Amar Sahay, Ph.D, is an Assistant Professor of Psychiatry at Harvard Medical School, Massachusetts General Hospital, Center for Regenerative Medicine, and Principal faculty at the Harvard Stem Cell Institute. His primary research interests include how plasticity mechanisms in the adult brain affect cognition and mood and how perturbations in neural circuits may contribute to psychiatric disorders. Among other honors, Dr. Sahay has been the recipient of the Society for Neuroscience Career Development Award, the NIMH Pathway to Independence Award, and the Ellison New Scholar Award. StemBook Editor, Lisa Girard, was able to talk with Dr. Sahay recently about the exciting advances in understanding how changes in neural circuitry, specifically adult hippocampal neurogenesis, affects cognition and mood.
Could you tell me a little bit about some of the projects going on in your lab right now?
My lab is primarily interested in adult hippocampal neurogenesis, which is the process by which stem cells in the hippocampus generate new brain cells, or neurons, throughout life. As part of this endeavor, we’d like to understand what the functions are of these adult born neurons and how this unique form of plasticity that persists in the adult brain might be harnessed to improve cognition and modulate mood. This is relevant for the treatment of psychiatric disorders because the hippocampus is found to be compromised or impaired in multiple psychiatric disorders, as well as disorders of memory such as Alzheimer’s as well as age-related cognitive decline. So as part of this goal, we have several different projects in the lab that are aimed at understanding different facets of adult hippocampal neurogenesis and how this form of plasticity might be harnessed.
We are interested in the mechanisms that maintain stem cell homeostasis because ultimately, if we were to harness adult neurogenesis, we also want at the same time to maintain the population of stem cells. We are also interested in the properties and the connectivity of the new neurons and how they relate to the functions of the hippocampus. Our work, as well as the work of others, has found that when you genetically disrupt neurogenesis by blocking it, or you genetically up-regulate the process without affecting the rest of the brain, you can produce selective effects on specific mnemonic processes like pattern separation. However, these studies really haven’t touched upon what the properties of the neurons are that are essential for their functions. We think that this is important because by identifying genes that encode for these properties, you might be able to target these genes and the specific property of the neuron as it relates to the function, and improve the function of the circuit into which these neurons integrate. By also identifying the properties of new neurons that are important for function, we may be able to rejuvenate the older neurons in the circuit and that is just as important because adult born neurons comprise 10-20% of the total number of neurons in the dentate gyrus, which is the circuit in which neurogenesis occurs. So for therapeutic potential, we may want to target the older neurons by essentially restoring them to a younger state, or we may want to boost our endogenous capacity for making new neurons by targeting survival of adult-born neurons or proliferation of stem cells and progenitors.
The third question in the lab has to do with understanding, at a systems level, how changes in the level of neurogenesis in the hippocampus affect the regulation of mood. The expression of fear and the regulation of stress responses are thought to be mediated by parts of the brain like the prefrontal cortex, the amygdala, and the hypothalamus. The prefrontal cortex is important for executive decision-making and also is important for regulation of fear, in that it controls the amygdala. So, if you think adult neurogenesis is important for the regulation of mood, and if you hypothesize that, then it is important for us to understand how changes in levels of neurogenesis in the hippocampus affect communication with these other brain regions that are connected with the hippocampus. This is why I say we need to employ a systems level approach because we need to modulate the interactions between the hippocampus and the other parts of the brain in mice, which is the model system we use, in which levels of neurogenesis are changed. Towards this goal, we are employing optogenetic approaches to selectively activate or silence extrinsic connectivity of the hippocampus.
Finally, we are interested in understanding how risk factors such as stress, which is known to be a catalyst for the development of disorders such as depression and PTSD, affects adult hippocampal neurogenesis and the properties of these new neurons and their functions.
In the course of these studies have you found some factors and pathways that have been previously implicated in other contexts to be involved in these processes as well?
Yes. So, in understanding the mechanisms that underlie stem cell homeostasis there are multiple pathways and factors that are conserved across physiological systems, specifically stem cells in other systems. On the other hand, there are factors that we have identified and a particular transcription factor that we are interested in that may be specific to neural stem cells, so I think there is some degree of conservation of pathways. I mean you have your usual suspects, comprised of the Wnt signaling pathway, TGF-beta signaling pathway, the Notch signaling pathway, but there are also, as I said earlier, likely to be distinct determinants of stem cell homeostasis that are found in the brain and not in other parts of the body. So we have a lot to gain from reading the works of our collaborators and colleagues because of these shared mechanisms, and even if there are shared mechanisms, there are likely to be nuances, but there is a high degree of mechanistic conservation across systems.
When you were telling me about some of the projects in your lab you mentioned pattern separation. Can you talk a little bit about what pattern separation is and about your assays for measuring pattern separation?
There is a growing consensus in the field of adult neurogenesis that has emerged from our work and that of others for a role for new neurons in the adult brain in pattern separation. From one perspective, this is not a surprise because the dentate gyrus-CA3 circuit, which is a part of the hippocampus, has been long proposed to play a role in pattern separation based on computational, neuroanatomical, behavioral and genetic studies in rodents, and now brain imaging studies in humans. What is surprising, though, is that by just changing the number of neurons we can essentially modulate levels of pattern separation. So, you might ask why pattern separation is important. Pattern separation is critical for our formation of memories of what, when, and where, or episodic memories. Our ability to discriminate between similar contextual environments is essential for us to identify what is new and what is previously experienced. And this is what enables us to determine if we are sitting in this office versus someone else’s office two days before or distinguish between your birthday last year and your birthday five years prior. But in addition to this implication for episodic memories, we recently formulated the hypotheses that pattern separation may also be important for the regulation of mood. This hypothesis stemmed from the idea that the way we respond to our environment is intimately dependent on the way we perceive it. An individual, for example, who has returned from combat and has experienced trauma may walk into an office building and may experience symptoms of PTSD like heightened fear, hyper vigilance, increased glucocorticoid levels, skin conductance, high blood pressure, and heart rate, because this individual thinks that the room that he is in in this building is similar to the room in which he may have experienced something terrible, such as torture. In the past, people have emphasized that this overgeneralization of fear in PTSD might be due to just an eroding of the fidelity of memory with the passage of time. We instead have recently emphasized that this may be due to impairment in pattern separation. In other words, the individual now essentially makes the mistake of thinking that he is actually in the room where he experienced the trauma just because it has some semblance to that room. We think this is an impairment of pattern separation and may be due to excessive pattern completion, which is the opposing mechanism by which our brain retrieves formed memories based on partial cues. So, we think that there is this balance of pattern separation and pattern completion that dictates the way we process environmental cues. An imbalance may result in improper retrieval of a previous memory, which in the soldier’s case could be the room in which he experienced the trauma. Once that happens, you then activate the limbic system, made up of the amygdala, which is important for fear, and the hypothalamus, which is important for the expression of glucocorticoid or stress hormones. The hippocampus, where you have adult neurogenesis, essentially acts as a break to restrain overactivation of other parts of our limbic brain, , and when you have encoding errors or impairments in pattern separation, or perceiving contextual cues, it relieves the break and you get hyperactivation of the stress and fear system.
Our emphasis is on understanding how we can stimulate adult hippocampal neurogenesis, but also how we can rejuvenate the older neurons, like I said, to make them younger or have the properties that confer increased pattern separation to the circuit, may have implications not just for improving memory function, age-related cognitive decline, which we know is accompanied by changes in pattern separation, as well as early stages of Alzheimer’s, but also for mood disorders such as PTSD and major depressive disorder.
How are fMRI and mouse assays for pattern separation run?
As the name of the mnemonic process indicates, we test animals for their ability to discriminate between overlapping representations and there are several ways that this can be done. It is important to note that the concept of pattern separation has its origins in the computational literature, on the basis of which, any circuit that contributes to pattern separation must generate more divergent output for similar input. And that’s one way the circuit enables discrimination between similar representations. So the task one uses to probe pattern separation in rodents must also integrate this key concept that underlies pattern separation. One way to do that would be to essentially vary the context of features and assay the animal’s ability to discriminate between contexts with slightly different features. The hypothesis is that pattern separation would be engaged when an animal is discriminating between overlapping features but when two contexts are very different you don’t really need pattern separation. So we use a fear-conditioning paradigm, in which we essentially give the animal a shock in a chamber and then we place it in a chamber that is similar but is safe and is not associated with a foot shock. Now the animal, because it is repeatedly exposed to these two contexts and everyday gets a shock in the context that is fearful, or rather becomes fearful, makes the association between the shock and the context and freezes. That’s the typical response that an animal shows once it learned that a context is fearful. On the other hand, if the animal has discriminated between the shock context and the safe context, then the animal will not freeze in the safe context, indicating that it has successfully discriminated between these two overlapping representations. Now a built in control would be to expose the animal to a context that is completely different from the fearful context and the animals should show normal levels of exploration and lack of freezing in this completely different context that’s safe but only impairment in a similar context. This allows us to infer that it is the dentate gyrus that is preferentially being engaged so that only in discrimination of similar contexts do you see an effect, but not in discrimination between very different contextual representations.
Now, we were very pleasantly surprised to learn of work that came out a few years ago in the human domain in which, using fMRI on individuals, several different groups have now found that individuals, when subjected to an incidental encoding task and are presented with a series of images, show preferential activation of the dentate gyrus only when they are presented images of similar objects, rather than very different objects. This is not the typical response to novelty that is seen in different regions of the medial temporal lobe upon first presentation of an object. It is only when you have a similar object that’s presented do you see activation of the dentate gyrus-CA3 circuit and this has lead researchers to propose this paradigm as a test for pattern separation. Other groups have now found that individuals with mild cognitive impairment that often proceeds Alzheimer’s disease, or individuals who have aged or are depressed are less efficient at this fMRI task discriminating between similar objects and they show an altered pattern of activation of the dentate gyrus CA3 circuit.
So these two paradigms in mice and in men actually have a lot of similarity in design and that is something we are really excited about because essentially it has enabled us, with collaborators in imaging and psychiatry at Massachusetts General Hospital to emphasize the construction of a hypothesis-driven pipeline in which in our laboratory we screen for molecules that may ultimately modulate the properties of new neurons to improve pattern separation in rodents, but the ultimate goal would be to use these noninvasive measures with imaging, to probe their efficacy on pattern separation in humans. Now, I should also point out that in addition to the imaging approach, we want to develop a behavioral assay that can be used in a high-throughput manner because getting an individual into a scanner is very expensive. Essentially, we’d like to present individuals with a series of objects such that the parametric properties are varied and as long as we know the assay has been validated by imaging showing preferential engagement of the dentate gyrus, we would use a behavioral assay as a readout for efficacy for compounds that we think might be pro-cognitive in that they improve pattern separation. These compounds may have therapeutic benefit not just for PTSD, but also for depression because depressed individuals typically show negative bias to contextual clues and this may have something to do with pattern separation. We know that the hippocampus is compromised in structure and there are changes in volume in disorders like PTSD and depression. We don’t know just yet if changes in neurogenesis may be accounting for these structural changes, but it would suggest that there might be impairments in pattern separation.
Have you visualized changes at the level of the synapse correlated with differences in pattern separation?
Yes, sort of, and that’s essentially what one project in the lab is focused on. We are interested in the properties of connectivity of the new neurons and how they causally relate to function. By that I mean we are looking for genes, and we have found genes, that we think are important for how these neurons make their connections with various cell types in the dentate gyrus-CA3 circuit. The challenge now is, using inducible genetics and viral tools, to probe the genetic requirement of these candidates in conferring certain circuit properties such as excitation-inhibition balance that we think are important for pattern separation. From one perspective, circuits in the brain should be treated as wiring diagrams so you can look at each part of the wiring diagram and ask, “Can we find a gene that encodes for that component?” and how does that component relate ultimately to the function of that wiring diagram or circuit. So, to come back to your question, that is what we are doing, and we think that that gene that is important for the function of new neurons may also enable us to, if misexpressed in the older neurons, change the properties of the synaptic connectivity of old neurons to those of the younger neurons because we know there are differences in both the properties and connectivity of the young adult born neuron as well as those of the neurons that were born throughout development. This gets back to the observation that just by changing levels of neurogenesis you can improve circuit performance despite the fact that the new neurons comprise only between 10-20% of the neurons, it has to be their properties, or the way they connect, that endow them with this capability So that’s why we are so interested in how these neurons do what they do.
So how can what we learn about differences in connectivity affect how we think about treating mood disorders?
I think it represents a fundamental shift in the way we think about mood disorders. It takes us from treating the brain as a black box where we use certain behavioral assays that have a high degree of predictive validity as a primary readout, to really focusing on a particular circuit in the brain and what it does, and using assays that are sensitive to the functions of that circuit. So, pattern separation is thought to be the property of a circuit that is primarily playing a role in the domain of cognition and here we are saying that this fundamental mnemonic process in addition to relevance to episodic memories, is important for the regulation of mood. So by targeting the same mechanism, we may have a much broader impact. This also gets at the idea of thinking of psychiatric disorders as a collection of circuit-based endophenotypes rather than a constellation of symptoms. I think when we start doing that, we will see that many disorders have circuit-based endophenotypes in common and impairments in pattern separation may be pervasive across the traditional diagnostic critera that currently stratifies these disorders as distinct based on the Diagnostic and Statistical Manual of Mental Disorders (DSM). I think this is an emerging theme in psychiatry and neuroscience. You have to think about circuits, which I think, will collapse the traditional boundaries that segregate these disorders. Ultimately we want to target circuit-based endophenotypes we can non-invasively and objectively quantify, and this will help assessment of risk/vulnerability and treatment responses. That’s the way I see the field moving.
Amar, thank you so much.
Sure, thank you.