A Compartmental Approach to the Mechanism of Calcification in Hermatypic Corals

Biocalcification is one of the most important biological process in the living world. Although an impressive amount of information has been produced in the last few years, calcification processes and, more specifically, the location and mechanism of Ca²⁺ transport largely remain a biological enigma (Wilbur and Simkiss, 1979; McConnaughey, 1989; Simkiss and Wilbur, 1989; Cameron, 1990). It appears that there are two possible routes for Ca²⁺ transport during calcification: (1) paracellular, where ions pass through intercellular spaces and (2) transcellular. The paracellular pathway has been hypothesized to explain the diffusional entry in algae (Borowitzka, 1982) and the movement of Ca²⁺ across the mantle of molluscs (Wheeler, 1992), while transcellular movement has been suggested in corals (Barnes and Chalker, 1990) and crustaceans (Roer, 1980; Cameron, 1990).

One of the major calcifying groups of organisms is the scleractinian corals. The rate of calcification of a coral reef is assumed to be around 10 kg CaCO₃ m^-2 year^-1 (Chave et al. 1975). It has been calculated that, at peak calcification rate, cells of the calicoblastic epithelium of the fast-growing distal regions of the staghorn coral (Acropora sp.) must transport each hour all the calcium contained in 50–100 times their own volume (Gladfelter, 1984; Barnes and Chalker, 1990). In spite of this high flux of calcium, mechanisms of Ca²⁺ transport are still poorly understood and Ca²⁺ compartments in tissue have never been studied. Several lines of evidence indicate that calcification in corals depends upon a specific, energy-requiring transport mechanism (Chalker, 1976; Krishnaveni et al. 1989). The presence of a high-affinity Ca²⁺-ATPase has been demonstrated (Isa et al. 1980; Ip et al. 1991). Carbonic anhydrase is also thought to control calcification rate (Silverton, 1991). However, kinetic studies of Ca²⁺ transport by coral tissues have been limited.

The main reason for this lack of data is the difficulty of using radioisotopes to study coral calcification. Buddemeier and Kinzie (1976) discussed the problems associated with the use of radioisotope techniques in the measurement of coral calcification. They identified the high variance of the results as the major experimental problem. This is due to isotopic exchange phenomena between ⁴⁵Ca in sea water and CaCO₃ in the porous aragonite skeleton during incubation (Clausen and Roth, 1975; Barnes and Crossland, 1977) and processing of samples after incubation (Crossland and Barnes, 1977). In order to overcome these problems, we have used cloned corals (‘microcolonies’) of the branching scleractinian coral Stylophora pistillata with no skeletal surfaces exposed to the radioisotope-labelled incubation medium and experimental protocols which enable us to obtain reproducible Ca²⁺ uptake measurements (Tambutté et al. 1995). These improvements have enabled us to demonstrate the presence of a large extracellular compartment in S. pistillata, which probably corresponds to the coelenteric cavity (Tambutté et al. 1995).

Kinetic analyses of Ca²⁺ influx and efflux have proved to be useful tools to study Ca²⁺ compartments in cell cultures (Borle, 1969a,b, 1990; Claret-Berthon et al. 1977) and in intact tissues (Van Breemen et al. 1979). In the present study, we used these two complementary procedures to characterize additional Ca²⁺ compartments within coral colonies and the movement of Ca²⁺ between them. The kinetic and pharmacological characteristics of these compartments are described here.

Microcolonies were propagated in the laboratory from small fragments of Stylophora pistillata collected at a depth of 5 m in front of the Marine Science Station, Gulf of Aqaba, Jordan. Colonies were packed in humidified plastic bags and transported to our laboratory (14 h transport time). Corals were stored in an aquarium (300 l) supplied with sea water from the Mediterranean Sea (exchange rate 2 % h^-1), heated to 26±0.1 °C and illuminated with constant irradiance of 175 μmol photons m^-2 s^-1 using metal halide lamps (Philips HQI-TS, 400 W) on a 12 h:12 h photoperiod. Microcolony propagation has been described by Tambutté et al. (1995). Briefly, terminal portions of branches (6–10 mm long) were cut from parent colonies and placed on a nylon net (1 mmX1 mm mesh) in the same conditions of light and temperature as parent colonies. After about 1 month, coral fragments became entirely covered with new tissues.

Two different protocols have been used to determine the kinetic characteristics of Ca²⁺ compartments in microcolonies. The first was based on uptake experiments. ⁴⁵Ca was measured in microcolonies after a 30 min wash-out of the coelenteric cavity. The second was based on efflux experiments, radioactivity being measured in the efflux medium.

Measurements were made at equivalent times of day in order to avoid possible variation caused by endogenous circadian rhythms (Buddemeier and Kinzie, 1976). Microcolonies were placed in plastic holders and incubated for 5–300 min in beakers (6 ml) containing 240 kBq of ⁴⁵Ca (as CaCl₂, 1.38 MBq ml^-1, New England Nuclear) dissolved in sea water filtered using 0.45 μm Millipore membranes (FSW) as described by Tambutté et al. (1995). Water movement was maintained by magnetic stirring bars. Incubations of varying duration were carried out on at least three samples under light and temperature conditions similar to those described during culture. Samples (100 μl) of sea water were removed during each incubation for the determination of specific radioactivity.

At the end of the labelling period, each holder and its microcolony was immersed for 20 s in a beaker containing 600 ml of FSW, then rinsed five times with 5 ml of ice-cold glycine high-calcium medium (50 mmol l^-1 CaCl₂, 950 mmol l^-1 glycine, pH adjusted to 8.2) to prevent further Ca²⁺ uptake and to reduce, by isotopic dilution, the ⁴⁵Ca adsorbed onto the external surface of both the microcolony and the holder. The total duration of the rinsing procedure was less than 1 min. Labelled microcolonies were then put in a beaker containing 20 ml of FSW for 30 min to monitor ⁴⁵Ca efflux. The total amount of radioisotope released into the FSW was used to determine the size of the coelenteric cavity (Tambutté et al. 1995). Upon completion of the efflux procedure, microcolony tissues were dissolved completely over a period of 20 min in 1 ml of 2 mol l^-1 NaOH at 90 °C. The supernatant (hereafter termed the ‘NaOH-soluble pool’) was collected and the skeleton was rinsed first in 1 ml of distilled water and, subsequently, five times in 5 ml of FSW. The first rinse solution was added to the NaOH pool, the remaining five washes were discarded since they did not contain proteins. Finally, the skeletons were dried and dissolved in 1.5 ml of 12 mol l^-1 HCl overnight (‘HCl-soluble pool’). Radioactive samples were added to 4 ml of Luma-gel (Packard) after neutralization and β emissions measured using a liquid scintillation counter (Tricarb, 1600 CA Packard).

In order to determine the calcification rate during short incubations, the rinsing procedure was bypassed by using EGTA. At the end of the labelling period, and after a short immersion in 600 ml of FSW, the microcolony was immediately immersed for 30 min in 10 ml of 5 mmol l^-1 EGTA in 1 mol l^-1 NaOH in order to chelate Ca²⁺ absorbed by the coelenteron and the tissues. After this treatment, the microcolony was processed as described above.

Microcolonies were incubated with ⁴⁵Ca for 3 h at 25 °C under an irradiance of 675 μmol photons m^-2 s^-1. Colonies were then transferred to one of the three efflux media, Ca²⁺-free artificial sea water (0Ca²⁺/ASW), 0Ca²⁺/ASW with 5 mmol l^-1 EGTA (0Ca²⁺/EGTA) or FSW. Effluxes into 0Ca²⁺/EGTA and FSW were conducted either on ice or at 25 °C. Efflux into 0Ca²⁺/ASW was on ice only. Efflux was performed by sequential incubations of microcolonies at increasing intervals over 180 min into beakers containing 20 ml of medium. Tissues were then removed as described above, the remaining skeletons were returned to efflux medium and transferred in a manner similar to the intact microcolonies. Skeletons were dissolved as described previously. Samples of all efflux solutions and the NaOH and HCl pools were counted in a liquid scintillation counter.

For experiments with varying Ca²⁺ concentrations, ASW was prepared from distilled water according to the method of Allemand et al. (1984). CaCl₂ was replaced by NaCl in order to maintain constant osmolarity. DIDS (4,4’-diisothiocyanatostilbene-2,2’-disulphonic acid), (+)-verapamil, (-)-verapamil, (+)-methoxyverapamil or (-)-D600, (-)-methoxyverapamil or (+)-D600, diltiazem, ethoxyzolamide, thapsigargin, cytochalasin B, colchicine and Bay K 8644 were dissolved in DMSO (dimethyl sulphoxide). Cadmium chloride, nickel chloride, NaCN and ω-conotoxin were dissolved in water. Nifedipine was dissolved in acetone. Lanthanum chloride was dissolved in ethanol; flunarizine was dissolved in methanol. The final solvent concentration was never more than 1 % (v/v). Preliminary experiments showed that this concentration had no effect on Ca²⁺ uptake (results not shown). Except for (-)-verapamil, (-)-D600 and Bay K 8644 (generous gift from Dr Barhanin), all chemicals were obtained from Sigma and were of analytical grade. When inhibitors were used, a 15 min pre-incubation step was included. The inhibitor was continuously present at the same concentration during pre-incubation, incubation (1 h) and efflux (30 min) periods.

Protein concentrations were measured using the method of Lowry et al. (1951) in an autoanalyzer (Alliance Instruments) using bovine serum albumin as the standard. Results are expressed as nmol Ca²⁺ mg^-1 protein in the NaOH-soluble pool (milligrams of protein per microcolony) and represent means ± S.D. for at least three measurements.

The rates of Ca²⁺ uptake by the coelenteric compartment, the NaOH-soluble pool and the HCl-soluble pool are depicted in Fig. 1. Ca²⁺ uptake by both the coelenteric compartment and the NaOH-soluble pool displayed saturable kinetics (Fig. 1A,B). Isotopic equilibrium was reached at about 10 min and 2 h in the coelenteron and in the NaOH-soluble compartment respectively. The equilibrium value corresponds to the size of the exchangeable Ca²⁺ compartment, i.e. 72.88±1.35 nmol Ca²⁺ mg^-1 protein for the coelenteric compartment and 7.12±0.72 nmol mg^-1 protein for the NaOH-soluble pool. The half-times of exchange (t_1/2) of these compartments, are, respectively, 4 and 20 min as determined by semi-logarithmic treatment of the data (insets of Fig. 1A,B). From the value of t_1/2, the respective rate constants can be calculated to be 0.17 and 0.034 min^-1. The initial rate of Ca²⁺ flux is obtained from the product of compartment size and k. The value determined for the initial rate of Ca²⁺ flux from sea water into the coelenteric compartment is therefore 12 629 pmol mg^-1 protein min^-1, and that from sea water (and the coelenteric compartment) into the NaOH-soluble pool is 224 pmol mg^-1 protein min^-1 (Table 1).

The time course of Ca²⁺ deposition in the skeletal compartment (HCl-soluble pool) was linear until at least 5 h (Fig. 1C). No lag phase could be detected under our experimental conditions. The calcification rate appears to be 975 pmol mg^-1 protein min^-1. In order to study the onset of Ca²⁺ deposition into the skeleton, we performed short-term incubations from 1 to 90 min. This procedure does not allow us to determine Ca²⁺ uptake by coelenteron and tissue. It can be seen in Fig. 2 that ⁴⁵Ca incorporation into the skeleton starts after a very short lag phase of less than 2 min and remains linear over the time range studied.

Fig. 3 shows ⁴⁵Ca uptake at various external Ca²⁺ concentrations ranging from 0 to 20 mmol l^-1. While Ca²⁺ uptake by the coelenteric compartment is linearly correlated to the external Ca²⁺ concentration (Fig. 3A), uptake into the NaOH-soluble pool displays biphasic kinetics (Fig. 3B). The Ca²⁺ uptake is curvilinear at low Ca²⁺ concentrations (below 4 mmol l^-1) and becomes linear at higher values. Over the entire substrate concentration range, calcium uptake follows a combination of Michaelis–Menten kinetics plus an apparent adsorption or more likely a diffusional component, since microcolonies had been rinsed in a high-calcium medium to remove unspecific labelling by isotopic dilution. By subtracting the apparent diffusion (determined as a regression line parallel to the total flux of Ca²⁺ between 5 and 20 mmol l^-1) from the total uptake, the flux can be broken into two components. Ca²⁺ deposition in coral skeleton showed typical saturable Michaelis–Menten kinetics, the plateau being reached at about 9 mmol l^-1 (Fig. 3C). These results support the view that a transport system for Ca²⁺ is involved in at least one step of the biocalcification mechanism.

In order to characterize the three Ca²⁺ compartments identified by uptake experiments and to determine the ion transport mechanism, we tested the effects of different drugs known for their ability to inhibit specific metabolic or transport mechanisms. The coelenteric compartment has previously been described as being insensitive to inhibitors of cell metabolism and ion transport (Tambutté et al. 1995). In the presence of 1 mmol l^-1 sodium cyanide, a well-known mitochondrial inhibitor, Ca²⁺ uptake by the NaOH-soluble pool was unchanged (t-test, P>0.05), whereas Ca²⁺ deposition in skeletal structures was inhibited (Fig. 4) (t-test, P<0.05). The inhibition of Ca²⁺ deposition demonstrates that the formation of calcium carbonate is an energy-requiring process.

To test whether Ca²⁺ uptake was through voltage-dependent Ca²⁺ channels, we tested the effect of organic Ca²⁺ channel inhibitors [100 μmol l^-1 (+)-and (–)-verapamil, (+)-and (–)-D600, diltiazem, nifedipine, 100 μmol l^-1 flunarizine, 2 μmol l^-1w-conotoxin] or heavy metals (200 μmol l^-1 lanthanum, 1 mmol l^-1 cadmium or nickel). Fig. 4 shows that most of these drugs stimulated by a factor of 1.2–2 the Ca²⁺ loading of the NaOH-soluble pool, whereas Ca²⁺ deposition in the HCl-soluble pool was almost entirely inhibited (t-test, P<0.05) except by w-conotoxin, flunarizine and lanthanum. No differences were found between the effects of the stereoisomers of verapamil and D600. A dose–response experiment was performed with (+)-verapamil. Half-maximal inhibition was obtained for a verapamil concentration of 16 μmol l^-1 (Fig. 5). Surprisingly, Bay K 8644, a Ca²⁺ channel agonist (Spedding and Paoletti, 1992), had no significant effect on Ca²⁺ uptake (t-test, P>0.05).

We also tested DIDS, an inhibitor of anion transport systems (Cabantchik and Greger, 1992), and thapsigargin, an inhibitor of endoplasmic reticulum Ca²⁺ uptake (Wictome et al. 1992).

It can be seen in Fig. 4 that 300 μmol l^-1 DIDS greatly inhibited Ca²⁺ deposition in the HCl-soluble pool (t-test, P<0.05) but had no effect on the loading of the NaOH-soluble Ca²⁺ pool (t-test, P>0.05). Thapsigargin (5 μmol l^-1) had no effect on either pool (t -test, P>0.05).

The effect of the carbonic anhydrase inhibitor ethoxyzolamide (300 μmol l^-1) was investigated. This drug did not affect the uptake of Ca²⁺ by the NaOH-soluble pool (t-test, P>0.05) but inhibited Ca²⁺ deposition in the skeleton (t-test, P<0.05). Finally, we tested the effect of inhibitors of microtubule and microfilament polymerization, colchicine (250 μmol l^-1) and cytochalasin B (20 μmol l^-1). A mixture of these drugs had no significant effect on the NaOH-soluble pool (t-test, P>0.05), but prevented Ca²⁺ incorporation into the skeleton (t-test, P<0.05).

A typical 400 min efflux curve is shown in the inset of Fig. 6. The data points (solid line) were obtained by sequentially adding back ⁴⁵Ca lost in the efflux solutions and NaOH-soluble pool to that remaining in the skeleton (HCl-soluble pool). The efflux curve was then ‘peeled’ by removing exponential components starting with the S2 (skeletal) compartment. A small S1 compartment (results not shown) was detected during the first few minutes of skeleton efflux. The colony (C) compartments are characterized after subtracting the S2 compartment. Fig. 6 shows the first 60 min of colony efflux to show the resolution of the C1 to C4 compartments. The values for t_1/2 and the sizes of each compartment are presented in Table 2 as measured under various efflux conditions. It should be noted that the t_1/2 values do not necessarily represent those of relevant physiological processes; they are used primarily to identify compartments. The C1 compartment has been associated with extracolonial water carried over on the plastic holders and on the outside of the colony. Similar rate constants were obtained with the holders alone (data not shown). The t_1/2 values for C1 were very similar for the different efflux conditions, averaging 14.6±1.9 s (mean ± S.D., N=4). The size of the C1 compartment varied, although this probably had little physiological significance. The C2 compartment had similar t_1/2 values (2.3±0.46 min) and also a similar size (56.0±10.9 nmol Ca²⁺ mg^-1 protein) regardless of efflux conditions. The rate constants were slightly higher (lower t_1/2 values) at 25 °C, consistent with the effect of temperature on diffusion. This compartment is thought to be coelenteric; it was insensitive to D600 (data not shown) and had a size consistent with that determined by uptake kinetics (62.5 nmol Ca²⁺ mg^-1 protein; see above). The C3 compartment was also insensitive to D600 and its size and time constant did not vary widely with efflux conditions. Again, rate constants were slightly higher at 25 °C; the mean value of t_1/2 was 15.3±8.1 min. The compartment size varied from 8.6 to 42.2 nmol Ca²⁺ mg^-1 protein.

The size of the C4 compartment was sensitive to 10 μmol l^-1 D600 (Fig. 7) and this compartment showed substantial variation in both rate constant and size depending upon the efflux conditions. In ice-cold 0Ca/EGTA, t_1/2 was 76.1 min; it was only slightly shorter at 25 °C (72.5 min). The size of this compartment was similar, around 130 nmol Ca²⁺ mg^-1 protein, in this medium at 0 °C and 25 °C. Elimination of EGTA decreased the rate constant and size of the C4 compartment: in FSW, t_1/2 values of hundreds of minutes were obtained and compartment size was much smaller. To determine whether the C4 compartment corresponded to the intracellular pool, its size was calculated at the time when tissue Ca²⁺ depletion was initiated (or the end of colony efflux; C4₁₈₀) and compared with the size of the NaOH-soluble pool (Table 3). The NaOH-soluble pool was always smaller than C4₁₈₀. The discrepancy was smallest when 0Ca/ASW was used as the efflux medium.

The skeletal effluxes yielded two components, a small, relatively fast S1 compartment and a larger, much slower S2 compartment. The t_1/2 value of S1 generally varied from 7 to 20 min: however, it was 1.3 min in 0Ca/ASW (Table 2). In the presence of EGTA, the compartment was larger (approximately 13 nmol Ca²⁺ mg^-1 protein) compared with that in media without EGTA (1.8±0.4 nmol Ca²⁺ mg^-1 protein). The S2 compartment also showed an effect of EGTA. When in the efflux medium, the t_1/2 value was shorter and the size smaller compared with these values in FSW. Although the rate constants were similar, the compartment size in FSW was smaller at 25 °C. The 0Ca/ASW medium produced an S2 component that had an intermediate rate constant and a size more similar to those obtained in FSW. In Fig. 7, it can be seen that the S2 compartment corresponds well with the HCl-soluble pool.

Using the 0Ca/EGTA efflux protocol, the effects of 10 μmol l^-1 D600 on the sizes of the NaOH-and HCl-soluble pools and their putative compartmental analogues were compared. D600 inhibited uptake into the S2 and HCl-soluble compartments by approximately 60 %; the absolute values of the pool sizes are also in close agreement (Fig. 7). The C4 Ca²⁺ content (at time=0) was much higher than that of the NaOH-soluble pool, but both were D600-sensitive. C4 Ca²⁺ efflux was inhibited by 64.2 % while efflux from the NaOH-soluble pool was inhibited by 49.0 %.

The Ca²⁺ transport mechanism of calcifying systems has received relatively little attention (Wilbur, 1976; Simkiss and Wilbur, 1989; Neufeld and Cameron, 1994). In scleractinian corals, most authors have focused their research on Ca²⁺ incorporation into the skeleton without studying calcium homeostasis in tissue. The present study shows that calcium reaches the site of calcification by an energy-requiring process through a transcellular pathway (see Fig. 8). Skeletal Ca²⁺ deposition displays saturable kinetics with respect to the calcium concentration in sea water, implying a carrier-mediated step. This result confirms a previous report (Chalker, 1976) suggesting an intracellular pathway for Ca²⁺ transport in corals, differentiating this mechanism from those found in the posterior caecum of the amphipod Orchestia cavimana (Meyran et al. 1984) or mollusc mantle (Wheeler, 1992).

Calcification is inhibited by organic or inorganic Ca²⁺ channel inhibitors (Figs 4, 5), as previously hypothesized (Mueller, 1984). The presence of voltage-dependent Ca²⁺ channels has been demonstrated in numerous marine invertebrates (Adams and Gage, 1979; Schmidt et al. 1982; Bilbaut et al. 1988) but the involvement of these channels in biocalcification has been demonstrated in only a few calcifying organisms such as echinoids (see review by Dubois and Chen, 1989) and red coral (Allemand and Grillo, 1992). Our present results show that such Ca²⁺ channels are also present in scleractinian corals and are responsible for the passive entry of Ca²⁺ into cells through which the bulk, if not all, of the skeletal Ca²⁺ must pass.

From pharmacological data, it appears that ⁴⁵Ca deposition into the skeleton is highly sensitive (inhibition greater than 90 %; IC₅₀ for verapamil of 16 μmol l^-1) to both stereoisomers of phenylalkylamines (verapamil and D600) and to benzothiazepines (diltiazem). Dihydropyridine (nifedipine) is also a very potent inhibitor (inhibition of about 80 %). In contrast, calcification is only weakly sensitive to typical inhibitors of N channels (ω-conotoxin) or T channels (flunarizine). This sensitivity pattern suggests that Ca²⁺ transport for calcification is performed by a Ca²⁺ channel whose pharmacological characteristics are close to those of the L-type of voltage-sensitive Ca²⁺ channel described in mammalian cells (Hosey and Lazdunski, 1988; Spedding and Paoletti, 1992). Because they display a prolonged opening time, Ca²⁺ channels of this type appear to be involved primarily in the regulation of intracellular Ca²⁺ homeostasis in a wide range of tissue including non-excitable cells (Miller and Fox, 1990; Pietrobon et al. 1990). The higher sensitivity to cadmium in comparison with nickel is also in agreement with this conclusion (Miller and Fox, 1990). Surprisingly, lanthanum did not affect calcification. Lanthanum-insensitive Ca²⁺ channels have also been described in fish (Marshall et al. 1992). Further studies are planned in order to characterize better the Ca²⁺ channel involved in calcification.

These data demonstrate that Ca²⁺ uptake by coral tissue is an obligatory step in skeleton formation. This conclusion contradicts the hypothesis of Johnston (1980), who proposed that Ca²⁺ was transported by vesicles via a paracellular route. This author suggested that the zonular cell junctions present between calicoblastic cells are dynamic structures that may open and re-form to allow passage of Ca²⁺-containing vesicles into the subcalicoblastic space. The absence of a diffusional component argues strongly against any continuity, even temporary, between the coelenteric fluid and the calcification site (Fig. 3C).

The transcellular movement of Ca²⁺ could be performed by transport via intracellular vesicles and their subsequent discharge at the mineralizing front (Hayes and Goreau, 1977; Johnston, 1980) or by Ca²⁺-binding proteins (Simkiss, 1976; Bronner, 1990). The insensitivity of the calcification process to thapsigargin, a specific inhibitor of the Ca²⁺-ATPase that is thought to load intracellular Ca²⁺ stores (Wictome et al. 1992), suggests that transcellular transport of Ca²⁺ is not achieved by intracellular vesicles. Furthermore, the sensitivity to Ca²⁺ channel inhibitors rules out the hypothesis that pinocytic vacuoles transport Ca²⁺. The presence of Ca²⁺-binding proteins as bulk intracellular carriers (Bronner, 1990) remains a viable hypothesis. Experiments are under way to address this point.

The anion carrier inhibitor DIDS was also a strong inhibitor of calcification. A similar finding was reported for spicule formation in echinoids by Mitsunaga et al. (1986a) and in red coral for spicule and axial skeleton synthesis by Allemand and Grillo (1992). As in these reports, the effect of DIDS on the calcification rate may be interpreted either as inhibition of an electrogenic Cl^- flux coupled with Ca²⁺ flux (Yasumasu et al. 1985) or as a decrease in the production of the carbonate moiety of CaCO₃ subsequent to the inhibition of HCO₃^- flux (Mitsunaga et al. 1986b; Allemand and Grillo, 1992). Further experiments are under way to investigate anion transport.

The involvement of carbonic anhydrase in biocalcification processes, including those in hermatypic corals, is well documented (Goreau, 1959; Istin and Girard, 1970; Mitsunaga et al. 1986b; Tuan et al. 1986; Kingsley and Watabe, 1987; Allemand and Grillo, 1992). Isa and Yamazato (1984) have shown using histochemical methods that carbonic anhydrase is localized in the calicoblastic epithelium. Our results support these studies and show that this enzyme is involved in Ca²⁺ deposition, whereas Ca²⁺ uptake into the NaOH-soluble pool appears to be insensitive to the sulphonamide tested (ethoxyzolamide).

The calcification rate is sensitive to a combination of two inhibitors of actin and tubulin polymerization, cytochalasin B and colchicin. Fujino et al. (1985) have previously shown that inhibitors of tubulin assembly, colchicin and podophyllotoxin, inhibit Ca²⁺ uptake and deposition by cultured primary mesenchyme cells of echinoderms. This suggests that actin and tubulin assembly are implicated in Ca²⁺ transport within the calcifying cells. It is known that microtubules are involved in vesicular transport activity across polarized cells (see review by Schaerer et al. 1991). However, nothing is known about intracellular movement of molecules such as Ca²⁺-binding proteins. While diffusion cannot be excluded, our results suggest that the cell cytoskeleton may play a role in the directed movement of these molecules.

The post-incubation washing procedure, described here for the first time, allows examination of at least two exchangeable Ca²⁺ pools relevant to calcification in Stylophora pistillata. Figs 1C and 2 show that the time course of Ca²⁺ deposition displays a short lag phase of less than 2 min. A linear rate of calcification was demonstrated in the scleractinian coral Galaxea fascicularis by Krishnaveni et al. (1989), whereas a biphasic pattern was found in more slowly calcifying animals, such as echinoderms (Nauen and Böhm, 1979; Donachy and Watabe, 1986; Dafni and Erez, 1987; Lewis et al. 1990) or octocorals (Allemand and Grillo, 1992), where the turnover rate of compartments is much slower. A linear uptake rate indicates that no intermediate compartment exists between the external sea water and the coelenteron, thus invoking a very short diffusional pathway. In the present study, the short lag phase should correspond to transit through an intermediate cellular compartment, since our results demonstrate that Ca²⁺ must pass through cells before reaching the site of calcification (see above). The short lag phase in the calcification process suggests that the cellular Ca²⁺ compartment that supplies skeletogenic Ca²⁺ (‘calcifying compartment’) is rapidly equilibrated with exogenous radioactive Ca²⁺. This means that this calcifying compartment has a rapid turnover rate or a small size or both characteristics. Our results suggest that the turnover rate of this hypothetical pool is in the region of 2 min.

What type of intracellular compartment could correspond to the calcifying compartment? The turnover rate of the NaOH-soluble pool (20 min), which anatomically corresponds to total tissue Ca²⁺, is too slow and the pool is too large (7.12 nmol mg^-1 protein). Furthermore, uptake of Ca²⁺ into the NaOH-soluble pool was not inhibited by Ca²⁺ channel inhibitors whereas calcification was. These results suggest that the tissue compartment identified as NaOH-soluble cannot be assumed to be the calcifying compartment. The gastrovascular pool exhibited kinetic characteristics indicative of a possible source for calcification. The very fast turnover rate of 2 min should correspond to that suggested for the calcifying compartment. However, its extracellular localization rules out its involvement as a direct source of Ca²⁺ for calcification. This leads us to hypothesize the existence of a third exchangeable Ca²⁺ compartment that our experimental approach cannot detect. This compartment should have (1) a rapid turnover rate, about 2 min, which causes it to be undetectable in the large gastrovascular pool, and (2) a small size. Ca²⁺ fluxes across this compartment should be high, 975 pmol mg^-1 protein min^-1 based on the calcification rate (Fig. 1C), in contrast to the flux into the tissue compartment which is only 224 pmol mg^-1 protein min^-1. The bulk of skeletal Ca²⁺ should come from this compartment (Fig. 8A) which could correspond anatomically to the calicoblastic epithelium (Fig. 8B). This compartment is probably supplied directly by the gastrovascular compartment: the theoretical rate of Ca²⁺ flux through the gastrovascular compartment calculated from kinetic parameters (see Results) is 12.6 nmol mg^-1 protein min^-1. This flux is 13 times larger than necessary to support the measured calcification rate (0.97 nmol mg^-1 protein min^-1) and, thus, more than sufficient in capacity (Fig. 8B). Such an high flux in the gastrovascular cavity could be accomplished through the mouth by muscular pumping or by active transport through the oral epithelial cell layers as shown by Wright and Marshall (1991) in Lobophyllia hemprichii and Plerogyra sinuosa. However, since in our experiments the flux is cyanide-insensitive (Tambutté et al. 1995), it is possible that Ca²⁺ movement could take place across the oral tissues via an intercellular pathway. This was recently confirmed in our laboratory in kinetic experiments with Heliofungia actiniformis using an Ussing chamber. These experiments indicated that Ca²⁺ crossed the oral tissue via an intercellular pathway with a flux of 2.3 nmol mg^-1 protein min^-1, which is higher than the calcification rate (Bénazet-Tambutté et al. 1996).

What is the nature of the NaOH-soluble Ca²⁺ pool? From an anatomical point of view, this compartment should correspond to the tissue compartment. If the cell water space is assumed to be similar in size to that this found in the sea anemone by Lopez et al. (1991), i.e. 2.54 μl mg^-1 protein, it can be calculated from the Ca²⁺ compartment size (7.12 nmol Ca²⁺ mg^-1 protein, see Table 1) that the total cellular Ca²⁺ concentration (not activity) is 2.80±0.3 mmol l^-1, a value typical of other cells (Carafoli, 1987). This result rules out the hypothesis of an extracellular location of this compartment since the Ca²⁺ concentration in sea water is about 10 mmol l^-1. However, this value should be taken with caution, since the tissues are highly heterogeneous and are made up of several cell types, including mucocytes, dinoflagellate symbionts, cnidocytes and ectodermal cells. The turnover rate of the NaOH-soluble pool is in close agreement with that found for intracellular compartments of exchangeable Ca²⁺ in other cell types (Borle, 1969a,b; Claret-Berthon et al. 1977). Surprisingly, Ca²⁺ channel inhibitors appeared to stimulate Ca²⁺ influx into this compartment (Fig. 4). Some authors have found that, in algae, organic channel antagonists such as nifedipine, bepridil or verapamil can increase the Ca²⁺ influx through voltage-dependent Ca²⁺ channels (reviewed by Reid and Tester, 1992). They explained this unexpected effect by an inhibition by these drugs of K⁺ channels, leading to hyperpolarization. However, the absence of voltage-dependent Ca²⁺ channels is not as unexpected in non-excitable cells (Sage et al. 1992).

The efflux experiments were designed to examine Ca²⁺ compartmentation further, particularly the nature of tissue compartments. The 0Ca/EGTA medium was employed to avoid the effects of isotopic dilution on specific activity during the efflux (or washing) period. Conducting the efflux on ice should reduce the redistribution of Ca²⁺ between compartments due to active transport. Comparing efflux measured in this way with efflux measured in filtered sea water, the faster components appear to correspond to similar compartments. The C1 component represents extracolonial Ca²⁺. The kinetic characteristics of C2 and C3 are close to those found in influx experiments for the gastrovascular cavity (Tambutté et al. 1995) and the NaOH-soluble pool (i.e. a tissue Ca²⁺ compartment) respectively. Caution must be exercised, however, in comparing uptake kinetics under physiological conditions with efflux under different conditions.

The D600-sensitivity of the C4 component indicates the presence of voltage-dependent Ca²⁺ channels. However, this compartment exhibits a very slow rate constant (72–742 min depending on the efflux protocol, Table 2), suggesting that it is located at a deeper site within the tissue (Claret-Berthon et al. 1977). When efflux experiments were performed in FSW, the C4 kinetic values were not very different from those of the NaOH-soluble pool, but the C4 compartment size was considerably larger than the amount of Ca²⁺ remaining in the NaOH-soluble pool when 0Ca/EGTA was employed (Table 3). The discrepancy was not as large, but was still substantial, when both EGTA and Ca²⁺ were absent from the efflux medium. These C4 values represent compartment size at the beginning of Ca²⁺ efflux (i.e. at the end of the incubation). While this value is ultimately of greatest interest, the C4 Ca²⁺ content remaining after efflux should be similar to that recovered in the NaOH-soluble pool. The C4 Ca²⁺ content remaining after efflux was always higher (Table 3). Is all tissue Ca²⁺ recovered by the NaOH extraction? Any unrecovered Ca²⁺ would still be associated with the skeleton. The S1 component is a possible indicator of such Ca²⁺. If one adds this amount of Ca²⁺ (Table 2) to the amount of Ca²⁺ in the NaOH-soluble pool (Table 3), the sum is consistently very close to the C4₁₈₀ Ca²⁺ value (Table 3), and a similar value is obtained in all efflux protocols except with FSW at 25 °C.

If one assumes that the C4 compartment is synonymous with a tissue compartment, then why do the apparent efflux rate constants for C4 differ between treatments, resulting in discrepancies in total C4 Ca²⁺ (C4₀)? It is possible that tissue Ca²⁺ or its specific activity changes during the efflux procedure. The results using the EGTA protocol suggest that skeletal Ca²⁺, or a labile pool thereof, is being redistributed and detected in the C4 compartment. This Ca²⁺, possibly represented by the S1 component, could be newly excreted Ca²⁺, perhaps associated with matrix proteins but not yet deposited as CaCO₃.

This argument may qualify the results obtained with D600 with regard to the C4 and NaOH-soluble pools (Fig. 7); the inhibition caused by D600 may reflect differences in calcification rather than in tissue Ca²⁺ influx. When FSW is used during the efflux, there could be a decrease in cellular specific activity when unlabelled Ca²⁺ influx must be assumed to occur. The relatively slow loss of label detected from the C4 compartment (reflected in the much shorter t_1/2 values) in FSW would reflect the exchange of ⁴⁵Ca for unlabelled Ca²⁺ rather than unidirectional efflux of Ca²⁺. Warmer temperatures appear to decrease the C4 (and NaOH-soluble fraction) Ca²⁺ values, and this could be due to continued operation of Ca²⁺ pumps during efflux.

The fraction of total colony Ca²⁺ uptake (sum of all fractions except C1) contributed by compartments C4, S1 and S2 is very similar in all treatments (75.5±3.3 %). Thus, all efflux protocols are similar in their ability to remove extracellular (coelenteron, etc.) Ca²⁺, but the remaining Ca²⁺ is distributed differently. The values for Ca²⁺ content in the S2 and HCl-soluble pool were in close agreement (data not shown) for all protocols. It was clear, however, that the presence of EGTA decreased the skeletal Ca²⁺ content. Comparing the S2 or HCl-soluble values (approximately 28 nmol Ca²⁺ mg^-1 protein h^-1, as determined from the values in Table 2) with calcification rates obtained using the influx protocol previously validated with the alkalinity technique (58.52 nmol Ca²⁺ mg^-1 protein h^-1; Tambutté et al. 1995), the EGTA method provides a substantial underestimate of calcification. The results obtained in FSW at 25 °C (S2=56.3 nmol Ca²⁺ mg^-1 protein h^-1, as determined from the values in Table 2) and with 0Ca/ASW (66.2 nmol Ca²⁺ mg^-1 protein h^-1, as determined from the values in Table 2) appear to provide the values closest to alkalinity values.

In conclusion, we suggest that the tissues of scleractinian corals should possess two types of cells according to their sensitivity to Ca²⁺ channel inhibitors. (1) Calcium-transporting cells, which anatomically belong to the calicoblastic epithelium and have L-type voltage-dependent Ca²⁺ channels. Ca²⁺ is transported across this epithelium by an energy-consuming mechanism (Ca²⁺-ATPase), the flux being coupled in some way to anion transport; the transcellular passage of Ca²⁺ is independent of intracellular Ca²⁺ stores, but is dependent on the cytoskeleton. (2) All other cells, which do not transport Ca²⁺ actively and which have voltage-dependent Ca²⁺ channels insensitive to the inhibitors tested or which lack this type of channel. Ca²⁺ used for building the skeleton equilibrates rapidly (t_1/2 ≈2–4 min) with the coelenteric compartment and with the Ca²⁺ within the Ca²⁺-transporting cells. Thus, skeletal Ca²⁺ should cross only one epithelium (Fig. 8B). Once excreted by the calicoblastic cells, Ca²⁺ seems to exist in a transitory state, perhaps associated with the organic matrix, before being firmly deposited as CaCO₃. As previously concluded (Crossland and Barnes, 1977), the post-incubation procedures are critical for evaluating calcification and the compartmentation of Ca²⁺.

The new methodologies applied in this study promise to add considerably to our knowledge of scleractinian coral physiology. The coral microcolony eliminates many of the problems that compromised previous calcification data and limited progress in this field. Coupled with a physiological approach, the microcolony provides a model system for the systematic study of calcification and its links to other physiological processes, including zooxanthellae photosynthesis.

We thank The Marine Station of Aqaba for facilitating the collection of corals. We thank Dr Barhanin (Institut de Pharmacologie Moléculaire et Cellulaire) for his generous gift of drugs and C. Emery for his technical assistance. We also thank Dr D. Zoccola for fruitful discussions. This study was conducted as part of the OOE 1991–1995 research program. It was supported by the Council of Europe (Open Partial Agreement on Major Natural and Technological Disasters), the Programme National Récifs Coralliens (PNRCO). Support for E.M. was provided by NSF-EPSCoR (award EHR-9108761) and the OOE.