Ca²⁺ signals coordinate zygotic polarization and cell cycle progression in the brown alga Fucus serratus

The coordination of cell polarization and cell cycle progression is crucial for proper zygotic development. In eukaryotes, both polarization and cell cycle progression rely on evolutionarily conserved mechanisms that include,respectively, the cytoskeletal actin nucleation machinery and cyclins/cyclin-dependent kinase (CDK) activity. These mechanisms are, in turn,coordinated by checkpoints that ensure asymmetric cell division events occur in their correct order (Hartwell and Weinert, 1989).

It follows that distinctive patterns of asymmetric division in different eukaryotic cell types may be explained largely by differences between the identity and order of checkpoints, rather than by differences between the basic mechanisms of polarization or cell cycle progression(Jensen et al., 2006; King et al., 1994). For example, in early Drosophila embryos, actin nucleation and DNA replication (Callaini et al.,1992; Raff and Glover,1988) are not interdependent and this checkpoint absence promotes synchronous nuclear division (Hartwell and Weinert, 1989; Sullivan et al., 1993). Other organisms, such as Saccharomyces cerevisiae, show more stringent coordination of polarization and cell cycle progression, with cytoskeletal actin nucleation only beginning once cells have passed through the `start' checkpoint of late G1/S phase(Moffat and Andrews, 2004; Singer et al., 1984). This actin nucleation is itself then monitored by a second, `morphogenesis',checkpoint (Lew and Reed,1995), which delays G2-to-M progression until cytoskeletal rearrangements have turned the parent cell into a characteristic `shmoo' shape(Keaton and Lew, 2006; McMillan et al., 1998).

It is particularly important to understand such checkpoint control during eukaryotic embryogenesis, because the first asymmetric cell division defines the embryonic axes that ensure correct multicellular patterning in the adult organism (Huynh and St Johnston,2004; Jenik and Barton,2005). To this end, the externally fertilized gametes of brown algae provide an excellent model for developmental studies of environmentally polarized embryos (Brownlee and Bouget,1998; Peters et al.,2008). In Fucus serratus zygotes, as in S. cerevisiae, polarization and DNA replication are both reliant on late G1/S phase CDK activity (Corellou et al.,2001b). However, the potential importance of actin and cytosolic Ca²⁺ ([Ca²⁺]_cyt) signals-which are known to regulate CDKs in many systems (Lew and Reed, 1995; Philipova and Whitaker, 1998)-has not been addressed in this, or in any other,multicellular plant or algal zygotic model.

Specifically, although it is well established that[Ca²⁺]_cyt elevations follow sperm-egg fusion in a number of algal (Roberts et al.,1994), animal (Gilkey et al.,1978) and flowering plant eggs(Digonnet et al., 1997), in many of these organisms continued [Ca²⁺]_cyt elevations are also known to integrate subsequent developmental events(Whitaker, 2006a). In Fucus zygotes, localized [Ca²⁺]_cyt elevations are known to be essential for fertilization(Roberts et al., 1994) and polarization (Pu and Robinson,1998; Speksnijder et al.,1989), but a potential role for post-fertilization[Ca²⁺]_cyt increases in initiating and co-ordinating cell cycle progression has not been explored.

In the present study, we test the hypothesis that[Ca²⁺]_cyt elevations coordinate actin nucleation,polarization and cell cycle progression during the first Fucuszygotic cell cycle. Previous attempts to study developmental[Ca²⁺]_cyt elevations in algal zygotes have been limited by the highly pigmented nature and scattering properties of the zygotes(Roberts et al., 1994), by difficulties in loading large numbers of eggs and zygotes with indicators or Ca²⁺ buffers, and by the absence of single cell markers for cell cycle events in vivo (Corellou et al.,2000). To overcome these problems, we use 2-photon microscopy and high-throughput biolistic loading to study how post-fertilization[Ca²⁺] elevations direct zygotic development. Our results suggest that distinct cytosolic and nuclear [Ca²⁺] elevations drive actin nucleation and cell cycle events, and that these [Ca²⁺] elevations are temporally and spatially distinct from the later polarizing[Ca²⁺] gradient. We also show that actin nucleation is essential for zygotic polarization, but not for cell cycle progression, presenting a system of asymmetric cell division control that is distinct from either the morphogenesis checkpoint-dominated yeast model or the uncoupled polarization and cell cycle model of Drosophila embryos.

Mature receptacles were cut from male and female Fucus serratus(L.) plants and dry stored at 4°C in the dark for up to 2 weeks. Gametes were released by placing male and female receptacles in separate dishes and covering them with filtered (0.45 μm) seawater at 14°C. After 1-2 hours, released gametes were collected and eggs were either used for dye loading or mixed with sperm, with fertilization occurring within 15 minutes of gamete mixing. Eggs or zygotes were filtered through a 100 μm nylon mesh,washed twice in filtered seawater, and maintained under unidirectional light at 14°C until use.

The GFP fused to the human PCNA gene and containing SV40 Nuclear Localization Signal at the N terminus was a kind gift from Dr C. Cardoso (Max Delbruck Center for Molecular Medicine, Berlin, Germany). The GFP-PCNA was subcloned into the BamHI/SacI restriction sites of the pCal-n expression vector and its correct insertion was verified by sequencing as previously described (Kisielewska et al., 2005; Philipova et al.,2005). The GFP-PCNA protein was then expressed in BL21 competent E. coli cells (Promega; www.promega.com)cultured at 37°C. The protein was purified on calmodulin affinity resin(Stratagene; www.stratagene.com)according to the manufacturer's protocol.

All dyes and Ca²⁺ buffers were from Invitrogen(www.invitrogen.com)or Sigma(www.sigmaaldrich.com). The Ca²⁺-sensitive fluorescent dye fura dextran (10 kDa; 4 mM), the Ca²⁺-insensitive dyes Texas Red dextran (10 kDa; 4 mM) and fluorescein isothiocyanate dextran (FITC; 10 kDa; 20 mM), the Ca²⁺buffers BAPTA dextran (10 kDa; 20 mM) and dibromoBAPTA (Br₂BAPTA;40 mM), and the green fluorescent protein constructs GFP (5 mg/ml) and GFP-PCNA (3 mg/ml) were coated onto gold particles (see below) and introduced in varying combinations into F. serratus eggs or zygotes using a biolistic method (Bothwell et al.,2006). Control experiments using the cell-permeant heavy metal chelator N,N,N′,N′-tetrakis(2-pyridylmethyl)ethylenediamine (TPEN)were conducted by supplementing the perfusing medium with 1-100 μM TPEN or the equivalent amount of DMSO (0.2% w/v).

Parameters for the biolistic loading protocol were, briefly, as follows:0.6 mg of 1.0 μm gold particles (Bio-Rad; www.bio-rad.com)were coated in the relevant solutions as previously described(Bothwell et al., 2006). If GFP-PCNA was being loaded, the macrocarriers were used immediately to prevent drying of the GFP-PCNA protein, otherwise macrocarriers were dried in a refrigerated drying chamber before use. Prior to biolistic loading, zygotes were transferred into filtered seawater supplemented with 0.6 M sorbitol. Around 100 μl of eggs or zygotes were then immediately transferred in a dense suspension (>100,000 cells/ml) to a 35 mm petri dish containing 3%seawater agar ∼5 mm deep and subjected to biolistic bombardment using a 2200 psi rupture disc (Bothwell et al.,2006).

After loading, zygotes were settled onto 35 mm culture dishes with 0.08-0.12 mm thick coverslip bases and left in filtered seawater under unidirectional light at 14°C until needed. Loaded eggs were left to recover in filtered seawater at 14°C for 1 hour before being mixed with sperm and were treated as zygotes thereafter.

To confirm that GFP-PCNA was not denatured by biolistic loading, the localization pattern of biolistically loaded GFP-PCNA was compared with that of biolistically loaded GFP (see Fig. S1A,B in the supplementary material) or microinjected GFP-PCNA (see Fig. S1C-E in the supplementary material). For microinjection, zygotes were superfused with filtered seawater supplemented with 0.7 M sorbitol to reduce internal turgor pressure and GFP-PCNA (3 mg/ml in 200 mM KCl, 10 mM Hepes, 550 mM mannitol, pH 7.0) was pressure microinjected using dry bevelled pipettes fabricated from 1.2 mm filamented borosilicate glass (Taylor et al.,1996). No significant difference was observed between the behaviour of GFP-PCNA introduced into zygotes by microinjection or biolistic loading (see Fig. S1C-E in the supplementary material).

Cell wall secretion was monitored as a proxy for fertilization using 0.0001% Calcofluor white (CFW; Sigma) to stain cell wall cellulose. CFW fluorescence was monitored with either UV epifluorescence or 2-photon excitation (see below).

To quantify zygotic DNA levels during early development, zygotes were stained with 100 μM Hoechst 33342 (Sigma) for 15 minutes and nuclear DNA visualized using 2-photon microscopy (see below). To measure rates of sperm pronuclear motion, sperm were stained with 100 μM Hoechst 33342 for 10 minutes and filtered through a 100 μm nylon mesh before being added to eggs. Progress of the sperm pronucleus through the egg cytoplasm was followed either with a graticule fitted to the eyepiece of an epifluorescence microscope (Swope and Kropf,1993) or with 2-photon imaging (see below).

Zygotes were fixed in buffer (2.5 mM PIPES, 1 mM MgSO₄, 1 mM EGTA, 0.75 M sucrose, pH 7.0 containing 0.3 mM m-maleimidobenzoyl-N-hydroxysuccinimide ester and 3.7% formaldehyde) for 30 minutes, then washed in filtered seawater and stained with 10 μM Texas Red-phalloidin for 30 minutes, before being washed twice more in filtered seawater and imaged using confocal microscopy (see below).

Dye-loaded cells were imaged with a Zeiss LSM 510 confocal/2-photon microscope(www.zeiss.com)and superfused using a gravity perfusion system. Loaded cells were imaged using a C-Apochromat 63×/1.2 n.a. water-immersion objective. All images were processed using Zeiss LSM Image Examiner software or Scion Image(www.scioncorp.com).

For confocal imaging FITC was excited using the 488 nm line of an argon/2 laser and Texas Red phalloidin excited using the 543 nm line, with emitted light bandpass filtered between 500-550 nm or 565-615 nm, respectively. Pixel images (512×512) were acquired in both cases, with four-line averaging and a pixel dwell time of 2.56 μseconds.

For all other fluorophores, the Zeiss LSM 510 was used in 2-photon mode with excitation provided by a tunable Titanium:Sapphire Mai Tai laser (Spectra Physics; www.spectraphysics.com). In all cases, 512×512 pixel images were acquired with a pixel dwell time of 2.56 μseconds and, unless stated, eight line summation. The CFW/fura dextran/Texas Red combination was excited at 780 nm with emissions monitored at 405-458 nm, 501-554 nm and 565-615 nm, respectively. Hoechst 33342 was excited at 765 nm and emitted light bandpass filtered between 435 and 485 nm. The GFP-PCNA/Texas Red marker combination was excited at 925 nm with emitted light bandpass filtered between 500-530 nm and 565-615 nm. DiOC₆(3)was excited at 950 nm, with emitted light bandpass filtered between 390-465 nm and 500-550 nm and using four-line averaging.

Two-photon excitation of fura dextran at 780 nm results in decreased fluorescence with increasing [Ca²⁺](Wokosin et al., 2004) and[Ca²⁺] was calculated from the R_min and R_maxvalues of the Texas Red/fura dextran fluorescence ratio (R) values(Grynkiewicz et al., 1985). Dye-loaded cells were superfused alternately with 50 mM Ca²⁺seawater and Ca²⁺ free seawater (with 0.1 mM EGTA) in the presence of 100 μM ionomycin (Calbiochem; www.merckbiosciences.co.uk)to obtain R_min and R_max, respectively. Nuclear fluorescence was quantified from projections of up to 10 different confocal nuclear sections.

Ca²⁺ buffer effects on cell cycle progression and zygotic polarization were compared by measuring three parameters 24 hours after fertilization: first, cytosolic Ca²⁺ buffer concentration([Ca²⁺ buffer]_cyt; biolistically loaded buffers were quantified by co-loaded FITC dextran fluorescence); second, the length/width ratio; and, third, a binary score reflecting cell cycle progression, with 0=one nucleus and no M-phase entry and 1=two nuclei and M-phase entry.

Zygotic polarization in F. serratus is known to be S-phase dependent (Corellou et al.,2001b), so we used analysis of residuals to test a second null hypothesis: that Ca²⁺ buffers had no effects on polarization that could not be explained by their effects on S-phase inhibition. As germinated rhizoids elongate at a constant rate for at least 24 hours after S-phase entry, it follows that zygote length/width ratios are directly proportional to cell cycle progression rates. Accordingly, logistic regressions (above) were used to predict cell cycle rates in each zygote, given their [Ca²⁺buffer]_cyt. Predicted length/width values were then subtracted from the measured length/width ratios to give length/width ratio residuals. Residuals were plotted against [Ca²⁺ buffer]_cyt (see Fig. 5B,C) and linear regressions tested for deviations from the null model fit, i.e. the horizontal line passing through the origin.

To monitor post-fertilization DNA replication in individual F. serratus zygotes, we took advantage of the fact that, during S phase,proliferating cell nuclear antigen (PCNA) forms a highly conserved sliding clamp that is associated with active DNA polymerase(Moldovan et al., 2007). Nuclear accumulation of a GFP-PCNA chimera(Leonhardt et al., 2000) can,therefore, be used as a proxy assay for zygotic S-phase onset(Kisielewska et al.,2005).

As in sea urchin embryos (Philipova et al., 2005), F. serratus eggs that were co-loaded with GFP-PCNA and the marker dye Texas Red dextran showed an initial nuclear localization of GFP-PCNA (Fig. 1A), which remained constant for up to 2.5 hours after fertilization (Fig. 1B). From 3.5 hours onwards, the pattern of Texas Red dextran fluorescence remained unchanged, but the GFP-PCNA fluorescence in the nucleus began to increase relative to that in the cytosol, remaining significantly elevated (F>5.73, P<0.03) for up to 4.5 hours after fertilization(Fig. 1A,B). The timing of the nuclear GFP-PCNA elevation-around 3 hours after fertilization-matches the onset of DNA replication, as measured by nuclear staining with the DNA-binding dye Hoechst 33342 (see Fig. S2 in the supplementary material) and broadly agrees with earlier work in populations of F. serratus zygotes, in which increases in Histone H1 kinase activity were used as a marker for DNA replication to indicate that S-phase onset occurred around 4 hours after fertilization (Corellou et al.,2001a).

As the present study represents the first use of GFP-PCNA in a non-animal system, we confirmed that nuclear GFP-PCNA accumulation reflected S-phase onset by noting that in sea urchin embryos the S-phase-associated nuclear accumulation of GFP-PCNA (Kisielewska et al., 2005) is reduced by the DNA polymerase inhibitor aphidicolin(Goscin and Byrnes, 1982; Ikegami et al., 1978). Accordingly, we confirmed that the post-fertilization increase in nuclear GFP-PCNA levels in F. serratus zygotes was completely inhibited(F=6.86, P<0.001) by aphidicolin (20 μM) applied continuously from 30 minutes after fertilization (Fig. 1C; see also Fig. S2 in the supplementary material). Taken together, these data show that GFP-PCNA can be used as a robust marker of S phase in individual F. serratus zygotes.

To investigate whether [Ca²⁺] elevations are associated with early embryogenesis in F. serratus, we used 2-photon measurement of biolistically co-loaded Texas Red/fura dextran ratios to monitor cellular[Ca²⁺] during the first 3 hours of development.

The first visible sign of fertilization and egg activation was the polarized secretion of cell wall components, which progressed around the zygote at around 90±10°/minute (n=8; linear regression with R²=0.95±0.02) and thus enclosed the zygote within 5 minutes (Fig. 2A). This was followed by a 30 minute rise in cytosolic and nuclear [Ca²⁺], to a peak of more than 300 nM (Fig. 2C,D). We did not observe any fast cytosolic [Ca²⁺]waves or polarized [Ca²⁺] elevations matching the polarized pattern of cell wall secretion.

The sperm pronucleus moved towards the egg pronucleus at a steady rate of 0.31±0.02 μm/minute (n=8; linear regression with R²=0.96±0.02) at 18-19°C(Fig. 2B,C), which compares with earlier estimates of 0.20±0.06 μm/minute at 14°C in the brown alga Silvetia compressa(Swope and Kropf, 1993). Given this rate of pronuclear migration, fusion of the sperm and egg pronuclei would have occurred from 2-3 hours after fertilization(Fig. 2B,C) and our results in Fig. 1 are consistent with S phase beginning after pronuclear fusion. Such post-pronuclear fusion onset of S phase has been proposed in sea urchin eggs(Longo and Plunkett, 1973),although more recent studies have shown that both PCNA(Kisielewska et al., 2005; Philipova et al., 2005) and the S phase cyclin E and cdk2 (Moreau et al., 1998; Schnackenberg and Marzluff, 2002) localize to male and female sea urchin pronuclei immediately after fertilization, suggesting that DNA replication may begin slightly before pronuclear fusion in some organisms.

After the sperm pronucleus reached the egg pronucleus, a rise in nuclear,but not in cytosolic, [Ca²⁺] was apparent, increasing to a peak of less than 300 nM in late G1/S phase, between 2 and 3 hours after fertilization(Fig. 2C,D).

To determine whether the [Ca²⁺] changes seen in Fig. 2 are required for zygotic polarization or cell cycle progression, we introduced the Ca²⁺buffer, BAPTA dextran (0.4-0.8 mM), into zygotes before S-phase onset (between 30 minutes and 1 hour after fertilization), and monitored subsequent pronuclear motion, actin nucleation and nuclear GFP-PCNA localization(Fig. 3).

BAPTA dextran had no effect (F=4.54, P<0.32) on sperm pronuclear motion, with sperm pronuclei migrating towards egg pronuclei at 0.22±0.03 μm/minute in FITC dextran-loaded eggs (n=10) and 0.26±0.02 μm/minute (n=7) in BAPTA dextran-loaded eggs(Fig. 3A,B).

Untreated control zygotes showed both polarization and polarized actin nucleation (Fig. 3C), but loading zygotes with BAPTA dextran (Fig. 3D) inhibited both zygotic polarization and the cortical actin nucleation on which polarization depends(Hable and Kropf, 2000; Pu et al., 2000).

Finally, BAPTA dextran significantly inhibited (F>5.82, P<0.05) S-phase onset, as measured by nuclear GFP-PCNA accumulation (Fig. 3F).

Because previous work has shown that inhibition of CDK activity had a similar effect to our BAPTA dextran studies on zygotic polarization(Corellou et al., 2001b) and DNA replication (Corellou et al.,2001a), we treated zygotes with the cyclin-dependent kinase inhibitor olomoucine (100 μM) continuously from 20 minutes after fertilization. Olomoucine did not prevent cortical actin nucleation(Fig. 3E), although it did prevent polarization. This inability of olomoucine to mimic the full effect of BAPTA dextran suggests that separate Ca²⁺ dependencies may exist for actin nucleation and the G1/S phase checkpoint. This hypothesis is discussed further in the final Results section, below.

To investigate whether actin nucleation or localization are necessary for DNA replication and to determine whether the inhibitory effect of the Ca²⁺ buffer BAPTA dextran on DNA replication was mediated through inhibition of actin nucleation, we studied the effects of cytoskeletal actin disruption on cell cycle progression.

The actin depolymerizing agent, latrunculin B, inhibited zygotic polarization (F=45.14, P<0.001)(Fig. 4B,F) relative to controls (Fig. 4A,F) at all concentrations tested when applied continuously from 20 minutes after fertilization. Surprisingly, bearing in mind the yeast morphogenesis checkpoint, zygotes underwent complete mitosis in the presence of 0.1 μM latrunculin B, at a rate that was not significantly different (F=0.01, P<0.92) from that of untreated zygotes(Fig. 4E), to give apolar binucleate zygotes (Fig. 4B). Moreover, even at higher latrunculin B concentrations (1.0 μM), zygotes still displayed condensed nuclei (Fig. 4C,D) that had progressed through G2/M to become arrested in M phase, although nuclei had been displaced towards the cell cortex from their normally central location (Fig. 4C).

Figs 2 and 3 suggest that zygotic S-phase entry in F. serratus requires a late G1/S-phase Ca²⁺elevation. To provide further evidence for the importance of post-fertilization Ca²⁺ signals, zygotes were loaded either before S phase (1 hour after fertilization) or after S phase [7 hours after fertilization (Corellou et al.,2001b)] with one of four separate treatments: the high affinity(K_d ∼200 nM) Ca²⁺ buffer BAPTA dextran together with FITC dextran; the moderate affinity (K_d ∼1.4 μM)Ca²⁺ buffer Br₂BAPTA together with FITC dextran; FITC dextran alone; or the very low affinity (K_d ∼40 μM)Ca²⁺ buffer-but very high affinity transition metal buffer-TPEN(Arslan et al., 1985).

In full support of our single-cell imaging results (Figs 2 and 3), logistic regressions fitted to these data (G² relative to full model=1.51, P=0.22)show that both moderate and high-affinity Ca²⁺buffers-Br₂BAPTA and BAPTA dextran-preferentially inhibit cell cycle progression when added before S phase, with BAPTA dextran being significantly more effective than Br₂BAPTA(Fig. 5A).

Although we have used BAPTA dextran and Br BAPTA as Ca²⁺buffers, they also show reasonably high affinities for other biologically important species: notably H⁺ and the heavy metal ion Zn²⁺. Although brown algal zygotes have very strong intracellular pH buffering mechanisms (Gibbon and Kropf,1993), we nonetheless used the cell-permeant buffer TPEN(Arslan et al., 1985) as a control against any non-specific effects of BAPTA-based buffers. TPEN has much higher affinities for heavy metal ions, much lower affinity for Ca²⁺ (Kao et al.,1990) and similar pKa values(Anderegg and Wenk, 1967) to the BAPTA-based buffers (Tsien,1980). Consistent with our hypothesis that the S-phase-specific effects of BAPTA-based buffers are mediated through Ca²⁺ binding,we did not observe preferential pre-S-phase cell cycle inhibition at TPEN concentrations with heavy metal buffering capacities equivalent to the BAPTA dextran and Br₂BAPTA concentrations used in the present study(Fig. 5A)(Kao et al., 1990).

Our single-cell imaging results show that Ca²⁺ buffers inhibit S-phase entry (Fig. 3F) and zygotic polarization (Fig. 3D)when introduced before S phase. Because S-phase entry and zygotic polarization in F. serratus are also dependent on late G1/S-phase CDK activity(Corellou et al., 2001b), it is possible that pre-S-phase Ca²⁺ elevations act solely to stimulate CDK activity (see also results for Fig. 3E, above). As no single-cell assay has been developed to measure CDK activity in individual F. serratus zygotes, it was not possible to test this hypothesis directly. Instead, we used multivariate analysis of residual length/width ratios (see Materials and methods) to assess the degree to which polarization and cell cycle progression relied on a common Ca²⁺-dependent process.

The measured length-to-width ratios for FITC- or Br₂BAPTA-loaded populations do not differ significantly (F<0.42, P>0.52) from the predicted values for full dependence of polarization on cell cycle progression, regardless of the time at which buffer was loaded into zygotes(Fig. 5B,C). FITC- and Br₂BAPTA-loaded zygotes that had reached M phase had, therefore,invariably polarized (Fig. 5D). However, BAPTA dextran loaded before S phase inhibited zygotic polarization(F=9.11, P<0.01) to a degree that could not be entirely explained by the effect of BAPTA dextran on cell cycle rate alone(Fig. 5B) and that, at its most extreme, manifested itself as a complete uncoupling of nuclear division and polarization (Fig. 5D). Crucially, this additional inhibitory effect was not observed(Fig. 5C,D) when BAPTA dextran was loaded after S phase, 7 hours after fertilization (F=0.712, P=0.492), ruling out effects on post-S-phase Ca²⁺dynamics, such as inhibition of the tip-high cytosolic [Ca²⁺]gradient which arises just prior to rhizoid germination(Berger and Brownlee,1993).

The heavy metal buffer TPEN was again used as a control and additionally inhibited polarization (F>24.65, P<0.001) when applied either before or after S phase (Fig. 5B-D), providing further evidence for an S-phase-specific Ca²⁺-mediated signalling event, as inhibited by BAPTA dextran, as opposed to a general nutrient requirement for polarized growth, as inhibited by TPEN.

Taken together, the simplest explanation for our data is one in which post-fertilization Ca²⁺ elevations trigger both the G1/S-phase checkpoint and a separate, G1/S-phase checkpoint-independent, early polarization pathway. Both processes are required for zygotic polarization,but the G1/S-phase checkpoint stimulates cell cycle progression on its own(Fig. 6A). Given our single-cell imaging results (Fig. 3), the most obvious candidate for the second, G1/S-phase checkpoint-independent, pathway is cortical actin nucleation, which is known to be required for zygotic polarization(Pu et al., 2000) and which we have shown to be Ca²⁺ dependent(Fig. 3D), but not CDK dependent (Fig. 3E). The observed uncoupling of cell cycle progression from polarization(Fig. 5B,D) also confirms the single-cell imaging results of Fig. 4 in indicating that there is no actin-based morphogenesis checkpoint for cell cycle progression in the F. serratus zygote.

Previous studies on cell cycle regulation in F. serratus zygotes have used whole-population assays to measure S-phase progression(Corellou et al., 2000) and CDK activity (Corellou et al.,2001b). Unfortunately, these bulk assays are not appropriate for intracellular Ca²⁺ signalling studies, because Ca²⁺buffers and indicators may only be loaded into a small fraction of the cells present in a population (Bothwell et al.,2006). To circumvent these problems in F. serratuszygotes, we have adapted a GFP-PCNA single-cell assay for DNA replication(Figs 1, 3) and have used statistical analyses of polarization and cell cycle progression to test models of coupling between Ca²⁺-dependent developmental pathways(Fig. 5).

In all eukaryotic systems studied to date, fertilization is accompanied by intracellular [Ca²⁺] elevations. In animal systems, these elevations generally take the form of fast [Ca²⁺]_cytwaves (Gilkey et al., 1978),which cross the newly fertilized egg at concentrations between 2 μM(Miyazaki, 1989) and 10 μM(Brownlee and Dale, 1990; Speksnijder et al., 1990), and at speeds of around 10 μm/second (Jaffe and Creton, 1998). These fast [Ca²⁺]_cytwaves trigger two groups of events: zygote construction and cell cycle reactivation (Rauh et al.,2005; Whitaker,2006a).

Evidence for such large and fast [Ca²⁺]_cyt waves has,however, been less forthcoming in non-animal systems, with slower[Ca²⁺]_cyt wave speeds of only around 1 μm/second having been recorded during in vitro fertilization in maize(Antoine et al., 2000; Digonnet et al., 1997) and with only localized cortical [Ca²⁺]_cyt elevations of around 300 nM having been previously reported in F. serratus(Roberts et al., 1994).

Here, we have used two-photon excitation for better depth penetration(Bush et al., 2007) into the highly pigmented F. serratus zygote to reveal the presence of a slow global post-fertilization [Ca²⁺]_cyt elevation(Fig. 2C,D). The polarized secretion of cell wall components that occurs within the first few minutes of fertilization may also indicate the presence of a low amplitude `fast',post-fertilization cortical [Ca²⁺]_cyt wave, possibly arising from the site of sperm entry, and we cannot exclude the possibility that we have been unable to detect such a [Ca²⁺]_cytelevation in F. serratus zygotes. However, the pattern of slow,global [Ca²⁺]_cyt elevation that we report here had a similar time course to the more localized [Ca²⁺]_cytelevations previously reported (Roberts et al., 1994) and is consistent with the slow[Ca²⁺]_cyt elevations known to be associated with developmental events in a number of other organisms(Jaffe and Creton, 1998).

We have demonstrated that BAPTA-based buffers are able to inhibit S phase(Fig. 3F) by buffering Ca²⁺ (Fig. 5A). This Ca²⁺ dependency of zygotic S phase comes after the sperm pronucleus has migrated to the egg pronucleus (Fig. 3A,B), but before any measurable DNA replication(Fig. 3F), and strongly suggests a Ca²⁺ signalling requirement for S-phase onset.

F. serratus eggs, like sea urchin eggs and somatic cells, are arrested in G0/G1 (Corellou et al.,2001a) and must be driven into S phase following fertilization. This task is performed by the cyclin E-cdk2 complex in somatic cells(Dulic et al., 1992; Koff et al., 1992) and probably in sea urchin embryos(Schnackenberg and Marzluff,2002). It is, therefore, interesting to note that[Ca²⁺] elevations are able to activate both partners in the sea urchin cyclin E-cdk2 complex. For cyclins, Ca²⁺-dependent PKC can switch on a Na/H antiporter (Epel,1990; Swann and Whitaker,1985), which alkalinizes the zygote cytosol(Johnson and Epel, 1976) and leads to pH-dependent cyclin synthesis(Grainger et al., 1979; Winkler et al., 1980). For CDKs, [Ca²⁺] elevations trigger the MAP kinase ERK1(Philipova and Whitaker,1998), which is thought to activate cdk2 to bind the cyclin E complex (Philipova et al.,2005). Similar MAPK activation is also needed for cdk2 activity in somatic cells (Keenan et al.,2001).

Despite this potential involvement of Ca²⁺ in driving cells out of G0/G1 arrest, [Ca²⁺] elevations associated with pronuclear fusion or S-phase onset have not been well documented. There is some evidence that [Ca²⁺] elevations are associated with embryonic S-phase onset in sea urchins (Poenie et al.,1985) and [Ca²⁺] elevations are required for, but have not been visualized in, somatic S-phase(Kao et al., 1990), leading to the suggestion that they may be restricted to microdomains(Whitaker, 2006b). Our observation of a localized nuclear [Ca²⁺] elevation around the time of pronuclear fusion (Fig. 2D)offers some support for this microdomain suggestion.

Furthermore, we propose that the timing of [Ca²⁺] elevations may have a bearing on the timing of other cell cycle regulatory events. For example, sea urchin zygotes enter S-phase around 20 minutes after fertilization and an immediate post-fertilization Ca²⁺ wave may be sufficient to activate their cyclin E-cdk2 complexes, which are present in an inactive state before fertilization. F. serratus zygotes, however,not only have much slower rates of cell cycle progression, but also demonstrate a unique translational regulation of CDKs, which are not synthesized until several hours after fertilization(Corellou et al., 2001a). This suggests that any immediate post-fertilization [Ca²⁺] elevation could not lead to subsequent S-phase activation by Ca²⁺-dependent MAP kinase stimulation of cdk2-cyclin, as the CDK partner would not be present. This may explain the need for a later, more prolonged[Ca²⁺] elevation of the type reported in Fig. 2D.

It is tempting to speculate that the post-fertilization[Ca²⁺]_cyt elevation that we have observed(Fig. 2C,D) acts to trigger one set of zygote activation events, such as actin nucleation(Muto and Mikoshiba, 1998),and the later nuclear Ca²⁺ elevation(Fig. 2C,D) acts to trigger another set, such as cell cycle progression. Nonetheless, although our Ca²⁺ buffer loading results suggest that the post-fertilization[Ca²⁺]_cyt elevation is not sufficient to drive S phase,the roles played by distinct Ca²⁺ elevations remain to be determined. Our results do, however, suggest that such distinct Ca²⁺-dependent pathways exist (Figs 3, 5) and the simplest model for our data is one in which actin nucleation and the G1/S-phase checkpoint have independent Ca²⁺ requirements(Fig. 6A).

In budding yeast, the joint activation of cell cycle progression and polarization by the G1/S-phase checkpoint is followed by their further coordination at the well-established morphogenesis checkpoint, in which actin localization at the bud collar is needed to bring about release of cell cycle arrest at G2/M (Keaton and Lew,2006). However, our results clearly indicate that a morphogenesis checkpoint does not operate in the F. serratus zygote. The evidence for this is twofold. First, exogenous Ca²⁺ buffers are able to inhibit actin localization and zygotic polarization while allowing cell division, albeit at relatively low frequencies(Fig. 5D), which indicates that there is no absolute requirement for an actin-localization based checkpoint. Second, there is no effect on cell cycle progression to M phase when actin polymerization is inhibited sufficiently to block polarization completely(Fig. 4), which is both direct evidence against the involvement of a morphogenesis checkpoint and a surprising finding, given the demonstrated ability of actin to regulate developmental Ca²⁺ elevations in F. serratus zygotes(Pu et al., 2000).

The absence of a strict actin-based polarity checkpoint in the F. serratus zygote presents a developmental model(Fig. 6B) that is distinct from both the tightly controlled yeast morphogenesis system(Fig. 6C) and that of Drosophila, in which early embryonic cell cycle checkpoints are absent (Fig. 6D). This emphasises the diversity of cell cycle control in relation to polarization in different organisms.

We are grateful to the BBSRC and NERC for financial support. M. Tristan Fabre provided help with the Texas Red phalloidin staining, Dr C. Cardoso (Max Delbruck Center for Molecular Medicine, Berlin, Germany) kindly provided the GFP-PCNA construct and we thank Prof. Michael Whitaker (University of Newcastle, Newcastle-upon-Tyne, UK) for a number of helpful discussions.J.H.B. is funded by a Leverhulme Trust Early Career Fellowship and J.K. is funded by a Royal Society Dorothy Hodgkin Fellowship.