A Blog Around The Clock

i-710d005c8660d36282911838843a792d-ClockWeb logo2.JPGThis is a summary of my 1999 paper, following in the footsteps of the work I described here two days ago. The work described in that earlier post was done surprisingly quickly – in about a year – so I decided to do some more for my Masters Thesis.

The obvious next thing to do was to expose the quail to T-cycles, i.e., non-24h cycles. This is some arcane circadiana, so please refer to the series of posts on entrainment from yesterday and the two posts on seasonality and photoperiodism posted this morning so you can follow the discussion below:

There were three big reasons for me to attempt the T-cycle experiment at that time:

A Phase-Response Curve Controversy

The genetic revolution in chronobiology was only about to begin back in 1995 when I was doing this. Since then, so many researchers have jumped on the molecular bandwagon, not to mention a huge influx of people into the field who had their training in genetics instead of chronobiology and not just did not know, but did not even know they did not know anything about the stuff I covered in my clock tutorials this week. So, today, very few people seem to care about such issues, but back in mid-1990s there was a battle going on behind the scenes.

In his 1965 book, Jurgen Aschoff described six different experimental protocols that can be used to derive a phase-response curve (PRC), suggesting that all six are equivalent to each other and that different species and different hypotheses will guide the researcher to pick one or the other among the six methods.

However, almost all of the PRCs in the field have been derived using the first method – later dubbed Aschoff’s Type I – which is the method I described in the series of posts yesterday. In short, animals are kept in constant darkness and brief light-pulses are applied at different phases of the cycle. The resulting phase-shifts are plotted to construct a PRC.

It appeared, though, that whenever someone chose to use another method, there was a price to pay – the reviewers did not like it, probably because they did not fully believe that the six methods are really equivalent. In the end, the dispute went public in 1996 when Nicholas Mrosovsky (the guy who invented the term ‘rheostasis’) published a paper with a title Methods of measuring phase shifts: Why I continue to use an Aschoff Type II procedure despite the skepticism of referees (Chronobiol International 13:387-392). The Journal ended up publishing two letters to the editor commenting on this paper, further highligting how flammable the issue was at the time.

I have just re-read Mrosovsky’s paper this morning, in which he describes Aschoff’s Type II method and explains why is it the most suitable method for the questions he is asking (effects of non-photic cus) in the species he is using (hamster) and also explains why the two methods are fully equivalent. I do not see anything controversial in what he states in there.

Measuring the phase-angle between the onset of a circadian rhythm and the onset of light in T-cycles is Aschoff’s Type VI method for constructing the PRC. As Kent Edmonds, a post-doc in our lab at the time (whose name may be familiar to the most careful readers of this blog as I described some of his subsequent quirky research), had just finished constructing a PRC using the “standard” Aschoff’s Type I method, this was a good opportunity for me to compare the methods head-to-head and see if they are equivalent.

Is circadian clock involved in photoperiodism?

Involvement of the circadian clock in the measurement of the seasonal changes in daylength has been definitively demnstrated in hundreds of species of vertebrates, invertebrates, fungi and plants. Three species, notably, were difficult in this respect and required much more work and much more creative work than expected to test this proposition.

Linda Hyde in our lab had just shown the previous year that one of those three species – the American chameleon (Anolis carolinensis) – also used the circadian system to measure photoperiod.

The Vetch aphid (Megoura viciae) was the second tough hold-out, but a nifty new mathematical model was published at that time which accounted for all the experimental data and apparently persuaded most people that this arthropod does, indeed, use the circadian clock to measure daylength.

This left the Japanese quail as the sole mysterious species yet to be shown to use the clock to regulate seasonal events. This is somewhat ironic, as this species has been used for studies of photoperiodism for almost a century!

The night-break (“skeleton photoperiod”) experiments by Tom Siopes were positive, but Resonance cycles done in Brian Follett’s lab were negative. It was high time for somebody to finally apply T-cycles, the most powerful experimental protocol for dealing with these questions, to the quail. It was not already done before mainly because it requires a lot of blood-sweat-tears, a lot of animals, and a lot of time. I had all three, so I got to work on this.

Distinction between External and Internal Concidence Models

As explained in my previous post this morning, the distinction between the two models of photoperiodism was not experimentally shown yet. The data from my first set of experiments (see this Tuesday morning’s post) gave us a hope that, having a better grasp on where exactly the two oscillators were located, we could possibly demonstrate that the Internal Coincidence Model was applicable at least in this species.

Related to this, we could also take a better look at the “internal coincidence” of two oscillators that determines the timing of ovulation and oviposition, a question we were excited about right after I finished my previous project. Use of T-cycles sounded like an excellent approach to this question.

So, I got to work and I acquired some great electrician’s skills, figuring out how to deliver light to each animal at an appropriate time outside of the 24-hour realm. What did we find?

Phase-Response Curve

Kent Edmonds constructed the PRC using the Aschoff Type I method. He applied 6-hour long (100lux) light pulses to quail kept in prolonged constant darkness (DD). The pulses elicited large phase-shifts in the body-temperature circadian rhythm (black is time when temperature is above daily mean, white when it is below, triangle denotes the beginning of the 6-hour pulse):
i-769f17f68a4594dd588ec69267ad5cfe-a1 - phaseshift.JPG

When the data were plotted, we saw a Type 0 Phase-Response Curve, which is quite unusual for vertebrate animals, especially with such a weak stimulus. It is as if a very cloudy, dark day in the middle of the winter completely and almost immediatelly reset the circadian system. No transients were obsaerved so the direction (advance or delay) was impossible to judge. In such cases all phase-shifts are plotted as phase-delays (bottom) and can be replotted as a Transient-Response Curve to show that every light-pulse resulted in a phase-shift to the same new phase, roughly around CT0 (bottom):
i-b51efe8466caaca3881d8cbc72536a51-a2 - PRC.JPG

I exposed cohorts of birds to T-cycles in which the light portion of the cycle was always a 6-hour (100lux) long pulse. In cycles in which T (Light + Dark) was longer than 24 hours (LD6:20, LD6:22, LD6:24, LD6:26 and LD6:30), the circadian rhythm of body temperature assumed a positive phase-angle in relation to the onset of the light pulse, as expected from theory, i.e., it preceded the onset of light (actographs folded at T, not at 24hours, i.e., the x-axis is not 2x24h, but 2xT):
i-11e6cb7bd6fb86b88c4c9d2cf6ecd517-a3 - long T.JPG

In cycles in which T (L + D) was shorter than 24 hours (LD6:12, LD6:14 and LD6:16), the phase-angles were negative (i.e., the onset of light preceded the beginning of the internal circadian cycle). Since light exerts masking effect on body temperature, I unmasked the phase-angles by releasing all the birds into DD (folded at T):
i-1918698715cd8f026cd2d950572e61ea-a4 - short T.JPG

When I plotted both the measured phase-angles and the calculated (from the PRC) phase-angles together, there was an excellent correspondence between the two, suggesting that Aschoff’s Type I and Type VI methods are indeed equivalent (black bars from T-cycles, grey bars from PRC):
i-b41026c8968c8cca9a8ef20b1070ed8f-a6 - model.JPG

Photoperiodic response

If you look at the previous image, in some of the T-cycles the light-pulse illuminated the subjective day in which no reproductive response is expected, while other light-pulses illuminated different portions of the subjective night, during which it was expected that the stimulation during the Photoinducible Phase would result in reproductive stimulation.

However, none of the T-cycles were reproductively stimulating. Birds introduced into the experiment while sexually regressed remained so throughout the experiment. Birds introduced into the experiment while sexually mature quickly stopped laying eggs and had completely regressed ovaries at the end of the experiment.

So, I did another experiment. I exposed cohorts of birds to T-cycles in which the light-portion of the cycle was 14 hours long (LD14:4, LD14:6, LD14:8, LD14:12, LD14:14, LD14:16, LD14:18). The birds introduced into the experiment while laying eggs continued to lay no matter which cycle they were exposed to. Most of the birds introduced into the experiment while sexually regressed were stimulated by every one of these cycles and laid eggs by the end of the experiment (folded at T; circles mark times of egg-laying).
i-d04296b313b227bfe22566bff336c826-a5 - T14.JPG

So, the results were negative – not just that there was no way to distinguish between Internal and External Coincidence models, but even the role of the circadian clock in photoperiodic time measurement suffered a severe blow. After all, the T-cycle experiment is supposed to be the most powerful protocol for testing this hypothesis.

But, there was something in the data that was telling. The positive phase-angles in birds exposed to T-cycles with 14-hours pulses were much smaller than phase-angles in birds exposed to T-cycles with 6-hour pulses with the equal total length of T. This (remember the formula – &tau – T = &Delta &Phi i.e., the phase-angle is equal to the difference between the period of the circadian clock and the period of the entraining cycle) suggested that the inherent period of the circadian rhythm was different in sexually regressed birds (all in 6-hour pulse cycles) and sexually mature birds (all in 14-hour pulse cycles).

We knew from previous work that the reproductive state in females and injections of testosterone into males both induce changes in the period quite dramatically.

So, in cycles in which the pulse was 14 hours, the duration of light was always long enough to, at least partially, illuminate the Photoiducible Phase (External Conicidence Model) or to place two oscillators into a stimulatory phase (Internal Coincidence Model).

On the other hand, with 6-hour pulses, some cycles were not inductive and remained so throughout the duration of experiment. The other cycles may have been initially inductive, but the gradual growth of ovaries and release of sex steroids resulted in the change of period. This, in turn, changed the phase-relationship between the circadian rhythm of photoperiodic photosensitivity and the light-cycle (External Coincidence Model) or the phase-relationship between two oscillators (one responsive, the other non-responsive to steroids; Internal Coincidence Model), resulting in cessation of stimulation and subsequent regression of ovaries.

Perhaps an intermediate duration of the light-pulse (e.g,. 10-12 hours) in a future T-cycle experiment will lead to a positive result – some cycles stimulatory and some not – as predicted from the theory. Several additional possible explanations were explored in the Discussion section of the paper – a good read, not just because I wrote it!

Timing of ovulation

Birds exposed to T-cycles with daily pulses of 14 hours were all laying eggs. Cycles with T of 22 hours or shorter resulted in an almost unfettered freerun of the ovulatory rhythm throughout the light-dark cycle, suggesting that in such short cycles the permissive phase is virtually spanning the entire cycle.

In cycles with T of 28 hours or greater, the ovulatory pattern exhibited clutching in a way we predicted beforehand.

In LD14:12 (T=26 hours), the ovulatory rhythm was fully entrained to the light-dark cycle. The strength of coupling between the two oscillators was stronger than that seen on 24-hour cycles, suggesting that the two-oscillator interaction has evolved to induce a clutching pattern, thus to control the size (total egg-number) of the clutch in this species.

Additional observations

While I was doing this, a paper came out from Brian Follett’s lab in the UK, in which his student exposed quail to a resonance cycle of LD6:30. While the cycle was not sexually stimulatory, the interpretation of their results was not satisfactory from our perspective. That is why I also added LD6:30 to my experiment. We exchanged several e-mails and talked in person with Follett and he urged us to go ahead and publish the disagreement with their findings. He was shutting down the lab (after accepting a position of the President of the University of Warwick) and had no chance to follow up on the work himself.

In their paper they argued that the circadian clock in quail acts like an hourglass clock. Our data suggested that the circadian clock in quail is a low-amplitude (Type 0 PRC), but self-sustainable oscillator.

The key difference between their work and our work was that they monitored locomotor activity as a measure of the phase of the rhythm. In our lab, we always monitor both the body temperature rhythm and the locomotor activity rhythm simultaneously. We almost alway use only the temperature data in our analysis because the activity rhythm in this species is messy and difficult to apply statistics to, while the temperature rhythm is clean and robust, usually providing clear-cut onsets and offsets of the rhythm. Actually, a French laboratory showed that it takes only about two generations of selective breeding to completely abolish a rhythm in behavior in quail, while physiological rhythms persist.

When I compared my LD6:30 data – both temperature and activity records – to their LD6:30 data, one thing became apparent. No positive phase-angles were seen in activity records from both labs. Yet, the positive phase-angles were clearly visible and easily measurable in body temperature records (folded at 2 x 36 hours; top is body temperature, middle shows locomotor activity above the daily mean, bottom shows all locomotor activity):
i-d32af089d67f08cbd69138d64b7f2228-a7 - LD6-30.JPG

Because they could not see anything but masking effects of light on activity, they deduced that the clock stopped and had to be restarted with every light-pulse. Since we saw the motion of the pacemaker independent from the masking effects of light on temperature, we could see that the clock is not stopping – it is entrained (not restarted) by light in each cycle.

Furthermore, their assumption that LD6:30 is a Resonance Cycle was erroneous as well. If it was a resonance cycle, the first light pulse would fall in the day, the second in the night, the third in the day, and so on. What we saw is that the quail’s clock fully entrained to the 36h cycle – it strecthed to accomodate such a long cycle in a way. Thus, each successive light pulse fell on the SAME phase of the rhythm, not two alternating phases. Thus, LD6:30 is a T-cycle, not a Resonance cycle.

This finding, coupled with the rarely-seen-in-vertebrates Type 0 PRC, showed that quail’s circadian clock had a very broad range of entrainment. It can, almost immediately, phase-advance at least 4.5 hours or phase-delay at least 13.5 hours. The quail CANNOT get jet-lagged (perhaps because they are long-distance migrants out in the field).


It does not sound very humble, but if I may say so, this is a very good paper – my best so far. Sadly, it is also one of the last papers in the field done and written in the tradition of “classical chronobiology”, kind of stuff that Pittendrigh would publish.

Since then, we in the field, myself included, have gradually moved away from this kind of research. Each experiment takes forever. It takes a lot of work. It takes a lot of animals. It sometimes requires invasive surgery. Thinking in terms of multiple interacting oscillators tends to make one’s brain hurt! So many reasons to move to something easier (and easier to get funding for).

Most people went molecular. Some of the elderly statesmen of the field could, at this stage of their careers, afford to move to ecological and evolutionary questions.

Since my Master’s defense (which included this work and this work), I initially did some more stuff in this mold, following up on some of the circadian and photoperiodic stuff. This will be included in my PhD Dissertation (about a quarter of it), although it looks a little out of place in a chapter of its own at the end, not really fitting well with the rest of the narrative.

I am not telling you any details until the stuff is published, but I can tell you that about another quarter of my work went down the levels to molecular and cellular neuroendocrinology. The rest, about half of the Thesis and the main story (also the most novel, creative and insightful part of it), went up the levels – looking at ontogenetic, behavioral, ecological and evolutionary questions.

Those experiments were faster and easier to do, required smaller numbers of animals and almost no invasive surgery – something that is more and more difficult to do these days both because of tightening IACUC rules and because it is hard for us to persuade ourselves that much of it is necessary any more. The approach was very fruitfull over the decades, but now it suffers from the effect of diminishing returns. It had to be done at the time when it revolutionized our understanding of circadian systems, less so now when it may just add new detail. Such experiments used to eliminate five out of ten alternative hypotheses, now more like one out of three – often not sufficiently important to actually decide to perform the study.

In any way, this paper in particular, as well as the stuff I described here and here (plus a review and a paper I’ll discuss tomorrow), helped me establish myself as a classical chronobiologist. It gives me authority to move on to other areas and other approaches without anyone questioning my credentials in the field (which some later entrants into the field may sometimes suffer, especially if they have not received classical training but entered horizontally from genetics or other fields).

What I need to do now is defend my Dissertation and try to get my foot back in the lab somewhere. It is not easy to get back in after beeing two years out of the world of science. During this time I taught biology and blogged about it so my brain is certainly not rusty, but in a competitive world, a two-year break looks suspicious.

On the other hand, I picked the best time to take a break. I was seriously burned out and needed a break just at the time when funding plummetted and everyone got nervous (and stingy with money). It is like science froze for a couple of years. Everyone is now adapting to the new funding situation and moving on – the work needs to be done. Even with a long break on my CV it may be easier for me to get back into research now than two years ago. And, unlike two years ago, today I am refreshed and raring to go (and semi-famous in the science blogosphere which may help). We’ll see what happens. I’ll let you know when it does.

BD Zivkovic, H Underwood, CT Steele, K Edmonds, Formal Properties of the Circadian and Photoperiodic Systems of Japanese Quail: Phase Response Curve and Effects of T-Cycles, Journal of Biological Rhythms, Vol. 14, No. 5, 378-390 (1999)


  1. #1 greensmile
    April 22, 2008

    whoa, this is interesting…I gotta go back and read the others now.

    BTW, Sci Am is getting on your turf a little: