A novel autopolysialylation activity of the ganglioside sialyltransferase ST8Sia5 regulates its secretion and enzyme activity

Abstract

Polysialic acid (polySia) is a linear polymer of sialic acid, which usually modifies N-glycans on the neural cell adhesion molecule (NCAM) mostly in the brain and is involved in the development of brain. PolySia is also associated with several diseases, including mental disorders and cancers. ST8Sia2 and ST8Sia4 are believed to be the only polysialyltransferases that synthesize polySia on NCAM (NCAM polysialylation). These enzymes are also autopolysialylated. In this study, we first found that ST8Sia5L, a ganglioside-specific sialyltransferase, has an activity to synthesize polysialic acid on ST8Sia5 itself, but does not exhibit the NCAM polysialylation activity. Notably, in silico and biochemical analyses revealed that ST8Sia5L contains a new polysialic acid-trapping motif (PSTM) that is essential for polySia elongation, instead of the conventional polysialyltransferase domain (PSTD) found in ST8Sia2 and ST8Sia4. We also found that autopolysialylated ST8Sia5L is secreted from the cells. To identify the autopolysialylation sites involved in secretion, we performed N-glycosylation site disruption experiments, and found that N92 and N277 in the five N-glycosylation sites are important for this phenomenon. Furthermore, the inhibitor experiments showed that certain metalloprotease(s), but not exosomal pathways, are involved in the secretion. Notably, the secreted autopolysialylated enzyme showed no ganglioside-sialylation activity; however, the activity was recovered when polySia was removed by sialidase treatment. Overall, we show that autopolysialylation of ST8Sia5L regulates both its secretion and the conventional sialyltransferase activity.

PolySia is a linear polymer of sialic acid that has been studied since its discovery in the bacterial glycocalyx in 1957 (1). The same epitope is also present in the brain of vertebrates (2) and in some cancer cells, and is recognized as an oncodevelopmental antigen (3, 4). PolySia is mainly expressed in the brain in a spatiotemporal manner. It is highly enriched in the embryonic brain in vertebrates. Although its expression persists even in the adult brain, it is restricted in the areas of neural regeneration or reorganization. Therefore, polySia is recognized as a marker of plasticity. Based on the abnormal phenotypes of polySia-impaired mice, polySia has been implicated in the regulation of long-term potentiation and long-term depression (5), learning and memory (6), social behavior (7), and the regulation of the clock (8).

Two special features are noted for the mechanism underlying the functioning of polySia (3, 9), namely, the anti-adhesive effect (10) and molecule-retention properties (3, 9). For the anti-adhesive effect, polySia physicochemically regulates the strength of cell–cell and cell–extracellular matrix interactions, owing to its large exclusion volume, and subsequently, signal transduction mediated by certain receptors near polySia is downregulated (10). For the molecule-retention property, polySia binds directly to some physiologically important molecules, such as neurotrophic factors, growth factors, and neurotransmitters, depending on its degree of polymerization (DP) (3, 9). Such unique-featured polySia is highly regulated by genetic factors that are related to polysialyltransferase genes (11, 12) or associated with mental disorders (13) and by environmental factors, such as stress (14) and medicines (14) and alterations in its expression are implicated in various diseases (3, 4). For example, impaired structures of polySia are reported in patients with schizophrenia and changes in blood polySia levels are also related to schizophrenia (15, 16, 17), although the responsible enzyme or the mechanism in the serum polySia expression is not well analyzed.

To date, six α2,8-sialyltransferases are known to synthesize the α2,8-linked oligomerized Sia structure (9). Members of the ST8Sia family are type II membrane sialyltransferases with a conserved luminal catalytic domain containing characteristic sialyl motifs. Although detailed structural information remains limited for the ST8Sia family, the crystal structure of human ST8SIA3 has revealed an electropositive surface around the catalytic region that provides a useful framework for considering how ST8Sia enzymes accommodate negatively charged oligo/polySia chains (18). ST8Sia1 and ST8Sia5 can synthesize the diSia or triSia structure on ganglioside. ST8Sia3 can synthesize the oligoSia and short polySia structure on the glycoconjugates. ST8Sia2 and ST8Sia4 are polysialyltransferases that are believed to be the key enzymes for polySia synthesis on glycoproteins. PolySia is present on N-glycans within the IgV domain of the neural cell adhesion molecule (NCAM), the major carrier protein in the brain (19). ST8Sia2 and ST8Sia4 are also involved in autopolysialylation (20). Polysialylation of the enzymes themselves has been clearly demonstrated (20, 21) and the autopolysialylation is not required for polysialylation of NCAM. ST8Sia3 has been reported to exhibit potential for autopolysialylation (22). The role of autopolysialylation has remained unclear, as yet.

In this study, we discovered a novel polySia synthetic activity of ST8Sia5, which is believed to be a ganglioside synthase (GT1a/GQ1b synthase) (23, 24). Recently, we found that the difference in localization of the ST8Sia 5 isoforms (S, M, and L) leads to the difference in activity (25). The only difference among the isoforms is in the length of their stem region. ST8Sia5S and ST8Sia5M localize to the ER and almost no production of the b-series gangliosides on cell surface is observed. On the contrary, ST8Sia5L localizes to the Golgi where b-series gangliosides are detected. The localization of ST8Sia5L to the Golgi depends on the stem region (25); however, its functions remain unclear. ST8Sia5L appeared to undergo autopolysialylation in this study but not polysialylate NCAM. In addition, the secreted autopolysialylated ST8Sia5 was shown to synthesize diSia on GT1b after de-polysialylation.

Results

To analyze the effects of α2,8-sialyltransferases on the malignant phenotypes of melanoma cells (25, 26), we transfected the coding sequence of each enzyme into B16 melanoma cells, which are incapable of synthesizing α2,8-linked oligo/polysialic acid because of the lack of α2,8-sialyltransferase (Fig. S1) (25, 26) and measured the amounts of polySia using the recently developed ELISA methods (15). Significant amounts of polySia were detected in the cells after transfection with known polysialyltransferase-coding plasmids (3, 9, 19), pcDNA-ST8Sia2 and pcDNA-ST8Sia4 (Fig. 1A). After measuring the expression of polySia in the cell lysates, we further analyzed the levels secreted from the cells using the same method. Notably, we detected polySia in the culture supernatant of ST8Sia5L-transfected cells (Fig. 1B, ST8Sia5L), although polySia expression was not observed in the cell lysates (Fig. 1A, ST8Sia5L). We did not observe significant polySia expression in the lysates and culture supernatants of ST8Sia5S- and ST8Sia5M-transfected cells (Fig. 1), which indicates that among the ST8Sia5 isoforms (Fig. S2A), only ST8Sia5L showed the polySia-synthesizing activity. This is consistent with the previous results (27) because ST8Sia5S and M localize to the ER, and L localizes to the Golgi.

Figure 1.

Expression of polysialic acid in the α2,8-sialyltransferase-expressing transfected melanoma cells and their culture supernatants.A, detection of polysialic acid in the cell lysates. The cells were transfected with the plasmids coding for α2,8-sialyltransferases, ST8Sia1 through ST8Sia6, and cell homogenates were analyzed for polysialic acid expression as described in “Experimental procedure.” B, detection of polysialic acid in the culture supernatant of the cells. The cell culture medium derived in A was used for the analysis of polysialic acid. Data are shown as mean ± SD from three technical replicates. Individual data points represent technical replicates. Statistical significance was evaluated using an unpaired two-tailed Student’s t test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

As ST8Sia5 is believed to be a GQ1b-ganglioside synthesizing enzyme and because there has been no report on the synthesis of polySia structures by ganglioside-synthesizing enzymes, we attempted to confirm polySia detection more specifically. We used an endo-N-acylneuraminidase (EndoN, endo-type sialidase) that hydrolyzes polySia to triSia or diSia for the conventional confirmation of the presence of polySia (27), although we established a new method to detect and quantify polySia precisely using an inactivated anti-polySia antibody as a control, which was shown to be the best method for detection (3). All the polySia detection in the ST8Sia2-, ST8Sia4-, and ST8Sia5L-transfected cells, which was confirmed based on the subtraction of absorbance for the anti-polySia active and anti-polySia inactive antibodies, completely disappeared after EndoN treatment (Fig. 2). We also analyzed polySia bands using western blotting and found EndoN-sensitive smear bands in the ST8Sia5 lane (100–140 kDa), similar to those obtained in the case of ST8Sia2-and ST8Sia-4-transfected cells, although the molecular mass of the bands at around 180 kDa or higher derived from ST8Sia5 did not completely match those from ST8Sia2 and ST8Sia4 (Fig. 3A). The band derived from ST8Sia5L-transfected cells disappeared after peptide-N-glycanase F (PNGaseF) treatment, indicating that the polySia structure was present on the N-linked glycans (Fig. 3B).

Figure 2.

Expression of polysialic acid in the ST8Sia2-, 4-, and 5L-transfected cells. Detection of polysialic acid in the culture supernatant of the cells expressing ST8Sia2, 4, and ST8Sia5L was performed as described in “Experimental procedure.” The culture supernatant with (+) or without (−) endo-N-acylneuraminidase (EndoN) treatment was analyzed for polysialic acid expression because EndoN cleaves polySia specifically to oligosialic acid that cannot be recognized by the anti-polySia antibody. Data are shown as mean ± SD from three technical replicates. Individual data points represent technical replicates. Statistical significance was evaluated using an unpaired two-tailed Student’s t test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Figure 3.

Western blotting for polysialic acid in the ST8Sia2-, 4-, and 5L-transfected cells.A, confirmation of the presence of polysialic acid in the culture supernatant derived from the cells expressing ST8Sia2, 4, and ST8Sia5L using western blotting. The culture supernatant with (+) or without (−) endo-N-acylneuraminidase (EndoN) treatment was analyzed for polysialic acid expression using an anti-polySia antibody based on the specificity of the enzyme that can specifically cleave polySia. B, confirmation of the presence of polysialic acid on N-linked glycan(s). The culture supernatant with (+) or without (−) PNGaseF treatment was analyzed using western blotting with anti-polySia antibody. The specific band is indicated with a molecular mass.

To identify the polySia carrier protein, we performed immunoprecipitation analysis. ST8Sia2 and ST8Sia4 are well-known enzymes that synthesize polySia not only on its specific substrate, NCAM, but also on themselves via autopolysialylation (20, 21). Therefore, we suspected that the polySia carrier protein might be ST8Sia5 itself because the lower part of the polySia-stained smear for ST8Sia2 and ST8Sia4 was the same as that for ST8Sia5L (Fig. 3A). To confirm that ST8Sia5 was autopolysialylated, we immunoprecipitated the polySia-containing glycoprotein with an anti-polySia or anti-His antibody, considering the presence of the His-tag in ST8Sia5L. We also used Mock-transfected cells as control. The immunoprecipitated glycoproteins were immunoblotted and analyzed using both anti-polySia (Fig. 4A) and anti-His (Fig. 4B) antibodies. We observed that the immunoprecipitated polySia was immunostained with anti-polySia (Fig. 4A, IP: anti-polySia, indicated by brace) and anti-His (Fig. 4B, IP: anti-polySia, indicated by brace) antibodies. We then immunoprecipitated the glycoproteins using anti-His and anti-polySia antibodies. The immunoprecipitate with the anti-His antibody was stained with both anti-His (Fig. 4B, IP: anti-His, indicated by brace) and anti-polySia (Fig. 4A, IP: anti-His, indicated by brace) antibodies. These results clearly showed that the immunoprecipitated molecule was polysialylated ST8Sia5L and that ST8Sia5L is the polySia-carrier protein secreted in the culture medium. In addition, the lanes containing unbound proteins immunoprecipitated with the anti-polySia and anti-His tag antibodies showed no staining (Fig. 4, A and B, lane u), indicating that ST8Sia5 is the major polySia carrier protein. We also analyzed the possibility of NCAM polysialylation, because NCAM is a well-known polySia carrier protein synthesized by ST8Sia2 and ST8Sia4. We performed an immunoprecipitation analysis; however, NCAM was not found to be a polySia carrier protein in ST8Sia5-transfected B16 cells (Fig. S3). Based on these results, we further reinforced our findings by performing proteomic analysis. The fractions immunoprecipitated with the anti-polySia and control antibodies were subjected to LC-MS-based proteomic analysis. Table 1 shows the abundance values for proteins detected in each immunoprecipitate. Only proteins with abundance values greater than or equal to the mean abundance in the anti-polySia immunoprecipitate were extracted, and the top-ranked proteins are shown in descending order according to the abundance ratio (anti-polySia antibody/control antibody). In the culture supernatant of ST8Sia5-transfected cells, ST8Sia5 was the protein that was most specifically detected in both the polySia antibody-immunoprecipitated fraction and the ST8Sia5 sample (Table 1). This finding was consistent with the results of Western blot validation of the immunoprecipitated products and clearly demonstrated the autopolysialylation of ST8Sia5L.

Figure 4.

Identification of polysialic acid carrier molecules using immunoprecipitation with specific antibodies. The culture supernatant of mock- or ST8Sia5L-coding plasmid-transfected cells was immunoprecipitated with anti-polySia (anti-polySia) or anti-His tag (anti-His tag) antibody. The control antibodies were also used (Control Ab). The immunopreicipates were analyzed via SDS-PAGE followed by western blotting. The membrane was visualized after immunoblotting with anti-polySia (A) or anti-His tag (B) antibody. i, u, and b indicate the input, unbound, and bound fractions, respectively.

Table 1.

The abundance values of proteins detected in immunoprecipitants

Abundance immunoprecipitated from polySia antibody	Abundance immunoprecipitated from control antibody	Abundance ratio (polySia antibody/control antibody)	Description	Gene Symbol	Accession
207812573.3	246373.9531	843.48	Alpha-2;8-sialyltransferase 8E	St8sia5	P70126
152386968.2	9881177.25	15.42	Threonine-tRNA ligase 1;cytoplasmic	Tars1	Q9D0R2
116293648.3	27346379.94	4.25	Growth.differentiation factor 15	Gdf15	Q9Z0J7
77509644.28	21006070.59	3.69	A disintegrin and metalloproteinase with thrombospondin motifs 1	Adamts1	P97857
76124411.28	27381556.63	2.78	Glutathione S-transferase P 1	Gstp1	P19157
58423501.38	25934975.13	2.25	Glutathione S-transferase P 2	Gstp2	P46425
36409383.44	16308023.42	2.23	Tubulointerstitial nephritis antigen-like	Tinagl1	Q99JR5
92644068.7	42939657.73	2.16	60S ribosomal protein L19	R9l19	P84099
44130956.02	21005745.55	2.10	60S ribosomal protein L26	Rpl26	P61255
100891160.1	48185022.1	2.09	Carbonic anhydrase 6	Ca6	P18761
81016611.75	39067745.11	2.07	40S ribosomal protein S6	RPs6	P62754
85081814.44	42963635.06	1.98	60S ribosomal protein L14	R9l14	Q9CR57
315245661.6	160310132.3	1.97	60S ribosomal protein L4	Rp14	Q9D8E6
80980623.19	41303018.8	1.96	60S ribosomal protein L27a	Rpl27a	P14115
170452144.8	87685043.56	1.94	60S ribosomal protein L6	R9l6	P47911

We discovered that ST8Sia5 exhibited autopolysialylation activity and attempted to identify the autopolysialylation position on the enzyme. To confirm the autopolysialylation sites in ST8Sia5, we compared the sequences of ST8Sia 2, 4, and 5L (Fig. 5). The N-glycosylation sites of ST8Sia2 and ST8Sia4 were 2-1 through 2-6 and 4-1 through 4-5, respectively. ST8Sia5L was found to have five potential N-glycosylation sites (5–1 through 5–5) (Fig. 5). Next, we prepared mutant ST8Sia5 enzymes in which the potential N-glycosylation sites (N-X-T/S) were mutated from N to Q, because polysialylation was found to occur on N-linked glycans (Fig. 3B). We analyzed the expression of the ST8Sia5 enzyme and polySia in cell lysates and culture supernatants (Fig. 6). As for the enzyme analysis using anti-His antibody, we digested the N-glycans with PNGaseF. In the case of polySia-modified glycoprotein, we did not observe the band clearly using the anti-protein antibody because polySia glycan sometimes shields the epitope of the anti-protein antibody. As shown in Figure 6A, the expression of all mutant enzymes (48 kDa), including the wild type (51 kDa), was observed in cell lysates (glycosylated form: 51 or 48 kDa, non-N-glycosylated form: 39 kDa). There was no change in the band intensity before and after PNGaseF treatment, indicating that large amount of polySia-modified enzyme is not present in the cell. The levels of polysialylated enzymes, observed as smear (100–140 kDa), in the cell culture medium were dramatically reduced in mutant 5-4 and 5-5 (Fig. 6B), although the expression of these enzymes in lysates was almost the same (Fig. 6A). We also analyzed the secreted enzyme and polySia expression in the culture supernatant and found that polySia and enzyme-staining were faint for 5-4 and 5-5 (Fig. 6, B and C). Non-polysialylated enzyme was not present in the culture medium because enzyme band (37 kDa) was only observed after PNGaseF treatment (Fig. 6C). In addition, N92 (5–4) and N277 (5-5) sites were clearly polysialylated (Fig. 6B) and the polysialylated enzymes were secreted from the WT and 5-1 through 5-3 cells because no enzyme band was observed before the PNGaseF treatment.

Figure 5.

Comparison of the amino acid sequences of ST8Sia5L, ST8Sia2, and ST8Sia4. Amino acid sequences of the three enzymes were aligned. L, S, III, and VS stand for the sialyl motifs long (L), short (S), III, and very short (VS), respectively. PBR and PSTD stand for the polybasic region and polysialyltransferase domain, respectively. The potential N-glycosylation sites for ST8Sia5L (5–1(N58), 5-2 (N64), 5-3(N73), 5-4 (N92), and 5-5 (N277)), ST8Sia2 (2–1(N60), 2-2 (N72), 2-3 (N89), 2-4 (N134), 2-5 (N219), and 2-6 (N234)), and ST8Sia4 (4–1(N50), 4-2 (N74), 4-3 (N119), 4-4 (N204), and 4-5 (N219)) are indicated.

Figure 6.

Effects of glutamine point-mutations in the N-glycosylation site of ST8Sia5L on the enzyme andpolySia expression.A, expression of ST8Sia5L wild type and its five mutants (5–1 (N58Q), 5-2 (N64Q), 5-3 (N73Q), 5-4 (N92Q), and 5-5 (N277Q)) in the homogenate of each transiently transfected cell line was analyzed via western blotting using an anti-His-tag antibody. For quantification, PNGaseF treatment was performed (PNGaseF+). Enzymes derived from WT or mutants (PNGaseF-) are indicated in the figure with molecular mass. The enzymes without N-glycans (PNGaseF+) were also indicated. The experiment was performed three times and representative data are shown. B, expression of polySia in cell culture medium mentioned in A was evaluated using an anti-polySia antibody. The secreted polysialylated enzymes are indicated with molecular mass. C, expression of ST8Sia5L in the culture supernatant. Secreted enzymes in the culture medium were analyzed with or without PNGaseF treatment (PNGaseF- or PNGaseF+) using an anti-His antibody. The secreted enzymes without N-glycans are indicated with molecular mass.

We also analyzed the relationship between autopolysialylation sites and secretion from ST8Sia2-and ST8Sia4-expressing cells, because ST8Sia2 and ST8Sia4 are typical autopolysialylated polysialyltransferases although the secretion of the polysialylated enzymes has not been analyzed. As shown in Figure 7A, the ST8Sia2 enzyme, including mutant enzymes, were expressed in cell lysates. PolySia staining in the lysates (Fig. 7B) differed from that of ST8Sia5 (Fig. 6B) because of the presence of polysialylated NCAM in the upper part of the gel (greater than 210 kDa). All the enzymes showed NCAM polysialylation (Fig. 7B), indicating that the N-glycosylation of the enzymes had no effect on the polysialylation activity for NCAM. This result is consistent with previous results (21). However, the enzymes, 2-3 (N89), 2-5 (N219), and 2-6 (N234), showed faint polySia staining (70–140 kDa) in the cell lysates (Fig. 7B), indicating that these three positions are the autopolysialylated sites, which is consistent with previous results (21). Next, we analyzed polySia staining in the culture supernatant; the 2-3 (N89), 2-5 (N219), and 2-6 (N234) enzymes showed the same results as obtained for the lysates in which autopolysialylated enzymes were absent (Fig. 7, A and B), indicating that autopolysialylation is highly related to secretion. In the case of ST8Sia4 (Fig. 7, C and D), 4-2 was identified as the key site for autopolysialylation and secretion.

Figure 7.

Effects of glutamine point-mutations at the N-glycosylation site of ST8Sia2 and ST8Sia4 on the enzyme andpolySia expression.A, expression of ST8Sia2. ST8Sia2 wild type and its six mutants (2–1 (N60Q), 2-2 (N72Q), 2-3 (N89Q), 2-4 (N134Q), 2-5 (N219Q), and 2-6 (N234Q)) in the homogenate of transiently transfected cells (Lysate) and culture supernatant (Culture medium) were analyzed in the absence or presence of PNGaseF treatment was performed (PNGaseF+) via western blotting using an anti-His-tag antibody. B, expression of polySia. Polysialylated glycoconjugates in cell homogenates (Lysate) and culture supernatant (Culture medium) were evaluated via western blotting using an anti-polySia antibody. For validation, EndoN treatment (EndoN- or EndoN+) was performed. C, expression of ST8Sia4. ST8Sia4 wild type and its five mutants (4–1 (N50Q), 4-2 (N74Q), 4-3 (N119Q), 4-4 (N204Q), and 4-5 (N219Q)) in the homogenate of transiently transfected cells (Lysate) and culture supernatant (Culture medium) were analyzed via western blotting using an anti-His-tag antibody in the absence or presence of PNGaseF treatment was performed (PNGaseF+). D, expression of polySia. Polysialylated glycoconjugates in cell homogenate (Lysate) and culture supernatant (Culture medium) were evaluated via western blotting using an anti-polySia antibody. For validation, EndoN treatment (EndoN- or EndoN+) was performed. The experiment was performed three times and representative data are shown. Specific bands are indicated with molecular mass.

We analyzed the characteristics of the secreted form of ST8Sia5. We purified the exosomal fractions from the supernatant, considering that some glycosyltransferases are secreted into the exosomes (28). We analyzed the expression of polySia-ST8Sia5 in both the exosome fraction and the exosome-deprived supernatant (Exo-sup) (Fig. 8). The exosomal fraction was confirmed by detecting exosomal markers (Fig. S4). Next, we analyzed the secretion of the enzyme and found polySia expression at 100 to 140 kDa (Fig. 8A) and enzyme staining at 37 kDa (Fig. 8B) in Exosup, but not in EVs, which indicated that ST8Sia5 was secreted into the supernatant in a soluble form, unlike GnT-V. The molecular mass of the enzyme in the lysate was 39 kDa. We also analyzed the secretion of the enzyme with GM6001 that inhibits the activity of ADAMs and MMP because MMPs (MMP2 and 14) and ADAMs (ADAM 9, 10, 12, and 17) were expressed in B16 cells (Fig. S5). B16 cells (Mock and ST8Sia5-transfected cells) were incubated with or without GM6001 at different concentrations. The cells and cell culture medium were subsequently harvested. To detect the enzyme bands, we performed PNGaseF treatment to remove N-glycans causing the smearing. As shown in Figure 8C, the secretion of the enzyme into the culture medium was completely inhibited by the addition of GM6001, which indicated that the cleavage is responsible for the activity of metalloprotease(s). To understand the GQ1b synthetic activity of the extracellular enzyme, we validated the enzymatic activity using ELISA employing a specific GQ1b antibody (Fig. S6) (29). As shown in Figure 8D, enzymatic activity was observed only after EndoN treatment, which indicated that the degradation of polySia is required for the ganglioside synthase activity. The docking simulation of the ST8Sia5, CMP-Neu5Ac and GT1a or GT1b using αFold3 (AF3) (30) revealed that the external monoSia structure of the ganglioside and CMP-Sia were located at the active site of ST8Sia5 (Fig. 8E).

Figure 8.

Localization of the secreted enzyme.A, expression of polySia in culture supernatant. Culture supernatant of mock- and ST8Sia5L-transfected cells was subjected to ultracentrifugation to obtain exovesicles (EVs) and the remaining supernatant (Exosup). EVs, Exsosup, cell lysate (Lysate), and culture supernatant (Input) were analyzed for polysialic acid expression via western blotting using an anti-polySia antibody. The samples were treated without or with EndoN (−/+). The arrows show the enzyme bands, and their molecular masses are indicated. B, expression of the enzyme in the culture supernatant. The same samples as in A were analyzed for enzyme via western blotting using an anti-His antibody. The samples were treated without or with PNGaseF (−/+) because polySia shield the protein epitope of the antibody. The arrows show the enzyme bands, and their molecular masses are indicated. The asterisk shows a non-specific band. C, the effects of the GM6001 on the secretion of the enzyme into the supernatant. The ST8Sia5L-transfected cells were incubated without (control) or with different concentrations of the GM6001 (10 and 20 μM). The DMSO control is shown as DMSO 10 and DMSO 20. The presence of the enzyme was confirmed using the anti-His antibody before or after PNGaseF treatment (PNGaseF −/+). D. GQ1b synthetic activity of the secreted enzyme. GT1b, a precursor of GQ1b, was immobilized onto plastic wells and the enzyme fraction was incubated with CMP-Neu5Ac at 37 °C for 24 h. The GQ1b synthetic activity of the enzyme was validated using a GQ1b-specific antibody. The concentrated supernatant of the cell (Mock and ST8Sia5) with (+) or without (−) EndoN were incubated and activity was analyzed. Data are shown as mean ± SD from three technical replicates. Individual data points represent technical replicates. Statistical significance was evaluated using an unpaired two-tailed Student’s t test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. E, docking of the ST8Sia5L and its substrates, GT1a/GT1b and CMP-Neu5Ac, using AF3. GT1a, GT1b, and CMP-Neu5Ac are shown in green, yellow and pink.

Finally, we tried to observe the conformation of the ST8Sia5L in comparison with ST8Sia2 and 4 using AF3 (30). The autopolysialylation position in the ST8Sia5 enzyme at N92 was the same as in ST8Sia2 (N89) and ST8Sia4 (N74) (Fig. 9). All these N-linked positions were autopolysialylated and involved in the secretion of the enzymes although the autopolysialylated positions N234 in ST8Sia2 and N219 in ST8Sia 4 were not involved in the secretion of the enzyme unlike N277 in ST8Sia5 (Figs. 6 and 7). We modeled the autopolysialylation state of the enzymes (Fig. 9B) and found that dimerization was required for the autopolysialylation of both N92 and N277 for reaching the N-linked glycan at the active site of the enzyme. Next, we docked the polysialic acid and CMP-Neu5Ac to the enzymes using AF3 (30). Docking simulation showed that the non-reducing terminal end of polysialic acid was successfully docked with the enzyme. In addition, polysialic acid was mainly bound with three basic amino acids (R289, R333, and K380) in ST8Sia5 and the direction of the reducing terminal end of polysialic acid was extended to the top of the enzyme. This situation was caused by spatial hindrance due to Y286. R333 is located in the corresponding polysialyltransferase domain (PSTD). In the case of ST8Sia2 and 4, polysialic acids were docked successfully at the non-reducing terminal end at the active site, and the direction of the reducing terminal end of the polysialic acid was extended beside the enzymes, unlike for ST8Sia5. The polysialic acid was mainly bound with four basic amino acids (R212, R292, R280, and R277 for ST8Sia2 and R197, R276, R265, and R236 for ST8Sia4) and these amino acids, except for R212 in ST8Sia2 and R197 in ST8Sia4, were located in the PSTD that is considered to be involved in the action of the polysialylation for NCAM (Fig. 5). Because the polysialic acid binding region of ST8Sia5 is not the same as PSTD in ST8Sia2 and 4, we searched for the PSTD sequences from ST8Sia5 based on the sequence from ST8Sia2, and ST8Sia4 (Fig. S7). The PSTD domain is the region where basic amino acids are involved and is present in front of the sialyl motif S. ST8Sia2 and ST8Sia4 have 11 and 9 basic amino acids, respectively, and ST8Sia5 has six basic amino acids in the corresponding region. In addition, we identified a new candidate region with an important basic amino acid cluster in the ST8Sia5 sequence (Fig. S7A, candidate1), and R333 was predicted to bind to polysialic acid (Fig. 10A). We mutated all the basic amino acids in candidates 1 and 2 (Figs. 10B and S7A) and analyzed polySia expression. As shown in Figures 10, B and C, and S7B, the predicted R333A (Fig. 10A) in candidate 2 showed dramatically reduced polySia staining. In candidate 1, polySia expression was reduced in R257A and R333A (Fig. 10A), indicating that candidate 1 is also involved in the action of polysialyltransferase. Finally, based on the AF3 docking results, we generated alanine-substituted mutants in which the basic amino acids R289, R333, and K380, predicted to contribute to the interaction with polySia, and Y286, which was predicted to enable the upward trapping of polySia, were each replaced with alanine. Expression of the WT and the four mutants was confirmed in cell lysates (Fig. 10B). Western blotting of the culture media using an anti-polySia antibody showed that all the mutants exhibited a dramatic reduction in polySia synthetic activity compared with the WT (Fig. 10C). These results indicate that R289, R333, K380, and Y286 play critical roles in polySia biosynthesis by ST8Sia5L, and suggest that ST8Sia5L synthesizes polySia through a mechanism distinct from the PSTD-dependent polySia synthesis reported for ST8Sia2 and ST8Sia4.

Figure 9.

Prediction of theautopolysialylation sites in ST8Sia5L, ST8Sia2, and ST8Sia4.A, the space-filling models of ST8Sia5L, ST8Sia2, and ST8Sia4 were generated using AF3. The schematic structures of the motifs and regions are colored. PBR is indicated in blue; sialyl motifs L, S, III, VS, and PSTD are shown in green, pink, yellow, light pink, light blue, and orange, respectively. Red indicates the N-glycosylation sites. N-glycans in ST8Sia 5, ST8Sia2 and ST8Sia4 are indicated in red, blue, and light green, respectively. The locations of N92 in ST8Sia5, N89 in ST8Sia2, and N74 in ST8Sia4 indicated in the circle are the same. B, homodimer modeling for the biosynthesis of the polysialic acid on the N-linked glycan of ST8Sia5. Left panel shows polysialylation on N92 and right panel shows polysialylation on N277. The colors are the same as in A.

Figure 10.

Polysialic acid-trapping motif (PSTM) critical forpolySia synthesis of ST8Sia5L.A, docking model among ST8Sia, polysialic acid, and CMP-Neu5Ac. ST8Sia5L, ST8Sia2, and ST8Sia4 were docked with CMP-Neu5Ac and (→8Neu5Acα2)₁₁→3Gal structure using AF3. Purple shows the polysialyltransferase domain (PSTD). Red amino acids are basic amino acids (R, K, and H) located within the 3.2 Å of the polysialic acid. Four mutants were created by substituting each of the four amino acid residues (R289, R333, K380, and Y286) in A, which were believed to contribute to ST8Sia5L polySia synthesis, with alanine. These mutants were transfected into B16 cells, and cell lysates (B) and culture media (C) were subjected to SDS-PAGE and western blotting. Cell lysates were treated with PNGaseF to remove N-glycans and detected using the anti-His-tag antibody. Culture medium was assayed using the anti-polySia antibody.

Discussion

Polysialyltransferases are key enzymes responsible for polysialylation in the brain—two such enzymes, ST8Sia2 and ST8Sia4, are known to exist (3, 4). Besides polysialylation of NCAM, these enzymes undergo autopolysialylation (19). However, the function of autopolysialylation remains to be elucidated. In this study, we identified, for the first time, that ST8Sia5L is a new polysialyltransferase that autopolysialylates its N-glycans. This finding raises three additional novel and important aspects of sialyltransferases: (1) one sialyltransferase has two enzyme activities with different specificities; (2) autosialylation of this sialyltransferase is involved in both secretion and suppression of the enzyme activity; and (3) it highlights the significance of enzyme isoforms in the regulation of their localization and activities.

The polysialyltransferase activity of ST8Sia5 was indeed surprising, because ST8Sia5 was hitherto recognized as a disialic acid-synthesizing enzyme with specificity for gangliosides (23, 24). ST8Sia5, also known as SAT-V, was cloned from mouse and human brains by two groups in 1996 (23, 24). It usually synthesizes GT1a/GQ1b from GD1a/GT1b. However, the synthesis of GD1c from GM1b may not occur under natural conditions, not only because of the reason described in our previous study (25), but also because the K_m value for the reaction is two orders of magnitude higher (23, 24). In this study, we demonstrated that ST8Sia5 exhibits polysialyltransferase activity on glycoproteins (Figs. 3 and 4) while maintaining its activity toward gangliosides; for example, it catalyzes the synthesis of GQ1b from GT1b (Fig. 8D). This is a novel finding wherein one enzyme modulates two different types of substrates: proteins and gangliosides. To date, no sialyltransferase that modifies glycans of both gangliosides and glycoproteins with different specificities (disialic acid or polysialic acid) has been reported, highlighting the novelty of our finding. NCAM is a major polySia-carrier protein in the brain, as is evident from the fact that only 5% of polySia remains in NCAM-KO mice (31). However, ST8Sia5 does not synthesize polySia on NCAM (Figs. 4 and S3) but it exclusively undergoes autopolysialylation. ST8Sia2 and ST8Sia4 exhibit autopolysialylation activity, and both these enzymes are well-known for catalyzing NCAM polysialylation. The functions of autopolysialylation in these enzymes in neural cells of the brains are not known; the new discovery of the polysialylated ST8Sia5 may address the outstanding questions. ST8Sia3 was reported to exhibit autopolysialylation activity (22), although it did not efficiently polysialylate NCAM. In this study, we did not detect a clear autopolysialylation signal for ST8Sia3 under our B16 melanoma cell expression conditions using anti-polySia antibody. This discrepancy may come from differences in experimental systems.

Two special regions, the polybasic region (PBR) and PSTD, are required for NCAM polysialylation (Fig. 5) (23). Previous structural analysis of human ST8SIA3 (18) suggested that these motifs collectively generate an extended electropositive surface around the active site for handling negatively charged oligo/polySia chains. The PBR of these enzymes binds to the acidic patch of the first fibronectin domain of NCAM, effectively enabling NCAM-specific polysialylation. Upon comparing the amino acid sequences of ST8Sia5, ST8Sia2, and ST8Sia4, we did not find PBR sequences in ST8Sia5 (Fig. 5). Thus, it is reasonable to assume that ST8Sia5 could not synthesize polySia on NCAM (Figs. 4 and S3). In ST8Sia2 and ST8Sia4, PSTD is thought to facilitate polySia elongation by contributing to a positively charged surface that coordinates the growing polyanionic chain near the catalytic site. We searched for the polysialic acid-binding site via docking simulation for both ST8Sia5 and the known polysialyltransferases ST8Sia2 and 4 using AF3 and found that the predicted polysialic acid-binding site in ST8Sia5 was different from that in ST8Sia2 and 4 (Fig. 10A). Polysialic acid was bound along with PSTD of ST8Sia2 and 4 (Fig. 10A) and R277, R280, and R292 for ST8Sia2 and R236, R265, and R276 for ST8Sia4 were found to be important. In addition, R212 and R197 were also found to be important. On the contrary, in case of ST8Sia5, polysialic acid was only located on R333 in the region corresponding to the PSTD in ST8Sia2 and 4; however, other basic amino acids R289 and K380 in other regions were found to be important. R289 was located in the predicted PSTD (we named it PSTD2), and its importance was confirmed via biochemical analyses (Fig. S7B). The direction of the extended polysialic acid was completely different and the position of the tyrosine (Y286) in ST8Sia5 was found to regulate the direction of polysialic acid toward the upper part of the enzyme via steric hindrance (Fig. 10A). We found that Y286, R289, R333, and K380 form the special motif for polysialylation of ST8Sia5, which we named “polysialic acid-trapping motif” (PSTM).

Furthermore, we found that ST8Sia5 was secreted as a fully autopolysialylated enzyme. Notably, autopolysialylated ST8Sia2 and ST8Sia4 were also secreted into the culture medium (Figs. 1, 2 and 7), which is the first such report for these enzymes, to the best of our knowledge. It is, thus, interesting to note that autopolysialylation may be related to the secretion of polysialyltransferases (polySTs), although this may not always be the case with ST8Sia2 and 4. ST8Sia5 is mainly expressed in the adult brain (23), and ST8Sia2 and 4 are the major polySTs in embryos. Therefore, secreted polysialylated enzyme may play important and ubiquitous roles in the brain. Based on our results, polysialylated ST8Sia5 can be extensively secreted; therefore, a specific binding counterpart molecule, that is, sialic acid-binding immunoglobulin-like lectins (Siglecs) (32, 33) are predicted. For example, Siglec11 on the microglia binds to polySia (34). In microglia, other Siglecs, such as CD33 and Siglec 8, 9, and 10, are present although their binding specificity for polySia remains unknown. Recently, new Sia-binding site were reported for Siglecs (35, 36). It is, therefore, important to clarify the specificity of these brain Siglecs toward polySia to understand the roles of autopolysialylated polySTs, namely ST8Sia5, ST8Sia2, and ST8Sia4 (Figs. 1 and 2). Although further analyses of secreted autopolysialylated polySTs are necessary, autopolysialylated polySTs may regulate the inactivation of Siglec by binding to its V-set domain. In particular, in the brain, the relationship between sialic acid, Siglecs, especially CD33, and Alzheimer’s disease is a hot topic of research (37), and in this context, secreted enzymes may be one of the regulators.

Besides the potential importance of polySia-ST8Sia5 in regulating the functions of the brain, we demonstrated that autopolysialylation of ST8Sia5 shielded the GQ1b synthase activity of the enzyme. Although autopolysialylated ST8Sia5 did not exhibit sialylation activity, it can become an active enzyme that sialylates GT1b to GQ1b only when polySia is degraded by sialidase (Fig. 8D) or under acidic conditions via lysosomal exocytosis (38). Notably, polySia has the ability to inhibit the ganglioside synthetic activity of ST8Sia5, which is a novel function of polySia. As an important regulator of polySia function, the secretion of sialidases and the subsequent degradation of polySia under acute stress have been demonstrated in the brain (14) as well as in immune cells (39). The sialidase-driven degradation of polySia or the lysosomal exocytosis (38) would not only induce the loss of polySia function but would also expose the hidden ganglioside GQ1b synthase activity of ST8Sia5. Sialidase or acidic conditions degrade key sialic acid-containing glycoconjugates, such as polySia (3, 4) and gangliosides, which regulate the receptors (3, 4).

GQ1b is an important molecule involved in the proliferation and differentiation of the brain cells (3, 4). In neurons, it is localized in the myelin around the Ranvier’s constriction ring of the cranial nerves (oculomotor, pulmonic, and abducens nerves) (40). The autoimmune antibodies directed against GQ1b sometimes attack leading to the Fisher syndrome. Therefore, GQ1b levels at this site are always kept constant, and the recovery of the levels is considered important when reduced. This recovery system is recognized as a ubiquitous resilient system. In this system, the activation of sialidase is usually transient, and long activation of sialidase should be avoided. Notably, the sialidase activity is rapidly decreased after secretion (39). ST8SIA5 is mostly expressed in the adult brain in both excitatory and inhibitory neurons, although it is also expressed in both astrocytes and oligodendrocytes (https://www.proteinatlas.org/ENSG00000101638-ST8SIA5/single+cell). Thus, the neuronal state-dependent secretion of sialidase could be a new regulatory mechanism that is important for enhancing signals and attaining resilience in the GQ1b-enriched brain regions.

We found that only ST8Sia5L showed polysialyltransferase activity, unlike the two other isoforms, ST8Sia5M and ST8Sia5S, present in the brain. In a previous study, we demonstrated that only ST8Sia5L exhibits the activity because ST8Sia5S and ST8Sia5M are localized in the ER, and not in the Golgi apparatus (25). This is consistent with the finding that only ST8Sia5L exhibits the polysialyltransferase activity. As ST8Sia5M can synthesize gangliosides when expressed in COS-7 (23), ST8Sia5S and ST8Sia5M may be ganglioside-synthesizing enzymes, whereas ST8Sia5L could be a dual-acting enzyme for secretion and targeting to sites other than the ER and Golgi (Fig. 10). Notably, glycosyltransferase isoforms determine the different destinations of enzymes, which is a new concept for sialyltransferase isoforms. In addition, the biosynthetic activity of polySia on the N-linked glycan of the enzyme is only achieved by the long stem region of the enzyme from the transmembrane domain, probably due to it reaching the active site of the enzyme (Figs. 9B and S2B). Deciphering the effects of glycans on the activity and localization of glycosyltransferase is an important topic for further research.

In summary, we found a new activity of ganglioside-synthesizing enzyme—dual activity to synthesize disialic acid and polysialic acid. In addition, the polysialylated enzyme was found to be secreted in keeping with the activity for which desialylation is required. All these results indicate the discovery of the new function of the secreted sialyltransferases.

Experimental procedures

Materials

The mouse melanoma cell line B16 was purchased from the RIKEN Cell Material Development Laboratory (Tsukuba, Japan). Dulbecco’s modified Eagle medium (Low Glucose), bovine serum albumin (BSA), Triton X-100, Tris (tris (hydroxymethyl) aminomethane), 2-mercaptoethanol, ethylenediamine-N, N,N,N′-tetraacetic acid disodium salt dihydrate (EDTA), acrylamide, penicillin G postassium, and streptomycin sulfate were purchased from Fujifilm Wako Pure Chemicals (Tokyo, Japan). β-Glycerophosphoric acid, sodium orthovanadate (Na₃VO₄), pyrophosphoric acid, 3,3,5,5-tetramethylbenzidine (TMB), and phenylmethanesulfonyl fluoride (PMSF) were purchased from Sigma-Aldrich. Protease inhibitor cocktail (PIC) contained 1 μg/ml leupeptin, 2 μg/ml antipain, 10 μg/ml benzamidine, 1 μg/ml pepstain, and 1 μg/ml aprotinin. Phosphate-buffered saline (PBS; pH 7.4) was prepared by dissolving 137 mM NaCl, 8.1 mM Na₂HPO₄, 2.7 mM KCl, and 1.5 mM KH₂PO₄ in water. PBST was prepared by adding 0.05% Tween-20 to PBS. The mouse-735-hIgG chimera antibody (c735 antibody) and mutant inactive antibody were prepared by fusing the variable portion of the 735 antibody, a mouse IgG that recognizes the interior portion of polysialic acid with a degree of polymerization greater than 11, with the stationary portion of human IgG (hIgG) (15). Anti-His mouse monoclonal IgG antibody (#66005-1Ig) was purchased from Proteintech (Rosemont, IL, USA). PEI “MAX” (polyethylenimine “MAX”) and COSMEDIUM005 (for hybridoma cell culture) were purchased from Cosmo Bio (Tokyo, Japan). Protein A Sepharose 4 Fast Flow was purchased from Cytiva (Tokyo, Japan). ECL (enhanced chemiluminescence) Prime (ECL prime) was purchased from Cytiva. Whatman 3 MM paper was purchased from Whatman and Pre-stained XL-Ladder Broad was purchased from Pharma Foods Apro Science Group. PNGaseF, an enzyme that specifically cleaves the binding site (GlcNAc-Asn linkage) between N-glycosidic sugar chains and proteins, and 10 × denaturing buffer, 10 × glyco Buffer 2, and 10% NP-40 used for the reaction were purchased from New England Biolabs. NucleoBondXtramidi, a bicinchoninic acid protein assay kit, was purchased from Takara Bio (Shiga, Japan). Endo-N-acylneuraminidase (EndoN), an enzyme that digests polySia chains with a degree of polymerization greater than 5, was prepared in our laboratory (41). GT1b and GQ1b were purchased from Calbiochem. Cytidine-5′-monophospho-N-acetyl neuraminic acid disodium salt (CMP-Sia) was purchased from Nacalai Tesque (Kyoto, Japan). 2-(N-morpholino) ethanesulfonic acid) was purchased from NACARAI CHEMICALS, Ltd. Nanosep with 10k Omega 100 was purchased from Nihon Pall Ltd (Tokyo, Japan). H₂O₂ was purchased from the Mitsubishi Gas Chemical Company (Tokyo, Japan).

Construction of the plasmids and establishment of cell lines

Mouse ST8Sia2, ST8Sia4, and ST8Sia5 cDNAs were cloned into the pcDNA3.1 vector (Takara Bio) containing a C-terminal V5 epitope tag and a His6 tag. The resulting plasmids containing the DNA sequences for ST8Sia2, ST8Sia4, and ST8Sia5 served as templates for the creation of all the NQ and alanine mutants. All mutants were created using KODplus-Neo and DpnI, according to the manufacturer’s instructions; the oligonucleotide primers used for this are listed in Table S1. To confirm the presence of mutations, the resulting DNA was sequenced at Eurofins Genomics in Tokyo.

Mouse melanoma B16 cells (2 × 10⁵) were seeded in six-well plates and cultured for 48 h. The cells in each well were transfected with 3 μg of each plasmid in 200 μl PBS containing 5 μg PEI Max. The cells were incubated with the transfection mixture for 6 h at 37 °C, which was followed by changing the medium to COSMEDIUM005. After 60 h, the cells were homogenized in a lysis buffer (1% Triton X-100, 1 mM PMSF, and PIC), 1 mM EDTA, 50 mM NaF, 10 mM β-glycerophosphate, 10 mM sodium pyrophosphate, and 1 mM sodium ortho-vanadate in PBS), the culture medium was collected and mixed with 1 mM PMSF and 1 μg/ml aprotinin. Homogenates were incubated on ice for 1 h and centrifuged at 10,000g for 15 min at 4 °C. The supernatant was collected, and protein concentration was determined using the bicinchoninic acid assay.

Bulk RNA-seq data analysis

Bulk RNA-seq data were obtained from the Tabula Muris Senis (https://tabula-muris-senis.sf.czbiohub.org/) resource and analyzed in R using the TabulaMurisSenisData, SingleCellExperiment, edgeR, dplyr, and ggplot2 packages. The bulk RNA-seq dataset was imported into R using the corresponding Bioconductor package. For this analysis, only samples from males were included, and samples from the brain, liver, lung, heart, kidney, and spleen were selected for visualization. Raw count data were normalized via the trimmed mean of M-values (TMM) method, and expression values were transformed into logCPM. ST8Sia5 expression was extracted from the dataset, and its distribution across the selected organs was visualized using boxplots overlaid with individual sample points. The organs were arranged in descending order according to the median expression level of ST8Sia5 in each organ. Data processing was performed using the SingleCellExperiment framework, and figure generation was carried out with ggplot2.

Quantitative real-time PCR (qPCR)

Mice (14-month-old female) were obtained from Chubu Kagaku Shizai (Nagoya, Japan) and maintained in a controlled environment (23 °C ± 2 deg and 50 ± 10% humidity, 12:12 light/dark cycle) with food and water available ad libitum. At least 1 week prior to the experiments, mice were habituated to our facilities. All procedures were approved by the Animal Care and Use Committee of Nagoya University (Permit Number: A240095–003). qPCR was carried out using cDNA derived from the cerebrum of 14-month-old female mice (Each reaction was prepared in a total volume of 20 μl, containing 4 μl of diluted cDNA, 10 μl of THUNDERBIRD SYBR qPCR Mix, 0.6 μl each of 10 μM forward and reverse primers, and 4.8 μl of nuclease-free water. PCR amplification was performed for 50 cycles under the following conditions: 95 °C for 15 s, 60 °C for 15 s, and 73 °C for 30 s. Fluorescence signals were collected during amplification, and Ct values were calculated from the amplification curves. Following amplification, melting curve analysis was conducted to assess product specificity. The temperature was raised from 60 °C to 95 °C in 0.5 °C increments every 30 s, and the melting curves and derivative melting peaks were examined to confirm the presence of a single specific amplicon. Absolute quantification was performed using pcDNA3.1-mouse ST8Sia5S, pcDNA3.1-mouse ST8Sia5M, and pcDNA3.1-mouse ST8Sia5L plasmids as standards.

ELISA

Fifty microliters of culture supernatant (diluted five times in PBS) was added to each well of a 96-well plate (MW96F Multisorp straight, Thermo Scientific, #467340) and allowed to solidify for 24 h at 4 °C. The plate was washed three times with PBS and blocked with 100 μl 0.5% BSA in PBS at 37 °C for 1 h. The plate was washed three times with PBS and incubated with 5 μg/ml anti-polySia antibody (c735) (15) or anti-polySia antibody mutant (735 mut) (15) at 37 °C for 2 h. The plate was then washed four times with PBST, and 50 μl of streptavidin-POD was added at a 10,000-fold dilution to each well, followed by incubation at 37 °C for 1 h. The plate was again washed five times with PBST, and 50 μl of a mixture of TMB, H₂O₂, and citric acid was added to develop color. The reaction was stopped by adding 100 μl of 2N H₂SO₄ at an appropriate time. The absorbance of solution in the wells was measured at 450 nm, and the value for 735 mut was subtracted from that for c735.

Western blotting

To remove N-glycans, some of the cell lysates used for detection with the anti-His antibody were incubated in 1 × Glycoprotein Denaturing Buffer (New England Biolabs) at 100 °C for 10 min. The de-N-glycosylation reaction mixtures contained 1 × Glycobuffer2 (New England Biolabs), 1% NP-40, and 20 U PNGaseF, and were incubated overnight at 37 °C. All samples were dissolved in Laemmli buffer containing 5% mercaptoethanol and were subsequently incubated at 60 °C for 20 min or at 100 °C for 3 min for detection with the anti-polySia antibody (c735) or anti-His antibody, respectively. Proteins from each sample (20 μg BSA) were separated via electrophoresis on a 7.5% sodium dodecyl sulfate-polyacrylamdide gel and blotted onto polyvinylidene difluoride membranes. The membrane was then blocked with PBST containing 1% BSA for the anti-glycan antibody or 1% skim milk for the anti-protein antibody at 25 °C for 1 h. Thereafter, the membrane was incubated overnight with primary antibodies, namely the anti-polySia monoclonal IgG antibody or the anti-His monoclonal IgG1 antibody (1:1000), at 4 °C. After washing with PBST, the membrane was incubated with a secondary antibody, HRP-conjugated anti-mouse IgG + M (0.2 μg/ml) or HRP-conjugated anti-human IgG (0.08 μg/ml), at 37 °C for 1 h. After washing with PBST, the immunoblots were developed using an ECL western blotting detection reagent.

Immunoprecipitation

In a 1.5 ml tube, 100 μl Protein A-Sepharose 50% slurry, washed three times with PBS, was suspended in 1.2 ml culture medium, and the volume was made to 1200 μl with PBS, followed by mixing by rotation at 4 °C for 1 h. The sample after mixing was separately stored as a 10 μl Input fraction before mixing. To a 1.5 ml tube, 100 μl Protein A-Sepharose 50% slurry, washed three times with PBS, was added, and 12 μg of anti-polySia antibody and anti-His tag antibody were added to it. The mixture was centrifuged at 200g for 5 min at 4 °C, and the precipitated gel was washed three times with 1 ml of PBST. The pre-cleaned sample was added to the precipitated gel and mixed overnight by rotation at 4 °C. Thereafter, the mixture was centrifuged at 200g for 5 min 4 °C, and the supernatant was collected as the unbound fraction. The precipitated gel was washed three times with 1 ml of PBST, and the precipitate was collected as the bound fraction.

LC–MS/MS sample preparation

In a 1.5 ml tube, 200 μl of a 50% Protein G-Sepharose slurry prewashed three times with PBS was incubated with 40 μg of anti-polySia or control antibody. After centrifugation at 200g for 5 min at 4 °C, the gel pellet was washed three times with 1 ml of PBST. The pellet was then incubated overnight with 1.4 ml of culture medium at 4 °C with rotation. After centrifugation at 200g for 5 min at 4 °C, the gel pellet was washed three times with 1 ml of PBST, and the precipitate was collected as the bound fraction. After the final wash, the beads were transferred to a new low-binding microcentrifuge tube. For proteomic analysis, proteins were partially digested on beads in 50 μl of digestion buffer (2 M urea, 50 mM Tris–HCl pH 7.5) containing Trypsin/Lys-C (1 μg) at 37 °C for 30 min with shaking. The supernatant containing partially digested peptides was collected after centrifugation, and the beads were further washed twice with 25 μl of digestion buffer supplemented with 1 mM TCEP. The fractions were combined and subjected to overnight digestion at 37 °C with shaking. Peptide samples were subsequently reduced with TCEP (final concentration 25 mM, 37 °C, 15 min) and alkylated with iodoacetamide (final concentration 25 mM, 37 °C, 30 min in the dark). The reaction was quenched by acidification with trifluoroacetic acid, and peptides were desalted using SDB-based solid-phase extraction prior to LC–MS/MS analysis.

Mass spectrometry analysis

nanoLC-MS/MS analysis of the digested peptides was performed on nano-flow reverse-phase LC (Dionex U3000) connected to a Q-Exactive Orbitrap mass spectrometer (Thermo Fisher Scientific) through a nano electrospray ion source (AMR Inc). The desalted peptides were loaded onto a separation capillary C18 reverse-phase column (NTCC-360/100-3-125, 120 × 0.1 mm; Nikkyo Technos, Japan). The mobile phase consisted of solvent A (water containing 0.5% acetic acid) and solvent B (80% acetonitrile containing 0.5% acetic acid). Peptides were separated at a flow rate of 0.5 μl/min using a linear gradient of 5 to 40% solvent B over 100 min, followed by a wash at 95% B and re-equilibration to 5% B, with a total run time of 115 min. The mass spectrometer was operated in the data-dependent acquisition mode. The analysis conditions were set as follows: capillary temperature: 250 °C; capillary voltage: 2.0 kV in the positive mode; S-lens RF level: 50. MS1 spectra were measured with a resolution of 70,000, an AGC target of 3 × 10⁶, and a scan range of 350 to 1800 m/z. HCD MS/MS spectra were acquired with a resolution of 17,500, an AGC target of 1 × 10⁵, an isolation window of 2.0 m/z, a maximum injection time of 60 ms, and a normalized collision energy of 27. Dynamic exclusion was set to 20 s. MS/MS spectra were processed and peak lists were generated using Proteome Discoverer 2.4.1.15 (Thermo Fisher Scientific). Database searches were performed with SEQUEST HT against the Mus musculus and the cRAP contaminant database. Search parameters were as follows: enzyme specificity, trypsin with up to two missed cleavages; precursor mass tolerance, 10 ppm; fragment mass tolerance, 0.02 Da; fixed modification, carbamidomethylation (C); variable modifications, oxidation (M), HexNAc (N, S, T), and dHex (N, S, T). Peptide-spectrum matches were filtered to 1% false discovery rate using the Percolator algorithm, corresponding to high-confidence peptide identifications. Phosphorylation site localization was evaluated using ptmRS. Identifications were further inspected to confirm peptide assignments and modification sites. Only proteins with abundance values greater than or equal to the mean abundance in the anti-polySia immunoprecipitate were extracted, and the top-ranked proteins are shown in descending order according to the abundance ratio (anti-polySia antibody/control antibody).

Preparation of exovesicle (EV) fractions via ultracentrifugation

Mouse melanoma B16 cells (1.5 × 10⁶ cells) were seeded in a 10 cm dish and cultured for 48 h. These cells were transfected using polyethylenimine “MAX”, incubated for 6 h and changed the medium. The culture supernatant was mixed with 40 μl of 100 mM PMSF and 8 μl of PIC per dish and centrifuged at 10,000g for 30 min to pellet the cells. The culture supernatant was then centrifuged at 100,000g for 70 min (Beckman Optima LE-80k, Ultra Centrifuge) and the obtained supernatant was used as the Exosup fraction. The pellet was washed with PBS and the suspension was centrifuged again at 100,000g for 70 min, followed by removal of the supernatant. Lysis buffer (45 μl) was added to the pellet and the mixture was incubated for 1 h on ice; the resulting supernatant was designated as the EV fraction.

Activity assay

Mouse melanoma B16 cells (1.5 × 10⁶ cells) were seeded in a 10 cm dish and cultured for 48 h. The cells were transfected using polyethylenimine “MAX”, incubated for 6 h and changed the medium. The culture supernatant was mixed with 40 μl of 100 mM PMSF and 8 μl of PIC per dish and centrifuged at 10,000g for 15 min to pellet the cells. For enrichment, the culture medium was concentrated 10-times using a Nanosep with 10k Omega 100. GT1b (6.25 ng per well) was added to a 96-well plate, incubated at 37 °C for 4 h, and dried. The plates were then washed three times with PBS, 1% BSA in PBS was added, and incubated at 37 °C for 1 h for blocking, followed by three washes with PBS. Concentrated and non-concentrated culture media were prepared with 0.2 mM CMP-Sia and 50 mM MES and incubated for 24 h. The plates were washed three times with PBS. Thereafter, they were washed five times with PBS, and 100 ng of anti-GQ1b antibody (FS3D) and anti-GQ1b antibody mutant (FS3Dmut) were added, followed by a 2 h incubation at 37 °C. The plates were then washed four times with PBST, incubated with HRP-conjugated anti-human IgG (0.08 μg/ml) at 37 °C for 1 h, and then with a mixture of TMB, H₂O₂, and citric acid. The reaction was stopped by adding 100 μl of 2N H₂SO₄ at an appropriate time. The recombinant anti-GQ1b antibody FS3D was cloned from cDNA derived from the FS3 hybridoma, which is known to produce anti-GQ1b antibodies (42). The cloned Fab gene fragments were subcloned into the pAb-HC and pAb-LC plasmids. A recombinant negative-control antibody, FS3Dmut, was generated via site-directed mutagenesis. pAb-FS3D(HC) and pAb-FS3D(LC), or pAb-FS3Dmut (HC) and pAb-FS3D (LC), were co-transfected into Expi293F cells, and the expressed antibodies were purified using protein G–Sepharose beads. As shown in Figure S6, the reactivities of FS3D toward GQ1b and GT1b were calculated by subtracting the corresponding signal obtained with negative antibody FS3Dmut as a negative control.

Protein structure prediction, alignment, and ligand docking

Amino acid sequences of mouse ST8Sia2 (O35696), ST8Sia4 (Q64692), and ST8Sia5L (P70126) were obtained from the UniProt database (https://www.uniprot.org) for the AlphaFold3 (AF3) study. Protein structures were predicted using a local implementation of AF3. The obtained 3D structures were visualized and structurally aligned using the open-source PyMOL (http://pymol.org). For the A2 N-Glycan structure, the exact stereochemical configuration was used as registered in PubChem. Docking predictions for protein and ligand were performed using AF3. The polySia ligand used in this study consisted of a galactose linked at position 3 to a single Neu5Ac residue, followed by a linear α2,8-linked chain of 11 Neu5Ac units (DP12). This custom ligand structure was constructed de novo using the Builder tool within the Molecular Operating Environment (MOE, Chemical Computing Group). Additional ligands, including GT1a, GT1b, and CMP-Neu5Ac, were sourced from PubChem using their respective SMILES notations for input.

Data analysis

Data are presented as mean ± SD. Statistical analyses were performed using Microsoft Excel. Comparisons between two groups were carried out using an unpaired two-tailed Student’s t test. No formal tests for normality were performed. No statistical methods were used to predetermine sample size, and no data points were excluded as outliers. Individual data points shown in the figures represent technical replicates, as indicated in the figure legends. Differences were considered statistically significant at ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001.

Data availability

The gene expression data reported in this paper have been deposited in the Gene Expression Omnibus (GEO) under accession number GSE327613. The proteomic data, the raw data, peak lists, and search results have been deposited under the following accession number: jPOST (JPST004560) and ProteomeXchange (PXD077130).

Supporting information

This article contains supporting information.

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.

Reviewed by members of the JBC Editorial Board. Edited by Robert Haltiwanger