Dr. Zachary Knight received his undergraduate degree in Chemistry from Princeton, where he developed chemical approaches for mapping protein phosphorylation sites. He then completed his doctoral research in Chemical Biology at UCSF. As a graduate student Dr. Knight synthesized some of the first selective inhibitors of PI3-K and mTOR and used these tools to show that PI3-Kalpha mediates the metabolic effects of insulin. In 2007 he co-founded a company, Intellikine, to help advance these compounds towards clinical trials, two drugs based on his graduate work are currently in clinical trials. Dr. Knight did his postdoctoral research at Rockefeller University, where he developed a technique for molecular profiling of activated neurons in the mouse brain. He has used this approach to identify new populations of neurons in the hypothalamus that control feeding. In November 2012, Dr. Knight joined the UCSF Department of Physiology. His lab is exploring the molecular organization of circuits that control innate behaviors and how these circuits are modified by drugs and disease.
Dr. Knight’s lab studies neural circuits in the mouse that control feeding and other motivated behaviors central to survival. The labs goal is to understand how these circuits are able to sense the needs of the body and then generate the specific behavioral responses that restore homeostasis. During the seminar, Dr. Knight described how the brain’s “hunger circuit” governs eating. Using newly developed fiber optic devices (video link) that allow one to record neural activity in awake behaving mice, his lab was able to measure the activity of AgRP and POMC neurons (two specific types of neurons associated with hunger sensing) in real time, when mice were given food after a period of fasting. The result was surprising, because the expectation was that if you give a hungry mouse some food, the hunger neurons would slowly drop their activity as the mouse eats. However, Dr. Knight’s lab found that almost immediately the AgRP-POMC neurons reversed their activation state, even before the mouse consumed any food. The initial drop in activity of the neurons occurred when the mouse first saw and smelled the food, as if the mere knowledge of available food satiates the AgRP-POMC neurons. They also discovered that the reaction of the AgRP-POMC neurons correlated with the palatability of the food offered: when mice were fed peanut butter or chocolate, there was a stronger and more rapid reversal of AgRP-POMC activity than when the mice were fed normal food. Dr. Knight also discussed possible implications of these findings, and why it may be necessary to broaden the concept of homeostasis in order to explain the dynamic mechanisms by which these circuits control behavior.
SAGE sat down with Dr. Knight to ask a few more questions…
Q: Can you talk about more these fiber optic devices? What is the advantage of this technology?
ZK: The technology to do these experiments has only existed for a few years, but it has quickly become widespread. It basically involves expressing within a specific neural cell type a protein called a GCaMP that emits fluorescence in response to neural activity. This fluorescence is then recorded through an optical fiber implanted deep in the brain. In this way you can optically monitor a neuron’s activity in freely behaving animals.
The reason this is such a major advance is that it allows you to look at the activity of specific, identified cell types, rather than random, unidentified neurons. This is important because the circuits we study are embedded within brain structures such as the hypothalamus which contain a vast diversity of neural cell types. Using traditional approaches in neuroscience, there is no way to isolate individual circuit elements for functional analysis. As a result we know remarkably little about the specific cell types, interconnections, and activity dynamics that drive these fundamental behaviors such as feeding. So one way we address this challenge is by using new optical tools to record the activity of specific cell types during behavior. A second way we approach this problem is by developing genetic tools that enable the use of RNA sequencing to identify molecular markers for functional populations of neurons in specific brain regions. We then use these molecular markers as entry points to visualize and manipulate the underlying cells, as I showed in the talk. Our long-term goal is to elucidate the structure and dynamics of these homeostatic circuits, so that we can begin to understand how they give rise to behavior and how they become dysregulated in conditions such as obesity.
Q: What kind of neurons can you target in the brain?
ZK: You can target any neuron you want as long as you know what it is, meaning its gene expression pattern (one can think of this as a unique barcode for that cell type). The challenge is that for most neurons in the brain we don’t have a specific molecular marker that identifies them. We know that these cells exist, but we don’t know their molecular identity. For example, the hunger neurons (AgRP and POMC) that I talked about today, we can study those because the AgRP and POMC genes specifically label them. There are many other hunger neurons that we can’t study because we don’t have such a specific label. However that will likely change as we learn more about gene expression in the brain.
Q: How many neurons are there in a mouse brain?
ZK: There are around 75 million neurons in a mouse’s brain. It is important to emphasize that although we know the total number of neurons, we don’t actually know how many different types of neurons are in the brain. Cataloging all the different types of neurons is the brain is one of the goals of President Obama’s Brain Initiative. As an example, in the hypothalamus, there are at least several hundred different types of neurons, maybe even a thousand depending on how you define a cell type. We know something about maybe a few dozen of these cell types, and most of them we haven’t studied in any detail. So there is a lot to learn.
Q: Do we know why the mice keep consuming food after the hunger neurons activity goes down?
ZK: This is still a mystery. Basically, what people previously thought was that the hunger neurons would stay active, driving the eating response, and then their activity would gradually decrease when you reach the end of the meal. It was expected that the activity of the hunger neurons would correlate with the subjective experience of hunger and satiation, but our data show that this is not the case. Our data show that when the mouse sees food, the hunger neurons activity goes down, but the mouse doesn’t lose the desire to eat, if anything, the desire to eat is increased. This is a paradox.
Q: Have you studied obese mice?
ZK: Many researchers have studied different types of obese mice and looked at different neurons, but no one has applied the specific technology that I discussed today to studies of obese animals. That would be interesting. My guess would be that the differences in the activity dynamics between lean and obese mice would be subtle, because I think that obesity is something that results from very small mismatches between energy intake and energy expenditure that accumulate over a long period of time. However we don’t know until someone does the experiment.
Q: There are some experiments in flies that show with dietary restriction, the flies live longer; but if the flies can sense the food even without access to the food, they don’t live longer. What would be your interpretation of this unusual result?
ZK: Although flies don’t have AgRP neurons, they probably have neurons that do something similar. One interpretation would be that in order to make dietary restriction work, it is necessary to have the hunger neurons chronically activated. However, our data indicate that being able to see and smell food, even if you can’t eat it, suppresses the hunger neurons’ activity. In this scenario, then perhaps the longevity extension associated with dietary restriction would also be decreased.
Q: Have other scientists looked the relationship between the mTOR pathway and these neural circuits?
ZK: Many scientists have looked at the role of mTOR signaling in these neurons as well as other nutrient sensors. It was initially thought that these neurons could sense nutrients directly via mTOR or AMPK (two different signaling molecules). One piece of evidence supporting this was the observation that genetic manipulations of these signaling pathways in specific hunger neurons could affect food intake and body weight. However I think the field has moved away from that idea to a certain extent. One reason is that proteins like mTOR are generally important for the biology of all cells, and so activating or inhibiting one of these proteins generally alters cell function or viability. If you do that in a hunger neuron, then of course you get a phenotype related to feeding. But that doesn’t mean that the activity of those neurons is regulated by mTOR dependent nutrient sensing. It just means that mTOR is an important protein for the health and viability of that cell, like most other cells.
Q: What is the big picture concept of your research?
ZK: Obesity is obviously a major problem. To better treat obesity, we need to understand the underlying mechanisms that control food intake, so we can find optimal points of intervention. It is pretty clear that the brain is the primary organ that regulates energy balance, but we know incredibly little about how it does that. So I think mapping out the neural circuits that control feeding is a necessary step toward the rational design of an obesity treatment. This kind of research won’t lead to the discovery of an obesity drug tomorrow, but it will create the knowledge base that is necessary in order to understand how the obesity drugs of the future will work.