A vesicular Na+/Ca2+ exchanger in coral calcifying cells

The calcium carbonate skeletons of corals provide the underlying structure of coral reefs; however, the cellular mechanisms responsible for coral calcification remain poorly understood. In osteoblasts from vertebrate animals, a Na+/Ca2+ exchanger (NCX) present in the plasma membrane transports Ca2+ to the site of bone formation. The aims of this study were to establish whether NCX exists in corals and its localization within coral cells, which are essential first steps to investigate its potential involvement in calcification. Data mining identified genes encoding for NCX proteins in multiple coral species, a subset of which were more closely related to NCXs from vertebrates (NCXA). We cloned NCXA from Acropora yongei (AyNCXA), which, unexpectedly, contained a peptide signal that targets proteins to vesicles from the secretory pathway. AyNCXA subcellular localization was confirmed by heterologous expression of fluorescently tagged AyNCXA protein in sea urchin embryos, which localized together with known markers of intracellular vesicles. Finally, immunolabeling of coral tissues with specific antibodies revealed AyNCXA was present throughout coral tissue. AyNCXA was especially abundant in calcifying cells, where it exhibited a subcellular localization pattern consistent with intracellular vesicles. Altogether, our results demonstrate AyNCXA is present in vesicles in coral calcifying cells, where potential functions include intracellular Ca2+ homeostasis and Ca2+ transport to the growing skeleton as part of an intracellular calcification mechanism.


Introduction
Coral reef ecosystems are valuable ecological [1] and economic resources [2] centered around the calcium carbonate (CaCO 3 ) exoskeletons deposited by scleractinian corals. The aboral ectodermis (also known as the calicoblastic epithelium or calicodermis) is directly above the subcalicoblastic medium (SCM) and the skeleton, and therefore is the tissue layer with the most direct role in calcification ( [3]; reviewed in [4]). However, the cellular mechanisms for coral calcification are poorly understood (reviewed in [5]).
Recent research indicates corals exert strong biological control on skeleton formation through intracellular calcification mechanisms. Calicoblastic cells express HCO 3 transporting proteins that likely supply dissolved inorganic carbon [5][6][7], as well as coral acidic rich proteins (CARPs) that can catalyze aragonite formation even at pH~7.6 [8][9][10]. Furthermore, amorphous CaCO 3 is present inside coral cells [8] and secreted at the mineralizing front together with HCO 3 -, CARPs, and several other proteins [11]. Those results suggest intracellular vesicles play an important role in coral skeleton formation. Another model proposes transcellular Ca 2+ transport to the skeleton by a combination of Ca 2+ channels that facilitate Ca 2+ entry from the coelenteron into the calicoblastic cells [12], and plasma membrane Ca 2+ -ATPases (PMCAs) that extrude Ca 2+ across the apical membrane into the SCM in exchange for H + (reviewed in [13,14]). However, we have recently reported PMCA is located throughout the cytoplasm of coral calcifying cells and not in the apical membrane, a pattern consistent with localization in intracellular vesicles [7]. Apical Na + /Ca 2+ exchangers (NCXs, SLC8A gene family) have also been proposed to secrete Ca 2+ from coral calcifying cells into the SCM for skeleton formation [15,16]. However, this model is largely based on osteoblasts from vertebrate animals where NCXs located in the cell plasma membrane mediate bone formation [17,18] and direct evidence for the presence and localization of NCXs in coral cells is lacking.
The goals of the current study were to establish if coral indeed have a protein homologous to the NCXs from osteoblasts, and characterize its expression throughout coral tissues and its intracellular localization in calcifying cells. This type of basic information is an essential first step for future functional experiments to elucidate coral calcifying mechanisms at the cellular level, and to be able to interpret responses to environmental stress based on "-omics" data.

Corals
Colonies of A. yongei were obtained from the Birch Aquarium at Scripps Institution of Oceanography (SIO) and maintained in flow through seawater (25˚C) and a 12:12 hour light:dark cycle in Hubbs Hall at SIO.

Sea urchin husbandry and gamete collection
Adult Strongylocentrotus purpuratus were collected in San Diego, California, and held in 11˚C (± 1˚C) in flowing seawater aquaria. Animals were spawned by intra-coelomic injection of 0.55 M KCl. Eggs were collected in filtered seawater, washed twice in 0.22 micron filtered seawater (FSW) and kept at 14˚C. Sperm was collected and stored at 4˚C as described previously [26]. mRNA synthesis, storage, and dilution mRNAs encoding C-mCherry Sp-ABCC9, Sp-ABCB6 [27] and C-mCerulean LCK were made with the SP6 mMessage mMachine kit (Ambion) according to the manufacturer's protocol and stocks stored at -80˚C as previously described [28]. For co-expression experiments, Sp-ABCC9 and Sp-ABCB6 mRNA were injected at 500 ng mRNA/μL injection solution and 100 ng/μL C-mCerulean AyNCX A , while C-mCerulean LCK and C-mCherry AyNCX A were injected at 50 ng/μL and 100 ng/μL, respectively.

Sea urchin zygote injections
Unfertilized S. purpuratus eggs were prepared for injection as previously described [29] stuck to 35 mm petri dishes (Fisher Scientific) coated with 0.25% protamine sulfate and fertilized. One-cell zygotes were then injected at between 2-5% egg volume with the mRNA mixture described above. Ampicillin (Sigma Aldrich, St. Louis, MO) at 100 μg/mL in FSW was added to the injection plate. Embryos were then cultured at 15˚C (±1˚C) for between 16 and 48 hours.

Imaging and image processing of sea urchin embryos
Injected S. purpuratus embryos were mounted on 1.5 coverglass (VWR, Radnor, PA) and imaged on either a Zeiss LSM 700 (Jena, Germany) or a Leica Sp8 (Wetzlar, Germany) confocal microscope (mCherry excitation 567 nm, emission 610 nm; mCerulean excitation 433 nm, emission 475 nm). Images were processed using the FIJI distribution of ImageJ [30] Antibodies Custom polyclonal antibodies were developed in rabbit and affinity purified (GenScript USA, Inc) against the peptide antigen sequence KDEDGKSVLRTGEG, which is present in AyNCX A but absent in AyNCX B.

Western blot
Coral tissue was removed from the skeleton and homogenized as previously described [7]. The homogenate was sonicated (3 x 10 second bursts, 30 seconds rest between pulses, on ice), and centrifuged at 500 x g for 15 minutes at 4˚C to pellet out Symbiodinium. Protein concentration in the crude homogenate was determined using the Bradford assay (Bio-Rad, Hercules, CA, USA). Samples were combined with 4x Laemmli buffer (Bio-Rad) with 10% beta-mercaptoethanol and heated at 70˚C for 15 minutes. 20 μg protein/lane were loaded into 10% polyacrylamide SDS-PAGE gels and run at 100V for 75 minutes at 4˚C. Proteins were transferred onto PVDF membranes on a TurboBlot machine (Bio-Rad) using the pre-programmed 30 minute "Standard Molecular Weight" protocol. PVDF membranes were incubated in blocking buffer (5% powdered fat-free milk in Tris-Buffered Saline + 0.1% Tween detergent (TBS-T)), on a shaker at room temperature for 1h. PDVF membranes were incubated overnight on a shaker at 4˚C with primary antibody (0.151 μg/ml; 1:100 dilution from the stock), primary antibody with 400x excess peptide on a molar base ('pre-absorption control'), or pre-Immune Serum (0.151 μg/ml) in blocking buffer. PVDF membranes were washed with TBS-T (3 x 10 min) and incubated 1 hour on a shaker at room temperature, with secondary antibody (goat anti-rabbit-HRP (BioRad) 1:10,000 in blocking buffer). After washing in TBS-T (3 x 10 min), bands were developed using ECL Prime Western Blot Detection Kit (GE Healthcare, Chicago, IL, USA) and imaged using a Chemidoc Imaging system (Bio-Rad).
Fixed and decalcified coral samples were also processed for immunohistochemistry on 400 nm cryosections [7,32]. Briefly, tissue samples were washed with 0.15 M glycine/phosphate buffer, embedded in 10% gelatin/phosphate buffer and infused with 2.3 M sucrose/phosphate buffer overnight at 4˚C. One cubic millimeter blocks were mounted onto specimen holders and snap frozen in liquid nitrogen. Ultracryomicrotomy was carried out at -100˚C on a Leica Ultracut UCT with EM FCS cryoattachment (Leica, Bannockburn, IL, USA) using a Diatome diamond knife (Diatome US, Hatfield, PA, USA). 400 nm sections were picked up with a 1:1 mixture of 2.3 M sucrose and 2% methylcellulose (15cp). Sections were permeabilized in PBS-T for three 10-minute washes, incubated in blocking buffer (2% Normal Goat Serum and 2% Bovine Serum Albumin in PBS) for 1h, and incubated with primary antibodies (NCX: 1.51 μg/mL, NKA: 10 μg/mL) overnight at 4˚C. Sections were then spot washed with 25μL wash buffer (0.1% Bovine Serum Albumin in PBS) three times, followed by three additional 5-minute washes. Sections were incubated in Alexa Flour 555 goat anti-rabbit secondary (4 μg/ ml) together with Hoechst stain (10 μg/mL) for 45 minutes. Sections were washed three times and then imaged. Peptide pre-absorption controls (7.55 μg/mL peptide) were run in parallel.

Transmission electron microscopy
Corals were fixed overnight in modified Karnovsky's fixative (2.5% glutaraldehyde and 2% paraformaldehyde in 0.15 M sodium cacodylate buffer, pH 7.4). Coral fragments were then decalcified and post-fixed in 1% osmium tetroxide in 0.15 M cacodylate buffer for 1 hour and stained en bloc in 2% uranyl acetate for 1 hour. Samples were dehydrated in ethanol, embedded in Durcupan epoxy resin (Sigma-Aldrich, St. Lewis, MO, USA), sectioned at 50 to 60 nm on a Leica UCT ultramicrotome (Leica, Bannockburn, IL, USA), and picked up on Formvar and carbon-coated copper grids. Sections were stained with 2% uranyl acetate for 5 minutes and Sato's lead stain for 1 minute. Grids were viewed using a JEOL 1200EX II (JEOL, Peabody, MA, USA) transmission electron microscope and photographed using a Gatan digital camera (Gatan, Pleasanton, CA, USA).

NCX isoforms are present in multiple coral species
We cloned two full-length transcripts encoding for putative A. yongei NCXs (AyNCX A and AyNCX B ). Similar to NCX proteins from vertebrates (reviewed in [33,34], AyNCX A has a predicted molecular weight of 101.7 kDa and 10 membrane-spanning helices. An alignment of AyNCX A and a mammalian NCX1 is provided in S1 Fig. In addition, AyNCX A contains a peptide signal with very high (0.995) probability to localize the protein to vesicles of the secretory pathway [25]. The other coral NCX that was cloned, AyNCX B , has a predicted molecular weight of~69.8 kDa and only five predicted membrane-spanning helices. We cloned three other cDNAs encoding from putative AyNCX B splice variants with predicted molecular sizes of 26.4, 41.3, and 61.7 kDa. BLAST searches in genomic and transcriptomic databases identified orthologous proteins for both AyNCX A and AyNCX B in multiple other coral species from both the Complex and the Robust clades. Phylogenetic analyses revealed coral NCX A proteins are more closely related to NCXs from vertebrate animals compared to coral NCX B proteins (Fig 1). All these analyses indicate AyNCX A is Na + /Ca 2+ exchanger with ion-transporting properties similar to NCXs from vertebrates. On the other hand, the function of AyNCX B is unclear. Thus, the rest of the experiments focused on AyNCX A .

Recombinant AyNCX A localizes in intracellular vesicles in sea urchin embryos
To further explore the localization of AyNCX A in polarized cells, we took advantage of sea urchin embryos, a recently established heterologous protein expression system for membrane proteins from marine animals [26,27,39]. In this system fertilized embryos (one cell stage) were injected mRNA coding AyNCX A fused to a fluorescent protein and then cultured for the next 16h, during which time the embryo develops a polarized epithelium with apical and basolateral membrane. During this time the exogenous mRNA is translated and the subcellular localization of the corresponding protein, which depends on signal peptides present in the protein of interest, is determined in the embryo by confocal microscopy.
Using this approach, we consistently found that the AyNCX A fusion protein was localized to small spherical intracellular structures~0.5-1 μm in diameter distributed throughout the cytoplasm. Localization to these structures, presumably vesicles, was regardless of the fluorescent protein to which AyNCX A was fused (mCherry or mCerulean) or the location of the fusion protein (C-or N-terminus) (S2 Fig). Additional controls demonstrated overexpressed mCherry protein is present in the cytoplasm and nucleus but not in vesicles (S3a Fig), and that autofluorescence in uninjected sea urchin embryos is minimal compared to fluorescence from mCherry tagged AyNCX A (S3B Fig). Co-expression of AyNCX A with the plasma membrane marker LCK (Fig 2B) demonstrated AyNCX A is not constitutively present in either the apical or basolateral cell membrane ( Fig 2C); however, discreet co-localization events were occasionally observed (Fig 2C) suggesting AyNCX A vesicles might fuse with the cell membrane.
Localization of AyNCX A in mitochondria was ruled out because it did not colocalize with SpABCB6, a mitochondrial protein ( [40,41] reviewed in [42]) (S4 Fig). On the other hand, AyNCX A did colocalize with the vesicular protein SpABCC9 (Fig 2E), a sea urchin ATP-binding cassette protein with a readily observable vesicular localization [26]. AyNCX A vesicular localization is consistent with the signal peptide prediction, as well as with the observations on coral calicoblastic cells described below.

AyNCX A in coral tissue
To determine AyNCX A localization in coral cells, we generated specific antibodies. Western blotting using anti-AyNCX A antibodies specifically recognized two major protein bands in A.
Immunostaining in 7μm histological sections revealed AyNCX A was present in cells from all tissue layers (Fig 3B). Immunofluorescent signal was absent in peptide pre-absorption controls ( Fig 3C) and pre-immune serum controls (not shown), further validating the specificity of anti-AyNCX A antibodies.
Next we looked at AyNCX A subcellular localization in more detail. In the oral ectodermis, AyNCX A was most abundant near the seawater-facing apical membrane of ciliated support cells. In gastrodermal and calicoblastic cells AyNCX A immunostaining pattern was punctate (Fig 4A). Fig 4B shows the corresponding bright field image (differential interference contrast, also known as Nomarski interference contrast). Immunostaining in 400 nm cryosections ( Fig  4C) again revealed punctate AyNCX A signal in calicoblastic cells, and clearly different from the basolateral localization of the Na + /K + -ATPase ( Fig 4D) (also compare with Fig 6b in [7]). The punctate AyNCX A immunostaining pattern in calicoblastic cells was also clear in confocal microscopy images (see S1 File, a 3D reconstruction in the Data supplement). In summary, AyNCX A immunofluorescence pattern, bioinformatics analysis, and heterologous expression experiments strongly suggest AyNCX A is present in the highly abundant 80-500 nm vesicles present in coral calicoblastic cells, which are readily visible by TEM (Fig 5).

Discussion and conclusions
Here we present the first characterization of NCX proteins in coral. We focused on AyNCX A , a~102 kDa protein with 10 predicted transmembrane domains, which is almost identical in structure to the well-characterized NCXs from vertebrate animals and therefore it almost certainly transports Ca 2+ across cellular membranes in exchange for Na + . Orthologous proteins are present in at least nine other coral species from the Complex and Robust coral clades (which diverged from one another over 350 million years ago [45][46][47]), suggesting NCX A plays a widespread important role in corals.
Unlike NCXs from vertebrate animals that are localized in the basolateral or apical membrane of polarized cells [17,18,37,48,49], AyNCX A is present in intracellular vesicles. This is supported by bioinformatics analyses that revealed a peptide signal typical of proteins present in secretory vesicles, heterologous expression of fluorescently tagged protein in sea urchin embryos demonstrating localization in intracellular vesicles, and immunofluorescence on A) The anti-AyNCX A antibodies recognize a~100 kDa and~75 kDa protein in homogenized A. yongei tissue. Both bands are eliminated when the antibody is pre-absorbed with the epitope peptide overnight, and neither band is present when the membrane is incubated with the pre-immune serum. All sample wells contain the same amount of protein and all three Western Blot images were taken at the same exposure. B) Immunofluorescence microscopy of A. yongei tissue reveals AyNCX A is present in all four tissue layers, including the calicodermis. C) Pre-absorption of antibodies with antigen peptide eliminates signal at the same exposure, confirming antibody specificity. https://doi.org/10.1371/journal.pone.0205367.g003 Vesicular NCX in coral cells native coral tissues using specific, custom made antibodies. Interestingly, Na + /Ca 2+ activity is also found in secretory vesicles of mammalian cells [50][51][52], suggesting vesicular NCXs may also exist in mammals.
Fluorescence microscopy indicates AyNCX A is expressed in all four tissue layers of coral; therefore, NCX A is likely involved in general Ca 2+ homeostasis processes such as Ca 2+ sequestration in vesicles. Maintaining a low Ca 2+ concentration inside cells is essential because Ca 2+ accumulation in the cytoplasm would precipitate phosphates, interfere with intracellular signaling pathways, and be generally toxic to cells (reviewed in [53]).
Additionally, the relative higher AyNCX A abundance in the calicodermis suggests this Ca 2+ transporter may be part of a calcification mechanism that relies on intracellular vesicles. Such mechanism used to be favored in early coral research based on the presence of numerous vesicles in the calicoblastic cells of multiple coral species [3,[54][55][56][57]. Moreover, it was proposed that those vesicles belonged to the Golgi secretory pathway, and that the vesicular membrane regulated Ca 2+ transport and CaCO 3 generation [56]. However, those vesicles were not always observed to contain mineralized structures within, and were not always observed fusing with the cell membrane [55]. Although those discrepancies could be at least partially explained by fixation artifacts [55], the model of coral intracellular calcification lost support and was replaced by transcellular Ca 2+ transport through the cytoplasm of calicoblastic cells (reviewed in [4,14]), paracellular transport of Ca 2+ regulated by septate junction between calicoblastic cells [58], or bulk transport of seawater to the site of skeleton formation [59]. Importantly, those mechanisms are not mutually exclusive, and all of them involve calicoblastic cells being exposed to high Ca 2+ levels and therefore the need for robust Ca 2+ homeostatic regulation.
More recent studies have revived the model for intracellular coral calcification. Amorphous CaCO 3 was detected as~400 nm particles throughout Stylophora pistillata tissues and then in the skeleton [8], and proteins such as CARPs and carbonic anhydrase were identified in the skeleton matrix and found to have secretory signal peptides [9,10,60,61] (implying exocytosis). In A. yongei, vesicles in the process of fusing with apical membrane of calicoblastic cells are readily visible by TEM [7], Fig 5). The abundance, size and localization of those vesicles are consistent with the punctate AyNCX A immunostaining pattern, and also matches PMCA's [7]). Unfortunately, our attempts of immunogold-TEM staining in coral calicoblastic cells have so far been unsuccessful (also see [7]), likely due to the fact that the harsh fixation essential to preserve their complex cellular morphology is not compatible with immunohistochemistry techniques. In fact, to our knowledge this technique has never been performed successfully in coral calicoblastic cells. Nonetheless, the combined evidence indicates AyNCX A is present in vesicles in calicoblastic cells, where it could be regulating intracellular Ca 2+ homeostasis, participating in skeleton formation, and most likely both.
From an environmental perspective, coral intracellular calcification would confer corals certain resilience to environmental changes in pH and [CO 3 -], as recently discussed in detail [8,11]. Thus, vesicular transport of Ca 2+ and amorphous CaCO 3 from coral calcifying cells to the skeleton has several important implications. Although the current study suggests NCX A is involved in an intracellular vesicular mechanism for coral calcification, this must be confirmed by functional studies. In this respect, the small size and convoluted morphology of coral calicoblatic cells, together with a lack of tools for studying coral cellular physiology are major limitations (reviewed in [5]). If NCX A was indeed important for coral calcification, it could be used as a biomarker for coral calcification responses to environmental stress; for example by quantifying its mRNA and protein abundance.  [62] are highlighted yellow and blue, respectively. The protein alignment was made using EMBOSS Needle [63]. (PDF) Vesicular NCX in coral cells

S2 Fig. Sea urchin embryos expressing coral AyNCX A tagged with fluorescent protein.
A-C) AyNCX A with Cerulean Fluorescent Protein (CFP) at the N-terminus. D-F) AyNCX A with CFP at the C-terminus. G-I) AyNCX A with mCherry fluorescent protein at the N-terminus. J-L) AyNCX A with mCherry fluorescent protein at the C-terminus. For each set of 3 images, the left image (A,D,G,J) shows a single z-stack at the base of the embryo, the middle (B,E,H,K) shows a z-stack through the middle of the embryo, and the right (C,F,I,L) is a z-project of all z-stacks. Embryos expressing CFP-tagged AyNCX A were imaged 16hpf, embryos expressing mCherry-tagged AyNCX A were imaged 24hpf.