Child Study Center Grand Rounds 02.09.2020

March 22, 2021

Information

Pregnancy Well-Being the Epigenome and Child Development

ID6313

To CiteDCA Citation Guide

00:00Here and and then Karen take it from there.
00:05So may I call Doctor Tara Thompson Felix.
00:09Take it away.
00:10Doctor Thompson feelings, yeah,
00:12so good afternoon everyone.
00:13My name is Tara Thompson.
00:15Felix I'm one of the first year child
00:18Psychiatry Fellows so I actually met
00:20Doctor O'Donnell a few months ago
00:22on virtually in one of our breakout
00:24sessions and grand rounds and just
00:27heard a lot about his research who
00:29really got me excited because I've
00:31done some research and in utero,
00:33and epigenetics and I really wanted
00:35the opportunity to kind of explore
00:37that more so Doctor O'Donnell
00:39has been awesome and discussing.
00:41Potential projects with me.
00:43Since then an I am very excited
00:46to announce that I will be a
00:48PhD student in his lab.
00:50Starting in July,
00:52so I'm very excited to pass it
00:55along to Doctor Karen O'Donnell.
00:58Thank you.
01:00But
01:00congratulations again, Tara Anne.
01:02I'm very happy that you got into
01:05the program and delighted that
01:06you'll be working in the lab,
01:08and I'll be looking forward to when you
01:11were doing Grand Ryans showing some.
01:13Hopefully you're very interesting data,
01:15fetal exosomes and how they are shaped
01:17by exposure to prenatal adversity
01:19and how they can inform on child in
01:22your development and Doctor Martin.
01:24Thank you very much for that
01:26kind introduction for not giving
01:28the game away about where I am.
01:30Chrome I thought I would start by giving
01:33you a little bit of a background on
01:36how and where I've come from an end to
01:38end up here at the Child study Center
01:41and as Doctor Martin mentioned, an.
01:42If this is an experiment between
01:44the Department of Citrix and
01:46the Child Study Center,
01:47I'm more than happy to be a subject in
01:50this study because it's such a pleasure
01:52to be acting as a bridge between these
01:55two Fantastic Department's an so as
01:57Doctor Martin at knows I actually,
01:59I'm from the West coast of Ireland.
02:01A very small village,
02:02around 200 people an on the
02:05western coast of Ireland,
02:06very close to a small town called Belona.
02:09It's right on the Atlantic coast.
02:11The red line that you're seeing on
02:14this map is what's called the Wild
02:16Atlantic way that has a roadway that
02:19hugs the Atlantic coast of Ireland.
02:21And for those of you who are
02:24avid cyclists like Doctor Martin,
02:26it's 1500 miles and that you
02:28can cycle around our lender,
02:30driver and arnensee some fantastic sites.
02:32Such as the Stone Age settlement
02:34that's around 20 minutes from my home,
02:37Amore Dan Patrick head.
02:38I can tell you it took a long
02:41time to take this photograph.
02:42This isn't representative of the
02:44weather that we have in Ireland,
02:46Ann,
02:47but there are some beautiful scenes
02:49to be had in the West Coast of
02:51Ireland and up until very recently
02:53that would have been the most
02:55famous thing about where I'm from.
02:58This wonderful coastline and
02:59this Stone Age settlement.
03:00And but then Joe Biden got elected.
03:03And his ancestral home is around 20 minutes
03:06from my home in a small town called Belle.
03:10And all you can see is distant relatives
03:13celebrating when the election was announced,
03:15and mural that remains in
03:17our small town and were,
03:19among other works.
03:21So where everyone was very excited by this,
03:24this is actually the 2nd president
03:26that belona can lay claim
03:28to because Mary Robinson,
03:30the first woman President of Ireland,
03:32also hails.
03:33From Belona,
03:34so a little fun fact,
03:36but as Doctor Martin mentioned,
03:38I my trading didn't occur in
03:40Ireland and I had to travel a
03:42little bit further East for that,
03:44and that was to London where I
03:47completed my undergrad Masters and
03:48eventually my PhD where I worked
03:50with Vivek Glover and who's an
03:53expert and perinatal cycle biology
03:54and also with Tom O'Connor.
03:56Andthat PhD was actually a little
03:59bit of an experiment at the time as well.
04:01It was in an NIH funded.
04:04PhD occurring in London using the Avon
04:06Longitudinal Study of Parents and Children,
04:09which is along the Tunal perspective
04:11cohort of around 15,000 pregnancies
04:13where these children have been
04:15followed up continuously there,
04:16now approaching their 30s themselves,
04:18having children of their own,
04:20and I'll talk to you a little
04:23bit about that cohort today.
04:25Now following my PhD,
04:26I moved back out West but much
04:29further West than where I'm from.
04:31An ended up at McGill
04:33University where I completed it.
04:35Post doctoral fellowship with Michael Meaney,
04:38who many of you will know and
04:40it's really been a pioneer in
04:43the field of social epigenetics.
04:45So how the environment can shape
04:47variation in the epigenome?
04:49When I talk about the epigenome,
04:51I'm talking about chemical marks
04:53or modifications that sit on or
04:56close to the genome that can change
04:58the way the genome functions and
05:01throughout all of my training.
05:03Really, what's been at the heart of my.
05:06Fascination,
05:06really with science is the idea that
05:09the early environment can shape health
05:11and disease across the lifespan.
05:14Ann and this notion or this idea has
05:17been described as the developmental
05:19origins of health and disease and which
05:22has led to a whole field of research.
05:25A SoC adore had society that was
05:27really an largely based on findings
05:30from the work of David Barker.
05:32In fact,
05:33it used to be referred to as the Barker.
05:37Hypothesis that and the fetal
05:39origins of disease,
05:40that the the origins of many
05:43types of disease could be traced
05:45back to the neutral environment
05:48and these observations initially
05:50stemmed from David's work,
05:52where he notice to parallel between
05:55high rates of infant mortality
05:57and subsequent rates of death
05:59from coronary heart disease,
06:01and these were largely in deprived
06:04areas in the United Kingdom.
06:07And what you're looking at here is
06:09the relationship of risk from death
06:11for death from coronary heart disease
06:13as a function of birth weight.
06:15I want David Barker noticed,
06:17was that lower birth weight was
06:19associated with an elevated or an
06:21increased risk from death from coronary
06:24heart disease before the age of 65,
06:26and you can see a greater decline in
06:28risk as we move to larger birth weights.
06:31These are birth weights
06:32expressed in pounds and answers,
06:34and then maybe a slight uptick.
06:37In at disease risk,
06:39and as I mentioned,
06:41this is really led to really an
06:43expansive literature on how the neutral
06:46environment can shape vulnerability
06:49for cardiovascular disease,
06:50and really a whole host of
06:54metabolic phenotypes.
06:55But of course,
06:56the question is if the cardiovascular
06:58system is so sensitive to the unusual
07:01environment into adversity in usual,
07:03what about the brain?
07:05And as we heard from Kartek last
07:08week and from Amanda study as well,
07:11we know that the prenatal
07:13environment can also shape variation
07:15in brain related phenotypes.
07:17So, for example,
07:18at the Dutch hunger,
07:20winter and the Holocaust have all
07:22been associated with increased
07:24risk for adverse.
07:25Mental health outcomes in the
07:27offspring and in the next generation,
07:30and these have largely been
07:32from retrospective studies where
07:33this evidence first emerged.
07:35And of course from our own work
07:37with the Avon Longitudinal
07:39Study of Parents and Children.
07:41What we found is that maternal
07:43prenatal anxiety and also
07:44maternal prenatal depression,
07:46associate's with an increased
07:48risk for adverse mental health
07:50outcomes in the child and what
07:52you're looking at here is the
07:54predicted population prevalence.
07:56Of a probable mental disorder at
07:59children from age 4 all the way
08:01up to age 13 an and what we can
08:05see is that those children born
08:07to women that experience high
08:09levels of anxiety in the prenatal
08:12period have approximately double
08:13the risk of ending up in the group
08:16that is likely to suffer from
08:19a probable mental disorder,
08:20and we see this elevated risk across
08:23childhood and into early adolescence,
08:25and indeed.
08:26As studies follow up,
08:28studies have been completed
08:29now into the early 20s,
08:31and Sean a similar pattern of
08:34Association between high rates of
08:36prenatal anxiety and depression
08:37and increase risk for adverse
08:39mental health outcomes.
08:41And I just want to point out that
08:43these effects are independent
08:45of socioeconomic status.
08:47So you may think that this may be
08:49confounded by maternal education
08:51or maternal age,
08:53or indeed taxol crowding or
08:55obstetric outcomes.
08:56Birth weight,
08:57gestational age.
08:57Because of the large sample
09:00size of this court,
09:01we can statistically control for
09:03the effects of those exposures,
09:06and we still see this independent
09:09Association with maternal prenatal anxiety.
09:11So for any of you that have
09:14heard me speak before,
09:16I always talk about 414 and 44 being
09:18one in four women that are likely to
09:22experience or struggle with their
09:24mental health in and around pregnancy.
09:27So I say one in four other estimates
09:30they went in five, one in six,
09:33and really what I think we're seeing
09:35from these epidemiological analysis,
09:38particularly the more recent
09:39epidemiological analysis,
09:40is increased rates.
09:41Of perinatal mental health problems.
09:44So,
09:44for example,
09:45Louise Howard Publishing in 2018 and
09:47one in four women struggling with their
09:50mental health in and around pregnancy.
09:53Rebecca Pearson,
09:53using the Avon Longitudinal Study
09:56of Parents and children using the
09:58second generation from that cohort.
10:00Showed a generational increase in
10:03rates of perinatal mental health
10:06problems with again one in four
10:09women experiencing perinatal
10:11mental health problems.
10:13So this is a common problem and
10:16what is challenging with this
10:18problem is that we're still not
10:20screening women effectively,
10:22and when we do screen women and
10:24they're still not receiving
10:26adequate treatment,
10:27so we know that around 25% of
10:30women who do experience perinatal
10:32mental health problems receive
10:34treatment and less than 5% of
10:36those women achieve remission
10:38or experience receive adequate
10:40treatment to reduce their symptoms
10:42down below clinical levels.
10:43So this is a common problem that we are
10:46not addressing sufficiently at the moment.
10:49It is also a costly problem,
10:51so we know that the per
10:53year costs of untreated
10:55perinatal mental health problems is
10:57around 14 billion US dollars per year,
11:00with around 40% of those costs attributed
11:03to the adverse effects on the child.
11:05So that's in the United States.
11:07What about the in the UK where the
11:10first cost estimate was produced?
11:12Well, we see that around.
11:14An 8 billion pounds is the cost
11:17associated with untreated paradata mental
11:19health problems in the United Kingdom,
11:22but in contrast to the United States,
11:25we see that 72% almost 3/4 of those
11:27costs are attributed to the adverse
11:30effects of untreated perinatal mental
11:32health problems on child outcomes,
11:35and you may ask,
11:36rightfully So what is the difference
11:38between these two cost estimates?
11:41Why is it 40% in the United States and 72%?
11:45In the United Kingdom or one
11:47of the explanations for that is
11:49because in the United States,
11:51costs were only calculated on
11:53child outcomes from zero to five,
11:56whereas in the United Kingdom costs
11:58were calculated from zero to 18.
12:00So I think you can appreciate that if
12:03we extend the follow up period in the
12:06United States that 40% an proportion
12:08of costs is likely to increase in
12:11addition to the total costs that have
12:14been reported from that cost analysis.
12:16In the United States.
12:19And of course,
12:20you may ask about what is the impact
12:22of the post Natal environment.
12:24And of course,
12:24we know that there is an effect
12:26of the post Natal environment,
12:28and in our studies from the Avon
12:30Longitudinal Study of Parents and Children,
12:32we see that perhaps,
12:34as you would expect,
12:35children that are exposed to high
12:37levels of anxiety in the pre and post
12:39Natal period are the children who do
12:41worse than children who are exposed
12:43to anxiety at one but not both.
12:45Time points end up somewhere in the middle.
12:48Now hasten to add, this is not a treatment.
12:51And study this is not an intervention.
12:54This is simply an epidemiological analysis,
12:56but I think it provides proof of
12:59principle that if we could pull down
13:01or reduce maternal symptoms of anxiety,
13:04ideally at both time points,
13:06we could improve child outcome.
13:10So again, just thinking about what are the
13:13consequences for untreated perinatal mental
13:15health problems on the next generation?
13:18Well, I showed you the effects on child
13:21outcomes from 4 to 13 years of age,
13:24but this is another study
13:26from the same cohort.
13:28the Avon Longitudinal Study of Parents
13:30and Children where they looked at the
13:33rates of prenatal depression in women
13:36that were born to ALS back moms who
13:39either didn't or did experience.
13:40Prenatal depression,
13:41and so when we look at the daughters of women
13:46who didn't experience prenatal depression,
13:48we see that around 16% of those
13:51women who want to experience prenatal
13:53depression in their own pregnancies.
13:56So what about the daughters from women
13:58who did experience prenatal depression?
14:01Or 54% of those women went on to experience
14:04prenatal depression in their own pregnancies.
14:08So I think you can begin to appreciate.
14:11Add there can be marked intergenerational
14:14effects of exposure to prenatal
14:16depression and this makes it critically
14:18important that we try to support
14:21pregnant women and their mental
14:23health both for the pregnant woman's,
14:25all mental and physical health,
14:27but also to potentially mitigate the
14:30effects of this intergenerational
14:33transmission of risk.
14:34But there are some unanswered questions
14:37and many of you will be looking at and
14:40some of the slides that I presented
14:42in some of the data that I presented
14:45and thinking about the advances
14:47that we've made in characterizing
14:48genetic variation and thinking,
14:50well,
14:50isn't this just all confounded
14:52by underlying genetic propensity
14:54for psychiatric disorders and and
14:55there have been many studies that
14:58have attempted to address this,
14:59perhaps indirectly,
15:00and perhaps the most well established or
15:02well known paper to address this question.
15:05Was at this study by Hanigan and
15:08colleagues based in the mobile
15:10quarter large Norwegian study of
15:13around 30,000 pregnant women,
15:15Ann and they used a children of children
15:19and sibling children of Twins design.
15:22Basically it's a twin study that
15:25looks at the offspring and their
15:28conclusion was that a genetic
15:30factor that explained relatedness
15:33between Twins and siblings was.
15:36The explanation for the effects of
15:38prenatal depression on child outcome.
15:40So they concluded that the entire
15:43fetal origins hypothesis was
15:45confounded by genetic variation using
15:47an indirect assessment of genetic
15:50variation using a twin design.
15:52So we wanted to revisit this question
15:55to really ask and provide a more
15:58direct test of this hypothesis
16:00as to whether or not there was
16:03confounding by underlying genetic
16:05propensity or genetic vulnerability.
16:07For adverse mental health outcomes
16:10now many of you will have heard from
16:13our recent ground round sessions.
16:16The use of polygenic risk scores
16:18to capture common variation that is
16:21associated with psychiatric disorders
16:23and the use of these genetic tools
16:26has really an been greatly facilitated
16:29by these incredibly large genome
16:31wide Association studies largely
16:33conducted by the Psychiatric Genomics
16:35Consortium where we have.
16:3710s or hundreds of thousands of
16:40individuals with a psychiatric disorder
16:42and where we look at the snips,
16:44the genetic variants that are associated
16:47with the psychiatric disorder
16:49and that provides us with an
16:51effect size for that snip and
16:53risk of psychiatric disorder.
16:55Now what we can do is take those
16:58effect sizes and we can count
17:00up and using our own data using
17:03genetic data from our own court,
17:05we can count up the number of.
17:08Risk snips that an individual carries,
17:11and we can wait.
17:12Each one of those snips by the
17:15effect size that has been derived
17:18from these very large scale genome
17:21wide Association studies and what
17:23you get is a simple summary score
17:26that reflects an individual's
17:28genetic vulnerability for adverse
17:30mental health outcomes,
17:31whether it be ADHD, schizophrenia,
17:33or depression.
17:34So the general principle is that once
17:37we calculate this summary score.
17:40And this polygenic risk,
17:41or you will generally see that
17:44cases or individuals that have
17:46high risk for psychiatric disorder,
17:48generally have a higher score
17:51than non cases or controls.
17:54So we use this methodology and
17:56we can talk about the limitations
17:58of this methodology and in the
18:01question period and but we use
18:03this methodology as what we think
18:05of as the best approach at the
18:08moment to capture genetic risk
18:10for psychiatric disorders.
18:11We calculated these polygenic risk
18:13scores in the Outback children
18:15around just over 5000 children.
18:17We calculated them for ADHD,
18:19schizophrenia and depression and then
18:21we used child mental health symptoms,
18:23derives from the strength.
18:25Difficulties questionnaire from age
18:284 to 16 years of age and then we use
18:32long to tude ainle model modeling.
18:35Generalized estimating equations to
18:36ask whether or not the prediction
18:39from maternal prenatal depression
18:42or maternal prenatal anxiety was
18:44confounded by child genetic risk for ADHD,
18:47schizophrenia, or depression.
18:49And.
18:49And the take home message from these
18:53analysis was that even when we
18:55adjusted for child genetic risk for ADHD,
18:58schizophrenia,
18:59or depression,
19:00we still saw a significant
19:02independent effect of maternal
19:03prenatal depression on child outcome.
19:05And this is just a representative
19:08figure using the ADHD polygenic
19:10risk score and what you can see is
19:13that children with a high burden of
19:16genetic risk for ADHD and exposed to
19:18high levels of maternal prenatal depression.
19:21Show increased symptoms relative to
19:23children with low genetic risk for
19:26ADHD and low and exposure to low
19:28levels of maternal prenatal depression,
19:30depression and we see this for
19:33the external Ising subscale.
19:34We see this for the total symptom
19:37scores and we see this at four years
19:41of age but also at 16 1/2 years of age.
19:45Now,
19:46one of the interesting observations
19:48from this study was that there
19:51was no interaction with time.
19:53We found a stable prediction from
19:56maternal prenatal depression overtime.
19:58Conversely, for both the schizophr.
20:00Any other polygenic risk score
20:02and for the depression polygenic
20:05risk or we found a significant
20:07interaction with time where the
20:09polygenic risk score for schizophrenia
20:12or depression strengthened as these
20:14children approached adolescence,
20:15so perhaps in the question period
20:18we can discuss how developmentally
20:20dynamic symptoms or phenotypes may
20:22require developmentally informed
20:24genome wide Association studies,
20:26but going back to the question
20:28at hand as to whether or not.
20:32Effects of the prenatal environment are
20:34confounded by child genetic variation.
20:36At least from this study,
20:39we can see that our best efforts to
20:41assess genetic risk for psychiatric
20:43disorders doesn't seem to confound
20:46the Association between maternal,
20:48prenatal depression or maternal
20:50prenatal anxiety and child outcome,
20:52and we do see an independent significant
20:55prediction from child genetic risk
20:58factors with the child ADHD PRS
21:00being the strongest predictor.
21:03Now for many years we spend a lot of
21:05time talking about the importance
21:07of maternal mental health,
21:09but of course maternal mental health
21:11can be associated with many other
21:13phenotypes that are also at risk
21:15factors for adverse mental health
21:17outcomes and one of the other
21:19exposures that we were particularly
21:21interested in assessing and perhaps
21:23especially relevant in the context
21:24of an ongoing global pandemic,
21:26was the role of maternal infection,
21:28and again with the idea of trying
21:31to understand whether or not there
21:33could be synergy between.
21:34Maternal prenatal anxiety or depression
21:37and maternal infection to produce
21:40an works outcomes for the child and
21:44what you're looking at here in this
21:47slide is symptoms from the social
21:49communication disorder checklist,
21:51which can be thought of as essentially
21:54symptoms related to autism like features,
21:57and we cut,
21:59characterized or assessed maternal
22:01infection in pregnancy,
22:02and we particularly focused on.
22:05Infections that may give rise to systemic,
22:08an inflammation and infection an
22:10and what we found was that the
22:12number of maternal infections was
22:14associated with increased symptoms.
22:16Scores for the social communication
22:19disorder checklist and the question
22:21was whether or not maternal
22:23anxiety would have an independent
22:24effect and multiplicative effect,
22:26and what we found was indeed
22:29an additive effect.
22:30An independent additive effect
22:32of maternal prenatal anxiety,
22:33and infection on child symptoms.
22:36We saw this first social
22:37communication disorder checklist,
22:39but also for symptoms of 80 HD.
22:42So,
22:42just to summarize this first part of my talk.
22:45I think that what we've documented
22:47using the OS backward is that there
22:49can be a persisting influence of the
22:51prenatal environment on child outcome,
22:53and we don't think that this is completely
22:56confounded by child genetic risk factors.
22:58Could it be amplified by
23:01genetic variation in the child?
23:03That's an open question and we have
23:06published papers previously showing
23:08evidence of gene environment interactions
23:11and the prediction of child outcome.
23:14Really highlights is that there are
23:17multiple opportunities to intervene to
23:19try and improve maternal mental health,
23:21ideally as early as possible in
23:23pregnancy and certainly early
23:25in the post Natal period.
23:27Of course,
23:28I think our data also speak to the
23:30importance of considering maternal
23:32physical health as another point of
23:35intervention to ensure that we can
23:38bolster both maternal well being but
23:40also potentially improve child outcome.
23:43Now one of the characteristics
23:45of this research area,
23:47the developmental origins
23:48of health and disease,
23:49is that there can be marked
23:51variation or marked individual
23:53differences in the effects of the
23:55prenatal environment on child
23:57outcome, and the question is,
23:59how can we better identify children
24:01that are at risk and to try and get at
24:04this question or address this question?
24:07And I moved to Montreal to
24:09study social epigenetics with
24:11Michael Meaney and that's really.
24:13Features heavily in my current
24:15research program because epigenetics,
24:17really, while it's heavily involved
24:19in cellular differentiation,
24:20there was a paradigm shift in
24:22the early 2000s where we began to
24:25appreciate that the environment could
24:27also shape epigenetic modifications.
24:29But before we get into that,
24:32I think it's helpful to start with a
24:35definition of epigenetics and I like
24:37add this definition that comes from
24:40the road map project and which is really.
24:43Markable initiative that sought to
24:45act as a parallel to the Human Genome
24:49Project and to characterize different
24:52epigenetic modifications across the
24:54genome and across different cells
24:56and tissues and integrate those
24:58data to provide a richer perspective
25:00and a deeper understanding of the
25:03epigenome across cells and tissues.
25:05Now,
25:06what many of you on the call
25:09will probably be aware of is the
25:12very controversial area of.
25:14Transgenerational epigenetic inheritance,
25:15which posits that epigenetics
25:17states can be transmitted across
25:19multiple generations with Fidelity,
25:21and the evidence for that in
25:23humans is lacking an.
25:25As I mentioned,
25:27it is a very controversial subject,
25:29and there is an excellent review by
25:32Edith Heard for any of you that are
25:35interested in getting a having a
25:38deeper dive into this controversy,
25:41but also the evidence we can see.
25:44Evidence for transgenerational
25:45epigenetic inheritance in C.
25:47Elegans.
25:47In certain plants,
25:48an Indra Sofala fruit flies and but
25:51again trying to establish that evidence
25:54in humans is particularly challenging.
25:57It doesn't rule out the possibility an,
26:00but there is no clear evidence for
26:03that in humans at the current time.
26:06But when I think about epigenetics
26:09and really the definition that
26:11I use in my work is different.
26:14The genetic states or epigenetic
26:16modifications that can alter the
26:18transcriptional potential of a cell,
26:20or indeed a system and what I mean by
26:23that is directly related to gene expression,
26:27so epigenetic modifications have the
26:29potential to alter gene expression,
26:31and that's one of the reasons that
26:34people are so interested in the
26:36epigenome trying to understand how
26:39these epigenetic modifications can
26:40alter the function of the genome.
26:43And as.
26:44Doctor Martin very kindly pointed
26:46out we've written a review on
26:49the evidence for and against the
26:52epigenome underlying the biological
26:54embedding of experience and what
26:56we conclude from this review is
26:59that there is quite a lot of a good
27:02correlational evidence suggesting
27:04that the epigenome may underlie the
27:07biological embedding of experience,
27:10but trying to establish causality
27:12does require model or.
27:14Organisms and I think,
27:16will be greatly facilitated by the
27:18advent of EPI genome editing technology,
27:21where we can actually directly
27:23manipulate in a site specific
27:24manner and different epigenetic
27:26States and establish functional
27:28associations with gene expression
27:30and different brain based phenotypes.
27:32Now the modification that
27:34I'm going to spend most of
27:36my time talking about today
27:38is that of DNA methylation,
27:41which is the addition of a
27:43methyl group are represented.
27:45Here in red to a cytisine,
27:47that's a C in the genetic code,
27:50Anne Anne, but I also want to point
27:52out from this figure from this
27:54review that this is one of many
27:57different epigenetic modifications.
27:59In fact, some people call them
28:01epigenetic systems that work in
28:03conjunction with one another,
28:05and as we make progress in our
28:07understanding of the epigenome,
28:09and indeed, social epigenomics,
28:11we're beginning to realize the
28:13importance of integrating different
28:14layers and levels of information.
28:16About the epigynum.
28:18To fully understand its impact
28:21on genome function.
28:22So let's think about this
28:24in more simple terms.
28:26I think it can be very helpful to think
28:29about the epigenome in terms of metaphor.
28:32And so some great metaphors exist to
28:34try and add describe the epigenome.
28:37I particularly like the idea of
28:39the epigenome as a conductor,
28:41so as sheet music and as a conductor.
28:44So we think about genes being the
28:46individual and instruments or musicians.
28:48And really,
28:49if we want to create a Symphony to create.
28:53A phenotype that makes sense.
28:55It's important that all of these
28:57different units and work together
28:59in a coordinated manner,
29:00and one of the ways that they
29:03do so is by following the signs
29:06of the signals of the conductor.
29:08One of the other metaphors that I
29:11love to use to describe how the
29:13epigenome influences the function
29:15of the genome is that of grammar,
29:18and so you can have all of the
29:21correct letters and text.
29:23In a book, but if you don't have punctuation,
29:27if you don't have grammar,
29:29then you lose all meaning and we all
29:32know that grammar can be critically
29:35important for our understanding of text,
29:38and similarly with the epigenome.
29:40Epigenetic modifications are critically
29:42important for placing emphasis on
29:44certain genes or silencing other genes,
29:47so really playing a functional role.
29:50Now,
29:50historically we've thought
29:51about DNA methylation as being
29:53a repressive modification.
29:55People have likened it to a light switch,
29:58so turning a gene on turning.
30:00Aging off and the evidence really
30:03to support DNA methylation
30:04as a repressive modification.
30:06Cones from X inactivation where DNA
30:09methylation plays a role in silencing
30:11one of the X chromosomes an in females,
30:15but also from an imprinting where
30:18there can be silencing of 1 copy
30:21of a gene for ad that occurs
30:23in a parent of origin fashion.
30:26But we've begun to realize that we
30:28as we add more deeply characterized
30:31DNA methylation.
30:32Is that it's Association with gene
30:34expression can be more nuanced.
30:36In some cases it can act like a dimmer
30:40switch, turning gene expression up,
30:41or Dan.
30:42Indeed in other situations,
30:44demethylation is not associated
30:45with gene expression,
30:46and in other cases still we can find
30:49the DNA methylation at certain sites
30:51within a gene can actually alter
30:54the product or the splice variant
30:56that's produced from a given gene.
30:58I think the take home message is that
31:01the context is critically important.
31:03Another cool curring epigenetic modifications
31:05can also have an impact on whether
31:08or not DNA methylation is negatively
31:10associated with gene expression or
31:12positively associated with gene expression,
31:15or indeed not associated
31:17with gene expression at all.
31:20Now, one thing to consider when we look
31:22at DNA methylation across the genome is
31:25that DNA methylation is a binary event,
31:28it's either on or it's off.
31:30But throughout my talk you'll hear me talking
31:33perhaps about percentage DNA methylation,
31:3590% DNA methylation, 60% DNA methylation,
31:38or 10% DNA methylation, and that is
31:40because when we look at DNA methylation,
31:43particularly in clinical studies,
31:44we're looking at an average across
31:47multiple cells, and so when we look within.
31:50Multiple cells we can see that there may
31:53be methylation at a given site in one cell,
31:56but not in another,
31:58and so when we report back DNA methylation,
32:01an results were talking about
32:03it as percentage metalation.
32:05Essentially,
32:05the number of metalation marks within
32:07the cells of your tissue of interest.
32:10And that brings me to one of the issues
32:13with epigenetics in clinical studies,
32:16and that is the rule of
32:18cellular heterogeneity.
32:19So one of the principle
32:21rules of the epigenome.
32:22Is to ensure that there is cellular
32:25differentiation and the maintenance
32:27of those cellular phenotypes,
32:29and in fact where you have disorders
32:32related to DNA methylation.
32:34Another epigenetic modifications.
32:36You can require pluripotency increase
32:38the stemness of these cells,
32:40giving rise to disorders and
32:43diseases such as cancer.
32:45But one of the other interesting
32:48features of the epigenome and one of
32:51the functions that is emerging for the
32:53epigenome is the idea of genomic priming.
32:56So this is the idea that there can
32:59be an exposure that gives rise to a
33:02change in an epigenetic state such
33:05as DNA methylation, and that am,
33:08instills,
33:08or instantiates the capacity to then
33:11have an even greater response to an
33:13exposure subject to subsequent exposures.
33:16That an individual or sell may
33:18experience as subsequently,
33:20and this is really nicely articulated
33:22in this paper from my colleague
33:24Nadine Provincal,
33:25working with Elizabeth ****** where
33:27they treated hippocampal stem
33:29cells with dexamethasone,
33:30which is a synthetic glucocorticoid.
33:32And you can think of it like
33:35a synthetic cortisol,
33:37an that produced widespread changes
33:39in DNA methylation and what was
33:41interesting about this particular study
33:43was that the changes in DNA methylation.
33:46Didn't always correlate with the gene
33:49expression response to dexamethasone,
33:51but the DNA methylation changes that
33:53did occur did predict the magnitude
33:56of response to subsequent exposures
33:58to dexamethasone,
33:59supporting this notion of genomic priming,
34:02and you may be asking,
34:04well, how could that occur?
34:07What would be the molecular mechanism?
34:09Well,
34:10one of the reasons that we're interested
34:13in steroid hormones such as cortisol,
34:16progesterone, estradiol,
34:17testosterone.
34:17Is because they are there.
34:19Their receptors are nuclear receptors.
34:22So when you have high levels of
34:24glucocorticoids such as cortisol,
34:26they can bind to the glucocorticoid
34:29receptor highlighted here in Gray,
34:31and the binding of that receptor
34:33to the DNA can
34:35result in DNA demethylation or changes
34:38in DNA methylation at the site that
34:41the transcription factor binds.
34:43So here you can see before
34:46exposure to glucocorticoids.
34:47You have higher levels of DNA methylation
34:50at this particular site or glucocorticoid
34:53response element then you have
34:55glucocorticoids binding to its receptor,
34:57resulting in changes in DNA methylation
34:59and then when you have subsequent
35:02exposures to glucocorticoids,
35:03you then have enhanced response
35:05to that exposure.
35:07And I think this is a particularly
35:10interesting hypothesis and model
35:12when we think about the effects
35:14of prenatal adversity or early
35:16adversity and how that may.
35:18Confer or prime the genome for subsequent
35:23exposures or responses to those exposures.
35:27So how do we analyze DNA methylation?
35:30Well,
35:30there are many approaches that we can use.
35:33We can use an epigenome wide
35:35Association study or metalation
35:37wide Association study an if we use
35:40whole genome bisulfite sequencing,
35:42we can assess roughly around 24 million
35:44CPG's more commonly because of cost.
35:47We're using an microarray based technology
35:49where we assess around 850,000 sites.
35:52Now, what you can quickly appreciate
35:54is that you're going to need very.
35:57Large courts to to adjust for
36:00multiple comparisons with so many
36:02sites and so what's promising in
36:04this regard is the PACE consortium,
36:07which is a consortium that's combining
36:10multiple different studies to
36:12perform meta analysis of prenatal
36:14exposures on DNA methylation.
36:16So they have performed a meta analysis
36:19of maternal prenatal smoking and
36:21DNA methylation and cord blood,
36:23and found over 2000 sites that
36:26survived genome wide.
36:27Adjustment and then the figure that
36:30you're looking at here on the right
36:32is showing all of the sites in
36:35blue and red across the different
36:37chromosomes in the human genome
36:39that were associated with infant
36:41birth weight in cord blood and
36:44from around 9000 participants.
36:46Now,
36:46one of the challenges with this approach is,
36:49as I said, you need very large sample sizes,
36:53but you also ideally would
36:55need to have longitudinal data.
36:57So for example in the birth weight
36:59study that I'm talking about here,
37:02they identified around 900 CPG's that
37:04were associated with birth weight
37:06for a subset of those participants.
37:09They then had longitudinal data and
37:11what they found was that of those
37:14900 sites only around 10% of them.
37:17We're still associated with
37:19birth weight at 7 years of age,
37:22and this highlights a complexity with
37:25epigenetic analysis that you don't have
37:28as it's not as strong a confounder
37:30with genome wide Association studies.
37:33This idea that there can be
37:35dynamic change in DNA methylation
37:37requiring longitudinal sampling.
37:40So what approaches can we
37:42take to overcome these issues?
37:44Well,
37:45one approach that has I've used extensively.
37:48Is that of an biomarkers?
37:50Epigenetic biomarkers that distill
37:52down and genome wide data into
37:55a single unit of measurements,
37:57and perhaps the most well
38:00established of these biomarkers,
38:01is that of epigenetic age,
38:04initially developed by Steve Horvath,
38:06an at UCLA Ann,
38:08and the idea with these epigenetic
38:10biomarkers is that we can identify
38:13sites that are predictive
38:15of chronological age,
38:17and we can create.
38:18A measure of epigenetic
38:20age for an individual.
38:22These clocks now exist with
38:24multi tissue predictors,
38:25so you can take any biological
38:27sample from anyone and you can then
38:30measure their epigenetic age and
38:32what we notice in population levels
38:34is that there are some individuals
38:37that you'll hire epigenetic age,
38:39relative chronological age and
38:40others that show lower epigenetic
38:42age relative to their chronological
38:44age and what's interesting is
38:46that those individuals with higher
38:48epigenetic age acceleration.
38:50Show increase risk for age related?
38:55Disorders including cardiovascular disease,
38:58but also all cause mortality.
39:01Now, one of the challenges with
39:03this epigenetic Clock from the
39:05multi tissue epigenetic Clock
39:07is that it was developed using
39:09primarily samples from adults and
39:11they ranged in age from zero to 100,
39:14but it was primarily samples from
39:16adult participants and the error in
39:18the prediction of the epigenetic
39:20Clock is around 3.6 years and
39:22which obviously is a very long
39:25time in the life of a child.
39:27So we set about creating a
39:30novel pediatric specific.
39:31Epigenetic Clock,
39:32which was published last year.
39:34We used approximately 2000 DNA
39:36methylome's and we simply asked what
39:38were the sites that were associated
39:41with chronological age in this cohort.
39:43This is data from longitudinal cohort
39:45where we use the original epigenetic
39:47Clock with longitudinal samples,
39:49and what you can appreciate from this is
39:52that the slopes are all over the place.
39:55An long digital samples that
39:57should be epigenetically older
39:59are appearing epigenetic.
40:01Younger and you can see this again here,
40:03and this simply reflects the error
40:05in the conventional epigenetic Clock.
40:07When we plot these data using the
40:09new pediatric epigenetic Clock,
40:11I think you can appreciate that the
40:13slopes become a lot more positive,
40:15so we brought the error in prediction of
40:18epigenetic age down to around six months,
40:21and many of you may be thinking,
40:23well, you know, that's great.
40:25You can just calculate someones
40:26age based on their date of birth.
40:29What is what value is this?
40:31An and so in this particular study,
40:34what we found was that children
40:36of the autism spectrum disorder
40:38had accelerated epigenetic age,
40:40an Association that we saw with the
40:42pediatric specific epigenetic Clock,
40:44but not with the conventional Horvath Clock.
40:47But of course,
40:48bringing us back to the topic of interest,
40:51the fetal origins of health and disease,
40:54we wanted to ask whether or not
40:56maternal prenatal anxiety would
40:58be associated with epigenetic age,
41:00acceleration,
41:00and to do that we made use of
41:03two longitudinal at courts,
41:05one from the Netherlands.
41:06That's primarily Caucasian one
41:08from Singapore,
41:09that's multi ethnic and what we found
41:11was that maternal prenatal anxiety
41:13was associated with accelerated
41:15epigenetic age at six years of age.
41:17In the 10 years of age in the Bible
41:20course and again we replicated
41:23this in the coastal court,
41:25finding that maternal prenatal
41:26anxiety was associated with
41:28accelerated epigenetic age,
41:29an effect that strengthens overtime
41:32is particularly pronounced at 48
41:35months of age. Now, one of the questions
41:37that again I'm very interested in is
41:40is trying to understand whether or not
41:41there are features or aspects of the
41:44Pulcinella environment that may be able
41:46to buffer or moderate the effects of the
41:48prenatal environment on epigenetic states.
41:50Because of course it's very
41:52depressing to to give a talk and say,
41:54well, it's all over at birth,
41:56and of course it's much more optimistic
41:58and positive to say that there are
42:00potential interventions that we can
42:02implement that may buffer or mitigate
42:04the effects of prenatal adversity.
42:06This is a paper from my PhD mentors
42:08showing that an infant attachment style,
42:11so each child's perception of
42:13the predictability an index of
42:16the quality of care in the pools.
42:18Naval environment moderates the
42:20Association between prenatal cortisol
42:21exposure and child cognitive development.
42:23Of course, other examples exist.
42:25This is from the Boukris Early
42:28Intervention Project,
42:28showing that secure an infant attachment
42:31can buffer or moderate the effects of
42:34early adversity on child psychopathology.
42:36So of course,
42:37the question we wanted to ask with
42:39this study was whether or not infant
42:42attachment would buffer or moderate the
42:44effects of maternal prenatal anxiety
42:46on child epigenetic age acceleration.
42:48And this is unpublished data.
42:49But what we find is that yes indeed in
42:52children that have secure attachment
42:53we see a positive but nonsignificant
42:55Association between maternal prenatal
42:57anxiety and child epigenetic age
43:00acceleration,
43:00but the effects of maternal prenatal
43:02anxiety on child epigenetic age
43:04acceleration are particularly
43:05pronounced in children.
43:06With an insecure attachment style.
43:09Again supporting this idea of a potential,
43:12pools Natal moderation of infant attachment.
43:16Now of course there are other M
43:19epigenetic biomarkers that we can
43:21use to try and probe our describe
43:23the effects of the environment
43:25on health related outcomes.
43:27This is one that we're starting
43:29to make use of.
43:31It's a second generation after genetic
43:33Clock and what is different about
43:35this epigenetic biomarker is that it
43:37incorporates information about plasma
43:39proteins that are associated with
43:41cardiovascular disease risk as well
43:43as sites that are associated with.
43:46Aging and we wanted to determine
43:48whether or not there was any Association
43:50between an early adversity and this
43:53epigenetic biomarker making use
43:55of the Nurse Family Partnership,
43:57which many of you will know is
43:59a randomized control trial of
44:01the perinatal intervention that
44:02targets vulnerable low income.
44:04First time moms and it provides
44:06nurse visitations have been shown to
44:09reduce child maltreatment an improve
44:10outcomes for both mothers and children.
44:13We published the first epigenetic
44:15analysis in this cohort.
44:16A collaboration with Jim Lechman and
44:19Elena Grigorenko when she was based
44:21here and we found that there was some
44:24preliminary Association between nurse
44:26Visitation and variation in DNA methylation.
44:29But really the take home message was
44:32that there was a profound effect of
44:35childhood maltreatment on DNA methylation,
44:37but we couldn't distinguish the
44:39effects of maltreatment from,
44:41say, for example,
44:42the effects of associated
44:44confounders like smoking.
44:45So what about?
44:47This measure of epigenetic age,
44:49acceleration in the context of the Nurse,
44:52Family, Partnership,
44:52or what we see is that children with
44:55a documented or substantiated case of
44:58child maltreatment show accelerated
44:59epigenetic aging using this disease.
45:02Relevant epigenetic biomarker.
45:03But what about when we break this
45:06down by an intervention group or
45:08what we find is that in the nurse
45:11visit a group in purple here and
45:14the yellow is the control group.
45:16We find no difference in
45:18epigenetic age acceleration.
45:19As a function in those individuals that
45:22don't have a history of child maltreatment.
45:25But when we look in the group that does
45:27have a history of child maltreatment,
45:30we see significantly increased
45:31an epigenetic age,
45:33acceleration and those individuals that
45:34have a history of child maltreatment
45:37that are in the control group.
45:39But it seems that exposure to
45:41nurse Visitation to that early
45:42intervention seems to be buffering
45:44the effects of child maltreatment.
45:46An epigenetic age acceleration.
45:48Now we can discuss.
45:50Potential explanations for this
45:51one possibility is that perhaps
45:53the severity of abuse was less in
45:56the nurse visited group that there
45:58was greater surveillance of abuse,
46:00and the nurse visited group an.
46:02An alternative hypothesis is that
46:04the early intervention is providing
46:06some buffering capacity,
46:07so even in the face of child maltreatment,
46:10there's less of an impact on
46:12epigenetic age acceleration just
46:14in the last couple of minutes.
46:16I just like to tell you about one of
46:19the biomarker that we're making use of.
46:22Which is a measure that relates
46:24to this paper I highlighted,
46:27previously speaking to this
46:28idea of genomic priming,
46:30and in this paper they created an epigenetic
46:33biomarker of glucocorticoid exposure,
46:35and so this essentially we can create
46:38a an index or a proxy measure for
46:41glucocorticoid exposure based on
46:43DNA methylation, and so we created.
46:46We use this array.
46:47Tested this out in a court
46:50where we had DNA methylation.
46:52Data upper than at one year of age in a
46:55cohort from the University of California,
46:58Irvine,
46:58and we also had structural imaging in
47:01this cohort and what we simply asked
47:03was whether or not the sites that
47:05were associated DNA methylation sites
47:07that were associated with maternal
47:09prenatal depression did they overlap
47:11with the sites that were identified
47:13to be glucocorticoid sensitive sites
47:15in that paper that I showed you.
47:17And indeed,
47:18we found significant enrichment
47:19of glucocorticoid sensitive sites
47:21in the sites that were associated
47:23maternal prenatal depression.
47:24And when we created this Google
47:27Corticoid exposure score,
47:28we saw a significant negative Association
47:30between maternal prenatal depression
47:32and this glucocorticoid exposure score.
47:34And interesting Lee,
47:35what we found was that this glucocorticoid
47:38exposure score at birth predicted
47:40lower hippocampal volume birth,
47:41and as you'll appreciate,
47:43the hippocampus is enriched
47:45for glucocorticoid receptors.
47:46So we find that the direction of
47:48this Association is consistent
47:50with a higher maternal prenatal
47:52liberation predicting a lower score.
47:55And a lower score predicting
47:58lower hippocampal volume.
47:59So, just to summarize,
48:01I think that with some of the
48:03studies that I've tried to
48:05highlight perhaps very quickly
48:06today, we can see that variation in
48:09DNA methylation is associated with
48:10variation in the early environment.
48:12I think as we move towards trying to
48:15make these findings clinically relevant,
48:17we need to move towards more integrative
48:19models where we're incorporating
48:20measures of genetic variation,
48:22and we're incorporating an greater
48:24measures of the social environment,
48:26and I think one way that we
48:28can really begin to probe.
48:30Causal associations between the social
48:32environment and epigenetic states
48:34is through the use of interventions,
48:36and this is an area that I'm
48:39particularly keen to do more work in,
48:42and one collaboration that I'm very
48:44excited about is a cluster randomized
48:46control trial of parental intervention
48:48that begins in early pregnancy that
48:51seeks to reduce prenatal anxiety
48:53and depression but also provide an
48:56information about nutrition and sleep,
48:58trying to reduce domestic violence, an.
49:00An increase female empowerment and
49:02we're doing this in rural Vietnam
49:04with my colleague James Fisher,
49:06where one in three women can
49:08experience or struggle with their
49:10mental health and pregnancy.
49:12We're just coordinating to receive
49:14samples from approximately 1200 mothers
49:17and their infants with biological
49:18samples at birth at 12 months and a 24
49:21months MA which have been collected
49:23in parallel with standardized
49:25measures of child newer development.
49:27And really the goal with these
49:29kind of studies and the goal of.
49:32Understanding epigenetic States and
49:34modifications and implementing them
49:36in clinical studies is really to
49:38try and understand how we can make
49:40interventions work from war individuals,
49:42so I'll leave it there with maybe
49:45just one kind of call to action.
49:47I was very pleased to be invited to
49:49take part in the Scientific Council
49:52of Postpartum Support International,
49:54and this is a plug for their
49:56national strategy on how we can
49:59improve perinatal mental health.
50:00And so I think this is a societal problem
50:03that requires a societal response,
50:06and I think we're all responsible
50:07for playing our part and trying
50:10to support perinatal mental health
50:12and recognizing that there are
50:14structural and societal factors
50:15that we can target and to try and
50:18improve perinatal mental health.
50:19And this isn't just all pregnant mothers
50:22and have another thing to worry about an,
50:25so I'll leave it with that and just
50:28thank you all for your attention and take.
50:31Any questions?
50:41Fantastic. Questions please.
50:43Just go for it or put it in the text.
50:51Hi, this is Flora. Do you hear me?
50:54Yes yes Laura hi sorry hi,
50:57how are you really? Nice talk.
50:59I had a question um so so as you
51:02know epigenetics are very much
51:04self type an organ specific so
51:07perhaps you can clarify for us.
51:10I mean of course studies in humans
51:13cannot be done in brain whereas
51:16studies in animals can and I'm
51:18assuming some of those that you.
51:21Elucidated or talked about where
51:23done in mouse or rat brains, right?
51:27So perhaps,
51:28given the course profound difficulties,
51:31you know?
51:32Same brain samples from humans
51:35living individuals.
51:37Is it been any study in animals that has?
51:41Um illuminated this concept to
51:43what extent peripheral samples
51:45like blood can inform us on what's
51:48actually happening in the brain or
51:50the individuals as they grow up,
51:52and an and develop.
51:55Yeah floor this is such
51:57a great great question.
51:58And as as you've shown with your own work,
52:02talking about somatic mutations,
52:03and we know that even genetic variants
52:06may not be shared across different
52:08tissues and so there have been
52:11attempts to address this problem.
52:13And so for example,
52:14there's a tool called Pecan
52:16developed by Michael Horror,
52:18Gustavo Tracking Michael Meaney,
52:19which actually does a paired comparison of
52:22DNA methylation in multiple brain regions,
52:24and unfortunately is just
52:26in peripheral blood.
52:27At the moment and looks at the
52:29correspondence between DNA methylation and
52:31in blood with different brain regions,
52:33and they identify CPG's that show a
52:36higher degree of concordance than others.
52:38I think your point is well taken,
52:41this is the idea that we can take
52:43a peripheral sample like blood and
52:46say that this is going to predict
52:48DNA methylation state in a neuron in
52:51the dentate gyrus of the hippocampus
52:53I think would is a stretch.
52:55I think that it's going to.
52:58Be very challenging to identify and
53:00sites where there is a high degree
53:03of correspondence in specific brain
53:05nuclei between brain and blood.
53:07Where I think we can begin to get
53:09a better understanding of pathways
53:11that are likely to be shared across
53:15brain and periphery is if we focus on
53:18specific regions in the genome where
53:20there may be snips that influence
53:22DNA methylation in the periphery that
53:25also are shared snips that influence DNS.
53:28Relation in central and we can use
53:30the peripheral tissue essentially as
53:32a model Organism to say look this
53:34proof of principle that this exposure
53:36influences DNA methylation or inclusion.
53:38Influences the relationship between the
53:40snip and DNA methylation in the periphery.
53:43And perhaps this could be
53:45occurring in the brain.
53:46But then we would need to
53:48document that experimentally,
53:49either in cell culture in ipsc's.
54:04Any other questions anyone?
54:12Well, it's. It is 2:00 o'clock,
54:14Kieran saved by the Bell.
54:16But thank you so much that was really
54:18a marvelous presentation and we learn
54:20so much and wonderful to have you
54:22here and we look forward to all that
54:24you'll teach us another ideal do.
54:26So thank you here and thank you
54:28very much. Thank you everyone.