Overexpression of Murine Pax3 Increases NCAM Polysialylation in a Human Medulloblastoma Cell Line*

Polysialic acid (PSA) is a developmentally regulated carbohydrate found primarily on neural cell adhesion molecules (NCAM) in embryonic tissues. The majority of NCAM in adult tissues lacks this unique carbohydrate, but polysialylated NCAM (PSA-NCAM) is present in adult brain regions where neural regeneration persists and in some pediatric brain tumors such as medulloblastoma, which show greater propensity for leptomeningeal spread. Pax3, a developmentally regulated paired homeodomain transcription factor, is thought to be involved in the regulation of neural cell adhesion molecules. Overexpression of murine Pax3 into a human medulloblastoma cell line (DAOY) resulted in an increase in NCAM polysialylation and a 2–4-fold increase in α2,8-polysialyltransferase type II mRNA levels. No difference was observed in α2,8-polysialyltransferase type IV message. The addition of PSA to NCAM changed the adhesive behavior of these Pax3 transfectants. Transfectants expressing high PSA-NCAM show much less NCAM-dependent aggregation than those with less PSA-NCAM. In addition, Pax3 transfectants having high PSA-NCAM show heterophilic adhesion involving polysialic acid to heparan sulfate proteoglycan and agrin. These observations suggest that a developmentally regulated transcription factor, Pax3, could affect NCAM polysialylation and subsequently cell-cell and cell-substratum interaction.

Polysialic acid (PSA) 1 is a large negatively charged linear homopolymer of ␣2,8-sialic acid residues mainly associated with neural cell adhesion molecule (NCAM). The kinetics of the homophilic adhesion mediated by NCAM correlate inversely with the degree of NCAM glycosylation and specifically with differences in the amount of ␣2,8-linked polysialic acid (1)(2)(3)(4)(5). Heterophilic binding of NCAM to extracellular matrix proteoglycans, on the other hand, appears to increase when PSA is present on NCAM (6). This suggests that PSA on NCAM promote cell migration. Migration of neural crest cells is critical for the developing embryo, failure of which causes neural tube defects. Studies done on brain tumors indicate that PSA may be a critical factor in facilitating neuroinvasive tumor metastasis in the brain (2). Whereas adult brain typically lacks this unique carbohydrate, medulloblastoma, neuroblastoma, and alveolar rhabdomyosarcoma are characterized by highly polysialylated NCAM (7)(8)(9)(10).
Pax3 is developmental transcription factor that may play a role in regulating cell adhesion molecules (11) and in oncogenesis (12,13). Neale and Trasler (14) detected abnormal sialylation of NCAM in Splotch mice, which exhibit a neural tube defect. These authors suggested that Pax3 might be involved in modifying NCAM post-translationally, thus affecting cell adhesion. Gain-of-function mutations in Pax3 have been shown to cause cancer in human cell lines (15). A chromosomal translocation of PAX3 (the human homolog of mouse Pax3) is implicated in the generation of pediatric solid tumor alveolar rhabdomyosarcoma (15)(16)(17). A chimeric transcription factor PAX3-FKHR, produced by this t(2;13)(q35;q14) chromosomal translocation in alveolar rhabdomyosarcoma binds to the NCAM promoter through its Pax3 homeodomain recognition helix. Although polysialylated NCAM isoform is expressed in rhabdomyosarcoma (10), the levels of NCAM are unaffected in alveolar rhabdomyosarcoma (18). Therefore, it seems plausible to suggest that PAX3/FKHR may not regulate the NCAM gene but could regulate NCAM polysialylation.
Although it is suggested that Pax3 may be involved in the post-translational modification of NCAM, the mechanism for this modification has not been identified. One possibility is that Pax3 may influence PSA levels on NCAM by altering the levels or activity of polysialyltransferase enzymes responsible for the synthesis of PSA chains. The primary objective of this study was to evaluate the role of Pax3 in NCAM polysialylation. In order to examine the effects of Pax3 on the expression of NCAM and PSA-NCAM, we transfected mouse Pax3 into a human medulloblastoma cell line, DAOY. The Pax3 transfectants showed increased NCAM polysialylation, as well as an upregulation of ␣2,8-polysialyltransferase (ST8SiaII/STX) message levels and activity. Additionally, cell adhesion was altered in the transfectants. Transfectants expressing high levels of PSA-NCAM showed lower Ca 2ϩ -independent aggregation than those with lower levels of PSA-NCAM. These transfectants also showed heterophilic adhesion, involving PSA to heparan sulfate proteoglycans (HSPG). Taken together, these observations suggest that Pax3 may affect NCAM polysialylation and as a result cell-cell and cell-substratum interaction.

EXPERIMENTAL PROCEDURES
Transfection of Pax3 cDNA into DAOY Cells-For the production of stable transfectants of mouse Pax3 in the human medulloblastoma derived cell line, DAOY, a 2.3-kb cDNA was inserted into the pcDNA3 expression vector (Invitrogen) at the EcoRI sites. The 2.3-kb Pax3 cDNA was prepared from pBH3.2, which was kindly, provided by Dr. Peter Gruss, Max Planck Institute, Göttingen, Germany (19). pBH3.2 was digested with EcoRI (Amersham Pharmacia Biotech) to remove the 2.3-kb Pax3 cDNA, this was then ligated into pcDNA3 at the EcoRI site. Restriction digests confirmed orientation of the cDNA insert. The pcDNA3/Pax3 construct, or pcDNA3 alone as the vector control, was then transfected into the cells using a cationic liposome system, DOTAP (Roche Molecular Biochemicals). Transfectants were selected by antibiotic resistance in cell medium containing 900 g/ml Geneticin G418 (Life Technologies Inc.). After 4 weeks in culture in the presence of G418, surviving colonies were tested for the presence of Pax3 mRNA.
Selection of Transfectants-Colonies were cloned by limiting dilution, cells were seeded in 96-well plates with a cell density of 1 cell/well. The cells were grown in DMEM supplemented with 10% fetal bovine serum (Life Technologies Inc.), penicillin (50 units/ml), streptomycin (50 g/ml), and L-glutamine (2 mM) supplemented with Geneticin G418 (900 mg/ ml), and incubated at 37°C in a humidified 95% air, 5% CO 2 incubator. Replicate plates of 96-well plates were made from the colonies formed from the initial cell in each well. When cells in the replicate plates were near confluence, mRNA was isolated using the Poly(A)Tract series 9600 system (Promega), following the instructions in the Promega manual. To synthesize cDNA, 5 l of mRNA from each well was added to each well in 96-well PCR reaction plates. The following were then also added to the wells in a total volume of 50 l for each PCR reaction (Perkin Elmer/Applied Biosystems, Inc.): 5 l of 10ϫ TaqMan buffer A (composed of 500 mM KCl, 100 mM Tris-HCl, 0.1 M EDTA, 600 nM passive reference dye, pH 8.3, at room temperature), 10 l of 25 mM MgCl 2 , 1.5 l of each dNTP (10 mM), 0.5 l of forward and reverse primers (10 M), 1 l of appropriate TaqMan probe (5000 nM), 0.25 l of AmpliTaq Gold supplied at 5 units/l and 0.25 l of Murine Moloney leukemia virus. RT-PCR cycle parameters were 48°C for 30 min, 95°C for 15 min, followed by 40 cycles at 95°C for 15 s and 59°C for 1 min. The primers and the probes used in the study were designed using Primer Express software (Perkin Elmer/Applied Biosystems Inc.) and synthesized by Life Technologies Inc., and are shown in Table I. Clones with a threshold cycle (C T ) value of 30 or less were selected as positive for Pax3. A definition of C T value is described in the section below, "Real-time Quantitative RT-PCR." Real-time Quantitative RT-PCR-The TaqMan probe consists of a specific oligonucleotide with both a 5Ј-reporter dye and 3Ј-quencher dye. The fluorescent reporter dye such as 6-carboxylfluorescein, is covalently linked to the 5Ј end of the oligonucleotide. The reporter is quenched by 6-carboxytetramethylrhodamine, typically located at the 3Ј end. The 5Ј nucleotidase activity of Taq DNA polymerase cleaves the probe during the PCR cycle when it hybridizes to the template. This results in an increase of reporter fluorescence signal that is detected during each cycle by the 7700 Sequence Detector System (Perkin Elmer/Applied Biosystems Inc.). Quantitation of the sample is then based on the cycle when the amplicon is first detected. A parameter, threshold cycle (C T ) is defined for each PCR reaction, which is the cycle number at which the reporter fluorescence generated by the cleavage of the sequence specific probes passes above a fixed baseline. This C T value has been shown by Higuchi et al. (20) to be a straight line when the log of the initial template is plotted versus C T .
Western Blotting-Pax3 transfectants, selected by the above method, were seeded into 75-cm 2 flasks containing DMEM/10% fetal bovine serum and Geneticin (200 g/ml). When cells were near confluence, monolayers were washed twice with cold PBS, and solubilized by the addition of 200 l of lysis buffer (50 mM Hepes, pH 7.4, 150 mM NaCl, 100 mM NaF, 1 mM MgCl 2 , 1.5 mM EGTA, and 1% Triton X-100). A separate aliquot of 25 g of protein as determined by the method of Bradford (21) was treated with 0.05 units of type X neuraminidase (Sigma) and incubated at 37°C for 1 h. Western blots were performed according to Towbin et al. (22). 25 g of protein was applied to a 10% polyacrylamide gel and subjected to electrophoresis and protein electrotransferred to an Immobilon P membrane. Incubating the membrane in 5% nonfat dry milk for 1 h at room temperature blocked nonspecific sites. The membrane was then incubated with respective antibodies at a dilution of 1:500: anti-NCAM, 5B8 (Developmental Study Hybridoma Bank), and anti-PSA (Sigma), and anti-MyoD (Santa Cruz Biotechnology Inc.) for 1 h at room temperature. After three washes in PBS containing 0.2% Tween 20, the membrane was incubated for 1 h at room temperature in the presence of 2 mg/ml peroxidase-coupled anti-mouse secondary antibodies. After additional washes, antibody binding was revealed by enhanced chemiluminescence (ECL) detection (Amersham Pharmacia Biotech).
Measurement of Total ␣2,8-Polysialyltransferase Activity-Polysialyltransferase activity was assessed as described by Sevigny et al. (23). Pax3 transfectants, pcDNA3 vector controls, and nontransfected DAOY cells were seeded in 75-cm 2 flasks in DMEM/10% fetal calf serum. After attaining 75% confluence, the cells were washed with PBS and then homogenized in homogenizing buffer (10% glycerol, 50 mM MES, pH 6.1, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 4.6 units/liter aprotinin, 40 g/ml leupeptin, 1 g/ml pepstatin A) in a Kontes glass homogenizer (20 strokes). The homogenates were centrifuged for 10 min at 1000 ϫ g and the supernatant removed and centrifuged for 30 min at 150,000 ϫ g. To enhance the polysialyltransferase activity relative to the monosialyltransferase activity, the pellet was subjected to two freeze-thaw cycles as described previously (23) and then resuspended in the homogenizing buffer supplemented with glycerol to 30% to preserve the ␣2,8-polysialyltransferase enzyme activity (23).
The enzyme activity then was measured as follows. Each incubation mixture contained the following components in 350 l: 1 mg of polysialyltransferase-enriched membrane fraction (P2), 10 mM MnCl 2 , 200 M CMP-[ 14 C]Neu5Ac (1.11 ϫ 10 6 dpm), 1 mM dithiothreitol, 50 mM MES buffer, pH 6.1. This buffer and pH maximized the ratio of polysialyltransferase to monosialyltransferase activities. After incubation for 3 h at 33°C, the reaction was terminated by the addition of EDTA to a final concentration of 50 mM. Each sample was then centrifuged for 30 min at 150,000 ϫ g at 4°C, to remove excess substrate. The pellets were resuspended in 100 l of 20 mM Tris-HCl, pH 7.4, and divided into two equal aliquots. Endo-N-acylneuraminidase (EndoN) was a generous gift from Dr. Fredrick Troy (Department of Biological Chemistry, University of California School of Medicine, Davis, CA). EndoN, a bacteriophage-derived endosialidase that specifically cleaves ␣2,8-linked polysialic acid chains, was added to one of the two split samples, and both samples were incubated at 37°C for 1 h (23). Each sample was subsequently spotted on Whatman 3MM paper. Descending paper chromatography was carried out overnight with ethanol and 1 M ammonium acetate, pH 7.5, in a ratio of 7:3. The papers were then allowed to dry. A 12 ϫ 1-inch 2 portion encompassing each origin was cut out from the paper, placed in vials containing scintillation fluid and quantitated in a Beckman scintillation counter. The difference in the counts between control and EndoN-treated samples were taken as a measure of the amount of sialic acid that had been enzymatically incorporated into ␣2,8-linked polysialic acid chains.
Cell Aggregation Assays-To evaluate aggregation in a Ca 2ϩ -independent manner, transfectants and parental DAOY cells were plated 24 h prior to the assay, so that the cells would be in their exponential growth phase when the assay was performed. Half of the dishes were then treated with EndoN (0.5 unit/ml) for 1 h at 37°C, and the other non-treated dishes were used as a control. For both conditions, the cells were washed with Hepes-buffered saline (HBS), containing 100 mM NaCl, 20 mM Hepes, pH 7.2, and then incubated with HBS containing 3 mM EDTA for 10 min at 37°C. Cells were collected by centrifugation at 500 rpm (800 ϫ g) for 5 min at room temperature. The cells were then washed in HBS/EDTA and collected as above, after which they were suspended in RPMI/10% fetal calf serum (2.5 ϫ 10 5 cells/ml). The viability of cells was ascertained by trypan blue dye exclusion. 200-l aliquots from EndoN-treated and untreated cells were incubated in plastic tubes at 37°C for 40 min, followed by four cycles of gentle up-down rotation to ensure sampling from a homogeneous population (24). Single cells, obtained by passing cells 10 times through a 25-gauge needle (25), were counted with a hemocytometer before and after allowing cells to aggregate at 37°C. Single cells were defined as cells that did not share a common border with neighboring cells (26). Particles that measured less than 4.5 m in diameter were excluded from evaluation.
Cell-substratum Adhesion Assays-The cell-substrate adhesion assays were performed essentially as described by Storms and Rutishauser (6). 0.25-ml aliquots of nitrocellulose solution (5-cm 2 nitrocellulose in 6 ml of methanol) were applied to each well in a Costar 12-well dish (Costar 3512) and allowed to air-dry for 30 min. The 2 l (0.25 l/l) of substrate solution (0.5 g/spot) was applied as a spot to the nitrocellulose and incubated for 1 h at room temperature. Each well had 3-4 spots, and the spots were labeled on the reverse side for identification. The nitrocellulose in each well was blocked with 0.5 ml of freshly prepared 1% BSA in PBS for 1 h at room temperature. To minimize nonspecific binding, the BSA solution was heat-inactivated at 56°C for 1 h and filtered through a 0.2-m Amicon filter. After blocking, the plates were washed twice with PBS.
Cells were then prepared in 0.1 mM EDTA for 5 min and gently removed from the culture dishes. Cells were then washed twice in PBS, centrifuged and resuspended in DMEM, counted, and adjusted to a final concentration of 2 ϫ 10 6 cells/ml in DMEM containing 0.1% BSA. Nearly confluent Pax3 transfectants as well as the vector and wild type control DAOY cells were treated with EndoN for 4 h. EndoN-treated and untreated cells (2 ϫ 10 6 cells/ml) were added to appropriate experimental dishes and incubated in a tissue culture incubator for 1 h. The experimental dishes were washed twice with DMEM, and the cells remaining in contact with the substrate were counted using an inverted phase-contrast microscope (magnification, ϫ200). Three 1-mm 2 fields of view were counted for each spot. Each experiment was repeated at least three times.

Expression of Pax3 Message and Functional Pax3 Protein in
Pax3 Transfectants-Pax3 message expression was confirmed in selected clones by real-time quantitative RT-PCR using the probe and primers for murine Pax3 (Table I). Amplification plots generated by the real-time quantitative RT-PCR analysis showed the presence of Pax3 in the transfected clones (Fig. 1A). Seven highly positive clones were identified in subsequent experiments. These were designated B9, E7, E11, E12, F8, G8, and H6. The threshold cycle values (C T ) of different Pax3 transfectants were normalized against ␤ actin in real-time quantitative RT-PCR. All the transfectants showed lower C T values ranging from Ϫ3.77 to Ϫ6.44 (␤ actin C T Ϫ Pax3 C T ) cycles greater than the ␤ actin (Table II). This corresponds to Pax3 expression levels of 150 -625 copies of Pax3 per 10,000 copies of ␤ actin. DAOY or vector controls showed no detectable murine Pax3 message. These observations indicated that mouse Pax3 message was expressed in transfected human DAOY cells. To investigate whether the Pax3 transfectants had functional Pax3 transcription factor, immunoblots were run with antibody against MyoD, a known downstream target of Pax3 (27). Immunoblots were performed using the seven Pax3 clones, pcDNA3 vector control, and DAOY wild type controls. The results of the immunoblots (Fig. 1) showed that MyoD was strongly expressed in six of the seven transfectants selected for our study. In one clone, H6, only very weak anti-MyoD staining was observed. Staining was also very weak to absent in the vector-transfected and DAOY wild type control cells. The presence of immunoreactive MyoD in Pax3 transfectants indicated that functional Pax3 was present in the transfected clones. A probable reason H6 does shows a very weak MyoD band could be the bifunctional transcriptional regulatory property of Pax3 (11). These authors have shown that at lower concentrations of Pax3 protein activates NCAM transcription, whereas at higher concentrations inhibited transcriptional activity. Thus, even though H6 clone does not show a strong MyoD band, Pax3 still could be active in this clone. The expression of murine Pax3 or human ␤-actin, STX, and PST were analyzed by RT-PCR reaction using the 7700 Sequence Detection System (Perkin Elmer) as described under "Experimental Procedures." The plot shows the fluorescence intensity (⌬Rn) at each PCR cycle. ⌬Rn is normalized reporter signal corrected for initial reporter signal. To calculate the ⌬Rn, the initial reporter signal is subtracted from the normalized reporter signal at each PCR cycle (20). A cycle threshold (C T ) is defined as the fractional cycle number at which the reporter fluorescence passes a fixed threshold value above baseline. Samples with higher message expression will have a lower C T value. The cycle threshold was set at 0.06 ⌬Rn for this experiment. The amplification plots for messages originating from the parental DAOY cells are shown in black lines, whereas the plots originating from clone F8 are shown with gray lines.   5Ј-TCT TAG AGA CGC AAC  CAT GGG-3Ј   5Ј-CCA ACC ATA TCC GCC  ACA A-3Ј   6 FAM-ATG GCA TTC GGC CTT  GCG TCA TTT-TAMRA   Human STX   5Ј-AGC TCT ATT AGA TTT  GCT ATG TAA GCT GTT-3Ј   5Ј-TCC TCA TCT TCG CAG  ACA TCT C-3Ј   6 FAM-TGC CTC CCG AAT TCC  CGA TTT  In the transfectants two immunoreactive bands were observed. In the pcDNA3-transfected and wild type DAOY controls, the anti-PSA immunoreactivity was observed, suggesting the presence of PSA in these cells. However, the upper immunoreactive band was not present in the controls (Fig. 2).
An increase in high molecular weight immunoreactive NCAM, as shown in Fig. 2, could be due to increased NCAM levels or additional forms of NCAM. To test this possibility anti-NCAM antibody, 5B8 (directed against the protein) was used on the immunoblots. Once again, higher molecular weight immunoreactivity was observed. After neuraminidase treatment, the antibody recognized a band at ϳ140 kDa only. The transfectants did not show any change in the form of NCAM expression. Overall NCAM levels did not appear to be significantly altered in any of the clones (Fig. 3). These results suggest that Pax3 did not alter NCAM protein type or levels in the transfectants. Most likely, the increased high molecular weight immunoreactivity of anti-NCAM was due to increased levels of PSA on NCAM in the transfectants.
Pax3 Transfectants Show an Increase in ST8SiaII/STX mRNA-The increase in NCAM polysialylation observed by Western blot could be due to an increase in the levels of polysialyltransferase message. To examine if the increase in NCAM polysialylation was due to an increase in polysialyltransferase message or an increase in NCAM message, we performed realtime quantitative RT PCR using probes and primers for human ST8SiaII/STX, ST8SiaIV/PST, and NCAM (primer and probe sequences are found in Table I). The NCAM probe and the primers were chosen to recognize all three molecular mass isoforms: 120, 140, and 180 kDa. Representative amplification plots for Pax3, ␤-actin, STX, and PST expression from the parental DAOY cells and clone F8 are shown in Fig. 1. The data for all the clones are summarized in Table II. The expression of genes of interest was normalized to ␤ actin message levels. The data are expressed as C T value relative to ␤ actin (␤ actin C T Ϫ gene of interest C T ). Approximately a two-cycle difference in the relative threshold values of ST8SiaII/STX message, which corresponds to a 4-fold increase, was observed in Pax3 transfectants as compared with the vector-transfected or wild type control DAOY cells (Table II). Mann-Whitney U statistical analysis of the ST8SiaII/STX data showed this increase to be significant (p Ͻ 0.0500). There was no significant difference in the threshold cycle values of ST8SiaIV/PST message between the transfectants and the controls (p Ͻ 0.6985). There was also a trend toward a difference in the threshold values for NCAM message in the transfectants, as compared with DOAY wild type (p Ͻ 0.0528). These results suggest that NCAM message expression is increased in the Pax3 transfectants as described previously (11). In addition, these data suggest that Pax3 also influences the message levels for ST8SiaII/STX.
Pax3 Transfectants Show an Increase in Polysialyltransferase Activity-In order to determine if approximately 4-fold increase in ST8SiaII/STX message level could account for the difference in NCAM polysialylation, total polysialyltransferase activity in polysialyltransferase enzyme rich membrane preparations were measured from the clones, the wild type, and vector-transfected controls. Clones B9, E7, and H6 showed a significant increase in the polysialyltransferase activity as compared with wild type and pcDNA3-transfected control cells. Student's t test showed the p values of Ͻ0.018, Ͻ0.005, and Ͻ0.0009, respectively, when B9, E7, and H6 were compared with vector pCDNA3-transfected DAOY cells (Fig. 4). These data showed that the increase in the level of ST8SiaII/STX message resulted in measurable increase in total polysialyltransferase activity for some of the clones. Since the polysialyltransferase activity assay cannot discriminate between ST8SiaII/STX and ST8SiaIV/PST activity, the lack of measurable increase with other clones may be due to the ST8SiaIV/PST activity. The clones with measurable increase in the polysialyltransferase activity also appeared to be the clones showing the greatest increase in PSA immunoreactivity (Fig. 2).
Pax3 Transfectants Show Decrease in Cell-Cell Aggregation-Polysialic acid on NCAM is known to alter cell-cell interaction. In order to determine whether the increased polysialylation in Pax3-transfected clones altered this cellular behavior, Ca 2ϩ -independent aggregation assays were performed. Fig. 5 shows the percentage of cell-cell aggregation in the Pax3 transfectants, parental DAOY, and vector control cells, with and without EndoN treatment. EndoN cleaves off ␣2-8-linked polysialic acid from NCAM; cell aggregation in the absence of ␣2-8-linked polysialic acid is an index of the aggregation due to NCAM alone. The results from the aggregation assays show that the parental and vector-transfected controls showed about 40 -60% aggregation as compared with the Pax3-transfected clones, which showed 10 -40% aggregation. Clones B9, E7, and H6, which had the highest levels of PSA on NCAM (Western blots; see Fig. 2), as well as the highest polysialyltransferase activity had the lowest aggregation. However, when the cell surface PSA was removed by EndoN treatment, cells aggregation returned to ϳ80% in all cells. Therefore, these observations suggest that the change in aggregation was due to increased PSA in Pax3 transfectants.
Pax3 Transfectants Show Increased Heterophilic Adhesion to Heparan Sulfate Proteoglycans-Polysialic acid was previously known to be a negative regulator of cell-cell adhesion (28).  a C T value is defined as the PCR cycle at which fluorescence increase above a predetermined baseline. RT-PCR was performed, and the base line was determined as described under "Experimental Procedures." b All data were normalized against ␤ actin expression. The numbers represent the cycle difference between ␤ actin C T and the gene of interest C T (␤ actin C T Ϫ gene C T ).
c Significantly different from wild type DAOY or pcDNA3-transfected DAOY cells by Mann-Whitney U test, p Ͻ 0.0500.
Recently, Storms and Rutishauser (6) have shown that PSA may act as a positive regulator of cell-extracellular matrix interaction. In this study, spot cell adhesion assays were used to measure the interaction of Pax3 transfectants with the two extracellular matrix proteoglycans used by Storms and Rutishauser (6), agrin and HSPG (Fig. 6). Laminin was used as a control substrate, where binding should be independent of PSA content on NCAM (6). Clones B9 and H6 showed highest interaction with agrin and HSPG. This increased adhesion was eliminated by EndoN treatment, suggesting a requirement for PSA. In general the Pax3 clones were less adhesive to laminin substratum, but this adhesiveness was not affected by EndoN treatment. The PSA-dependent adhesion to agrin was determined by subtracting the endoN-treated adhesion (Fig. 6B, open bars) from the total adhesion (Fig. 6B, hashed bars). Three clones (F8, H6, and B9) showed significant PSA-dependent binding to agrin. There appeared to be a positive correlation between agrin binding and the PSA content of these clones (Fig. 6D). The total PSA on NCAM content was determined by densitometric measurement of the anti-PSA immunoblot shown in Fig. 2. Clone B9 had the highest PSA content and also showed the greatest PSA-dependent adhesion to agrin. In contrast, the parental DAOY cells showed the least PSA-dependent adhesion. These data are consistent with the observation made by Storms and Rutishauser (6) that increased PSA on NCAM promotes heterophilic adhesion of neural cell adhesion molecule with extracellular matrix molecules like agrin and HSPG.

DISCUSSION
Several studies have indicated that Pax3 may be involved in regulating NCAM expression (11) and post-translational modification (14). Our in vitro data on Pax3 transfection in DAOY cells showed that Pax3 overexpression resulted in an increase in PSA-NCAM as observed on Western blots. Using real-time quantitative RT-PCR, the message levels of NCAM, ST8SiaII/ STX, and ST8SiaIV/PST were determined in Pax3 transfectants. These data indicated an approximately 4-fold increase in the level of ST8SiaII/STX message. There was no change in ST8SiaIV/PST expression. All of the NCAM expressed in these cells appeared to be the 140-kDa isoform. In the clones that showed the greatest increase in PSA-NCAM, there was a measurable increase in total polysialyltransferase activity. Thus, the increase in NCAM polysialylation may be explained by a Pax3-dependent increase in ST8SiaII/STX message levels, resulting in increased polysialyltransferase activity. These changes in the message and activity of polysialyltransferase also appeared to be significant enough to alter NCAM-mediated cell-cell and cell-substratum adhesion.
Previously, Chalepakis et al. (11) have shown that Pax3 could regulate NCAM expression. Our data also indicate an increase in NCAM message expression in the transfectants. Ginsberg et al. (18) reported that a highly sialylated NCAM isoform was expressed in rhabdomyosarcoma (10), but the levels of NCAM itself were not changed. Rhabdomyosarcoma contains the fusion protein PAX3/FKHR transcription factor. This  protein extracts from various  transfectants, labeled B9, E7, E11, E12,  F8, G8, and H6, the vector pcDNA3-transfected control (VEC) and DAOY wild type controls (WT) were probed with anti-NCAM, anti-PSA, and anti-MyoD antibodies at 1:500 dilution. The immune complex was visualized with anti-horseradish peroxidase secondary antibody at 1:5000 dilution and using the ECL detection system as described under "Experimental Procedures." The position of molecular weight markers, either prestained or horseradish peroxidase-conjugated, are indicated on the extreme right of the figure.
FIG. 3. Neuraminidase treatment of NCAM from Pax3 transfectants. 25 g of protein extracts from various transfectants were treated with 0.05 units of type X neuraminidase (ϩ) and incubated for 1 h at 37°C. The neuraminidase-nontreated samples (Ϫ) were incubated identically without the addition of the enzyme. Immunoblots of neuraminidase-treated and nontreated samples was performed using anti-NCAM antibody, 5B8 (directed against the protein), at a dilution of 1:500. The immune complex was visualized with anti-horseradish peroxidase secondary antibody at 1:5000 dilution and using the ECL detection system as described under "Experimental Procedures." The position of pre-stained molecular weight markers is indicated on the extreme right of the figure.
fusion protein is known to regulate Pax3 downstream targets. The NCAM message levels in rhabdomyosarcoma were not significantly higher than the message levels in other tumors, but NCAM sialylation was significantly increased. From these data Ginsberg et al. (18) suggested that the PAX3/FKHR fusion protein may regulate not only NCAM but also NCAM polysialylation. Our findings are consistent with the findings of Ginsberg et al. (18) that Pax3 may have a significant impact on NCAM polysialylation.
The increased expression of ST8SiaII/STX suggests that it may be a downstream target of Pax3. Other molecules known to be regulated by Pax3 are NCAM (11), MyoD (27), c-Met (29), a component of replication licensing factor, cdc46/MCM (30), Dep-1 (31), and myelin basic protein (32). The minimal essential promoter of the STX gene (33) does not appear to contain the commonly described ATTA(N) n GTTCC Pax3 binding site. On the other hand, the DNA-binding sites for Pax3 do not appear to be the same in all Pax3 downstream target genes (34). It is premature to assume that Pax3 may not bind directly to the ST8SiaII/STX promoter. Alternatively, it is possible that Pax3 may bind to the promoter elements of other transcription factor, such as the MyoD, and then the MyoD transcription factor may bind to MyoD response elements in the STX promoter. The minimal essential promoter of ST8SiaII/ STX (33) does not contain the MyoD response elements either. There is, however, one MyoD binding site upstream, Ϫ448 to Ϫ438 (CGTCAACTGCTGAAG), of the minimal essential promoter region, suggesting that it may play a role in ST8SiaII/ STX expression.
ST8SiaII/STX and ST8SiaIV/PST play different roles in the biosynthesis and expression of PSA. Using various glycosylation site mutants of NCAM, Angata et al. (35) reported that ST8SiaIV/PST preferred the sixth N-glycosylation site (Asn-478), which is closer to the transmembrane domain, over the fifth site (Asn-449). These observations suggested that ST8SiaIV/PST preferred the sixth site, whereas ST8SiaII/STX added polysialic acid on the fifth as well as on the sixth site.
The size of the polysialic acid chain synthesized by ST8SiaII/ STX was smaller than that synthesized by ST8SiaIV/PST, even though the total amount of polysialic acid synthesized by ST8SiaII/STX was comparable to that synthesized by ST8SiaIV/PST (35). Preferential increase in the message of ST8SiaII/STX in Pax3 transfectants could mean increased polysialylation on the fifth as well as sixth glycosylation site. This may explain the appearance of the upper and lower bands in the Pax3 transfectants. The upper band may represent PSA chains at both fifth and sixth glycosylation sites, whereas in the lower band only one of these sites is used. Interestingly, the presence of both upper and lower bands correlates with increased adhesion to agrin and HSPG substrates (clones B9 and H6).
The significance of these findings is that Pax3 may regulate ST8SiaII/STX, which confers polysialic acid on NCAM. Mouse Pax3 is expressed with a distinct spatiotemporal pattern beginning between day embryonic day 8 (E8) and embryonic day 9.5 (E9.5). Splotch phenotype results from the failure or partial failure of Pax3 to execute a specific genetic program (36). NCAM expression starts around E11 in mouse embryos, and reaches its peak around E18, whereas ST8SiaII/STX expression starts around E14 and reaches its peak around E20 (37,38). ST8SiaIV/PST expression is fairly constant in the developing embryo. Using multivariate statistical analysis on the expression of Pax3, NCAM, and other genes, Bennett et al. (37) suggested that these genes do interact during those gestation times and that this is crucial for the migration of neural crest Aggregation assays of Pax3 transfectants were performed as described under "Experimental Procedures." Where indicated cells were treated with EndoN (0.5 unit/ml) for 1 h at 37°C. Single cells, obtained by passing cells 10 times through a 25-gauge needle, with or without EndoN treatment, were counted with a hemocytometer before and after allowing cells to aggregate at 37°C for 1 h. The number of single cells remaining after allowing for 1 h of aggregation was used as an index for the percentage of aggregated cells. The percentage of aggregated cells after 1 h of aggregation was calculated using the number of single cells before the aggregation assay as 100%. Single cells were defined as cells that did not share a common border with neighboring cells (26). Particles that measured less than 4.5 m in diameter were excluded from evaluation. Mock-1 and mock-2 represent two different pcDNA3-transfected DAOY clones. Pax3-transfected cells are indicated by their clone number. Each bar represents the mean Ϯ S.E. percentage aggregation of four separate experiments done in duplicate. Open bars represent aggregation after EndoN treatment, and hatched bars represent aggregation without EndoN treatment. cells. Our present study provides experimental data that are consistent with their analysis.
Not only are Pax3, NCAM, and ST8SiaII/STX involved in embryonic development, they also appear to be involved in oncogenesis and may be predictors of clinical outcomes of pediatric brain tumors. Overexpression or aberrant regulation of the Pax3 gene can transform fibroblasts into tumor cells in nude mice (12,13), suggesting that this gene may act as a proto-oncogene. Changes in the amount of polysialic acid on NCAM are related to invasive and metastatic growth potential of human tumors (37,39,40) and linked to the modulation of adhesive properties of NCAM. Highly polysialylated NCAM is a cell surface marker for medulloblastoma (41)(42)(43), neuroblastoma, and rhabdomyosarcoma (42)(43)(44). This polysialylation may give the tumor the propensity for leptomeningeal spread to other areas of the brain and adversely affect tumor outcomes, because the polysialylated NCAM on tumor cell has less cell-cell adhesion and more cell-extracellular matrix proteoglycan adhesion.
In sum, the present findings demonstrate that Pax3 overexpression in DAOY cells resulted in an increase in NCAM polysialylation. Increased polysialylation was due to an increase in the levels of ST8SiaII/STX message and subsequently in the activity of polysialyltransferase enzyme. Increased sialylation of NCAM caused a decrease in cell-cell aggregation and an increase in heterophilic binding to extracellular matrix proteoglycans. It is significant that Pax3 regulates the function of NCAM by regulating a developmentally regulated enzyme, FIG. 6. Adhesion of DAOY and Pax3 transfectants to extracellular matrix components. Cell-extracellular matrix adhesion assays were performed as described under "Experimental Procedures." EndoN-treated cells (open bars) were compared with untreated cells (shaded bars ), to determine the contribution of PSA to cell adhesion with extracellular matrix components. Panel A show adhesion to HSPG, panel B shows adhesion to agrin, and panel C shows adhesion to laminin. Panel D shows dose dependence of polysialic acid on heterophilic adhesion to agrin in select clones that showed significant binding to agrin, as seen in panel B. Densitometric measurements of the PSA-specific immunopositive bands from the DAOY cells, F8, H6, and B9 clones (from Fig. 2) were recorded in measured percent transmission and calculated optical density derived the formula OD ϭ Ϫlog%T/100% where OD is optical density and %T is measured percentage of transmission. Images of the gel were captured by a Zeiss S.E.-IPS image processing system with a Data Series 68 camera. Arbitrary optical density units of PSA was plotted against PSA specific binding (obtained by subtracting the ϩEndoN value from ϪEndoN value from panel B). BSA was used as a negative control for all substrates, but data are not included in the figure. Cellular binding to BSA, when it was used, was never more than 6 -10 cells/mm 2 . The cells remaining in contact with the substrate were counted using an inverted phase-contrast microscope (original magnification, ϫ200). Three 1-mm 2 fields of view were counted for each spot. The bars shown are the average number of cells/mm 2 Ϯ S.E. of 4 -7 different spots from three independent experiments. Cells transfected with vector alone are designated as pcDNA, and Pax3-transfected cells are identified by their clone number.
ST8SiaII/STX, that confers polysialic acid on NCAM. Improper expression or a mutation in Pax3 could adversely affect NCAM expression and polysialylation. This may partially explain the defects in Pax3 homozygous embryos. It may also affect tumor outcomes in medulloblastoma (41)(42)(43).