AepG is a glucuronosyltransferase involved in acidic exopolysaccharide synthesis and contributes to environmental adaptation of Haloarcula hispanica

The attachment of a sugar to a hydrophobic lipid carrier is the first step in the biosynthesis of many glycoconjugates. In the halophilic archaeon Haloarcula hispanica, HAH_1206, renamed AepG, is a predicted glycosyltransferase belonging to the CAZy Group 2 family that shares a conserved amino acid sequence with dolichol phosphate mannose synthases. In this study, the function of AepG was investigated by genetic and biochemical approaches. We found that aepG deletion led to the disappearance of dolichol phosphate-glucuronic acid. Our biochemical assays revealed that recombinant cellulose-binding, domain-tagged AepG could catalyze the formation of dolichol phosphate-glucuronic acid in time- and dose-dependent manners. Based on the in vivo and in vitro analyses, AepG was confirmed to be a dolichol phosphate glucuronosyltransferase involved in the synthesis of the acidic exopolysaccharide produced by H. hispanica. Furthermore, lack of aepG resulted in hindered growth and cell aggregation in high salt medium, indicating that AepG is vital for the adaptation of H. hispanica to a high salt environment. In conclusion, AepG is the first dolichol phosphate glucuronosyltransferase identified in any of the three domains of life and, moreover, offers a starting point for further investigation into the diverse pathways used for extracellular polysaccharide biosynthesis in archaea.

The attachment of a sugar to a hydrophobic lipid carrier is the first step in the biosynthesis of many glycoconjugates. In the halophilic archaeon Haloarcula hispanica, HAH_1206, renamed AepG, is a predicted glycosyltransferase belonging to the CAZy Group 2 family that shares a conserved amino acid sequence with dolichol phosphate mannose synthases. In this study, the function of AepG was investigated by genetic and biochemical approaches. We found that aepG deletion led to the disappearance of dolichol phosphate-glucuronic acid. Our biochemical assays revealed that recombinant cellulosebinding, domain-tagged AepG could catalyze the formation of dolichol phosphate-glucuronic acid in time-and dosedependent manners. Based on the in vivo and in vitro analyses, AepG was confirmed to be a dolichol phosphate glucuronosyltransferase involved in the synthesis of the acidic exopolysaccharide produced by H. hispanica. Furthermore, lack of aepG resulted in hindered growth and cell aggregation in high salt medium, indicating that AepG is vital for the adaptation of H. hispanica to a high salt environment. In conclusion, AepG is the first dolichol phosphate glucuronosyltransferase identified in any of the three domains of life and, moreover, offers a starting point for further investigation into the diverse pathways used for extracellular polysaccharide biosynthesis in archaea.
Haloarcula hispanica is an extremely halophilic archaeon originally isolated from a solar saltern in Spain (20). Previously, we showed that H. hispanica can produce an acidic EPS that protects the cells against harsh environments (21). Although two glycosyltransferase-encoding genes (HAH_1662 and HAH_1667) involved in the synthesis of the acidic EPS were identified (21), the complete pathway used for the synthesis of this EPS remains to be delineated.
Analysis of the H. hispanica genome revealed that the HAH_1202-HAH_1214 cluster contains several genes encoding putative glycosyltransferases, including HAH_1206 (22). In this report, the function of the HAH_1206 was investigated. Our results revealed that the protein encoded by HAH_1206 is a dolichol phosphate (DolP) glucuronosyltransferase needed for the incorporation of glucuronic acid (GlcA) and sulfated-GlcA into the acidic EPS. We also found that HAH_1206 was required for the adaptation of H. hispanica to high salt. As the first DolP glucuronosyltransferase shown to be involved in EPS synthesis, HAH_1206 and its protein product were renamed aepG (archaeal exopolysaccharide glucuronosyltransferase) and AepG, respectively.
encoding homologs of proteins involved in N-glycosylation in Haloferax volcanii, another haloarchaeon where the pathway of N-glycosylation has been delineated (23)(24)(25)(26). This H. hispanica gene cluster also includes HAH_1206, which encodes a putative protein (HAH_1206) previously annotated as a CAZy (carbohydrate-active enzyme) Group 2 glycosyltransferase (GT-2) family protein (27,28 1C) (29,30). These similarities include the DADXQX signature motif, the acceptor loop, and the active site of family 2 inverting glycosyltransferases (31). Additionally, HAH_1206 is also similar to H. volcanii HVO_1613, a DolP glycosyltransferase not involved in N-glycosylation (32). Given all these enzymes are glycosyltransferases, catalyzing the transfer of the sugar from a nucleoside-charged diphospho-sugar donor to a phosphorylated polyprenol lipid carrier, it is likely that HAH_1206 is also a glycosyltransferase involved in the production of a lipid-linked sugar (Fig. 1C).

Deletion of H. hispanica HAH_1206
Quantitative RT-PCR analysis revealed that HAH_1206 was transcribed during log phase and that its level of expression was 78% of that of the 16S RNA gene, suggesting that HAH_1206 is expressed in H. hispanica. To investigate the function of HAH_1206, the gene was deleted as described in Experimental procedures. PCR amplification confirmed that only a 1048-bp fragment was amplified from the ΔHAH_1206 mutant instead of the 2083-bp fragment present in the parental H. hispanica strain. The deletion of HAH_1206 was further confirmed by Southern blot. When genomic DNA was digested with Ava I and incubated with an 800-bp probe, a 2186-bp fragment was detected in the ΔHAH_1206 mutant instead of a 3221-bp fragment seen in the parental strain, thus confirming HAH_1206 deletion.

HAH_1206 is not involved in N-glycosylation of the S-layer proteins
Previously, we showed that the two S-layer proteins of H. hispanica are decorated with an N-linked branched 6-sulfo-QuiNβ-(1,6)-[Glcα-(1,2)-]Gal trisaccharide and an O-linked Glc-1,4-Gal disaccharide (33). As HAH_1206 lies within the gene cluster containing an aglB homolog encoding the archaeal oligosaccharyltransferase, we first asked whether glycosylation of the S-layer proteins was affected in the ΔHAH_1206 strain. Accordingly, the S-layer proteins were extracted from H. hispanica parental or mutant cells cultivated in AS-168 medium at 37 C, separated by SDS-PAGE, and subjected to silver and Periodic Acid-Schiff's (PAS) staining. As shown in Figure 2, although silver staining revealed some proteins unique to either the parent or mutant strain or differing in amount between the two strains, the S-layer proteins from the ΔHAH_1206 mutant were found at similar levels in both strains and migrated identically on the SDS-PAGE gel ( Fig. 2A). It should be noted that SDS-PAGE does not separate the two S-layer proteins well (33). Furthermore, glycosylation of the S-layer proteins was not affected in the ΔHAH_1206 mutant, as reflected by the similar PAS staining in the parental and mutant strains (Fig. 2B).
In a second experiment, glycans were released from the S-layer glycoproteins by the One-Pot Release and Labeling (OPRAL) method (34). The released N-and O-glycans were separated by RP-HPLC using an Agilent HC-C18 column (250 × 4.6 mm, 5 μm) and analyzed by online RP-HPLC-UV-ESI-MS/MS. As shown in Figure 3, the glycans at m/z 898 and m/z 673 were identified as a di-1-phenyl-3-methyl-5pyrazolone (PMP)-modified N-linked trisaccharide and an O-linked disaccharide by MS/MS, respectively. Both glycans also appeared in samples prepared from the ΔHAH_1206 strain, indicating that HAH_1206 does not participate in N-or O-glycosylation of the H. hispanica S-layer glycoproteins.
HAH_1206 contributes to the appearance of DolP-linked hexuronic acid and sulfated hexuronic acid Since glycosylation of the H. hispanica S-layer glycoproteins was not affected in ΔHAH_1206 cells (Fig. 3), it was not surprising to detect in the deletion strain DolP-hexose, DolPdihexose, and DolP-QuiNS, all involved in S-layer protein glycosylation, in the deletion strain (Fig. S1). In contrast, analysis of the membrane fraction extracted from ΔHAH_1206 cells by normal-phase LC-MS in the negative ion mode revealed that neither DolP-HexA nor DolP-HexA-SO 4 ( Fig. 4) was detected in the mutant, despite being present in the parent strain. This suggests HAH_1206 is involved in the appearance of these DolP-bound sugars in H. hispanica.
H. hispanica produces an acidic EPS consisting of the neutral sugars glucose (Glc), galactose (Gal), and mannose (Man) and an unidentified acidic component (21). EPSs extracted from the parental strain consists of GlcA, Man, Gal, and Glc at a 4:3.6:1.5:1 M ratio. However, analysis of the EPS from the ΔHAH_1206 mutant showed only trace amounts of GlcA could be detected (35). To further assess whether HAH_1206 contributes to acidic EPS synthesis, DolP-linked sugars in the parental and mutant strains were extracted from the cell membrane, released with hydrofluoric acid, and analyzed by HPAEC-pulsed amperometric detection. These experiments detected GlcA in the membrane extract of the parental strain but not in the extract from the mutant (Fig. S2). Together, these results suggest that HAH_1206 is responsible for the synthesis of DolP-GlcA, the likely donor of acidic sugar residues in the acidic EPS of H. hispanica. Accordingly, HAH_1206 was renamed aepG.

In vivo and in vitro demonstration of AepG activity
To further address the role of AepG in DolP-GlcA synthesis, the deletion strain was complemented to express cellulosebinding domain (CBD)-tagged AepG (Fig. 5). Such complementation resulted in the appearance of both DolP-HexA and DolP-HexA(SO 3 ) (Fig. 4).
The CBD-AepG expressed in the complemented strain was next used in an in vitro assay to test for glucuronosyltransferase activity. As C 60 -DolP is not commercially available, the cell pellet from the mutant containing this lipid was used as substrate. As shown in Figure 6, the addition of CBD-AepG to the reaction mixture led to both dose-and time-dependent increases in the amount of DolP-HexA. No DolP-HexA was generated when the assay was performed in the absence of CBD-AepG.
Together, these in vivo and in vitro results confirm that AepG is a DolP glucuronosyltransferase responsible for the synthesis of DolP-GlcA in H. hispanica.

Phenotypes of the ΔaepG mutant
To evaluate the significance of the aepG gene, the mutant strain was grown under high salt conditions. On medium containing 3.4 M NaCl (normal growth conditions) or 4.7 M NaCl, the mutant showed significantly reduced growth (Fig. 7A). Transmission electron microscopy revealed the parental cells to be irregularly shaped and dispersed as single cells in the high salt medium. In contrast, ΔaepG cells formed aggregates (Fig. 7B). These observations suggest that AepG is required for proper growth in high salts. We showed that deletion of aepG resulted in the loss of DolP-HexA and DolP-HexA-SO 4 (Fig. 5), the sugar donors for the negatively charged sugars of the acidic EPS. Indeed, only trace amounts of GlcA could be detected in the EPS from the ΔHAH_1206 mutant (35). Therefore, it is reasonable to conclude that AepG is required for the synthesis of the negatively charged sugars of the acidic EPS, which is required for the separation and dispersal of H. hispanica cells in hypersaline surroundings.

Discussion
Among microbial EPSs, bacterial and fungal EPSs have been most extensively studied (1,4). Most bacterial and fungal EPSs are hetero-polysaccharides containing three or four different monosaccharides arranged in groups of 10 or less to form repeating units. Although some EPSs are neutral macromolecules, the majority are polyanionic due to the presence of uronic acids or ketal-linked pyruvate or inorganic residues. In bacteria, three pathways of EPS production have been reported, namely the Wzx/Wzy flippase-dependent pathway, the  ABC transporter-dependent pathway, and the synthasedependent pathway (9,36,37). None of the enzymes used to synthesize anionic uronic acid residues in EPSs have been identified to date.
In recent years, the study of extremophiles has gained increasing attention, not only from an evolutionary point of view but also for exploring the EPSs that allow these microorganisms to survive in extreme environments (19). Various haloarchaea produce EPSs (14)(15)(16)(17)(18)(19), yet little is known about archaeal EPS biosynthesis. H. hispanica can produce an acidic EPS consisting of GlcA, Man, Gal, and Glc at a 4:3.6:1.5:1 M ratio (21,35), by a biosynthetic pathway that remains unknown.
The H. hispanica HAH_1202-HAH_1214 gene cluster was predicted to encode N-glycosylation-related proteins based on homologies to components of the well-defined H. volcanii N-glycosylation pathway (22,33). Specifically, HAH_1214 was predicted to encode a homolog of H. volcanii AglJ that adds the first sugar of an N-linked pentasaccharide to a DolP carrier (32), HAH_1202 is thought to encode the oligosaccharyltransferase AglB (38), and HAH_1203 was predicted to encode a homolog of the H. volcanii hexuronic acid transferase AglG that adds GlcA to glucose-charged DolP (24). The aepG (HAH_1206) gene was predicted to encode a Group 2 family glycosyltransferase, specifically, a homolog of DolP mannose synthase (39). Based on these observations, we initially thought that aepG (HAH_1206) might be involved in H. hispanica protein glycosylation. However, analysis of glycans decorating the S-layer proteins of H. hispanica ΔaepG cells revealed that protein glycosylation was not affected in the mutant.
We have shown that GlcA residues were not detected in the EPS of the ΔaepG mutant (35), which also lacked DolP-linked GlcA. Using recombinant CBD-AepG and the cell lysate of the mutant, we determined AepG activity in vitro. These results, together with complementation studies, confirmed that AepG is a DolP glucuronyltransferase responsible for the synthesis of DolP-GlcA, the donor of acidic sugar residues in the H. hispanica EPS. Finally, we showed that deletion of aepG led to retarded growth and aggregation of H. hispanica cells. As AepG is responsible for the production of DolP-GlcA, presumably a donor of negatively charged sugars in the acidic EPS, it is reasonable to conclude that separation and dispersal of H. hispanica cells in a high salt environment involve the negatively charged GlcA and GlcA-SO 3 residues of the EPS. Therefore, our results demonstrate that AepG is essential for the adaptation of H. hispanica to hypersaline surroundings. Additionally, we recently showed that the EPS from H. hispanica exhibited antiviral activity against SARS-CoV-2 but not the EPS from the ΔaepG mutant, suggesting an important role for the acidic sugar residues of the EPS in this antiviral activity (35). In summary, we report the first DolP glucuronyltransferase in archaea and also the first DolPmodifying glycosyltransferase contributing to EPS biosynthesis. Our results expand the diversity of pathways of EPS biosynthesis and may have a biotechnological potential for the biosynthesis of functional acidic polysaccharides.

Growth conditions
Strains and plasmids used in this study are listed in Table 1.

Construction of the aepG deletion mutant
The aepG deletion mutant was created by homologous recombination, as previously described (40). One 532-bp Figure 6. In vitro AepG activity assay. AepG activity was assayed as described in Experimental procedures. In the assay, the reaction was initiated upon addition of 1 unit of CBD-tagged AepG into 1 volume of substrate in 5 ml of cell lysis buffer containing DNase I. After a reaction at 37 C for 12 h, lipid-linked sugars were extracted by methanol and chloroform. Sugars were released from the lipids by HF and analyzed with HPAEC-PAD. A, effects of substrates or CBD-AepG on activity. Various amounts of substrates or purified CBD-AepG (insert) were added to the reaction mixture. After incubation at 37 C for 12 h, the amount of DolP-linked GlcA was determined. B, effect of reaction time on activity. Three units of CBD-AepG were added into three volumes of substrates. After reactions at 37 C for 3, 6, and 12 h, the amounts of DolP-linked GlcA were determined. aepG, archaeal exopolysaccharide glucuronosyltransferase; CBD, cellulose-binding domain; DolP, dolichol phosphate; GlcA, glucuronic acid; HF, hydrofluoric acid; PAD, pulsed amperometric detection.  AepG is a dolichol phosphate glucuronosyltransferase fragment containing the upstream flanking sequence of the aepG gene was amplified by PCR using the primers HAH_1206GLF and HAH_1206GLR (Table 2). A 516-bp DNA fragment containing the downstream flanking region of aepG was amplified by PCR using the primers HAH_1206GRF and HAH_1206GRR (Table 2). These two PCR products were sequenced and cloned into plasmid pHAR to generate plasmid pHARΔHAH_1206 ( Table 1). The resulting plasmid was then transformed into the H. hispanica parental strain to delete the aepG gene by double-crossover homologous recombination. Deletion of aepG was confirmed by PCR using the flanking sequence primers HAH_1206GLF and HAH_1206GRR. Southern blot analysis of the mutant was performed using an 800-bp hybridization probe. Probe labeling and color detection with NBT/BCIP were performed using a DIG-High Prime DNA Labeling and Detection Starter Kit I (Roche).

Complementation of the ΔaepG mutant
For complementation of the mutant, plasmid pWL-CBD-SecE containing an archaeal constitutive promoter PrP16 and the CBD from Clostridium thermocellum was employed (41). The aepG gene was amplified from genomic DNA using primers HAH_1206REF/HAH_1206RER (Table 2), in which Nde I and Kpn I restriction sites were introduced at the 5 0 -and 3 0 -ends, respectively. The secE gene in pWL-CBD-SecE was then replaced by the aepG gene to generate plasmid pWL-CBD-HAH_1206. ΔaepG cells were transformed with plasmid pWL-CBD-HAH_1206 using the PEG precipitation method (42). Transformants were confirmed by PCR analysis using primers HAH_1206GCF/HAH_1206GCR (Table 2) and DNA sequencing. Expression of CBD-tagged AepG was confirmed by Western blot.

Western blot
Capture and immunoblot of CBD-tagged protein were performed as described previously with slight modifications (42,43). Briefly, 2 ml aliquots of the cells were harvested (3000 rpm in a microfuge for 10 min). The cells were resuspended in 1 ml of solubilization buffer (1% Triton X-100, 1.8 M NaCl, 50 mM Tris-HCl, pH 7.2) containing 10 μg/ml DNase I and incubated at 37 C for 2 h. After solubilization, 50 μl of 10% (w/v) cellulose were added and the mixture was incubated at room temperature for 60 min. Then, the cellulose pellets were collected by centrifugation at 12,000 rpm for 10 min, washed twice with washing buffer (2 M NaCl, 50 mM Tris-HCl, pH 7.2), separated on 7.5% SDS-PAGE, and transferred to a nitrocellulose membrane. The CBD-tagged proteins were detected with polyclonal rabbit anti-CBD antibodies (1:2000 dilution). Alkaline phosphatase-conjugated anti-rabbit antibodies (Bio-Rad) were used as secondary antibodies (1:5000 dilution). Antibody binding was detected using BCIP/ NBT Western blotting detection reagent (Pierce).

Isolation of S-layer proteins
H. hispanica cells cultivated in 100 ml of AS-168 medium at 37 C were collected and resuspended in a basal salt solution containing the same ionic composition as AS-168 medium. After sonication, the supernatant was collected by centrifugation at 7000g for 20 min and further ultracentrifuged at 250,000g at 4 C for 1 h. The pelleted membrane proteins were dissolved in deionized water and quantified by Bradford assay (44). Membrane proteins (36 μg) from the parental or mutant were separated by 7.5% SDS-PAGE and visualized by silver staining (Sangon Biotech) or PAS reagent (Thermo Fisher Scientific).

Release and labeling of glycans from the S-layer proteins
H. hispanica cells cultivated in 100 ml of AS-168 medium at 37 C were collected by centrifugation at 7000g for 20 min and resuspended in 10 ml washing buffer. The S-layer proteins were extracted by the addition of 1 ml 0.5 M EDTA (pH 8.0). After incubation at 37 C for 3 h, the mixture was centrifuged at 7000g for 20 min. The supernatant was collected, precipitated with 15% trichloroacetic acid on ice for 30 min or 4 C overnight. The precipitated S-layer proteins were collected by centrifugation, washed with ice-cold acetone, and air-dried for further use.
Glycans on the S-layer proteins were released and labeled by the OPRAL method with slight modifications (34). Briefly, 0.35 g PMP were dissolved in 0.5 ml of a 1:1 (v/v) mixture of methanol and aqueous ammonium hydroxide (26%-28%, v/v) in a 1.5 ml centrifuge tube by heating and shaking in a water bath at 80 C. Once the PMP had dissolved, a 1:1 (v/v) mixture of methanol and aqueous ammonium hydroxide was added to bring the final volume to 1 ml, followed by rapid mixing with a multivortex mixer. Subsequently, lyophilized S-layer proteins were dissolved in 1 ml of the reaction solution in a 1.5 ml screw-cap tube, followed by incubation at 80 C for 16 h. When the reactions were finished, the reaction mixture was neutralized by adding glacial acetic acid drop-wise with shaking.
Glycans obtained using the OPRAL method were transferred into a new centrifuge tube, and a three-fold volume of dichloromethane was added, followed by violent shaking and centrifugation at 12,000×rpm for 10 min. This step was repeated until no intermediate layer appeared. The aqueous layer was transferred into another 2 ml centrifuge tube and concentrated to complete dryness using a SpeedVac concentrator. Finally, the dried glycans were redissolved in 1 ml of AepG is a dolichol phosphate glucuronosyltransferase deionized water and loaded onto a Sep-Pak C18 cartridge prewashed with 25 ml of 25% acetonitrile (ACN) containing 0.1% TFA and then 5 ml 0.1% TFA. When the sample was fully absorbed, the column was washed with 3 to 5 ml 0.1% TFA to remove hydrophilic impurities, and the target glycans were eluted using 1 ml of 25% aqueous ACN containing 0.1% TFA solution. The elution fraction was concentrated and stored at −20 C for further use.

Glycan analysis by HPLC-MS/MS
The PMP-labeled glycans were dissolved in ddH 2 O. HPLC-MS/MS analysis was conducted on an Agilent 1290 HPLC System using an Agilent HC-C18 column (250 × 4.6 mm, 5 μm) with a flow rate of 1.0 ml/min. MS data were collected in the auto-MS/MS mode. The solvent system used was a linear gradient of 25% to 29% solvent B (ACN) in solvent A (10 mM aqueous ammonium acetate, titrated to pH 5.5 with glacial acetic acid) over a period of 20 min. The column temperature was set at 40 C. The sample injection volume was 100 μl. The column was cleaned using 100% ACN for 5 min. Data acquisition was performed using Agilent MassHunter Workstation Software-Qualitative Analysis.

Analysis of total lipid extracts
Total lipid extracts from H. hispanica parental, ΔaepG mutant, and complemented strains were analyzed by normalphase LC-ESI MS as described previously (45). MS spectra of the major species were obtained in the negative ion mode.

In vitro assay of AepG activity
The mutant cells were grown in 100 ml AS-168 medium and harvested by centrifugation when the A 600 reached 1.0. The cells were resuspended in 5 ml of cell lysis buffer (5 mM MgCl 2 , 0.13% NP-40, 50 mM Tris-HCl, pH 8.0) containing DNase I and incubated at 37 C with shaking at 200 rpm for 2 h to release the cell lysate, which was considered as one volume of substrates for use in assaying AepG activity.
The complemented strain was grown in 100 ml AS-168 medium to an A 600 of 1.0. The cells were harvested by centrifugation, resuspended in 10 ml of solubilization buffer (1% Triton X-100, 1.8 M NaCl, 50 mM Tris-HCl, pH 7.2,), and incubated at room temperature for 2 h, followed by the addition of 0.15 g cellulose. After 60 min incubation at room temperature, the cellulose pellets were collected by centrifugation at 12,000 rpm for 10 min, washed twice with washing buffer (2 M NaCl, 50 mM Tris-HCl, pH 7.2), and defined as 1 unit of CBD-AepG for use in assaying AepG activity.
The assay was initiated by adding 1 unit of CBD-AepG into 1 volume of substrates (in 5 ml cell lysis buffer containing DNase I). After a reaction at 37 C for 12 h, the mixture was centrifuged (8000g for 10 min). Total lipids in the supernatant were extracted with H 2 O:MeOH:CH 3 Cl (3.8:1:1), air-dried, and then hydrolyzed with 250 μl hydrogen fluoride on ice for 3 h. The released lipid-linked monosaccharides or glycans were dried with a hair dryer and dissolved in H 2 O:MeOH:CH 3 Cl (3.8:1:1). The water phase was collected, vacuum-dried, dissolved in water, and subjected to HPAEC-PDA analysis.

Analysis of monosaccharides by HPAEC-pulsed amperometric detection
Standard and samples were separated by HPAEC and detected by pulsed amperometric detection (Dionex) using a CarboPack PA10 (2 × 250 mm) column (LCPackings). Acidic GlcA and GalA were eluted by 10 mM NaOH and 10 mM CH 3 COONH 4 with 1 ml/min flow rate at a retention time of 25 min.
Transmission electron microscopy H. hispanica parental and mutant cells in 3.4 M NaClcontaining buffer were fixed in 2.5% glutaraldehyde in 0.1 M phosphate buffer, pH 7.0, for 4 h or overnight at 4 C, washed three times in 0.1 M phosphate, postfixed in 1% osmium tetroxide in 0.1 M phosphate for 2 to 4 h, incubated in 30%, 50%, 70%, 85%, 95%, and 100% methanol for 15 to 20 min at each concentration, and post-fixed in 2% uranyl acetate-30% methanol. The cells were rinsed, dehydrated, and embedded in Epson 812 for the floating sheet method. Sections were examined with an H-600 electron microscope (Hitachi).

Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.
Supporting information-This article contains supporting information.