Cell Division Rhythmicity in Wild-Type and Mutant Euglena

The CDCs of many algae, fungi and protozoans exhibit persisting CRs of cell division, or “hatching”.²⁴ Division occurs only at a certain phase of the circadian cycle—often times (“subjective” nights) in constant darkness (DD) or light (LL) corresponding to the dark intervals in a synchronizing LD cycle. This “gating” phenomenon, reflecting an interaction between a circadian oscillator (CO) and the CDC, has been intensively investigated in Euglena.² Because it will serve as a basis for the analysis of the mechanism whereby coupling between the clock and cell cycle is achieved, we treat this class of rhythm in depth.

Singularity Point

Mathematical studies have predicted that a CO might be rendered arrhythmic—characterized by a phaseless, motionless state—by a critical pulse of a certain strength and duration given at a specific time (termed the “singularity point,” S^*) in the circadian cycle.^86,87 As stimulus strength is increased, the transition from Type 1 (weak pulse) to Type 0 (strong pulse) resetting is necessarily discontinuous at one special phase point, the “breakpoint,” corresponding to this unique singularity. This prediction hasbeen demonstrated for light and chemical perturbations in several circadian systems,⁸⁸ and the clock underlying the cell division rhythm is no exception.

The threshold intensity of illumination for phase-shifting the division rhythm in Euglena by 3-h light signals has been examined.⁸⁴ Perturbations having an intensity of 700 to 7500 lux generated the same phase shift (Δ℘) at a given CT (fig. 2A). Signals of lower intensity, however, elicited different responses. Thus, a 40- to 400-lux pulse given at CT0.4 (the approximate location of the breakpoint at about CT23) induced arrhythmicity (fig. 2B), the population reverting to asychronous, exponential growth. The intensity of this annihilating pulse and the CT at which it was imposed were quite specific: A 300-lux stimulus given at CT21.5 (fig. 2C) merely generated a phase delay (-Δ℘) of the same magnitude found for 7500-lux signals. Different degrees of asynchrony were observed as one approached the boundaries (lux and CT) of the critical pulse. The location of S^* at approximately CT24 (CT00) corresponds to the steepest part of the Type 0 resetting curve (see Ref. 83) and probably lies on the axis of the three-dimensional phase resetting surface of Euglena.

Figure 2

Localization of a singularity point in the behavior of the CO underlying the free-running rhythm of cell division in cultures of photoautotrophically grown Euglena gracilis (Z) perturbed by a “critical” 3-h light signal. Three initially (more...)

The existence of this “critical pulse” and its corresponding S^* not only further supports the hypothesis that a CO regulates the CDC in Euglena but also suggests that the pacemaker may have limit cycle dynamics. Inasmuch as the results were obtained with populations of cells, we cannot deduce the state of the oscillator(s) in individual cells. Arrhythmicity in the culture may be the result of a dispersion of the phases of individual CDCs (cellular incoherence), entailing that the pulse differentially phase shift the cells which were synchronous at the time of the pulse; or, the critical pulse may have stopped each circadian clock comprising the population (but not the CDCs, which continue to run).

Temperature Compensation

A remarkable property of CRs is that their period, but not their amplitude, is only slightly affected by the ambient steady-state temperature over the physiological range.⁸⁹ This is just what one would anticipate in a functional biological clock measuring the solar day. In contrast, the duration of the CDC is highly dependent on temperature, and, indeed, this is true for Euglena.⁸ A study of the effects of different constant temperatures ranging between 16 and 32°C on the generation time (g) of the wild-type Z strain (as well as the DCMU-resistant ZR line) photoautotrophically batch-cultured in LD:3,3 clearly illustrates this dependency (fig. 3A).⁸⁵

Figure 3

Contrasting effects of temperature on generation time (g) and the free-running period (τ) of the circadian rhythm of cell division in E. gracilis. A) Influence of temperature on the average g of two different strains (Z, wild type; ZR, DCMU-resistant) (more...)

In a more extensive comparative study of temperature compensation of the free-running period in the Z and ZR strains of Euglena maintained in LD: 3,3 at different steady-state temperatures,⁸⁵ a Q₁₀ of 1.05 was found in the former (fig. 3B), indicating that it was virtually unaffected by changes in temperature over a 10°C range (22 to 32°C). (The circadian clock was not as well compensated in ZR, in which an average value for the Q₁₀ of 1.23 was observed over a temperature range of 18 to 28°C. Nevertheless, even the CO of ZR was affected only modestly by temperature as compared to typical biochemical reaction rates and other biological processes, such as membrane transport.) Finally, a critical permissive temperature of 22°C was found for the Z strain (18°C for ZR): Although slow exponential growth occurred at lower temperatures, synchrony was not seen.)

Photoinduced Commitment and Photoperiodic Control

Earlier work with Euglena and the green alga Chlamydomonas^90-94 has suggested that the CDC comprises two consecutive processes: a precommitment stage and a commitment stage. Progress through the former during the G₁ phase of the CDC would be dependent on photosynthetic growth. The A point denotes the time at which the progression of G₁-phase cells is light-dependent, whereas point T marks the time when cells become able to (i.e., are commited to) complete the CDC, even in darkness.⁹⁰ If cells in the precommitment stage are transferred to DD, they immediately become arrested somewhere between the A and T points, whereas cells in the commitment stage, when transferred to DD, proceed through the remainder of the CDC until they stop at A in the following cycle. This photosynthetic regulation of CDC progression in algal cells resembles the nutritional control of heterotrophs. Nutritional deprivation in Saccharomyces cerevisiae or Schizosaccharomyces pombe arrests the CDC in either the G₁ or G₂ phase,^95-97 while yeast cells that have progressed beyond “Start” (at the G₁/S border) or a size-control checkpoint are committed to complete the CDC even if they are undergoing starvation.⁹⁸

In a study of the effects of light and darkness on CDC progression during log-linear photoautotrophic growth of Euglena, Hagiwara et al³⁰ demonstrated both dark-induced CDC arrest and photoinduced commitment to cell-cycle transition. Light-dependent restriction points were found, not only in the G₁ (as previously described for Chlamydomonas), but also in the post-G₁ phases as well. Thus, if cultures were transferred from LL to DD, some (committed) cells in the G₁, S or G₂ phases were able to undergo one or two CDC transitions even in DD, eventually becoming arrested in S, G₂ or the subsequent G₁ after mitosis. Other (uncommitted) cells were not able to proceed in the CDC. Committed cells that received light of higher intensity were more likely to undergo cell cycle transition in DD and to commit earlier at each CDC phase. Commitments were not achieved at fixed phase-points in the G₁, S or G₂ phases of the CDC.

The commitment to CDC transit in Euglena, therefore, is regulated by the maturity of the cell-cycle phases that is comprises and by light intensity. In addition, Hagiwara et al³¹ have demonstrated that a CO regulates these light responses of commitment to CDC phase transitions. When algal cells were entrained by LD:14,10 and then transferred to DD at the eighth hour of the last LD photoperiod, cells were arrested in G₁, S or G₂. Subsequent exposure of these dark-arrested cells to a 6-h light pulse at different CTs allowed them to further proceed in the CDC in a manner that was dependent on the CT of the pulse (fig. 4). Maximum photoinduction occurred around dusk (CT12). These results demonstrate a circadian gating of photoinduction of commitment to cell-cycle transitions—not just of cell division, as discused earler in this section—and the photoperiodic regulation of cell reproduction. Results from inhibitor experiments with diuron (DCMU), 2,5-dibromo-3-methyl-6-isopropyl-p-benzoquinone (DBMIB), methyl viologen, N,N,N',N-tetramethylphenylenediamine (TMPD) and carbonyl cyanide-m-chlorophenylhydrazone (CCCP) suggested that the photoinduced commitment of G₂ cells to cell division required light for a signal originating upstream in noncyclic photosynthetic electron transport, particularly cytochrome b₆-f, but not for the metabolic energy required for the process.³¹

Figure 4

Circadian gating and photoperiodic control of photoinduced commitment to cell division in E. gracilis Z. Cultures were entrained to LD:14,10 and then given a final photoperiod that was shortened to 8 h before being released into DD. Subsequently, the (more...)

Mutant Strains

In order to eliminate a cellular compartment in which the clock might lie and to obviate possible signaling influences of the LD cycles previously used to elicit rhythmicity and provide energy for growth, we have chosen the achlorophyllous ZC mutant of Euglena—an obligate organotroph—for a series of studies parallel to those carried out in the wild-type strain.¹⁶ Division rhythmicity could be entrained by LD:12,12, or even by a one-pulse skeleton photoperiod (LD:1,23), and the rhythm free-ran in ensuing DD for at least 8 days (τ = 25.5 h), providing that the overall growth rate, or generation time (g) was >24 h (see fig. 5). Similar, though less extensive results had been obtained earlier with other photosynthesis mutants of Euglena, such as the white, heat-bleached, achlorophyllous W₆ZHL strain and the ultraviolet light-induced P₄ZUL mutant.^1,13,15,26 A Type 0 phase-response curve (PRC) for light signals in the ZC mutant strain was generated from 15 phase-shifting experiments¹⁶ (fig. 5); this PRC (fig. 6) resembled that previously obtained with the wild-type strain.⁸³ These studies have provided even more conclusive evidence for the role of a CO in the control of the CDC and have effectively circumvented the problem of the dual use of imposed LD cycles: as an energy source, or “substrate,” for growth and as a timing cue (Zeitgeber) for the underlying clock.

Figure 5

Phase-shifting the free-running rhythm of cell division, previously entained by LD:12,12, by one-hour light (3000 lx) signals given at different circadian times in DD in the achlorophyllous ZC mutant of Euglena gracilis (strain Z), batch-cultured at 16.5°C (more...)

Figure 6

A,B. Phase-response curve (A) for the resetting action of light (3000 lx) signals on the free-running rhythm of cell division of the achlorophyllous ZC mutant of E. gracilis maintained in DD. The steady-state phase-shifts (±Δ℘) (more...)

Thus, the CO plays a key role in the control of the CDC in Euglena, taken to be representative of other eukaryotic cells. Mitosis would not be an essential part of the oscillator but would lie downstream from it. Blockage of cell division should not stop the system from oscillating. ^29,99 Therefore, cell division would be a “hand” of the underlying clock. We have tested this hypothesis in the wild type in two ways: (i) If the free-running division rhythm was stopped due to low initial levels of vitamin B₁₂, and if this inhibition subsequently was released by readdition of B₁₂ to the medium, the rhythm started up again in phase with an unperturbed control. (ii) If a pulse of lactate was given to a free-running culture, temporarily accelerating the CDC and overriding the CO,¹³ the phase of the rhythm when it was finally restored after the supplemental substrate had been depleted was in phase with that of an unperturbed control. These results are consistent with those found previously for the in-phase restoration of rhythmicity in the P₄ZUL mutant free-running in LL by addition of sulfur-containing compounds to the medium.¹⁵

Respiratory Activity in Colorless Euglena Mutants

A CR of photosynthesis has been reported²³ in cultures of wild-type E. gracilis Z cultures, in which the concentration of free O₂ in the medium was continuously measured with a Clark electrode. Oxygen production peaked in the middle of the day, and the rhythm persisted in LL for a least 8 cycles, thus confirming earlier findings^41,44 (see Table 1). These data were extended to colorless mutants lacking functional chloroplasts,²³ but with this strain the phase of the rhythm was the inverse of that in the wild-type strain, with the highest values of oxygen concentration in the medium occurring during (subjective) night (fig. 7). The rhythm in the mutant cells responded directly to the light of more complex LD regimens, and red light (>598 nm) and blue light (<550 nm) were equally effective in regulating respiration activity as compared to white light if given as LD:8,8 (which synchronized the cells). Thus, the demand for energy, provided by either photosynthesis or respiration depending on the cell type, is under both light and circadian clock control. The origin of the rhythm in photosynthesis may lie in the light-induced synthesis of light-harvesting proteins (LHCPs) asssociated with the two photosystems, whose synthesis is clock-regulated, showing an up to 20-fold increase in the subjective day (see following section).⁷⁶

Figure 7

Circadian rhythm of respiratory activity in a colorless mutant of Euglena. Normalized data of the free oxygen concentration within the medium of a culture entrained by LD:12,12 and then released into LL. Reprinted with permission from: Wolff D, Künne (more...)

Circadian Synthesis of Light-Harvesting Proteins

By the in vivo labeling of proteins with [³⁵S]methionine in cultures of E. gracilis grown in LL, Künne and de Groot⁶⁴ confirmed earlier observations⁶³ that total protein synthesis was circadian, and, further, that the τ was only slightly influenced by temperature. Three proteins (separated by SDS-PAGE), having molecular weights of 17, 24 and 60 kD, were found to be synthesized rhythmically (τ ∼26 h).⁷⁶ The 60 kD protein had a maximal rate of synthesis at about CT11, while the two smaller proteins showed peaks at approximately subjective noon (CT05–08) (fig. 8). Their rate of synthesis oscillated by up to 20-fold when maxima and minima were compared. Over temperatures ranging between 16 and 27°C, τ differed by only 3 to 5 h, reflecting fairly good temperature compensation of the free-running period. Synthesis of the two smaller polypeptides was inhibited by cycloheximide but not by chloramphenicol, implying that synthesis occurred on 80S ribosomes. Once synthesized and processed, they were fairly stable in LL for about 120 h, although they were slowly degraded in darkness.

Figure 8

Circadian rhythms in the syntheses of light-harvesting proteins in E. gracilis cultured in LD:12,12 and released into LL. Oscillations in the rate of synthesis of the 17-, 24- and 60-kD proteins were calculated from densitometric traces obtained by scanning (more...)

Subsequently, the two smaller proteins were identified as belonging to the family of light-harvesting-chlorophyll-proteins (LHCPs) of class I (17 kD) and class II (24 kD) by N-terminal sequencing of proteins isolated by two-dimensional PAGE and by detection with an anti-LHCP II serum.⁷⁷ Although the total amount of LHCPs remained almost constant in LD:12,12, the amount of newly synthesized 17 and 24 kD proteins varied by about 20-fold, with maximum synthesis in the light phase. In contrast, their specific mRNAs varied only slightly, a finding (confirmed by using inhibitors of transcription) that suggests that the circadian control of LHCP synthesis occurs at the translational level, in contrast to most higher plants. It may well be that these rhythmicities in the synthesis of LHCPs play a significant role in the circadian control of photosynthesis in Euglena (see Table 1).

Circadian Rhythms in Enzymatic Activity

The activity of a number of enzymes have been shown to undergo circadian variation in Euglena (Table 1). Some more recent representative examples follow.

NAD Kinase, NADP Phosphatase and the Mitochondrial Calcium Cycle

The in vivo levels of NAD⁺, NADP⁺ and NADPH were measured in synchronously dividing and in very slowly dividing cultures of E. gracilis Z grown photoautotrophically in LD:3,3 and were found to oscillate (τ = 27 h), with an amplitude not correlated with the CDC (fig. 9).⁶⁸ There was also a CR in the activity of NAD⁺ kinase (peak at CT00) in extracts of Euglena with a phase relationship such that it could induce the rhythm in the in vivo level of NAD⁺ (see Goto¹⁰⁰). No circadian oscillation of the ratios of either NADH/(NAD⁺ + NADH) or NADPH/(NADP⁺ + NADPH) occurred. Finally, a 3-h light pulse applied at CT18 to the free-running division rhythm not only generated the expected Δ℘, but also similarly reset the phase of the oscillation in total NAD⁺.⁶⁸ These results suggest that the circadian oscillation in the in vivo level of NAD⁺ could be ascribable, as for the duckweed, Lemna,¹⁰⁰ to that of the conversion between NAD⁺ and NADP⁺, but not to that of reduction–oxidation between NAD⁺ and NADH.

Figure 9

(Upper panel) Circadian variations in the intracellular content of NAD⁺ and total NADP(H) in very slowly dividing autotrophic cultures of Euglena gracilis maintained in LD:3,3. Two out-of-phase, free-running cultures (#x25A1; and ?) were sampled at different (more...)

To determine whether NAD⁺ constitutes a possible clock “gear,” small seed cultures displaying a free-running circadian rhythm of cell division were given 2-h pulses at various CTs with NAD⁺ or NADP⁺—expected to directly elevate their own in vivo levels and then to increase or decrease the rate of the reactions catalyzed by NAD⁺ kinase and NADP⁺ phosphatase —or p-nitrophenylphosphate, a competitive inhibitor of NADP⁺, and then resuspended in fresh medium (effectively terminating the pulse) for subsequent monitoring of the division rhythm.⁶⁸ PRCs were derived for each of these agents. Furthermore, a 2.5-h pulse of NAD⁺ given at CT21.7, which caused a Δ℘ of 4 h in the division rhythm, also phase-shifted the CR in the in vivo levels of of NAD⁺, NADH, NADP⁺, and NADPH. These results suggest that NAD⁺ (or NADP⁺, or NADPH), NAD⁺ kinase, and NADP⁺ phosphatase might represent components of the underlying CO.

What is the element that regulates these enzymes? Inasmuch as Ca²⁺-calmodulin activates NAD⁺ kinase in many plants, including green algae, and in the sea urchin, and because the CRs in the activities of NAD⁺ kinase and NADP⁺ phosphatase appear to be generated by a rhythm in the in vivo level of this complex in Lemna,¹⁰⁰ it would be a likely candidate for a clock element in Euglena also. Attempts were made to perturb the clock either by directly causing a transitory decrease in [Ca²⁺] with 2- to 3-h pulses of chlortetracycline, a membrane-permeable chelator of Ca²⁺, or by transitorily inhibiting Ca²⁺–calmodulin by similar short pulses of W7 and chlorpromazine, both calmodulin inhibitors.⁶⁸ All three drugs phase-shifted the division rhythm. Their PRCs suggested that both cytosolic Ca²⁺ and calmodulin constitute clock gears. If so, there should be another gear directly regulating the level of Ca²⁺ in Euglena. The main regulatory sites for many noncircadian systems are known to be the plasmalemma, the endoplasmic reticulum, and the mitochondria. To test the possibility that the mitochondrial Ca²⁺ -transport system might be a clock gear,⁶⁸ cultures were pulsed (2 to 3 h) with either nitrogen, dinitrophenol, or sodium acetate, all of which affect electron transport, ATP hydrolysis and mitochondrial Ca²⁺-efflux. Each of these agents phase-shifted the division rhythm.

Based on these findings, Goto et al⁶⁸ proposed a model in which NAD⁺, the mitochondrial Ca²⁺-transport system, Ca²⁺, calmodulin, NAD⁺ kinase and NADP⁺ phosphatase would represent clock elements that in ensemble, might constitute a self-sustained, circadian oscillating loop—not “the” clock but an oscillator in what is most probably a cellular ”clock-shop.” This proposed feedback loop can autonomously oscillate because the “cross-couplings” are always of opposite sign. The rhythmic activities of the two enzymes would then oscillate 180° out of phase because of the circadian changes in cytoplsmic concentration of one of their effectors, the Ca²⁺–calmodulin complex, which would activate NAD kinase but inhibit NADP phosphatase.

Subsequent in vitro studies^9,20 detected circadian variations in the activities of NAD kinase and NADP phosphatase in the soluble and membrane-bound fractions of both synchronously dividing and nondividing cultures of the ZC mutant of Euglena in DD. Bimodal CRs in total NADP phosphatase activity were found in dividing cells (peaks at CT00 and CT12). The peak observed at CT00–03 disappeared when the cells had ceased dividing, a result that suggests that it might be regulated by the CDC. NAD kinase activity displayed unimodal CRs (peak at CT12) in dividing cells, which persisted with the same phase after the culture had entered the stationary growth phase. The relative velocities of the reactions catalyzed by the enzymes under both K_m and saturating conditions was measured at different CTs, and their ratios were calculated.²⁰ The values obtained for NADP phosphatase exhibited a complex pattern of rhythmicity, while those for NAD kinase displayed circadian variations strongly correlated with the rhythms in enzyme activity. The curves for showed troughs at CT00–04 both in dividing and nondividing cells and peaks at CT18–20 or at CT08–14 in cells sampled, respectively, from a dividing or a stationary culture. Such variations are indicative of changes in the kinetic properties of the enzyme, which may reflect modifications in its affinity either for effectors (such as Ca²⁺–calmodulin) or for its substrate, NAD⁺. This may be due to (i) the expression of different isoenzymes at different CTs; (ii) different posttranslational modifications of the enzyme; or (iii) concentrations of effectors varying in a circadian manner.

Adenylate Cyclase, Phosphodiesterase, Cyclic AMP and Cyclic GMP

A bimodal circadian rhythmicity (peaks at CT09–14, CT19–22) in total cyclic AMP (cAMP) content has been found in dividing and nondividing photoautotrophic cultures of wild-type E. gracilis Z under either entraining (LD:12,12) or free-running (LD:1/2,1/2,) regimes.72 These findings confirm that the bimodal rhythm of cAMP content is regulated by the CO and is not dependent on the CDC.

Similarly, cultures of the achlorophyllous ZC mutant, displaying free-running CRs of cell division in DD¹⁶ (cf. fig. 5), also exhibited bimodal circadian variations in their level of cAMP¹⁷ (fig. 10). Maximum cAMP levels occurred at the beginning of the light period (CT00–02, when cells are in G₁) and at the onset of dark (CT12–14, corresponding to the onset of M). These variations appeared to be CDC-independent, since they persisted after cultures had reached the stationary phase (fig. 10B,C), although more recent results indicate that the oscillations in cAMP level may become attenuated in cell populations during late stationary-phase conditions.⁷² These oscillations in cAMP concentration could be phase-shifted by light signals¹⁷ in a manner predicted by the PRC previously derived for the phase-resetting of the division rhythm by light signals¹⁶ (fig. 5).

Figure 10

Entrained and free-running circadian oscillations of the cAMP level in synchronously dividing and in nondividing cultures of the achlorophyllous ZC mutant of Euglena gracilis (strain Z), grown at 16.5°C on a mineral medium supplemented with ethanol (more...)

Key factors in the cAMP metabolic pathway are two enzymes responsible for its generation and degradation, namely, adenylate cyclase (AC) and phosphodiesterase (PDE). In LD:12,12 these enzymes were found to undergo bimodal, circadian variation of activity in both dividing and nondividing cultures of the ZC mutant of Euglena,²¹ although more recent results suggest that the amplitudes of the rhythms are damped in later-stationary-phase populations. Maximal AC activity occurred 2 h after the onset of the light interval (CT02) and at the beginning of darkness (CT12–14); these times corresponded to the acrophase profile for the rhythmic changes in cAMP content (cf. fig. 10) that have been reported previously.¹⁷ The activity of PDE also exhibited a daily oscillation, but with an inverse phase pattern. Both the AC and PDE activity rhythms persisted in DD (fig. 11).

Figure 11

Free-running CRs in AC activity (A) and PDE activity (B) measured over a 40-h timespan in stationary cultures of the ZC mutant of Euglena maintained in constant darkness (DD). Cultures, synchronized by LD:12,12, were transferred to DD shortly before the (more...)

The activity of AC was activated significantly in vivo by forskolin,²¹ a highly specific activator of AC that can fully stimulate AC activity directly, independently of G proteins. Forskolin stimulation of AC did not result in uniform effects throughout the 24-h day but varied with CT. The degree of AC potentiation by an in vivo forskolin pulse appeared to be inversely dependent on the basal level of AC. Since forskolin appeared to stimulate AC activity at different CTs to the same maximal value regardless of its initial basal level, we conclude that the oscillatory activity of the enzyme derives from modulatory cellular effectors emanating from the CO rather than by changes in the amount of enzyme protein itself.

PDE is an important component of the signal system because it is responsible for the destruction of cAMP after a pulse of synthesis and, thereby, permits the cell to recover from a refractory state induced by cAMP. We found that the addition of 50 μM 3-isobutyl-1-methylxanthine (IBMX) to intact cells at CT20, corresponding to the peak of the rhythm in PDE activity (see fig. 11B), depressed PDE activity by about 51%, while at CT12 (the trough phase) IBMX had only a slight (8%) effect.²¹ These CT-dependent inhibitory effects of IBMX on PDE observed in Euglena extracts may be the resultant of its action on several PDE species, whose relative amounts may also vary with CT. These results indicate that the rhythms of both AC and PDE are key factors in generating the circadian oscillations of cAMP content.

Cyclic AMP and cyclic GMP (cGMP) are second messengers that may act in opposition as a dualistic “Yin Yang” type of system in biological regulation.¹⁰¹ To determine if cGMP plays a role in the mediation of circadian rhythmicity of the AC–cAMP–PDE system in Euglena, the levels of cAMP and cGMP were monitored in synchronized cell populations, and the effects of the cGMP analog 8-bromo-cGMP (8-Br-cGMP) and the cGMP inhibitor LY 83583 (6-anilinoquinone-5,8-quinone) on the activity of AC and PDE, as well as on the level of cAMP, were measured in vivo.²² A bimodal, 24-h rhythm of cGMP content was found in both dividing and nondividing cultures in either LD:12,12 or DD. The peaks and troughs of the cGMP rhythm occurred 2 h in advance of those of cAMP. The addition of 8-Br-cGMP at different CTs increased the cAMP level in vivo by two to five times, while LY 83583 reduced the amplitude of the cAMP rhythm so that it disappeared. The effects of 8-Br-cGMP on the activity of AC and PDE were CT-dependent, consistent with the changes in cAMP content. These findings suggest that cGMP may serve as an upstream effector that mediates the cAMP oscillation by regulation of the AC–cAMP–PDE system.

Coupling between Circadian Oscillator and Cell Cycle Clock

We have reviewed evidence for circadian clock control of the cell division cycle of Euglena (as well as many other microorganisms). This “gating” phenomenon is thought to reflect an interaction between a CO and the CDC. This coupling may be effected by cAMP, which has been implicated both in the regulation of the CDC and in circadian clock function, as a coupling link in the input and output pathways, if not as an element of the oscillator itself.¹⁸

The CDC of eukaryotes is a complex cascade of events that culminates in cell duplication. ¹⁰² These transitions from one regulatory state to another initiate the modification of substrates that determine the physical state of the cell and are themselves feedback-controlled. The basic mechanism, elucidated by genetic and biochemical analyses, has been conserved in organisms ranging from unicells such as yeasts, to Xenopus, clams, starfish eggs, cultured human cells and higher plants (reviewed in ref. 103).

Two crucial transition control points exist at the G₁/S and G₂/M boundaries of the CDC. The M phase is characterized by the activation of a kinase [MPF, or Maturation (M phase)-Promoting Factor, so named for its ability to induce cell division when injected into Xenopus oocytes], which consists of two component subunits. One is the p34^cdc2 protein (the product of the S. pombe gene cdc2, or its homologue, CDC28, in S. cerevisiae), which is able to phosphorylate casein and histone H1 in vitro and has maximal kinase activity at mitosis. The other is cyclin, whose cellular concentration fluctuates during the CDC. It accumulates gradually during interphase, forms a complex with p34^cdc2, and activates the protein kinase; it is then degraded during mitosis, turning off kinase and MPF activity. The cyclin–cdc2/CDC28 (MPF) system thus behaves like an oscillator, or “clock,” which is reset to its interphase state during mitosis and which appears to constitute a universal cell-division-cycle “engine.”¹⁰³ We now know that there are entire families of cyclins with which p34^cdc2 periodically associates to create successive waves of protein kinase activity that regulate the progression of the CDC and that distinct classes of cdc2-related genes are differentially expressed during the CDC in plants and animals.

Circadian Rhythmicity in Cyclic AMP Levels

We have discussed the CDC-independent, bimodal CRs of cAMP levels observed in cultures of both wild-type Euglena^50,51 and of the achlorophyllous ZC mutant⁵² (cf. fig. 10) displaying free-running CRs of cell division. Maximum cAMP levels occurred at the beginning of the light period (CT00–02, when cells are in G₁) and at the onset of dark (CT12–14, corresponding to the onset of M). Similar fluctuations of cAMP that are tightly connected to cell cycle progression in synchronized suspension cultures of tobacco BY-2 have recently been reported. ¹¹² Cyclic AMP peaks were observed during the G₁ and S phases. Application of indomethacin, a drug that inhibits adenylyl cyclase in animal cells, at early S phase resulted in the loss of the S-phase cAMP peak and inhibited mitotic division by blocking cells at the G₂/M border.

Collectively, these results suggest that cAMP levels are controlled by the same endogenous clock as that regulating the CDC and that the periodic, bimodal cAMP signal (initiated by the CO, or in conjunction with it) may participate in the gating of CDC events to specific phases of the circadian cycle.

Perturbation of the Cell Division Rhythm by Cyclic AMP

If periodic cAMP signals play a role in the generation of cell division rhythmicity in Euglena, then conditions that disturb or override the the control of the cAMP level by the CO should also cause derangements in downstream events.¹⁸ Perturbations of the cAMP oscillation (see fig. 10) by exogenous cAMP resulted in the temporary uncoupling of the CDC from the circadian timer, causing either a shortening or a lengthening of the CDC, depending on the phase of the CDC when the drug was applied (fig. 12A). Delays of the next synchronous division step (up to 9 h) were obtained when cAMP was given between CT03 and CT09, and advances were observed when cAMP was given between CT16 and CT22. Maximum effects were obtained at CT06–08 and CT18–20, corresponding to times when endogenous cAMP levels were minimal (see fig. 10). Dose-response curves were also derived at CTs when maximum +ΔΦs or -ΔΦs of the CDC had been obtained. As little as 1 nM cAMP was enough to perturb CDC transit. The CO, in contrast to the CDC cyclin clock, and unlike excitable tissue in which cAMP does phase-shift the output rhythm,^113,114 was not reset by addition of cAMP: the division rhythm returned to its original phase (fig. 12B).

Figure 12

Phase-response curves for the perturbation of the free-running rhythm of cell division by cAMP in the achlorophyllous ZC mutant of Euglena. The curves were derived from 17 separate experiments. (A) Advances or delays of the first cell division step following (more...)

The ΔΦs of the division steps observed in the growth curves following cAMP injection reflected real changes in the rate of CDC progression. Measurement of cell DNA content by flow cytometry¹⁸ indicated that cAMP injected at CT06–08 delayed progress through S phase, and perhaps through M (fig. 13A). If added at CT18–20, cAMP accelerated the G₂/M transition (fig. 13B). These effects on the cell division rhythm were only transitory: After 48 h division occurred at the same phase as in unperturbed controls (see fig. 12B).

Figure 13

Effects of cAMP on cell cycle progression. Cultures of the ZC mutant of Euglena synchronously dividing in DD were divided into two halves, one of which received cAMP (white histograms), while the other was used as a control (black histograms). Cells were (more...)

That the endogenous circadian variations of cAMP level were of sufficient amplitude to have similar effects on CDC progression was demonstrated by testing the effect of drugs that reduce the amplitude of the cAMP oscillation, or that keep cAMP at a level such that all cAMP receptors should be permanently saturated: We would expect such drugs to prevent the expression of division rhythmicity. Indeed, when the cAMP analog forskolin, an activator of AC that maintained cAMP at an abnormally high level,²¹ was added to a culture of the ZC mutant free-running in DD, a rapid desynchronization of the cell population occurred and division rhythmicity eventually was lost.¹⁸

These findings indicate that although cAMP signals do not reset the CO in Euglena (thus, cAMP is unlikely to represent a “gear” of the clock), they do regulate CDC progression, acting downstream from the oscillator—perhaps in a parallel pathway, as Zatz¹¹⁵ has suggested for chick pineal cells.

Downstream Pathway: Cyclic AMP-Dependent Kinases

What are the coupling links between the oscillatory cAMP system and the CDC? Whatever be the cell cycle regulatory pathways that are affected, one must explain how identical signals can perturb different pathways depending on the CT at which they are applied, so that cell cycle delays are obtained during the subjective days, and advances during the subjective nights. One possibility is that cAMP acts through different “receptors” that selectively modulate one or the other of the two regulatory pathways.

We have demonstrated that the ZC mutant of Euglena contains two types of cAMP-dependent kinase (cPKA and cPKB), which have different affinities for cAMP and for several cAMP analogs.¹¹⁶ Cell extracts were found to contain two cAMP-binding proteins, which bound cAMP with a high affinity (Kd values of 10 nM and 30 nM) and which could be separated by DEAE-cellulose chromatography. Protein kinase activity was assayed using Kemptide as a substrate (specifically phosphorylated by cAMP-dependent kinases from mammalian cells). Stimulation of kinase activity by cAMP was observed after partial purification by DEAE-cellulose chromatography. Two peaks of activity were resolved, corresponding to enzymes with different cAMP-analog specificities. Thus, cAMP signalling in plant cells may proceed by phosphorylation of target proteins by cAMP-dependent kinases in a manner similar to that of animal cells.

The differential activation of the two kinases identified in Euglena extracts by cAMP analogs afforded a tool for the study of their respective roles in the control of CDC progression; we determined the mimimum doses of cAMP or of cAMP analogs that caused ΔΦs of the division rhythm during the subjective day, or during the subjective night.¹⁸ Different results were obtained at CT06–08, and at CT18–20, suggesting that the effect of cAMP at the different CTs was mediated by two different cAMP kinases. Thus, 8-benzylamino-cAMP (8-BZA-cAMP), which selectively activates cPKA, induced -ΔΦs when added at CT06–08, but had no effect at CT18–20. Reciprocally, 8-(4-chlorophenylthio)-cAMP (8-CPT-cAMP), which specifically activates cPKB, induced +ΔΦs and a loss of division rhythmicity when added at CT18–20 but did not perturb the division rhythm when added at CT06–08.

There also was a correlation between the doses of cAMP, 8-BZA-cAMP, 8-CPT cAMP, and N⁶ -monobutyryl-cAMP (6-MBT-cAMP) that caused perturbations of the CDC at CT06–08 and the K_a values of cPKA for these analogs, a finding that suggests that cPKA mediates the delaying effects of cAMP at these CTs. Similarly, there was a correlation between the doses of the same nucleotides that caused perturbations of the CDC at CT18–20 and the K_a of cPKB for these analogs, a result indicating that cPKB mediates the accelerating effects of cAMP at CT18–20. A simple explanation for these findings would be that these kinases are expressed at different stages of the CDC, as has been described for type I and II cAMP-dependent kinases in mammalian cells.

Model for Circadian Control of the Cell Division Cycle

These results were incorporated into a model (fig. 14) for the coupling of the CDC to the CO.¹⁸ We have proposed that the cAMP surge at CT00–02 delays DNA synthesis and then holds the cells at a restriction point in G₂ to prevent cell division during the subjective day. The cells are released from this blockage after cAMP levels subside, which Grieco et al¹¹⁹ have shown is associated with the tyrosine-dephosphorylation-induced activation of MPF, and the G₂/M transition, or mitosis itself, is accelerated by the second cAMP peak, at CT12–14, so that mitosis is phased to the subjective night. The delaying effects of cAMP on CDC 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's 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 CDC progression during the subjective day, and cPKB during the subjective night.

Figure 14

Model for the “gating” of cell division by the CO. We propose that the cAMP surge at CT00-02 delays DNA synthesis (and, perhaps, holds the cells at a restriction point in G₂) to prevent cell division during the subjective day. The cells (more...)

Interface with the Cyclin Oscillator

How does the CO, acting via (or in conjunction with) the AC–cAMP–PDE system and the cAMP-dependent kinases (cPKA and cPKB) that we have identified, ultimately couple to the MPF clock regulating the CDC? Recently, we have run western blot analyses of cyclin B (the regulatory subunit of MPF) using monoclonal antibody to human clb1 protein and have found that the abundance of the homolog in protein extracts from dividing ZC cultures varied quantitatively during both the circadian cycle and the CDC,73,74 with a peak at CT00 and a trough at CT12. These circadian fluctuations in cyclin B in dividing cells ceased, however, during the nondividing phase in stationary-phase cultures.

Preliminary western blot analyses of protein extracts from dividing and stationary-phase cultures of the ZC mutant in DD^73,74 confirm the presence of a homolog of p34^cdc2, the catalytic subunit of MPF, reported earlier in wild type Euglena and in other algae^117,118 and suggest that its abundance throughout the CDC is invariant, decreasing to a much reduced level when cells cease dividing. Further, a relative shift in electrophoretic mobility of the monomer was observed to occur in a cycle-dependent manner: A more slowly moving band disappeared at CT09 and reappeared at CT18. Such changes in mobility have been attributed to phosphorylation and dephosphorylation. Remarkably, this rhythm of phosphorylation of cdc2 persisted with a circadian period (τ = 26h), even in stationary-phase cultures. We determined the nature of these changes by using potato acid phosphatase to completely dephosphorylate cdc2, which then ran as a single band with no anti-phosphotyrosine immunoreactivity.^73,74 These initial findings suggest that the activity of this polypeptide depends on posttranslational mechanisms.

More recent results⁸⁰ using untreated extracts have revealed, in addition to dephosphorylated protein, two phosphorylated forms of cdc2. The more slowly moving (more phosphorylated) form predominated during the subjective day and was the only form recognized by anti-phosphotyrosine. The faster moving, less phosphorylated form appeared during the subjective night concomitant with a decline in the more slowly moving form. The faster moving, phosphorylated form lacked tyrosine phosphorylation and, in fact, may be threonine-phosphorylated, as is the case in other species. By the end of the subjective night, however, this form disappeared, while the amount of the more slowly moving form, now phosphorylated at both residues, became elevated before subjective dawn. We conclude that reversible tyrosine phosphorylation of cdc2 is coupled to the CO.

Finally, we note that Bjarnason and coworkers^119,120 have examined the circadian expression of clock genes in human oral mucosa and skin and their association with specific CDC phases. The relative RNA expression of hClock, hTim, hPer1, hCry1 and hBmal1 was determined in biopsies taken throughout one timespan of 24 h and was found to exhibit a circadian profile consistent with that found in the SCN and peripheral tissue of rodents. hPer1, hCry1 and hBmal1 were rhythmic, peaking in the early morning, late afternoon, and at night, respectively, whereas hClock and hTim were arrhythmic. In concurent oral mucosa biopsies, thymidylate synthase activity (a marker for DNA synthesis), had a circadian variation with an early afternoon peak, coinciding with the timing of S phase. The major peak in hPer1 expression occurred at the same time of day as the peak in the G₁ phase in oral mucosa, suggesting a link between the CO and the mammalian CDC. Similarly, the CDC proteins, p53, cyclin E, cyclin A and cyclin B1 were found to vary rhythmically, with their respective peaks' occurring in late G₁, G₁/S, G₂ and M. Likewise, in an analysis of some 68 CDC-related genes in regenerating liver of mice, using DNA microarray and Northern blot techniques, Matsuo et al¹²¹ have shown that the circadian clock directly controls the expression of wee1, which in turn modulates the expression of active cyclin B1–Cdc2 kinase and the entry of cells into mitosis. These findings have important clinical implications for the timing of both chemotherapy and radiotherapy, as illustrated by a recent study¹²² reporting that proapoptotic drugs such as docetaxel displayed least toxicity and highest antitumor efficacy following dosing during the circadian rest phase in mice, a finding that suggests that the CDC and apoptotic processes (e.g., expression of the proapoptotic BCL 2-associated X protein) could be regulated by the circadian clock.