Normal vision can compensate for the loss of the circadian clock

Circadian clocks are thought to be essential for timing the daily activity of animals, and consequently increase fitness. This view was recently challenged for clock-less fruit flies and mice that exhibited astonishingly normal activity rhythms under outdoor conditions. Compensatory mechanisms appear to enable even clock mutants to live a normal life in nature. Here, we show that gradual daily increases/decreases of light in the laboratory suffice to provoke normally timed sharp morning (M) and evening (E) activity peaks in clock-less flies. We also show that the compound eyes, but not Cryptochrome (CRY), mediate the precise timing of M and E peaks under natural-like conditions, as CRY-less flies do and eyeless flies do not show these sharp peaks independently of a functional clock. Nevertheless, the circadian clock appears critical for anticipating dusk, as well as for inhibiting sharp activity peaks during midnight. Clock-less flies only increase E activity after dusk and not before the beginning of dusk, and respond strongly to twilight exposure in the middle of the night. Furthermore, the circadian clock responds to natural-like light cycles, by slightly broadening Timeless (TIM) abundance in the clock neurons, and this effect is mediated by CRY.

system [10]. During dark nights, flies appear to cover only smaller distances, which appears reasonable since they cannot see anything. In the commercial Trikinetics system used by most laboratories, even such small movements are registered, which might be explained by the small tube volume that allows IR beam crossings with little activity. Indeed, even at 20°C, flies recorded by the Trikinetics system show anticipatory M activity [10].
We conclude that M anticipatory activity depends on environmental temperature as well as on the recording system and is not an adequate criterion for clock-controlled activity (see also discussion in [10]).

Do per 01 and tim 01 mutants have a residual clock that facilitates wild-type activity patterns under more natural-like conditions?
The next question to be answered is why the clock mutants per 01 and tim 01 have even more anticipation of lights-on than WT flies (Fig. 1). If the ability to anticipate lights-on is taken as proof for an endogenous clock, the mutants must at least have a working M oscillator. Indeed, there are old observations that point to a residual M oscillator in per 01 mutants [11][12].
According to these findings the M oscillator of per 01 mutants is not strong enough to drive locomotor activity rhythms under constant conditions but it is strong enough to promote activity before lights-on in the morning; moreover, M-activity-onset depends on the Zeitgeber period being earlier under long periods [12]. WT flies show the same dependency of M-activity-onset on Zeitgeber period, but they always begin activity ~2h after per 01 mutants. This result fits nicely to the here observed stronger anticipation of lights-on in per 01 mutants. Several other observations point to the presence of a residual clock in per 01 and tim 01 mutants: individual mutant flies showed bimodal activity with a real siesta and other individuals exhibited even rhythmic activity for some days under constant conditions [4,[12][13][14][15]. In the present study we could also see bimodal activity in some tim 01 mutants under LD and this is reflected in the average activity profile shown in Figure 2A. per 01 ;cry b double mutants have even been reported to show more robust bimodal activity patterns under LD conditions than per 01 single mutants, perhaps because the absence of functional CRY facilitates photoreceptor input from the compound eyes, thereby provoking a more clear-cut synchronization of the residual oscillator [14].
The molecular basis of the residual oscillator is not completely clear. It is possible that the per and tim genes, respectively, have some residual function in the absence of one of them (see [14]). However, even per 01 ;tim 01 mutants show some residual rhythmicity [4], suggesting that the second molecular feed-back loop comprising vrille and Pdp1 (reviewed in [16]) can still show some rhythmic activity in absence of per and tim. The Clk gene is central to both feedback loops [16]. Therefore, Clk JRK mutants should not show any rhythmicity. This is indeed the case [4]. Thus, it is well conceivable that the residual clocks of per 01 and tim 01 mutants may have contributed to their quasi-normal activity under LDR2 and outdoor conditions [3]. Most importantly, twilight simulations considerably improve synchronization of several species to light cycles [16][17][18][19][20][21][22][23][24][25]. Not only the precision of the synchronized rhythm is higher, but also reentrainment to shifted LD-cycles occurs faster and the animals can adapt better to unusually long photoperiods and Zeitgeber cycles with different periods. Twilight may therefore have improved rhythmic output in animals with a weak residual clock such as per 01 and tim 01 mutants.

Supplementary Figures
Figure S1