Corey and Steve talk with MSU Neuroscientist A.J. Robison about why females may be more likely to suffer from depression than males.
- 01:18 – Link between antidepressants, neurogenesis and reducing risk of depression
- 13:54 – Nature of Mouse models
- 23:19 – How you tell whether a mouse exhibits depressive symptoms
- 32:36 – Liz Williams’ serendipitous finding and the issue of biological sex
- 45:47 – A.J.’s research plans for circuit specific gene editing in the mouse brain and a start up’s plan to use it to tackle human cardiovascular disease
- 59:07 – Psychological and Neurological Effects of Ketamine. Testosterone and Deep Brain Stimulation
- Androgen-dependent excitability of mouse ventral hippocampal afferents to nucleus accumbens underlies sex-specific susceptibility to stress.
- Neurogenesis and The Effect of Antidepressants
- Integrating Interleukin-6 into depression diagnosis and treatment
- Sub-chronic variable stress induces sex-specific effects on glutamatergic synapses in the nucleus accumbens.
- Prefrontal cortical circuit for depression- and anxiety-related behaviors mediated by cholecystokinin: role of ΔFosB.
- Emerging role of viral vectors for circuit-specific gene interrogation and manipulation in rodent brain.
What is Manifold?
Steve Hsu is Professor of Theoretical Physics and Computational Mathematics, Science, and Engineering at Michigan State University. Join him for wide-ranging conversations with leading writers, scientists, technologists, academics, entrepreneurs, investors, and more.
Steve: Thanks for joining us, I’m Steve Hsu.
Corey: And I’m Corey Washington. And we’re your hosts for Manifold. Our guest today is A.J. Robison. A.J. is an associate professor of physiology at Michigan State University specializing in the use of mouse models to study the role of changes in gene expression and protein function in the mesolimbic dopamine system, also known as the reward system and hippocampus. A.J. Studies behaviors as varied as the comorbidity of opioid abuse with PTSD and special learning in memory. His methods include the use of genetically modified mice, viral expression vectors, biochemistry, electrophysiology and complex microscopy. His studies range in scale from gene expression in single cells to macro molecular complexes wholesales, all the way up to whole animal behavior. Today we’re talking about his recent article about sex differences in models of depression, androgen dependent excitability of mouse ventral hippocampal afferents to nucleus accumbens underlies sex specific susceptibility to stress. Welcome to Manifold A.J.
AJ: Thank you so much for having me.
Corey: So, major depression is a substantial problem in the US and worldwide. I recently was writing up a proposal, so I did a little bit of background research. Estimates suggest that around 16 million Americans suffer from major depressive disorder and according to the WHO, depression is the leading cause of disability worldwide. Among the US workforce, the prevalence of MDD has been estimated at 7.6% that’s in 2015 and to cost approximately $210 billion per year, a 21% increase from 2005. A depression diagnosis at age 15 is associated with a life expectancy reduction of 14 years in men and 10 years in women. Depression treatment can be improved, it could obviously have substantial public health impacts. I think we’ve learned a fair amount in the past couple of decades about the neuro biological basis of depression and perhaps you can give us a brief survey of what you think the basic findings are.
AJ: Well, I think you’re correct in saying that we have learned a lot over the past couple of decades, but there’s still a long way to go. What we’ve currently begun to understand are the parts of the brain which appear to function differently in depressed individuals compared to non-depressed individuals and how the connections between those parts of the brain may also be dysfunctional. And what we’re trying to understand better now are what the molecular mechanisms are underlying that the cellular and region specific dysfunction and how we can leverage those molecular mechanisms to come up with better pharmacological treatments for these diseases. So if you’re asking about what some of those mechanisms at the cellular and molecular level might be, I can maybe summarize a little bit of that.
Steve: Can I ask a question?
Steve: Are we at the point where if you make measurements on someone’s brain that you can tell they’re either in a depressed state or at elevated risk for depression just from those measurements?
AJ: We do know that having…. Well, I’ll give you an example of that. If your hippocampus is of a different size, smaller size or is asymmetrical, it does mean you’re at a higher risk for developing depression. Also other neuropsychiatric disorders like schizophrenia and some other diseases. So we do have information that can tell us things like that, but that would be a poor predictor, much better predictors are family history, early life adversity, genetic predictors are even in some ways useful. But for the most part a diagnosis of depression doesn’t come from looking at a person’s brain, but looking at their behavior and with a physician understanding their struggles.
Corey: This is something I hope to get to during our discussion because I probably know there’s a big fight happening at NIMH about how to understand psychiatric diseases between the kind of conventional psychiatrist approach of talking to people about their symptoms and experiences and classifying things and a desire to go more towards neurobiology and genetics.
Corey: And so I think the hope is eventually we can have a deeper understanding of these often rather vague frameworks if we could actually get at it genetically. But let’s not begin with that. Let’s begin with your issue, which is trying to unpack sex differences in depression. So, how does depression prevalence differ between men and women?
AJ: This is a really important and great question. So women are almost twice as likely as men to have a depressive episode in the course of their lifetime. And this is true across all cultures that we can measure and at multiple ages as well. And multiple income levels, education levels it remains true. So because the social differences in different cultures or different income levels or different education levels or different ages can differ greatly, we think it’s very likely that this difference in diagnosis of depression doesn’t stem simply from the societal differences between men and women or normal behavioral differences between men and women, but must stem from their physiology. That there are physiological differences between men and women that must underlie this disparity.
AJ: And what my group and many other groups all over the world are trying to understand are what are those physiological differences and are they things we could leverage for general treatment of depression, but even better for precision treatment of depression that depends on the person’s background, on the person’s sex and genetics and things like that. So to understand this, my lab uses preclinical models, uses rodent models and compares the function of the brain in male versus female animals and tries to pull apart differences in the function of the brain between the two sexes and determine whether they underlie behavioral differences between the sexes that can be linked to diseases like depression.
Steve: A quick question. I think you mentioned that having a small hippocampus or oddly shaped hippocampus makes you at higher risk for depression. Is there a gender or sex difference in the size and shape of hippocampi between men and women?
AJ: That’s a great question. I should be clearer that of course this is in proportion to the size of the entire brain and men are generally larger in most body parts than are women and so we have to normalize for this. But when that is normalized for, then yes it’s not that women have a smaller hippocampus than men that is probably not the driving factor. One good theory about the reason that a smaller hippocampus can be linked to some of these diseases is the theory of adult neurogenesis. So the dogma for many decades was that people produced new neurons from when they’re born to the post pubescent state and that that is involved in the growth of the brain, but also in the wiring and the structuring the hardware of the brain.
AJ: But that at a certain point, maybe in late adolescence or early adulthood, we stop producing new neurons and our neurons are senescent cells that do not divide. And so you’re stuck with the neurons you have for the rest of your life. There remains controversy about this, but most groups now believe that adult neurogenesis occurs in humans. It certainly occurs in most of the animal models we use. In humans there is good evidence that it does occur, although there is some evidence that it may not and it certainly occurs less in adults. But that having been said, lack of neurogenesis may be tied to a smaller hippocampus. The hippocampus is one of the places where adult neurogenesis is most prominent in animals and where we have the best evidence for it in humans. And the key finding in animal models very strongly and then perhaps also in humans is that antidepressants drive neurogenesis.
AJ: And if we prevent neurogenesis in animal models, antidepressants do not have their effects on the behaviors that we would link to depression in animals.
Steve: So Corey, don’t let me hijack your outline on this, but I just want to make sure I’m understanding what you’re saying. So it sounds like you’re saying that the shape or the size of the adult hippocampus can change in time. It sounds like you’re linking the size of it to neurogenesis that happens right in adulthood?
AJ: Maybe the evidence for that in humans again is very controversial and difficult. I can say for certain that in mice, the neurogenesis that happens during development and in early adulthood can be impaired and that can lead to a smaller hippocampus in adulthood. And that adult neurogenesis can also contribute to that size of the hippocampus. We have some evidence that that’s true in humans and that antidepressants can I don’t want to say correct, but can reverse a deficit in neurogenesis, both during development in animals and in adulthood in animals and perhaps in humans. And that that correction of the lack of neurogenesis may drive some of the antidepressant properties. Another critical news that it’s true cross multiple classes of antidepressants making the evidence stronger.
Steve: Right. And just to clarify an earlier point, so when you say an individual is at high risk because maybe they have a smaller than typical hippocampus, you’re normalizing it to the average for that sex, is that how you define big or small?
AJ: Sure. For that sex, for that size of that person’s brain is what really matters.
Steve: Okay, so relative to brain size.
Steve: But once you normalize to brain size, is there a second normalization for sex or not?
AJ: I don’t know it, I think it depends on who’s doing the studies. But my understanding is that once you normalize for total brain size, there shouldn’t be a difference between the two sexes [crosstalk].
Steve: So that’s the question. So if you had a population of men and women that have the same total brain size, is the hippocampus relatively larger say in males or smaller?
AJ: I am not steeped in that literature, but I would say no.
Steve: Okay, thanks.
Corey: I just want to address a question readers or listeners may have at this point. The evidence I think is fairly strong that there’s a biological difference that may count for depression. I was recently reading a meta analysis, which you may have seen, which again found something like a two fold increase in depression diagnosis risk for men to women. And the difference was actually higher in countries with hire gender equity according to this study, which seems like it actually provides some evidence that it may not be social. It’s actually different across the lifespan. Greatest around adolescence lessened over time, but it seems that that evidence is fairly strong, whatever the basis might be.
Corey: You said anatomical evidence. Is there any clear physiological differences between… say physiological predictors of depression? People often think of depression as sort of lower levels of activity in the brain, but I’ve heard theories that actually may result from overactive regions of the brain, perhaps the amygdala.
AJ: It’s interesting that you bring that up. The paper that you mentioned that we published recently suggests that an increased activity in a particular brain circuit does indeed drive the depressants levels [crosstalk].
Corey: I was thinking prior to yours, right?
AJ: Sure. I just have to plug that as is fitting with your hypothesis that that having been said as far as physiological predictors based on brain activity, I don’t really know if there are good ones. I do know that there are many other physiological measurements that can be made that are predictive. There was a wonderful study by a friend of mine named Scott Russo. The first author was Georgia Hodes or Hodes so also a good friend and is now in Virginia. They showed that a particular interleukin which is a signaling molecule critical for the immune system that can be measured directly from blood correlates with susceptibility to depressive states in humans and in mice and in the model that they were using to study mice. And it’s the same model we happen to use in the paper that you mentioned of mine. So I think there are biomarkers that can be used as predictive tools and potentially as helpful diagnostic tools that might be a lot easier to use than the brain itself which is a little less accessible.
Corey: It seemed like these would be extremely useful because you have to intro for a lot of things, if you’re talking about depression diagnosis. We know for example, just coming back to the initial question, women are more likely to seek psychiatric care than men are and so you need to control for that fact and trying to estimate prevalence. And it seems like if you have a biomarker you can do that by taking a random sample of the population whereas you can’t do that if you’re having an observational study where people are coming in with symptoms.
AJ: No, I think you’re exactly right. That’s, very much the case. I think the one issue with using a biomarker like interleukin six is that we’re not entirely certain that changes in interleukin six aren’t sex differential, aren’t sexually dimorphic as well. So trying to relate that directly to sex differences in depression isn’t something that I know has been done yet but could certainly be useful.
Corey: So I’d like to talk a little bit about mouse models because as you know, it’s been controversial over the past couple of decades. Many people have questioned the validity of them, how well they generalize to people. And you’re using a mouse model of a psychiatric disorder, which seems like it may be a little more controversial because of course you can talk to a person and ask them how they feel, you can ask them whether they lack desire for things, whether they feel unmotivated, very hard to do in a mouse.
AJ: We can ask them, but they don’t answer.
Corey: Okay, it’s reasonable. But I want to say that your paper starts with a pretty shocking observation about research in mice and let me read it to you. “Effective disorders such as major depression disproportionally affect women, but studies investigating depression related behaviors animal models that include both female and male subjects are lacking. Their sex differences in brain regions that regulate reward and motivation following some stress paradigms including sub quantum variable stress, chronic mild stress, social defeat stress.” I’ll use an alliteration. Unfortunately, most circuit specific animal models… animal model studies investigating depression related to behaviors have not included female subjects.
AJ: That is very much a true statement.
Corey: That’s kind of surprising. You have a phenomenon which is more common in female humans than male humans, yet the models you’re using are male mice, why would that be?
AJ: So as a field for many decades, more than 90% and I think I’m being generous going as low as 90% of the studies using mice as a model or using rats as a model or many others used exclusively males. And the reasoning behind this, which I think most researchers today would agree is faulty reasoning, was that the estrous cycle in female animals is more difficult to account for and we’re not looking for differences between the sexes, so let’s just eliminate that variable and use male animals for everything.
AJ: And that was standard practice for many decades. It has since been realized and it has been mandated at the level of the NIH that the use of female and male animals in studies of this nature and of virtually all nature that the NIH will fund is beneficial to the field, is beneficial to our understanding of normal baseline physiology and health and is also critical to our understanding of how diseases work and how we can create better treatments for the diseases that we’re attempting to model using these animals.
Steve: I don’t know Corey, I mean the jump from mouse to human seems much bigger and more uncertainty inducing than just the male female issue because male humans do get depressed as well. Right?
Corey: Yeah, but also looks like depression in female humans is linked to hormonal changes, puberty, postpartum, et cetera, et cetera. So it’s a phenomenon. It’s face that looks like it’s hormonal and to focus on males. Look, they’re both issues, right?
Steve: It doesn’t invalid… So if you said I happened to mainly be studying a model for male human depression and I’ll use male mice and that’s my problem. You can study female mice as a model for female human depression. That’s a sort of different things, but it doesn’t invalidate them.
Corey: No actually not but I think people were not making the kind of qualification that you are making there.
Corey: I think in these studies or in any other studies, for the most part, something people have learned.
Steve: I just want to know how can you tell when the mouse is depressed?
AJ: So I think the first thing I would say is that-
Corey: It was the next question.
AJ: Is that the national Institute of mental health would frown very strongly upon me ever saying that a mouse is depressed.
Steve: Oh, okay.
AJ: Our mice are used as a model for the study of depression, that is the language that the NIMH prefers and that’s the language that we use. So we do not believe we have depressed mice, we believe that we have mice that we can use genetic pharmacological or behavioral tools to manipulate them in such a way that they exhibit some behaviors that are analogous to some of the aspects of some diseases. That’s a really long winded way of saying we can study specific things about what might be happening in depression and we can model those things in mice, but I don’t believe that my sorry model of depression. They are a model for us to study some things about depression.
Steve: So what are those things?
AJ: So depression in people is characterized by a constellation of symptoms, but I’ll just name a couple which include anhedonia, which is no longer feeling pleasure in response to stimuli that used to be pleasurable. For instance, you no longer enjoy listening to the music you used to enjoy or you don’t feel like going out and spending time at the job that you used to like doing because of the depression. That’s anhedonia.
AJ: Social withdrawal. So a lack of enjoyment in interaction with other people, changes in weight, body weight. Lots of these things happen in people who have depression in the mice-
Corey: Up and down?
AJ: In body weight yes, that’s correct. With social interaction it’s almost always down and with feelings of pleasure or pursuit of reward, it’s almost always down. And so with the mice that we study, we can induce behaviors that mimic that in these mice. So for instance, the study that we were talking about earlier uses a task called sucrose preference where a normal mouse if given the choice between two bottles, one of which is filled with water and the other is filled with water that has some sucrose in it, a normal mouse will drink almost exclusively the sucrose water, it tastes better. My kids would do the same thing, but a mouse in which we have made these manipulations will prefer the two water bottles about the same, will no longer seem to be finding pleasure. Now we’re anthropomorphizing that mouse’s behavior if we’re saying that that’s a lack of pleasure or that’s depression, but it’s certainly anhedonia.
AJ: The mice will also interact less with other mice in controlled situations and so we say that this is a model of social withdrawal and also the mice, their body weight is decreased. So in this case we’re mimicking one possible aspect of depression and a physiological change. There are also many others. We can look at behavioral despair in some other tasks, things like that. So we can model very specific aspects of depression that are present in humans and are present in a similar way in our mice. But of course that doesn’t mean that the disease is happening in the mice or that it’s identical to what’s happening in the humans.
Corey: You induce, I’m not sure you even call it right now, you induce depression like behavior in your mice.
AJ: You’re doing great Corey that’s what the NIMH wants you to say.
Steve: Or you can say depression associated behavior.
AJ: That one works too.
Corey: My legal mind is beginning to operate here. It seems what’s interesting in the back of my head is running this whole narrative that NIMH has designed language set so that you can’t actually jump on them when these studies don’t generalize to people. Cause if you’re just studying kind of depression associated behaviors or depression like behaviors, there’s actually no direct implication for human beings. And thus when someone points out, “Hey, this study didn’t generalize to people,” they can say, “Hey yeah we weren’t actually studying people at all we’re studying depression like behavior in mice.” Anyway, that’s a little side.
AJ: I also don’t think you’re wrong. I think that might be a defense mechanism that we’re using as a field. So that’s not a bad thought. That having been said, I think the reasoning behind changing the language wasn’t so much to cover our behinds as it was to be more exact with what we were trying to say. So that when different studies apparently contradict each other, it’s not because there’s actually one study was wrong and one study was right, it’s more because the models they were using were different and the behaviors they were measuring were different and that explains the differences between the studies.
Steve: Yeah, I actually don’t want to view it as CYA behavior. It just seems like you’re being very clear about what’s going on. So at base it seems like you are able to study behaviors in mice and some perhaps causes of changes in behavior in mice and that may be of interest all by itself.
AJ: I think that’s true too. Being entirely honest with you guy. My personal interest is in how the neurons work. That’s what makes me come to work everyday and what gets me excited. And when we find differences in how neurons work and we understand the molecular and cellular level, how that drives behavior in an animal that to me is in and of itself exciting. The fact that some of that may have implications with regards to human health and medicine to me is very interesting and exciting as a separate idea. But even if it’s not true, the studies we’re doing themselves are intrinsically interesting to me.
Corey: I always had that sense too about many animals. I was very interested in ants for a long time and I think they’re just facinating organisms. But the research is hard to fund because people always want to see the implications. People meaning funders, the public, government wants to see implications for people. But let’s drive into your studies. How do you induce depression like behaviors in the study that we’re talking about?
AJ: So in this particular study we used a protocol called subchronic variable stress or SCVS. Subchronic means just a little bit and variable means it’s different each time and stress means we’re doing something that the mice don’t like. So the stressors we use are a mild shock to the foot, which makes them uncomfortable but doesn’t cause pain or injury. A confinement to a small space, which they don’t like when they’re not choosing it. Of course if you let a mouse choose, it will hide in a small space. But if you confine it there with light it doesn’t like that idea. And then also being held by the tail, they don’t like that either.
AJ: So over the course of six days, the mice undergo each of those different stressors in an order that they can’t predict. And at the end of that six days, the female mice have some of the behavioral differences that we were talking about earlier. The one we focus on in this particular paper is anhedonia. So they no longer prefer to drink sucrose water as much over non sucrose water. So they seem to have this lack of pleasure in their normal behaviors and the male mice do not have this. So this is not something we invented. Others use this paradigm before us, but we see the same thing that has been shown by others.
Corey: Well do not have… Let’s be more specific about that. You find a significant difference in the females after the intervention than before. You do not find a significant difference in males. But what’s the effect size? And as we know, we don’t want to focus exclusively on P values, right?
AJ: Sure. The effect size, so the typical mice and I can’t remember exactly what the numbers are in this paper, but the typical-
AJ: And I can’t remember exactly what the numbers are in this paper, but the typical mice will have something around an 85% to 90% sucrose preference, and the female mice that are affected here drop down to that 60%, 65% sucrose preference. Which is quite big when you think about what that means about their behavior, how much they’re actually consuming compared to in the two different conditions.
Steve: What fraction are actually affected?
AJ: Of the mice?
AJ: We don’t divide the mice into affected and non-affected the way you might with a social defeat experiment where we have resilient and susceptible animals. So this was an overall group effect. Maybe another key point that I should have made was that we don’t measure sucrose preference before the stressor and then measure it again after in the same mouse, which might get more at your question and that’s something that we could try. But in this case, the overall group of female mice that underwent the stress had a significantly lower sucrose preference than the female mice that did not undergo the stress. The male mice did not have that effect, and a two-way ANOVA shows that that’s-
Corey: So what percentage of male mice actually have decreased sucrose preference after the…
AJ: Again, that’s not easy to determine because what we’re looking at is a group-
Corey: So take the groups, right.
Steve: I think there may be measuring an aggregate consumption rate.
AJ: That’s a much better way to say it.
Corey: What’s the aggregate consumption rate for male mice after the intervention?
AJ: The same as the male mice that did not have the intervention.
Corey: Exactly the same or non statistically significant difference?
AJ: No statistically significant difference. And if you look at the paper, their not even… They don’t trend to be lower by more than a couple percent of that at all, they might even be higher. But I don’t remember, I’d have to look at the graph.
Steve: And how big are these groups? How many mice are involved in the experiment?
AJ: I believe in most of these we’re looking at two cohorts of 16 mice per cohort. So the group size is pretty big. It depends on the experiment, some of them were two cohorts of 12, some are two cohorts of 16. But we always do each experiment twice, multiple cohorts and we make sure that we see the effect in both cohorts to ensure that it’s replicable.
Steve: It sounds like to me just intuitively it’s a very small N.
AJ: With regards to mouse behavior, that’s a pretty good sized experiment, especially when you consider that we have four groups overall, we’re talking about a lot of mice. And then we haven’t talked about any manipulations yet. That’s just a control model. Now we have to do all the circuit manipulations that we did, all the testosterone manipulations that we did. This study used many hundreds of mice to do all of that.
Steve: So I’m just curious for result, let’s take your results specifically if for example, it became very popular, this particular experimental design and by five years from now a hundred other groups had repeated same experiment with each of them using 10 times as many mice. What do you think is a probability that a meta analysis of that work would not show your effect or show your effect with significantly decreased effect size?
AJ: I think we could already do it. We weren’t the first people to use this protocol. This was modeled after Georgia hoses work, but Debbie Bangasser has done it, other people in Scott Russo’s lab have done it and I’m leaving out 10 other labs. So that meta analysis could already be done. Now the problem with that as an idea though, to me, is that when people don’t get this behavioral paradigm to work or other behavioral paradigms, they don’t publish it. So if we just look at the papers, I don’t think that’s going to be an accurate representation of actually doing the experiment that many times.
Steve: let me rephrase. So under ideal circumstances, if this became the number one priority of the EU to test your result, what do you think are the chances that it would fail to replicate?
AJ: Just the fact that SCVS has some specific effect on, on sucrose consumption in female mice and not in males. I think the chances are incredibly low.
Steve: So you’re 99.9% confident…You would?
AJ: Okay. I’d say 99%, sure. I’ve seen enough replication of that that I feel pretty confident in it and we’ve done enough times in my lab. The problem with working with mice is that my mice aren’t the same as your mice. And my Institute isn’t the same as well. In this case, you and I are at the same Institute, but for many of the people doing these experiments were not the same place. The environment the mice are in isn’t the same. The vendors we’re getting them from isn’t the same, the food they’re getting, the people who are giving them the food and what perfume they wear, we could go on and on. So there are other things that are huge factors in stuff like this. But you can say that about almost any kind of experiment you want to do.
Steve: Well, I’m a physicist, so I would say there are different kinds of experiments without those kinds of uncertainties.
Corey: And not to make a joke here, but you know, we know that the gender of the experimenter has an effect on mouse behavior from the last couple of years, electrons have no gender. And so there’s various degrees of variability that just are not present.
AJ: So all electrons behave the same way?
Steve: Yeah, they do actually, they all seem to be exactly identical to the others. I’m just trying to get a sense of your calibration of… because obviously there are a lot, as you just pointed out, there are huge uncertainties, right, in dealing with mice. And the EU lab that does this, they might be using a different breed of mouse slightly or different circumstances with the experimenters. So I’m just trying to calibrate, given the numbers and the size of the experiment that you’ve done, how confident are you that the effect that you’ve discovered is real? That’s all I’m trying to understand.
Corey: He didn’t discover this effect.
AJ: Well, I didn’t discover this effect.
Corey: He’s using this as a baseline for his further studies.
AJ: This is a model adopted that many other labs have been using for a long time.
Steve: Oh I see.
Corey: Okay. But the men analysis you described, Steve could easily be done even in spite of the foul drop program because people-
Steve: But it would be a meta analysis of published results.
Corey: But no, could also do some great literature. These studies, they contact everyone who’s ever been heard to run a study. They get stuff that often isn’t published and they put that into the meta analysis. So meta analysis often include unpublished literature, at least in the best run studies.
Corey: So let’s hop in. You begin to dig into this particular finding, physiologically and through manipulations of testosterone. So the obvious question is, your hypothesis is that it’s testosterone driven, the male resistance to the effect of the stress on sucrose preference.
AJ: What we found was that if we remove testosterone from males, they exhibit the same decrease in sucrose preference in response to these behavioral stresses that females exhibit. And if we give testosterone to females, they no longer exhibit those same responses. So our best explanation for those data is that testosterone in this particular type of experiment is driving the resilience to stress that prevents a decrease in sucrose consumption.
Corey: And whenever you have a finding like this, right, you of course begin to look back and see how plausible it is in light of previous findings. What did we know about the effect of testosterone on circuitry that you think is either consistent or inconsistent with your findings? Before you had you clearly done a lot of research on this before you…
AJ: Sure. Although the finding was actually serendipitous, the initial finding where this came from was observing that there was a circuit that was different between them, in excitability between the male and female animals and we didn’t expect to find that. If you want, maybe I’ll take one minute to describe how that happened. It’s kind of cool.
Steve: And maybe explain what you mean by circuit.
AJ: Sure. As I know you’re aware there are many different parts of the brain and they’re connected to each other. So a circuit in the context that we’re talking about it today is the connection between one part of the brain and the other. In this case, we’re talking about an excitatory connection, meaning when the cells fire here, they make the cells fire there, more likely to happen, right. As opposed to an inhibitory connection where they’d make them less likely to fire.
AJ: So in this case, we’re talking about an excitatory connection between that part of the brain that we mentioned earlier, the hippocampus in a specific part of it, the ventral bottom part of the hippocampus. And an area of the brain called the nucleus accumbens, which we’ve known for many decades, is involved in reward and has changes in these models that we study.
AJ: So we were studying that particular circuit in male mice. And one of my postdoctoral fellows was doing a lot of research on that circuit, on the excitability of that circuit doing other genetic manipulations. And we had a newer MD PhD student in the lab, her name was Liz Williams. And she was assisting Andrew Eagle, the postdoc on those studies and was starting to do some of her own. And she came to lab meeting one day with her recordings from this circuit. And unlike the postdocs recordings, hers were terribly variable. The data had a huge variance and I was disappointed when I saw them. And said that, “This doesn’t look right. I don’t know what you’re doing wrong in your recordings.”
AJ: And what we found out was that what she was doing wrong was that she was using both male and female mice, not wrong at all, and she was lumping all the data together. And when she separated the data, which was her idea, not mine and two male and females, the data were very tight. The variance was very low, but the males and the females were different. And that’s how we found out that this circuit differs and excitability in the male and female mice. Then we pursued this model because we knew this circuit-
Corey: So stop for a second because this is a classic example of science often works.
AJ: Yeah, serendipity.
Corey: You write things up kind of in reverse order to make it look like there’s a hypothesis that you’re investigating, blah, blah, blah, when in fact there’s an observation that puzzles you. I just want to have you kind of elaborate on this. It’s something that I think really drives science and it’s underappreciated. We tend to present science as it’s presented in articles and in the way you have to write them in grant proposals like a hypothesis. But we interviewed Stuart Firestein recently about his two books, How Ignorance Drives Science and How Failure Drives Science. By failure, he partly had in mind serendipity, these ideas that you think are going to go one way and then they fail in the sense that you get an unexpected result. And that’s, I think a primary driver science, unappreciated. But your result here is a classic example of that. I take it it’s probably not the first time this has happened to you.
AJ: No, but, I think it’s my favorite. I liked this finding very much, no pun intended, we were very excited when we found this difference in excitability and we didn’t plan on it. But what I’m most pleased about is that Liz Williams who has since graduated and is finishing med school, MD PhD students, she got her PhD finishing the MD and Claire Manning, who is the PhD student also just graduated started her postdoc at Stanford just two months ago. The two of them together, followed up on this idea and did the right kinds of experiments, had the right thoughts along. I’ll take some credit, I had a couple of ideas too, about how to turn this serendipitous discovery into a better project than the okay ideas we had before we made the serendipitous discovery. So I think you’re exactly right. The best science we did, we did because of that observation, not because of all the thinking and grant writing I had done before that.
Corey: So how did you follow up this finding?
AJ: So when we first saw this difference in excitability in this circuit between male and female animals, our thought was a study that had been done by Rosemary Baggott, who was a postdoc with Eric Nestler. And the study had come out a couple of years before and showed that increased excitability in this particular circuit drove a social withdrawal behavior in male mice, she only used males. So we already knew that this circuit could be linked to behaviors in mice that were used to model depression. But no one had studied this particular circuit in females before. So we also knew that the males and females in humans differ in depression diagnosis and that male and female mice could differ in these models of depression in things like sub chronic variable stress.
AJ: So our first thought was that one of the things that might underlie these behavioral differences in the male and female mice might be this newly discovered difference in the circuit. So the way we wanted to follow that up was one, determine whether that circuit was actually driving the behavioral difference and two determine what aspect of sex was driving this difference in the circuit. When I say aspect of sex, I mean is it a hormonal difference? Is it a genetic difference having two X chromosomes versus having an X and a Y, is it something else? And so those were our two main questions and we did experiments to, to answer both of them and I think we can some pretty good answers.
Corey: So when you went to characterize the circuit, I don’t want to get too far in the weeds here. What techniques did you do to try to understand? We talked generally about the circuit, but tell us a little bit about what you did to try to, in a sense, identify the particular connections and see how they differed?
AJ: So we use a tool, we use two tools to identify the circuit visually. The mice that we use are genetically modified such that they express a fluorescent protein in any cell in which we make a specific genetic manipulation that we can make with a viral vector. And we have these viral vectors we can inject into the mouse’s brain that wherever we inject them, any cell that projects to that area will now have the manipulation. So by combining these two tools, we could make that injection in the nucleus accumbens and all of the cells in the ventral hippocampus. That project to that place will be labeled bright green with this fluorescent protein. So we could see them that way. So then we were able to do our normal electrophysiological recordings from those specific cells that we knew were projecting to nucleus accumbens and observe that difference.
AJ: And then we could also use other tools to manipulate the excitability of those cells and then observe whether that changed the behavior. The tools we used to do that are called chemo genetic tools. The specific name is dreads. These special receptors that let us kind of tune up or tune down the excitability of the cells pharmacologically and specifically just that circuit. And when we tuned up the excitability of that circuit in male mice to make the circuit seem like a female circuit, the male mouse’s behavior became like a female mouse’s behavior. They had that reduced sucrose preference in response to stress. When we tuned down the excitability of the circuit in the female mouse, their behavior became like a male mouse, they when stressed no longer had this reduction in sucrose consumption. Please.
Steve: Just a technical question, for this fluorescent labeling, do you need to sacrifice the mouse in order to read it out?
AJ: Yes. So our electrophysiology that we use to look at the excitability of this circuit is a whole cell slice physiology. So yes, the animal has been sacrificed. We slice the brain to get the region we care about. In this case the ventral hippocampus and we make our recordings there.
Steve: Right, so the region is identified in one set of mice, but then once you know what that connection pattern architecture looks like, then in the other mice that you’re monitoring, you just monitor those areas. Is that correct?
AJ: Sure. We can identify the difference in the circuit by doing what we just mentioned, making the green fluorescent protein show up there doing our slice electrophysiology and seeing this difference. To do the manipulation where we changed the excitability of the circuit, we do multiple surgeries, we use an intersecting viral strategy. I know that you’re aware. And that is done on an intact animal, the animals not sacrificed. Then we can observe the behavior. Then post-hoc after the animal has finished the behavior, we can then sacrifice the animal and check our placement.
Steve: Check. Got it.
AJ: Because that’s really critical. Sometimes you miss when it’s four surgeries in a single mouse.
Corey: Can you do this in a behaving animal, actually alter the functioning of the circuit physiologically while the animal is acting?
AJ: Great question. The method that we used is a slower, longer-term method, this chemogenetics. It can be used while the animal’s behaving, but the time during which you start to increase the change in excitability takes 10 minutes for the drug to come on and then it washes out after an hour or two hours. So your temporal control is not very specific. However, it’s not very invasive. We just give the animal the drug and then do the behavior.
AJ: The method that I think maybe you’re alluding to that would give us precise, very specific millisecond temporal control in the excitability of the circuit is optogenetics. The advantage of this is that the temporal control would let us change the excitability of the circuit while the animal is behaving and watch the change in behavior in real time.
AJ: The disadvantage of this is that the mouse has to have a fiber optic cable implanted into its head through its skull to shine that light on that circuit to change the activity in that millisecond way. But then its behavior is of course a little bit altered by the fact that it has a big fiber optic cable sticking out of the top of its head.
AJ: And that kind of thing is absolutely done and it’s a very exciting and viable technique and we would like to use it. Our friend Alex Johnson here at Michigan State is using it now and we are working with him to do so as well. But that wasn’t the approach we wanted to take here. We wanted a less invasive approach and we weren’t as concerned with the millisecond timescale in these initial experiments.
Steve: So, just to summarize my understanding, you have enough understanding here that you feel you can actually make interventions which alter behavior in the mice. And interestingly the interventions point to actual sex differences in behavior of mice.
AJ: I think that’s a true statement, in the context of our study, yes.
Steve: Are people in your field sensitive about revealing differences between male and female mouse brains? Is that an issue of sensitivity? Because if I just replaced the word mouse with human, then I think people would be very sensitive about it.
AJ: Yeah. I think what people talk about in our field is biological sex as a very specific variable. With humans this is a much more nebulous issue. It might be nebulous with mice too we just can’t observe things about gender and gender fluidity and things like that in mice.
Steve: But you can determine biological sex in humans too, right-
AJ: That’s true.
Steve: … in terms of just defining it by chromosome?
AJ: From a genetic standpoint, we absolutely can. And some statements can be made about the biological sex of humans and what that means about their physiology. These are very obvious things like if you have two X chromosomes, then you may be able to get pregnant. If you have an X chromosome and a Y of chromosome-
Steve: No chance.
AJ: … it is very unlikely that you’d be able to get pregnant. And I don’t think anyone would find that statement touchy or problematic. That’s just physiology.
Steve: Right. I could imagine statements that you might make about differences in architectures of brains or circuit structures in brains between men and women that people would be quite sensitive about. And all I’m asking is whether that sensitivity has leaked into the mouse world at all?
AJ: I think that most people are sensitive to the issue of sex and gender and how that can affect how others feel. But when we’re trying to describe the results of our experiments in a scientific journal, I think we’d simply try to use the most precise language possible and allow the reader to interpret it as they choose. I think that’s the best answer I can give.
Steve: No, that’s fair.
Corey: So I think we’ve seen a little bit about where you may be going with this research in the future. You’re considering optogenetic studies to look at a behaving animal.
AJ: That’s possible.
Corey: Are there other ways in which you plan to follow up on these findings?
AJ: Yeah. Again, I hate to keep using the word excited because it seems like such a pun in this context, but we are really excited because we’re doing some really neat things right now. So we have a very good friend who’s moved into the new ISTB building with us or is in the process of moving there now. Aritro Sen, who is part of the development group joining us and our neuroscience group and Anna Moore’s precision medicine group there. And Dr. Sen is an expert in molecular manipulation of the androgen receptor and what we showed in this study-
Corey: Can we have a definition of the androgen receptor?
AJ: Let me do that right now. What we showed in this study was that testosterone appeared to be driving some of these effects, but we didn’t show in this study for sure that the mechanism by which testosterone was driving these effects was by binding to its specific receptor on these specific cells that we’ve been measuring. That receptor is typically the androgen receptor, that’s the testosterone receptor, although it does other things and testosterone does other things.
AJ: What we want to know is whether the way testosterone is changing the behavior of these mice and the excitability of this circuit is through binding to androgen receptors on these very cells. And the tools that Dr. Sen has provided for us already let us knock out the androgen receptor only in these cells of this circuit and not in others. And then we can perform similar experiments like giving testosterone to females or taking testosterone away from males or stressing the animals. And determine whether the testosterone is acting through the receptor on the cells to drive these behaviors.
AJ: The reason for doing that is if it is, then we have a lot of downstream effects that we know that the androgen receptor performs in the cell that are druggable. The actual pathways downstream of the androgen receptor in the cell cytosol, meaning the body of the cell and at the genetic level because the androgen receptor binds to DNA and directly modulate gene expression. We can have gene targets and cellular protein targets that are druggable that we can then test whether those targets might be potential areas to leverage to invent new drugs for specific treatment of depression or other diseases.
Steve: Can you just elaborate on the method for how you actually knock out the androgen function in a particular group of cells?
AJ: Terrific question. So the tool that Dr. Sen has provided for us already is what’s called a fluxed androgen receptor mouse. What this means is that the mouse is completely normal, its behavior and its physiology and its cells are just like any other mouse. But in its genetic code it has a couple of small engineered mutations within the androgen receptor gene and those mutations are benign, they do nothing to the expression of the androgen receptor in a normal mouse. But when we introduce a very specific protein called Cre recombinase, that protein then goes to those engineered sequences and cuts out a piece of the gene defined by those sequences and basically makes the gene nonfunctional. And Dr. Sen and his prior group before he came to MSU, made this mouse and have and have provided this mouse for us.
AJ: Now, what that would normally let us do is express this Cre recombinase in a certain cell type or a certain region of the body and study the effects of this receptor and those regions that are in that cell type. But here we’re doing a circuit specific manipulation. This again involves injecting two viral vectors into the brain that only when they are combined, so we can inject into the target region and the location where the cell body is. And only when the cell gets both viral vectors in the same place, same cell do we introduce Cre and knock out the gene?.So that’s what allows us to do circuit specific knockdown of the androgen interceptor in a living mouse and then study its behavior.
Steve: So you’re at the point where you can really knock genes out in specific cells that you want.
AJ: This is one method to do that. If you don’t mind I’ll tell you one other method. I have an RO1 from a National Institute of Mental Health to do crisper editing in a circuit specific manner. And our first paper showing this exact technique working is now under a revision for Nature Communications.
AJ: And what we were able to show is that we can put a retrograde virus into one brain area, and a local virus in the other brain area. And they have two components, the CRISPR-Cas9 enzyme and the guide RNA. And only in cells that give both of those components do we edit the genome in a very specific site specific way.
AJ: And we picked a gene that I’ve been studying for many years called FOSB. We edited that gene in the specific circuit. And we got a specific behavioral result and a specific molecular result. And we edited in a different circuit and we get a different behavioral result. And so we think we have circuit specific gene editing using this technology as well.
Steve: So you’re able to test the functional impact of that. But are you able to directly measure the accuracy of the localization of your edit?
AJ: In the animal’s brain? No. That would be really difficult. What we can do is show that that protein, the protein expressed by that gene, is knocked down in that circuit and only that circuit. But that’s a very indirect measure, as I’m sure you’re thinking. In cell culture, we can do what you’re suggesting, and have done so.
Steve: So you have a sense of the spatial specificity of the technique.
AJ: That’s correct. Initially, we worked with Rachael Neve who was at MIT initially, and now is at Harvard Mass General, to develop these tools. And then her husband and collaborator, Rick Boyce, helped us to do the kind of checking that you’re talking about, the kind of appropriate controls, to show the specificity of these tools.
AJ: And we did that in cell culture with him. And that’s a process, using cell culture, allows us to do these manipulations the same way, in many, many cells at the same time. And then make the kind of measurements you’re suggesting. Something that we don’t yet have the technology to do in a circuit.
Steve: Got it. It’s really amazing stuff. It’s incredible technology, actually.
AJ: Oh, I’m so glad you’re excited about it. We are too. We think the power of this is that you can use this viral technology to edit any gene you like, if you have the right guide RNA, in any circuit you like. If you do the surgery you want, in a wild type mouse, you don’t need a genetic tool. So any mouse, or any other organism, if you have the right guide RNs, we can use this technology in.
Steve: There’s a researcher in the area of computational genomics that I work in, which we look at a lot of data, and we build genomic predictors for a elevated disease risk for certain conditions, like for example, coronary artery disease. So we now know what, say, the top 100 or top 1,000 genetic variants are, that elevate risk for coronary artery disease.
Steve: And there’s actually a very well funded startup, which is actually planning to use CRISPR to actually directly edit the cells in people’s hearts, to actually reduce the risk of coronary artery disease. So when I heard about this person who’s in my area was actually heading off into this more direct kind of application of the knowledge, I was amazed. I said, “Wow, what a risky thing to be trying to do.” But now that you’ve told me that you’re doing this in a tiny little mouse brain, it seems like, “Wow, if you can do that, then the human heart is a much bigger target to go after.”
Corey: Hold on. Do mice clearly have analogs of CVD, cardiovascular disease, coronary artery disease? I assume these things. Whatever startup your friend is running, they must be based on the mouse studies of some kind.
Steve: No, no, no. Everything I’m talking about is in humans. So all the data analysis and the variant-
Corey: But the physiology, why would he … He would have to have some physiological theory of what’s going to happen when you start editing these genes in somebody’s heart. Clearly, they’re not going to allow … FDA is not going to allow you to.
Steve: So, for sure, there’s a question of causality. These variants that we identified, is it a causal effect? Or are they just tagging something else which is causing the elevated risk? And I’m not going to make a statement about that. But these guys are aggressively pursuing and have filed a lot of patents around the idea that eventually they’re going to know what to CRISPR. And they’re going to do localized CRISPR-ing of your heart. And then, after that, you’ll have zero risk, or much-
Corey: Yeah. I get that. But for that to work, you have to have very convincing models of this phenomena, or they’ll never get approved. So are they jumping the patent line? Are they filing patents for something where there are no clear models for it already?
Steve: That’s a good question. So that question of what FDA would require, before they allowed you to do something like this, I’m not an expert on that. And you’re right. Maybe there’ll be many, many years of tears, trying to do things in mouse models, before you-
Steve: Yeah, decades.
AJ: I would point out, there are many mouse models of cardiac myopathy, and things like that that can be used. I worked on one as a grad student. But they’re not CRISPR based. But CRISPR could certainly be applied to them.
Steve: Yeah. So, that step, I don’t know about the regulatory process.
Corey: It’s just the research process, right? They’re advocating something in people. Presumably, you have to do … All the questions you had about AJ’s work, jumping from mouse to people, of course apply here too. So it’s an exciting idea, but it just seems like-
Steve: Yeah, in some ways, I acknowledge all your points. I want to reemphasize that when I heard that he was heading into that direction, I was amazed. Because it seemed like such a leap from just the computational side of things, which is mapping out the genetic architecture, to actually trying to do interventions on people.
Steve: The one thing I just wanted to say, in reaction to our guest’s description of his work, is it makes me feel like the actual ability to do the editing in a very localized way seems less science fiction-y. It seems less speculative now than before. This question of establishing causality, and getting the FDA to approve it, I don’t know. I don’t have any view about how hard that will be. It sounds very hard, I agree with you.
AJ: But, if it works, cardiac disease might just be one of many, many, many things that could be affected by this kind of technology.
Steve: Yeah, I mean, it-
AJ: I certainly love the idea.
Steve: Interestingly, Corey, if I move you from top 1% risk on the predictor to bottom 1% risk, I could be changing by a 100 X.
AJ: Sure. No, of course. Of course.
Steve: Your risk. And if I could actually do that actively to somebody, that’s pretty amazing, right?
Corey: It is pretty amazing. This is an entirely separate debate, as you know. It’s a real question as to when you want to move the needle on cardiovascular disease, you focus on genetics or just basic public health, like get some exercise, eat reasonably. I’m not saying they’re separate, right? But I’m saying as far as investment goes.
Steve: I think the reason these guys were able to raise over a $100 million right away to get this funded is the idea that this is heavily patentable. If they get it working, they’ll be the ones who control the intellectual property. And everyone in the world who’s high risk for heart disease will want this.
Corey: Is the patent picture that clear yet? Because I thought there’s a lot of stuff [crosstalk].
Steve: No, nothing is clear. But they’re betting on the ability to get to control that intellectual property.
Corey: How much is Editas raised? Because if these guys are raising that much, people really must think that you’re going their patents on it.
Steve: Yeah. Editas has also raised just unbelievable amounts of money. Yeah.
Corey: We’ve diverted a little bit.
AJ: I think this is not an unrelated conversation. Because what you’re talking about is taking these basic molecular ideas and applying them to human health, in exactly the way I try to sell my research. Even if that’s not my personal long-term goal, it’s the reason why I’m able to do the research that I’m able to do. So I like this discussion.
Steve: Yeah. It’s tour de force stuff that you can literally change a circuit in a mouse, and you’re using these genetic tools to do it. And it’s very precise. And you’re talking about very small regions of the brain. And then you show that there’s actually a difference in behavior. It’s incredible stuff.
AJ: I will put you down as a potential reviewer for our next paper.
Steve: Yeah, sure. They’ll never let me review it. But, no, I just think that as a demonstration of what technical capabilities our civilization is gradually acquiring, I think it’s amazing.
AJ: Yeah, I think you’re exactly right. And I’m sure you are both familiar with this concept. And I’m preaching to the choir when I say the acceleration in those technical capabilities is asymptotic. We are really changing. Every two years, it’s a new world in our field. And trying to keep up with it is challenging, but also makes it fun every day.
Steve: Yeah. Coming back to this stuff that we spent a lot of time on in the initial discussion, this question of whether the model is actually relevant for human depression, it sort of to me becomes less important when I just realized, “Wow, you’re able to do this stuff in a mouse. You’re able to actually change the behavior of the mouse by active intervention.”
AJ: Steve, I think I mean this is a very much a compliment when I say I think you’re a techie. You like the cool stuff we can do. And I’m glad because we like it too. We also think that the biological questions that we’re answering are interesting. And we enjoy that part of it. But I also like the fact that we can just reach inside a mouse’s brain and change the genes.
Steve: To rephrase, I would just say this. Even if it is later determined that the mouse is really kind of an awful model for depression in humans, I would still be extremely impressed that you guys were able to do what you’ve done.
AJ: Right. And it’s not wasted time. Because the technology that we’re building and doing, this is technology that’s applicable in many other ways, that might not apply to mice at all.
Corey: Yep. So I want to ask a question that’s a little bit outside of your area of expertise. And just feel free to say it’s outside your area of expertise. You’re not interested, don’t know about it. But one of the most exciting developments recently in depression research has been discovery of ketamine as a viable treatment of treatment resistant depression. And it’s thought to operate through the glutamate receptor.
AJ: One of the glutamate receptors, yeah. There’s a bunch of them.
Corey: One of the glutamate receptors?
AJ: Yeah, one specific one, the NMDA receptor.
Corey: Exactly. Yeah. So I’m just curious as to whether you think is there any link between the NMDA receptor and the kind of sex differences you’ve observed?
AJ: Oh, boy. I hadn’t really thought about this very much, but it’s a really terrific question. Because the exact differences we’re observing are inexcitability. And so they could potentially be driven by a difference in glutamatergic input onto the cells.
AJ: Now, one thing I can say is that we did not observe a difference in glutamatergic input onto these cells between males and females. What we observe is a difference in the intrinsic excitability of the cell itself. So it’s getting the same amount of glutamate from its friends that are trying to excite it, but it responds more strongly in the female than the male. This female is more excitable.
AJ: So we don’t think that it’s driven by a change in NMDA receptors, that then make it more responsive to glutamate. It’s the membrane and the sodium and potassium, and calcium channels in the membrane, that are probably driving the effect we’re looking at. That doesn’t mean the other effects don’t exist. It simply means that our particular effect is not driven probably by the NMDA receptor.
AJ: That having been said, the NMDA receptor is super exciting as a target for depression treatment, for a variety of reasons. So ketamine works by antagonizing the NMDA receptor. And that, in the past, has been thought of as one of the mechanisms by which it is a hallucinogen, and a mechanism we can use to reduce consciousness. So as an anesthetic, it’s used as an anesthetic. That’s at high doses. At low doses, the modulation of the NMDA receptor is much less. And it might have other effects at other molecules. We don’t entirely know that yet. But it appears that its anti-depressant properties are mediated through the NMDA receptor.
AJ: So the question then becomes why is it affecting the NMDA receptor strongly? Has affects like hallucination and loss of consciousness. But affecting the NMDA receptor at a different dose, in a different way, can have these anti-depressant properties. And if we can figure out what’s different downstream of the NMDA receptor, how its signaling is different under these two conditions, that can lead us to really understand a lot more about the etiology of the disease, and other ways to treat it.
AJ: The last thing I’ll say is, and we’ve got a great new drug, especially for people who are treatment refractive for some of the traditional antidepressants. Ketamine’s working really beautifully for some of those people.
Corey: And I’m not quite sure what the history of ketamine as a treatment is, but it struck me, it must have been serendipitous. Somebody must’ve noticed over time, I’m guessing, right, that-
AJ: I don’t know the story either, but I’m pretty sure you’re correct.
Corey: Yes. I’ve been depressed for a long time. I go to the club, I take ketamine, I feel better.
AJ: Ah, that might be true. It might also be post-surgical. Because it is used as an anesthetic in surgeries in people. So I don’t know which one it was. It is an abused drug. So it could be the way that you described, but it could also have been medically.
Steve: I wanted to go back to the effect of testosterone on the system that we were discussing earlier.
Steve: So how much do we know about … It sounded like there were experiments where you’re maybe elevating the testosterone level in a female mouse up to the male level, and then looking for changes in the circuit that you’ve been studying. Is that fair?
AJ: That’s true. And we did see that that was able to reduce the excitability of the circuit, so that it looked like a male circuit.
Steve: And is the interpretation of that, that having an elevated testosterone level for a female mouse for some period of time, is actually altering the function of that brain circuit?
AJ: That’s great. So what you just said is exactly dead on, because you said for a certain period of time. We don’t think that the effects that we’re observing are acute. And I can say this from a couple specific experiments in that paper, where we used that chemogenetic method to tune up or tune down the excitability of the circuit. And when we did that briefly, during the behavior only, we didn’t see the change.
AJ: Only when we changed the excitability of the circuit for a couple of weeks, during the stress and leading up to the behavior, did we see the behavioral change. And the same was true of removing testosterone from the males, or adding testosterone to the females, required us a period of time that was weeks leading up to the behavior, to have the behavioral effect.
AJ: So we do think this is a long-term change in the excitability of the cell, that’s driven by testosterone, not an acute effect of the testosterone binding and changing excitability.
Steve: Interesting. Because this is jumping a little bit from the little furry guys to us.
AJ: Of course.
Steve: But there have been very interesting essays, and only a few that I’ve found over the years, but there are a few, by people who for some reason are taking testosterone. Either, for example, maybe they’re undergoing a sex change, or maybe they’re just taking it for some reason. And they’ve written about the psychological effects, or at least via introspection, what the psychological effects are of taking testosterone.
Steve: And I find it quite interesting. And it seems to conform to this kind of crude idea that it makes you more physically aggressive, more confident, etc, etc. But I think that those interpretations are controversial. And this is really just self-report. So I’m curious what is actually known in the case of mouse, for the actual real physiological, observable physiological effects in the brain from testosterone.
AJ: Yeah, that’s also a great question. Before we go to the mouse, I will kind of comment on what you just said about the human condition. To me, the studies that I find most reassuring about the potential physiological value of our observations in this paper, are that men who have reduced testosterone are more likely to have depressive episodes than men who have normal levels of testosterone.
AJ: So I think that that kind of supports this as an idea. Now how, for instance, anabolic steroids might play into this as an idea, or testosterone treatment in humans, that’s tougher. What I think you asked about was how testosterone changes the brain of the mice, and what we know about that. What I know about is the particular circuit that we talked about in this paper. But lots of other people have studied many other things.
AJ: And someone that you know very well, Cheryl Sisk here at Michigan State, much of her career has been spent on understanding the effects of sex hormones on the brain. And what she and many others have shown is that sex hormones program the development of the brain across the lifespan, in ways that are really important for the adult, in ways that make the brain of mice or rats, or some of Cheryl’s studies are in hamsters, in ways that make the mouse or rat or hamster brain different in male and female adults. Those differences go away, if we don’t allow sex hormones to have their effects throughout development.
AJ: So the effects we might be observing here could be due to the hormones acting during adulthood. And that’s what our experiment suggests. But the way the hormones affect the mouse’s brain don’t occur in isolation. They occur across the development of the organism, from when it’s born to when you observe the behavior. And so changes in those sex hormones can drive lots of physiological changes in the brain, that are linked to lots of behaviors of the animal.
Steve: I think in some of this self-report, these sort of autobiographical essays that I’ve read, I’ve actually seen descriptions of acute effects, where I took my weekly testosterone shot on Monday, and I felt totally different after that. But it could be psychosomatic. So I think, obviously, there are likely effects through development, but also possibly even acute effects.
AJ: I think we have pretty strong evidence that that’s the case, not just anecdotal, but in the clinic, that treatment of testosterone, or for instance with anabolic steroids, does change people’s mood, and does change their behavior.
Corey: Yeah. One last question. Oh, sorry.
AJ: Before we move on, I do want to make the one caveat that what our study suggests, and almost every other study I know of in this area, is that curing depression will not amount to giving testosterone to people. So let’s make sure that that’s very clear. Sorry. Go on, Corey.
Corey: I have a question about excitability. As you know, another treatment for fractory depression is deep brain stimulation. What’s the relationship between DBS and the kind of excitability phenomenon that you observed, if any?
AJ: So I see the connection you’re making, and it’s a smart one. Because deep brain stimulation does involve putting an electrode into a particular brain region, and injecting current that will make cells fire. And what we’re looking at in this particular paper is injecting current, and watching how cells fire, and increasing or decreasing that firing. So I can see that connection.
AJ: That having been said, the way that deep brain stimulation works isn’t simply to increase the activity of a brain region. It’s to change the activity of that brain region, in ways that aren’t necessarily well understood. Deep brain stimulation often uses excitation patterns, that are modeled on observed patterns from actual humans. And it’s that frequency that very much determines-
Corey: So the oscillatory [crosstalk].
AJ: That’s exactly right. And the frequency of stimulation that determines the effects of the stimulation on the person, not just for the treatment of depression, but for disorders of movement, or many-
Corey: Like dystonia, for example.
AJ: Exactly. And it’s particularly effective in some people with dystonia. And when they want to treat a person who has dystonia, they put the electrode into that particular part of the brain that they think is problematic. And they start firing the electrode. And they slowly change the frequency, until they find the one that best helps the person.
AJ: That’s very different from just increasing the excitability of a brain region. This is something that speaks to the complexity, and the intricate wiring and connectivity of all of these regions, and how little we really understand about all of that.
Corey: So I think, in the case of dystonia, you’re inserting the electrode into the basal ganglia, as far as I understand [crosstalk].
AJ: That can be true, yes.
Corey: Do you know, in the case of depression, are you putting it into the hippocampus, or the nucleus accumbens?
AJ: So you mentioned nucleus accumbens is one of the main sites. This is also called the ventral striatum. And this is actually the part of the brain where our projections from hippocampus we’re going to, in the paper we’re talking about. But this part of the brain, the nucleus accumbens, has been tied to mood and reward for at least 40 years or more. And so the fact that deep brain stimulation there can have effects on human mood is not surprising.
AJ: It fits with our limited understanding of neuroanatomy and neurophysiology. But it’s not the only place that can be stimulated to have effects on mood. And why it has antidepressant properties, rather than changing other aspects of behavior, such that it increases riskiness behavior, or whatever you would like to think about the nucleus accumbens is also involved in. Why it has this particular effect is not really well understood. And we need to know more, so that we can get better at having that effect.
Corey: This was another example where I guessed that it was just some accidental finding that led to a lot of research.
AJ: You could be right. Although, I don’t know how many times that people just randomly stick electrodes into other people’s brains to see what happens. So-
Steve: Accidental DBS? Is that what you-
Corey: Well, I think my guess is you probably may … This is a hypothesis about history, right? They tried it in mouse, they found some kind of phenotype. They tried it in people, for some kind of phenomenon, and found unexpected phenotype. But maybe we’ll just leave this as a historical question mark.
AJ: It may be true.
Steve: You might be able to Google it.
Corey: I think, yeah, maybe we will. A.J, thanks for coming by. This has been a great conversation.
AJ: What a pleasure. Thank you, Corey. Thanks, Steve. I really appreciate being here.