Biosynthesis and expression of polysialic acid on the neural cell adhesion molecule is predominantly directed by ST8Sia II/STX during in vitro neuronal differentiation.

We have reported recently that ST8Sia II/STX as well as ST8Sia IV/PST-1 is a neural cell adhesion molecule (NCAM)-specific polysialic acid (PSA) synthase (Kojima, N., Tachida, Y., Yoshida, Y., and Tsuji, S. (1996) J. Biol. Chem. 271, 19457-19463). To investigate which of two PSA synthase (ST8Sia II and IV) are involved in the biosynthesis of PSA associated with NCAM, the expressions of PSA, PSA synthase activity, and the genes of two PSA synthases during in vitro neuronal differentiation of mouse embryonal carcinoma P19 cells were determined. PSA was not expressed on undifferentiated cells (day 0) or cell aggregates (days 1-3) induced with retinoic acid. Expression of PSA began after cell aggregates had been dissociated and re-plated on a dish (day 4) and increased up to day 7. The expression of the mouse ST8Sia II gene was negligible in both undifferentiated and aggregated cells, it beginning at day 4, then dramatically increasing, and reaching the maximum level at days 6-7. On the other hand, transcription of the ST8Sia IV gene remained at a very low level throughout the entire period, a significant increase in its expression during differentiation not being observed. PSA synthase activity was not detected in undifferentiated or aggregated P19 cells, it increasing in parallel with ST8Sia II gene expression during differentiation. In addition, the cells at day 7 were stained with an anti-mouse ST8Sia II antiserum. Similar up-regulation of the ST8Sia II gene were observed during the differentiation of rat MNS-8 cells, which were derived from E-12 rat neuroepithelium of the neural tube and shown to differentiate into neurons. These results indicate that ST8Sia II predominantly directs PSA expression during neuronal differentiation rather than ST8Sia IV.

Polysialic acid (PSA 1 ) is an unusual carbohydrate, which is mainly associated with the neural cell adhesion molecule (NCAM), and modulates the homophilic adhesive properties of NCAM (1). The expression of PSA on NCAM is highly regulated during embryonic development, and the attenuation of cell interactions is important in the pathfinding and target innervation of axons and migration of neuronal cells in the brain (2)(3)(4)(5). Although the properties and functions of PSA on NCAM were investigated in detail, knowledge of the mechanism underlying PSA expression remained limited. It was shown recently, using chick embryo brain and ciliary ganglion neurons, that the level of PSA is probably regulated by the level of PSA synthetic enzyme(s) (6,7). We cloned two ␣2,8-sialyltransferases named ST8Sia II and IV from mouse, both of which can synthesize PSA on the ␣2,3-linked sialic acids of N-glycans without any initiator ␣2,8-sialyltransferase (8 -10). Mouse ST8Sia II exhibits 99.2% identity to rat STX and 56.0% identity to ST8Sia IV. On the other hand, mouse ST8Sia IV and hamster PST-1 exhibit 99.2% identity (11).
PST-1 has been shown to be involved in the biosynthesis of PSA associated with NCAM (11). On the other hand, there was no evidence of whether or not ST8Sia II/STX specifically synthesizes PSA on NCAM in vivo as well as in vitro, because PSA was synthesized on some glycoproteins, such as fetuin in vitro (9), and the transfection of the human STX gene into NCAMnegative cells also caused the expression of PSA on the cell surface (12). However, we recently demonstrated that ST8Sia II directly transferred all ␣2,8-sialic acid residues on the ␣2,3linked sialic acids of N-glycans of specific NCAM isoforms to yield PSA-NCAM and that the polysialylation did not require any initiator ␣2,8-sialyltransferase, but did depend on the carbohydrate and protein structures of NCAM (13). These results indicate that two distinct enzymes (ST8Sia II and IV) are potentially involved in the biosynthesis of PSA associated with NCAM in mammalian cells and tissues.
Northern blot analysis of mouse tissues indicated that expression of the ST8Sia II gene was restricted to the brain and was well regulated during brain development, like the expression of PSA (8), while the ST8Sia IV gene was only weakly expressed in the mouse brain (10). Developmentally regulated expression of the ST8Sia II/STX gene has been observed in not only mouse brain but also in human and rat brains (14,15). Other groups demonstrated that the PST-1 gene is expressed much more abundantly in newborn than adult brain (11,16). There is no doubt that both ST8Sia II and IV are expressed in the same tissue, i.e. the brain, at the same developmental stage (e.g. postnatal day 1). Therefore, it is important to identify the PSA synthase(s) responsible for the biosynthesis of PSA during neural differentiation and development to understand the regulation of PSA expression.
Mouse embryonal carcinoma P19 cells are multipotential stem cells, which differentiate into a variety of cell types, including neurons, and therefore are used as model cells for differentiation. In addition, P19 cells have been shown to express PSA during differentiation (17). In this study, we examined the expression of two PSA synthase genes as well as PSA expression and the activity of PSA synthase during the neural differentiation of P19 cells induced with retinoic acid. During neuronal differentiation of P19 cells, only the ST8Sia II/STX gene was up-regulated in parallel with the expression of PSA and PSA synthase activity.

EXPERIMENTAL PROCEDURES
Materials and Antibodies-The anti-mouse NCAM monoclonal antibody (mAb), H.28 (rat IgG1), was obtained from Immunotech, the anti-PSA mAb, 735 (mouse IgG) (18), was kindly provided by Dr. R. Gerady-Schahn, Institut fü r Medizinishe. Mikrobiologie, Hanover, Germany and the anti-microtubule-associated protein 2 (MAP-2) mAb, HM-2, was from Sigma. Anti-mouse ST8Sia II and anti-mouse ST8Sia IV rabbit antisera were prepared using bacterially expressed mouse ST8Sia II and IV, respectively. Endo-neuraminidase (endo N) purified from bacteriophage K1F, which only cleaves PSA, was kindly provided by Dr. F. A. Troy, University of California, Davis, CA (19). Poly-D-lysine, ␤-estradiol, and poly(2-hydroxyethylmethacrylate) were purchased from Sigma, basic fibroblast growth factor was from Boehringer Mannheim, and protein G-Sepharose was from Pharmacia Biotech Inc. The plasmid containing cDNA encoding the soluble human neural cell adhesion molecule fused with the Fc region of human IgG1 (NCAM-Fc) (20) was provided by Drs. D. L. Simmons and P. Crocker (Oxford University). The plasmid was transfected into COS-7 cells and then the cells were cultured in serum-free medium for 72 h. The medium was then collected and the NCAM-Fc was purified with a protein A-Sepharose column. The concentration of NCAM-Fc was 25 g/ml, as estimated by SDS-PAGE followed by densitometry, with bovine serum albumin as a standard.
Differentiation of Cells-Mouse teratocarcinoma P19 cells were cultured and then induced to differentiate into neuronal cells as described previously (21). At day 0, P19 cells were seeded into and aggregated in bacteriological-grade dishes in the presence of 1 M retinoic acid at the cell density of 1 ϫ 10 5 /ml. At day 3, the aggregates were trypsinized, planted into tissue culture-grade dishes, and then cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, and at day 4, the medium was changed to a serum-free medium (Opti-MEM I; Life Technologies, Inc.), followed by culturing for 3 days (from day 5 to 7) in serum-free medium to enrich the neuronal population. During the culture period, glial cells did not appear. For differentiation into muscle cells, P19 cells were induced with 0.5% (v/v) Me 2 SO, as described by Boer (22).
MNS-8 cells were cultured in Dulbecco's modified Eagle's medium/ F-12 (1:1) supplemented with 10% fetal bovine serum and 5% horse serum. For differentiation into neuronal cells, the cells were seeded into a poly(2-hydroxyethylmethacrylate)-coated dish in the presence of 20 ng/ml basic fibroblast growth factor and 1 M ␤-estradiol and cultured for 3 days. Then the cell aggregates were transferred to poly-D-lysinecoated dishes in a serum-free medium and cultured for 4 days (23).
Northern Blotting and RT-PCR-One or five g of poly(A) ϩ mRNA prepared from cells at days 0, 3, 4, 5, 6, and 7, respectively, was fractionated on a denaturing formaldehyde-agarose gel (1%) and then transferred to a nylon membrane (Nytran; Schleicher & Schuell). The full-length mouse ST8Sia II gene (9) and the full-length mouse ST8Sia IV gene (11) were radiolabeled and used as probes. As a positive control, the glyceroaldehyde-3-phosphate dehydrogenase transcript was measured in the same mRNA samples.
Immunocytochemical Study-P19 or MNS-8 cells, that had been seeded onto poly-D-lysine-coated Lab-Tek chamber slides, were fixed in PBS containing 1% paraformaldehyde for 30 min at room temperature. For PSA staining, the fixed cells were treated with 5 g/ml of the anti-PSA mAb, 735, followed by treatment with fluorescein-conjugated anti-mouse IgG. For MAP-2 staining, the fixed cells were further fixed in 95% ethanol, 5% acetic acid at Ϫ20°C for 2 min, washed with PBS, and then treated with the anti-MAP-2 antibody.
For immunoblot analysis, the cells were sonicated on ice in an extraction buffer (20 mM Tris-HCl, pH 8.0, 1% Nonidet P-40, 150 mM NaCl, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride), incubated for 15 min on ice, and centrifuged at 15,000 ϫ g for 15 min and then the protein concentration in the supernatant was measured by the BCA method (Pierce). The lysate (100 g of protein) was immunoprecipitated with the anti-NCAM mAb, H. 28, and protein G-Sepharose, followed by SDS-PAGE on a 5% gel (24). The proteins were then transferred to Zeta-Probe (Bio-Rad) filter paper. The filter was blocked for 60 min with PBS containing non-fat dry milk and then incubated with the anti-PSA mAb, 735, overnight at 4°C, followed by incubation with horseradish peroxidase-conjugated anti-mouse IgG. The filter was visualized using Konica Immunostaining HRP-1000 (Konika) PSA Synthase Assay-The cells were sonicated for 5 s in 25 mM MES, pH 6.0, and centrifuged at 3000 ϫ g for 10 min, and then the supernatant was centrifuged at 100,000 ϫ g for 30 min. The precipitate was suspended in 25 mM MES, pH 6.0, and used as the enzyme after the protein concentration had been measured. The PSA synthase activity was measured in a reaction mixture containing 0.1 mM CMP-[ 14 C]Sia (0.25 Ci), 10 mM MgCl 2 , 25 mM MES, pH 6.0, 0.5 g of NCAM-Fc, and 50 g of membrane protein, at 37°C for 4 h. After incubation, NCAM-Fc in the reaction mixture was recovered by adding protein G-Sepharose to the supernatant. The NCAM-Fc was then divided into two, one-half being treated with endo N, followed by SDS-PAGE. The radioactivity incorporated into NCAM-Fc was visualized with a BAS2000 image analyzer (Fuji Film) and counted. PSA synthase activity was estimated as the difference in radioactivity between before and after treatment with endo N (13). not differentiate into glial cells, because cells were not stained with the anti-glial fibrillary acidic protein mAb (data not shown).

PSA Was Expressed during the Neuronal Differentiation of
The expression of PSA was first observed at day 4. Almost all the MAP-2-positive neurons expressed PSA on their cell bodies and axons at day 6. On the other hand, PSA as well as MAP-2 was not expressed during the differentiation into muscle cells induced by Me 2 SO. Therefore, PSA expression was up-regulated during P19 cell differentiation into neurons. To analyze the expression of PSA associated with NCAM, cells at days 0, 3, 4, and 7 were lysed, and the resulting lysates were immunoprecipitated with the anti-NCAM mAb and then subjected to immunoblotting with the anti-PSA mAb. PSA associated with NCAM was slightly expressed on the cells at day 4 and dramatically increased at day 7 ( Fig. 2A).
The ST8Sia II Gene but Not the ST8Sia IV Gene Was Upregulated during Neuronal Differentiation of P19 -Two sialyltransferases, both of which exhibit ␣2,8-sialic acid polymerization activity in vitro and have been shown to be directly involved in PSA synthesis in vivo, have been cloned, namely ST8Sia II/STX and ST8Sia IV/PST-1 (8 -11, 13, 15). In order to determine which of the two enzymes is involved in the PSA expression during the neuronal differentiation of P19 cells, mRNAs were prepared from cells at days 0, 3, 4, 5, 6, and 7, and the expression of the ST8Sia II and IV genes was analyzed by Northern blotting and the RT-PCR method. As shown in Fig.  2B, the ST8Sia II and IV genes were both expressed at very low levels at days 0 and 3, at which time the PSA expression was negligible. ST8Sia II transcripts began to appear at day 4, then dramatically increased, and reached a maximum level at days 6 -7. Throughout the period of cell differentiation, the transcriptional profile of the ST8Sia II gene was similar to the expression profile of PSA associated with NCAM. The signals for the ST8Sia IV gene were hardly detected if 1 g of mRNA was used for analysis, but clearly detected when 5 g of mRNA was applied. The ST8Sia IV transcripts remained at a low level at days 4 -7, a significant increase in the gene expression during differentiation not being observed. The lower expression of the ST8Sia IV gene than that of the ST8Sia II gene was confirmed by the RT-PCR method (Fig. 3). When primers, which gave almost the same intensities of amplified products for both the ST8Sia II and IV genes, were used, the signals for the ST8Sia IV gene at days 4, 6, and 7 were hardly detected on 26 cycles of PCR, but those for the ST8Sia II gene were clearly observed in an increasing manner. With 30 cycles of PCR, the signals for the ST8Sia IV gene were detected, but no significant increase in the signals between days 0 and 7 was seen.
Expression of PSA Synthase Activity Is Correlated with Expression of the ST8Sia II Gene but Not the ST8Sia IV Gene during P 19 Cell Differentiation-The relative gene expression levels, which were normalized as to transcription of the glyceroaldehyde-3-phosphate dehydrogenase gene, and the PSA synthase activity of the cells are shown in Fig. 4. PSA synthase activity increased about 50-fold between days 3 and 6, in which time the ST8Sia II gene expression also increased about 60fold. In contrast, the increase in the ST8Sia IV gene expression was only slight (1.5-2-fold) at day 6, and the expression of the ST8Sia IV gene at days 0, 4, and 7 was almost the same. In addition, the cells at day 7 were stained with an anti-mouse ST8Sia II antiserum, but not with anti-mouse ST8Sia IV (Fig.  4C). Thus, the gene expression of ST8Sia II, but not that of ST8Sia IV, was correlated with the PSA synthase activity in cell lysates, like PSA expression. We then examined the requirement of cations for the PSA synthase activity in P19 cells, because our recent results indicated that recombinant ST8Sia II was activated by Mn 2ϩ and Mg 2ϩ , whereas recombinant ST8Sia IV was activated by Ca 2ϩ (13). As shown in Fig. 5, the activity of PSA synthase observed in differentiated P19 cells was clearly enhanced in the presence of Mg 2ϩ and Mn 2ϩ but not Ca 2ϩ , like that of recombinant ST8Sia II.
Up-regulation of the ST8Sia II Gene during Neuronal Differentiation Was Also Observed in Rat MNS-8 Cells-Similar re-

FIG. 2. Expression of PSA-NCAM and the ST8Sia II and IV genes.
A, cells at days 0, 3, 4, and 7 were lysed, and NCAM was immunoprecipitated with anti-mouse NCAM mAb, H.28, from each lysate (100 g of protein). The immunoprecipitated materials were then subjected to SDS-PAGE, Western blotting on a Zeta-Probe, and immunostaining with anti-PSA mAb, 735, as described under "Experimental Procedures." D denotes Me 2 SO-treated cells. B, 1 or 5 g of poly(A) ϩ mRNA prepared from cells at days 0, 3, 4, 5, 6, and 7, respectively, was fractionated on a denaturing formaldehyde-agarose gel (1%) and then transferred to a nylon membrane. The full-length mouse ST8Sia II and IV genes were radiolabeled and used as probes. D denotes Me 2 SOtreated cells. GAPDH, glyceroaldehyde-3-phosphate dehydrogenase.

FIG. 3. Quantitative analysis of expression of the ST8Sia II and IV genes by RT-PCR.
A, the standard ST8Sia II or IV cDNA (1.1 kilobases), which was inserted in a vector plasmid (pCMV, 5.5 kilobases) (0.05, 0.1, 0.5, 2, and 10 pg of plasmid), was subjected to 26 cycles of PCR as described under "Experimental Procedures." The amplified products were separated by electrophoresis in a 2% agarose gel and visualized by staining with ethidium bromide. Note that the sensitivity of PCR for the ST8Sia II and IV cDNAs was almost the same, and the amounts of the amplified products were linear up to 10 pg of cDNA as a template. B, mRNAs prepared from cells at days 0, 4, 6, and 7 were converted to cDNAs and then the same amount of each cDNA (as estimated with 20 cycles of PCR for the glyceroaldehyde-3-phosphate dehydrogenase (GAPDH) gene) was subjected to 26 or 30 cycles of PCR. sults were obtained for another rat cell line, MNS-8, which was established from embryonic rat neuroepithelium by introducing the mycer fusion gene and was shown to differentiate into a variety of neural cells including neurons (23). In monolayer cells, weak expression of rat ST8Sia II was detected on RT-PCR, but expression of PSA was not detected. In aggregated cells induced with inducing agents, ST8Sia II gene expression had disappeared. After the differentiation of MNS-8 cells into neurons, expression of PSA and up-regulation of the rat ST8Sia II gene were detected (Fig. 6). During the entire period, expression of the rat ST8Sia IV gene was not observed, even with the RT-PCR method. It should be noted that the rat ST8Sia IV gene was detected in mRNAs prepared from postnatal day 1 rat brain under the same RT-PCR conditions.

DISCUSSION
The present study showed that the gene expression of one of the NCAM-specific PSA synthases, ST8Sia II/STX, was well correlated with expression of PSA and PSA synthase activity of the cells in timing and amounts during the neuronal differentiation of mouse P19 and rat MNS-8 cells, as in vitro models for neuronal differentiation. On the other hand, such correlation between the expression of ST8Sia IV gene and PSA expression was not observed during the neuronal differentiation of these cells.
We have shown recently that both recombinant ST8Sia II and IV can specifically synthesize PSA on the recombinant NCAM (13), indicating that the two enzymes could potentially be involved in the biosynthesis and expression of PSA associated with NCAM in mammalian cells, particularly neuronal cells. P19 cells are a line of multipotential stem cells derived from a mouse teratocarcinoma, and many studies have shown that induced differentiation of embryonal carcinoma cells in vitro closely resembles events occurring during mammalian embryogenesis. When P19 cells were induced to differentiate with retinoic acid, the differentiated cells expressed MAP-2, indicating they had differentiated into neuronal cells. At the time of MAP-2 expression (days 4 -7), ST8Sia II gene expression dramatically increased, at which time PSA expression and PSA synthase in the cells activity increased. In contrast, expression of the ST8Sia IV gene in differentiated P19 cells remained at the very low level, which was almost the same level as that in nondifferentiated P19 cells. In addition, expression of ST8Sia II was determined by the immunostaining with anti-ST8Sia II anti-serum. Recently, we compared the two PSA synthase activities using soluble recombinant ST8Sia II and IV and showed that ST8Sia II and IV can be distinguished enzymatically (13). ST8Sia II was activated in the presence of Mn 2ϩ or Mg 2ϩ , whereas ST8Sia IV was activated by Ca 2ϩ . The PSA synthase activity in differentiated P19 cells was clearly activated by Mn 2ϩ or Mg 2ϩ , as seen in the case of chick embryo brain (6), suggesting PSA synthase activity in differentiated P19 cells are derived from ST8Sia II rather than ST8Sia IV. These results strongly suggested that PSA expression of the in vitro neuronal differentiation of P19 cells was predominantly FIG. 4. Changes in polysialic acid synthase activity and gene expression. A, membrane fractions were prepared from cells by sonication of cells followed by centrifugation at 100,000 ϫ g, and the polysialic acid synthase activity in the membrane fractions was analyzed. Enzymatic assays were carried out as described under "Experimental Procedures." SE, three batches of cells in duplicate. B, the relative expression of the ST8Sia II (closed triangles) and IV (open triangles) genes, based on expression of the glyceroaldehyde-3-phosphate dehydrogenase (GAPDH) gene, was calculated from the radioactivity on Northern blot analysis, as shown in Fig. 2B. The data indicate the averages for two batches of cells. C, cells at day 7 were fixed, treated with 0.1% Nonidet P-40 in PBS for 15 min, and then stained with the anti-ST8Sia II rabbit antiserum (upper panel), or anti-ST8Sia IV rabbit antiserum (lower panel) (each, 1:500 dilution), followed by with fluorescein-conjugated anti-rabbit IgG. directed by the expression of ST8Sia II rather than ST8Sia IV. On the other hand, when the cells were induced to differentiate into muscle cells with Me 2 SO (22), the ST8Sia II gene as well as the ST8Sia V gene was not up-regulated, the expression remaining at a very low level.
We showed previously that expression of the ST8Sia II gene was observed at 14 embryonic days (E-14) in mouse brain, reaching a maximum at E-20, then decreasing, and almost completely disappearing by 10 postnatal days, while expression of the ST8Sia IV gene was very low throughout the development of the mouse brain (8,10). In addition, the mouse ST8Sia IV gene was expressed in lung and heart rather than brain (10). Similar ST8Sia II/STX gene expression in a developmentally regulated manner was observed in rat and human (14,15). On the other hand, other groups demonstrated that the expression of the ST8Sia IV/PST-1 gene in hamster and human was greater in embryonic than adult brain, like PSA expression, and therefore PST-1 was responsible for the biosynthesis of PSA associated with NCAM in the brain (11,16). The results that expression of the ST8Sia II gene, but not the ST8Sia IV gene, increased in parallel with the expression of PSA and PSA synthase activity during the in vitro neuronal differentiation of mouse P19 are consistent with our previous findings in mouse brain (8 -10).
Since the embryonal carcinoma cell differentiation is not exactly the same as the embryogenesis, we examined rat MNS-8 cells as the other model. MNS-8 cells are derived from the E-12 rat neuroepithelium of the neural tube (23), from which the three types of neural cells in mammalian central nervous systems, i.e. neurons, astorocytes, and oligodendrocytes, originate (25). In the case of MNS-8 cells, only the ST8Sia II gene was expressed in differentiated cells, in which PSA was expressed, while expression of the ST8Sia IV gene was negligible in differentiated and PSA-expressing cells, as indicated by the RT-PCR method. Taken together with the results obtained from P19 and MNS-8 cells, PSA associated with NCAM are predominantly synthesized by ST8Sia II, not by ST8Sia IV, during the initial stages of neuronal differentiation. To further confirm the involvements of ST8Sia II and IV in the PSA synthesis during neuronal differentiation and brain development, experiments on the effects of blocking of their mRNAs are required. Studies along these lines are currently in progress.