It was mostly descriptive. The PER gene had been cloned in fruit flies. I looked at plants for the PER gene, but I couldn’t find it. There were no PER gene homologues, but nobody had described genes being regulated in a circadian fashion.

The gene product is pretty boring. It’s a light-harvesting protein. It binds chlorophyll and holds it into the right orientation to channel light energy into photo-reaction centers. It acts like an antenna, which is what it is—it’s the rabbit ears of the complex. We were studying it because it’s a gene beautifully regulated by light. It was providing the paradigm of how we could begin to understand in plants how light regulates the development in genes, by switching genes on and off.

Plants have amazing arrays of photoreceptors that detect light of different colors, just as we do. But instead of forming an image, they transduce light signals into different suites of genes, coming on and off. It was a complete piece of serendipity that we stumbled onto the fact that it was a clock regulating this gene we were studying.

That’s true. I felt it was an opportunity to really turn this into a project. The problem was that in Drosophila, you could use rhythmic behavior to screen for mutants of circadian rhythmicity. That’s how the PER gene was originally identified. In plants, you had a model plant, Arabidopsis, but no way to apply genetics to discover not just the genes that responded to the clock, like CAB, but genes that formed the core part of the clock, the cogs rather than the hands. We needed a way to do plant genetics.

I came up with this crazy idea that you can regulate the sequence of the CAB gene and attach it to the luciferase gene from fire flies. So we were able to make transgenic Arabidopsis that glowed on and off. Nobody ever believed we would get this to work; it wasn’t easy to do. But by 1991-92, we had Arabidopsis glowing on and off, and we were ready to do mutant screens.

At that point I moved from Rockefeller to the University of Virginia, where we obtained a huge grant from the NSF to form the Center for Biological Timing under the direction of Gene Block. The idea was that people studying circadian rhythms in many different organisms and from many different institutions could get together and try and break open the field. Even up until the early 1990s, the field was largely descriptive, measuring a gazillion different rhythms in different organisms. With the exception of the fly work, and some in Neurospora, nobody was making much progress in identifying any kind of molecular mechanics.

So we took these glow-in-the-dark plants, and we took mutant genes and—hey, presto—we discovered clock genes. To this day, we still use that tool. Now we’re beginning to have a pretty good picture of what the clock looks like in Arabidopsis, and the clock gene in Arabidopsis, and they look quite different than the genes described in flies or mice. This raises the hypothesis that circadian rhythms have risen multiple times during evolution.

Well, during this period, I was really lucky to meet an amazing guy named Jeff Hall. And Jeff basically came to me and said, "Look, this is a fantastic thing you did in plants with luciferase. We should try this in Drosophila." So Jeff and I started collaborating; we began to learn fly genetics and began to do fly genetics in my lab, and we discovered some pretty interesting things.

We first discovered, much to our surprise, that clocks were all over the fly body. One of our other more highly cited papers is a 1997 paper in Science reporting this (Plautz JD, et al., "Independent photoreceptive circadian clocks throughout Drosophila," Science 278[5343]:1632-5, 28 November 1997). And the reason why it’s surprising, especially in the mammalian field, is that the whole focus had been on how rhythms describe behavior. People had identified a sub-region of the hypothalamus called the SCN, which was known to contain essentially a master clock that controlled behavior. And so the whole focus in behavior was this clock in the brain.

Then we came along and showed that you could culture bits and pieces of different fly cells and they’d all show these beautiful circadian rhythms, completely in the absence of the brain. What was even more surprising was that if you shined a light on these tissues in a test tube, you could reset the clock with light, just like you can reset it in vivo.

These circadian rhythms we had identified could persist independently of the light-dark cycle. That told us there was a real clock there, but to be useful, the hands of the clock have to be bumped each day, generally by light in the morning and the evening. The clock has to adjust to changes in daylight all year long. Here we could show this happening in the test tube.

We then went on and did a mutant screen for Drosophila, and we ended up discovering a gene that my lab had been working on in plants called cryptochrome, a blue light photo-detector, and so this was pretty cool. Actually then it subsequently has turned out cryptochrome seems to be involved in circadian rhythms in many different organisms. It’s the one protein that circadian clocks appear to have in common.

We wanted to understand, of course, the molecular mechanism of clocks in more detail. And here’s a strange twist of fate. A lot of clock genes in mice have been identified by looking for homologues of genes first identified in mutant screens in Drosophila. That’s the classic way of doing things. You do mutant screens in Drosophila, you clone the genes, and sure enough, there’s usually an orthologue in mice.

Well, what turned this on its head came from another guy who was part of the Center for Biological Timing—Joe Takahashi at Northwestern. Joe did something pretty brave; he said, "Look, we can do these types of genetic screens in mice. It’s expensive, a gigantic project, but I’m up for it, and I’m going to do it." So Joe did mutant screens, and he wanted to isolate the same thing we had seen in flies. He used a classical laboratory assay: lab mice will hop onto a running wheel around dusk and will run, on average, five miles a night. And, unbelievably, mouse number 25 was defective. It had this defective circadian rhythm. Joe called this mouse mutant clock, and he called me up and said, "We cloned this gene in mice and it’s a particular type of transcription factor called a bHLH-PAS transcription factor." That immediately blew me away, because we knew that the PER protein was a PAS protein—PER is short for "period," by the way.

So we did the opposite of what most people were doing at the time. We took Joe’s mouse gene and we screened libraries of fly genes. Now you do this on the computer, but then we actually had to do it the old-fashioned way. And we isolated a fly version of Joe’s mutant clock gene. And we found then that the fly clock gene interacted with another gene in fly that is called BMAL1. They interact and switch on this PER gene. And at the same time, they switch on another gene called TIM, which stands for "timeless."

So these two proteins switched on PER and TIM. They stick together and come back around to the nucleus and switch their own genes off, stopping the action of clock and BMAL. It’s a classic negative feedback loop. And we were able to reconstruct this in a cell line. In collaboration with Chuck Weitz at Harvard, we put in the fly genes, clock and BMAL, and we showed that they induced PER and TIM expression. And when we took PER and TIM and put them back into the cell it prevented expression of clock and BMAL. That’s why the paper is called "closing the loop" (Darlington TK, et al., "Closing the circadian loop: clock-induced transcription of its own inhibitors PER and TIM," Science 280[5369]:1599-1603, 5 June 1998).

Because it describes for the first time the actual molecular loop that is the core of this timing mechanism. It actually says, "This is how you make a clock inside of a cell using proteins." You can set up this clock as a negative feedback loop of transcription, because of the negative feedback loop. As PER and TIM levels rise, they switch off their own expression, and when they fall far enough, the expression is switched back on. And they start to rise again.

The paper provided a molecular picture of how this set up an oscillation. Probably another reason it’s cited so much is that a similar mechanism was identified shortly after in mammalian cells. Although the thing about the mammalian clock is that TIM doesn’t seem to play a role.

Yes, various people, including ourselves, have shown that your liver, skin cells, kidney cells, gut cells, cardiovascular cells, the lining of your arteries, all have these beautiful circadian rhythms. The paradigm we had set in the flies now turned out to be reiterated in mammals, too.

It’s really interesting. To try to summarize it, the type of biology we were doing, where we were describing these networks of gene interactions, has now been given a buzzword. It’s called systems biology. I didn’t know I was a systems biologist when I was doing this work, but apparently I am.

With these systems approaches, people treat cells like they’re Boeing 777s. They want to understand the intricate wiring, and now we’ve discovered in flies and mammals and plants that the clock is not just made up of one loop, but multiple interacting loops. If you imagine one loop being a ring, the clock now looks a bit like the Olympics emblem—multiple interacting loops defining a complex circuitry.

Well, one final point I’m making here is that even though you don’t see a lot of conservation of genes themselves, between these model organisms, we’re starting to see conservation in the way these circuits are constructed. In all these organisms, we’re seeing these elaborate, interconnected loops, presumably there to provide robustness and precision to the system. The interesting thing here is that the comparative genomics of circadian rhythms is beginning to teach us not so much about the secrets of the genes doing the work, but about the ultimate architecture of the circuitry that could have been a substrate of convergent evolution.

Well, our lab has moved on quite a bit. A few years ago we decided that we were probably pretty much done with any contributions we could make in Drosophila. We finished up by discovering some genes in the brain—a potassium channel called slowpoke—that seem to be important for rhythmic behavior. And what I decided was that we would move increasingly into mammals, where the issues of how the clock is constructed in different tissues and how different tissues interact are a real interest to me.

Our lab is currently half Arabidopsis and half mouse. And in Arabidopsis, we’re beginning to discover this elaborated circuitry we’ve seen in other organisms. We’ve now made the link between 24-hour systems and seasonal timing of flowers. So we’re beginning to make real strides in understanding how this 24-hour clock is used as seasonal timer. I find that very interesting. In mammals, we’re very much interested in whether the clocks in all these various organs and cells are constructed in the same way. Are we seeing tissue-specific heterogeneity? And what does that heterogeneity control? That’s what we hope to understand.