Oscillator control of cell division in Euglena: cyclic AMP oscillations mediate the phasing of the cell division cycle by the circadian clock

Cells that grow with a generation time greater than 24 h often exhibit circadian rhythms of cell division. Typically, mitosis occurs at a certain phase of the circadian cycle, at times (subjective nights) frequently corresponding to the dark intervals in a synchronizing LD:12,12 cycle (the alternance of 12 h of light and 12 h of darkness). This phe-nomenon, which has been described in unicellular algae, fungi and protozoa, as well as in mammalian cells in vivo (Edmunds and Laval-Martin, 1984; Edmunds, 1988), is thought to reflect an interaction between an autonomous cir-cadian oscillator and the cell division cycle (CDC).

Circadian cell division rhythmicity in the algal flagellate Euglena gracilis has been studied extensively in our labo-ratory (Edmunds and Laval-Martin, 1984). We are inter-ested in elucidating the biochemical basis of the circadian clock and the signal transduction pathway(s) that mediate(s) the control of the cell division cycle by this oscillator. In order to circumvent the dual use of light (for photosynthe-sis, and as a time cue for the circadian clock), we have made use of an achlorophyllous strain (the ZC mutant) that has been shown to exhibit cell division rhythms that per-sist in constant darkness (DD) for up to 7 days with a period (τ) that only approximates 24 h (Carré et al., 1989a).

We have hypothezised that second messengers, which control many cellular activities, may play a role in the phas-ing of mitosis and of other processes by the circadian oscil-lator. We found (Carré et al., 1989b) that the cyclic AMP level in the ZC mutant exhibits bimodal circadian oscilla-tions, which are independent of cell cycle progression (since they are also observed in nondividing cells), and which per-sist in DD with a period similar to that of the cell division rhythm. Peak cyclic AMP levels occurred at circadian time (CT) 00-02 (at a time when most cells were in the G₁ phase of the CDC) and at CT 12-14 (corresponding to the onset of mitosis). Transient surges of cyclic AMP levels, which were correlated with the initiation of DNA synthesis and with the onset of mitosis, have also been observed during the division cycle of Euglena cells that had been blocked in S phase in the absence of vitamin B₁₂, then released syn-chronously from this restriction point following the addi-tion of the vitamin (Carell and Deardfield, 1982).

The role of cyclic AMP in the CDC is a subject of con-troversy, since exogenous cyclic AMP has been found to stimulate mitosis in some cell types, and to inhibit it in others (Dumont et al., 1989). Interestingly, however, tran-sient increases of the cyclic AMP level are correlated with cell cycle transitions at both the G₁/S and the G₂/M bound-aries in many cell types (for reviews, see Boynton and Whitfield, 1983; Whitfield et al., 1985, 1987). In Saccharomyces cerevisiae, genetic evidence indicates that cyclic AMP is essential for cells to initiate DNA synthesis (Mat-sumoto et al., 1983). Cyclic AMP also has been found to inhibit MPF (mitosis promoting factor), an activity that is found in the cytoplasm of mitotic cells and is characterized by its ability to induce nuclear envelope breakdown and cell division upon injection into maturing Xenopus oocytes (Maller, 1985).

On the basis of this evidence, we hypothesized that cyclic AMP may play a role in the cell cycle of Euglena, to reg-ulate the transition through the G₁/S or the G₂/M bound-aries, or both. Thus, periodic cyclic AMP signals, under the control of a circadian oscillator, may act to initiate certain portions of the CDC at specific phases of the circadian cycle. To test for this possibility, we examined the effect of exogenous cyclic AMP signals, and of agents that per-manently increased the cellular cyclic AMP level, on the cell division rhythm in the ZC mutant.

We show that exogenous cyclic AMP perturbs cell cycle progression in a manner that depends on the circadian time of its addition. Cyclic AMP appears to act downstream from the circadian oscillator, in the output pathway for the con-trol of the CDC, since in most experiments the cell divi-sion rhythm quickly returned to its original phase (before the perturbation). Finally, we investigated the role of the two cyclic AMP-dependent kinases previously identified in the ZC mutant of Euglena (Carré and Edmunds, 1992), using cyclic AMP analogs that have been shown to specif-ically activate either of the two kinases.

The achlorophyllous ZC mutant of Euglena gracilis Klebs (strain Z) was obtained from Dr R. Calvayrac (Laboratoire des Mem-branes Biologiques, Université Paris VII, France). It was derived from the wild-type strain by action of 2.5×10⁻⁵ M diuron (DCMU) in a 33 mM lactate medium (pH 3.5) under illumination and anoxia (Calvayrac and Ledoigt, 1976).

Axenic, aerated, magnetically stirred 4 l batch cultures were grown at 16.5(±0.5)°C in environmental chambers, on a modified Cramer and Myers’ (1952) medium supplemented with vitamins B₁ and B₁₂ (as previously described by Edmunds and Funch, 1969), and containing ethanol (0.1 to 0.4%, v/v) as a carbon source. Cysteine and methionine (10⁻⁵ M), were added to the experimental cultures in order to improve cell division rhythmic-ity. Sulfur-containing coumpounds have been shown to allow a better coupling between the CDC and the underlying circadian oscillator in other chloroplastidic mutants of E. gracilis (Edmunds et al., 1976). Illumination (3000 lx) was provided by clock-pro-grammed, cool-white fluorescent bulbs. Cell number was moni-tored every 2 h by a miniaturized fraction collector and a Coul-ter Electronic Particle Counter (Edmunds, 1964).

Cultures growing in the infradian mode (g >24 h) were obtained at 16.5(±0.5)°C. Cell division could then be entrained to a 24 h period by imposition of LD:12,12. Typically, the cells divided during the dark intervals, and the onset of mitosis occurred at the onset of darkness (12 h after light onset). The circadian rhythm of cell division persisted after transfer of the culture to DD, with a period (τ) of 25±2 h (Carré et al., 1989a). The onsets of cell division were taken as phase reference points and were consid-ered to fall at CT 12, which corresponds to the onset of darkness in a LD:12,12 reference cycle.

Cultures that had been pre-entrained in LD:12,12 were placed into DD. These ‘free-running’ cultures were perturbed by the addition of cyclic AMP (or of other drugs) to the culture medium at different CTs (corresponding to different phase points of the CDC) during the second day in DD. Advances or delays of the CDC were measured as the difference between the onset of the subse-quent cell division step and the predicted one, projected from the last unperturbed cycle with a period of 25 h. The steady-state phase shifts (Δϕ), indicative of possible perturbations of the cir-cadian oscillator itself, were also measured after transients had subsided (in 2 or 3 days). Phase advances were arbitrarily desig-nated +Δϕ and phase delays −Δϕ (Pittendrigh, 1965). In all cases, both CT and data for Δϕ were normalized to 24 h in order to facil-itate comparison among cell cultures which had different free-run-ning periods (τ). A plot of the Δϕ (if any) engendered by a cyclic AMP signal as a function of the CT at which the pulse was admin-istered yielded a phase-response curve for the rhythm of cell divi-sion (see Edmunds et al., 1982).

For each time point, approximately 2×10⁶ cells were harvested by centrifugation (3000 g for 10 min) and resuspended in 5 ml of 70% ethanol. Cells were washed once in 5 ml of 70% ethanol and stored in the same medium for up to 10 days at 4°C. On the day of the assay, cells were treated with RNAse A and stained with propidium iodide (50 μg/ml), as described by Yee and Bartholomew (1988, 1989). The cell suspension was then filtered through Nytex screens (60 μm pores) and analyzed with a Becton Dickinson FACS flow cytometer. Total cellular DNA content also was estimated colorimetrically by the diphenylamine reaction (Burton, 1955).

Approximately 10×10⁶ cells were harvested by centrifugation (10 min, 7700 g) of the cell suspension. The pellet was resuspended in 0.7 ml of distilled H₂O, and was extracted in 7.5% (final con-centration, w/v) TCA for 20 min at 4°C. Pelletable material was eliminated after centrifugation (10 min, 39000 g). The supernatant was then extracted 5 times with an equal volume of water-satu-rated diethylether. The remaining ether was boiled off at 80-90°C until bubbling stopped. The extracts were frozen in liquid nitro-gen and were stored at −70°C until the day measurements were performed.

The amounts of cyclic AMP were measured by a competitive protein-binding assay, modified from Gilman (1970) and Døske-land and Ogreid (1988). Assays were carried out in test tubes kept in iced water, into which were successively injected: 50 μl of cell extracts (or of standard solutions of cyclic AMP), then 100 μl of an incubation mixture containing 40 μg of protein kinase inhibitor (Sigma) and 3 pmol (60000 d.p.m.) of [³H]cyclic AMP (Amer-sham) in sodium acetate (50 mM, pH 4.0). The binding reaction was started by the addition of either 50 μl of a protein mixture (containing approximately 10 μg of protein kinase (Sigma) and 40 μg BSA) or of 50 μl H₂O (for the measurement of nonspecific binding). The samples were incubated at 0°C for 90 min, and the reaction was stopped by dilution with 1 ml of 80% saturated ammonium sulfate and 0.05 mM Hepes (pH 7.0). The reaction mixture was filtered through cellulose ester Millipore filters (HAWP025). The filters were rinsed three times with 3 ml of 60% saturated ammonium sulfate and 0.05 mM Hepes (pH 7.0) and then were transferred to scintillation vials. 1 ml of 2% SDS was added in order to solubilize the proteins bound to the filters, prior to the addition of scintillation fluid (Scintiverse, Fisher) and count-ing for 10 min in a Packard Tri-Carb liquid scintillation spec-trometer (Model 3320). The cyclic AMP concentrations per 10⁶ cells were extrapolated from a standard curve that was obtained simultaneously by exactly the same procedure applied to known quantities of cyclic AMP (ranging from 0.5 to 20 pmol).

To facilitate the understanding of the relationship between circadian time (CT) of the rhythm of cell division in a pop-ulation of Euglena and the discrete phases of the cell divi-sion cycle (CDC), we show (Fig. 1) as a reference one division step of a culture in LD:12,12 (see Edmunds, 1964). The plateau of the growth curve lasts for approximately 10 h (from CT 02 to CT 12). Cellular DNA content starts to increase earlier in the day (approximately CT 04). The rate of DNA synthesis reaches a maximum at (or just before) the beginning of the night (CT 10-12). The onset of cell division in the culture corresponds approximately to CT 12. The maximum division rate is observed at the end of the night (CT 18 to CT 24), in parallel with a sharp drop in cellular DNA content. After the cultures are transferred to constant darkness (DD), CTs are determined by projection from the onset of cell division, taken as a phase-reference point for CT 12.

To ask whether the periodic cyclic AMP signals generated by the circadian oscillator may play a role in phasing cell division, we assayed for the effect of exogenous cyclic AMP signals on the timing of the subsequent cell division steps in cultures of the ZC mutant that were dividing rhythmically in DD with a period (τ) of 25 h. The incubation the of ZC mutant for 1 h in the presence of exogenous cyclic AMP increased the cellular cyclic AMP content in a concentration-dependent manner (not shown). We chose to test first the effect of 250 μM cyclic AMP, which after 1 h increased cellular cyclic AMP concentration from 20 to approximately 60 pmol per 10⁶ cells. We found that cyclic AMP added to synchronously dividing cultures induced advances or delays of the following burst of cell division, depending on the CT at which the drug was given. Four typical experiments are shown in Fig. 2. The injection of cyclic AMP at CT 12 caused a 5 h delay of the subsequent division burst (Fig. 2A). Although cyclic AMP was added in a permanent manner (tonically), its effect on the cell division rhythm was only transient. The following period was shortened, bringing the rhythm back to its original phase, an indication that the underlying circadian oscillator had not been perturbed by the cyclic AMP signal. The same experiment repeated at CT 13.9 did not induce any perturbation, even transiently, of the cell division rhythm (Fig. 2B). Injection of cyclic AMP between CT 18 and CT 22 had dramatic effects on the oscillation. The plateau following the cyclic AMP signal was shortened (indicating the advanced division of some of the cells); the pattern of cell division then became arrythmic (Fig. 2C,D). In two experiments only, rhythmicity was restored after 36 h (as in Fig. 2C,D), and the cell division rhythm appeared to have been phase-shifted by 10 to 11 h.

The free-running rhythm of cell division exhibited by the ZC mutant was systematically scanned by cyclic AMP signals, given at different CTs during the second free-running period in DD. Advances or delays of the cell division steps were measured as described for Fig. 2 (see Materials and Methods). From these data, two phase-response curves were generated, for (a) the immediate perturbations of the first division step following the addition of cyclic AMP (which reflect the effect of cyclic AMP on cell cycle progression), and (b) the steady-state effects of cyclic AMP on the cell division rhythm (which might indicate perturbations of the underlying circadian timer).

Fig. 3A shows the phase-response curve for the perturbation of the cell cycle by cyclic AMP in the ZC mutant. Cyclic AMP injected between CT 06 and CT 12 caused delays (4 to 6 h) of the following burst of cell division, whereas the same signal given between CT 18 and CT 22 induced some of the cells to divide before the following CT 12 time-point. The advance was estimated to be as great as 8 h. Cyclic AMP had no effect when injected at CT 00-02 and at CT 12-14 (when the endogenous cyclic AMP level is at its peak). These results indicate that cyclic AMP signals may modulate cell cycle progression in Euglena.

To test for a possible role of cyclic AMP as an element or ‘gear’ of the circadian oscillator, we also looked for permanent phase-shifting effects of cyclic AMP on the cell division rhythm. Fig. 3B shows the steady-state resetting effects of cyclic AMP on the cell division rhythm. No significant lasting phase shifts were observed at any circadian time (with the exception that in two experiments apparent phase delays of 10 to 12 h surprisingly were elicited by cyclic AMP administered at approximately CT 20; see Fig. 2C,D). Similarly, no steady-state phase shifts were obtained when synchronously dividing cultures of the ZC mutant were exposed to 500 μM cyclic AMP for 1 h, then diluted 50-fold, thus reducing the cyclic AMP concentration down to 10 μM (Fig. 3B, open symbols).

In order to determine whether the perturbations seen on the growth curve reflected real changes in the rates of cell cycle transitions, rather than changes in cell motility or settling, we measured cellular DNA content in control Euglena cultures and in cultures that had been perturbed by cyclic AMP, using flow cytometry techniques.

A typical DNA histogram from a nonsynchronized culture of the ZC mutant growing exponentially in constant light is shown in Fig. 4A. Approximately 41% of the cells had a single (1C) DNA content and constituted the G₀/G₁population. An estimated 37% of the cells had a double (2C) DNA content and thus were either in G₂ or in mitosis. Cells having an intermediate DNA content (approximately 22%) constituted the S phase population. The mean fluorescence of the G₂ peak was only 184% of that of the G₁ peak, due to the poor accessibility of DNA for dye in G₂ cells, as previously described in Euglena (Bonaly et al., 1987). The high percentage of cells with a 2C DNA content (normally half of that of the percentage of cells in G₁ in rapidly cycling cultures) suggested a blockage point in G₂. The existence of such an arrest point was confirmed by examining the DNA content in cells from stationary phase cultures (Fig. 4B). No cells were found with an intermediate DNA content, a finding that indicated that cells had stopped cycling. Cells were found arrested with either a 1C or a 2C DNA content (59% and 41%, respectively). Thus, in Euglena, major cell cycle control points appear to operate in both the G₁ and the G₂ phases of the CDC.

To test for the effect of cyclic AMP on cell cycle progression, cultures of the ZC mutant were synchronized by LD:12,12 and then were transferred to DD. At the time of the experiment, the cultures were divided into two halves, one of which received cyclic AMP or a cyclic AMP analog, while the other half was used as a control. Fig. 5A shows the effect of cyclic AMP (1 μM) injected at CT 06 (plateau of the growth curve) on cell cycle progression. In the control culture, from t=0 to t=12 h, the G₁ peak decreased, while the G₂/M peak increased, an observation indicating the progression of cells from G₁ into S and G₂. By t=22 h (CT 04), the G₁ peak was markedly increased, a finding that indicated that a large portion of the cell population had just completed mitosis. We noticed the slower accumulation of cells with a 2C DNA content (visible from t=4 to t=12) in the presence of cyclic AMP (1 μM, added at CT 06), results that were indicative of an inhibition of the progression of cells through S phase. By t=22 h, most cells had left G₁ and were in late S phase or in G₂, at a time when cells from the control experiment were already going back to the G₁ phase. Thus, the delay of cell division observed in growth curves after the addition of cyclic AMP at CT 06-08 was due to an inhibition of DNA synthesis. In the experiment shown in Fig. 5A, we were unable to detect any effect of cyclic AMP on the progession of cells through mitosis. In another experiment (not shown), however, in which 8-BZA-cAMP (8-benzylamino-cyclic AMP) was used, we observed the accumulation of cells with a fully replicated DNA, a finding that indicated an inhibitory effect of 8-BZA-cAMP on the G₂/M transition.

Fig. 5B shows the results of a similar experiment performed at CT 20 (in the middle of the division phase). In the control culture, the G₁ peak increased from t=0 to t=8 h, an observation demonstrating that cells proceeded through mitosis. At the same time, cells began to escape from G₁ into S phase. From t=12 h to t=22 h, the accumulation of cells with a 2C DNA content was found. The effect of 1 μM cyclic AMP injected at CT 20 was clear 12 h after the cyclic AMP signal: the proportion of cells with a 2C DNA content was markedly decreased as compared to the control experiment. The progression of cells through G₁ and S phases, however, did not seem to be affected since the proportion of cells in early S phase increased normally between hour 8 and hour 12. Thus, the decreased cell population with a double DNA content was due to a shortening of the G₂ phase rather than to a blockage in an earlier phase of the CDC. This observation was consistent with the advance of cell division observed in growth curves (Fig. 2C,D).

We have demonstrated that exogenous cyclic AMP signals cause lengthenings or accelerations of the CDC, depending on the CTs when the drug is added. We still needed to determine whether the endogenous circadian variations of cyclic AMP levels have sufficient amplitude to have similar effects on cell cycle progression. Our approach was to test the effect of drugs that may reduce the amplitude of the cyclic AMP oscillation, or that keep cyclic AMP at a level such that all cyclic AMP receptors should be saturated in a permanent manner. We would expect such drugs to prevent the expression of cell division rhythmicity.

Forskolin, which has been shown to stimulate adenylate cyclase in Euglena at times when the activity of the enzyme is minimal (Tong et al., 1991), would be expected to have such an effect on the cyclic AMP oscillation. When forskolin was injected into a culture of the ZC mutant that was maintained in LD:12,12, the cyclic AMP level was increased by as much as 3-fold, as compared to an untreated control (Fig. 6). Maximum increases of cyclic AMP level corresponded to times when cyclic AMP in the control culture was minimum; forskolin had little effect at times when cyclic AMP was at its maximum. As a result, the cyclic AMP level in the presence of forskolin never exceeded the range of cyclic AMP concentrations in control cultures, but oscillated with a reduced amplitude, and at a higher level. When forskolin was added to a free-running culture of the ZC mutant (previously synchronized by LD:12,12) at the time of its transfer to DD (at approximately CT 12), we observed a rapid desynchronization of the cell population (Fig. 7A). The cells appeared to go through one synchronous division step, then resumed nearly exponential growth. Very low-amplitude waves of cell division were still visible, however, for a minimum of 3 more days, a sign that that the circadian timer was running unaffected. Such a rapid loss of cell division rhythmicity sometimes occurs after transfer of control cultures to DD, but heretofore has never been observed in cultures that had been maintained in 24-h LD cycles. When forskolin was added to a culture that was entrained by LD:12,12, a rapid damping of the cell division rhythm was also observed (Fig. 7B). This effect was not phase-dependent: similar results were obtained when the drug was added at CT 12 (Fig. 7A,B) or at CT 04, 08, 15, or 00 (not shown). These results suggest that the amplitude of the cyclic AMP oscillations determine the amplitude of the cell division rhythm.

We have demonstrated earlier (Carré and Edmunds, 1992) that two cyclic AMP-binding proteins are present in the ZC mutant of Euglena, which comigrated with two peaks of cyclic AMP-dependent kinase activity (cPKA and cPKB) during DEAE-cellulose chromatography. The distinct cyclic AMP analog-specificity of the two cyclic AMP-dependent kinases identified in Euglena extracts enabled us to ask whether only one or both of these enzymes play a role in cell cycle control. Thus, we tested the effect of different concentrations of cyclic AMP, 8-BZA-cAMP, 8-CPT-cAMP (8-(4-chlorophenylthio)-cyclic AMP), and 6-MBT-cAMP, (N⁶-monobutyryl-cyclic AMP), at times (CT 06-08 and CT 18-20) when 250 μM cyclic AMP induced maximum perturbations of the cell division rhythm.

We found that a 2-to 5-fold increase in cyclic AMP or in cyclic AMP analog concentration was sufficient to elicit maximum perturbation of division rhythmicity. Table 1 gives the minimum concentrations of cyclic AMP and of cyclic AMP analogs that induced advances or delays of cell cycle progression that were comparable to those previously obtained with 250 μM cyclic AMP. Surprisingly, different results were obtained at CT 06-08 and at CT 18-20. Thus, 8-CPT-cAMP, which specifically activated cPKB in vitro (Carré and Edmunds, 1992), had no effect on cell cycle progression when given at CT 06-08, up to a 1 μM concentration; in contrast, very low concentrations (0.1 nM) of the same analog were able to accelerate mitosis when given at CT 18-20. Furthermore, 8-BZA-cAMP (which specifically activates cPKA; Carré and Edmunds, 1992) caused large delays of cell division when added to the culture at CT 06-08 at a 100 nM concentration, but was ineffective up to a 1 μM concentration at CT 18-20. Finally, the minimum concentrations of cyclic AMP and cyclic AMP analogs that caused perturbations of the cell division rhythm at CT 06-08 were correlated with the K_a values of cPKA for these ligands (compare columns 1 and 2, Table 1), and the concentrations that were effective at CT 18-20 corresponded to the K_a values of cPKB for these ligands (columns 3 and 4, Table 1). These results suggested that the delaying effects of cyclic AMP observed during the subjective day were mediated by cPKA, and the accelerating effects of cyclic AMP during the subjective night by cPKB.

We have investigated the possible role of the bimodal circadian oscillation of cyclic AMP levels, previously demonstrated in the ZC mutant of Euglena (Carré et al., 1989b), in the generation of cell division rhythmicity. Our rationale was that if the periodic cyclic-AMP signal does play a role in phasing cell division, perturbations of the cyclic AMP oscillation by drugs that affect cyclic AMP metabolism should change the timing of the following burst of cell division. We showed that the permanent addition of cyclic AMP to the cultures of the ZC mutant, synchronously dividing in DD, induced transient advances or delays of the following cell division step, depending on the CT at which the signal was given (Fig. 2). Maximum delays were obtained at CT 06-08 and maximum advances at CT 18-20, both times at which the endogenous cyclic AMP level is minimal (Carré et al., 1989b). Cyclic AMP did not perturb the cell division rhythm at times when its endogenous level was at its peak (CT 00-02 and CT 12-14), a finding that suggests that the cellular cyclic AMP concentration was already saturating.

In most experiments, the perturbation of the cyclic AMP oscillation by exogenous cyclic AMP did not induce permanent phase shifts of the cell division rhythm (Fig. 3B), an indication that the circadian oscillator was unaffected by the signal. A complete loss of cell division rhythmicity was observed following cyclic AMP injection at CT 18-20, however, a finding that suggests a profound disturbance of the coupling pathway for the control of the CDC by the circadian oscillator (Fig. 2C,D). In two cultures, the cells resumed rhythmic cell division 48 to 72 h after the cyclic AMP signal, with a 10-12 h phase shift. It is possible that cyclic AMP is only able to interact with the circadian oscillator at precisely this time and brings it close to its singularity point. Another hypothesis is suggested by the bimodal nature of the cyclic AMP oscillation: since cyclic AMP signals occur at 12 h intervals, it is possible that two solutions exist (180° out of phase with each other) for recoupling the CDC to the circadian oscillator. This would explain the fact that only 10-12 h phase shifts were obtained.

We believe that it is unlikely that cyclic AMP serves as an element of the circadian oscillator in Euglena. If the pacemaker consists of a biochemical feed-back loop, one would indeed expect exogenous cyclic AMP signals to induce both permanent advances and delays of the cell division rhythm. It could be that some of the targets of cyclic AMP or of cyclic AMP-dependent kinases in excitable tissues in which cyclic AMP does phase-shift the output rhythm (Eskin et al., 1982; Prosser and Gillette, 1989) interact with the circadian oscillator. Such indirect resetting of the clock by cyclic AMP – if, indeed, this be the case – does not occur in Euglena (except possibly around CT 20).

The advances and delays of cell division steps observed in growth curves following cyclic AMP injection reflected real changes in the rate of cell cycle progression, as shown by DNA flow cytometry (Fig. 5). Addition of the drug at CT 06 (Fig. 5A) caused a delay in the progression of cells through S phase. This finding is reminiscent of results obtained in mammalian cells, showing that, although a cyclic AMP signal is necessary for G₁/S transition, preventing it from subsiding blocks the initiation of S phase, or prematurely terminates DNA replication (Boynton and Whitfield, 1983). In another experiment, utilizing 8-BZA-cAMP, cells also appeared to be blocked in the G₂ phase or in mitosis, since the accumulation of cells with a 2C DNA content was observed 12 h after the addition of the drug (not shown). A similar inhibitory action of cyclic AMP and cyclic AMP-dependent kinase on cell cycle progression late in the division cycle has been demonstrated in animal oocytes arrested in the first prophase of meiosis. Reinitiation of cell division by fertilization in starfish, or by specific hormones in amphibian and in mammalian oocytes, is accompanied by a drop in cyclic AMP concentration (Meijer and Zarutskie, 1987; Maller, 1985; Schultz et al., 1983; Maller and Krebs, 1977). Thus, cyclic AMP may play a similar role in Euglena, possibly by mediating the inhibition of MPF activity (Maller, 1985).

A cyclic AMP signal given between CT 18 and 20 (Fig. 5B) had the opposite effect on synchronously dividing cultures: 12 h after cyclic AMP injection we observed a marked decrease in the amount of cells with a 2C DNA content, as compared to the control, unperturbed half of the culture. The depletion of G₂ cells was due to an acceleration of mitosis, rather than to an inhibition of DNA synthesis, since no delay was observed in the progression of cells through S phase. This result indicates that cyclic AMP also plays a stimulatory role during the cell cycle in Euglena. It has not yet been demonstrated unambigously whether cyclic AMP plays such a role in mitosis. It has been suggested, however, that cyclic AMP-dependent phosphorylation plays a role in the regulation of mitosis through the regulation of spindle assembly and (or) of microtubule function. The subunits of cyclic AMP-dependent kinase have indeed been localized on the cytoplasmic microtubules, on centrosomes, and on the mitotic spindle (Browne et al., 1980; Tash et al., 1981; Nigg et al., 1985; De Camilli et al., 1986). Furthermore, the microinjection of the heat-stable specific inhibitor of the cyclic AMP-dependent kinase into sea urchin oocytes 20 to 45 min after fertilization blocks the assembly of the mitotic spindle, without preventing nuclear envelope breakdown (Browne et al., 1990). In the ZC mutant of Euglena, cyclic AMP did not seem to have any stimulatory effect on the progression through G₁ and S phases, although it does in other systems (Boynton et al., 1983). It is possible, however, that our assay was not sensitive enough, or that the degree of synchrony of the cell population was insufficient to detect such an effect.

These results are consistent with the hypothesis that periodic cyclic AMP signals act downstream from the circadian oscillator to control cell cycle progression (and other cellullar activities). If this hypothesis holds true, however, permanent increases in cyclic AMP level would be expected to have an uncoupling effect, inducing a permanent loss of cell division rhythmicity. Such a permanent effect was not observed following the permanent addition of cyclic AMP or of phosphodiesterase-resistant cyclic AMP analogs of such as (Sp-cAMP) to the culture medium. One possible explanation is that the cells may have become desensitized (for example, by becoming impermeable to the drug), thus allowing the circadian oscillator to regain control of the CDC.

Nevertheless, we were able to reduce the amplitude of the cyclic AMP oscillation by using forskolin (Fig. 6), which activates adenylate cyclase in Euglena at CTs when its activity is minimal (Tong et al., 1991). When added to LD-synchronized, rhythmically dividing cultures of the ZC mutant, forskolin (10 μM) also caused a rapid loss of cell division rhythmicity, both in populations that were maintained in LD:12,12 and in those that were subsequently transferred to DD (Fig. 7A,B). This desynchronizing effect of forskolin on the cell division rhythm in Euglena strongly indicates that the bimodal, circadian oscillation of cyclic AMP levels exhibited by the ZC mutant is necessary for the phasing of cell division by the circadian oscillator.

We have previously shown that the ZC mutant of Euglena gracilis contains two types of cyclic AMP-dependent kinases (cPKA and cPKB), which have different affinities for cyclic AMP and for several cyclic AMP analogs (Carré and Edmunds, 1992). A correlation between the potency of a cyclic AMP analog in activating one type of kinase and in causing physiological responses can provide evidence for a role of this enzyme in mediating the effect of cyclic AMP (Beebe et al., 1988). The differential activation of the two kinases identified in Euglena extracts by these cyclic AMP analogs provided us with a tool for the study of their respective roles in the control of cell cycle progression, and we determined the mimimum doses of cyclic AMP or of cyclic AMP analogs that caused perturbations of the cell division rhythm during the subjective day, or during the subjective night.

Interestingly, different analogs were effective at CT 06-08 and at CT 18-20, results that suggest that the action of cyclic AMP at the different CTs was mediated by two different cyclic AMP receptors. In addition, there was a correlation between the doses of cyclic AMP, 8-BZA-cAMP, 8-CPT-cAMP, 8-Br-cAMP, and 6-MBT-cAMP that caused perturbations of the CDC at CT 06-08, and the K_a values of cPKA for these analogs (Table 1), a finding that suggests that cPKA mediates the delaying effects of cyclic AMP at those CTs. Similarly, there was a correlation between the doses of the same nucleotides that caused perturbations of the CDC at CT 18-20 and the K_a values of cPKB for these analogs (Table 1), an observation indicating that cPKB mediates the accelerating effects of cyclic AMP at CT 18-20.

These results have been incorporated into a model for the coupling of the CDC to the circadian oscillator (Fig. 8). Bimodal, out-of-phase variations in the activities of adenylate cyclase and phosphodiesterase cause the cellular cyclic AMP level to oscillate (Tong et al., 1991), with peaks corresponding to CT 00-02 and CT 12-14 (Carré et al, 1989b). We propose that the cyclic AMP surge at CT 00-02 delays DNA synthesis and holds the cells at a restriction point in G₂, preventing cell division during the subjective day. The cells are released from this blockage after the level of cyclic AMP subsides, and the G₂/M transition, or mitosis itself, is accelerated by the second cyclic AMP peak, at CT 12-14, so that division is phased to the subjective night. The delaying effects of cyclic AMP on cell cycle progression during the subjective day would be mediated by the activation of cPKA, and the stimulation of mitosis during the subjective night by the activation of cPKB. Activation of either of these kinases would cause the phosphorylation of a different set of targets, and perturb different cell-cycle control pathways. cPKA and cPKB may be expressed at different phases of the CDC. Alternatively, the level of these enzymes might exhibit circadian variations, with cPKA being expressed during the subjective day and cPKB during the subjective night. Another possibility is that the level of one of their downstream targets oscillates, so that only cPKA activation has an effect on cell cycle progression during the subjective day, and cPKB during the subjective night. Future studies of the regulation of cyclic AMP-dependent kinases during the CDC (or during the circadian cycle), and the identification of targets that are selectively phosphorylated by these enzymes, will be necessary to ascertain this part of the model.

In contrast to the model of Homma and Hastings (1989) for the circadian control of cell division in Gonyaulax polyedra, which supposes the existence of a circadian timer within the CDC that controls the exit of cells from G₁, we show that the cell cycle and the circadian oscillator in Euglena are two distinct mechanisms that can be temporarily uncoupled, as suggested by previous results obtained for this unicell (Edmunds, 1964; Edmunds et al., 1976; Edmunds and Laval-Martin, 1984). The model proposed here closely resembles the one proposed by Adams et al. (1984), which formally proposes two different timers (a circadian pacemaker and a cell cycle oscillator) that can either run independently from each other or interact, depending on the growth conditions. This paper, however, is the first to suggest a biochemical basis for the interaction between these two complex processes.

The model that was presented may be of a more general interest, since other cellular activities might be gated by the circadian clock in a similar manner. The further investigation of the signal transduction pathway for the control of the CDC by cyclic AMP signals may permit the identification of cell cycle regulators that modulate the timing of cell division in response to signals from the circadian clock, or to other signals, and should provide important information about the role(s) of cyclic AMP in the CDC.

 
• DD

  continuous darkness

 
• LD

  light-dark cycle

 
• LD:xy

  a repetitive light-dark cycle comprising x hours of light and y hours of dark

 
• τ

  average period of a free-running rhythm under constant conditions here DD

 
• CT

  circadian time (CT 00 indicates the phase-point of a free-running rhythm that has been normalized to 24 h and that corresponds to the identical phasepoint that occurs at the onset of light in a reference LD:12,12 cycle the onset of cell division in a dividing culture occurs at approximately CT 12).

This work was supported by National Science Foundation grants DCB-8901944 and DCB-9105752 to L. N. Edmunds, Jr. We thank Dr J. Tong for helpful discussion and J. Simone for performing the flow cytometry analysis. Some of this work was reported at the Third Meeting of the Society for Research on Biological Rhythms, 6-10 May 1992, Amelia Island, Jacksonville, FL, USA (Abstract 99).