Loading summary
Transcript82 lines
- [00:04]
A
Welcome to Brain Science, the show where we explore how recent discoveries in neuroscience are helping unravel the mystery of how our brain makes us human. This is episode 150 and I'm your host, Dr. Ginger Campbell. Before I tell you about today's guest, I want to correct a mistake I made in episode 148 when I accidentally said that that October 1st was the deadline for signing up for our trip to Australia. I had misunderstood the instructions from my travel agent. October 1st was actually the first day that you could put down your deposit, so there's still plenty of time to sign up. Today's guest is Dr. Seth Grant from the University of Edinburgh in the UK. This is his fourth time on Brain Science, which is a record, but the reason I keep having him back is that he keeps making new, mind blowing discoveries. When we first Talked back in 2008, he introduced us to the evolution of the synapse and the surprising fact that vertebrates have more proteins in their synapses than invertebrates. Today we are going to take that discovery even further as he tells us about a new paper that describes the first ever synaptome, which is a map of the synapses and it is the synaptone of the entire mouse brain. Don't worry if that sounds esoteric because Dr. Grant is an excellent science communicator. And no matter what your background, this will make sense to you by the end of the interview. As always, you will find complete show notes and episode transcripts@brainsciencepodcast.com and you can send me feedback@brainsciencepodcastmail.com of course. I'll be back after the interview to review the key ideas, so please listen all the way to the end.
- [02:06]
B
Seth, it is so great to have you back on the show.
- [02:10]
C
It's a pleasure to be here.
- [02:11]
B
I was looking at the old episodes and I realized it's been almost 10 years ago since we had our very first talk 100 episodes ago. And I'm glad that we're filling our goal to talk again in only a year since our last conversation was episode 137.
- [02:32]
A
So we did good.
- [02:33]
C
Yeah. And we've got a totally different story for everybody today.
- [02:37]
B
Right. We're going to focus on your group's most recent paper that came out in Neuron. But for the sake of new listeners, you start out by just telling us a little bit about genes to cognition.
- [02:50]
C
Yes. That was a program of research that I started back in the early 2000s, and this was at a time at which the human Genome had recently been sequenced, and my laboratory was focusing very much on trying to understand the genes and the proteins and how they control cognitive functions, including learning and memory and other behaviors. And it was all very well to be thinking about genesis and their effect on behavior. But what you want to do is sort of fill in the gap between that. It's also finding out where the proteins work, which parts of the brain they conduct, their functions, and ultimately how the nerve cells fire and process information. And from that, we wanted to try to have an integrated research program. So it brought together a big team of scientists who all had this sort of common objective, overarching objective, but who were very much focused on thinking about it Right from the level of the gene or the genome. Right.
- [03:50]
B
So your background's in molecular biology?
- [03:53]
C
That's right. I've worked on the molecular biology for over 30 years now. And in particular, the molecular biology of synapses in the central nervous system, and with respect to synapses is actually on the post synaptic side, which is the side of the synapse where the information comes in, which is because neurotransmitter is released from the presynaptic side, and it impinges on the postsynaptic side and causes those postsynaptic neurons to process that information.
- [04:23]
B
So the proteins that are on the postsynaptic side are the ones that really determine what happens to the message. I mean, a neurotransmitter could hit a different receptor and get a different response, correct?
- [04:38]
C
That's right. And, you know, for a very long time, the view of the synapse has been a rather simplistic one, which is, namely that the synapse is a connector between nerve cells. And in fact, that is in fact, what the name synapse means. I believe it was coined by Charles Sherrington at the turn of the 19th to 20th century. And it is derived from the Greek word to clasp. It is the connector. But we think of it in a very different way today, because inside synapses, there are very large numbers of proteins, and those proteins are doing far more than just serve the simple job of being a connector. What those proteins are doing is processing information. And what I mean by that is that on this postsynaptic side, which is where information comes in, that information is really kind of simple. It's actually just pulses of neurotransmitter separated by time. Sort of like a code, like Morse code, You know, goes dot, dot, dot, dash, dash, dash kind of thing. You get a different pattern and frequency of these Stimuli occurring, and these post synaptic proteins are reading that information and deciphering that information. And thus the synapse we now think of as a molecular computer.
- [05:54]
B
Right. And it's been a long time since I talked about this on my show. So I wanted to emphasize the fact that it's not the neurotransmitter really that determines what happens. It's what's. I mean, it's part. But a big part is what's happening on this side that you are studying.
- [06:11]
C
Indeed. And in this genes to cognition program, one of the things that we did was to make a lot of gene mutations in those genes that encode those postsynaptic proteins and then ask, how does this information processing capacity, the synapse, change? And how does behavior change as well? And we found that there was a very beautiful molecular combinatorial system for decoding information. In other words, you have all these different molecules in your synapses and some of them read certain types of information and other ones are reading other types of information. And collectively they're capable of reading a sort of a large vocabulary of information as it comes in.
- [06:55]
B
We talked, I can't remember which year it was about the experiment with the knockout mice or one of the experiments with knockout mice where you were able to show that learning was affected by changing just one of these proteins in the synapse. So I'm thinking that what we're going to talk about today is almost a follow on to that in some ways.
- [07:19]
C
Yeah. What we're going to talk about today is not so much as a study where in the past, as you quite clearly said, is that where we mutated genes and see how behavior changes. What we've done here is to genetically engineer these synapse proteins in the mouse in a very special way. What we've done is to modify them so that when the gene makes the protein, it doesn't just make that important protein that goes into the synapse, which is used for information processing, but it has made it visible by the fact that we fused protein with one of these fluorescent proteins. And so then we can visualize the synapses with microscopes and look at where those proteins are expressed and how much of them is in any individual synapse. And in this study, we did that for two of these very important proteins. They had different colors. And so when you have mice with these different color proteins in their synapses and you see these most beautiful and spectacular pictures of the brain, where all of the individual synapses are lighting up but with different amounts of these two different colors. Right.
- [08:26]
B
And so, as we start to talk about your new paper, what other kind of background information do my listeners need?
- [08:34]
C
Well, I think some of the key background information is just to understand, in a way, our historical understanding of how the brain works. And in the sort of traditional model that's been around for at least a century, it's been thought that each and every behavior that a person or an animal has is because they have a certain circuit of nerve cells for each of those behaviors. So, in a sense, your brain is made up of a large number of different circuits. And every time you do a particular thing or you perceive a particular image or say a particular word, you're using a special set of nerve cells. In other words, one circuit equals one behavior, if you like, kind of a simple model. But in all of that, the synapse is really kind of a simple thing. It's just this connector, and it's necessary to hold together all of those different circuits, if you like. But I mentioned to you that we have been changing our position because we know that the proteins in the synapses are very complicated. There's lots of them, and they're doing this very sophisticated computational job. And what we're now leading into in this new study is this idea that synapses are, in fact, very different. So the traditional model has been basically, all the synapses are more or less the same. All of the excitatory synapses are the same. And there's also inhibitory synapses. There's very few synapses. That's our traditional model. But in our recent work, we have found evidence that, in fact, there may be a very huge number of different types of synapse, Great synapse diversity. In fact, it may even be that no two synapses are identical.
- [10:14]
B
Wow, that's surprising.
- [10:16]
C
Well, I think it's a surprising and beautiful thing, because once you start to think like that, and, well, not even think like that, once you see the evidence that supports that, you begin to realize that if they're different to one another, that means they have different functions. And we think that these different molecular compositions of these synapses are relevant to different behaviors. So just again, to help the listeners, the standard model is that for every behavior, you have a particular circuit of nerve cells. That's the old model. And in our model, we think that each behavior preferentially uses different synapses of particular molecular compositions. And so you have all this diversity in the brain of synapses, and somehow during your behaviors, you're sort of selecting these ones for this behavior and other synapses for those behaviors and so on. Now, I should point out that these two historic, the historical model and our model actually can work together. They're not mutually exclusive, because what we would now say is that our different synapse types are in fact, being distributed on these different circuits, and they help even explain how this traditional model really works.
- [11:25]
B
Right. So then in your new paper, architecture of the Mouse Synaptome, this is a sort of a first step in this whole approach.
- [11:35]
C
One of the key things about this study. Now, let's leave aside behavior for the moment and let's just talk about what we have seen and done here. Because we could light up these synapse proteins and see the synapses, we reasoned that what we could do is examine synapses across the whole brain of the mouse. And what I mean by that is that we could make thin brain sections of the mouse, and on a microscope slide, and then using very fast microscopes at very high magnification, we could scan all of those sections and look at each individual synapse. We looked at around about a billion of them. And for every single one of those, we looked at how much of these different proteins and the light was being emitted and their size and shape. And this allowed us, for the first time in any organism that we know of, to create a synapse map across the whole nervous system. And that's why we call it a synaptome map. It's like the word genome. The genome is the complete set of genes. The synaptome is the complete set of synapses. And so we've created this first real set of synapses across the brain of an organism. And this really opens up all sorts of new directions for us.
- [12:49]
B
And in this study, you've labeled particular types of protein machinery that's in the synapse, so that you've studied the distribution of just two of these proteins, and there's actually, what, thousands of them?
- [13:02]
C
That's right. And that's quite a fascinating thing. So it's really quite simple. Imagine now we had a traditional model of the brain where all synapses are equal. And if it were that there's a thousand proteins that can be found in synapses, then, well, all synapses would have each of those thousand proteins, and they'd all be the same. That'd be kind of a traditional model. But what we have seen here, by just looking at two proteins, we can get not Just those with protein number one and other synapses only with protein number two and some that have both. But there's actually even more varieties because some of them are big and some of them are little and different shapes. So, in fact, you can, from just two proteins, you can get as many as 37 types of synapses. And now, if you would imagine if you had three proteins or 10 proteins, if you had 10 proteins, you could have enough different synapses in the mouse brain. So just only with 10. But as you have rightly pointed out, there's over 1,000 proteins. That's why we think that there's potentially a limitless number of different types of synapses within the brain of an animal.
- [14:11]
B
So do you want to just tell us the key findings? I mean, you've sort of alluded to.
- [14:15]
C
Them, but this paper is really comprised of different sorts of sections. In the first part, we present the fact that you can actually produce whole brain synaptome maps, and we present data from many hundreds of different brain regions where we count up and quantify all of these different types of synapses. So in the first part, it's a sort of a bit of an accounting exercise, but then we move into asking how these different types are distributed and are they organized in any particularly interesting way? And indeed, they're organized in many, many different interesting ways. I'll just give you a couple of examples. You find that if you look at an individual dendrite on a nerve cell, that some of these subtypes are distributed in a very organized fashion, where the synapses might be quite small, closer to the cell body, but they get bigger as you go away from the cell body. Then if you go up again in scale, you might say, let's look, in different regions of the brain. Do they have different compositions? And we found that more or less every brain region has its own signature or composition of these subtypes of synapses, and that gives them obviously, their differential properties. But another thing is that we found that these differential signatures are sort of organized on a global scale across the whole brain, where some areas of the brain are very diverse with respect to synapse types, and others are very sort of limited. And one really interesting and little provocative finding that comes out of that is that those parts of the brain that are involved with really these higher cognitive functions, cortex and hippocampus, for example, those parts of the brain involved with higher cognitive functions have the most diverse collections of these synapse types, which would make sense when you Consider that the different types of synapses give more functionality and presumably they're contributing to cognition in that way. So that's something on the global architecture. But then we also related this. We asked the question, could all of these synapses be in any way related in their molecular composition, related to how the brain wiring is all connected up? And indeed it is. And we could show that essentially the molecular composition of individual synapses will sort of tell you in a sense where that axon came from. It came from the cortex. It'll have this particular composition of proteins. If it came from the brainstem, it would have this particular composition of proteins. In other words, there's a kind of a molecular code for the wiring diagram of the brain that you can find in these synaptome maps. And then we related all of that to functional brain imaging, because the connectome of the brain that is all a wiring diagram is thought to be important for the way you can see the brain active when you stick a head in a scanner like an FMRI machine. And we used FMRI data to correlate with our synapse maps. And we found that there was a very nice correlation suggesting that these molecular signatures and organization of the synaptome is in fact, ultimately relevant to the firing and overall global neural activity of the brain. So that was some of the stuff that we did in terms of brain anatomy.
- [17:26]
B
Did you use any of the information from the Allen brain Atlas, since you're. Aren't you both working with mice?
- [17:35]
C
Yes, indeed. And the Allen Institute for Brain Science has made a wonderful contribution by their generation of a reference atlas for the mouse brain, which essentially gives you sort of coordinates that define the different regions and sub regions of the brain. And we took that sort of reference coordinate map and aligned all of our synapse data onto that. And we produced a very nice website and tools by which people can now analyze and view our data using this Allen Reference Atlas data. And we also use the data, the Allen Reference Atlas data. It's very handy for being able to connect to other data sets. For example, these brain imaging data sets are sort of annotated according to this reference atlas.
- [18:22]
B
So every time somebody like you does work, hopefully they put it into a place where other people can use it, because this is hard stuff to do, I would think.
- [18:32]
C
Absolutely. And it's very important that these data sets are made completely available, but not only available, but usable. And we went to some lengths in the lab to produce a very nice tool we call the Synaptome Explorer. Which your readers can find or listeners can find online, which is a tool whereby you can look. It's sort of like the Google Maps view of the synapses of the brain. You can sort of look at a brain section and then you can sort of zoom right in and you can see individual synapses, and you can see these vast and beautiful fields of these different colored synapses. And then you can turn on and off various buttons that will tell you which types of synapses they happen to be and tell you exactly where they are. Now, this is not just a pretty thing to look at, but it's also very useful because obviously lots of people who are sticking electrodes into the brain and recording from it, they now need to take into consideration the fact that where they place their electrode, they'll be recording from certain types and subtypes of synapses that we've classified in our synaptome maps.
- [19:35]
B
Your work really seems to indicate that this complexity of the synapse means that also when you're going, say, between vertebrates and invertebrates, because most of this synaptic complexity seems to be vertebrates. Right. How does this affect, say, Eve Marder is doing her work in the lobster and does a lot of stuff with circuits. You would expect that if she went in and tried to do a synaptome on the lobster. Of course, that's not exactly the same because you don't really have the true kind of the brain's not the important part in the lobster. But let's say she did that on her somatogastric ganglion. You'd expect those synapses to be less complex.
- [20:18]
C
Yes, we would. As you quite rightly pointed out, we identified a number of years ago that the different types of proteins that you find in mammalians and other vertebrate synapses, there's many more types of proteins than there are in the invertebrates, including the lobster. And that's because there were in the early vertebrate lineage, two genome duplication events that essentially took the same ancient machinery of the invertebrate and multiplied it up so you have varieties of the same genes. We call those paralogs. Now, what we did in this present study, these two proteins that we studied, one of them called PSD95 and the other one called SAP102, those are two copies of a gene which was in the invertebrates. There was only a single copy of those. And in fact, these are examples of duplicated genes. And by virtue of the fact that we readily see that the PSD 95 and the SAP 102 have different maps across the brain, then it follows that the duplication event led to the generation of a more complex synaptome map and more diversity of synapses, which is a characteristic of the vertebrate. So if Eve, Marta or others were to go back and map the invertebrate nervous system, I expect they would find a very much lower level of synapse diversity and thus complexity in their nervous system.
- [21:49]
B
But even so, one of the things that Marder has done is showed that there's an awful lot of diversity even among a lobster. I mean, you can't even just say, I found this in this lobster, it's everybody. So I would think the kind of diversity that she's discovered is going to be like magnified. Looking at even the mouse.
- [22:11]
C
Oh, most certainly it will be magnified or it is magnified. There's no doubt about that. But you see, the notion that we're now pushing toward is that the synapse diversity is actually an important and rather fundamental aspect of how the brain operates. And we all know that the way the brain operates and much of the work of Dr. Marder and others has focused on the firing patterns of nerve cells, their electrical activity. And one of the aspects of our current study that we performed was to ask, how do these different synapses, which have different molecular compositions, respond to different patterns of activity? And we used computational approach done by Professor Eric Franzen from Stockholm, and he demonstrated that these different synapses can respond, give a different output when you present them with the pattern of activity. So the sort of simple concept, the sort of bigger concept that comes from that is that you have many different types of synapses, synapse diversity, and that produces different responses to patterns of activity. And this is a beautiful thing, really, because even though it sounds sort of simple and maybe sort of rather trivial, it leads into a new model for how memories can be stored and recalled.
- [23:30]
A
Right.
- [23:31]
B
Because it would be recorded differently according to which synapses it hits.
- [23:36]
C
Well, yes, and let me put it like this. If the molecular composition of synapses controls how it responds to patterns of activity, then if I change the molecular composition of a synapse, then it'll change how it responds to a pattern of activity. Now you hear that I'm emphasizing response to a pattern of activity. The standard model of learning and memory is one where that isn't the issue. What is the issue is that it's the long term stable strength. And on your other podcasts, From Eric Kandel and others, they will tell you quite a lot about the studies that have looked at long term stable strength and its role in learning and memory. But what we are saying here is something quite different, which is that when a pattern of activity comes in, it might have, let's just say, a simple pattern like a 5 Hz train. Synapse number one might respond very effectively to that, but synapse number ten might not. But then you have a different frequency like a 2 Hz train. Well, it might be that other synapses then respond preferentially to those. So if I now have lots of synapses in the brain, I can modify different ones so that when patterns of activity come in, you can recall information that's been stored in this map of the brain.
- [24:53]
B
So it's an entirely different way of thinking about memory.
- [24:57]
C
It is a different way of thinking about memory. And I think it has certain similarities with the standard model and some major differences. I just want to emphasize the similarity is that in the sort of standard model where long term strength is a key thing, and in our model, where these patterns of activity are the key thing, both of them are dependent on the molecular composition of synapses. That's the first thing. So there's a commonality there. But what we are saying is that this issue of recall, you see, that's a really an interesting problem in learning and memory research. Most people study how information is stored. But frankly, I think it's quite mysterious how information is actually recalled. I mean, how does it pop out when you recognize something?
- [25:37]
B
Or not pop out?
- [25:39]
C
Yeah. Or not pop out. Well, our studies indicate that it's when a pattern of activity goes into this field of synapse diversity. And I emphasize the diversity of synapses is essential in our model that when you have diverse synapses, you can instantly recall the information out of that. And we show in our paper, in a sense, how when an animal is doing different behaviors, you get these sort of different responses of the synaptome map. And we have a nice video in our paper as well, which sort of illustrates that rather nicely.
- [26:10]
B
So how would you go about testing this hypothesis, I mean, to like say someone's going to figure out which model is closer to reality. What would be the next step?
- [26:22]
C
Yeah, that's a fun question. To some extent we've been going down that track because we have used genetics and we have some work which we hope to publish in the near future, which basically asks the following type of question. If we make gene mutations, can we See any evidence that the long term stable strength is the key thing, or this pattern detection is the key thing. And there's a general strategy used by geneticists and indeed other kinds of scientists, which is called genetic dissociation or genetic dissection. And Paul Nurse, the former head of the Royal Society, gave a nice description of it once on television when he tried to explain to people that if you want to look at a person who has lung cancer, you might notice that they have yellow fingers that correlates with their lung cancer, but it doesn't mean that it caused their lung cancer. You know, the yellow fingers didn't cause cancer. And obviously we know that cancer is caused by gene mutations in cells in the lung. But people can often be fooled by correlations. And what you're looking for is dissociations of correlations. So in effect, in the case of lung cancer, you might like to show, here's a person who doesn't have yellow fingers, but they still have lung cancer. And that would in a sense prove that those things can be dissociated. So we did this big genetic study where we made lots of gene mutations in lots of mice, and we studied learning in all of these mice, and we studied this long term change in the synapses, something called LTP, and we found that we could dissociate that with 45 different genes, which is a very, very robust dissociation. And what that suggests to us and is certainly consistent with, is the notion that the long term change in stable strength isn't really the key thing, and it's this other potential model that we're suggesting. So this will be something that will be hotly debated in the future, I'm sure, But at least now we've got a new model on the table for consideration when thinking about how storage and recall occurs and the important role of synapse diversity.
- [28:25]
B
So can you talk a little bit more about how you use the mutations in your study discoveries? I think there are some good examples in this paper.
- [28:33]
C
Yeah. In this particular paper, one of the things that we did was to take two different lines of mice, one of which has a mutation in a gene which is a recurrent mutation in human schizophrenia. And another one was a mutation that in humans produces autism and intellectual disability. In this particular study, what we did here was not to measure the behavior of the animals. We have indeed done that in previous studies in these mice, but we actually measured this synaptome map. In essence, what we asked was how does the gene mutation that causes this intellectual problems does it change the maps of the brain? Does it change synapse diversity? And the answer to both of those questions is yes. And moreover, the changes we saw because of our ability to do this whole brain mapping, we could find those parts of the brain that were damaged in these disorders, which is actually something that's very, very difficult to do. But we now have a simple and straightforward method whereby with a gene mutation, we can find where in the brain the synapses are affected. And we found that the changes are really very widespread. So this implies then that these particular disorders, schizophrenia, autism, for example, they don't just have a little change in a few synapses in a small part of the brain. They actually have rather large widespread changes. And this presents the question now for, for therapeutic approaches, is there some way we might reverse and rectify that? And that's something we're interested in.
- [30:01]
B
So what was the biggest surprise of the work that led to this paper?
- [30:07]
C
I think the biggest surprise, or the most enlightening point, if you like, for me, was it pushed us into considering the importance of synapse diversity at the molecular level. And although I haven't gone into this, one of the key things in our paper was not just to say, hey, look, there's a lot of synapse diversity, but what we really wanted to do was to explain how the genome produced that. And that was achieved by choosing these two proteins. I mentioned they were not chosen at random. They were very carefully chosen. And that's because we know quite a lot about the biochemistry and molecular biology of the proteins. These two proteins physically assemble these sort of molecular computers, and each of these two proteins assembles different molecular computers to one another. We therefore, by looking at these brain maps, could see the distribution of all of these molecular computers. So this really has given us some really sort of deep insight into what is ultimately going to be the genetic programming of behavior through these synapse maps. And I find this the most fascinating and deep question.
- [31:17]
B
So one of the things I didn't really realize is the current resolution of these high resolution microscopes. Could you talk. You talked very briefly about that in the paper because it was something you had to use. But I don't think that the average person realizes the kind of resolution that's now possible. Could you talk about that just for a second?
- [31:38]
C
There have been marvelous advancements in light microscopy over the last 15 years. And one of the methods that we use is a method called confocal microscopy. And in this case, the sort of resolution of that gets down to about say 2 to 300 nanometers in scale. So if one thinks about, say, a human hair, for example, that would probably be like a hundredth of the width of a human hair or a thousand, some very small amount. And a synapse is really quite small compared to a normal cell. Synapse measures about 1 micron. That's 1 micrometer across, that's a thousandth of a millimeter across. And in each one of those, we can then detect the presence of these proteins. But I should point out to you that there's another form of microscopy which we've used in other studies, which goes to even higher level of resolution. It's called super resolution microscopy. And that allows you to see, not to the resolution of say, 2 to 300 nanometers, but sometimes less than 100 or maybe less than even 50 nanometers, depending on the particular technique. And this has been a wonderful and important sort of advancement which has allowed scientists like myself and others to see where the proteins are within synapses. And you can find some proteins are sort of over here and a little bit of some of them are closer to the center and some of them a bit further away. So we can actually now be getting down to this sort of nanoscale, as it's called in terms of the molecular architecture.
- [33:16]
B
So if I'm looking at a, say in your paper where there's a diagram of what the particular protein looks like, all those little globules, I mean, that looks strange to me as a non biologist. How close is that to what it really looks like?
- [33:31]
C
We certainly can't be resolving the level of the actual structure of the protein or even the structure of the protein complex, which is the protein complex is where several different proteins bind together into a little molecular machine.
- [33:45]
B
So that's still extrapolated from the data.
- [33:48]
C
That's right, yes. We can say it's there or approximately in that very location, but we can't tell anything about the particular shape of the molecules or their protein complex. Is there other types of microscopy that are in the sort of electron microscopy sort of range that are relevant to that?
- [34:04]
B
Could you talk a little bit more about how this relates to the connectome, since that's a subject talked about often on this show. The synaptome's relationship to the connectome.
- [34:16]
C
Yeah, let's talk about that. The connectome is really sort of the wiring diagram which is this part of the brain or this cell connects to the other cell and it traces out the axons and the dendrites, but it's actually the synapses, which are at the point of the connections between them. And that's what we're looking at. And we're looking at the molecular composition of the synapses. So some of these synaptome mapping methods, for example, electron microscopy methods that have been used, and recently there was a paper published showing a very detailed electron microscopy mapping of the drosophila fruit fly brain. They sort of tell you where the synapses are, but they don't tell you anything about how they're built and thus really much about their functions. In our case, we're actually telling about the molecules that are within them. Now, this is an important distinction because most connectome studies are anatomical studies and not molecular, whereas ours is a combination of molecular and anatomical. And because we're studying proteins and we're modifying genes and so on, ours give us a way of the synaptome we can connect to the genome, whereas at present, the connectome isn't really connected to the genome at all. And I think this is really a very, very important bridge to cross because the genome science has been one of the great victories of the last 15 years as a result of the Human Genome Project. I mean, every day, every week in these journals is publishing new gene mutations important for diseases of the brain and so on. And we need to know in which cells in the brain, those diseases work, in which synapses in the brain they work. And you're only going to know that by looking at the molecular composition. The connectome will tell you virtually nothing about that. So that's a really important functional distinction. But I would just want to emphasize that they're really very complementary. Our molecular synapse maps, obviously linked to this wiring diagram and connectome. And that's one of the things that we demonstrated in our current paper.
- [36:10]
B
You've talked about this, but could you talk some more about the role of the mutations in terms of, say, give us an example from the paper of how we would do a mutation and learn from that with this tool.
- [36:24]
C
Yeah, using. It's a good point that you made about how you use this tool. So this synaptome mapping, we developed a pipeline which we call synmap. And the pipeline means a kind of a standard approach to doing the mapping where you take the fluorescent mice, you make the brain sections in a certain way, you capture the images in a particular way, then you use a suite of tools that we developed to look at that and analyze and produce the data output. And that pipeline, that synmap pipeline, can be applied to all sorts of different studies for the future. There's really a limitless number of applications. And one obvious application is to look in the context of gene mutations of disease relevance. And as I mentioned earlier, we use that to study two different gene mutations. And we showed how these brain map changes. But I'd just like to return to the issue of brain activity and how sort of recall of behaviors work and how these patterns of activity and nerve cells change. You remember that I said the different synapse types have particular responsivities to different patterns of activity in the brain. And if now we have a gene mutation, like a schizophrenia mutation, if the synapses change and the map of them change, which they do, and we showed that in the paper, then the question might be, does the brain now respond differently to patterns of activity? And indeed, we demonstrated that in the paper using a sort of a computer simulation. And this is really quite important because it means that your synapse maps, if they're sort of set up by your genes, and if you now have gene mutations that sort of mess up your synaptome maps, then your response to the outside world, to whatever experiences you have, will now be different to normal people. And this is, I think, really going to be a very interesting way to sort of move forward in understanding how all of these different genetic disorders work and all these other diseases, including diseases like Alzheimer's, will change these synaptome maps.
- [38:21]
A
Yeah.
- [38:22]
B
So I can imagine then that in the case of schizophrenia, this, you could say perceptual distortion could lead to an hallucination.
- [38:31]
C
That's exactly right. That's very much the way we see it, too. In our paper, we show a diagram where effectively these patterns of activity come into these. Now, in a sense, schizophrenia synaptome map, and it produces a perceptual distortion. As you say, the map responds quite differently. And I think there's no reason to suppose that isn't a kind of a substrate or a model for how hallucinations may in fact work.
- [38:57]
B
As you were talking, I was thinking of an analogy, and I want to test it on you to see if it fits. I was thinking of the difference between, like, if I was taking a picture with a black and white camera versus a color film, I would get, my output would be black and white picture, color, picture. Is that at all similar to what you're talking about in terms of these different synaptone responses?
- [39:23]
C
Yes, it is. To the extent that in the color one, you have more information per pixel by virtue of Having many colors as opposed to perhaps just either black or white or a grayscale to correct in that sense. But there is an important distinction between the traditional model and our model. And I'll try to put it like this. In the traditional model, where the long term stable strength is the key sort of information, if you like, then you could imagine making a photo out of pixels which are either black or white. And now let's just say I had a photograph of you that was made like that, and I now have a pattern of nerve cell activity come in like a 1 Hz train. Every time an action potential were to come in, it would in a sense produce this picture of you. But if I was sending in a train of activity at 1Hz every second, there's a particular new impulse. It would generate the same photograph of you every time one of these action potentials came in, no matter how rapidly they came in or how slowly they came in. And that would be the traditional model. But now I'm going to explain to you how in our model you get much more than a snapshot, you get a movie. And what would happen would be in ours, when an action potential comes in, yes, it would generate a picture of you. But now when the next action potential comes in 10 milliseconds later, then it might give it a slightly different picture of you. And if it were that, it was now another 5 milliseconds or 25 milliseconds in different patterns like that, we could now get other pictures coming out of the map. We might get a picture of a cat or a dog or a house or a building. In other words, in this model of synapse diversity, we can store many more images in the same set of synapses than can be stored in the standard model. And this is, I think, a very important thing. You get more information, but you also can get a continuously changing movie in your brain. And I think you'll appreciate, and the listeners will appreciate that when you think of things in your brain, you generally don't think of them in a static snapshot like way. The memories and everything you have, you tend to think of them as a changing image, which is at least consistent with our molecular model.
- [41:41]
B
Right. And it also makes sense in the sense that, well, take short term working memory, when I'm remembering a number just long enough to write it down or something, there's not even time for all this other stuff to happen and then it's gone.
- [41:57]
C
Yeah. And in this sort of setting, although we haven't really fathomed this out in sufficient detail yet we've been considering how the sort of temporal components of this synaptome mapping might be relevant to working memory, these phenomena known as episodic memory, linking it to time and so on. But this is really a very interesting aspect. Our model really opens up this sort of time dimension in a whole new way, which the standard model doesn't quite do.
- [42:25]
B
So if you look back on your career, can you give us a sense of where this paper fits in the scheme? Because you're going in an exciting direction again?
- [42:37]
C
Yeah, I think it's important to change direction every few years because otherwise things get a bit stale. But the direction there is a sort of uniformity to this sort of trajectory in the sense that about 30 years ago I was working on making mice carrying gene mutations and asking how they cause changes in the physiology of nerve cells and behavior. But I was very much focused then on understanding at the molecular level, and thus the biochemistry of the synapse, and asking then how mutations change the biochemistry. And in the year 2000, we applied this method, known as proteomics, to studying the synapse and were very surprised to find that synapses were far more molecularly complex than anybody was thinking at the time. And then over the next decade, we did a lot of work on characterizing the complexity of molecular complexity, that is the numbers and types of proteins. We did this in humans and in mice and in fish and Drosophila and different parts of the brain and so on. And we sort of really charted out all of that complexity at that level. But now what we're wanting to do is go down to the single synapse resolution level through these sort of microscopy based methods and genetic manipulation methods. And in doing that, we didn't just want to look at a few synapses in this little part of the brain or that little part of the brain in the same way that we did the proteomic studies. We wanted to do it really systematically and quantitatively and comprehensively. Because if you know all the parts lists, it's like knowing the genome. If you know all the genes in there, at least you know what you're working with. And in this case, it's very good that we can now have ways to look at all the synapses across the brain and know how diverse they are and exactly what we're working with. And it's opened up this really surprising diversity issue. So in other words, there's a nice connection between, or nexus between the molecular complexity, that is the numbers of proteins you find in synapses and their diversity. And obviously the complexity generates diversity. It's really quite simple. And I think that's really going to be a very important thing in biology. In biology, we've seen the importance of diversity in the context of the immune system. The fact that the immune system can generate vast numbers of immunoglobulins is a key to understanding how the body is defended against all of these infections and antigens and so on. It's because of the ability to select from that diversity. And we think there's some sorts of selective processes going on in this synaptic diversity as well.
- [45:02]
B
So what's next?
- [45:03]
C
Well, it's a good question. One thing next is that the technology that we've developed, we want to make it very much available. You also flagged earlier on that. And we give away all of these tools and resources to everybody. But we also recognize that there's going to be a need to do a lot more of this type of mapping with these two proteins that we studied. They both have very different maps, and we think that probably all synapse proteins have their own map. So in theory, you could generate many hundreds, if not a thousand or more of these different maps. Now, that's actually tractable and doable, and it's going to be absolutely vital and essential if we're going to understand where diseases act in the brain and which synapses are affected. So I see no reason why that shouldn't be done. And so I'm interested at the moment in trying to set up a plan to do that. The other thing, of course, is to study the problem of how synapse diversity is responsible for controlling our perceptions and the recall of information. I think that's really a wonderful problem because I think it could lead to a key to understanding a whole diversity of different sorts of behaviors of all kinds.
- [46:10]
B
And is there, Is there anything else about the paper that I left out that you would like to share?
- [46:17]
C
No, I'd just like to say that many people might like to know how this sort of work is actually done. And in this paper I want to emphasize the fact that we've brought together again, it's the sort of theme of this program I started a number of years ago, which clearly you've got to bring together some really, not only talented and hardworking people, but people who have complementary skill sets. For example, molecular biologist Dr. Naburu Komiyama was really a key individual for building some of these mice that we generated. And then two PhD students, Fei Zhu and Melissa Cizeron did a lot of the microscopy and the actual mapping and data analysis. And then at the computational level, Dr. Zhen Kui, who is an image analysis scientist, developed a whole suite of fantastic analytical tools and really made it possible to make these whole brain synaptome maps. And so you can see by bringing together these sorts of people, you can put together a project that no individual could possibly do. And it's a real challenge to do that, to bring all of these people together. And that's always very exciting to work with them as a team. So I think they deserve really all the credit for this.
- [47:25]
B
So neuroscience, both in the UK and the US is facing similar challenges in terms of funding and challenges for young researchers getting their careers going. Does this influence the advice that you give students who are considering careers in the field?
- [47:44]
C
Well, having been around the block a few times now, one thing I would encourage everybody who's listening, who's a student, to realize is that funding agencies are, by their very structure, very conservative. And what I mean by that is that they typically have a committee, and a committee will always have some people on it who will say, well, I don't think it's very good, and therefore committees are not good at funding really innovative projects. If I look back on the papers that I've published over the years that have had the greatest impact, I would say to you that every single one of those has been met prior. When I wrote out the projects to be done, they were met with rejection. When I've written things that have been fairly sort of mediocre, they have been well accepted. So I think that's a real pattern, and I find that's true for many, many other scientists as well. So I think one of the big challenges for the funding agencies is really being able to see through these sorts of complexities and being able to see the sort of vision that many individual scientists have. The agencies don't seem to be able to replicate that kind of vision. But having said that, now with a method like this synaptome mapping that we have, and of course it's very easy to say retrospectively when you see a nice big PA with lots of data and tools and demonstration, I would hope that this would attract funding because it's going to be a very powerful set of technology that can be used really very widely.
- [49:00]
B
So if a student was interested in this, could they go into the tools that you have out there and do say a little project, or would they need other things?
- [49:13]
C
You can indeed do that. And I have quite a Few students who come into my lab, Even masters and PhD students who are doing their own very specific sort of project, they might be interested in, say, some aspect of behavior, for example, or some aspect of aging or some particular disease. And they take these tools and they essentially do the particular studies on these fluorescent mice, and then they capture the images on the microscope and then use some of the data analysis. One of the main problems at the moment is the data analysis, which is why we're putting a lot of effort into making the tools more usable and more sort of freely available. It's a bit like genome analysis tools. There's a lot of need for a lot of bioinformatics. And of course, there's hundreds, if not thousands of scientists who've been developing those tools over the last several decades. There really needs to be quite a bit more investment into scientists who are going to develop many, many of these tools. But as I mentioned, we have an outstanding individual in my group, Dr. Zhen Kui, who has done a marvelous job to the state. Now that other scientists can go and take away these tools and do the analyses by themselves. We also have been training students and people from other labs to use them. So we hope to spread it as widely as possible.
- [50:23]
B
Well, that must be really exciting.
- [50:26]
C
I hope it's exciting for the students. It's certainly a pleasure to see this stuff being used by others, from my point of view.
- [50:32]
B
So what's the best place for listeners to go if they want to learn more about your work?
- [50:37]
C
I don't have a popular website or anything quite like that. And I'd have to direct really just to the manuscripts at the present time, which we have made freely available. They don't have to have subscriptions because we have them all on open access, so they should be able to download them without much trouble. Right.
- [50:53]
B
That was what I was about to say is this paper is freely available, so I will link to it in the show notes on the Genes to Cognition website. There's some information.
- [51:04]
C
Yeah. And there's actually a website in the paper that we have a link to, which is to the synaptome map aspects itself. You might have a look in the paper for that.
- [51:14]
B
Okay, I'll do that. Well, it was great getting to talk to you again. I hope we can talk again soon.
- [51:25]
A
It was great to talk with Seth Grant again, but don't feel bad if you are a little overwhelmed by today's discussion. If you are a new listener, don't hesitate to go back and listen again. Quite a few of our regular Listeners do that habitually. But I'm also going to take a few minutes to review the key highlights. As you listen to brain science, you will probably come to appreciate that how our brains work can be explored at many different levels. And Dr. Grant works at the most basic level currently accessible, which is at the level of molecular biology. He explores the relationship between the genes and the protein molecules that the genes code for, specifically on what is known as the postsynaptic side of the synapse. The synapse is the most common way that neurons connect with one another. As Dr. Grant noted, the synapse was once thought to be a relatively simple structure. But due in part to his decades of work, we have come to appreciate that it's actually quite complex and that the complexity increases as one goes from invertebrates to vertebrates to mammals. Dr. Grant's latest paper is focused on the synapses in the brain of the mouse. He picked two important postsynaptic protein complexes, and his team developed a way of tagging these with a fluorescent protein that makes it possible to determine exactly where these proteins appear throughout the mouse brain. The key finding was that their distribution varies throughout the brain, and it's actually most complex in the parts of the brain associated with higher cognitive functions, such as the hippocampus and the cortex. We also talked about how these findings challenge our traditional ideas, both about the structure of the synapse and about how memory works. With regards to synapses, the key idea is that not all synapses are identical. Vertebrate synapses are more complex than those of invertebrates, and it's entirely possible that every synapse is unique. Why does this matter? Well, if the synapses are not identical, this implies that their structure may contain unexpected computational power. It means that identical inputs to different synapses could lead to different outputs. On a practical level, this could mean a difference in how sensory input is interpreted, or it could be an important element in memory recall. This is just a high level overview of this very dense interview. If you want to learn more, I strongly encourage you to go to brainsciencepodcast.com where you will find the links that Dr. Grant talked about, including his papers and the free tools that his team has developed. But don't forget that if you are a premium subscriber or a Patreon supporter, you will automatically get a transcript of every episode, no extra charge. If you aren't a subscriber, you can get single episode transcripts, and you can get this episode transcript@brainsciencepodcast.com for only $2. If you're interested in learning about next year's trip to Australia, please go back and listen to the short announcement I posted a few weeks ago. If you email me@brainsciencepodcastmail.com I will send you a PDF with all the details. Meanwhile, I want to mention for those of you who are listening on Android devices, that Google recently released a new app called Google Podcasts. I strongly encourage you to use this app, especially if you've been using Google Play. Up until now, the best way to listen to Brain Science on Android has been through the free Brain Science mobile app, which I still encourage you to share. Especially if you have friends who are new to podcasts and doesn't matter if they're on iPhone, Android or Windows Phone, the Brain Science app is available. The free mobile app is also the easiest way to get premium content. Finally, it is important to me that you subscribe to the show either in Apple Podcasts, Google Podcasts, Stitcher, Spotify, or whatever your favorite podcasting app happens to be. Reviews are also greatly appreciated. You can also like the show on Facebook. Speaking of Facebook, the next Brain Science Live will be on November 1st at 8:00pm Central Time. This will be posted on our Facebook Fan page, and to be honest, I'm really not sure exactly how it's going to turn out because this is during the time I'm going to be in Boston. So if you're a fan of improvising, you might want to tune in. Until then, I look forward to your emails. Brainsciencepodcastmail.com and I will be back next month with a new episode. It's scheduled to come out on the Friday after Thanksgiving in the United States. Thanks again for listening. Brain Science with Dr. Ginger Campbell is copyright 2018 to Virginia Campbell, MD. You can copy this show to share it with others, but for any other uses or derivatives, please contact me@brainsciencepodcastmail.com.