to keep in mind
• 09:33a limited amount of information
• 09:35in order to guide their
• 09:36behavior.
• 09:37And they did this had
• 09:38people do this task with
• 09:39an EEG reporting, and what
• 09:41Ray observed is that bilaterally
• 09:43over the frontal lobes
• 09:45in unaffected
• 09:46comparison subjects
• 09:48during the delay period of
• 09:49the task, there was a
• 09:50heightened bump in the power
• 09:53of oscillations at thirty to
• 09:54forty hertz,
• 09:56gamma frequency oscillations. And as
• 09:57you can see on the
• 09:58right hand side, people with
• 10:00schizophrenia
• 10:01failed to show
• 10:02that increase in gamma band
• 10:04power and then Cam Carter,
• 10:06subsequently replicated
• 10:08exactly the same study in
• 10:09antipsychotic naive subjects.
• 10:11Well, this struck me as
• 10:13really fascinating because I was
• 10:14aware of work from the
• 10:15late John Lisman who had
• 10:17actually done intracranial
• 10:19recordings in humans while they
• 10:21performed a particular working memory
• 10:23task, the Sternberg task.
• 10:24In this task, individuals
• 10:26are shown a set of
• 10:27four stimuli
• 10:28with the instructions remember these.
• 10:30Then after a delay period,
• 10:32a probe card comes up
• 10:33and the individuals have to
• 10:34respond. Does that match one
• 10:36of the four stimuli? And,
• 10:37of course, then after they
• 10:38respond, they've gotta forget it
• 10:39all because the task will
• 10:40start over with a new
• 10:41set of stimuli.
• 10:43And what,
• 10:44John Lisman showed is that
• 10:46as working memory load went
• 10:48up, the number of items
• 10:49to be retained,
• 10:50gamma band power intra, in
• 10:53the prefrontal cortex increased, stayed
• 10:55high while the information was
• 10:56being retained, and then as
• 10:58soon as individuals responded and
• 10:59no longer needed that information,
• 11:01gamma band power dropped.
• 11:04Chen et al. Went on
• 11:05to show using exactly the
• 11:06same task that that in
• 11:07people with schizophrenia, just as
• 11:09in the Cho study, they
• 11:10don't demonstrate
• 11:12an augmentation in gamma band
• 11:14power.
• 11:15Now support for the idea
• 11:17that gamma oscillations is may
• 11:19actually be a neural substrate
• 11:21for working memory as opposed
• 11:22to what some people have
• 11:23called the exhaust
• 11:25of having the network activate
• 11:26that way. Comes from a
• 11:27set of studies. I'll just
• 11:29review some in, mouse models.
• 11:31This is work from
• 11:33by Kasohol at UCSF
• 11:34where they showed that disrupting
• 11:36gamma synchrony impairs working memory
• 11:39performance in mice. And really
• 11:41strikingly, when they restored gamma
• 11:43oscillations
• 11:44via optogenetics,
• 11:45they could improve working memory
• 11:47in a manner that persisted
• 11:48past manipulation.
• 11:50And similarly,
• 11:52Peter Yulhas demonstrated
• 11:54in humans
• 11:55that prefrontal gamma activity
• 11:57actually codes for the number
• 11:59of relevant items maintained in
• 12:01working memory, very similar to
• 12:03the Listman results.
• 12:05So we began to reason
• 12:06then, okay. Maybe as a
• 12:08way to reduce
• 12:10the illness to something that's
• 12:11studied, but we can say
• 12:12we're gonna focus on the
• 12:13impairment in working memory as
• 12:15a core feature.
• 12:16We're gonna correlate that with
• 12:18lower gamma oscillation power, and
• 12:19then we're gonna ask,
• 12:21can we identify
• 12:23a pathological
• 12:24entity
• 12:25in the prefrontal cortex, an
• 12:26alteration in a circuit that
• 12:28generates gamma oscillations that might
• 12:30be the substrate
• 12:32for these abnormalities?
• 12:33Well-to-do that first I need
• 12:35to give you a little
• 12:35journey
• 12:36into the circuitry of the
• 12:37human prefrontal cortex. So left
• 12:40hand side here is the
• 12:41right view of a post
• 12:41mortem human brain. Prefrontal cortex
• 12:44is circled. If we made
• 12:45a coronal cut along the
• 12:46dotted line, you'd see the
• 12:47image on the right. And
• 12:48then if we zone in
• 12:50on a portion of the
• 12:50prefrontal cortex in purple there,
• 12:53look at tissue section put
• 12:54on a microscope slide, stained
• 12:56for initial substance, you see
• 12:57the six layers of the
• 12:58cortex from the outer pial
• 12:59surface to the underlying white
• 13:01matter, layers defined by the
• 13:03relative size and packing density
• 13:04of the constituent cells. And
• 13:06then if we turn up
• 13:07the magnification and look at
• 13:08layer three, you see the
• 13:09image on the right. So
• 13:10you see some unlabeled non
• 13:12neuronal cells
• 13:13and then the two main
• 13:14neurons of the cortex, excitatory
• 13:16pyramidal cells, which are present
• 13:18in about a four to
• 13:19one ratio with inhibitory GABA
• 13:21neurons.
• 13:22And what's particularly striking in
• 13:24deep layer three of the
• 13:26macaque and human cortex
• 13:28is there is an enrichment
• 13:30for a particular population of
• 13:32GABA neurons called the parvalbumin
• 13:34basket cell.
• 13:36And these cells with pyramidal
• 13:38cells form a striking circuit.
• 13:40So the pyramidal cells have
• 13:41a principal axon that goes
• 13:43elsewhere in the brain, but
• 13:44they give off a lot
• 13:45of local axon collaterals.
• 13:48About half of those collaterals,
• 13:50Darlene Wilczewski showed,
• 13:52target the spines of other
• 13:54pyramidal cells. And we think
• 13:55this provides a basis for
• 13:57recurrent excitation, which I'll come
• 13:59back to in a minute.
• 14:00Not necessarily one to one
• 14:01between two cells as shown
• 14:02here, but a way in
• 14:04which pyramidal cells can keep
• 14:06each other,
• 14:07in sustained activity.
• 14:09The other half of those
• 14:11outputs
• 14:12in deep layer three go
• 14:13on to the dendrites of
• 14:15parvalbumin basket cells,
• 14:16which have some few axons
• 14:18that spread very far, but
• 14:20a high density of axons
• 14:22that are arborized nearby.
• 14:24And those axons
• 14:26target
• 14:27the parasympathetic region of the
• 14:29pyramidal cells from which they
• 14:30receive excitatory input, providing a
• 14:32means for feedback
• 14:34inhibition.
• 14:36Well, this circuit
• 14:39seems to be relevant to
• 14:40gamma oscillations for a number
• 14:42of reasons. So first, you
• 14:43know, Pat Goldwyn, Rakesh, and
• 14:45Amy Arnstein demonstrated
• 14:46that working memory depends upon
• 14:49recurrent excitation
• 14:51in layer three in the
• 14:52prefrontal cortex.
• 14:54Work done here under Pat
• 14:56showed that inhibition in layer
• 14:57three shapes pyramidal cell activity
• 14:59during working memory tasks.
• 15:01Franze Conde and our group
• 15:02demonstrated that these PV basket
• 15:04cells are the predominant inhibitory
• 15:06neuron in this laminar location.
• 15:09And then work from Bob
• 15:10Desimone
• 15:11and Eric Miller demonstrated in
• 15:13monkeys that gamma oscillations and
• 15:15especially gamma bursting
• 15:17during the delay period of
• 15:18working memory task occurs principally
• 15:21in layers two and three
• 15:22and not in any layers
• 15:24deeper than that. And then
• 15:25finally more recently,
• 15:28human fMRI high resolution showed
• 15:30that layers two and three
• 15:31in the prefrontal cortex are
• 15:33preferentially
• 15:34active during the delayed period
• 15:35of working memory tasks. So
• 15:36all
• 15:38evidence, you know, that converges
• 15:39upon the idea
• 15:41that this circuit
• 15:43is critical
• 15:44for the generation of gamma
• 15:46oscillations
• 15:47and the performance of working
• 15:49memory task. But how does
• 15:50it generate the oscillations?
• 15:53Well, here's another view of
• 15:54the circuit. So here I'm
• 15:55just showing three pyramidal cells
• 15:57interconnected with one PV basket
• 15:59cell. And you can imagine
• 16:00the situation when there's no
• 16:02active stimulus or no demand
• 16:04for working memory that the
• 16:05pyramidal cells are just firing
• 16:07asynchronously.
• 16:08But then something happens,
• 16:10causes the PV basket cell
• 16:11to spike,
• 16:13and the inhibitory powers of
• 16:15these cells are so great
• 16:17that all of the pyramidal
• 16:18cells that it innervate are
• 16:19hyperpolarized.
• 16:21That hyperpolarization
• 16:23wears off synchronously.
• 16:25Pyramidal cells can spike together,
• 16:27send a volley of inputs
• 16:28back to the PV basket
• 16:29cell. It fires again, hyperpolarizes
• 16:32the pyramidal cells. So it's
• 16:33a beautiful way for creating
• 16:35synchrony among pyramidal cells with
• 16:37a certain periodicity
• 16:39where the periodicity depends upon
• 16:41the decay kinetics
• 16:42of the GABA inhibition of
• 16:44the pyramidal cell. And at
• 16:45this particular synapse based upon
• 16:47the GABA a receptors there,
• 16:48it's about twenty five milliseconds
• 16:50giving rise to a forty
• 16:51Hertz gamma frequency oscillation.
• 16:55So if this is the
• 16:56network, the local circuit that
• 16:58generates gamma, then the question
• 17:00obviously becomes, okay. Now we
• 17:01can ask. What might be
• 17:03wrong in this circuit that
• 17:04gives rise to the,
• 17:06deficits in gamma and working
• 17:08memory and schizophrenia.
• 17:11So can we employ the
• 17:12postmortem human brain to actually
• 17:14probe the integrity of the
• 17:16circuit even though we can't
• 17:17measure the function of it
• 17:18postmortem?
• 17:19Can we just see if
• 17:20there are alterations in it?
• 17:22So the first place we
• 17:23wanna start with, though, is
• 17:26what we call the rule
• 17:27of the five critical c's.
• 17:29And this I'm gonna take
• 17:31some time to talk about
• 17:32this because I think this
• 17:33is in the spirit, doctor
• 17:35Henninger, that we gotta know
• 17:36what we know and how
• 17:37well we know it. And
• 17:38so it really depends on
• 17:40the quality of the work.
• 17:41Our five critical c's are
• 17:43if you see something
• 17:45that's different in the brain
• 17:46of a person with schizophrenia
• 17:47from someone who doesn't have
• 17:48it, could that represent an
• 17:50upstream cause in the disease
• 17:52process?
• 17:53Might it be a downstream
• 17:54detrimental consequence of a cause?
• 17:57Or because the brain is
• 17:58highly,
• 17:59plastic and driven to maintain
• 18:01homeostasis?
• 18:02Could it be a compensatory
• 18:04response to a cause or
• 18:05consequence?
• 18:06Or because as I told
• 18:07you, there's a lot of
• 18:08comorbidities in schizophrenia,
• 18:09maybe it doesn't reflect the
• 18:11disease process. It's just something
• 18:12else that's shared by people
• 18:13with the illness. And then,
• 18:15of course, obviously,
• 18:16we always gotta pretend attend
• 18:18to potential confounds.
• 18:20So before we get to
• 18:21the interesting stuff of cause,
• 18:22consequence, and compensation, I wanna
• 18:24give you a few examples
• 18:25of how we have attempted
• 18:27to
• 18:27understand comorbidity
• 18:29and confounds.
• 18:31So first,
• 18:32lots of comorbidities to talk
• 18:33about, but the two major
• 18:34ones that people worry about
• 18:36are medications and marijuana.
• 18:38K. Medications,
• 18:40principally antipsychotic
• 18:41effects. So the strategy we've
• 18:43used here is what we've
• 18:44called a triangulation
• 18:45approach, you know, thinking about
• 18:47GPS
• 18:48that the more angles
• 18:50the more satellites you have
• 18:51signals from, the more precise
• 18:52you know your location.
• 18:53So we've used three strategies.
• 18:56One is to compare individuals
• 18:57with schizophrenia
• 18:58who are on and off
• 18:59antipsychotics at the time of
• 19:01death, informative, but has some
• 19:02limitations.
• 19:03Another is to compare individuals
• 19:05with mood disorders
• 19:07who had been exposed or
• 19:08not exposed to antipsychotics,
• 19:11some advantages, but also some
• 19:12limitations. And then third and
• 19:13the only way we can
• 19:14actually rigorously control it is
• 19:16to compare monkeys
• 19:18exposed to drugs or not.
• 19:20So I'll give you three
• 19:21examples of these. Here's an
• 19:23example from a study from,
• 19:26Sam Dino in which
• 19:28I've highlighted here
• 19:29antipsychotics,
• 19:31offer on. The dash line
• 19:32there is basically a normalized
• 19:34value for where people with
• 19:36schizophrenia should be, and you
• 19:38can see in both groups
• 19:40whether they're on or off,
• 19:41the magnitude of the alteration
• 19:43is the same. And, likewise,
• 19:44you'll notice from some of
• 19:45the other comorbidities we've looked
• 19:47at here
• 19:48in this particular study
• 19:50don't see an effect of
• 19:51those suggesting that what we're
• 19:52actually indexing is the disease
• 19:54process.
• 19:55Likewise, we can look at
• 19:56other disorders. I'll talk about
• 19:58somatostatin
• 19:59later. In this particular study,
• 20:01Sam looked at a combination
• 20:02of bipolar and major depressive
• 20:04disorder subjects, you know, roughly,
• 20:07divided between,
• 20:08on and off antipsychotics, a
• 20:09good sample size in each.
• 20:11Again, no evident effect of
• 20:13the drugs.
• 20:14And
• 20:15in the third arm, we,
• 20:18exposed monkeys
• 20:19to either haloperidol
• 20:20or olanzapine
• 20:22in a manner that mimics
• 20:23clinical use. So we gave
• 20:25the drugs exactly in the
• 20:26same way that humans receive
• 20:28the drug.
• 20:29We dosed them to the
• 20:30point where they had therapeutic
• 20:32trough levels in the range
• 20:33that are known to be
• 20:34therapeutic in humans.
• 20:36We bandaged the antipsychotic side
• 20:38effects in the same way
• 20:39we would in humans,
• 20:40and we can expose them
• 20:42for two, years to the
• 20:43drug. And as you'll see
• 20:44the studies on the right,
• 20:45we did find some abnormalities
• 20:47in these animals, but they
• 20:48were principally around
• 20:49nonneuronal cells.
• 20:51And so just to cut
• 20:53to the chase,
• 20:54using this triangulation approach
• 20:56in the data that I'll
• 20:57present to you today, we
• 20:58have failed to find evidence
• 21:00that it's explicable
• 21:01by
• 21:02antipsychotic drugs.
• 21:05So what about marijuana?
• 21:08Well, to address this, in
• 21:09people with schizophrenia
• 21:11commonly use marijuana, tends to
• 21:13elicit, relapse.
• 21:16So to address this issue,
• 21:19we got a cohort of,
• 21:21adolescent male monkeys
• 21:23and,
• 21:24divided them into two groups.
• 21:25But every day, everybody went
• 21:26to school, sat in the
• 21:28train a a chamber, did
• 21:29working memory tasks. After school,
• 21:31half of them got an
• 21:32attacks intoxicating dose of THC.
• 21:35The other half got got
• 21:35sham. They went back to
• 21:36the dorm. They watched movies
• 21:37all afternoon, and then we
• 21:39did this
• 21:40for a year.
• 21:41And what we demonstrated
• 21:43is that repeated THC administration
• 21:46did
• 21:49impair working memory.
• 21:50But what we were really
• 21:52struck this is a whole
• 21:52separate story, is that with
• 21:54persistent training, the animals could
• 21:56eventually
• 21:57recover.
• 21:59But when we looked at
• 22:00the tissue from these animals,
• 22:01we failed to find evidence
• 22:03of any of the things
• 22:04I'm gonna show you today
• 22:05are present in schizophrenia.
• 22:07But I do wanna shape
• 22:08that that one of the
• 22:09last things that,
• 22:10Sam did before coming here
• 22:12for residency
• 22:13was he looked,
• 22:15by RNA Seq in layer
• 22:16three
• 22:17of thirteen pairs of these
• 22:19THC and sham animals
• 22:21and found
• 22:22an upregulation
• 22:24of genes that mitochondria use
• 22:25for make energy. This is
• 22:26gonna come up to be
• 22:27important later because what we
• 22:28actually observe in schizophrenia is
• 22:30a substantial downregulation.
• 22:32So, again, it got a
• 22:32way of saying it's probably
• 22:34not
• 22:35marijuana that's causing it.
• 22:37Okay. What about confounds? Again,
• 22:39critically important,
• 22:41unfortunately frequently overlooked.
• 22:43So what we've tried to
• 22:44do to address confounds
• 22:46is
• 22:47to reduce biological variance between
• 22:49groups and control for technical
• 22:51variance
• 22:53by how we design studies
• 22:55with the subjects that we
• 22:56include.
• 22:57So a lot of the
• 22:58data I'll show you today
• 22:59come from this cohort.
• 23:00Sixty two individuals with schizophrenia
• 23:02match one to one with
• 23:03an unaffected comparison subject,
• 23:06for sex and age. The
• 23:08important point is that when
• 23:09you look at the group
• 23:10values,
• 23:11virtually identical for both mean
• 23:12and standard deviation. So we're
• 23:14capturing the variance in equally
• 23:16in both groups that contribute
• 23:18to this.
• 23:19We only use subjects that
• 23:20have excellent tissue,
• 23:22quality metrics, which I've listed
• 23:23there, and then we always
• 23:25process
• 23:26pairs of samples together. So
• 23:28whatever the effect of batch
• 23:30might be or technical variance,
• 23:32it's equally captured
• 23:34in both groups.
• 23:36Another factor that's been a
• 23:38problem in
• 23:40postmortem human research is the
• 23:42design of studies that actually
• 23:43lead to uninterpretable
• 23:45results,
• 23:46and this has been the
• 23:47case for one of our
• 23:48important cell populations of parvo
• 23:50of human cells. If you
• 23:51if you Google,
• 23:53parvo of human cells and
• 23:54schizophrenia, you'll come up with
• 23:55a lot of studies in
• 23:57humans and in rodents saying
• 23:59they're lost,
• 24:00They're missing.
• 24:02And this was partly supported
• 24:03by the fact that we
• 24:04and multiple others have shown
• 24:05that PV message levels are
• 24:07lower, PV protein levels are
• 24:09lower. But when you look
• 24:10at the cell count data,
• 24:11you see a very mixed
• 24:12literature.
• 24:13And we thought that part
• 24:14of the problem and and,
• 24:16is that there was an
• 24:18experimental confound in that parvalbumin
• 24:20is used as both the
• 24:21independent and the dependent measure.
• 24:23So you can't tell. Are
• 24:25the cells not there or
• 24:26is there less parvalbumin per
• 24:28cell such that it fell
• 24:29below the level of detectability
• 24:30or both?
• 24:32So as part of his
• 24:33dissertation, Sam had a great
• 24:34idea for how to address
• 24:35this, and he developed
• 24:37basically a six label fluorescent
• 24:38in situ hybridization approach where
• 24:40you can see on the
• 24:41left, he could label all
• 24:42cells with DAPI.
• 24:43He could identify the lipo
• 24:45fusion, which is a degenerative
• 24:46protein that autofluoresces
• 24:47and found a way to
• 24:49to screen that out. And
• 24:50then he reasoned that we
• 24:51could identify the neurons of
• 24:53interest by using two labels
• 24:56importantly that are not altered
• 24:57in schizophrenia.
• 24:58Vesicular GABA transporter,
• 25:00which labels all GABA neurons,
• 25:02and SOC six which labels
• 25:04MGE neurons which includes the
• 25:06two classes of neurons of
• 25:07interest, parvalbumin and somatostatin cells.
• 25:09And then as you can
• 25:10see on the right, he
• 25:11could pick out the vGAT
• 25:12soc six cells
• 25:14and not only count the
• 25:15cells, but also quantify
• 25:17the level of gene expression
• 25:19per neuron.
• 25:20And what he found I'll
• 25:21just show you the parvalbumin
• 25:22data. On the left,
• 25:24absolutely no difference
• 25:26in the density of neurons
• 25:27in either layers two and
• 25:29four.
• 25:29But on the right,
• 25:31no difference in gene expression
• 25:32in layer two,
• 25:34but a large effect size
• 25:36over one of a deficit
• 25:38in parvalbumin
• 25:38message
• 25:40in layer four. And these
• 25:41findings actually were convergent with
• 25:43a previous study that had
• 25:44used
• 25:45a technique to show that
• 25:46it was layer four and
• 25:47not layer two that was
• 25:49altered.
• 25:49So the important point of
• 25:51this is it really gave
• 25:52us some interpretive power.
• 25:54So it could say, the
• 25:55neurons are there. They're all
• 25:57there.
• 25:58So it was evidence against
• 25:59two hypotheses that had been
• 26:01persistent in the field. One
• 26:03of which was that these
• 26:04cells died due to apoptosis
• 26:06or neurodegeneration,
• 26:07and the other was that
• 26:08they failed to migrate to
• 26:09their normal home. I think
• 26:11this study was really a
• 26:12stake through the heart of
• 26:13both of those hypotheses,
• 26:14but it also told us
• 26:16that they're since they're there,
• 26:17they're a potential target for
• 26:19therapeutic interventions.
• 26:21And then we also think
• 26:23it's critical
• 26:24to know
• 26:25that we have technical validation,
• 26:27replication and biological. So here,
• 26:30Sam used two different cohorts
• 26:31of subjects, two different techniques,
• 26:33but found pretty much exactly
• 26:35the same finding. In this
• 26:36case, lower levels of somatostatin
• 26:38message
• 26:39in most individuals with schizophrenia
• 26:41in the superficial layers of
• 26:43the illness. Okay.
• 26:44So
• 26:46all by way of trying
• 26:47to
• 26:49hopefully convince you that what
• 26:50I'm about to say, you
• 26:52can believe. Okay?
• 26:54That we've attended these things.
• 26:56And for the most part,
• 26:57they'd all show you, we've
• 26:58looked at all those issues.
• 26:59But now we get to
• 27:00the interesting stuff. Cause, consequence,
• 27:02or compensation.
• 27:06So a reason first because
• 27:07of the important work of
• 27:08Pat and Amy that we
• 27:09should look. Is there evidence
• 27:11of impairments in recurrent excitation?
• 27:14And one way to do
• 27:15that, because dendritic spines on
• 27:17pyramidal cells which protrude for
• 27:18them as illustrated in this
• 27:19beautiful
• 27:20electron micrograph on the right,
• 27:22are the main site where
• 27:23the excitatory inputs
• 27:25arrive on pyramidal cells that
• 27:27we would count dendritic spines.
• 27:29And what we found
• 27:31using classic Golgi technique is
• 27:33on the basilar dendrites of
• 27:34layer three pyramidal cells, you
• 27:35can see at the top
• 27:36in an unaffected subject.
• 27:38To me, I always just
• 27:39I love looking at anatomy
• 27:40and these beautiful spines sticking
• 27:42off always,
• 27:43make me excited.
• 27:44Then you see two extreme
• 27:45examples on the bottom.
• 27:47And when Lisa Glance quantified
• 27:49these, she found about a
• 27:51twenty percent deficit in the
• 27:52density
• 27:53of spines on the basilar
• 27:55dendrites of layer three pramyl
• 27:56cells in people with schizophrenia
• 27:58relative to both unaffected
• 28:00and subject of major depression.
• 28:02And then in the same
• 28:03subject, she and Newton Taluri
• 28:04went on to look at
• 28:06pyramidal cells across the layers
• 28:07and noticed that the magnitude
• 28:09of the effect is only
• 28:10in layer three. It's strongest
• 28:12in deep layer three and
• 28:13it's absent in layers five
• 28:14and six.
• 28:16Furthermore, they showed that the
• 28:17length of dendrites on layer
• 28:18three parameters was shorter, meaning
• 28:20that the total deficit
• 28:22in excitatory inputs is probably
• 28:24greater than the deficit suggested
• 28:26by the change in density.
• 28:30So is this evidence
• 28:31of deficient excitatory drive
• 28:34selectively
• 28:35to layer three pyramidal neurons
• 28:36in schizophrenia
• 28:38neurons in which recurrent excitations
• 28:40is essential
• 28:41for working memory?
• 28:43Well, the literature seems to
• 28:45suggest it is because here's
• 28:46a meta analysis
• 28:48using a variety of different
• 28:49approaches.
• 28:50They found that basically a
• 28:51way of indexing these postsynaptic
• 28:53elements in spine
• 28:54lower significantly so in the
• 28:56cortex, but not in subcortical
• 28:57structures in schizophrenia,
• 28:59and strikingly,
• 29:00a huge effect size in
• 29:02layer three.
• 29:03So supporting
• 29:04the idea across studies that
• 29:06there's something going on on
• 29:08layer three pyramidal cells.
• 29:10So we could reason then,
• 29:11maybe we can put together
• 29:14a potential
• 29:15pathological entity, a deficit in
• 29:17dendritic spines on these pyramidal
• 29:19cells, and maybe that's the
• 29:21upstream problem for gamma oscillations
• 29:23and working memory. But what's
• 29:24a reason to think that
• 29:25this is maybe primary?
• 29:27Well, I'll give you a
• 29:28couple of lines of evidence,
• 29:29one of which comes from
• 29:31the GWAS and and, CMB
• 29:33studies in schizophrenia
• 29:34where there's at least a
• 29:36relative enrichment along with other
• 29:37things for genetic liabilities and
• 29:39act in regulation,
• 29:40which is critical for the
• 29:41structural
• 29:42components of neurons. And then
• 29:44work that Dibs Dada did,
• 29:45when he was in Pittsburgh
• 29:47showing that in the CDC
• 29:49forty two signaling pathway, which
• 29:51is also critical for spine
• 29:52maintenance and formation,
• 29:54there seemed to be a
• 29:55cell type
• 29:57specific alteration that is more
• 29:58enriched in layer three pyramidal
• 29:59cells.
• 30:01And then most recently, actually
• 30:03just in the last week
• 30:04or so, a paper in
• 30:05neuron from the Lieber group
• 30:06showing that schizophrenia risk gene
• 30:08convergence is strongest
• 30:10in layer two, three prionylons.
• 30:12So none of this conclusive,
• 30:13but all of which gives
• 30:14us, you know, reasonable premise
• 30:15to saying, yeah, maybe this
• 30:17deficit in dendritic spines in
• 30:19these cells is indexing
• 30:21primary pathology
• 30:22related to the ideological factors
• 30:24of the illness.
• 30:26So we can be in
• 30:27now to set up some
• 30:27postulates and predictions to go
• 30:29after.
• 30:30So postulate,
• 30:31so I've just sent, there's
• 30:32genetic alterations and regulate actin
• 30:35that these are most prominent
• 30:37in layer three due to
• 30:38certain cells,
• 30:39or genes that are selectively
• 30:40expressed there.
• 30:43If this is true, fewer
• 30:44spines, fewer excitatory inputs would
• 30:46reduce excitatory
• 30:47activity of these neurons. And
• 30:49so the prediction
• 30:50is that these layer three
• 30:51pyramidal cells
• 30:52are chronically hypoactive in the
• 30:54illness
• 30:55and they have less drive
• 30:57for mitochondrial energy production since
• 30:59it is neuronal activity that
• 31:01principally tells neurons,
• 31:03to make energy.
• 31:05So is there evidence to
• 31:06support this? Well, I'll give
• 31:07you a couple lines. You
• 31:09know, they're they're they're insufficient,
• 31:10but they're, I think, suggestive.
• 31:13The first is we looked
• 31:14at activity dependent genes expressed
• 31:16in pyramidal neurons. Great example
• 31:17of that is neuronal contraction
• 31:19two, which is expressed by
• 31:20pyramidal neurons in response to
• 31:22neuronal activity
• 31:23and it's secreted from presynaptic
• 31:25axon terminals at glutamatergic
• 31:27synapses onto p v basket
• 31:28cells. So it's particularly interesting
• 31:30for our circuit.
• 31:31What I show you on
• 31:31the left hand side is
• 31:32a study in these sixty
• 31:34two pairs. What's plotted for
• 31:36each dot is that's the
• 31:37value for a matched pair.
• 31:39So the x axis tells
• 31:40you the value for the
• 31:41comparison subject, the y axis
• 31:43the sub value for the
• 31:44schizophrenia subject.
• 31:47The, diagonal line is the
• 31:48unity line, so any value
• 31:51below the unity line means
• 31:53in that pair,
• 31:54the subject with schizophrenia
• 31:56has a lower value than
• 31:57it should.
• 32:00And in this study, it
• 32:01was over eighty percent
• 32:02of people with schizophrenia
• 32:04have abnormally low levels of
• 32:05neural patraxin two. Very similar
• 32:07findings for BDNF, which is
• 32:09also an activity dependent pyramidal
• 32:12neuron, transcript. And then we
• 32:13recently showed that the effect
• 32:15size of both of these
• 32:16transcripts predicts
• 32:18alterations in GABA neurons, which
• 32:19which I'll come back to
• 32:20later.
• 32:22What about energy production? Well,
• 32:23to address that, Dominique Arianne
• 32:25used laser microdissection to individually
• 32:27capture layer three pyramidal neurons
• 32:29in thirty six pairs of
• 32:30subjects, pulled the neurons, subjected
• 32:32them to microarray,
• 32:33and we were struck with
• 32:35the fact that the gene
• 32:37pathways
• 32:38that were most altered in
• 32:39these
• 32:40all include
• 32:41genes that mitochondria use to
• 32:43make energy.
• 32:44And this
• 32:46deficit
• 32:47seemed to be enriched in
• 32:48layer three pyramidal cells because
• 32:49the magnitude of the alteration
• 32:50in the illness was much
• 32:52greater than it was in
• 32:54gray matter from the prefrontal
• 32:55cortex as a whole.
• 32:58So we can expand the
• 32:59pathological entity now to say
• 33:02fewer dendritic spines,
• 33:04and we infer that it
• 33:05results in hypoactive layer three
• 33:07pyramidal neurons.
• 33:09So vast to our postulates
• 33:10and predictions, we've got a
• 33:12cause,
• 33:13fewer spines, a consequence,
• 33:16less activity, reduction in,
• 33:19mitochondrial energy production. And I
• 33:21would just add into the
• 33:22side, you know, since we're
• 33:23always just looking at associations
• 33:25in these findings, that the
• 33:27GWAS
• 33:28data really tend to suggest
• 33:30that the mitochondrial genes are
• 33:32not involved as risk factors
• 33:34in the illness the same
• 33:35way that the structural and
• 33:37postsynaptic
• 33:38genes are.
• 33:39So the prediction would be,
• 33:40as I mentioned at the
• 33:41outset, given the brain
• 33:44tends to maintain
• 33:45balance,
• 33:46we would predict
• 33:48that this reduction in excitatory
• 33:50activity
• 33:51would be paralleled by some
• 33:53kind of reduction in inhibition
• 33:55to maintain some level
• 33:56of excitatory inhibitory balance.
• 33:59So we wanted to know,
• 34:00is there evidence of reduced
• 34:02inhibition
• 34:03at the parval of that
• 34:04basket cell input to pyramidal
• 34:06cells?
• 34:07I'll quickly show you several
• 34:08studies in which,
• 34:10Alison Curley and then, Ken
• 34:12Fish
• 34:13measured protein levels of GAD
• 34:14sixty seven in parvalbumin basket
• 34:16terminals
• 34:19and found
• 34:20it's lower in people with
• 34:21schizophrenia.
• 34:22And since GAD sixty seven
• 34:24is the principal
• 34:26regulator of GABA levels consistent
• 34:29with
• 34:29lower inhibition on the presynaptic
• 34:32side, and then Joe Glaussier
• 34:33looked at the postsynaptic receptor
• 34:35of this synapse and found
• 34:38lower levels of the receptor.
• 34:39So a combination
• 34:41pre and postsynaptic
• 34:42downregulation
• 34:43of inhibition consistent with lower
• 34:45strength here. And I would
• 34:47add the fact that it's
• 34:48down on both sides,
• 34:49I think tends to rule
• 34:51out the idea that the
• 34:52primary problem is in the
• 34:53GABA neurons. Because if the
• 34:54primary problem was less GABA,
• 34:55you would have expected the
• 34:56receptors perhaps to be upregulated.
• 35:00So, again, approximate cause,
• 35:02consequence.
• 35:03Now I've suggested a compensation
• 35:06in inhibitory neurons, and we're
• 35:08pleased that this is supported
• 35:10by work done, by Adams.
• 35:11I think, Wally was here
• 35:13from computational modeling
• 35:15showing that
• 35:17you can explain
• 35:19the the data better if
• 35:20you pause it a primary
• 35:22disturbance in excitatory inputs
• 35:24with a compensatory response
• 35:26in interneurons rather than vice
• 35:27versa.
• 35:30But
• 35:32as always, it's not that
• 35:33simple, and compensations can be
• 35:35accompanied by negative consequences. You
• 35:37know, we were all
• 35:39exposed to this in a
• 35:40way we didn't wish in
• 35:41the early days of COVID
• 35:42when, you know, people were
• 35:44dying, not so much of
• 35:45the virus, but at the
• 35:46cytokine
• 35:47storm that was a compensatory
• 35:49response to the virus.
• 35:51And so we've wondered in
• 35:52the brain,
• 35:54are there homeostatic
• 35:56responses
• 35:57to maintain inhibitory, sedentary balance
• 35:59that are really I I'm
• 36:00gonna be anthropomorphic here where
• 36:02the brain is saying, you
• 36:03know,
• 36:04I don't want stupor. I
• 36:05don't want seizures. I'm gonna
• 36:06do anything
• 36:07to keep out of that.
• 36:09And if it costs me
• 36:10some working memory capacity,
• 36:12that's a good trade off.
• 36:13Right?
• 36:14So to try to
• 36:17ask
• 36:17whether these compensatory
• 36:20creative compensatory changes
• 36:22in PV cells
• 36:24might actually also be contributing
• 36:26to the impairments in gamma
• 36:27oscillations,
• 36:28we took the studies of
• 36:30lower strength there, which about
• 36:32a ten percent deficit,
• 36:34in combination with two studies
• 36:35that Wange Chung did. First,
• 36:37he found that there were
• 36:38fewer inputs, excitatory inputs to
• 36:40PB basket cells, which is
• 36:41consistent with the deficit in
• 36:43NPTX two that I showed
• 36:44you earlier. And he also
• 36:45found a striking increase in
• 36:47variance
• 36:48of the strength of these
• 36:49synapses.
• 36:50And so while he and,
• 36:52Matt Jaramita were residents,
• 36:54they teamed up,
• 36:55to do a,
• 36:57computational
• 36:58study where they were able
• 37:00in computational model
• 37:01to vary
• 37:03the mean strength of inhibitory
• 37:05inputs to excitatory neurons, which
• 37:06is shown here
• 37:08on the x axis, the
• 37:09number of excitatory inputs to
• 37:10inhibitory neurons, the connectivity on
• 37:12the y axis, and the
• 37:14variance in strength
• 37:15of those inputs,
• 37:16CV,
• 37:17on the,
• 37:19z axis.
• 37:20And,
• 37:21you know, we as I
• 37:22said, these are all kind
• 37:24of relatively small change. And
• 37:25in fact, when they altered
• 37:26only one of these in
• 37:27the model, they got relatively
• 37:29small declines in gamma power.
• 37:31But when they put all
• 37:32three in the model, there's
• 37:33a mag a large deficit
• 37:35in gamma power more on
• 37:36the order of magnitude that
• 37:37we see in people with
• 37:39schizophrenia.
• 37:40So it's a nice use
• 37:41of computational modeling to provide
• 37:43proof of concept evidence
• 37:45that multiple modest alterations in
• 37:47cortical circuitry, including
• 37:49the changes
• 37:50in the parvo of human
• 37:52basket cell to pyramidal cell
• 37:54circuit that we thought of
• 37:56as compensations
• 37:57could actually contribute to the
• 37:59disturbances
• 37:59in gamma power.
• 38:01So here, add a little
• 38:02bit more to the pathological
• 38:04model. If you respond slower
• 38:06activity, detrimental compensation,
• 38:09Then we begin to wonder,
• 38:10well, can we find out
• 38:12which cell types are affected
• 38:14and when do these alterations
• 38:15arise?
• 38:17This is important because if
• 38:18you look at the longitudinal
• 38:20data from the Dunedin study,
• 38:22at age seven, people with
• 38:24schizophrenia
• 38:24underperform cognitively as I mentioned
• 38:26earlier, but principally on areas
• 38:28of concept formation and reasoning.
• 38:30But it's only later in
• 38:32development
• 38:33when the impairments in attention
• 38:34and working memory emerge. So
• 38:36these working memory impairments appear
• 38:38to arise or to at
• 38:40least increase during the post
• 38:41puberty to adult transition.
• 38:43So we wonder, is there
• 38:45a cell population
• 38:46that's actively
• 38:48developing its spines
• 38:50during that time. It could
• 38:51be vulnerable
• 38:52and the existing data or
• 38:54or recent data suggests that
• 38:55it might be those cells
• 38:56in layer three that project
• 38:58colossally.
• 38:59This is interesting
• 39:00because
• 39:01that intrahemisphere
• 39:02connection is critical
• 39:04for the transfer of working
• 39:05memory traces in monkeys.
• 39:07And what Dominique Arian showed
• 39:09in the paper that was
• 39:10just published last month that
• 39:11the transcriptome
• 39:13of closely projecting layer three
• 39:15pyramidal neurons matures after puberty
• 39:18later than other cell types
• 39:19that are their neighbors in
• 39:21the same layer and region.
• 39:23And this late maturation
• 39:25involves expression shifts in multiple
• 39:27gene pathways involved in synaptic
• 39:29function.
• 39:30And then,
• 39:31in these same monkeys,
• 39:33Guillermo had demonstrated that that's
• 39:35when spines
• 39:37decline,
• 39:38during normal development.
• 39:40So this led to the
• 39:41hypothesis
• 39:42that we're actively working on
• 39:44that these layer three colosally
• 39:46projecting neurons are preferentially vulnerable
• 39:48to excessive prune of spines
• 39:49and excitatory synapses in schizophrenia.
• 39:52And this converges with recent
• 39:54work,
• 39:55from Dibsdata and Amy Arnstein
• 39:57where
• 39:58the CP transcriptome and parabola
• 40:00neurons that we identified
• 40:02in their RNA seek data,
• 40:04they think these cells look
• 40:05like what they called the
• 40:06cuts two b neurons that
• 40:08are enriched in calbindin calcium
• 40:09related and stress responsive proteins.
• 40:11It's contact with their earlier
• 40:12work. They make these cells
• 40:14preferentially vulnerable.
• 40:16And so what this sets
• 40:17up is an opportunity to
• 40:18test this hypothesis in postmortem
• 40:20studies. So can we get
• 40:21better cellular resolution?
• 40:23Alright. So now stay in
• 40:25the model a little bit,
• 40:27but here's the caveat.
• 40:29And this comes from the
• 40:30Nobel laureate Daniel Kahneman who
• 40:32said, our confidence in an
• 40:33explanatory story. That's what I've
• 40:35been trying to pitch to
• 40:36you today. We've got an
• 40:37explanatory story.
• 40:38It rests not in an
• 40:40evaluation of the evidence and
• 40:41its quality. Now I will
• 40:42say I've tried to demonstrate
• 40:44to you at the outset
• 40:45that you can trust the
• 40:46quality, but I haven't given
• 40:47you all an opportunity to
• 40:49examine the evidence because we're
• 40:50moving pretty fast, but rather
• 40:52in the coherence of the
• 40:53story. So if I told
• 40:54you a story that fits
• 40:55together,
• 40:56But, of course, he says,
• 40:58it's easier to construct a
• 40:59coherent story when there are
• 41:00fewer pieces to the puzzle.
• 41:02The less you know, the
• 41:03easier it is to explain.
• 41:05Right. So what happens if
• 41:07we add more pieces to
• 41:08the puzzle? And I'm just
• 41:09gonna show you one, other
• 41:10GABA neurons.
• 41:12So here's a schematic model
• 41:13on the right. Our layer
• 41:14three pyramidal cell, there's the
• 41:16parablebumin basket cell that I've
• 41:17talked about. And there are
• 41:18two other major cell populations
• 41:20in the cortex, not exclusively
• 41:22this, that are located more
• 41:23superficially,
• 41:25the calretinin
• 41:25containing neuron, and a population
• 41:27of somatostatin cells.
• 41:30We and multiple other groups
• 41:31have shown that calretinin message
• 41:33just doesn't change in schizophrenia,
• 41:35but somatostatin
• 41:36message is substantially downregulated.
• 41:38And what's striking to us
• 41:40about that is that those
• 41:41calretinin cells do not get
• 41:43input
• 41:44from layer three pyramidal cells.
• 41:47But Darlene demonstrated
• 41:49that somatostatin
• 41:50cells in the superficial layers
• 41:52get these local axon collateral
• 41:53inputs or at least the
• 41:55evidence suggests that they do
• 41:56in the same way that
• 41:57the parvalbumin basket cells do.
• 42:01So we ask the question
• 42:02and,
• 42:03find Sam's final paper, we're
• 42:05still in revision,
• 42:06is is somatostatin neuron inhibition
• 42:10to pramilar neuron dendrites lower
• 42:12in schizophrenia? We can't address
• 42:13that at the level of
• 42:15the actual synapse, but we
• 42:16can get an inference of
• 42:17it by, again, looking at
• 42:19the expression of key genes,
• 42:20in this case, somatostatin
• 42:21and gad sixty seven selectively
• 42:23in somatostatin neurons.
• 42:25And what Sam demonstrated
• 42:27in a discovery cohort is
• 42:29that both of these transcripts
• 42:31are downregulated.
• 42:32He replicated that in another
• 42:34cohort, combined the cohorts,
• 42:36and found that the two
• 42:37transcript levels were highly correlated.
• 42:40And what's shown here
• 42:42are scores that are, kinda
• 42:44normalized for where they should
• 42:45be relative to unaffected comparison
• 42:47subjects.
• 42:48So any circle in the
• 42:49lower left quadrant
• 42:51refers to a person with
• 42:52schizophrenia who has lower levels
• 42:54of both transcripts.
• 42:55So then we decide, well,
• 42:57what the heck? Let's create
• 42:58a presynaptic
• 42:59index of dendritic inhibition, which
• 43:01is the combination of these
• 43:02two transcripts.
• 43:03And here you can see
• 43:04that more than eighty percent
• 43:06of people with schizophrenia
• 43:08show a deficit
• 43:09in this index.
• 43:10Why this is really interesting
• 43:12is that somatostatin dampens the
• 43:13excitability of targeted neurons and,
• 43:13of course,
• 43:14somatostatin dampens the excitability of
• 43:15targeted neurons and of course,
• 43:17gad sixty seven regulates
• 43:19the amount of GABA available
• 43:20for fast inhibition.
• 43:22So together the findings suggest
• 43:25that the input to the
• 43:26distal dendrites of layer three
• 43:27pyramidal cells
• 43:29from an inhibitory basis of
• 43:31the somatostatin cells is lower
• 43:33in schizophrenia.
• 43:35Interestingly, work that Monica Vigneto
• 43:37had done earlier than that
• 43:38for a different reason
• 43:40found that the postsynaptic
• 43:41receptors, the SSR
• 43:43t r two and the
• 43:44GABA a five receptor were
• 43:46also lower in schizophrenia.
• 43:48So this is reminiscent of
• 43:49what I'd show you about
• 43:50parvalbumin basket cells that we've
• 43:52got downregulation
• 43:54presynaptically
• 43:55and postsynaptically,
• 43:58suggesting that it again is
• 43:59secondary to layer three pramelone
• 44:01hyper hypoactivity.
• 44:03And because computational models
• 44:05suggest that dendritic inhibition of
• 44:07pramel neurons inhibits
• 44:08or dampens the effects of
• 44:10distractors during working memory tasks,
• 44:12perhaps the alteration in somatostatin
• 44:14cells
• 44:15gives us a convergent explanation
• 44:17for working memory impairments because
• 44:18we know people with schizophrenia
• 44:20can't resist distractors during a
• 44:22working memory task in the
• 44:24same way with unaffected.
• 44:25So maybe the PV basket
• 44:26cell disturbance is contributing to
• 44:28the gamma oscillation,
• 44:30the somatostatin
• 44:31disturbance to the failure to
• 44:32resist distractors.
• 44:34Okay. So added a little
• 44:35bit more, to the pathological
• 44:37entity here,
• 44:39but
• 44:40to conclude, are we approaching
• 44:42are we near the final
• 44:44proof? Right? Well, I just
• 44:45wanna
• 44:46check the explanatory story I've
• 44:48been telling you. So first,
• 44:50try to convey to you
• 44:52that the findings are reliable.
• 44:54They're replicated across studies, excluding
• 44:56false positive. They're convergent across
• 44:58methods, excluding technical artifacts.
• 45:00Believe that they're rigorous and
• 45:01that they index the disease
• 45:03process because we've, to the
• 45:04extent possible, excluded
• 45:06comorbidities. And I haven't shown
• 45:08you the chronicity data, but,
• 45:09again, these findings don't look
• 45:10like they're related to how
• 45:11long people have been ill.
• 45:14Well, do they all really
• 45:15fit with the explanatory story?
• 45:17Well, I think they do,
• 45:18but there's some other stories
• 45:20that still have to be
• 45:20considered. One is maybe
• 45:23the clinical illness emerges only
• 45:25when alterations in both inhibitory
• 45:28and
• 45:29excitatory neurons occur.
• 45:31And one reason to pay
• 45:32more attention to that is
• 45:33a study that's in press
• 45:35from Laramie,
• 45:36Duncan at Stanford where she
• 45:39looked at cell type
• 45:43specific expression,
• 45:45of all the two hundred
• 45:46eighty seven loci
• 45:48that are risk genes for
• 45:49schizophrenia
• 45:50across the four hundred and
• 45:51sixty one cell types. And
• 45:53busy slide to explain, but
• 45:55what she was struck and
• 45:56we're struck by is that
• 45:57you look at the top
• 45:57of that list as our
• 45:59favorites, somatostatin and parvalbumin,
• 46:02saying that maybe these cells
• 46:03are in fact harboring some
• 46:04of the genetic risk, and
• 46:05so we should be thinking
• 46:06about them perhaps not just
• 46:09as secondarily compensated.
• 46:11And then, we're also pleased
• 46:13to see that layer two,
• 46:14three pramyl neurons pop out,
• 46:16but the calretinin cells, they're
• 46:17not here.
• 46:18So, again, kind of a,
• 46:20you know, proof of concept
• 46:21that maybe we're onto the
• 46:22right circuit.
• 46:24And, of course,
• 46:26a whole other story. No
• 46:27time to talk about it
• 46:28today, but we've also gotta
• 46:29think about you know, I've
• 46:30acted like
• 46:32schizophrenia is one brain region,
• 46:34three neurons.
• 46:35You know?
• 46:36It's easy.
• 46:39But, you know, we've also
• 46:41been looking at whether some
• 46:42of these findings
• 46:44might be either secondary to
• 46:45or convergent with other disease
• 46:47processes such as alterations in
• 46:48other cortical regions or in
• 46:50the projections from the thalamus.
• 46:52But then the real bottom
• 46:54line is, does this maybe
• 46:55advance us anywhere towards restoration,
• 46:57the goal that doctor Henninger,
• 46:59you know, dedicated his career
• 47:01to.
• 47:02Well, I think they suggest
• 47:03some really interesting possibilities.
• 47:06One is at the level
• 47:07of the synapse.
• 47:09Maybe there's strategy
• 47:10being called neuroplasticens
• 47:12for
• 47:13enhanced
• 47:15specific types of excitatory inputs.
• 47:17So avoid off target
• 47:19problems getting at,
• 47:21increasing excitatory tone
• 47:23at particular synapses.
• 47:25It's also recent development of
• 47:27being called sonogenetics
• 47:28with the idea that you
• 47:29can genetically sensitize
• 47:31specific cell types to ultrasound
• 47:33stimuli.
• 47:33So combine a
• 47:35biological manipulation with a device
• 47:37based simulation. And then at
• 47:39the circuit, you know, there's
• 47:40some promising proof of concept
• 47:42studies
• 47:43that we can actually manipulate
• 47:44the circuit with external
• 47:46modulation. So Cam Carter has
• 47:48a proof of concept study
• 47:49that's showing prefrontal
• 47:51transcranial direct current stimulation increases
• 47:53gamma band power in people
• 47:55with schizophrenia and enhances cognitive
• 47:56control. And then there was
• 47:58this more recent study showing
• 47:59that transcranial
• 48:00alternating current stimulation,
• 48:02modulating synchronous activity improve working
• 48:04memory. So still to come,
• 48:06but I think a reason
• 48:08for sober optimism.
• 48:10So here's
• 48:12the model we're working on.
• 48:14Will understanding the disease process
• 48:16as we've illustrated here ultimately
• 48:18lead to rational and personalized
• 48:20interventions? Well, story still to
• 48:22be told. But I hope
• 48:23and I will conclude with
• 48:25the concluding sentence
• 48:27of Hecker's monograph
• 48:29over a hundred and fifty
• 48:30years ago in which he
• 48:31said,
• 48:32perhaps I might hope that
• 48:34my work has made a
• 48:35not unwelcome contribution toward furthering
• 48:37clinical research in psychiatry,
• 48:39and I think that's,
• 48:40my hope today.
• 48:42So I wanna acknowledge the
• 48:43many people who've contributed to
• 48:45this work and especially
• 48:46acknowledge and thank
• 48:48the many family members
• 48:50who had a very difficult
• 48:51time in their lives,
• 48:53generously caved consent for brain
• 48:54tissue donation,
• 48:56and then continue to work
• 48:57with us through a set
• 48:58of exhausting and exhaustive interviews
• 49:01so that we would have
• 49:02the privilege of reconstructing
• 49:04the life history of their
• 49:05loved one in order to
• 49:06make not only the tissue
• 49:08available for study, but also
• 49:10have rich,
• 49:11metadata to interpret it. So,
• 49:13again, thank you very much
• 49:14for being here, Fudge.