Differential Biosynthesis of Polysialic or Disialic Acid Structure by ST8Sia II and ST8Sia IV*

ST8Sia II (STX) and ST8Sia IV (PST) are polysialic acid (polySia) synthases that catalyze polySia formation of neural cell adhesion molecule (NCAM) in vivo and in vitro . It still remains unclear how these structurally similar enzymes act differently in vivo . In the present study, we performed the enzymatic characterization of ST8Sia II and IV; both ST8Sia II and IV have pH optima of 5.8–6.1 and have no requirement of metal ions. Because the pH dependence of ST8Sia II and IV enzyme activities and the p K profile of His residues are similar, we hypothesized that a histidine residue would be involved in their catalytic activity. There is a conserved His residue ( cf. His 348 in ST8Sia II and His 331 in ST8Sia IV, respectively) within the sialyl motif VS in all sialyltransferase genes cloned to date. Mutant ST8Sia II and IV enzymes in which this His residue was changed to Lys showed no detectable enzyme activity, even though they were folded correctly and could bind to CDP-hexanolamine, suggesting the importance of the His residue for their catalytic activity. Next, the degrees of polymerization of polySia in NCAM catalyzed by ST8Sia II and IV were compared. ST8Sia IV catalyzed larger polySia formation of NCAM than ST8Sia II. We also analyzed the


ST8Sia II (STX) and ST8Sia IV (PST) are polysialic acid (polySia) synthases that catalyze polySia formation of neural cell adhesion molecule (NCAM) in vivo and in vitro.
It still remains unclear how these structurally similar enzymes act differently in vivo. In the present study, we performed the enzymatic characterization of ST8Sia II and IV; both ST8Sia II and IV have pH optima of 5.8 -6.1 and have no requirement of metal ions. Because the pH dependence of ST8Sia II and IV enzyme activities and the pK profile of His residues are similar, we hypothesized that a histidine residue would be involved in their catalytic activity. There is a conserved His residue (cf. His 348 in ST8Sia II and His 331 in ST8Sia IV, respectively) within the sialyl motif VS in all sialyltransferase genes cloned to date. Mutant ST8Sia II and IV enzymes in which this His residue was changed to Lys showed no detectable enzyme activity, even though they were folded correctly and could bind to CDP-hexanolamine, suggesting the importance of the His residue for their catalytic activity. Next, the degrees of polymerization of polySia in NCAM catalyzed by ST8Sia II and IV were compared. ST8Sia IV catalyzed larger polySia formation of NCAM than ST8Sia II. We also analyzed the (auto)polysialylated enzymes themselves. Interestingly, when ST8Sia II or IV itself was sialylated under conditions for polysialylation, the disialylated compound was the major product, even though polysialylated compounds were also observed. These results suggested that both ST8Sia II and IV catalyze polySia synthesis toward preferred acceptor substrates such as NCAM, whereas they mainly catalyze disialylation, similarly to ST8Sia III, toward unfavorable substrates such as enzyme themselves.
Recently, two cDNAs encoding polysialyltransferases have been cloned. These enzymes, ST8Sia II (STX) and ST8Sia IV (PST), showed 59% sequence identity at the amino acid level (20 -24). Of particular interest, both ST8Sia II and IV could synthesize polySia on the ␣2,3-linked Sia residues on NCAM in vivo and in vitro without an initiator ␣2,8-sialyltransferase (21)(22)(23)(25)(26)(27)(28)(29). Increased levels of ST8Sia II expression were observed in the embryonic brain, whereas ST8Sia IV was continuously expressed in various tissues throughout development (19,28,30). The respective roles of these enzymes in polySia biosynthesis of NCAM remain unknown (31). Detailed information on the enzymatic properties of these respective enzymes is also lacking, although there was a recent report concerning characterization of ␣2,8-polysialyltransferase activity in the embryonic chick brain (32).
Both ST8Sia II and IV enzymes are type II membrane proteins and have conserved sequence motifs, namely sialyl motif L, S, and VS, which are present in all eukaryotic sialyltransferases characterized to date (20,33,34). Mutagenesis analysis within sialyl motif L of ST6Gal I showed that this region is involved in the binding of donor substrate, CMP-Sia (35). A recent study showing that mutation of sialyl motif S of the same enzyme caused changes in K m values for both the donor and acceptor substrates suggested that sialyl motif S is involved in the binding of both substrates (36). However, neither of these motifs seems to contain a catalytic amino acid that could be involved in the catalytic inverting mechanism (34). Furthermore, there have been no mutagenesis studies within sialyl motif VS.
A recent study using Fc-NCAM mutants lacking the fourth plus fifth or sixth N-glycosylation sites clearly showed that ST8Sia IV formed larger polySia chains than ST8Sia II (31). We still cannot exclude the possibility that the lack some Nglycosylation sites Fc-NCAM caused conformational changes that could affect the polysialylation rate by ST8Sia II or/and IV. Unfortunately, the polySia chains of wild-type Fc-NCAMs polysialylated by ST8Sia II and IV have not yet been characterized in detail. The ST8Sia IV enzyme itself was shown to be polysialylated in vitro, and this modification seems to be important for the enzyme activity (37). Subsequently, it was shown that both ST8Sia II and IV underwent polysialylation in vivo and that polysialylated ST8Sia II and IV were localized to the Golgi and cell surface, respectively (38,39). These polysialylated enzymes on the cell surface could modulate cell adhesion as suggested by Colley and colleagues (38). However, there have been no reports concerning the actual lengths of the polySia chains of these enzymes.
In this study we compared the enzymatic properties of ST8Sia II and IV. Then we examined the importance of the His residue in sialyl motif VS of these enzymes for their catalytic activity. Next, the DPs of polySia in NCAM catalyzed by ST8Sia II and IV were analyzed. Furthermore, we re-examined the actions of both ST8Sia II and IV on themselves, and we found that in this case NeuAc␣2,8-NeuAc␣2,3-R was the major product instead of ␣2,8-polysialylation.

Materials
Tissue culture media and reagents were purchased from Life Technologies, Inc. Nitrocellulose membranes and protein molecular weight standards were from Bio-Rad. CMP-[ 14 C]NeuAc, IgG-Sepharose, and protein A-Sepharose were purchased from Amersham Pharmacia Biotech. Qiagen columns for DNA purification were obtained from Qiagen Inc. (Chatsworth, CA). 35 S-Express protein labeling mix was purchased from PerkinElmer Life Sciences. Peptide-N-glycosidase F was purchased from Roche Molecular Biochemicals. A vector plasmid containing human Fc-NCAM cDNA was a gift from Drs. P. Crocker and D. L. Simmons, Oxford University, Oxford, UK (40). CDP-hexanolamine-Sepharose was a gift from Dr. K. J. Colley, the University of Illinois, Chicago. All other chemicals were purchased from Sigma.

Methods
ST8Sia II-H348K and ST8Sia IV-H331K Proteins-ST8Sia II-H348K-bs ϩ and ST8Sia IV-H331K-bs ϩ were generated using a Quick Change Site-directed Mutagenesis kit (Stratagene). Primer 1 (TC-CCAGGCCAGCCCCAAAACCATGCCCTTGGAA) and primer 2 (TTC-CAAGGGCATGGTTTTGGGGCTGGCCTGGGA) were used to mutate the His residue at position 348 of ST8Sia II to a Lys residue. Primer 3 (TCCAACGCCAGTCCTAAAAGGATGCCATTAGAA) and primer 4 (TTCTAATGGCATCCTTTTAGGACTGGCGTTGGA) were used to mutate the His residue at position 331 of ST8Sia IV to a Lys residue. ST8Sia II-348K and ST8Sia IV-H331K constructs were then cloned into the pCDSA expression vector at the XhoI site.
Preparation of Recombinant Proteins-COS-7 cells were grown on 150-mm tissue culture dishes and transiently transfected using Lipofectin with the vector plasmids designated as pCDSA-ST8Sia II, pCDSA-ST8Sia II-H348K, pCDSA-ST8Sia IV, and pcDSA-ST8Sia IV-H331K, containing cDNAs encoding IgG binding domain of protein A-fused soluble mouse ST8SiaII, ST8Sia II-H348K, ST8SiaIV, and ST8Sia IV-H331K, respectively (24,30). Aliquots of 20 ml of the medium were collected 48 h after transfection, and the soluble enzymes were adsorbed to 20 l of IgG-Sepharose resin (50% suspension in PBS), and then used as enzyme preparation as described previously (30). Soluble enzymes adsorbed to IgG-Sepharose resin were analyzed by SDS-polyacrylamide gel electrophoresis and silver staining. Aliquots of 2 l of each enzyme preparation used for sialyltransferase assay were calculated to contain 3.0 pmol of the enzyme.
Fc-NCAM protein was prepared as described previously (40). Briefly, COS-7 cells were transfected with a vector plasmid containing a cDNA encoding human NCAM fused with the Fc region of human IgG1. After 72 h, the medium was collected and applied to a protein A-Sepharose column (1 ϫ 1.2 cm). After the column was washed with 10 volumes of PBS, Fc-NCAM was eluted with 0.1 M sodium citrate buffer (pH 3.0), and then neutralized immediately with 1 M Tris-HCl (pH 9.0). Fc-NCAM protein was precipitated with 3 volumes of ice-cold ethanol, suspended with water, and used for sialyltransferase assay. Protein concentration was determined by BCA assay (Pierce).
Enzyme Assay-Enzyme assays were carried out in the presence of 50 mM MES buffer (pH 6.0), 0.5% Triton CF-54, 100 M CMP-[ 14 C]NeuAc, 0.6 g of Fc-NCAM, and 2 l of enzyme preparation, in a total volume of 10 l, as described previously (30), unless described otherwise. For ST6GalNAc IV enzyme assay, 10 g of asialofetuin was used as an acceptor substrate. After 2 h (for SDS-polyacrylamide gel analysis) or 16 h (for peptide N-glycanase digestion and for mild acid hydrolysis) of incubation at 37°C, the reaction mixture was terminated by the addition of Laemmli sample buffer (for SDS-polyacrylamide gel analysis) (41) or by quick freezing at Ϫ80°C (for PNGase F digestion and for mild acid hydrolysis). Product Analysis-For product identification by SDS-polyacrylamide gel analysis, the reaction mixture containing Laemmli sample buffer was denatured at 65°C for 20 min, loaded onto the gel, and analyzed with a BAS2000 radioimage analyzer (Fuji).
For product analysis by Mono-Q HPLC following PNGase F digestion or mild acid hydrolysis, enzyme reaction was carried out essentially as described above, but on a 5-or 10-fold larger scale. The reaction mixtures containing 14 C-polysialylated Fc-NCAM were diluted with 0.5 ml of PBS, and then 50 l of protein A-Sepharose (50% suspension in PBS) was added to adsorb 14 C-polysialylated Fc-NCAM. For PNGase digestion, the resin was washed with PBS five times and was suspended with 100 l of 10 mM Tris-HCl (pH 7.4), 50 mM EDTA, 0.2% SDS, 1% Nonidet P-40, and 20 mM ␤-mercaptoethanol, and then 10 l (10 unit) of PNGase F was added to it (42). After incubation at 37°C for 16 h with gentle shaking, the sample was centrifuged, and the supernatant containing free N-linked glycan chains was collected for further analysis by Mono-Q HPLC (31,43,44). The sample was diluted to 2 ml with water, and then applied to the Mono-Q HR 5/5 column (0.5 ϫ 5 cm, Amersham Pharmacia Biotech) pre-equilibrated with solvent A (10 mM Tris-HCl (pH 8.0)). By using solvent A and solvent B (1 M NaCl, 10 mM Tris-HCl (pH 8.0)), two-step linear gradients were generated with a Tosoh HPLC system. Elution was performed for the first 5 min with 0% of solvent B, 0 -30% solvent B for the next 20 min, and 30 -55% solvent B for the next 75 min. The elution rate was 1 ml/min, and fractions were collected every minute with a connected fraction collector. As an internal reference, mild acid hydrolysates of colominic acid were co-injected with the sample. The elution patterns of oligo/polysialic acid chains and 14 Cpolysialylated compounds were monitored by UV absorption at 210 nm and scintillation counting, respectively. Before the next sample was injected, the column was washed with solvent B for 10 min and then re-equilibrated with solvent A. 14 C-Polysialylated ST8Sia II adsorbed to IgG-Sepharose was subjected to controlled mild acid hydrolysis in 50 mM sodium acetate buffer (pH 4.8) at 37°C for 4 days. The method has been previously established to detect sialic acid dimer and oligomers (43, 50). After incubation, the sample was centrifuged, and 90% of the supernatant was used for the Mono-Q HPLC analysis which was described as above. The remaining sample was used for TLC analysis on 0.2-mm thick silica gel

FIG. 1. Polysialylation of ST8Sia II and IV themselves in vitro.
Fc-NCAM was purified using a protein A-Sepharose column from the medium of COS-7 cells transiently expressing this molecule. ST8Sia II and IV, which were fused with IgG binding domain of protein A, were purified using IgG-Sepharose resin and were incubated in the presence or absence of Fc-NCAM. ST8Sia II (A) and ST8Sia IV (B) were incubated with (ϩ) or without (Ϫ) purified Fc-NCAM and with CMP-[ 14 C]NeuAc as described under "Experimental Procedures." SDSpolyacrylamide gel electrophoresis was carried out using a 5% polyacrylamide gel followed by radioimage analysis.
Determination of Optimum pH and Metal Ion Requirements-For determining optimum pH, the enzyme activities were measured in one of the following buffers: 50 mM MES, pH 5.5, 5.8, 6.1, 6.4, or 6.7; or 50 mM PIPES, pH 6.7, 7.0, 7.3, or 7.6. To determine metal ion requirements, the enzyme activities were determined by addition of 1 mM EDTA or 10 mM MnCl 2 , MgCl 2 , or CaCl 2 .
Kinetic Analysis-To determine the K m values of ST8Sia II and IV for N-CAM substrate, various concentrations of Fc-NCAM (0.1-15 M) and a fixed concentration of CMP-NeuAc (100 M) were used for sialyltransferase assays. Kinetic constants were obtained from Lineweaver-Burk plots. In all cases, assays were performed in triplicate.
CDP-hexanolamine-Sepharose Affinity Chromatography-COS-7 cells transiently expressing ST8Sia II, ST8Sia II-H348K, ST8Sia IV, or ST8Sia IV-H331K cDNA were labeled with 35 S-Express protein labeling mix for 1 h in methionine-and cysteine-free DMEM and then chased for 6 h in DMEM, 10% fetal bovine serum. Each medium fraction was collected and applied to a 0.5 ml of CDP-hexanolamine-agarose column. The column was washed with buffers E (10 mM sodium cacodylate, pH 6.5, 0.1% Triton CF-54, 0.15 M NaCl) and H (20 mM sodium cacodylate, pH 5.3, 0.1% Triton CF-54, 0.15 M NaCl) and then eluted with buffer H containing 5 mM CDP (45,46). IgG-Sepharose resin (20 l) was added to each fraction to pull down protein A-fused enzymes and washed with buffer H. The bead fractions were analyzed by SDS-polyacrylamide gel electrophoresis and with a radioimage analyzer.

Discrimination of Polysialylated Fc-NCAM from Polysialylated ST8Sia II and IV Themselves by SDS-Polyacrylamide Gel
Electrophoresis-When we used Fc-NCAM as an acceptor substrate and protein A-tagged ST8Sia II or IV as the enzyme for enzymatic characterization, it was necessary to discriminate between the two kinds of reaction products, i.e. polysialylated Fc-NCAM and ST8Sia II or IV themselves. The reaction products analyzed by SDS-polyacrylamide gel electrophoresis and radioimage analysis are shown in Fig. 1. When we used ST8Sia II for enzyme assay in the absence of Fc-NCAM, we observed a broad band of radioactivity extending from 60 to 90 kDa (Fig.  1A). In the presence of Fc-NCAM, we observed an additional band of Ͼ220 kDa that was thought to be polysialylated Fc-NCAM. When we used ST8Sia IV, heterogeneous bands were also observed from 100 to 240 kDa (Fig. 1B). On addition of Fc-NCAM as a substrate, we observed a polysialylated Fc-NCAM at Ͼ300 kDa. Migration patterns of the polydisperse bands in the absence of Fc-NCAM were similar with those of polysialylated ST8Sia II (70 -131 kDa) and ST8Sia IV (105-190 kDa) reported previously in Chinese hamster ovary cells (38). Taken together, these polydisperse bands were most likely polysialylated ST8Sia II and IV, respectively. In the present study, we discriminated between 14 C-polysialylated Fc-NCAM and 14 C-polysialylated ST8Sia II and IV by their migration rates on SDS-polyacrylamide gels.
Effects of pH and Divalent Cations on ST8Sia II and IV Enzyme Activities-Optimal pH for ST8Sia II and IV activity was determined over the pH range of 5.5-7.6 ( Fig. 2 and Table  I). Maximum enzyme activities of ST8Sia II and IV were observed at almost the same pH range, pH 5.8 -6.1, with Fc-NCAM as an exogenous acceptor. Next, the effects of 1 or 10 mM Mn 2ϩ , Ca 2ϩ , or Mg 2ϩ ions on ST8Sia II and IV activities were tested (Table I). At 1 mM, none of these ions showed any significant effects on enzyme activity. ST8Sia II activity was inhibited slightly by the addition of 10 mM MnCl 2 , CaCl 2 , or MgCl 2 , whereas these ions showed lesser effects on ST8Sia IV activity. It was also shown that 1 mM EDTA had little effect on ST8Sia II or IV activity (data not shown).
Kinetic Analysis of ST8Sia II and IV-Although it is widely known that both ST8Sia II and IV can polysialylate NCAM in vitro and in vivo, no quantitative data concerning the affinity of these enzymes with NCAM have been reported. Next, protein A-fused soluble enzymes, ST8Sia II and IV, were used to determine the K m values of ST8Sia II and IV for Fc-NCAM. Enzyme assays were performed using various concentrations of Fc-NCAM (0.1-15 M) as described under "Experimental Procedures." The K m value of ST8Sia II for Fc-NCAM was 3.2 M, which was almost one-third of that of ST8Sia IV (8.9 M) ( Fig.  3 and Table I). These values were significantly lower than those for any other glycosyltransferases for their acceptor substrates reported previously. pK profile of His are similar, we hypothesized that a His residue conserved in these enzymes may be involved in their catalytic activity. All sialyltransferase genes cloned to date have three consensus motifs, i.e. sialyl motif L, S, and VS. Previous reports suggested that sialyl motif L is involved in binding of CMP-NeuAc and that sialyl motif S is related to the binding of both donor and acceptor substrates. By using degenerate primers based on highly conserved sialyl motif L and S sequences, many sialyltransferase genes have been cloned (48). In contrast, sialyl motif VS has been poorly characterized. Interestingly there is a conserved His residue in sialyl motif VS of all sialyltransferase genes cloned to date (Fig. 4). To determine whether the His residue in this motif is critical for the catalytic activity, we constructed mutant ST8Sia II and IV in which the His residue of sialyl motif VS was changed to a Lys residue. As shown in Fig. 5A, both ST8Sia II and IV mutant proteins, ST8Sia II-H348K and ST8Sia IV-H331K, were secreted into the medium from COS cell transfectants, suggesting that the mutants were folded correctly. Each enzyme protein fused with protein A was prepared using IgG-Sepharose from medium and used for enzyme assay as described under "Experimental Procedures." Interestingly, we found no detectable enzyme activity for ST8Sia II-H348K or ST8Sia IV-H331K mutants (Fig. 5B). Not only polysialyltransferases but also other types of sialyl-   35 S-Express protein labeling mix in methionine-and cysteine-free in DMEM and then chased for 6 h in DMEM, 10% fetal bovine serum. As each enzyme protein was fused with IgG binding domain of protein A, the enzyme was pulled down using IgG-Sepharose resin from medium and analyzed by SDSpolyacrylamide gel electrophoresis and radioimage analysis. B and C, sialyltransferase assays were performed as described under "Experimental Procedures" using wild-type polysialyltransferases, ST8Sia II, and ST8Sia IV and their respective mutants, ST8Sia II-H348K and ST8Sia IV-H331K (B), and using wild-type ST6GalNAc IV and ST6GalNAc IV-H332K mutant (C). transferase such as ST6GalNAc IV lost enzyme activities by the replacement of His residue in sialyl motif VS with Lys residue, as shown in Fig. 5C. The result suggested that the mutation in His residue of sialyl motif VS is not specific for polysialyltransferases but is specific for sialyltransferases in general. As ST8Sia II-H348K and ST8Sia IV-H331K mutants were shown to be inactive, we next examined whether these mutant enzymes still have binding affinity with donor substrate, CMP-NeuAc. CDP-hexanolamine, which is chemically similar to CMP-NeuAc, has been used extensively to purify sialyltransferase (46) and has also been used to determine the affinity of ST6Gal I dimer for CMP-NeuAc (45). We used a CDP-hexanolamine-Sepharose column to determine whether these mutants differed in their affinity for this matrix. The medium containing ST8Sia II-H348K and ST8Sia IV-H331K protein from COS cells was applied to the column, and unbound material was removed by extensive washing. Specifically bound material was then eluted by 5 mM CDP. IgG-Sepharose resin was added to both wash-through and eluted fractions to pull down enzyme proteins, and then each bead suspension was washed and analyzed by SDS-polyacrylamide gel electrophoresis and with a radioimage analyzer. Both ST8Sia II-H348K and ST8Sia IV-H331K mutants were detected in the eluted fraction as well as ST8Sia II and IV, suggesting that these mutant proteins have affinity for CDP-hexanolamine (Fig. 6). Therefore, it seems that the lack of catalytic activity of ST8Sia II-H348K and ST8Sia IV-H331K mutants was not due to their inability to bind CMP-NeuAc. Due to technical problems, it was not possible to determine whether these mutants could bind to the acceptor substrate, NCAM. However, the mutations did not appear to affect binding to NCAM for the following reasons. (i) A recent report suggested that the amino-terminal region of the active domain of ST8Sia IV is involved in its NCAM recognition (49). (ii) Mutation in sialyl motif VS of ST6GalNAc, whose substrate specificity is different from that of polysialyltransferases, also caused the loss of catalytic activity at all as shown in Fig. 5C.
Analysis of 14 C-Polysialylated Reaction Products by ST8Sia II and IV-It has been reported that [ 14 C]polysialic acid resi-dues on NCAM synthesized by ST8Sia IV are longer than those synthesized by ST8Sia II as determined from the migration patterns of 14 C-polysialylated Fc-NCAM on SDS-polyacrylamide gels (cf. Fig. 1). To obtain more direct evidence, we performed PNGase F digestion of wild-type Fc-NCAM products 14 C-polysialylated by ST8Sia II and IV to obtain free N-linked glycan chains. These samples were then subjected to Mono-Q anion exchange HPLC analysis. As shown in Fig. 7, A and B, [ 14 C]polySia chains derived from Fc-NCAM product synthesized by ST8Sia II were indeed shorter than those by ST8Sia IV as reported previously (31). Nevertheless, both [ 14 C]polySia chains synthesized by ST8Sia II and IV were much longer than those previously reported for mutated Fc-NCAM, and these [ 14 C]polySia chains eluted by Mono-Q HPLC chromatography had peaks at DP 40 for ST8Sia II and 100 for ST8Sia IV, respectively. Fractions released from protein A-Sepharose by PNGase F digestion did not contain any detectable amounts of glycopeptides due to possible contaminating protease activities in PNGase enzyme preparations (data not shown). Unlike polysialylated NCAM, polysialylated ST8Sia II and IV have not yet been characterized biochemically. Next, using the same procedure, we analyzed the length of polySia of ST8Sia II and IV themselves. Even though previous reports showed that the presence of polysialylation of ST8Sia II and IV themselves (37)(38)(39), our biochemical analysis indicated that 14 C-disialylated component was the major product for both 14 C-sialylated ST8Sia II and IV, even though 14 C-polysialylated components were also observed (Fig. 7, C and D). We performed further analysis to detect [ 14 C]disialic and [ 14 C]oligosialic acid directly from the 14 C-sialylated ST8Sia II product. 14 C-Sialylated ST8Sia II adsorbed to IgG-Sepharose was subjected to controlled mild acid hydrolysis in 50 mM sodium acetate buffer (pH 4.8) at 37°C for 4 days. The method has previously been established to detect sialic acid dimer and oligomers (43,50). After the treatment, more than 85% of 14 C-radioactive components were released from Sepharose beads, and these were identified as [ 14 C]sialic acid dimer as major component, and [ 14 C]sialic acid oligomers (DP Ͼ 2) by thin layer chromatography (Fig. 8A) and HPLC analysis (Fig. 8B). Small amounts of [ 14 C]sialic acid monomer were detected as minor components (4.7%). These results confirmed that the frontal peaks in Fig. 7, C and D, which represent sialyl oligosaccharide released by PNGase F, are indeed disialylated and polysialylated polysaccharides. Our results suggested that ST8Sia II and IV themselves are not preferential acceptor substrates as compared with NCAM and that both enzymes behaved like ST8Sia III to such substrates. A possible mechanism for disialylation of the enzyme itself instead of polysialylation is discussed below. Again, [ 14 C]polySia chains in ST8Sia IV were found to be somewhat longer than those in ST8Sia II.

DISCUSSION
The present study demonstrated similar enzymatic properties between ST8Sia II and IV. Both enzymes showed similar pH dependence for their activities and similar K m values for NCAM substrate. Furthermore, we showed that both enzymes required no metal ions for their activities as shown in Table I. Even though our results were different from those reported previously showing that ST8Sia II was activated by Mn 2ϩ and Mg 2ϩ , whereas ST8Sia IV was activated by Ca 2ϩ (26), another result in which 1 mM EDTA showed little effect on ST8Sia II or IV activities confirmed the lack of a metal ion requirement for these enzymes.
PolySia chains synthesized by ST8Sia IV were longer than those by ST8Sia II, as suggested previously by Fukuda and colleagues (31). Nevertheless, information about wild-type NCAM polysialylation was lacking, because they used mutant NCAM proteins in which one or more than one N-linked glycosylation site were mutated to examine polysialylation by ST8Sia II and ST8Sia IV. In this study, we used wild-type NCAM as a substrate for ST8Sia II and ST8Sia IV, and then we analyzed the polysialylated compounds of reaction products using Mono-Q HPLC chromatography. We showed that polysialylated compounds synthesized by ST8Sia IV had a peak at DP 100, whereas polysialylated compounds produced by ST8Sia II had a peak at DP 40. Both of these peaks were eluted at higher DP than those in case of mutant NCAMs.
We also analyzed the polysialylation of ST8Sia II and ST8Sia IV themselves when NCAM was not added as a substrate. Interestingly, disialylated compound was the major product for both polysialylated enzymes. This result suggested that neither ST8Sia II nor ST8Sia IV are preferred substrates as compared with Fc-NCAM and that they catalyze ␣2,8-monosialylation of such substrates similarly to ST8Sia III. It has been suggested that "initiase," ␣2,8-sialyltransferase activity, is necessary for subsequent ␣2,8-polysialylation by "polymerase," which is believed to act on ␣2,8-linked sialic acid residues. This hypothesis was based on the finding that during trout oogenesis, disialylated O-linked glycan chains were always detected together with polysialylated O-linked glycan chains of polysialoglycoproteins (47). In contrast, accumulation of disialylated structures has not been detected in polysialylated NCAM. In this study, when we analyzed the polysialylation of ST8Sia II and ST8Sia IV enzyme themselves, we found large amounts of disialylated structures in addition to small amounts of polysialylated structures. We did not detect significant amounts of oligosialylated structures except disialylated structures. Based on this finding, we proposed the following model, which is illustrated in Fig. 9. ST8Sia II and ST8Sia IV can form enzymesubstrate complex with ␣2,3-linked N-glycosylated proteins with a relatively broad specificity. (a) If stable enzyme-sub-FIG. 8. Mono-Q HPLC analysis of the products of mild acid hydrolysis of 14 C-polysialylated ST8Sia II. ST8Sia II that was fused with IgG binding domain of protein A was 14 C-sialylated under polysialylation conditions. The reaction product that was adsorbed to IgG-Sepharose was subjected to controlled mild acid hydrolysis in 50 mM sodium acetate buffer (pH 4.8) at 37°C for 4 days. A, mild acid hydrolysates of the sample (S) were subjected to TLC analysis and visualized with BAS2000 radioimage analyzer. A series of sialyl oligomers were also carried out and visualized with the resorcinol method. Positions of NeuAc mono-, di-, and trimers were indicated by arrows. B, the sample was also analyzed by HPLC on a Mono-Q column. Eluted positions of NeuAc mono-, di-, tri-, and tetramers were indicated by arrows.
FIG. 9. Proposed model of polysialylation or disialylation by ST8Sia II and ST8Sia IV. Polysialic acid synthases such as ST8Sia II and IV form enzyme-substrate complexes with ␣2,3-sialylated substrates. A, if acceptor substrate is preferred by these enzymes (e.g. NCAM), stable enzyme-substrate complex produces polysialylation. B, if the substrate is not preferred by the enzymes (e.g. ST8Sia II and ST8Sia IV themselves), the complex is somehow broken after ␣2,8monosialylation and therefore no more polysialylation occurs. In this case the disialylated structure is the major product. strate complex is formed (e.g. NCAM is utilized as a substrate), subsequent polysialylation occurs. (b) If substrates are not preferred by these enzymes (e.g. enzyme themselves are utilized as substrates), enzyme-substrate complex somehow becomes unstable after a ␣2,8-monosialylation, and this complex would be easily broken resulting in termination of polysialylation, and therefore disialylated structures would be the major product. At present, it is not clear if the peptide moiety and/or specific carbohydrate structures such as ␣1,6-linked fucose residue attached to di-N-chitobiose of N-linked glycan chain on NCAM are preferentially recognized by these enzymes. The recently identified disialylated structure in the N-linked glycoprotein of porcine embryonic brain may be ␣2,8-monosialylated by ST8Sia II or IV, both of which are highly expressed in the embryonic brain (51,52).
In this study, we also demonstrated an indispensable role of the His residue in sialyl motif VS for the catalytic activity of ST8Sia II and ST8Sia IV. Sialyl motif VS that is located at the carboxyl terminus of sialyltransferases has two amino acid residues, Glu and His, that are conserved among all mammalian sialyltransferases characterized to date. It seems that catalytic amino acid residues are located in neither sialyl motif L nor sialyl motif S. Although Geremia et al. (34) suggested that the conserved Glu residue in sialyl motif VS might be involved in catalysis, we postulated that the His residue could be a catalytic amino acid residue based on the similarity between optimum pH of ST8Sia II and ST8Sia IV enzymes and the pK value of the His residue. When we changed the His residue to a Lys residue, no catalytic activity was observed for both ST8Sia II and ST8Sia IV, although they were folded correctly and secreted, and they showed affinity for CMP-NeuAc. It seems that this His residue in sialyl motif VS has a critical role in the catalytic activity of sialyltransferases, because other sialyltransferases such as ST6GalNAc IV also lost its enzyme activity following mutation of this His. It is necessary to determine three-dimensional structure of sialyltransferases by crystallization to confirm whether the His residue within sialyl motif VS is indeed a catalytic residue.