Coccolith-bearing cells of Pleurochrysis (Chrysotila) haptonemofera were grown photoautotrophically, as described previously (Fujiwara et al., 2007). The cells were cultured in Daigo’s IMK medium (Nihon Pharmaceuticals, Tokyo, Japan) or artificial seawater-based Eppley’s medium (SLEP; Endo et al., 2016) at 20°C under continuous illumination at 40 μmol photon m⁻² s⁻¹ with constant bubbling of air.

A full-length PhTGS mRNA sequence was obtained by 5′-RACE and RT-PCR using primers designed based on a Kre6 homolog sequence that was found in the EST database Pleurochrysome (Yamamoto et al., 2016) and in de novo assembled sequences from RNA-seq data. Total RNA of P. haptonemofera was prepared using Isogen with a Spin Column (Nippon Gene, Japan), and then, cDNA synthesis and 5′-RACE were performed using the SMARTer^® RACE 5′/3′Kit (TaKaRa Bio, Japan). For 5′-RACE, a nested PCR was carried out, using PhTGS-5′RACE-1 and PhTGS-5′RACE-2 (Supplementary Table S1) as gene-specific primers for the first and second rounds of PCRs, respectively, and the PCR products were cloned in the plasmid pRACE. For the confirmation of the internal sequence, the coding region was amplified using primers PhTGS-pET-F and PhTGS-pET-R (Supplementary Table S1), and the PCR products were inserted into pET-15b (Novagen) by In-Fusion cloning (TaKaRa Bio). The insert in the obtained plasmid pET-15b-PhTGS was sequenced, as well as those of the 5′-RACE products (DDBJ; accession number, LC773529).

Searches for the signal peptide and transit peptide were conducted using SignalP 4.1 and ChloroP 1.1, respectively. For the search for the transit peptide, the sequence from which the signal peptide was removed was applied. A domain search was performed by NCBI’s Conserved Domain Search. The membrane protein and transmembrane helix were predicted using ExPASy ProtScale and SOSUI and TMHMM (v. 2.0) servers, respectively. InterPro was utilized to search the protein family membership, domain, homologous superfamily, and GO terms. WoLF PSORT was used for the prediction of cellular localization.

Sequences already annotated in the previous study were taken, as well as sequences from P. haptonemofera; we subsequently enriched using homologous research with BlastP against sequences of the non-redundant protein sequence database of the NCBI and sequences from other databases (MMETSP and data publicly available). We then aligned the sequences using MUSCLE (Edgar, 2004); block selection was carried out using BMGE with a block size of 4 and the BLOSUM30 similarity matrix. Trees were generated using IQ-TREE with 100 bootstrap repetitions using the LG4X model.

Three kinds of double-stranded RNA (dsRNA) were transferred into P. haptonemofera protoplasts by the method developed by Endo et al. (2016). For dsRNA preparation, DNA having the T7 RNA polymerase consensus binding site at the 5′-terminus was synthesized by PCR using PhTGS-ND-1F and-1R (for targeting 5′ half of the CDS), PhTGS-ND-2F and -2R (for targeting 3′ half of the CDS), and PhTGS-ND-3F and -3R (for targeting the region around 5′ end of the CDS) as primer pairs (Supplementary Table S1), and then, in vitro transcription was performed by T7 RNA polymerase (TaKaRa Bio) using the DNA fragment as a template. The synthesized RNA molecules with complementary sequences were annealed to synthesize dsRNA. P. haptonemofera protoplasts were prepared using the hypoosmotic K⁺ treatment method (Takayanagi et al., 2007) modified by Endo et al. (2016). The three kinds of dsRNA and control dsRNA (GFP dsRNA) were transferred into the protoplasts by the PEG-mediated method (Endo et al., 2016), and the dsRNA-introduced cultures were analyzed as KD-tgs_1, KD-tgs_2, KD-tgs_3, and control, respectively.

Total RNA isolated from cells at 3 days after RNAi using Isogen (Nippon Gene) was treated using the TURBO DNA-free™ Kit (Ambion, United States), and then reverse-transcribed using the QuantiTect Reverse Transcription Kit (QIAGEN, Germany). qPCR was performed using the Rotor-Gene SYBR^® Green PCR Kit (QIAGEN) using a Rotor-Gene Q Real-time PCR System (QIAGEN) and primer sets PhTGS-qPCR-F and -R and Actin-qPCR-F and -R. The PCR thermal profiles comprised 45 cycles of 95°C for 5 s and 60°C for 10 s.

Small-scale β-glucans were measured according to Hirokawa et al. (2008)’s method with a slight modification. Briefly, cells in 2 mL of cultures at 3 days after RNAi were harvested by centrifugation (18,500 × g) and then suspended in 300 μL of the McIlvain buffer (pH6.0). After sonication, methanol was added to the suspension, and the mixture was incubated at 4°C for more than 4 h. After centrifugation, the β-glucan-containing pellet was dried up and then solved in 25 μL of the McIlvain buffer. The β-glucan was converted to glucose by treatment using westase (TaKaRa Bio), and the total glucose content was determined using hexokinase (Oriental Yeast Co., Ltd, Japan) and glucose-6-phosphate dehydrogenase (Roche Diagnostics K.K., Germany), according to Bergmeyer et al. (1974). Westase is an enzyme cocktail, including β-1,6-glucanase and β-1,3-glucanase, for yeast protoplast preparation, and it is demonstrated to completely degrade the β-glucan of P. haptonemofera into glucose (Hirokawa et al., 2008).

Approximately 0.5 μL of BODIPY 493/503 (Life Technologies, United States) stock solution (1 mg/mL DCMU) was added to 100 μL of culture. After mixing and incubation for a couple of minutes, cellular fluorescence was detected via fluorescence microscopy (BX53 using the NIBA filter, Olympus, Japan) and flow cytometry (SH800, Sony, Japan). For flow cytometry, the fluorescence of BODIPY 493/503 and that of chlorophyll was detected using FL2 and FL4 detectors with filter sets for FITC (518 nm) and 7AAD (647 nm), respectively. Signals from 5,000 particles that were gated using FL4-area were collected using a logarithmic scale, according to a previous report (Takayanagi et al., 2007).

For the induction of the recombinant protein expression, logarithmic-phase cells of E. coli BL21 (DE3) carrying pET-15b-PhTGS were further cultivated in the presence of 0.5 mM IPTG at 20°C for 16 h. The cells collected and resuspended in PBS without KCl were disrupted by sonication using a Sonifier 250D instrument (Branson, United States), and then, a soluble fraction was obtained by centrifugation (20,000 g for 10 min). The soluble and insoluble fractions were applied to a 10% SDS-PAGE gel, and the proteins were visualized using Coomassie Brilliant blue R-250 stain (CBB).

Using the soluble fraction of rPhTGS-expressing E. coli, β-glucan-synthesizing activity was measured. Approximately 1.3 mg/mL of the protein-containing soluble fraction was incubated at 20°C for up to 24 h in a total of 100 μL of a reaction solution [0.01 mM DTT, 0.2 mg of P. haptonemofera β-glucan (Hirokawa et al., 2008), 3.2 mM UDP-glucose, and 0.75 nmol [U-¹⁴C] UDP-glucose (1.2 × 10¹⁰ Bq mmol⁻¹)]. The reaction was stopped by adding 2 mL of 70% ethanol. The precipitate was filtered using a glass-fiber filter (Whatman GF/C), rinsed with 15 mL of 70% ethanol, and dried at room temperature. The incorporated ¹⁴C was counted using a liquid scintillation counter. Protein concentrations were determined according to the method of Bradford (1976) with BSA as the standard, using the Coomassie protein assay reagent (Pierce Chemicals, United States).

Based on the Kre6 homolog sequence found both in the EST database Pleurochrysome (Yamamoto et al., 2016) and in de novo assembly sequences of RNA-seq, 5′-RACE and RT-PCR were performed, and a full-length RNA sequence was determined. The gene product was deduced to be a 64-kDa protein with a pI of 5.11. The protein was predicted to have a signal peptide of 22 amino acids but not a transit peptide to the plastid (Figure 1A). The domain search showed a conserved domain in the center of the protein [SKN1 domain, 48–461 aa; β-glucan-binding site, 245–250 aa (Figure 1B)], followed by a transmembrane region and a cytosolic tail. This structure is very similar to TGS1 and TGS2 of the diatom Phaeodactylum tricornum but different from that of yeast Kre6 (Huang et al., 2016). Thus, the Kre6 homolog in P. haptonemofera was designated as PhTGS. The C-terminal structure is not homologous compared to those of ochrophytes, but two types of sorting motifs necessary for transport to the trans-Golgi network (Hanton et al., 2005; Hwang and Robinson, 2009; Robinson and Pimpl, 2014), the di-basic motif RR and the Tyr-based motif YXXΦ, are found in PhTGS, as in P. tricornutum TGS2 and Ectocarpus siliculosus TGS (Figure 1C; Huang et al., 2016), suggesting its transport to the vacuole or the cytoplasmic membrane via the trans-Golgi network. In addition, WoLF PSORT software predicted that PhTGS is localized in the vacuole (vacuole: 7, extracellular: 3, chloroplast: 2, ER: 2). Thus, the protein is probably localized in the vacuole as in ochrophytes (Huang et al., 2016).

The molecular phylogenetic tree of GH16_2 suggests that haptophyte TGSs form a sister clade to stramenopile TGSs, and the clade consisting of haptophytes and stramenopiles TGSs is sister to the cell wall-forming GH16_2s of fungi and Chlorophyta (Figure 1D). Haptophytes maintaining only β-storage glucan metabolism might be evolved from a common ancestor of haptophytes and stramenopiles possessing both α- and β-storage glucan metabolism, as in Gyrista, consisting of Oomycota and Ochrophyta (Chabi et al., 2021).

To know whether PhTGS is actually involved in β-glucan synthesis, we performed a knockdown of PhTGS by RNAi. Three kinds of double-stranded RNA for targeting to the 5′ half, 3′ half, and around 5’ end regions of the CDS were prepared and introduced to P. haptonemofera protoplasts by the PEG method to obtain reliable results. The PhTGS mRNA level was decreased by around 30%–50%, when compared to the control (GFP dsRNA) (Figure 2A). At the same time, the cellular β-glucan amount was decreased by approximately 30%–50% (Figure 2B). Significant differences were detected between control and KD-tgs_1 and KD-tgs_3 in 2a, and KD-tgs_3 in 2b (the p-values were <0.05), although differences between the control and KD-tgs_2 in 2a and KD-tgs_1 and KD-tgs_2 in 2b were not significant (the p-values were 0.051∼0.076). These relatively high p-values may be due to the high SEs in the RNAi experiments. On the other hand, it was clearly shown that there was no significant difference among the knocked-down cells tested (the p-values among KD-tgs_1, _2, and _3 were 0.3∼0.9). These findings suggest that PhTGS plays an essential role in β-glucan synthesis.

In α-glucan-accumulating microalgae and β-glucan-containing diatoms, it is reported that an impaired storage glucan synthesis leads to an increase in the lipid content and lipid body (Siaut et al., 2011; Daboussi et al., 2014; Hildebrand et al., 2017; Chen et al., 2021). Thus, to determine whether a similar phenomenon is observed in the β-glucan containing haptophytes, we observed PhTGS knocked-down cells after BODIPY staining that stains the lipid body. Fluorescence microscopy indicated that the lipid body increased in PhTGS knocked-down cells even under N-replete conditions, although the number of lipid bodies is less compared to those in normal cells under N-starved conditions (Figures 3A, B). To confirm this observation, we performed the flow cytometry of the cells (Figures 3C, D). The fluorescence of BODIPY 493/503 and that of chlorophyll were detected using FL2 (518 nm) and FL4 (647 nm) detectors, respectively, and signals from cells that were gated by FL4 were collected on a logarithmic scale. The flow cytometry demonstrated that the amount of oil droplets in the cell increased in PhTGS knocked-down cells even under N-replete conditions, although the amount of oil droplets is less compared to those in normal cells under N-starved conditions (Figures 3C, D), similarly to fluorescent microscopy (Figures 3A, B). Significant differences were not observed among the three kinds of double-stranded RNA (Figure 3D). These findings suggest that the carbon metabolic flow is largely changed between glucans and lipids depending on their synthetic activities and C/N balance in β-glucan-accumulating haptophytes, as in α-glucan-accumulating microalgae (Siaut et al., 2011) and β-glucan-accumulating diatoms (Daboussi et al., 2014; Hildebrand et al., 2017; Chen et al., 2021).

To know if the expression pattern is related to β-glucan synthesis, we investigated the effect of light on the PhTGS mRNA level. Logarithmic-phase cells that were actively synthesizing β-glucans were used for the experiments. The amount of β-glucans steadily increased in the cultures under continuous light, while it gradually dropped when cultures were placed in the dark (Figure 4A). Meanwhile, the PhTGS mRNA level was not significantly changed in the light, while it drastically decreased in the dark (Figure 4B). Taken together, the co-occurrence of active PhTGS gene transcription and β-glucan biosynthesis in the presence of light strongly suggests that PhTGS activity plays a role in the β-glucan anabolic path of P. haptonemofera, as in the diatom P. tricornutum. In P. tricornutum, genes involved in the β-glucan biosynthesis, PtUGP2, PtBGS, and PtTGS1 have a coordinated diel expression pattern, with an increase at the beginning and decrease at the end of the light period (Skogen et al., 2013).

To understand the enzyme activity of PhTGS, we carried out the in vitro assay using the recombinant protein. The full-length CDS of PhTGS was inserted into an expression vector pET-15b, and an E. coli strain BL21 (DE3) was transformed with the plasmid. Expression of PhTGS in the transformant was induced, as described in Materials and Methods. The desired protein with a molecular mass of about 66 kDa (the His-tag including up- and downstream region, 2 kDa + PhTGS, 64 kDa) was detected as a band of approximately 70 kDa in the soluble fraction by SDS-PAGE (Figure 5A). The slightly smaller mobility on SDS-PAGE gels may be due to a positive charge of the His-tag.

The glucan synthase activity was measured in the soluble protein fraction at 20°C, using ¹⁴C-UDP-glucose as a substrate and β-glucan of P. haptonemofera as a primer. ¹⁴C-UDP-glucose was used because its incorporation into β-glucans was detectable with very high sensitivity. Specifically, the reaction was stopped by adding ethanol at appropriate times, and the ¹⁴C-glucose incorporated into the precipitate, including β-glucans, was counted in a liquid scintillation counter. The incorporation of the ¹⁴C-glucose moiety of UDP-glucose into β-glucans was significantly increased when compared with the soluble protein fraction of an empty vector-carrying E. coli (control) (Figure 5B). This finding suggests that PhTGS has an activity for the elongation of β-glucans using UDP-glucose. Furthermore, this fact infers that β-glucan-synthesizing organisms including oomycetes and ochrophytes might also have a glucan-branching mechanism utilizing TGS, which is different from that of α-glucan-containing organisms, where branching enzymes catalyze the formation of branch points in α-glucans by cleaving α-1,4-glucosidic linkages and then transfer the released oligosaccharide chains to C-6 hydroxyl groups. To confirm this hypothesis, further experiments for the clarification of the PhTGS reaction mechanism are necessary, including the determination of the substrate specificities and the product structures, as well as the detection of other lineages’ TGS activities.

In summary, a gene (pHTGS) for 1,6-β-transglucosylase involved in storage β-glucan synthesis was identified in the haptophyte P. haptonemofera. The soluble recombinant protein was demonstrated to have activity-catalyzing glucan elongation using UDP-glucose as a substrate. This soluble enzyme system could be used to produce various structures of β-glucans, including biologically active glucans (Karuppusamy et al., 2022), as well as for basic studies on molecular mechanisms and evolution.