The representation of stress resilience in neuronal networks
This is a modal window.
Das Video konnte nicht geladen werden, da entweder ein Server- oder Netzwerkfehler auftrat oder das Format nicht unterstützt wird.
Formale Metadaten
Titel |
| |
Serientitel | ||
Anzahl der Teile | 23 | |
Autor | ||
Lizenz | CC-Namensnennung 3.0 Deutschland: Sie dürfen das Werk bzw. den Inhalt zu jedem legalen Zweck nutzen, verändern und in unveränderter oder veränderter Form vervielfältigen, verbreiten und öffentlich zugänglich machen, sofern Sie den Namen des Autors/Rechteinhabers in der von ihm festgelegten Weise nennen. | |
Identifikatoren | 10.5446/62854 (DOI) | |
Herausgeber | ||
Erscheinungsjahr | ||
Sprache |
Inhaltliche Metadaten
Fachgebiet | ||
Genre | ||
Abstract |
|
Leibniz MMS Days 202318 / 23
00:42
Computeranimation
03:53
Computeranimation
15:03
Computeranimation
15:42
ComputeranimationBesprechung/Interview
17:09
ComputeranimationVorlesung/Konferenz
Transkript: Englisch(automatisch erzeugt)
00:00
Basically, I will start our institute for resilient research is obviously we are dealing with stress resilience, but I hope that at the end of my talk I will be able to convince you that at the end of the day also there's a lot of common grounds, but I will start
00:21
now with an introduction to resilience and then move into real neural networks and at the end a little bit of a glimpse and here of course talking to the expert on what we are currently doing in terms of AI driven data analysis.
00:42
So resilience, so the concept of resilience actually dates back to the 1950s where Ms Werner conducted a study on the Hawaiian island of Kauai.
01:01
At this time in the 50s it was maybe not that idyllic as it is now because there were a lot of underprivileged native populations and she conducted a study of about 700 children and she compared the local population of children to a population in the continental US and
01:32
she looked at these children from a kind of risk environment. These children endured poverty, physical and mental abuse and she did a follow up study
01:47
and with a strong expectation of course these children will do poor in life, they will have less economic status, they will maybe also exhibit abuse, but to her surprise yes of
02:02
course she found this subpopulation of children, two thirds of all who did experience difficulties in school, they had criminal records and were also depending on welfare and adulthood. But about a third of these underprivileged children did not end up in poverty, they did
02:27
end up having college degrees and actually getting into adult life without a criminal record. So there seems to be a subpopulation of children even though they grew up in the same kind
02:49
of negative underprivileged circumstances who are resilient, who kind of overcome these detrimental conditions.
03:03
In these early studies she put forward so-called resilience factors, so she looked at those children who are resilient in terms of their negative upbringing and it so seems that these children actually exhibit a high social competence, they can actively cope with a negative situation
03:28
in life and they have a rather close supportive social environment. So from these early studies what did we learn is that we can actually derive resilience
03:41
factors on three different levels. First of all the so-called resilience factors, a lot has something to do with optimism, self-efficacy, also experience positive emotions, hardiness, self-esteem and then on the more abstract level also having an idea of seeing meaning in life and also a sense of coherence so that
04:07
your own actions are actually embedded and meaningful in the greater things. So from these studies there was a theory that okay these are so-called traits so you
04:27
are either born resilient or not, you either kind of see this meaning in life, the sense of coherence or you're not. But what we are actually now more and more understand is that this is not the case of
04:42
course these are factors but they are so-called resilience mechanism which then turn to a resilient outcome and here at the, not here, in faraway mines, not as far as Kaiserslautern but also quite a ride, we are actually studying these resilient factors leading to resilient
05:06
mechanism with a resilient outcome and we study here in mice and men. We have here clinical studies which I'm not talking about today and we actually also as
05:22
of course as the Leibniz Institute we are also interested in offering resilience teaching meaning workshops in which you can also learn to improve your outcome. Today though I wanted to slowly move now to the world of neuroscience so let me just
05:44
bring up a more modern definition of resilience that we say resilience as the maintenance or recovery of mental health during or after adverse life circumstances. So if we look at this here, now my first graph, that of course there has to be a stressor
06:04
or traumatic event. In the case of the children in Hawaii it can be physical abuse or kind of growing up in poverty and then you look at trajectories and of course it's normal
06:20
that there's a deflection in mental health. It's completely normal that you react to a traumatic event. It would be abnormal if you had, God forbid, a car accident and you walk from there and said okay I don't care what happened. Of course you will react. Of course. But now the difference is how is your trajectory in terms of your mental health afterwards.
06:43
After this initial kind of deflection the two-thirds of adults here they actually go back and the word of resilience comes from the Latin phrase of resiliere which comes
07:00
from mechanics so kind of a spring goes back to this initial point and another 20 percent here they have a so-called delayed recovery phase so they stay in a kind of psychological dysfunction and then revert back. But 10 to 20 percent actually show either the following
07:22
trajectory that they do not return or basically that they do not react in the first place again which is also pathological if you come up from a car accident and you don't feel anything. Most of these people actually then go into mental dysfunction later. So in a
07:43
contrast to many studies focusing on many groups focusing on for example on those groups here who develop long lasting developmental dysregulation we here at the Leibniz Institute
08:02
we focus on those who are resilient and want to know what differentiates these people who react in a resilient way from those who are not. So in a way it's health research and how is actually the majority of people here protected from getting sick and indeed if
08:26
you then look at this trajectory in a lifetime you will experience in lifetime everyone will experience stresses here and the point is that we do not react the same time. So resilience
08:44
cannot be a trait because there are many examples upon which someone reacts resilient to the first stressor and then there comes a second hit and the person develops a kind of a chronic mental disorder. So basically we define resiliency as an outcome and then we want to look also
09:09
at the neural mechanisms which guide a resilient outcome. And with that let's dive into neurons and into preclinical research in animals. So what do we do here? How do we model resilience
09:30
in animals? So we do it the following way. We have a white aggressor mouse here, X breeder is a very strong mouse and we kind of introduce a smaller mouse to the aggressor and these
09:49
are two male mice and males also male mice what they do if they bring them together they fight and here of course this leads to so called social defeat. So this mouse here is
10:03
being beaten up and we kind of continue that for a phase of 10 days they will not beat each other up for 10 days but just for a minute or so and then we separate them with a mesh so that they can have still sensory interaction. So this is the traumatic event. So what do
10:26
we ask then? What we do then is that we change the context. Now here the aggressor is enclosed here in it's still in one arena in one cage but now the aggressor here is enclosed so
10:42
now the mouse here is safe. And what we do now is that we measure the explorative drive of this mouse and this is a very important part of non-resilient behavior. So given you had a bad experience with a man in a black suit in a given context a generalization would
11:08
now mean that I'm afraid of every man in a black suit and so if this, translating it here if this mouse here is only crouching here in the corner it means that the mouse
11:23
generalized it has a generalization of fear and cannot discriminate that now I'm in a different context I'm not in the park at night I'm somewhere else there's a man sitting in a black suit this is a different contract I do not have to be afraid I can still talk to you without being afraid. And to our surprise actually there's a huge spread even though
11:49
these are inbred mice there are mice which we then deem resilient and this is so called social interaction score so how long does this mouse here interact with this aggressor
12:03
and the higher it is the more resilient it is. You can of course argue that the same as resilience in humans maybe not maybe it is to a given extent but we believe that there are neural mechanisms at play which are similar across species. So now we have a mouse model
12:28
of resilience so what do we do next? So what's the advantage of using mice? Yeah the advantage is that we can look into their brain and how do we do that? Let's use light. And so as
12:52
all of you know brain is a highly scattering tissue so how can I look even in a small Homer Simpson or mouse's brain you can argue what is larger and what we do is that we use this
13:06
technique of two photo microscopy so if you look at the scattering properties and finally I come up with my first equation you will see that the scattering is highly dependent on the wavelength at first so if you move to higher wavelength you can penetrate deeper
13:25
and second of all if you kind of excite I hope that you can see it maybe we can switch off the light somewhere that'd be great. I think then these images will be better yeah it will be much better if you could switch off the light so because what you cannot see
13:45
here is that you have here an objective and that here is a fluorescent tissue if you use normal one photon excitation actually you have here the beam of light and if you then want to I think that's better thank you if you then want to collect this light of course
14:04
then you have a problem that you have also out of focus light which you then collect. If we do a two photon excitation due to the very high very low likelihood to get the two photon effect actually you will only get excited here a flow for at the focal spot
14:25
what does it mean first of all you can go to get deeper because you can use longer wavelength double of the wavelength and second of all you do not care whether there's scattering from the mitten light because you know that all of my light is anyway coming from a focal point so you can still do kind of a scanning approach a microscopy even in highly scattering
14:45
tissue because you can just collect all the light emitting from this point even if it's being scattered by its way back here to the objective yeah this is the main advantage of two photo microscopy and therefore it has revolutionized preclinical neuroscience research
15:03
so what we are doing we perform now two photon so called calcium imaging so we have a two photo microscope and what we also have here is that we have fluorescent indicators for calcium calcium is a very important ion in which elevates in the cell if a neuron is
15:27
active so if we use green fluorescent protein based indicators we can now actually observe activity of individual neurons by optical means in the living animal yeah and yeah this
15:43
is how it for example looks like if you look at the activity just at rest of a large part of the brain and you will see that the brain is constantly active and of course there is stimulus evoked activity but now unfortunately now talking to you only three or four percent
16:03
of your neurons will be tasked with what I'm actually doing here ninety five percent of your neurons are doing something completely else and it's like like throwing a stone into already a wavy sea and so we look at these activity states as both spontaneous and and
16:22
stimulus evoked and basically now we have here the mouse the mouse is here sitting here on a so called jet ball so this had fixed but the mouse is kind of can still move and can still navigate here for example through a virtual maze and here on the right side
16:41
what you see here these are individual neurons and I hope you can just see a little bit if I if I repeat this here that you can see that here this neuron for example is increasing its first emission meaning this neuron is here active so what I can do now is I get
17:00
a kind of an optical readout of a local neural net while the animal is moving and now we come to the part which is maybe then we come slowly to common ground what I'm doing actually now is that I assign a number to each of these of these neurons here neuron one to twenty
17:22
five and then looking at the intensity trace and then you see here these red peaks are those time points here where this neuron kind of linked up and then I can show that these peaks here actually are associated to a neuron firing so at the end of the day I have an
17:42
activity matrix which and this is a kind of analysis routine I will quickly guide you through so we have our raw images we do a segmentation approach where we segment out semi-automatically all of these neurons we have the traces in time we can remove artifacts
18:03
and then we binarize at the end of the day we have a binarized matrix with the advantage that we can still assign to each of these point one point in space and can also then look at connectivity within this neural network here you have it right and this is how you
18:21
get from these images to to these points and you can do that now in these animals which react differently to the cage aggressor based on the social interaction score and what you see there yes there are differences so there are here these so-called transients here and
18:43
if you look at this here the non-resilient one is firing more than the other one and you can quantify this of course here this is a cumulative probability distribution here and you can see there is a significant difference between here those animals which are differing in a very high
19:02
cognitive domain if you see and that's actually also one reason why I'm here because really you can can use these kind of psychological theories at the end of the day there's a difference in the temporal spatial coherence of these neural nets and to our surprise what is really surprise if we
19:24
look at the non-stressed or non-traumatized animals they look very much similar to the susceptible animals meaning that in a way a resilience means that you are being changed by
19:40
the stress area you are different you you are not unchanged right you react to your god forbid car accident right but you act in a plastic plastic way you so you come up in a way stronger and and this is what you see here for example also in these of course compared to your methods
20:00
very very rudimentary circuit diagrams you have here very local and very strong connection between connections between individual points and and you see here that particularly also the susceptible network is very ill-connected so what does it mean to to what the what how does it relate all the
20:23
way back to to resilience in humans so if we stimulate these mice now with these with these we can stimulate these mice as you saw earlier with this kind of drifting gratings and you can look at the so-called visual cortex and what do the neurons do they encode for different orientation
20:42
what does it mean so if I show you like an angle moving from left to right there will be a few neurons in your visual cortex only firing if I'm showing you this angle not firing if I'm showing you this thing at the end at the higher cortical circuits you kind of put everything together so you can look at this here in this cook so-called polar plot and the sharper here the plot is the
21:06
less promiscuous a neuron is and if you compare this now this is the resilient one to the susceptible and an untreated control and you look at the so-called circular variance you will see that the
21:23
there's a significant difference meaning that the resilient animals are very good in sensory discrimination and there you have kind of the circuit equivalent what I told you earlier that I do not have to be afraid of my dear colleague at Vias just because he has a dark dark suit
21:46
because because because I can better sensory discriminate right I can better know now I'm in it now I'm in a different environment and this is actually now to be to be found here and this can
22:02
be is also stable in time and so we are neurophysiologists and I'm biophysicist as training but nonetheless we are not AI expert and we we are aware of that and what we are doing and what
22:24
we are doing is that I'm we found it I must speak of this so-called initiative for systems analysis neuroscience where we kind of want to want to introduce these AI driven methods in all levels
22:41
of observation and neuroscience from genes to cells to network and behavior and we have here the different methodological columns and what we do is that we offer teaching Python courses for example and also offer software solutions where then particularly also the students and postdocs
23:04
can take part and and this is actually has been a great success because as I just mentioned earlier in our discussion I think that that it will be more and more important also for neuroscientists that this is as important as being able to hold a patch pipette to get an education and these methods
23:27
and therefore also these meetings here are also so important to us and what we did in here in this context for example and which is just a very poor beginning and but I showed you how we do the analysis
23:41
of these local micro circuits at the end we have here these calcium transients which are kind of which are related to room activity but at the end if you have the cells on this axis and frames on this you have this kind of binary activity matrix and and again we use k-means clustering and with the elbow method
24:07
we derive now so and to look for how many activity states are now in these data and if we do that you can see that there are besides the first more random cluster there is a structure in the data
24:27
in terms of functional motifs which are always visited and what you can then do if you can ask how many motifs functional states do I have in these resilient and non-resilient animals and how do they transition
24:41
and ultimately also how stable are these states and whether they're attractors or not I mean this is I think open to debate but what is important is that we also see here differences that these resilient animals have deeper and less attractors in the neuronal activity space and this has for us important implications
25:07
because maybe what we then at the end of the day can do is that we can shift then first animals and then humans from a kind of maladaptive attractors state to the other and of course then we have to know where
25:24
in which state is the current network where is an intervention best so that I have a kind of a psychologic intervention not when you are deep in your maladaptive attractors then I can save the time but when you are susceptible to a change and then maybe transgress to a more adaptive state
25:44
so I think there's a great use of also here system theory which I of course do not have to explain here but I would like to give this audience also in this and understanding these trajectories and with that I would like to summarize so we can conceptualize stress resilience as a dynamic outcome after stress exposure
26:06
governed by a resilient mechanism we can model resilience by the chronic social defeat and social interaction paradigm we can use two photon excitation and combination with calcium imaging to optically record cortical networks
26:25
in the awake behaving mouse and we actually do see that resilient networks exhibit an adaptive and plastic behavior which are better than those animals which are not stressed and just one sentence it has also been shown that
26:41
if you overcome a stress I mean I don't like this word what doesn't kill you makes you stronger because there are a lot of people suffering miserably after stress but there are a few who are coming out stronger and maybe we can use this knowledge to increase the population of those who are coming out better
27:02
and with that I would like to thank you very much for your attention Thank you for listening