Aragonite Precipitation by “Proto-Polyps” in Coral Cell Cultures

The mechanisms of coral calcification at the molecular, cellular and tissue levels are poorly understood. In this study, we examine calcium carbonate precipitation using novel coral tissue cultures that aggregate to form “proto-polyps”. Our goal is to establish an experimental system in which calcification is facilitated at the cellular level, while simultaneously allowing in vitro manipulations of the calcifying fluid. This novel coral culturing technique enables us to study the mechanisms of biomineralization and their implications for geochemical proxies. Viable cell cultures of the hermatypic, zooxanthellate coral, Stylophora pistillata, have been maintained for 6 to 8 weeks. Using an enriched seawater medium with aragonite saturation state similar to open ocean surface waters (Ωarag∼4), the primary cell cultures assemble into “proto-polyps” which form an extracellular organic matrix (ECM) and precipitate aragonite crystals. These extracellular aragonite crystals, about 10 µm in length, are formed on the external face of the proto-polyps and are identified by their distinctive elongated crystallography and X-ray diffraction pattern. The precipitation of aragonite is independent of photosynthesis by the zooxanthellae, and does not occur in control experiments lacking coral cells or when the coral cells are poisoned with sodium azide. Our results demonstrate that proto-polyps, aggregated from primary coral tissue culture, function (from a biomineralization perspective) similarly to whole corals. This approach provides a novel tool for investigating the biophysical mechanism of calcification in these organisms.


Introduction
The potential effects of global warming and ocean acidification on coral calcification have received significant attention in recent years [1][2]. However, despite the broad interest in coral calcification [3][4][5][6] and the potential for climate-driven adverse effects, the actual calcification mechanisms are still poorly understood at the cellular level. A lack of mechanistic understanding of processes that lead to and control calcification limits our ability to predict corals' response to increasing atmospheric CO 2 [6]. Thus, if we are to truly predict the effects of increasing atmospheric pCO 2 , it is of critical importance to understand these mechanisms. This will also provide a stronger basis for interpreting geochemical proxies in corals.
Corals (phylum, Cnidaria) belong to one of the oldest invertebrate phyla. They are the earliest metazoans that possess an organized body structure. Their body plan consists of two cell layers: an ectoderm and an endoderm, separated by the mesoglea, which is a non-cellular gelatinous matrix [7]. In symbiotic corals, the endoderm hosts unicellular algae of the genus Symbiodinium. These tissues are mechanically anchored to the skeleton by desmocytes [8]. The aboral ectoderm in contact with the skeleton, referred to as the calicoblastic epithelium, is involved in the extracellular production of the aragonite (orthorhombic CaCO 3 ) coral skeleton under its apical membrane [9]. The subcalicoblastic space between the skeleton and the calicoblastic epithelium, containing the ''calcifying fluid'' in which new skeletal material is precipitated, does not exceed a few nanometers [3]. Accessing this extremely small space is very difficult and has prevented detailed physiological studies of sub-calicoblastic chemical variability. The lack of mechanistic understanding of the calcification process also constrains our ability to interpret geochemical proxies in corals (e.g., d 11 B for reconstructing seawater pH) using calibrations based solely on correlative geochemical signatures of intact living corals [10][11][12].
Previous studies that attempted to quantify the in vivo chemistry at the cellular level rely on microsensor pH measurements within the calcifying fluid or fluorescent dyes that give information about pH and/or calcium concentrations [13,14]. Using microelectrodes, Al-Horani et al. [10] found, in the coral Galaxea fascicularis, large shifts in pH from 9.3 to 7.5 under light and dark conditions, respectively. Similarly, higher pH in the calcifying fluid than in the ambient seawater was reported in the coral Stylophora pistillata based on a pH-sensitive fluorescent probe (SNARF-1) [14]. The diurnal pH shift reported in the latter study was lower under comparable light and dark conditions (from ,8.7 to ,8.4). However, given that these two studies were conducted on different coral species and with different methodologies, identical pH values should not be expected. Based on their pH measurements, Venn et al. [14] assumed a very high alkalinity for the calcifying fluid (.5000 mmol/kg) and estimated the aragonite saturation state in the sub-calicoblastic space (V arag ) to be ,20 in the light and ,11 in the dark. Further support for elevated pH below the calcifying tissue of corals comes from boron isotope measurements of coral skeletons [10][11][12]. These studies suggest that the pH in the calcifying fluid is apparently about 0.4 pH units higher than that of ambient seawater, which could indicate the regulation of pH in the calcifying fluid by the coral. Boron isotopic composition, recorded in the skeleton of the tropical coral Cladocora caepetos [15], suggests that corals may adjust the pH of the calcifying fluid by up to ,1.4 pH units, depending on ambient seawater pH. If true, corals could maintain a similarly high V arag over a range of environmental conditions. This observation raises an apparent paradox: if corals can regulate their internal pH, why would they be vulnerable to ocean acidification? A comprehensive answer to this question requires manipulation of the chemistry of calcifying fluids independent of ambient conditions.
The presence of an extracellular organic matrix (ECM) in coral skeletons is well documented [9,15,16]. Muscatine and Cernichiari [17] showed that carbon fixed by Symbiodinium is transferred to the host and incorporated into ECM. Additionally, isotopically labeled dissolved free amino acids in seawater and in particulate food sources have been shown to accumulate as an organic component in coral skeletons [18]. Skeletal ECM is rich in aspartic and glutamic acid leading to the hypothesis that acidic amino acids are important in the calcification process [19]; however, the sequences of proteins involved in calcification and pathways leading to their accumulation in the calcifying space remain to be determined.
In principle, in vitro investigations of coral calcification can be studied at the cellular level because of the simplicity of the cnidarian organization and the small number of cell types. Helman et al. [20] previously reported a primary coral cell culture method that can potentially be used to analyze the mechanism of calcification in these organisms. In that study, the authors showed that the primary cell culture exhibits the fundamental processes involved in coral calcification, including production of extracellular matrix, skeletal organic matrix and evidence for extracellular calcification [20]. However, if this method is to be used as a model system for studying the calcification mechanism in corals, it is necessary to demonstrate the feasibility of dissociated coral cells to re-aggregate and to form a functional calcifying organism. Likewise, it is critical to show that skeletal crystals generated by this system are comparable to those produced in vivo.
In this paper, we present the first evidence that primary cell cultures of Stylophora pistillata, a well-studied species, re-aggregate to form functional organized cell culture (''proto-polyps''), which produces an extracellular organic matrix (ECM). Moreover, we demonstrate that these proto-polyps produce extracellular aragonite crystals. This system will provide a critical tool for studying calcification mechanisms through in vitro manipulations.

Cell Cultures
Coral fragments from the zooxanthellate coral, Stylophora pistillata, were obtained from nubbins growing in an 800 L, custom-designed aquarium as described previously [21]. Cell cultures were prepared following previously published procedures, with small modifications [20]. Briefly, small fragments of coral were excised from parent colonies and incubated for 3 to 5 h with gentle shaking in calcium-free artificial seawater supplemented with a 3% antibiotics-antimycotics and 20 mg ml 21 chloram-phenicol solution (GIBCO). Fragments were then transferred to 35610 mm Primaria culture dishes containing 3 mL of culture medium. The medium was prepared in two steps: N Dulbecco's Modified Eagle Medium (DMEM) with no glucose (Invitrogen) supplemented with the following major seawater ions: 0.35 g l 21 KCl, 1.1 g l 21 CaCl 2 , 1 g l 21 MgSO 4 7H 2 O, 18.1 g l 21 NaCl, 0.052 g l 21 taurine and 25 mM HEPES buffer [22]. N A mixture of artificial seawater (Instant Ocean sea salt, Aquarium Systems 34 p.s.u.), 12.5% modified DMEM (from step I), 20 mg/mL aspartic acid, 2% heat-inactivated FBS (Invitrogen),1% antibiotic-antimycotics solution (GIBCO), 0.1 mM glucose, and 50 mg ml 21 L-ascorbic acid. Cultures were exposed to the same atmospheric conditions as in the aquarium. Therefore, in order to avoid evaporation of the media and to keep the growth conditions similar to that in the aquarium, the cultures were maintained in an environmental growth chamber (Percival scientific, INC, USA) on a 12:12 h light:dark cycle at 26uC and 80% humidity, and the media was replaced every 7 days. The carbonate alkalinity of the final medium solution (without the HEPES buffer) was 2215 mmol kg 21 , as measured by Gran titration [23]. We note that within the analytical error of the method the titration of the artificial seawater only yielded similar carbonate alkalinity to that determined on whole media, suggesting a minimal effect of the organic components. The pH of the medium was maintained at 8.060.1 by the HEPES buffer. We calculated V arag to be 3.7 using an online program that calculates in situ CO 2 conditions [24].
These conditions mimic open ocean surface waters and the carbonate saturation state in our aquaria. After 48 h, the coral skeleton was removed from the plate, and the medium with the detached cells was centrifuged (85006 g for 10 min). The pellet was re-suspended in fresh media and filtered through 20 mm nylon mesh to remove cell aggregates and debris. In order to rule out the possibility that observed aragonite crystals were formed by inorganic precipitation in the saturated culture medium the following controls were set: (1) a plate with medium but no culture cells; and (2) a primary cell culture treated with 15 mM sodium azide, that through respiratory inhibition, prevents reduction of O 2 to water by the heme cofactor of cytochrome oxidase in mitochondria Chlorophyll a concentrations were measured by the method of Jeffrey and Humphrey [22] on a DW2000 spectrophotometer after culture samples had been extracted in 90% acetone overnight at 220uC. A rapid assessment of photosynthetic activity of zooxanthellae was determined by use of a Fluorescence Induction and Relaxation (FIRe) fluorometer [25,26].

Microscopy Imaging
Microscopy imaging was carried out with an inverted IX71 epifluorescent microscope (Olympus) and a Zeiss LSM 710 confocal microscope. For confocal images, cells were grown on glass bottom dishes (WillCo-Dish). Excitation wavelengths at 490, 488 nm with emission wavelengths at 510, 667 nm were used for green fluorescence proteins (GFP) and chlorophyll fluorescence measurements, respectively.
For field emission scanning electron microscopy (FE-SEM) and energy-dispersive X-ray spectrometry (EDS), cells were grown on Millicell cell culture inserts in 6-well tissue culture plates (Millipore). These cultures were fixed for 2 h with 2% glutaraldehyde in 0.05 M phosphate buffer (pH 7). Some of the samples were gently washed with distilled water and the rest were incubated in 1 M NaOH at 90uC for 20 minutes to denature cellular membranes, followed by dehydration with an ascending ethanol series (50-100%) and critical point drying with liquid CO 2 . All samples were then coated with gold and platinum and observed on a Zeiss Sigma FE-SEM equipped with Gemini column and Oxford Instruments with an IncaPenta FET-X3 detector.
All PCR reactions contained 0.1-0.4 mg of template DNA, 10 mM total dNTP, 16 REDTaq reaction buffer, 0.1-0.5 mM of each primer and 0.05 unit mL 21 of REDTaq polymerase (sigma #D4309) in a total volume of 25-50 mL. Amplifications were performed using a Perkin Elmer-Cetus 480 Thermal cycler with the following thermal profile for the S. pistillata and algae-specific primers respectively: 40 cycles of 30 sec at 94uC, 30 sec at 40uC, 90 sec at 72uC, and 35 cycles of 1 min at 94uC, 2 min at 55uC, 3 min at 72uC. PCR products were cloned using the TOPO TA cloning kit (Invitrogen) and transformed into Escherichia coli (Top10). Plasmids were purified by the QIAprep spin miniprep kit (Qiagen) and sequenced by Genwiz sequencing service (http:// www.genewiz.com/).

Characterization of CaCO 3 and ECM
Aragonite mineralogy was confirmed for 7-10 days old culture by X-ray diffraction (XRD) on a Bruker/Siemens Hi-Star detector. Contents of 5-10 wells of each treatment (see above) were scraped with a rubber policeman, combined, and centri-fuged. The pellet was rinsed with Milli-Q deionized water and then resuspended in 1 M NaOH and heated to 90uC for one hour. Samples were again centrifuged and the NaOH-insoluble pellet was rinsed with Milli-Q water, dried at 60uC, and stored until analysis. Additionally, skeleton samples of the mother colony were bleached in 1% commercial bleach at room temperature for 4 h, rinsed extensively, dried, and ground to a fine powder with mortar and pestle made of agate. The powder was soaked again for 4 hours in 1% bleach, rinsed extensively, and dried. Samples were mounted onto borosilicate capillary tubes and analyzed by XRD. We added ,1 mg LaB 6 (NIST SRM-660a), a calibration standard, to the culture samples to confirm accuracy of peak identification; this is common for very small samples (i.e.; ,1% of the aragonite mass used in mother colony skeleton analyses). Resulting spectra were processed in GADDS software (Bruker AXS).
Amino acid composition of ECM proteins was determined by high performance liquid chromatography (HPLC) after acid hydrolysis of proteins extracted from decalcified skeletons of S. pistillata and from NaOH-insoluble culture ECM. Briefly, skeletons were cleaned as described above for XRD, then subsamples of clean skeleton powder and NaOH-treated culture pellet were decalcified in 1N HCl for four hours. The solutions were neutralized and water-soluble and insoluble proteins were separated by centrifugation. Soluble proteins in the supernatant were concentrated by centrifugal filtration (35006 g) on Amicon Ultra filters (Millipore, 3 kDa cut-off), and both solubility fractions were then lyophilized. Dried organic matrix samples were hydrolyzed in 6 N hydrochloric acid at 110uC for 18 hours and then neutralized with NaOH. Subsamples were analyzed according to the modified methods of Mopper and Lindroth [28]. Hydrolyzed amino acids were combined with o-phthaldialdehyde/N-acetyl-L-cysteine in 0.8 M borate buffer (1:3) and allowed to react at room temperature for 2 min. The derivatization solution was run on a Shimadzu HPLC fitted with an ODS Rexchrom column (Regis, 5 mm) and sodium acetate/methanol mobile phase.

Results and Discussion
Self-assembly of proto-polyps Within 2 days of culture initiation, S. pistillata tissue, which had been pre-incubated for 3 to 5 h in calcium-free seawater, spontaneously dissociated from the skeleton into separate, discrete cells. Dissociated cells consisting of a mixture of cell types including free Symbiodinium sp. and individual endoderm and ectoderm cells (Fig. 1A), were placed in the culture dish. These cells do not appear to morphologically dedifferentiate (i.e., they do not form stem cells), however, transferring the coral fragment to the culture medium suppressed photosynthetic activity in the Symbiodinium sp. After 24 hours, photosynthetic efficiency, measured as F v /F m , was undeterminable compared with intact coral fragments (,0.50). It should be noted, however, that the chlorophyll concentration did not change during this time period. The observed inhibition of photosynthesis could have been the result of the glucose added to the culture medium that was utilized as an exogenous organic substrate (in lieu of photosynthetic fixation) and subsequently respired by the Symbiodinium sp. [20,29].
The presence of coral cells and Symbiodinium sp. in the cultures was confirmed by polymerase chain reaction (PCR) using anthozoan-and Symbiodinium-specific primers and blasting the PCR-derived sequences against the National Center for Biotechnology Information nucleotide database (www.ncbi.nlm.nih.gov) (Fig. S1).
Within 48 h, individual cells in the culture assembled into organized cell clusters, consisting of 3 layers (Fig. 1B-D, Figs. S2,  S3). The viability of these ''proto-polyps'' remained .80% over a period of ,5 weeks, as was quantitatively assessed by using Sytox green [20], after which microbial contamination became unavoidable. Therefore, all calcification and metabolic measurements were preformed on 5 to 10 day-old uncontaminated cultures. Cells adhered to the Primaria dish substratum through ECM that mediated cell-cell, as well as, cell-substratum adhesion. In addition, assembly of larger spherical aggregations ,80 mm thick, which detached from the substratum, was observed ( Fig. 2A). Cell cultures treated with 15 mM sodium azide did not aggregate, ruling out the possibility of spontaneous aggregation.
Aggregates incorporated all individual cell types (ectoderm or endoderm cells, Symbiodinium sp., and nematocysts). Symbiodinium sp. cells, as indicated by chlorophyll florescence, were located in the middle part of the aggregation (20-50 mm). In contrast, animal cells, as indicated by GFP fluorescence, were located throughout the aggregate (Fig. 2B, S3). Associated with all aggregates, we observed the formation of aragonite crystals (Fig. 2D).

Aragonite precipitation
Relatively large, extracellular crystals were detected after 10 days in cell culture, on the surfaces of both adherent and nonadherent proto-polyps (Fig. 3A). Crystal growth was not observed on individual cells. Crystals on proto-polyps formed distinct flowershaped bundles, originating from the upper surface facing the media, and were attached to ECM (Fig. 3A). Once the animal's cells were removed from the proto-polyp by 1 M NaOH, the structure of the ECM was revealed (Fig. 3D). In contrast, no calcium carbonate crystals were detected by XRD or EDS in the control treatments, treatments made of culture medium without cells, or with addition of 15 mM sodium azide to healthy cultures.
The elongated shape and morphology of crystals are similar to aragonite crystals observed in the septa of the S. pistillata mother colony (Fig. 3C), which are one of the primary sites of CaCO 3 precipitation [30]. Indeed, XRD analysis of these crystals confirmed that the calcium carbonate polymorph precipitated in our healthy coral cell cultures was aragonite (Fig. 4). Aragonite crystals formed in cell cultures were about 10 mm long. In comparison, aragonite crystals from the mother colony of S. pistillata, growing in the in-house aquarium, were smaller than those produced in vitro despite the similarity in saturation levels (Fig. 3). Previous studies describe similar morphology and orientation for aragonite crystals at active growing regions of corals as those observed in our tissue cultures [31,32].
Chemical analysis by energy-dispersive X-ray spectroscopy (EDS) confirmed that crystals observed were, indeed, calcium carbonate. A typical X-ray spectrum of a crystal in the protopolyp, the mother colony skeleton and of the ECM reveals a prominent calcium peak (Fig. 3). In addition to the calcium peak, a second prominent sulfur peak can be seen in the crystal deposit in the proto-polyp (Fig. 3a). However, sulfur peaks were not observed on the cell section of the proto-polyp (Fig. 3b), but rather peaks corresponding to nitrogen, carbon and oxygen that are associated with the organic matrix were detected. As the initial crystals emerged from the ECM, the sulfur peak observed in those crystals, and not in the mother colony skeleton, may result from relatively abundant sulfate-bearing organic compounds at centers of calcification but lower organic matter concentrations in the bulk skeleton, as suggested by Cuif et al. [33]. These results support the Cuif et al. [33] proposed model of crystal growth that involves a step-by-step growth of aragonite fibers where each step is initiated and guided by a sulfated organic matrix sheet.

ECM composition
The total ECM was separated by centrifugation into soluble and insoluble fractions. Total aspartic acid plus asparagine content (Asx), measured by HPLC, was ,20% of total ECM skeletal protein content for both fractions. The second most abundant amino acid was glutamic acid plus glutamine (Glx), also accounting for ,20% of the total. Thus, these two sets of amino acids account for ,40% of the total in both soluble and insoluble ECM fractions both in the mother colony and cell culture ECM (Table 1). Similar to our study, Young [15] reported 12-23% of Asx in skeletons from 14 different species of scleractinian corals. Specific functions of the soluble and insoluble ECM fractions in the calcification processes have yet to be determined. While some differences have been found between the two fractions, many similarities have been indicated, including polar amino acid content and protein size and activity [34,35]. Differences between the two fractions may be ascribed to their roles in either structural framework (insoluble ECM) or crystal nucleation (soluble ECM) [36][37][38]. However, as only one ECM protein has been fully sequenced, the distinctive and/or overlapping nature of soluble and insoluble ECM proteins remains to be seen [39].
Sulfur has been shown to be integral to centers of calcification in coral skeletons [33]. We have confirmed a lower amount of sulfur containing material in ECM compared to primary crystals by EDS (Fig. 3). In addition, we measured 0.5-1% methionine residues in S. pistillata mother colonies by HLPC (this method does not allow quantification of cysteine).

Final Remarks
In this study we demonstrate that a primary culture of disaggregated coral cells can re-assemble into proto-polyps from primary, differentiated cells. Each proto-polyp contains three cell  layers: a basal layer of ectodermal cells that adheres to the plate is covered with a second layer of endodermal cells containing zooxanthellae (Fig. 5). The top layer contains ectodermal calicoblastic cells exposed directly to medium. In coral colonies aragonite is precipitated on ECM scaffolding that is secreted into the calicoblastic space by ectodermal cells facing the skeleton [2]. In contrast, in our culture the calicoblastic cell-layer is exposed to the artificial calcicoblastic fluid surrounding the proto-polyp (Fig. 5). Under these culture conditions, we propose that the calicoblastic cell layer secretes ECM with sulfate rich calcifying centers [33] onto which the aragonite crystals precipitate (Fig. 3). We note that the ECM amino-acid composition produced by the cell culture is similar to that of the mother colony. Aragonite crystals in cell cultures have similar structure and chemical composition to the mother coral colony skeleton; however, they are substantially larger than those in the mother coral (10 mm vs. 2 mm, respectively). The difference is likely due to the fact that in the culture the new skeleton formation is not constricted by the small calicoblatic space as it is inside a coral colony, therefore crystals extend to greater lengths and are lower in density (Fig. 3). Regardless of morphological differences, however, our results clearly suggest that coral cells can catalyze carbonate crystal formation under ambient concentrations of DIC. How that process occurs remains fundamentally unknown. This study demonstrates the potential of coral cell culture for studying physiological mechanisms, including calcification and cell differentiation, at the cellular level. This will provide a critical tool for mechanistically understanding how shifts in ocean pH will impact calcifying corals.