Sixteen T. squamosa weighing 550 ± 150 g were imported directly from Vietnam through Xanh Tuoi Tropical Fish Co. Ltd. On arrival, the specimens were distributed into three tanks, each with a dimension of 92 cm (L) by 62 cm (W) by 62 cm (H) containing approximately 320 l of seawater at Salinity 30–32 and 26°C. Artificial seawater was prepared with Red Sea salt (Red Sea, Houston, TX, United States). The salinity and temperature of the seawater were monitored using a Pro30 conductivity meter (YSI Incorporated, Yellow Springs, OH, United States). The pH of the seawater was maintained at 8.2–8.4; the hardness at 143–179 ppm; the calcium content at 380–420 ppm; the phosphate content at <0.28 ppm; the total ammonia and NO₃^– contents at 0 ppm. Each tank was illuminated with two sets of four feet Aquazonic T5 lighting systems, and each system consisted of two white light tubes and two actinic blue light tubes. The underwater light intensity (photosynthetic photon flux density; PPFD) reaching the clams was ∼115 − 125 μmol photons m^–2 s^–1 (400−700 nm) as determined by a SKP 215 PAR Quantum sensor connected to the SKP 200 display meter (Skye Instruments Ltd, United Kingdom). This level of irradiance mimicked the light intensity received by T. squamosa in its natural habitat at a depth of ∼20 m (Jantzen et al., 2008). No food was supplied to the giant clams during the 1 month of acclimatization under a 12 h: 12 h dark: light regimen. Approval on the use of giant clams in this study was exempted by the Nanyang Technological University Institutional Animal Care and Use Committee.

After 1 month of acclimatization, individuals of T. squamosa (n = 13) that had been exposed to light for 3 h were sampled randomly from the three tanks. Three hours of light exposure is chosen as giant clams are known to display light-enhanced phenomena, and the transcript levels of many transporters and enzymes could be enhanced after 3 – 6 h of illumination (see Ip and Chew, 2021 for a review). To minimize stress on the clams, they were anesthetized in 0.2% phenoxyethanol prior to tissue sampling. The shell valves were forced open to sever the adductor muscle. Samples of the outer mantle, inner mantle, foot muscle, hepatopancreas, and ctenidium (gill) were excised. Excised tissue samples were blotted dry, freeze-clamped in liquid nitrogen, and stored at −80°C until further processing. Separately, tissue samples of the outer mantle were collected from three other individuals of T. squamosa for immunofluorescence microscopy (n = 3). Excised outer mantle tissues were fixed in 3.7% paraformaldehyde prepared with seawater for 18 h at 4°C.

The total RNA of a tissue sample was extracted using TRI Reagent™ (Sigma-Aldrich Co., St Louis, MO, United States), and purified with a PureLink™ RNA Mini Kit (Thermo Fisher Scientific, Waltham, MA, United States). The concentration of the purified RNA was determined using a NanoDrop ND-1000 spectrophotometer (Nanodrop Technologies Inc., Wilmington, DE, United States), and the RNA integrity was checked by agarose gel electrophoresis. A RevertAid first strand cDNA synthesis kit (Thermo Fisher Scientific) was used to convert the purified RNA into cDNA.

The partial sequences of NRT2 from Symbiodinium, Cladocopium, and Durusdinium were obtained using a set of genus-comprehensive PCR primers (Forward: 5′-GCACT GTTCAGCAGAATCC-3′; Reverse: 5′-GGCTGTGAGTTGTC CACCA-3′) designed at the homologous regions of nine NRT2 sequences obtained from various dinoflagellate databases (Supplementary Table S1). The PCR reaction was performed using a 9902 Veriti 96-well thermal cycler (Thermo Fisher Scientific) with the cycling conditions: 94°C for 3 min, followed by 40 cycles of 94°C for 30 s, 57°C for 30 s, 72°C for 1.5 min and a final extension at 72°C for 10 min. The pGEM^®-T Easy Vector system II (Promega, Madison, WI, United States) was used to clone the PCR products obtained. Sixty clones were picked randomly and sequenced. The partial sequences obtained were identified by comparing with NRT sequences available in multiple dinoflagellate databases. A major sequence of NRT2 was identified for each genus of Symbiodinium, Cladocopium, and Durusdinium. To obtain the full coding sequence of the major Symb-NRT2, Clad-NRT2 and Duru-NRT2, 5′ and 3′ Rapid Amplification of cDNA Ends (RACE)-PCR were performed using the SMARTer RACE cDNA amplification kit (Clontech Laboratories, Mountain View, CA, United States). Three sets of RACE primers were designed specifically for Symb-NRT2 (5′ RACE: 5′-AGCTGGAACTGCTGCACAGTCCGTGA-3′, 3′ RACE: 5′-TGTCATCACGGACTGTGCAGCAGTTCCA-3′), Clad-NRT2 (5′ RACE: 5′-GCAGTCCCACTCGCTAAAATGTCC-3′; 3′ RACE: 5′-ACTGCAATCCCGTGCCACAGGACATT-3′) and Duru-NRT2 (5′ RACE: 5′-GTGCAGTAGCATTGTT GGCGATATCGG-3′; 3′ RACE: 5′-TTGAAGTACAAGAA CATTTCCACGACGG-3′). The full coding sequences of Symb-NRT2, Clad-NRT2, and Duru-NRT2 were deposited into Genbank.

The ExPASy Proteomic server¹ was used to deduce the amino acid sequences of Symb-NRT2, Clad-NRT2, and Duru-NRT2 from their respective nucleotide sequences. TMpred provided by Expasy², was used to identify the transmembrane regions and pore lining amino acid residues. The identities of Symb-NRT2, Clad-NRT2 and Duru-NRT2 were confirmed by conducting a phenogramic analysis together with NRT2 sequences obtained from various databases. The phenogram was generated using Maximum Likelihood analysis using the program RaxML 8.2.5 (Stamatakis, 2014) with 2000 bootstraps. Using ModelGenerator v0.85 (Keane et al., 2006), the best-fitting evolutionary model for NRT2 was determined to be WAG + G + F (Whelan and Goldman, 2001).

Three sets of genus-specific qPCR primers, one each for Symb-NRT2 (forward: 5′-TGAAGACAGGTCTGGAGTA-3′; reverse: 5′-CGCATATGGGCTCTTCT-3′), Clad-NRT2 (forward: 5′-AGAATGATGATACCAATCCCAC-3′; reverse: 5′-CAAA CACAGTCCGCCAG-3′), and Duru-NRT2 (forward: 5′-GAAGTACAAGAACATTTCCACGAC-3′; reverse: 5′-AAAC GCACTTGGACAGCAC-3′) were designed by aligning Symb-NRT2, Clad-NRT2, and Duru-NRT2 with nine NRT2 sequences selected from various dinoflagellate databases (Supplementary Table S2). In order to verify the specificity of the designed Symb-NRT2 primer, Clad-NRT2 primer, and Duru-NRT2 primer, efforts were made to generate three different plasmid clones, each of which contained the insert of the amplicon region of Symb-NRT2, Clad-NRT2, or Duru-NRT2 following the method of Hiong et al. (2017). Then, qPCR was performed using these three plasmid clones as substrates to confirm that each set of genus-specific primers would only react with the plasmid containing the corresponding insert. The amplification efficiencies of the qPCR primers for Symb-NRT2, Clad-NRT2, and Duru-NRT2 were 102.4, 92.9, and 101.9%, respectively. Genus-specific qPCR primers designed by Poo et al. (2021) were adopted to quantify the transcript levels of Symb-rbcII, Clad-rbcII, and Duru-rbcII from T. squamosa. The amplification efficiencies of the primer set for Symb-rbcII, Clad-rbcII, and Duru-rbcII were 95.1, 95.1, and 112.0%, respectively.

qPCR was performed using a 96-well StepOnePlus™ Real-Time PCR System (Thermo Fisher Scientific). Each reaction, in a total volume of 10 μl, consisted of 5 μl of qPCRBIO SyGreen Mix Hi-ROX (PCR Biosystems Inc., Wayne, PA, United States), 0.3 μl of forward primer (10 μmol l^–1), 0.3 μl of reverse primer (10 μmol l^–1), and an appropriate amount of cDNA. The qPCR cycling conditions included a 20 s denaturation and enzyme activation at 95°C, followed with 40 cycles of 95°C for 3 s and a specific temperature for a certain gene for 30 s. The specific temperature that lasted 30 s for Symb-rbcII, Clad-rbcII, and Duru-rbcII were 58, 56, and 55°C, respectively. For Symb-NRT2, Clad-NRT2, and Duru-NRT2, the respective temperature used in the 40 cycles were 57, 60, and 60°C. The dissociation curve obtained after each run was analyzed to verify the homogeneity of the PCR product and the specificity of the PCR reaction. Three standard curves were constructed using the three different plasmid clones as standards for Symb-rbcII, Clad-rbcII, and Duru-rbcII. The transcript levels Symb-rbcII, Clad-rbcII, and Duru-rbcII in a sample were calculated based on these three standard curves and expressed as copies of transcripts per ng of total RNA.

A genus-comprehensive anti-Zoox-NRT2 antibody was custom-made by Genscript (Piscataway, NJ, United States) based on the epitope sequence of MADFKLKVDESNKA, which was selected from a highly conserved region of six NRT2 sequences retrieved from various dinoflagellate databases (Supplementary Table S3). This epitope sequence, corresponding to residues 1−14 of Symb-NRT2, Clad-NRT2, and Duru-NRT2 with similarity of 100, 71.4, and 78.6%, respectively. Thus, anti-Zoox-NRT2 could possibly bind with NRT2s of all phylotypes of Symbiodinium, Cladocopium, and Durusdinium.

The fixed outer mantle sample was dehydrated in ethanol and cleared using HistoChoice Clearing Agent (Sigma-Aldrich Co.) before embedding in Paraplast Plus (Sigma-Aldrich Co.). Sections of 5 μm was prepared using a Leica RM2125 RTS microtome (Leica, Wetzlar, Germany) and mounted on Menzel Gläser SuperFrost Plus Adhesion slides (Thermo Fisher Scientific). The deparaffinized section was treated with citraconic anhydrase (Nacalai Tesque, Kyoto, Japan) at 95°C for 5 min, followed with 1% SDS solution at 25°C for 10 min for the retrieval of antigen. The section was then washed with TPBS containing 0.2% Triton-X, 10 mmol l^–1 Na₂HPO₄, 1.8 mmol l^–1 KH₂PO₄, 137 mmol l^–1 NaCl, and 1.8 mmol l^–1 KCl at pH 7.4. To reduce autofluorescence, the section was treated with 0.1% Sudan Black B (Sigma-Aldrich Co.) in 70% ethanol for 10 min and washed three times with TPBS. Blocking was performed with 1% bovine serum albumin in TPBS at 25°C for 1 h. Thereafter, the section was incubated with the anti-Zoox-NRT2 antibody (2.5 μg ml^–1 diluted with Signal Enhancer HIKARI Solution A obtained from Nacalai Tesque) at 25°C for 1 h and rinsed three times with TPBS. The section was incubated with 2.5 μg ml^–1 of fluorochrome-coupled goat anti-rabbit gamma globulin diluted with Signal Enhancer HIKARI Solution A (Alexa Fluor 488; Thermo Fisher Scientific) for 1 h at 25°C. Finally, they were mounted in Prolong Gold antifade reagent (Thermo Fisher Scientific Inc.) for microscopy.

The mounted section was examined under a fluorescence microscope (Olympus BX43F; Olympus Corporation, Tokyo, Japan) and the images were acquired using an Olympus DP80 camera and the cellSens Imaging software (Olympus). Differential interference contrast microscopy (DIC) was applied to examine tissue structures and orientation. The red autofluorescence of the plastids of dinoflagellates was examined using the U-MWIG Interference Green Fluorescence Filter with an excitation wavelength of 520–550 nm. The green fluorescence resulting from the staining by the primary antibody and Alexa Fluor 488 was acquired using the Olympus U-WNIBA Blue Fluorescence Filter with an excitation wavelength of 470–490 nm. Overlaying of the images and adjustment of brightness were performed using Adobe Photoshop CC (Adobe Systems, CA, United States).

Values were reported as means ± SEM unless otherwise stated. The non-parametric Friedman test of differences followed by the Wilcoxon Signed-Rank Test were applied for data expressed as percentages or ratios in Tables 4–8 and Figure 4. Differences obtained among means were considered statistically significant with p-values < 0.017 after Bonferroni adjustment. For Table 3, One-way Analysis of Variance (ANOVA) was used for comparison among the means of the transcript levels of the genus-specific NRT2 or those of genus-specific rbcII in a particular organ while Levene’s test was used to assess the homogeneity of the variance. This was followed by Dunnett’s T3 test as the variance was assumed not to be equal. On the other hand, the paired t-test was used to compare the means between the genus-specific NRT2 and the corresponding genus-specific rbcII in a particular organ in Table 3. The differences between the two means were regarded as significant when the p-value was <0.05. All comparisons were performed with the use of SPSS Statistics software v26 (IBM Corporation, Armonk, NY, United States).

The complete cDNA coding sequences of Symb-NRT2, Clad NRT2, and Duru-NRT2 obtained from the outer mantle of T. squamosa comprised 1614, 1659, and 1635 bp, respectively. They have been deposited into GenBank with the respective accession numbers of MZ014639, MZ014640, and MZ014641. A comparison of these three NRT2 sequences with NRT2 contigs from multiple dinoflagellate databases revealed that Symb-NRT2 and Clad-NRT2 had the highest similarity (98.6%) to the NRT2 sequence of Symbiodinium tridacnidorum (ITS2 type A3; Shoguchi et al., 2018; 98.6%) and the NRT2 sequence of Cladocopium goreaui (ITS2 type C1; Davies et al., 2018; 99.5%), respectively (Table 1). For Duru-NRT2, it had the highest similarity to the NRT2 sequence of Durusdinium trenchii (ITS2 type D1a; Bellantuono et al., 2019; 99.3%; Table 1).

The deduced amino acid sequences of Symb-NRT2, Clad- NRT2 and Duru-NRT2 contained 538 (∼58.5 kDa), 553 (∼60.2 kDa), and 545 (∼59.4 kDa) residues, respectively. A multiple alignment of Symb-NRT2, Clad-NRT2, and Duru-NRT2 with other NRT sequences from Genbank demonstrated that they consisted of 12 predicted transmembrane regions (TMs), with an intracellular N-terminus and an intracellular C-terminus (TM 1−TM 12; Figure 1). They contained a conserved MFS motif (Forde, 2000) between TM 2 and TM 3 (G-x-x-x-D/N-R/K-x-G-R-R/K) (corresponding to amino acid residues 169–178 in Symb-NRT2, 166–175 in Clad-NRT2, and 170–179 in Duru-NRT2). Two nitrate signature motifs (Trueman et al., 1996; Unkles et al., 2004a,2012) were present in Symb-NRT2 (residues 229–249 and residues 456–476), Clad-NRT2 (residues 226–246 and residues 453–473), and Duru-NRT2 (residues 230–250 and residues 457–477). In addition, the putative substrate-binding site that could form hydrogen bonds with NO₃^– (Yan et al., 2013) was also present in Symb-NRT2, Clad-NRT2, and Duru-NRT2. This binding site consisted of two positively charged amino acid residues (corresponding to Arg-165 and Arg-382 in Symb-NRT2, Arg-162 and Arg-379 in Clad-NRT2, and Arg-166 and Arg-383 in Duru-NRT2) and two polar residues (corresponding to Asn-246 and Tyr-338 in Symb-NRT2, Asn-243 and Tyr-335 in Clad-NRT2, and Asn-247 and Tyr-339 in Duru-NRT2). The Glu residue involved in the symport of H⁺ and NO₃^– (Akhtar et al., 2015; Jacquot et al., 2017) was conserved (corresponding to Glu-345 in Symb-NRT2, Glu-342 in Clad-NRT2, and Glu-346 in Duru-NRT2), implying that Symb-NRT2, Clad-NRT2, and Duru-NRT2 could act as H⁺-dependent symporters.

The identities of Symb-NRT2, Clad-NRT2, and Duru-NRT2 were supported by phenogramic analysis (Figure 2), which grouped them with NRT2 sequences of algae, plants, and dinoflagellates, but distinct from NRT1. Notably, they were clustered with other NRT2 sequences from the same genus of dinoflagellate.

NRT2 immuno-labeling was detected at the plasma membrane of the coccoid dinoflagellates in the outer mantle of T. squamosa, in support of a possible role in transporting NO₃^– across the plasma membrane (Figure 3). NRT2 immunofluorescence was also detected in certain intracellular vesicles inside the cytoplasm of the coccoid dinoflagellates.

The genus-specificity of the primer sets designed for Symb-NRT2, Clad-NRT2 and Duru-NRT2 were validated with three plasmid clones, each containing the insert of one of the three targeted amplicons. Indeed, a specific set of primer reacted positively only with its related plasmid clone (Table 2). Poo et al. (2021) had verified the genus-specificity of the primer sets designed for Symb-rbcII, Clad-rbcII, and Duru-rbcII previously.

A comparison of genus-specific NRT2 and rbcII in T. squamosa indicates that the mean transcript levels of Symb-NRT2 were significantly higher than those of Symb-rbcII in the outer mantle, foot muscle and hepatopancreas, but they were comparable in the inner mantle and ctenidium (n = 13; Table 3). Of note, the most prominent difference in transcript level of Symb-NRT2 and Symb-rbcII was observed in the outer mantle. By contrast, the mean transcript levels of Clad-NRT2 were either significantly lower than (as in the outer mantle and ctenidium) or comparable to (as in the inner mantle, foot muscle and hepatopancreas) the transcript levels of Clad-rbcII. The transcript levels of Duru-NRT2 were significantly lower than those of Duru-rbcII in these five organs (n = 13; Table 3).

For rbcII, the mean transcript level of Duru-rbcII was significantly higher than those of Symb-rbcII and Clad-rbcII (n = 13; Table 3) in all the five organs studied, indicating that Durusdinium was the major dinoflagellate genus in the individuals of T. squamosa obtained from Vietnam (Figure 4). Based on ANOVA, the transcript levels of Clad-rbcII were not significantly different from those of Symb-rbcII in the five organs because equal variance could not be assumed (n = 13; Table 3). Overall, the percentage proportions of transcript levels of rbcII derived from the three genera of Symbiodiniaceae dinoflagellates in the five organs of these 13 individuals of T. squamosa were Duru-rbcII > > Clad-rbcII = Symb-rbcII (Figure 4).

For NRT2, the mean transcript level of Duru-NRT2 was also significantly higher than those of Symb-NRT2 and Clad-NRT2 in four of the organs studied, except the hepatopancreas (n = 13; Table 3), corroborating the proposition that Durusdinium was the major dinoflagellate genus in these T. squamosa individuals (Figure 4). Furthermore, the mean transcript levels of Clad-NRT2 were statistically comparable to those of Symb-NRT2 in the outer mantle, inner mantle, foot muscle, and ctenidium (Table 3) due to the high variation of transcript levels among different individuals (n = 13). By contrast, the mean transcript level of Symb-NRT2 was significantly higher than that of Clad-NRT2 in the hepatopancreas (n = 13; Table 3). Hence, unlike rbcII, the percentage proportions of transcript levels of NRT2 derived from the three genera of Symbiodiniaceae dinoflagellates varied among the five organs of T. squamosa (Figure 4). For instance, the estimated proportions of Symb-rbcII transcripts in the outer mantle, hepatopancreas and ctenidium were 0.8, 3.0, and 1.9%, respectively, which were considerably lower than the estimated proportions of Symb-NRT2 transcripts in these three organs (outer mantle, 10.5%; hepatopancreas, 28.3%; ctenidium, 4.4%; Figure 4).

The mean transcript ratios of Symb-NRT2/Symb-rbcII (N = 13) for the outer mantle (Table 4), inner mantle (Table 5), foot muscle (Table 6), and hepatopancreas (Table 7) were > 1.0 because the transcript levels of Symb-NRT2 were generally higher than those of Symb-rbcII in these organs (Table 3). The mean transcript ratios of Symb-NRT2/Symb-rbcII (n = 13) for the outer mantle (8.4; Table 4) was significantly higher than those for the inner mantle (4.8; Table 5), foot muscle (4.0; Table 6) and hepatopancreas (4.3; Table 7). For each of these four organs, the mean transcript ratios of Symb-NRT2/Symb-rbcII (n = 13) were significantly higher than the mean transcript ratios of Clad-NRT2/Clad-rbcII and Duru-NRT2/Duru-rbcII, which were statistically comparable to each other (Tables 4–7).

For the outer mantle, while 12 of the 13 individuals examined displayed values > 1.0 for Symb-NRT2/Symb-rbcII, clam 3 had a distinctly low Symb-NRT2/Symb-rbcII value of 0.41 (Table 4). Of the 13 individuals examined, eight had ratios of Clad-NRT2/Clad-rbcII ranging between 0.39 and 0.88, but clams 2, 7, 9, 11, and 13 had noticeably lower Clad-NRT2/Clad-rbcII ratios of 0.018, 0.010, 0.022, 0.004, and 0.007, respectively (Table 4). In comparison, the transcript ratios of Duru-NRT2/Duru-rbcII were relatively constant in the outer mantle of these 13 individuals.

For the inner mantle, 10 of the 13 individuals examined displayed Symb-NRT2/Symb-rbcII values > 1.0, but clams 2, 3, and 10 had values < 1.0 (Table 5). Of the 13 individuals examined, three (clams 1, 4, and 8) had Clad-NRT2/Clad-rbcII transcript ratios > 1.0 (ranging from 1.0 to 1.5; Table 5). For the other 10 individuals, six of them (clams 2, 6, 7, 9, 11, and 13) had values close to or below 0.10. The transcript ratios of Duru-NRT2/Duru-rbcII in the inner mantle were again relatively constant among the 13 individuals.

For the foot muscle, out of the 13 individuals examined, 11 of them displayed Symb-NRT2/Symb-rbcII ratios > 1.0, except for clams 9 and 11 (Table 6). One individual (clam 8) had a Clad-NRT2/Clad-rbcII > 1.0 and clams 4, 6, 7, 9, 10 and 11 had values ≤ 0.17 (Table 6). In comparison, the transcript ratios of Duru-NRT2/Duru-rbcII in the foot muscle were relatively constant among the 13 individuals.

For the hepatopancreas, two (clam 3 and 10) of the 13 individuals examined had Symb-NRT2/Symb-rbcII values < 1.0 (Table 7). One individual (clam 9) had a Clad-NRT2/Clad-rbcII > 1.0, and five individuals (clams 2, 4, 7, 11, and 13) had values ≤ 0.16 (Table 7). In comparison, the transcript ratios of Duru-NRT2/Duru-rbcII in the hepatopancreas were relatively constant among the 13 individuals.

For the ctenidium, the mean transcript ratio of Symb-NRT2/Symb-rbcII (n = 13; Table 8) was <1.0 because the transcript level of Symb-NRT2 was generally lower than that of Symb-rbcII (Table 3), which was dissimilar to the other four organs. Nevertheless, the mean transcript ratio of Symb-NRT2/Symb-rbcII (n = 13) was significantly higher than that of Clad-NRT2/Clad-rbcII, but comparable to that of Duru-NRT2/Duru-rbcII (Table 8).

Out of the 13 individuals examined, only clams 1 and 13 had Symb-NRT2/Symb-rbcII values > 1.0 (Table 8). As a result, the mean transcript ratio of Symb-NRT2/Symb-rbcII for the ctenidium (0.77; Table 8) was the lowest among the five organs studied (Tables 4–7). The transcript ratios of Clad-NRT2/Clad-rbcII in the ctenidium of all 13 individuals were <1.0, with five of them having particularly low values (<0.10). In comparison, the transcript ratios of Duru-NRT2/Duru-rbcII in the hepatopancreas were relatively constant among the 13 individuals.

Symb-NRT2, Clad-NRT2 and Duru-NRT2 were probably derived from S. tridacnidorum, C. goreaui, and D. trenchii, respectively. Similar to other members of MFS, Symb-NRT2, Clad-NRT2, and Duru-NRT2 had 12 TMs, consisting of two sets of six transmembrane domains linked by a cytosolic loop (see review by Forde, 2000). In addition, Symb-NRT2, Clad-NRT2, and Duru-NRT2 contained a conserved MFS motif between TM 2 and TM 3 (Forde, 2000). A distinguishing feature of the NNP among members of the MFS is the presence of two glycine-rich nitrate signature motifs (Trueman et al., 1996; Unkles et al., 2004a,2012; Yan et al., 2013), and these motifs were present in Symb-NRT2, Clad-NRT2 and Duru-NRT2. By contrast, members of the NPF family have only one signature motif (F-Y-x-x-I-N-x-G-S-L) in TM 5 (Steiner et al., 1995). Symb-NRT2, Clad-NRT2, and Duru-NRT2 could form hydrogen bonds with NO₃^– because of the presence of the putative NO₃^– binding site (Yan et al., 2013). The presence of the residue equivalent to Glu-330 in NrtA of Aspergillus nidulans (Unkles et al., 2004a,b), which is known to be involved in the symport of H⁺ and NO₃^– (Akhtar et al., 2015; Jacquot et al., 2017), in Symb-NRT2, Clad-NRT2, and Duru-NRT2 implies that they are H⁺-dependent symporters.

NRT2 is localized at the plasma membrane of the coccoid dinoflagellates of T. squamosa, indicating that it can transport NO₃^– from the luminal fluid of the zooxanthellal tubule into the symbionts. The intracellular vesicles that display NRT2 immunofluorescence could be involved in the transfer of NRT2 from the internal membranes to the plasma membrane as suggested previously for Arabidopsis thaliana (Wirth et al., 2007). Hence, the transcript levels of Symb-NRT2, Clad-NRT2, and Duru-NRT2 can be appropriately used as molecular indicators to estimate the potential of NO₃^– transport (or uptake) in Symbiodinium, Cladocopium, and Durusdinium, respectively.

Molecular characterization indicates that Symb-NRT2, Clad-NRT2, and Duru-NRT2 probably rely on an inwardly directed [H⁺] gradient to drive the active absorption of NO₃^–. Recently, Mani et al. (2021) reported that coccoid dinoflagellates of T. squamosa could excrete H⁺ into the luminal fluid of zooxanthellal tubules through the merging of intracellular vesicles containing vacuolar-type H⁺-ATPase (VHA) subunit B (VHA-B) with the plasma membrane. The excreted H⁺ could augment the dehydration of luminal HCO₃^– and promote the absorption of CO₂ by the photosynthesizing symbionts. The inwardly directed [H⁺] gradient generated by VHA across the plasma membrane could also drive the transport of NO₃^– through NRT2 into the symbiont. The absorbed NO₃^– could be reduced to NH₄⁺. Then, a portion of the carbon fixed during photosynthesis could be utilized to assimilate NH₄⁺ into amino acids catalyzed by glutamate dehydrogenase, glutamine synthetase and glutamine oxoglutarate aminotransferase (Padgett and Leonard, 1996). Some of the amino acids synthesized could be donated to the host in support of its growth and development. Of note, glutamine synthetase and glutamine oxoglutarate aminotransferase have been cloned from dinoflagellates of T. squamosa (Fam et al., 2018; Teh et al., 2021).

Cladocopium has been reported as the major genus of dinoflagellate found in T. squamosa from Japan (Ikeda et al., 2017), which is temperate and subtropical. However, based on the transcript levels of Symb-rbcII, Clad-rbcII, and Duru-rbcII, Poo et al. (2021) reported that T. squamosa (n = 4) obtained from Vietnam contained mainly Durusdinium (85–95%). Although we also worked with T. squamosa from Vietnam, the individuals involved were different from those examined by Poo et al. (2021). In general, the transcript levels of Symb-rbcII, Clad-rbcII, and Duru-rbcII reported by Poo et al. (2021) were higher than the corresponding results obtained in this study, which could be due to differences in the environmental conditions in the natural habitats of these two batches of T. squamosa. Nonetheless, in agreement with Poo et al. (2021), results of this study (n = 13) also denoted Durusdinium as the dominant genus of dinoflagellates in our experimental animals based on the transcript levels of NRT2 (78–92%) and rbcII (88–94%). This could be related to the tropical environmental conditions of and the availability of Symbiodiniaceae dinoflagellate phylotypes in Vietnam waters. Durusdinium is generally tolerant of a variety of environmental stressors (Brown et al., 2002; van Oppen et al., 2009), including high light intensity that would result in bleaching (Kemp et al., 2014) and turbid reef environments (Ulstrup and Van Oppen, 2003; LaJeunesse et al., 2010; Tonk et al., 2013). Thus, the high abundance of Durusdinium in the outer mantle of T. squamosa from Vietnam could have provided these clams with an advantage under stressful conditions, particularly at elevated temperature in the tropics. Based on the transcript level of Symb-rbcII, individuals of T. squamosa obtained from Vietnam contained relatively small populations of Symbiodinium (1–13%, Poo et al., 2021; 0.1–3%, this study), which could be due to the rarity of Symbiodinium in Indo-Pacific waters as compared to the Red Sea and the Caribbean waters (LaJeunesse, 2002; Baker, 2003).

There are discrepancies in the percentage proportions of Symb-rbcII and Symb-NRT2 transcripts in three organs of the 13 individuals of T. squamosa examined in this study. The percentage proportions of Symb-rbcII in the outer mantle, ctenidium and hepatopancreas (0.8, 1.9, and 3.0%, respectively) were apparently lower than the percentage proportions of Symb-NRT2 in the same organ (10.5, 4.4, and 28.3%, respectively). A logical explanation is that some phylotypes of Symbiodinium in these three organs expressed more transcripts of NRT2 per dinoflagellate cell than phylotypes of Cladocopium and Durusdinium. Hence, our results indicate the presence of different phylotypes of Symbiodinium with different expression levels of NRT2 in the five organs T. squamosa. They also signify that some Symbiodinium phylotypes had a higher potential of NO₃^– transport than the phototrophic potential. Thus, it was crucial to analyze the transcript ratios of NRT2 to rbcII.

The transcript ratio of NRT2/rbcII estimated for each genus of dinoflagellates (e.g., Symb-NRT2/Symb-rbcII) could indicate whether that genus would have a high or low potential of NO₃^– transport relative to the potential of phototrophy. Any difference in the transcript ratios among the three genera of dinoflagellates would suggest different potentials of NO₃^– transport and therefore different abilities to assimilate NO₃^– for the production of amino acids and other nitrogenous compounds. The mean transcript ratios of Symb-NRT2/Symb-rbcII were significantly higher than those of Clad-NRT2/Clad-rbcII and Duru-NRT2/Duru-rbcII in the outer mantle, inner mantle, foot muscle and hepatopancreas of T. squamosa, and the greatest differences were observed in the outer mantle. Hence, it can be concluded that the phylotypes of Symbiodinium generally had higher potential of NO₃^– transport than the phylotypes of Cladocopium and Durusdinium in these four organs. For the ctenidium, Symb-NRT2/Symb-rbcII was also significantly higher than Clad-NRT2/Clad-rbcII but comparable to Duru-NRT2/Duru-rbcII. These results imply that the composition of Symbiodinium phylotypes present in the ctenidium of T. squamosa were dissimilar to those present in the other four organs, and particularly distinct from those in the outer mantle.

Symbiodiniaceae dinoflagellates are phototrophic and express high transcript levels of rbcII (Poo et al., 2020, 2021). Yet, the mean transcript levels of Symb-NRT2 in the outer mantle, inner mantle, foot muscle and hepatopancreas of 13 T. squamosa individuals were substantially higher than the corresponding mean transcript levels of Symb-rbcII, resulting in the mean Symb-NRT2/Symb-rbcII ratios of 4.0–8.4 in these four organs. In the outer mantle, the mean Symb-NRT2/Symb-rbcII ratio of 8.4 was much greater than the mean Clad-NRT2/Clad-rbcII ratio of 0.39 and the mean Duru-NRT2/Duru-rbcII ratio of 0.63. These results are unusual as they indicate for the first time that certain phylotypes of Symbiodinium could specialize in NO₃^– absorption, and hence NO₃^–assimilation, more so than photosynthesis. They also indicate that the potentials of NO₃^– transport in these Symbiodinium phylotypes were greater than those in phylotypes of Cladocopium and Durusdinium in general.

Of the 13 individuals examined, 11 had ratios of Symb-NRT2/Symb-rbcII that ranged between 7.2 and 13.1 in the outer mantle of T. squamosa, but the ratios for clam 3 (0.41) and clam 10 (1.34) were distinctly < 7. These results indicate that the phylotypes of Symbiodinium in clam 3 and clam 10 had strikingly lower potential of NO₃^– transport than those in the other 11 clams. Hence, there could be two physiologically distinct phylotypes (species) of Symbiodinium as defined by high or low potential of NO₃^– transport.

To analyze our results, a Symb-NRT2/Symb-rbcII ratio of 1.0 (i.e., equal transcript levels of Symb-NRT2 and Symb-rbcII) was set arbitrarily to differentiate the Symbiodinium phylotypes with high potential of NO₃^– transport from those with low potential of NO₃^–. In the 13 individuals examined in this study, the numbers of clams that displayed a ratio of >1.0 for the outer mantle, inner mantle, foot muscle, hepatopancreas and ctenidium were 12, 10, 11, 11, and 2, respectively. These results indicate that the proportion of phylotypes of Symbiodinium with high potential of NO₃^– transport varied among the five organs, and that the ctenidium contained mainly Symbiodinium phylotypes with low potential of NO₃^– transport. It is noteworthy that while muscle tissues can be found in the other four organs, the ctenidium comprises mainly epithelial tissues dedicated for gases exchange, membrane transport and acid-base balance. It could be that the demand for using NO₃^– to form amino acid is low in the symbionts of the ctenidium in comparison with the symbionts of the other four organs. It could also imply that the symbionts of the ctenidium prefer to use other source of nitrogen, e.g., ammonia, for nitrogen metabolism. On the other hand, the predominant presence of Symbiodinium phylotypes with high potential of NO₃^– transport in the outer mantle facilitates the effective coupling of NO₃^– transport and assimilation with photosynthesis. Hence, it can be concluded that the phylotype compositions of Symbiodiniaceae dinoflagellates could vary among organs of T. squamosa based on the host’s physiological needs. Nonetheless, how the relevant phylotypes of Symbiodinium find the way to and establish in different organs of T. squamosa needs to be elucidated in the future.

Based on Clad-NRT2/Clad-rbcII, the outer mantle of T. squamosa contained two physiologically distinct phylotypes of Cladocopium. Of the 13 individuals examined, eight had ratios of Clad-NRT2/Clad-rbcII ranging between 0.39 and 0.88. However, clams 2, 7, 9, 11, and 13 had distinctly lower Clad-NRT2/Clad-rbcII ratios (<0.1) and hence lower potential of NO₃^– transport than the other eight individuals. An examination of Clad-NRT2/Clad-rbcII ratios in the inner mantle, foot muscle, and hepatopancreas indicates the presence of some phylotypes of Cladocopium with Clad-NRT2/Clad-rbcII slightly > 1. The inner mantle, foot muscle, hepatopancreas, and ctenidium apparently contained dissimilar proportions of Cladocopium phylotypes with either high or low potentials of NO₃^– transport. Overall, Cladocopium phylotypes in the ctenidium, with a mean Clad-NRT2/Clad-rbcII ratio of 0.21 (n = 13), appeared to have lower potentials of NO₃^– transport than those in the other four organs (0.32–0.57). Again, this might indicate the relative low demand for using NO₃^– to form amino acids in the ctenidium as compared with the other four organs.

Using NRT2 and rbcII as molecular indicators, our results indicate for the first time that Symbiodiniaceae dinoflagellates harbored by T. squamosa could have different potentials of NO₃^– transport, and offer insights into the functional diversity among coccoid dinoflagellates at the genus level (Symbiodinium, Cladocopium, and Durusdinium). A corollary of some phylotypes of Symbiodinium having a preference of NO₃^– uptake and assimilation is that different members of Symbiodiniaceae could be specialized in absorbing and utilizing distinct types of nitrogenous compounds (e.g., ammonia, urea and glutamine) furnished by the host. An important implication is that it would not be possible for the clam host to regulate its general symbiont population by simply reducing the availability of endogenous ammonia to them through the recycling of metabolic ammonia into non-essential amino acids, as has been suggested for scleractinian corals (Falkowski et al., 1993; Cui et al., 2019; Xiang et al., 2020). This is particularly the case for those Symbiodinium phylotypes that can effectively absorb and assimilate NO₃^–. Rather, the clam host might be able to regulate the population of a specific Symbiodiniaceae phylotype in a certain organ by controlling the translocation of a particular nitrogen-containing compound from the hemolymph through the tubular epithelial cells into the luminal fluid of the zooxanthellal tubules.

Importantly, our results signify that, in the coccoid stage, certain phylotypes (species) of Symbiodinium might specialize in nitrogen metabolism more so than photosynthesis as reflected by their high potential of NO₃^– transport with transcript ratios of Symb-NRT2/Symb-rbcII much greater than one. This implies that different phylotypes of Symbiodiniaceae dinoflagellates could have different abilities to donate amino acids and carbohydrates to the host. As the host requires ∼20 different types of amino acid for growth and reproduction, it is highly unlikely that these amino acids are donated equally by all the phylotypes of dinoflagellates. Rather, there could be a division of labor among different phylotypes of coccoid dinoflagellates in the supply of nutrients (e.g., specific types of carbohydrate or amino acid) to the host. Therefore, efforts should be made in the future to develop a large variety of molecular indicators (e.g., transporters of ammonia, urea and glutamine, as well as enzymes involved in the syntheses of various amino acids) in order to elucidate the distinct physiological roles of various Symbiodiniaceae phylotypes in the giant clam-dinoflagellate holobiont.