The 2020 Pittendrigh/Aschoff Lecture: My Circadian Journey

Abstract

The circadian field has come a long way since I started as a postdoctoral fellow ~30 years ago. At the time, the only known animal clock gene was period, so I had the privilege of witnessing, and participating in, the molecular revolution that took us from the discovery of the circadian clock mechanism to the identification of pathways that link clocks to behavior and physiology. This lecture highlights my role and perspective in these developments, and also demonstrates how the successful use of Drosophila for studies of circadian rhythms inspired us to develop a fly model for sleep. I also touch upon my experiences as a non-white immigrant woman navigating my way through the US science and education system, and hope my story will be of interest to some.

I took the liberty of giving this lecture a title. I don’t think I had to do that but my circadian journey is what I will be talking about. There will be a couple of scientific threads in here and it’s also going to be my personal experiences along the way. This is really a trip down memory lane for me. So where did it all begin?

My initiation into science came with going to graduate school. I was not one of those people who grew up wanting to be a scientist. Very very far from it. It was very much a case of, I finished my studies in India and had to do something with my life, so you know, grad school it is. I came to the United States to visit family and ended up staying on here and going to grad school, to the only school that would have me, which was Cornell University in New York City or Weill Cornell as it’s now called. Cornell University admitted me to grad school but they provided no financial support. I got off to a rough start and, during that initial semester, ramen noodles is what I lived on; you could get six packets for a dollar. It was particularly rough as a foreign student because I wasn’t allowed to work off-campus either, but fortunately after the first semester, Cornell provided me aid and it was pretty much fine after that.

I went through the usual rotations people do in grad school and ended up picking for my PhD thesis a young faculty member who was just at that point moving to Cornell and getting started with his lab, Moses Chao. And it was he who inculcated in me a love of science, who showed me that this was a mystery just like those mystery stories that I love to read. He encouraged me to take risks in science. And he gave me all these tickets to operas and ballets; it was great to be his first student! Moses, even in starting his own lab, went in a pretty risky direction and I was happy to join him on that. We cloned the first mammalian (actually human) neuronal growth factor receptor, which was p75 (Chao et al., 1986). It was an exciting time but working with a human gene, I became very acutely aware of the limitations of research with the human system. Basically, there was no genetics, so you couldn’t put the gene in somewhere or knock it out and see what it did to function.

So I decided that for my postdoc, I had to work with a genetic model. Genetics was actually one of the aspects of science that I was interested in even before I went to grad school, probably the only aspect in fact. I decided that the model for me was going to be Drosophila, and this is where Mike Young came in. Here (Figure 1a) is a picture of Mike Young from that time. You will notice that he looked exactly the same then as he does now. Whoever said only women benefit from being blonde! Mike’s lab at the time worked not only on per, which they had cloned, but also on the notch gene. Half his lab at the time was neurodevelopmental and it was the neurodevelopment stuff that I was interested in because that is what fit with the work I had done in grad school. However, at the end of my interview with Mike, he gave me a paper they had in press on per and said, “I know you’re not interested in clocks but maybe you want to take a look at this.” Not only was I not interested in clocks, I knew absolutely nothing about clocks. I went home and I read this paper, and I was fascinated. The idea of studying the molecular basis of behavior really appealed to me. In fact, my “Prelim” in grad school was on this topic, but I trained as a molecular biologist at a time when everything was qualitative. You either had a band on a gel or you did not. I felt very uncomfortable with the behavioral work I saw out there, where people were looking at learning and memory mutants and the phenotypes weren’t that strong and you would need to do a lot of quantitation and statistics to convince yourself that the effect was reliably different from control. But the circadian work wasn’t like that. This was a very robust behavior and the mutants were very very penetrant and robust. Meantime they had molecular biology now to go with it. I called Mike and said, “I changed my mind. I think I want to work on per.” That is how I moved on to circadian frontiers.

Figure 1:

Getting started on circadian rhythms in Drosophila. (A) Sehgal’s postdoctoral mentor, Michael Young, in the early 1990s. (B) Results of the historic genetic screen for circadian mutants, conducted by Ron Konopka and Seymour Benzer (from Konopka and Benzer, 1971). The phenotype of the different per mutants is evident. (C) Sehgal and her postdoctoral collaborator, Jeffrey Price.

I got to Mike’s lab, raring to go, ready to do genetics. And that led me to undertake a forward genetic screen, like the one shown here (Figure 1b) in which the historic period (per) mutants were isolated by Ron Konopka in Seymour Benzer’s lab (Konopka and Benzer, 1971). Basically, we asked, can we identify other clock mutants? Now I should say that I was very green when I started my postdoc—I was really green when I started grad school too, but let’s focus on the postdoc now—and plus, I’ve always had this tendency to jump into things without thinking much about them. Sometimes that serves me well and maybe this was one of those times because, had I stopped to think about this, I would have said, “in 20 years there’s only been per, what are the chances that there is another clock mutant to be found?” As it was, I didn’t think about it, so I set out to do a screen. The first thing I had to do was build the setup for monitoring rhythms. When I got to Mike’s lab, they didn’t have the system in place because when Rob Jackson cloned per, during a sabbatical in Mike’s lab, he set the system up and then he took it with him to his own lab (Jackson et al., 1986). So I called Rob and said, “How do I do this,” and he was very helpful. He told me I needed to buy these antiquated Apple IIe computers, which were very hard to find, but I was eventually able to track down used versions of them at a garage in Shreveport, Louisiana. This was before the days of the Internet, of course. Then I spent many hours on the phone with Rob asking him what to do next. The typical conversation was Rob saying something like, “you need a transwarp card” and me going, “a trans what?.” At which point Rob would say, “I can tell you’re not a Star Wars fan.” Star Wars, Star Trek, I don’t know, all the same to me! It was all a challenge but we managed to get the system set up.

A year into the project, a new postdoctoral fellow, Jeff Price, came along (Figure 1c). He joined me on the project to screen for new mutants and this was a great collaboration. Jeff is a terrific colleague, a great collaborator. I really enjoyed working with him and I appreciate everything he did including even the little tests that he would sometimes put me through. Jeff would do things like take a new mutant we had, code it, and stick it back into the primary screen to see if I’d pick it up again. I’m happy to say I passed the test! Right about the time we got this mutant, a film crew from a PBS series called Infinite Voyage did an episode on circadian rhythms, and they came to the lab to talk to Mike and to film Mike interacting with some of us. The video clip I’m playing here wasn’t recorded with the best technology, but if you listen carefully, you can hear Mike say, “Jeff tells me you have a new mutant” (actually this might be the first time that Mike heard about the mutant). I answer, “yes it’s on the second chromosome.” I’m glad you can’t see the rest of me because I was quite pregnant at the time.

The mutant referred to in the video was timeless (tim) and here (see Figure 2) is the timeless phenotype. The original allele we got for timeless was arhythmic. Here’s your wild-type fly (Figure 2a), rhythmic in light:dark and rhythmic in constant dark. And here is timeless (Figure 2a), which becomes arhythmic in DD, shown also by the periodogram analysis. So that was the good news. There was a mutant. And then everything that could go wrong went wrong. We had done a P transposable element screen to isolate new mutants, expecting then that the phenotype would be linked to the P element, making it easy to clone. But when I started to cross this mutant out, I realized that the behavior was not segregating with the P element. Initially I thought I had contaminated my flies and wasted a few months just trying to clean up the stock, but eventually I realized it was not linked to the P element and so we were going to have to clone it the painful way—in other words, map the mutation, do a chromosome walk, and positionally clone it. In the meantime, we thought “What can we do with this?” Given that per was the only known gene at the time, it made sense to see how per was affected. And Paul Hardin in Michael Rosbash’s lab showed, right around then, that per RNA is expressed rhythmically (Hardin et al., 1990). So I set out to look at whether per RNA cycles in tim mutants.

Figure 2:

The discovery of timeless (tim): (A) Behavioral phenotype of the tim mutant. From Sehgal et al, 1994. (B) Loss of per RNA cycling in tim mutants. From Sehgal et al, 1994. (C) Data from (B) included in the Nobel poster.

This is where I have to tell you that I have always had terrible hands. I never ever generated a pretty experiment and this was no exception. Here you are looking at the cycling of per RNA over a 4-day period (Figure 2b). In wild-type, per cycles so robustly that even I couldn’t screw it up. But when it came to the timeless mutant, what we saw was erratic expression; per RNA went up, it came down, but with no obvious pattern. We couldn’t convince ourselves that it was cycling and we couldn’t convince ourselves that it was not cycling. The story of my life is that I always had to do experiments many more times than normal because they never looked good! Indeed, I ended up doing the RNA cycling experiment over many more days than people typically did or do, so here you’re looking at a 6-day experiment (Figure 2b). Transcript cycling at the time was assessed through RNAse protection experiments, which were kind of torturous because the RNA extraction required shaking (fly) heads in a −20 ° room, so I’d have to come in equipped with paraphernalia designed for the North Pole, and then you’d grind them in a mortar and pestle in liquid nitrogen. We eventually convinced ourselves that per RNA was not cycling in the tim mutant, but I was always very embarrassed by these data. Imagine my shock when more than 20 years later I got to accompany Mike to Stockholm as his guest and, lo and behold, on this poster put out by the Nobel assembly I saw these data right in the middle (Figure 2c)! As I remarked to somebody, “If somebody with good hands had done this experiment, it might have been a straight line in the timeless mutant, but because it was me it’s all over the place.” But the conclusion, though, was correct. And this is a lesson for all you trainees out there—you never know when what you do today may be important tomorrow or 25 years from now.

At the time though, I was very insecure. It was the grad student/postdoctoral spiral of doubt plaguing me. We decided to hold off on publishing the mutant until we could say something meaningful about it, and I kept wondering whether things were going to pan out. It didn’t help matters any that people were quite skeptical about this new mutant, perhaps rightfully so because there hadn’t been anything other than per in a long time. About that time, I went to the Drosophila neurobiology conference at Cold Spring Harbor where I presented a poster on the isolation of the mutant and sadly nobody came to my poster (except for one person who I’ll tell you about in a minute). This was the glory days of developmental biology, so while I had this very drab–looking black-and-white poster with activity records and periodograms, there were these gorgeous developmental posters, and one of those posters, unfortunately, was right next to mine, and that poster was mobbed. There was perpetually a crowd around it, and people kept elbowing me out of the way. It got to be really embarrassing, so at some point, I just pretended I was part of the crowd looking at that other poster. The saving grace, and actually a highlight of my life, is that Seymour Benzer came to my poster.

Somewhat tired of being a postdoc, I decided to apply for faculty positions. My insecurity about our findings, strangely enough, did not stop me; I believe my thinking at the time was that I should find a job before we discovered that the mutant was not a mutant! Because I was geographically restricted, I applied to a handful of places most of which rejected me. I got rejected from places that you’ve never heard of. It so happens that the furthest place I applied to was the only one that gave me an interview and that turned into a job, which I of course took, and that was the University of Pennsylvania. It wasn’t as though the people at Penn were interested in my new mutant, or in clocks for that matter. A couple of people on the search committee knew my graduate mentor and they were excited about the work I did in grad school. In fact, these people said, “We hope you’ll work on nerve growth factor when you come here.”

Starting my lab, I had to do all the tasks that we never get trained for but then are required of us as faculty members—set up a laboratory, teach, recruit, get funding. Many people here have been through this and I don’t need to tell you it was stressful for me. The stress was compounded by the fact that (a) the bulk of my postdoc work wasn’t published and (b) I had a very small start-up package because I was not in a good negotiating position. It turns out that I was not the only one who was nervous about my chances. As a woman, the only tenure-track woman in the department at the time, and a woman with a family at that, I was somewhat of a novelty for my department. They, my chair in particular, liked introducing me at seminars by saying that I not only did science but also had two kids that I had to take care of, so I really had two jobs. He said it the nicest possible way but I found out much later, thankfully not at that time, that he told other members of the department he was worried about whether a woman with two kids could meet the high bar for tenure at Penn. I should say, though, that once things started to go well, he actually put me up early for tenure and promotion, even though I felt I wasn’t ready, so at the end, it was all good. Initially though, it was a struggle, and it was particularly hard to get funded. The NIH (National Institutes of Health) would not fund me, even after we published the mutant, because they wanted to see the gene cloned. This is where you remember those helping hands that reach out to you. One of these was Marty Zatz. Marty Zatz was at the NIMH (National Institute of Mental Health) and he arranged for me to get invited to ad hoc at a study section so I could see what reviewers were looking for in grants. Thank you Marty!

Eventually we published this work. We published the mutant (Sehgal et al., 1994; Vosshall et al., 1994), and the following year, we published the cloning of the gene. We cloned the gene positionally and also collaborated with Chuck Weitz who pulled it out of a two-hybrid screen based on its interaction with PER (Gekakis et al., 1995; Myers et al., 1995). The collaboration with Chuck Weitz that Mike and I entered into was initiated at an SRBR (Society for Research on Biological Rhythms) meeting and this, as you know, is one of the great things about this meeting. (This particular meeting is no exception. Lots of ideas have come up over the past 2 days and I can see myself entering into new collaborations as a result of these.) Based upon our initial findings with the timeless gene, we proposed the idea that the clock consists not just of per but these two genes that are co-regulated and act together to negatively regulate their own transcription (Sehgal et al., 1995). In fact, we had a pretty detailed model for how we thought things worked (Figure 3). We only had data for timeless (tim) RNA but we also made predictions about the protein, which have largely held up with time. We went on then to look at the protein.

Figure 3:

Model for a circadian clock: The model postulates that per and tim are co-regulated, and the proteins partner to allow nuclear entry and negative feedback by PER. From Sehgal et al, 1995.

One of the first things we discovered about TIM was that the protein gets degraded by light. This work was done by my first graduate student, Melissa Hunter-Ensor, who raised antibodies to TIM and examined its expression in the brain at different times of day. She discovered it was really low at night, so she pulsed with light to see if light would make TIM go away (Hunter-Ensor et al., 1996). It did indeed, as you all know now (Figure 4a). We suggested that the way the clock resets in response to light is via light-induced degradation of TIM. We went on to show that the degradation of TIM correlates with the circadian light response (Yang et al., 1998). Dose response experiments using different durations and different intensities of light revealed that with increasing light, there was increasing degradation of TIM and increased behavioral resetting, suggesting that the two were causally linked (Figure 4b). This study was published back-to-back with a similar study from the Rosbash lab (Suri et al., 1998).

Figure 4:

Resetting the clock with light via TIM degradation: (A) Immunocytochemistry of the fly head, including the photoreceptor cells, whose apically located nuclei express PER and TIM. PER and TIM are also expressed in brain clock cells (lateral neurons) that co-stain with PDF. With increasing amounts of light, PDF expression remains but TIM goes away in the photoreceptor cells and also in the central clock cells. From Hunter-Ensor et al, 1996. (B) the TIM response to light correlates with the behavioral response. With increasing intensity or duration of light, more TIM gets degraded and is accompanied by a larger shift of the behavioral rhythm. From Yang et al, 1998. (C) The jetlag mutant is rhythmic in constant light although wild type flies are arrhythmic. From Koh et al, 2006. (D) Model for how CRY and TIM respond to light. JETLAG is the E3 ligase that targets TIM to the proteasome.

In subsequent work, we showed that TIM is degraded by the proteasome. At this point, you might say, “Sure, the protein is degraded. Why wouldn’t it be the proteasome?” But at the time, these things weren’t so clear-cut. Using larval brains, which could be incubated in pharmacological agents, we showed that proteasome inhibitors prevented degradation of TIM by light (Naidoo et al., 1999). We also asked how light was transmitted to the TIM. Using mutants of the visual system, we found that visual photoreceptors were not required for TIM degradation, although they did appear to influence it (Yang et al., 1998). We speculated that a dedicated circadian photoreceptor, perhaps a cryptochrome like molecule, mediated effects of light on TIM (Yang et al., 1998).

Cryptochromes were on my mind because my colleague and friend at Penn, Tony Cashmore, had discovered these in plants. And Tony had suggested to me that they could be relevant for circadian entrainment in other species. Lo and behold, the Hall and Rosbash labs, specifically Patrick Emery and Ralf Stanewsky working in those labs, identified cryptochrome in Drosophila and showed that it is a circadian photoreceptor (Emery et al., 1998; Stanewsky et al., 1998). We found that cryptochrome is also degraded by light in a proteasome-dependent fashion (Lin et al., 2001), leading us to propose the following—light activates cryptochrome, which transmits a signal to TIM to target it for degradation by the proteasome and then cryptochrome itself gets degraded by the proteasome.

Somewhere along the way, I started to go out on the road and get invited to give talks. My first seminar at a university was at UMass and Bill Schwartz was my host; and my first invited talk at a circadian meeting was at a circadian conference at Dartmouth organized by Jay Dunlap, thank you Bill and Jay. But then you lose some. For instance, I got invited to give a talk at the Brain Research conference, which is a prestigious conference, so I was excited about it. But then a few days later, the organizer called me back and rescinded the invitation because, in the meantime, Michael Rosbash had accepted the invitation that they first extended to him. I note that I have never again been invited to the Brain Research conference. It was not Michael’s fault, of course. Given that I was coming out of the lab of the rival Mike Young, Michael was initially the enemy. Thankfully, this did not last. We may have had our ups and downs at first, but Michael went on to become a close colleague, and I’m very very grateful for the support I’ve received from him over the years.

Back to timeless! I told you that in response to light, TIM gets targeted for degradation by the proteasome. But what targets it? We, quite fortuitously, stumbled upon the molecule that does the targeting. This was work done by Kyunghee Koh, a postdoc in the lab, who now has her own lab. Kyunghee found that one of the lab stocks she was working with had reduced sensitivity to light. Wild-type flies are rhythmic in constant dark and they lose rhythms in constant light because TIM (and subsequently PER) gets continuously degraded. This particular stock was rhythmic in constant light, suggesting that it was less sensitive to light (Figure 4c). The stock also had extended jetlag, such that it took a long time to adjust to a new light-dark cycle, so Kyunghee called the mutant “jetlag.” She found that the phenotype of reduced sensitivity to light was also evident in a few other lab stocks. Kyunghee went ahead and mapped the lesion and positionally cloned the jetlag gene, which turned out to encode an E3 ligase (Koh et al., 2006). Sequencing this E3 ligase in the stocks that had reduced light sensitivity, Kyunghee identified mutations that accounted for the aberrant light response. She also showed that JETLAG is the E3 ligase that targets TIM for degradation in response to light (Figure 4d). This is the story of the TIM light response.

We’ve also worked on other aspects of TIM like its phosphorylation and regulation and its impact on PER. A quite recent TIM story that I want to touch upon addresses a long-standing question in the circadian field. We all know that clock genes encode RNAs that cycle. The RNA peaks at a specific time of day and then the protein encoded by that RNA comes along and inhibits the transcription. You have to have delays built into the system because if RNA synthesis were followed immediately by RNA inhibition, you would reach equilibrium between synthesis and inhibition and you wouldn’t have a cycle. So you need a delay in the feedback by the protein. One of these delays is between the peak of the RNA and the peak of the protein (Figure 5a). This is true for tim RNA and protein, for per RNA and protein and also true for mammalian clock genes. We could explain how you get that delay in Drosophila in light:dark cycles because per/tim RNA builds up during the light phase, but light-induced degradation of TIM prevents accumulation of the proteins until the light goes off. However, we know that circadian rhythms persist in constant darkness, and the molecular mechanism persists in constant darkness. How do you get delayed accumulation of TIM in constant darkness? Clues to that came from this recent work.

Figure 5:

Accounting for the lag between tim RNA and protein. (A) The peak of TIM protein lags behind the peak of the RNA by 6 hours. In light:dark cycles (shown) this lag likely arises from light-induced instability of TIM, (B) Alternative splicing of tim at the tim-tiny intron contributes to the lag in constant darkness. When tim RNA is first made in a daily cycle, the intron is retained, so truncated unstable TIM is made. Later, the intron is spliced, allowing accumulation of stable full-length TIM. From Shakhmantsir et al, 2018.

This work was done by a graduate student, Iryna Shakhmantsir, and it built upon our discovery of phosphatases, specifically PP2A and PP1, as part of the clock mechanism (Fang et al., 2007; Sathyanarayanan et al., 2004). Iryna was interested in kinase activity targeted by these phosphatases and so she looked for new kinases that affect circadian rhythms and identified a role for Pre mRNA Processing Factor 4 (PRP4). This particular molecule is a kinase but it is also a splicing factor, and Iryna found that PRP4 functions in the clock as a splicing factor. We did an unbiased screen to identify genes targeted by PRP4, and a top hit was tim. I’m going to jump to the model to tell you how PRP4 affects the splicing of tim and how that then contributes to the delay that I’m talking about (Figure 5b). Basically, an intron in tim that we’re calling tim-tiny is alternatively spliced by PRP4. When this intron is retained, the reading frame is truncated to generate an unstable TM protein; when the intron is spliced out, you get normal full-length and stable TIM. Retention of the tim-tiny intron is cyclic; in other words, the alternative splicing is cyclic. During daytime, or subjective daytime in constant dark, the intron is retained, so even though tim RNA is high, the protein does not build up because only unstable truncated TIM is made. Later on, during the night, the intron is spliced out and normal full-length TIM is generated, which then also allows the accumulation of PER. This is a mechanism by which you can get a delay between RNA and protein in the clock mechanism (Shakhmantsir et al., 2018).

That is the last tim story I’m going to mention and, as Katja (the moderator) already alluded, over the years, we’ve worked on several different things. At some point, we moved from looking at clock mechanisms and the response of the clock to light to understanding how clocks drive behavior and physiology. We mapped circuits in the fly brain that drive behavior (Barber et al., 2016; Cavanaugh et al., 2014; Erion et al., 2016; King et al., 2017); we identified clocks in peripheral tissues, in the fly fat body, which is like the fly liver, and in the prothoracic gland that drives ecdysone (Myers et al., 2003; Xu et al., 2008); we discovered a clock in the blood-brain barrier that controls interactions between the brain and the periphery (Zhang et al., 2018); and, also as Katja mentioned, we’ve started to translate some of this work to mammals, specifically mouse models and human cells. We’ve also initiated studies to address consequences of circadian disruption and the propensity that they might cause to disease, in particular, cancer (Lee et al., 2019).

We also became interested in the question of sleep, and this came largely from my interactions with sleep researchers at Penn. The question we asked was whether flies could be used to study sleep. This work was driven by Joan Hendricks, a professor at the Penn vet school who worked on sleep apnea in bulldogs. You might know that bulldogs are a great model for sleep apnea because the entire breed has this disorder. Joan came to my lab for a sabbatical to learn molecular biology, but while she was in the lab, we started to think about whether flies sleep. We were motivated, in part, by what was happening in the sleep field at the time. The circadian and sleep fields have evolved independently with separate journals, separate meetings. The sleep field, at that point, was realizing that the circadian field had made a lot of progress in understanding basic mechanisms, and there was this idea that maybe small animal models that provided insight into circadian rhythms could be employed to study sleep. From our studies of circadian rest:activity behavior in flies, we knew that flies have rest phases in-between bouts of activity, so Joan set out to determine whether fly rest constitutes a sleep-like state. She based her studies upon the pioneering work of Irene Tobler who, in the 1980s, had asked if rest in small animal models meets criteria for sleep (Tobler, 1983). The most important of these criteria is that there should be a homeostatic need for the state; in other words, if you deprive an animal of sleep, it should have the need to make up the sleep lost because sleep serves essential functions. We may not know exactly what those functions are but it’s essential.

I’m going to tell you a bit more about homeostatic mechanisms in a minute but I also want to point out that additional features of Drosophila rest, which are not necessarily criteria for sleep, are reminiscent of sleep states. Notably, fly rest or sleep responds to the same drugs that affect human sleep. In fact, we’ve managed to use the fly to do a screen for novel small molecule regulators (Nall and Sehgal, 2013). Homeostatic regulation in sleep is best illustrated by the response to sleep deprivation. Typically, sleep is the best known example of a circadian rhythm, and it occurs at a specific time of day. For us, that would be the night time but if you don’t sleep well at night or you go to sleep late, then in the morning, even though your circadian clock and your alarm clock are telling you to wake up, you will want to sleep on; that’s the sleep homeostat at work. We showed that this homeostatic regulation exists in flies (Hendricks et al., 2000). Initially, there was a lot of resistance to the fly model as the sleep field was very wedded to the idea that sleep only occurs in animals in which you could monitor EEG. But I’m happy to say that they eventually came around, and now fly sleep is a regular feature at sleep meetings and we get invited to those meetings. There are easily about 50 labs that currently study fly sleep.

The Drosophila sleep model was followed by the development of other small animal models so there are now Caenorhabditis. elegans, zebrafish, Aplysia and, most recently, Jellyfish models of sleep (Nath et al., 2017; Prober et al., 2006; Raizen et al., 2008; Vorster et al., 2014). So what do we want to do with the fly model for sleep? We are asking two main questions: how is the drive to sleep generated and why do we sleep? We are approaching the question of why we sleep in several different ways. Our data indicate that sleep is required for metabolic homeostasis. This includes the clearance of metabolic wastes, which could occur on a cellular level and also on a brain-wide level through the blood-brain barrier (Artiushin et al., 2018).

For the last few minutes of my talk, I’m going to tell you about our efforts to understand how the drive to sleep is generated. In the early days when we developed the fly model for sleep, we tried to determine whether the neuromodulators that affect mammalian sleep affect fly sleep and indeed they do. Gamma aminobutyric acid (GABA) is potently sleep promoting and dopamine is arousal promoting (Sehgal and Mignot, 2011). But we didn’t develop a fly model to say that what is true in mammals is true in flies. You want to be able to go beyond that and discover novel genes through unbiased approaches, specifically forward genetic screens, which worked so well for us with respect to circadian rhythms. The other point I would make here is that there’s a specific advantage of flies for sleep and that is that they have less redundancy and less compensation. That is important because although circadian rhythms are a very robust behavior, as I mentioned earlier in my talk, sleep tends to be very variable from individual to individual and also at different points for a particular individual. It changes with age, with environment and so on and so forth. To have a mutant that you can use to dissect mechanism, you need a very strong phenotype that’s cleanly separated from the wide distribution of the mean. The lower redundancy/compensation in the fly allows us to get such mutants, one example provided by the sleepless mutant, in which sleep amount is reduced by 80% (Koh et al., 2008) (Figure 6a).

Figure 6:

Short-sleeping mutants in Drosophila: (A) Sleep profiles are shown for male and female flies (averaged from a population). Males have a large afternoon siesta, which is reduced in females. Sleep is greatly reduced in the sleepless (sss) mutant; quantification on right. From Koh et al, 2008. (B) Short-sleeping mutants isolated through forward genetic screens in Drosophila (Cirelli et al., 2005; Kume et al., 2005; Liu et al., 2014; Pfeiffenberger and Allada, 2012; Rogulja and Young, 2012; Shi et al., 2014; Stavropoulos and Young, 2011; Wu et al., 2008)

Hence we can get strong mutants, and here are the other mutants that have come out of genetic screens done by us and by others (Figure 6b). Many of them affect the neuromodulators that I mentioned earlier. Importantly, the sleep-regulating pathways we’re finding here are conserved in humans. To date, the studies that have been done in human sleep to identify genetic factors, be they genome-wide association studies or studies of human syndromes associated with insomnia, are implicating the same pathways that we are finding through forward genetics in flies (Allebrandt et al., 2013; Cornelius et al., 2011). In addition, I’d like to note a heroic screen done by Masashi Yanagisawa and Joseph Takahashi, in which they pulled out two sleep mutants, corresponding to two different genes, in mice (Funato et al., 2016). We were excited about this paper, especially the fact that we had previously identified one of their new sleep genes as a sleep regulator in flies (Joiner et al., 2013). The other mouse gene was known to be a sleep regulator in worms (van der Linden et al., 2008). Thus, mechanisms underlying sleep are conserved from worms to mammals, including humans.

The genes that we found (Figure 6b) came out of screens for short sleeping mutants. The thinking was that if you mutate a gene and you decrease sleep, you should increase sleep when you overexpress the gene. That turns out to not be the case for most of the sleep genes we found in flies. We think that these genes are permissive for sleep. They’re required to allow the brain to sleep in response to an upstream signal but they’re not instructive. We asked if we could find something that is instructive, something that can drive sleep. And so, Hirofumi Toda, a postdoc who now has his own lab, did a humongous screen where he overexpressed most Drosophila genes one at a time and asked whether they affect sleep in males and in females. From this screen of >12,000 lines, we got 7 lines that correspond to six genes. One gene was isolated twice through two independent lines, which suggests that the screen is close to saturating. Five of these genes decrease sleep. Only one, which we named nemuri, proved to be an instructive signal that increases sleep (Toda et al., 2019).

The nemuri gene encodes a secreted antimicrobial peptide that, under normal conditions, is expressed at very low levels. It is induced under conditions that promote sleep need, which could be sleep deprivation or infection. We all sleep more when we are sick and the increased sleep is restorative. The gene nemuri is required for sleep in animals that are infected and need to sleep more and animals that haven’t slept for a long time. This would suggest that the mechanisms underlying those two types of sleep need are similar. To summarize our findings with nemuri, it is necessary for sleep under conditions of high sleep need, it is sufficient to drive sleep, and its levels correlate with sleep need. The reason I’m making a point of this is because those of you who’ve been in the circadian field long enough remember that we, at one point, had developed criteria for how you would define a circadian clock component; it should be expressed with a circadian rhythm, its loss should affect the rhythm, its overexpression should affect the rhythm. Now, we’re developing criteria for sleep homeostasis factors. We believe nemuri is such a factor, which is why I chose to highlight it here. I felt that it nicely illustrates how we’re using lessons from circadian biology and applying them to the study of sleep. The gene nemuri links sleep and immune functions (Figure 7). It kills microbes in the periphery. In the brain, it is expressed in specific neurons, which, under conditions of high sleep need, act through the major sleep promoting structure of the fly brain to drive sleep. Thus, nemuri has two functions, one of which is related to the immune response. The other function (sleep promotion) is not, but both the functions serve to promote recovery.

Figure 7:

The gene nemuri links sleep and immune function: nemuri encodes an anti-microbial peptide that is expressed in the brain and periphery. It kills bacteria and is also induced under conditions of high sleep need to increase sleep through a major-sleep-regulating region of the fly brain. Both functions likely promote survival.

In closing, I want to end with something non-scientific. This is a message to the young people out there, in particular, to young women. It is important to be heard. I’m no longer the only tenured woman in my department, which is really great but I often find myself in situations where I’m the only woman or one of the few women in a room full of people. You may think that at this point in my life and my career, it is not hard for me to be heard but that’s actually not the case. Of course, it doesn’t help matters when the other people are a bunch of clinical chairs and they gang up on me. It is easy to get intimidated under those conditions. There are also times when I speak up and I feel like nobody has really paid attention to what I’ve said. Later I’ll see somebody else applauded for something I’ve said before. Somehow my ideas sound a lot better when they’re coming out of the mouths of men. As a very prominent female neuroscientist once said to me, “It’s very hard for women to be heard. The physical presence is relevant and, for that reason, you have it particularly hard because you are short, you are dark, you are foreign and you have a high voice.” But the message is that you have to keep trying. This is my message to you all, be heard.

Finally, I would like to acknowledge all the trainees I’ve had over the years. I am fortunate to have had tremendous predoctoral and postdoctoral trainees who’ve gone on to successful careers. Many of them now have their own labs. I think the most important thing we do is to train the next generation of scientists. I am super proud of my trainees, past and present (Figure 8). I’d also like to acknowledge my funding, past and present. One of my current grants came, in part, out of a mini-sabbatical I did with Martha Merrow, which speaks to the tremendous community we have in this field.

Figure 8:

The Sehgal laboratory, and current sources of funding. (Abbreviation: HHMI = Howard Hughes Medical Institute.)