Epigenetics, Chromatin Plasticity & the Neural Basis of Memory | Giulia Santoni | #169

About the guest: Giulia Santoni, PhD is a neuroscientist who obtained her PhD at the EPFL in Switzerland, where she studied epigenetic influences on memory formation.

Episode summary: Nick and Dr. Santoni discuss: transcription & gene regulation; synaptic plasticity; learning & associative memory; epigenetics, histones, DNA methylation, and mechanisms of gene regulation; chromatin plasticity & the neural basis of memory formation; and more.Related episodes:

* Emotion, Cognition, Consciousness, Behavior & Brain Evolution | Joseph LeDoux | #73

* Cognitive Neuroscience, Cognitive Flexibility & Control, Attention, Working Memory, Multitasking & Behavior | Tobias Egner | #130

*This content is never meant to serve as medical advice.

* Support M&M if you find value in this content.

* Full audio only version: [Apple Podcasts] [Spotify] [Elsewhere]

* Full video version: [YouTube] [Odysee]

* Episode transcript below.

Full AI-generated transcript below. Beware of typos & mistranslations!

Giulia Santoni 3:42

I come from Italy, so this is where I did my bachelor's and my master's. I am a biologist, and during my master's, I switched to neurobiology because I was very intrigued by how the brain works and and really the question that drove my studies and my formation was this very existential question of, like, what makes us who we are. And so I thought that by understanding our body and our mind, I could get a little bit closer to that question. And yeah, so after my end of my master in neurobiology, where I still had, like, a very strong background in basic molecular biology and neuroscience. Then I decided to join the lab of Johannes graph for my PhD, because I had these two main curiosity, one being epigenetics and the other one being memory. And so in his lab, I was like, okay, I can, can I do like a project where I can combine both fields? And he was super cool because he said, Yeah, of course, I'm also very interested in these two aspects of science. And so we had a lot of fun over the last years. So like trying to figure out how to study epigenetics memory at one. Is

Nick Jikomes 5:00

so on the question of what makes us who we are, one thing we can talk about is what makes a cell what a cell is. So obviously, we start from a single cell, and then, you know, the embryo develops, and we go from one cell to many, many, many cells. And there's many different cell types. There's neurons, there's skin cells, there's different types of neurons, there's different types of skin cells, and so on and so forth. At a very basic level, what actually differentiates one cell from another? They all have the same genome. So what determines if something's, you know, a neuron type A around type B and so forth.

Giulia Santoni 5:42

Yes, yeah, this, when I first understood this, really blew my mind, because essentially, almost every cell in our in our body has this two meters long stretch of DNA inside. And so, as you well said, the DNA is equal in all cells, and yet we have such a massive diversity and and this is because this two meters long DNA has to be wrapped and condensed in in an environment that is microscopic. And this implies that during the wrapping, the conformation will make it so there will be parts of the genome that are totally inaccessible, so where the chromatin will be very condensed, and whereas other parts of the regions will be more accessible, the DNA over these parts are is more relaxed, and therefore the genes that are left In this more open, accessible regions will be the genes that characterize the identity and the functions of the cell. And this is what epigenetics stand for. Stands for For in in a very microscopical sense, because it's really like allowing for these three dimensional confirmation of the DNA into such a small space, like the nucleus

Nick Jikomes 7:02

I see. So there's enough DNA in our cells that the if it was just stretched out linearly in a straight line, you'd be talking about meters of DNA. And yet somehow all of that has to fit inside of every single one of our cells, and so that DNA is bound up or wound up very tightly. And I guess, I guess a picture that could offer an analogy here would be like a fishing pole. You know, you've got a lot of line, but it has to, like, wrap very tightly in one little space. And so, based on which length of the the DNA segments are available or unwrapped, that determines which genes could be used by the cell to make proteins. Exactly,

Giulia Santoni 7:47

yes. And of course, there are. So what the analogy you used is very, very good. There is also this analogy, sometimes of the necklace of some ladies with pearls. So essentially, these proteins you so well introduced are called histones, and they really serve to wrap DNA around it. And what is interesting is that these histones are not fixed. I mean, they can be displaced. They can be changed into different variants depending on many different factors and and so there are some regions that we know are called constitutively open regions. So those usually bear genes that are really making up the most of a cell, the most of the identity of a cell, whereas there are other regions that are more variable, they can be open, they can be closed, and this really is more what determines the function of that cell and the epigenetic factors that are like, essentially epigenetic factors like chemical groups. So we have so many different chemical groups that can be harbored on the pace of these systems and depending on the combination, and some people refer to our epigenetic code because they can be so many different and, and, and so, and also organizing so many different loci of the histone space different amino acids. So depending on the code that they that they will they will compose, then some codes will promote more openness, and therefore more accessibility for the genes that are found around those histones be read and then expressed, whereas some codes instead promote more closure. And so even though these regions can be accessible, it doesn't mean that they will be accessible for that specific cell at that specific moment. So

Nick Jikomes 9:41

it sounds like the openness of the DNA, the interaction between the DNA and the histone proteins around which the DNA is wrapped, there's, there's, there's got to be a balance here, because, on the one hand, you don't want all of the DNA available and turned on, because then you would be able to maintain the identity of a cell. So a neuron. Be able to stay a neuron, if, if everything just turned on. But on the other hand, you you can't have everything completely static. You have to have some room to turn some genes on and off, because that's what's ultimately going to enable that cell to change somewhat. So in the context of the nervous system, you know, it could build stronger synapses. It could get rid of synapses. Neuroplasticity would require that some of these genes can turn on and off to some extent, while at the same time maintaining the overall identity of the cell. Yeah,

Giulia Santoni 10:28

exactly. And this is the beauty of epigenetics as well, because there are like these chemical groups that are added on these histone or, for example, methylation, can also be added on the cytosine groups of the of the DNA. So this chemistry can be modified by two classes, while several classes of enzymes. We have erasers and we have writers that are really found that this dichotomous, like they have this very opposite role. So while writers, they really add different chemical groups on this amino acid found on the histones the erasers. On the other hand, they will remove these chemical groups. And so this really like talks, in favor of this balance, in a way, because the system somehow by balancing the activity and the content in the cells of these classes of enzymes. It will therefore balance the presence of the chemical groups on the histones, and therefore the accessibility, or not of the chromatin to transcriptional machineries.

Nick Jikomes 11:39

And so, so you've got, so what are, what are some of these reading and writing capabilities? So you mentioned DNA methylation, so you can add a little chemical group to certain letters in the alphabet of the DNA. What are some of the other mechanisms there that do this reading

Giulia Santoni 11:56

and writing? So in terms of methylation, is, is this one special mark? Because, as you said, can be present in the DNA and as well on the histone tail. So in terms of histone post translational modification. So those are called, like these chemical groups added on the histone tails, we have methylations. I, in my personal like research, I will focus particularly on histone acetylation, because it's really like a splendid tool to promote openness of of the DNA, of the chromatin structure, because histone acetyl groups, they managed to, by repulsion, really Stretch and open and relax the chromatin and make it accessible. But there is also phosphorylation right now. I mean, we have, this is really one, a wonderful time to be studying epigenetics, because there are so many new PTMS, so post translational modifications being discovered every day. There is serotonin, serotonin, serotonin, lactose, elation, and all these new chemical groups seems to us contribute to the overall dimensional organization of the DNA in the nucleus. I

Nick Jikomes 13:15

see So, so you said post translational modification. So, so this would mean that proteins get made within the cell. So histones would be one example of a protein where it's a sequence of amino acids that's got a shape to it, and after the protein is actually made, after it's manufactured, translated, you can modify it with various chemical residues. So the one that you mentioned was acetylation. So the idea is just there's proteins in the cell. They have an amino acid sequence, they have a three dimensional structure. Then you can modify them after that, after they're already made and functioning, and that can determine things like the DNA winding, yeah,

Giulia Santoni 13:53

perfect. Yeah. It's like a Christmas tree. You know, you have your yeast, and it's like the bear Christmas tree. And then you can add all sorts of stuff on top.

Nick Jikomes 14:03

And so what's the relationship between acetylation and transcription, or turning genes on which one?

Giulia Santoni 14:10

Yeah, acetylation is one of these modifications that we decided to study in the context of memory, because it has been associated in the nucleus to neuronal activity, for example. So this, this histone modification is very interesting because it is linked to to the energy levels of a cell. These acetyl groups, they derive from, for example, acetyl covid, which is one of these byproduct of the glycolysis or of many other metabolic pathways, actually. And so whenever you have this pool of Acetyl CoA, they kind of somehow correlate with the amount of acetylation you would have in the nucleus. And this is interesting because we see, I mean, a neuron has to use a lot of energy to fire and to function. And. And and therefore we see that like, for example, after, let's call it synthetic neuronal activation, or chemically induced neuronal activation. For example, sometimes we do experiments in pattern dishes where we cultivate and grow neurons. And there, if you use specific chemical compounds, we can induce neuronal activity. And this has been shown both in vitro system or also in vivo, using mice and stimulating them both either with chemical compounds or with like behavior, like, for example, Latino mice exploring new arena, or, for example, learn something new. So both the chemical induced neuronal activation, as much as the natural induced neuronal activation is correlating with higher histone acetylation levels. And these high histone acetylation levels have been associated to higher level of gene transcriptions. And there have been plenty of studies really showing how, by promoting his acetylation, you really relax the the portions of the genes called promoters, as well as enhancers that are more like distal regulatory regions of the genome that really manage to loop, go to the promoter and carry and complete these transcriptional machineries that will be essential for allowing robust gene expression. And yeah, I

Nick Jikomes 16:32

see so so cells have, actually, they've coupled their bio energetic supply of fuel in the form of its ability to make ATP, to its to its potential for doing gene transcription through acetylation, because so So you mentioned that these histones become acetylated, so you added an acetyl group to them. This enables more genes to be turned on. That's very energetic process, inherently, right? You have to use proteins to transcribe the genes. You've got to fold the proteins, then you've got to shuttle them to maybe the far end of of a neuron, for example. But that requires a lot of fuel, so it sounds like there's, there's sort of something elegant going on here, where the the actual signal that opens up or that modifies the histones to open up the potential for transcription is actually coming from a Acetyl Co A itself, which is basically the a precursor to make ATP in the cell. Yeah, yeah.

Giulia Santoni 17:28

As I said, I think right now, we are still at the correlative level. I think in in in other studies link more like, like fundamental biology studies on yeast, for example, they really saw, they really investigated deeper the role and contributions of metabolism to gene expression and epigenetics profiles in in neuroscience, we are still not at this like causality level, but I think that there are plenty of studies showing how Like, metabolism really supports urinal activity. And really, like defines the type of epigenetic landscapes that you have. So for sure, like knowing that the building blocks, so this bricks of a simulation derived from metabolism, even though we are still lacking the precise experiments to test how much and from which pathway then would change which of the genome we have a pretty solid yeah background to really rely on and say, Yeah, for sure, they are linked.

Nick Jikomes 18:32

And I want to give people a sense for the the basic steps involved in, say, doing something like building a synapse, or the connection between the nucleus of a neuron and the synapse of a neuron. So neurons, there's sort of a weird cells, in many ways. They've got a weird shape a synapse, or connection between a given neuron and one of its partners might be very, very far away from the nucleus of that, that neurons, where the DNA is. And so, you know, if, when we're talking about transcription and turning genes on and off, you know, the genes are in the nucleus, they have to be turned on. They have to be made into proteins. But then, you know, if you want to build the synapse, for example, you've got to shuttle that protein all the way, you know, quite, quite a distance to the synapse, and that probably takes a lot of machinery and a lot of energy. So can you give us a sense for how how expensive and how costly it is to do stuff in the nucleus and then and then ship it all the way to the synapse?

Giulia Santoni 19:35

So I just want to make a distinction, because we know that not all the synapses need to so for sure, for sure, let's say, if we would start to build the neuron from scratch, then the nucleus would be fundamental. But let's say, once the neuron is already set up, there will be it. It is already somehow connected, and there are already some points, like with some junctions between. Neuron. So this sign up there are synapses that are constantly like maintained at the synaptic, synaptic level, because in the synapse, there are still like rather ribosomes, sometimes even mRNA, so which is usually this, like spectra messenger RNA, that is the direct copy of the gene you want to express from the nucleus, from the DNA, it can so mRNA copies can translocate and and be like allocated in the synapses at the bottom of them. And so this means that for very fast and rapid signals and communications, the synapses can buy their stuff without too much of a contribution of the nucleus respond to this activity, local activity, and get strengthened, become bigger. So so this is very rapid and very efficient for like, for example, also for these short, short term forms of memory. So when, whenever people have to remember number for a few seconds, this is what would happen. You have like this brief strengthening of that specific circuit that will help you retain that information, but eventually forget about it in four three minutes time, whereas anything that involves the activity of the nucleus has to derive by a very strong type of synaptic inputs. So this is why, whenever we call or we talk about long term memories, we really have to have a strong signal that from the synapse already, there starts all the journey back to the new nucleus, and this is done through a series of past molecular cascades. So usually, many different receptors at the synaptic level are involved. For example, we have this NMDA receptors that would be among the leaders in in this calcium cascades and like secondary molecular cascades that will, that will fall on protein that are really good, like this come kinase protein class is really good in shuttling from the from the periphery of a cell. So either from dendrites on, from the from, yeah, where the spine activated. Spines are. They can travel up to the new place, and they follow these crazy highways like this. This acting, acting inside the skeleton get will guide them until the nucleus were there. They will induce changes, being like activating other transcription factors, gathering them and then binding the DNA. And only there you have all these changes being in acetylation, in methylation, phosphorylation and so on. So I only focus in acetylation, but what I can see is that some promoters will be more acetylated, and therefore the transcriptional machine can be locked into this acetylated site and start its journey along the gene and the gene body. And when we will have this this new model, new copy of mRNA for all the genes that are needed to support the new formation of of spines, or the new formation, or maybe just an enlargement of the of the we call it this, yeah, dendritic spine. Then all this messenger RNA will be transferred to the cytosol, where it will become so translated into proteins. And these proteins, from there will start their their journey, as you said, towards the periphery again and and what I find always interesting is that also in this journey, it's not really yet super clear what tells a protein to stop there. So there must be signal as well involved in this can really guide this traffic, trafficking of proteins and enzymes all around the right spot, to the spot that needs to be and to, yeah, to be strengthened. So I in terms of how many molecules of ATP this takes, I cannot tell. I haven't measured them all, but I am pretty sure quite, quite many.

Nick Jikomes 24:24

It's probably quite expensive, yeah, yeah. So, um, yeah. I mean, if people think about it, right, in our own body, there are, there are neurons that are, you know, meters long, you know, there's neurons that go all the way from your head down through your spinal cord. So there can be quite a distance that has to be traversed by some of this machinery. But sounds like what you're saying is, you know, roughly speaking, there's some short term things that can happen and some long term things that can happen. So at many synapses, you've got mRNAs that are already at the synapse, you've got ribosomes that are already at the synapse, all of. Stuff needed to turn the

Giulia Santoni 25:02

mitochondria. So this like, yeah, ATP sources mitochondria are also at the synapses. Yep. So there's

Nick Jikomes 25:09

mitochondria. There's basically everything the cell needs to turn that mRNA into a protein and stick it somewhere can already be there at a synapse, and so that can help mediate some short term things. So, so if you imagine, I think you use the example of just remembering a phone number, briefly, everything you might need to do something like that in the short term is already there, oftentimes. But then also there's you know, as we all know from experience, we can encode long term memories that that we store for weeks, months, years, and the these long term, longer term changes often do require communication between the nucleus and the synapse. And so there are mechanisms in place, even if we don't understand what they all are, that that allow some of the short term stuff to happen and some some of this long term stuff to happen that involves communication with the nucleus when we think about strengthening a synapse for the long term, say, in the service of encoding a new long term memory. From from a neurobiological standpoint, can you talk a little bit about LTP and give people a sense for the essence of what LTP is? This

Giulia Santoni 26:18

is not an easy question, because there is a thing, sila, food debate, and I want to precise. I'm already sorry, because I formation, so I might say plenty of things that are not 100% correct, so from my understanding. So what I imagine when I think about LTP, which is this long term potentiation of a sign up is is a readout of how strong assignments is because and how much response it will evoke once it's stimulated. And so what we know is that, for example, this this type of of of readout of measurement has been associated to memory for a very long time. There was almost like this, this, this kind of sense that nltp. So if one finds an LTP type of phenomenon at a sign up, it is an indication of a memory event. Because we know that memories, long term memories, in theory or any practice, they do make our system stronger, like the memory circuit stronger. And so what we saw is that after a very strong activation of pathway, we see that the synapses involved in that pathway will be strengthened. And this means that if you go in at a certain time and you stimulate that sign ups again, the response the same stimulus will be much stronger. So it's, it's, it's, let's say, a synonymous of an event that previously happened, that strengthened a spine that before, when you record that needed certain, certain current to be activated in a certain way, whereas even a sub threshold active type of activation will lead to a similar response this time, because that sign ups has made has become more strong. So this was, this is what I imagine. I imagine all the cascade of events that in my from my field. So as a as a molecular biologist, I do think that because a previous event of signaling has induced a lot of changes that which mean that they led to a restructuring of the synapse, and this time, the Synapse will bear more channels, like AMPA receptors and mga receptors. There will be more like even the size of the spine will be larger to harbor more of this molecular and proteins, molecular factors and proteins that make it more stable. Then, if I sub threshold, if, if I use a sub sub threshold simulation, these sign ups will be able to respond equally strongly as the first time that I really needed a lot of a lot of current I hope this makes some sense for you, too. Yeah. So

Nick Jikomes 29:23

basically, if we simplify it a little bit, you can imagine a synapse receives a message. That message can be relatively weak, it can be relatively strong, and when the message it receives from are the neurons is particularly strong, this can activate that synapse in ways that it is not activated when a weaker signal comes in. And this involves thing, you know, the variety of things, including calcium influx into that synapse. And like, what are some of the hallmark things that happen when you get a very strong stimulus, something that's. Is capable of inducing LTP. That's capable of making that synapse bigger and stronger. I know that one of the things that happens is some special genes get ultimately turned on. And so what are some of those things that happen in response to those very strong stimuli all the way back?

Giulia Santoni 30:14

Yeah, so, so, as I said, like this very strong simulate stimulations will induce this secondary messages like they will induce this cascade that will activate the genes. And so this means that it's not one gene, one LTP, so, but however, there have been like, for example, hubs of genes that were mastered by some transcription factors. And among like, the most famous in the in the learning and learning and memory fear field, there is this crap, which is like, it's, it's like as, yes, a cre binding element. So this transcription factor can recognize some element of DNA and bind to those elements. And these elements are found in the proximity of the genes of some of the genes that are really important to support this, like synaptic activity and transmission. So so here we see how much like the importance sometimes of one factor, one transcription factor can really, can really have on multiple types of chips that are all necessary to support a stronger activation. But LTP in the for example, has been challenged by using protein inhibitors, for example, or transcription inhibitors, and in both cases, have been found to be like defective so if you inhibit the transcription genes or the translation proteins, then you cannot have the LTP, which really speaks in favor of this fact that you need this communication from the signups good nucleus and that in order to have this potentiation of that synapse. So

Nick Jikomes 32:00

it sounds like, when you have a very strong input stimulus that can stimulate LTP, there is a signal that's created that leads to some coordinated changes in gene activity. So an individual, you know, the example you gave was creb, c, r, e, v, there's an individual protein that can turn on multiple genes at the same time. And I would imagine that the reason you have that kind of coordination is there's multiple genes encoding multiple proteins that are going to be necessary for synaptic strengthening. So what are some of those genes? Are they things like ion channels that will make the synapse more sensitive and things like that. Yeah.

Giulia Santoni 32:41

So I can talk about like, for example, what we find, what we found in our study. So what I was looking for were genes that would be like, would compose, somehow, a signature of what would make a neuron more like, easily predisposed to become part of the memory network. So what we thought was okay, if acetylation is important for gene expression, and if memories are like long term memories are really they necessitate protein expression and gene expression, then what if there is like, disconnection between acetylation and memory formation. What if the neurons that have to become part of the memory network have a higher content of acetylation, really, as a baseline, compared to other neurons? And what we saw is that whenever we manipulate so, we artificially increase the acetylation level, levels of random neurons, that, of course, we pick regions that we know will be important for the memory networks we want to investigate. Then whenever we increase the levels of acetylation, we see that no matter how so by using different types of enzymes that can do this, that can mediate this increase in acetylation, we see that the regions sense the DNA region sensitive to these increases in acetylation are highly similar. So that speaks in favor of the fact that these variable regions of the DNA that can be found either in closed or more open position, they are very conserved across different neurons, because no matter how globally you induce increases in acetylation, the specific regions that will be sensitive to it and therefore become more open will be always the same. And as you said, these regions, what we found was very interesting, because, indeed, they are related to potassium channels. Because we know potassium channels are really important in the rectification of the action potentials so and which means that, essentially, they contribute to. The frequency of the spice of a neuron. So potassium channels are able to give a pace on the action potential. So these firing events that happen in neurons once they are stimulated, and in the context of memory, we can think this phenomenon being very, very relevant. Because, of course, if a neuron is more sensitive and sense the environment the stimuli, the incoming stimuli, better than the possibility of spiking, and the intensity of spiking will be higher. And for memory, we can imagine, okay, it's really important if we want to encode a random event that our new neurons are ready for, it are ready to activate all these processes that eventually will lead to the storage of of this specific event we're talking about?

Nick Jikomes 35:57

Yeah, I was just gonna say, okay, so part of the puzzle here is there are many neurons in different regions of the brain that could potentially become part of a memory, but not all of them do become part of any given memory. So the question is, well, why is it some neurons, but not other neurons? Before zooming back into the molecular and cellular level, I want to zoom out and just give people a sense for the system you're working in. So when you, when you're when you're talking about memories, and you're talking about these experiments, what's the experimental system? Like, what animal are you using, and how are they being made to remember something? Yeah,

Giulia Santoni 36:31

so we use mice. So we use this black six mice, which are, yeah, which are lab mice that we know are very able to form memories because animals, I mean, in general, mice, they have to learn a lot in already in their wild environments, right? So they have to know how to search for food. They have to remember where there was a danger, a predator, so they recognize smell. So their memory system is quite developed, and so we used, in particular, we kind of, yeah, made usage of the fear memory network that mice have, because, as we said, they have to respond to fear. They have to be scared of predators, because in that way, they increase their survival rates, right? If you are completely negligent to danger, then your success rate drops. So the fact that the mice can learn what's scary what is not scary, and how to search for things implies that they have a very well developed and defined memory circuit. And there is plenty of studies really dissecting exactly which regions in the brains are important for the different aspects of memories in mice. And we use, in particular, this ear memory network, and in my case, an auditory type of ear memory network, where we use, like this very traditional Pavlovian type of paradigm where a food shop, in my case, very mild, because I wanted to use, like, not sometimes you just want to make sure that your mouse remembers something, and therefore you might use very, very well established protocols that your 100% sure will elicit a behavioral response that we can translate into a yes or no type of output in yes memories, there or not, and in the context of fearful memories, the freezing is one of the types of behaviors that is, one that is very easy to detect in mice, and therefore there isn't much interpretation from the experiment nowadays, we can even use, like this, machine learning type of strategies to be even more blind to the type of scoring we give device and make sure that it's not our bias that says yes the most Yeah,

Nick Jikomes 38:58

yeah. But in this case, it is, it is very obvious, right? Like, if you've seen this with your own eyes, if you pair the tone, the auditory stimulus, with the shock to a mouse, they will freeze. And it's, it's quite dramatic, the freezing, yeah,

Giulia Santoni 39:12

and it's instant. And so as soon as the tone is played, and you really see that the mouse first was chilling, sniffing around, and then it really stops, and kind of also, like closes up and high, like not hides, because sometimes there is no space for hiding. But really goes more to the periphery of the chambers where we usually run these experiments, and says in the corners that it's also quite a sign who like of of danger and of being scared and and, yeah, yeah. And so in my case, I use this instead, this milder version, where the, let's say, the US, we call it the unconditioned stimulus. So this shock was not obviously extremely dangerous. We wanted to have a platform where we could see but when playing with acetal. Levels, whether this could change the salience of the memory. So if this has tolation had any impact either on increasing the memory, making it stronger in the expression type of behavior, or maybe weakening the memory and so but the process, procedure of the protocol, was the same, so these mild shocks were associated to sounds on day one, and then on the following day, you change the subject to make sure that it's really like a type of auditory memory and not like a contextual memory. So you change the chamber, you change the order, you change the shapes around the chamber, so that then by playing the tone again, you check whether the mice recognize it or not, by by by changing their behavior and increasing or decreasing their freezing

Nick Jikomes 40:49

and in which part of the brain are you looking at here? Where is this stuff happening?

Giulia Santoni 40:53

So we use as a target this region called lateral amygdala, and amygdala in Greek means, like almond. So because it's an almond shape type of region, we have two Amygdalas in our brain, and so do mice, and they are really, like at the bottom, lateral, bottom part of their brain. And this region is, has been shown by many, many brilliant studies, to be really fundamental in the formation of these auditory fearful memories, because it's the recipient of auditory projections, so auditory inputs that come from either the auditory cortices or the thalamic auditory in neurons, and therefore, at the same time it also receives this painful like dangerous type of signaling. So it's an incubator, and it's this hassociative hop that by receiving both inputs auditory so the stimulus and the painful stimulus, then it can combine both stimuli in neurons that will now be able to respond even to only one of them. And this is why we say that here, there must be some type of potentiation, because at the time of the recall so the next day, when the mice are probed for their memory, what we do is to only play with one of these two stimuli. We only induce a response of the auditory cortices, or these auditory inputs, so there will be only that type of activation on the lateral amygdala neurons, so without no painful stimulus. And now we see that by only having that one activity these neurons, they can still be activated again and trigger this defensive behavior, this being afraid type of behavior called freezing.

Nick Jikomes 42:52

I see So, so there's a part of the brain, within the amygdala, where there's a convergence of information. You get the auditory information that tells you about the tone that the mouse is hearing, and then you get the information about the foot shock, the physical sensation that's coming in. And then, obviously, these two things are being paired together. So there must be some kind of reconfiguration in this brain region that enables the animal to make this association and remember it, and then sort of connecting that to what we were saying before. There's a bunch of neurons in the lateral amygdala that could underlie this memory, but only some of them actually kind of get hooked up into the memory, so to speak. And so there's this question of why some neurons and not other neurons, and it sounded like that had to do with the acetylation of these histones and how ready they were to turn on certain genes.

Giulia Santoni 43:42

This was precisely our hypothesis. So to give you a little bit of a background, I do remember so it's thought, I mean, there are plenty of studies have explored this region in detail, and what they saw is that there is around 70% of the principal neurons, which means this excitatory neuronal population that can receive both these inputs, so the foot shock information and the auditory stimulation. So among all the neurons in the LA of the excitatory neurons, 70% is perfectly suited to become part of the memory network, of this associative memory network. However, when we have like tools to really detect how many are actually responding and forming these memory traces, and what we see is that usually between five to 20% are actually recruited at that moment. So in the last two decades, really, like many scientists were wondering, how come, like, how come neurons are selected? How come only so little, and how what's the what's the rule? Is there a rule? And initially, of course, anything we don't understand, we call it random. So initially, this. It's called memory allocation. So before it was thought to be a random process, so randomly, a memory has to happen, and some neurons are selected. And we didn't even think it was an active process, this was a passive, random type of mechanism through which neuronal selection was was taking place. However, then many groups, like, for example, Sheena, Jocelyn, Paul Franklin, dalcino, Silva, really started to dig deeper into into the process of memory allocation, to try to understand whether it is passive or active, whether there are some rules and what they really try to understand is whether there was a type of signature that some neurons had that made them better at becoming part of the memory network. And they found that there were two features that would be predisposing some neurons to better be to be better placed for memory encoding, one feature being intrinsic accessibility. So we come back to this concept of readiness of a neuron to respond to stimuli, right? So intrinsic acceptability is this process. Is this somatic? So somatic being this body area of a nucleus. So there we have again, channels sodium and potassium channels that define how ready is a neuron to fire, how much current do I need to inject in a neuron to see a spike? And we see that not all the neurons have the same type of intrinsic acceptability. So it seems that if a neuron has it a bit higher, then its likelihood to become part of the memory trace will be higher, and equally, again, they also tested this with creb, so again, that important transcription factor mediating many different genes. And they also saw that crab levels can be different, and the neurons that have more crab levels are also predisposed for memory formation.

Nick Jikomes 47:03

I see So neurons that are more excitable, more sensitive to their inputs, to receiving information, they're more likely to become hooked up into the memory trace. Yeah,

Giulia Santoni 47:15

exactly. And and this for us. I mean, while I was thinking about, Okay, what do I do with my PhD? I mean, what kind of project will I ever have? I really got hooked by this, by these studies, because I really thought, okay, here we're talking about neurons that are seemingly homologous. They can be twins. You know, they can do like. They are like, Yeah, almost identical. However, one response better than others and and the fact that there were like physiological signatures such as intrinsic acceptability or higher crab levels, to me, was really speaking in favor of somehow a super identity, like a new a new level of identity. On top of being the same type of neuron in the same region, connected in the same way, you also have an extra tag, an extra lab type of label that tells you, okay, you're even more excitable. You're even more like, Yeah, ready to transcribe genes with prep, right? So I really thought, what if the spot that you randomly have some neurons having these higher features would derive by higher acetylation levels in those neurons? Maybe those neurons, they can be more intrinsically excitable because they have a DNA confirmation that is more open, and therefore that allows for the transcription and the maintenance of this higher intrinsic acceptability states. So this is really the beginning of the story,

Nick Jikomes 48:53

okay? And so basically walk us through some of the major findings here. So the idea was, okay, you're in the you're in the lateral amygdala. This is a place where we know that information converges. That's really important for memory. It's very well established that this region is important for these type of Pavlovian conditioning fear memories that you're inducing in the mice. We know that you've got a bunch of neurons there that are seemingly identical. They're the same type of neuron in the same type of place receiving the same type of input, but only a fraction of these neurons for any given memory are actually going to be recruited into the memory. And you have these ideas around, okay, well, maybe the neurons that are recruited have their chromatin more accessible, so they're, they're they're more they're more readily able to turn on the genes necessary to, you know, strengthen synapses and do things like this. And so how did you approach this, and what did you

Giulia Santoni 49:45

find? So first of all, we have to answer a very, very simple question, which was, do we see that the lateral amygdala neurons have different acetyl. Levels like, do they differ in their content of acetylation? Because, of course, if you would see that all the neurons in the LA had the same amount of histone acetylation, then there would have been no point right in pursuing this, like big, big research. And so we did some, some studies where we tested different types of epigenetic marks, some that would decode and attack like more closed chromatin, like pieces of heterochromatin, such as HP one, which is a protein quite known in epigenetics to really identify closed area of the of the genome. And we also use different kinds of histone acetyl marks to really check whether it was because, as we said, histone acetylation can be very complex. So depending on where, on which type of amino acid histone acetylation happens, then you might have different responses. So we had a look at different epigenetic marks, and this was the first time where we saw that we see different levels of histone acetylations And in general, chromatin confirmation in seemingly identical neurons in a structure. So in the LA, this was the first time that we could explore the diversity of the overall content. So this was not specific to any genes was really like a like brute force type of approach to really go and look at the abundance of some marks. And what was, is also interesting, was that we could correlate which neurons were active after a memory encoding. Which means memory encoding. We say it when we want to stop the paradigm exactly at the moment where the mice learned the association. So this is day one. The mice went for the first time in this chamber. They had to learn to associate the foot shock to the tone. And immediately after, they are probed for the for the presence of a protein called C FOS, which is a gene that is like a rapid response genes that is really highly present in neurons that are activated at a certain time. So by Yeah, by correlating which neurons would be marked by this gene with the highest acetylation, we could really see that after a memory is encoded, the neurons that are more active are the ones that have more acetylation

Nick Jikomes 52:28

I see here. So just at any given time, you sort of look at the acetylation levels of neurons in this part of the amygdala. And just it just happens to be that some, some cells have more unbound available DNA. Some cells have less so then you induce the memory, and right at the beginning of the memory, you look for the signal, this gene transcription, signal of plasticity, basically. And you find that, okay, the neurons where the signal turns on also tend to be the ones that just have more open and accessible DNA. Yeah.

Giulia Santoni 53:01

Wonderfully. Sad, exactly. And this was a fantastic entry point for us, because we could say, okay, at least at the correlative level, not only do we see diversity and heterogeneity in the acetylation profile of excitatory neurons, but we also do see that there is this correlation, which was exactly what we wanted to test, right? We really wanted to now move from a correlation to more of a causation type of approach, where we could assess whether, by over expressing the enzymes that can increase the levels of histone acetylation, we would see that only the neurons in which we induce the like this increase in acetylation, those would be the ones that, upon a certain protocol for memory formation, would become the one harboring the memory trace. So the experiment now went from an observational type of approach to really more experimental and manipulate. We manipulated the system because we went into the middle of some mice. We only sparsely modified some of the neurons that we wanted to see whether they would become more like a prone to be, to be part of the memory trace and afterwards. So once we increase the levels in a sparse population of neurons, we could check okay, if we now subjectimize The usual memory protocol, what does happens to those neurons? Do they become part of the memory? Trace more or not? And what we could see is that we would really like compared to chance levels, because these experiments are always done with like a control where you just kind of do all the same procedure, but instead of changing the acetylation, you just add like a fluorescent protein that just tells you that that neuron was touched, essentially. So compared to this control group, we doubled the chance of neurons to be. The lab by just playing with the acetylation I

Nick Jikomes 55:03

see, so you're able to inject something into this part of the brain sparsely, so that there is an increase in the number of neurons that have chromatin in the open configuration. And you find that when you do this, those neurons that you end up manipulating, are more likely to be recruited into the memory. Basically, yeah,

Giulia Santoni 55:24

yeah, precisely. And what we wanted to do with that was also to make sure that we, I said sparsely, for a reason, because if we would just load the overall structure with more acetylation, you couldn't really see, like the network changes within the same animals, right? Instead, by sparsely playing with only a few, then you can kind of keep your manipulation consistent to this five to 10% of neurons touched, which is like the same percentage of neurons that usually shine up upon a memory formation event. Yeah. I mean,

Nick Jikomes 55:59

I would imagine that might also result in memory changes above and beyond what naturally happens, like if all of the neurons get recruited, maybe the animal, I don't know, maybe it can't remember it as well. Or maybe it

Giulia Santoni 56:13

remembers there are, like, plenty of studies. For example, you can play with epigenetics also using some drugs, right? So there is, like it's also an FDA type of drug, like approved drug the h tag. So H DAG is this class of enzymes that removes acetylation. And if you use inhibitors for this type of class, which implies that, then if you remove or inhibit the enzymes that remove acetylation, the outcome is that you have higher acetylation level. So this is global in the whole brain, and yet, you see that the mice tend to have a little bit more improvement in memory retention. So I think the brain is great, because it can always compensate somehow, even intrusions, you know. So, so who knows if we would plot the overall system? Who knows what would happen by the by this study, we see that it's not so easy to disrupt brain activity and brain functioning.

Nick Jikomes 57:13

And so, okay, so you can, you can render a subset of neurons in this part of the brain more likely to be recruited into the memory by changing acetylation, by changing how open the DNA is inside of those neurons, and then you see that they are recruited into the memory. Basically, what about sort of the other side of this? Can you disrupt those neurons and get rid of the memory?

Giulia Santoni 57:41

So we did sev