Down-regulation of Polysialic Acid Is Required for Efficient Myelin Formation*

Oligodendrocyte precursor cells modify the neural cell adhesion molecule (NCAM) by the attachment of polysialic acid (PSA). Upon further differentiation into mature myelinating oligodendrocytes, however, oligodendrocyte precursor cells down-regulate PSA synthesis. In order to address the question of whether this down-regulation is a necessary prerequisite for the myelination process, transgenic mice expressing the polysialyltransferase ST8SiaIV under the control of the proteolipid protein promoter were generated. In these mice, postnatal down-regulation of PSA in oligodendrocytes was abolished. Most NCAM-120, the characteristic NCAM isoform in oligodendrocytes, carried PSA in the transgenic mice at all stages of postnatal development. Polysialylated NCAM-120 partially co-localized with myelin basic protein and was present in purified myelin. The permanent expression of PSA-NCAM in oligodendrocytes led to a reduced myelin content in the forebrains of transgenic mice during the period of active myelination and in the adult animal. In situ hybridizations indicated a significant decrease in the number of mature oligodendrocytes in the forebrain. Thus, down-regulation of PSA during oligodendrocyte differentiation is a prerequisite for efficient myelination by mature oligodendrocytes. Furthermore, myelin of transgenic mice exhibited structural abnormalities like redundant myelin and axonal degeneration, indicating that the down-regulation of PSA is also necessary for myelin maintenance.

Polysialic acid (PSA), 3 a linear homopolymer of N-acetylneuraminic acid, is a posttranslational modification of the neural cell adhesion molecule (NCAM) (1). NCAM occurs in vari-ous isoforms that are generated by alternative splicing (2). The three major isoforms of NCAM are the two transmembrane proteins NCAM-180 and NCAM-140 and the glycosylphosphatidylinositol-linked NCAM-120. The latter is the predominant isoform in oligodendrocytes (3). PSA is synthesized in the Golgi apparatus by two sialyltransferases, ST8SiaIV (4,5) and ST8SiaII (6,7). Although in the mammalian brain, polysialylated NCAM (PSA-NCAM) is mainly found in neurons, other cell types within the central nervous system synthesize PSA-NCAM as well. Migrating oligodendrocyte precursor cells (OPC) express PSA, which is, however, down-regulated after these cells reach their final destination and start to differentiate into mature myelinating oligodendrocytes (8 -10). The presence of PSA seems to be functionally important, since migration of OPC is impaired after endoneuraminidase treatment, which specifically hydrolyzes PSA (11). More recent data suggest that PSA is not required for cellular motility in general but for directional movement of OPC in response to platelet-derived growth factor (PDGF) (12). In addition to its effect on migration, PSA seems also to inhibit differentiation of OPC into mature myelinating oligodendrocytes, because removal of PSA by endoneuraminidases accelerates oligodendrocyte differentiation (13).
PSA is also down-regulated in neurons at the time when their axons become myelinated. Axons that remain nonmyelinated in the adult (e.g. mossy fibers in the hippocampus) in contrast remain PSA-positive (14,15). These observations together with the fact that premature removal of PSA accelerates myelin formation led to the hypothesis that PSA expressed on the axonal membrane is an inhibitor of myelination (16). This is supported by the observation that premature removal of PSA by endoneuraminidase increases myelin formation in vivo and in vitro (16). The re-expression of PSA in multiple sclerosis lesions (17) and hypomyelinated mice (18) might therefore be one factor inhibiting (re)myelination.
Accelerated oligodendrocyte differentiation and myelin formation by premature removal of PSA suggests but does not prove that the normal postnatal down-regulation of PSA is a prerequisite for myelination. In order to further address this question and to discriminate between PSA-dependent effects in oligodendrocytes and neurons, we have generated transgenic mice overexpressing the polysialyltransferase ST8SiaIV in neurons and oligodendrocytes. In the present study, we asked whether down-regulation of PSA in OPC is a secondary effect of the differentiation process or whether PSA down-regulation is a prerequisite for myelin formation. Using transgenic mice overexpressing the polysialyltransferase ST8SiaIV in oligoden-drocytes via the proteolipid protein (PLP) promoter, we show here that the continuous expression of PSA in mature oligodendrocytes causes hypomyelination.

EXPERIMENTAL PROCEDURES
Generation of PLP-PST Transgenic Mice-Mouse ST8SiaIV was amplified by PCR using oligonucleotides PSTsense (5Ј-GCAAGCTTCCATGCGCTCAATTAGAAAACG-3Ј) and PSTanti (5Ј-GCTCTAGATTATTGCTTCATGCACTTTCC-3Ј), digested with HindIII and XbaI (introduced restriction sites are underlined in the primer sequences), treated with Klenow enzyme (fill-in reaction), and ligated into the PmeI site of the PLP promoter cassette (19) (kindly provided by W. B. Macklin (Cleveland Clinic Foundation)) (see Fig. 1A). Correct sequence of ST8SiaIV cDNA was confirmed by DNA sequencing. In order to generate transgenic mice, vector sequences were removed from the plasmid by digesting the DNA with NotI and ApaI, followed by agarose gel purification. The promoter-cDNA-poly(A) fragment was isolated and purified using Qiaquick gel purification (Qiagen, Hilden, Germany) and Elutip purification (Schleicher & Schuell). Pronucleus injection into fertilized eggs from F2 C57BL/6ϫCBA mice was performed at the Karolinska Center for Transgene Technologies (Stockholm, Sweden). Founder mice were identified by Southern blotting using a 32 P-labeled ST8SiaIV cDNA probe as described (20). Several founder mice were obtained, from which two independent transgenic lines (tg246 and tg281) were established and maintained on a mixed C57BL/6ϫCBA genetic background. Genotyping of mice was done by PCR on genomic tail DNA using oligonucleotides PSTsense and PSTanti as primers, which resulted in a 1090-bp product in the presence of the transgene.
Northern Blot Analysis-Brain RNA was isolated using Trizol reagent (Invitrogen) as described by the manufacturer. Total RNA (10 g) was separated in 1% agarose formaldehyde gels and transferred onto Hybond-N ϩ nylon membranes (Amersham Biosciences). cDNA probes were generated with [␣-32 P]dCTP by random priming using the Megaprime DNA Labeling System (Amersham Biosciences) following the instructions of the manufacturer. Hybridization was done as described (24). Membranes were first exposed to Bio-Imager screens followed by exposure to x-ray films.
In Situ Hybridization-cRNA probes for PLP and PDGF ␣ receptor were synthesized as described recently (20,25). Mouse ST8SiaIV was subcloned into the pcDNA3 vector (Invitrogen), and cRNA probes for MAG were transcribed from a MAG cDNA in pCR-Blunt II TOPO vector (clone IRAM p995E019Q obtained from RZPD, Berlin, Germany). MBP cDNA was amplified from murine brain RNA by reverse transcription-PCR using oligonucleotides MBPsense (5Ј-GCCTGGATGTG-ATGGCATCAC-3Ј) and MBPanti (5Ј-AGGTGCTTCTGTC-CAGCCATAC-3Ј). The PCR product was cloned into the EcoRV site of pBluescript SK(Ϫ). Antisense and sense probes were transcribed with T7 or SP6 RNA polymerase and digoxigenin-RNA labeling mix following the instructions of the manufacturer (Roche Applied Science). In situ hybridization on brain sagittal paraffin sections or cryosections was done as described (20). Bound probes were detected with alkaline phosphatase-conjugated anti-digoxigenin antibody (Roche Applied Science) and nitro blue tetrazolium/5-bromo-4chloro-3-indolyl phosphate as substrates. Sections were examined with a Zeiss Axiovert inverted microscope (Carl Zeiss, Jena, Germany).
Immunofluorescence of Brain Cryosections-Brains were removed, immediately frozen in cold isopentane, and stored at Ϫ80°C. Coronal or sagittal cryosections were cut at 10 m on a cryostat (Leica CM3050S cryotome) at Ϫ20°C and postfixed in 4% paraformaldehyde in PBS for 10 min. Cryosections were stored with desiccant at Ϫ80°C. Sections were permeabilized with 0.5% Triton X-100, and nonspecific binding sites were blocked with 1% bovine serum albumin, 5% fetal calf serum in PBS for 1 h at room temperature. PSA and MBP were stained with antibodies 735 and rabbit MBP antiserum (Chemicon, Hofheim, Germany) in blocking solution. Primary antibodies were visualized with anti-mouse Ig Cy2-conjugate and antirabbit Ig Cy3-conjugate in blocking solution. Nuclei were counterstained with 4Ј,6-diamidino-2-phenylindole (Sigma). Sec-tions were mounted in Kaiser's gelatin (Merck) and examined with a Zeiss Axiovert fluorescence inverted microscope attached to an Axiocam camera. Photographs were taken with the Axiovision program (Carl Zeiss, Jena, Germany).
Detection of Apoptotic Cells-Apoptotic cells were detected using bromodeoxyuridine incorporation with terminal deoxynucleotidyltransferase using a 5-bromodeoxyuridine triphosphate-FragELT DNA fragmentation detection kit (Merck) according to the manufacturer's instructions. Brain cryosections were postfixed in 4% paraformaldehyde in PBS, treated with 3% H 2 O 2 for 5 min, and incubated with dNTP, 5-bromodeoxyuridine triphosphate, and terminal deoxynucleotidyltransferase in reaction buffer for 30 min at room temperature. After washing, bromodeoxyuridine was detected using antibromodeoxyuridine-biotin/streptavidin peroxidase conjugate and 3,3Ј-diaminobenzidine as substrate.
Myelin Purification-Myelin was isolated from the forebrains of wild-type and transgenic mice at various time points as described by Norton and Poduslo (26). Briefly, brains were homogenized in 10.5% (w/v) sucrose using an Ultra-turrax homogenizer (Janke & Kunkel, Staufen, Germany). Homogenates were centrifuged for 45 min with 17,000 ϫ g. The myelincontaining upper part of the pellet was resuspended in 30% sucrose, overlaid with 10.5% sucrose, and centrifuged for 50 min with 68,000 ϫ g. Myelin was recovered from the 30%/10.5% interphase, washed twice with water, and subjected to a second sucrose gradient in order to remove contaminating axolemma. Purified myelin was again washed twice with water, resuspended in water, and stored in aliquots at Ϫ80°C. In order to determine the total amount of myelin in the brain, the purified myelin was lyophilized, and the dry weight was determined. For protein assays and Western blotting, myelin samples were solubilized in 1% SDS. Protein concentrations were determined using the Bio-Rad DC protein assay. In some experiments, phosphatase inhibitors were included in all solutions: 1 mM sodium ortho-vanadate, 1 mM sodium pyrophosphate, and 1 mM sodium fluoride.
Histology and Electron Microscopy-Intracardial perfusion with glutaraldehyde, tissue processing, and electron microscopy was done as described (29).

Generation of PLP-PST Transgenic
Mice-Transgenic mice expressing the polysialyltransferase ST8SiaIV under the control of the PLP promoter were generated (Fig. 1A). Several transgenic founder mice were identified by Southern blot analysis (Fig. 1B), and two transgenic lines, tg246 and tg281, were established and maintained on a mixed C57BL/6ϫCBA genetic background. Expression of the transgene was examined by Northern blotting using a ST8SiaIV-specific cDNA probe (Fig.  1C). These experiments showed a significant up-regulation in the transgene expression between postnatal day 1 (P1) and adult mice (7 weeks). In contrast, the endogenous ST8SiaIVexpression was down-regulated. In situ hybridizations of 4-week-old mouse brains with a digoxigenin-labeled ST8SiaIV cRNA probe indicated expression of the transgene in oligodendrocytes in the white matter tracts of forebrain, thalamus, and cerebellum ( Fig. 1D), as expected. Detection of the endogenous ST8SiaIV expression required much longer incubation times and is therefore not visible here (data not shown).
Size and morphology of the brain appeared not to be altered in the transgenic mice, and we did not observe changes in the postnatal development of the olfactory bulb, which is reduced in size in NCAM-deficient mice (30,31).
Western blot analysis of brain homogenates revealed that PSA was increased in the adult brains of transgenic mice (Fig.  1E), although PSA-expression was also considerable in wildtype mice in accordance with previous observations (32,33). Reprobing the membranes with NCAM-specific antibodies showed that the NCAM-120 bands were strongly reduced in the transgenic brains because of the molecular mass shift due to the presence of PSA. Western blot analysis of forebrain total membranes pretreated with endoneuraminidase (endoN) to remove PSA indicated that the level of all NCAM isoforms was not changed in the transgenic mice (Fig. 1E) and confirmed that most NCAM-120 (which is the major NCAM isoform in oligodendrocytes) in the brain of transgenic mice was polysialylated.
The presence of PSA in oligodendrocytes was further examined by immunofluorescence on coronal and sagittal cryosections of adult (3-month-old) brains using PSA-and MBP-specific antibodies. PSA in the forebrain of wild-type mice was restricted to the subventricular zone around the lateral ventricle ( Fig. 2A, wt), where PSA-positive neural precursors are generated in the adult brain (34), but was absent from white matter tracts (e.g. the corpus callosum). In contrast, PSA was strongly expressed in the corpus callosum ( Fig. 2A, tg) and other white matter tracts in the striatum (Fig. 2B, tg) and cerebellum (Fig.  2C, tg) of transgenic mice. Thereby, PSA partially co-localized with MBP (Fig. 2D), indicating the presence of PSA-NCAM in myelin. No co-localization of PSA and MBP in wild-type brains was observed. Data shown are from the line tg246; however, similar results were obtained with transgenic mice of the line tg281.
Presence of PSA in Purified Myelin of PLP-PST Transgenic Mice-In order to address the question of whether the presence of PSA interferes with myelination, MBP protein levels were determined in the forebrain of adult (3-month-old) wild-type and transgenic mice (Fig. 3A). Densitometric evaluation clearly showed that compared with wild-type mice, MBP levels were significantly reduced in both transgenic lines. Northern blot analysis of total RNA from brains of transgenic and wild-type mice at different ages indicated that the down-regulation of MBP protein was (at least in part) due to reduced MBP mRNA expression (Fig. 3C). MBP mRNA levels were decreased in young transgenic mice up to 12 weeks of age but reached normal levels in older animals. In contrast, expression of the oligodendrocyte marker cerebroside sulfotransferase (CST), which starts to be expressed earlier during oligodendrocyte differentiation, was similar in wild-type and transgenic mice at all time points analyzed (Fig. 3, C and E).
In order to confirm reduced myelin content in transgenic mice, myelin was purified from the forebrains of transgenic and FIGURE 1. Generation of ST8SiaIV-transgenic mice. A, schematic representation of the transgenic construct. The PLP promoter cassette has been described (19) and consists of the 5Ј-flanking region, first exon and intron, and the splice acceptor site of exon 2. The original ATG start codon in exon 1 is mutated to ensure efficient translation from the ATG of the cDNA insert. The murine ST8SiaIV cDNA was subcloned into the PmeI site and is shown by a solid box. B, Southern blot analysis of genomic tail DNA from two transgenic founder mice (246 and 281) and a nontransgenic mouse (245). Genomic DNA was digested with AccI and probed with the ST8SiaIV cDNA probe indicated in A. C, Northern blot analysis of transgenic mice and nontransgenic littermates at P1 and 7 weeks of age. Total brain RNAs (10 g/lane) were separated by gel electrophoresis, transferred onto nylon membranes, and hybridized to an ST8SiaIV-specific probe. Positions of endogenous (eg) and transgene (tg) ST8SiaIV mRNAs are indicated. In order to clearly visualize the weak endogenous ST8SiaIV band in RNA from adult mice, a longer exposure time was required, resulting in a strong background staining in the case of tg246 transgenic mice. EtBr staining served as a loading control. D, in situ hybridization. Sagittal brain sections of the lines tg246 and tg281 and wild-type (wt) mice (age 4 weeks) were hybridized to a digoxigenin-labeled mouse ST8SiaIV antisense cRNA probe. Control hybridizations with a ST8SiaIV sense probe gave no specific staining. Note that under the conditions used here (short staining time) to detect transgene expression, the expression of the endogenous STSiaIV was below the detection limit. cx, cortex; cc, corpus callosum. E, Western blot analysis of PSA and NCAM expression in total membranes isolated from the forebrains of transgenic (tg246 and tg281) and wildtype (wt) mice. Part of the samples were treated with endoneuraminidase N F (ϩendoN) in order to remove PSA prior to SDS-PAGE. PSA levels were slightly increased in transgenic brains as compared with wild-type mice. Probing the membranes with anti-NCAM antibody H28 showed that NCAM-120 was almost undetectable in line tg246 and very weak in line tg281. After endoN treatment, the amounts of all NCAM isoforms were similar in wild-type and transgenic mice. These data clearly show that most NCAM-120, the predominant NCAM isoform in oligodendrocytes, was polysialylated in the transgenic mice. Total NCAM levels were, however, not affected. wild-type mice at different time points (between 1 week and 12 weeks of age), and its dry weight was determined (Fig. 4A). We found a significant reduction in the myelin content at P21 and 10 -12 weeks. The difference of the mean at P10 was not signif-icant, most likely because of the low myelin content at this age. To examine possible changes in the myelin composition, Western blot analysis was performed. These confirmed the presence of polysialylated NCAM-120 in myelin (Fig. 4B). The total amount of NCAM-120 in myelin was, however, not reduced in the myelin of transgenic mice (Fig. 4C). Examination of several other myelin proteins, Fyn, CNP, MBP, did not reveal obvious differences in the amount of these proteins in purified myelin (Fig. 4D). Furthermore, silver staining of SDS-polyacrylamide gels did not indicate obvious changes in the myelin protein composition between transgenic and wild-type mice (data not shown).
In order to rule out the possibility that overexpression of ST8SiaIV affects other sialylation reactions because of the increased utilization of CMP-sialic acid for polysialylation, brain homogenates, membranes, and purified myelin were examined by Western blotting using the sialic acid binding lectin from Tritrichomonas. No significant differences in the lectin staining pattern were observed (Fig. 5, A and B). Furthermore, we examined gangliosides isolated from the forebrain and purified myelin of adult (8-month-old) transgenic and wild-type mice. Alkaline-stable lipids were separated by HPTLC. In the forebrain samples, we found a significant reduction in the ganglioside GM1 in the transgenic mice. In addition, the levels of two other myelin-specific sphingolipids, galactosylceramide and sulfatide, were decreased (Fig. 5C). In purified myelin, however, the level of galactosylceramide, sulfatide, and ganglioside GM1 was not decreased in transgenic mice as compared with wild-type controls (Fig. 5D). We therefore conclude that the overexpression of ST8SiaIV did not inhibit sialylation of sphingolipids but that the reduced GM1 level in total forebrain samples reflects the hypomyelination and progressive demyelination (see below) observed in ST8SiaIV-transgenic mice.   0.05, t test). C, Northern blot analysis of MBP and cerebroside sulfotransferase (CST) mRNA expression at different time points in wild-type and transgenic mice (line tg246). Membranes were rehybridized to an actin-specific probe as a loading control. D, densitometric quantification revealed a reduction in the MBP mRNA level (normalized to actin) in younger mice. In contrast, mRNA levels of cerebroside sulfotransferase (E), which is up-regulated much earlier during oligodendrocyte differentiation, were similar in wild-type and transgenic mice.  (n ϭ 3). Asterisks indicate significant differences between transgenic and wild-type mice (t test; p Ͻ 0.05). B, Western blot analysis of purified myelin with PSA-specific antibody 735 indicated presence of PSA-NCAM in the myelin of transgenic mice at all time points analyzed. In contrast, PSA was hardly detectable in the myelin of wild-type mice after the period of active myelination (14 and 24 weeks). C, Western blot analysis of NCAM. Treatment of samples with endoneuraminidase (ϩendoN) to remove PSA before SDS-PAGE showed that the amount of NCAM-120 was not reduced in transgenic mice. D, Western blot analysis of purified myelin for various myelin-associated proteins (Fyn, CNP, and MBP) indicated no significant differences between wild-type and transgenic mice. Western blots shown are from 24-week-old mice. Similarly, silver staining of purified myelin did not indicate major differences in the protein composition (data not shown). The asterisks indicate a significant difference of the mean (p Ͻ 0.05, t test). dendrocytes, or formation of the myelin sheath. Because PLP is first expressed in mature oligodendrocytes, we expected that differentiation of oligodendrocytes should not be affected in the transgenic mice. In order to prove this hypothesis, we quantified the number of mature oligodendrocytes in the forebrain at different time points (between 1 and 4 weeks of age) by in situ hybridization of sagittal sections of PLP-PST transgenic mice and wild-type littermates using digoxigenin-labeled PLP cRNA probes (Fig. 6A). Interestingly, the number of PLP-positive cells in the forebrain was significantly reduced in transgenic mice at 1 and 2 weeks of age (Fig. 6B). At 4 weeks of age, reduction in the number of PLP-positive cells was less prominent and was no longer significant in the transgenic line tg281 (Fig. 6B). Additional hybridizations were done with a cRNA probe of the PDGF ␣ receptor as a marker for OPC (Fig. 6C). Here we did not observe a decrease in cell number in the transgenic lines at 1 and 2 weeks of age (Fig. 6D). This indicates that the number of OPC was not reduced in the transgenic mice, which was expected, since the ST8SiaIV transgene should not be expressed at the PDGF receptor-positive stage of differentiation. Because reduced numbers of PLP-positive cells could be due to increased cell death, TUNEL staining was performed in order to detect apoptotic cells. TUNEL-positive cells were rarely detectable in the forebrains of both wild-type and transgenic mice at P14 (Fig. 6E), and an increase in the number of apoptotic cells in the transgenic brains could not be detected. In order to confirm our results, we repeated the in situ hybridizations with another marker for mature oligodendrocytes, MAG. One-week-old transgenic mice (line tg246) and wild-type littermates were hybridized to MAG-specific probes (Fig. 6F). The number of MAG-positive cells in the forebrain was significantly reduced in transgenic mice (Fig. 6G).
The observation of a reduced number of PLP-positive cells in transgenic mice was unexpected, because the ST8SiaIV transgene should only be expressed in PLP-expressing mature oligodendrocytes. One possible explanation is that the initial induction of ST8SiaIV transgene expression inhibits further differentiation from immature to mature oligodendrocytes and thereby further up-regulation of PLP expression such that the PLP mRNA level remains below the detection limit of the in situ hybridization.
In Vitro Culture of Oligodendrocytes from Transgenic Mice-We next asked whether presence of PSA affects morphological differentiaton of oligodendrocytes in vitro. Therefore, mixed primary cultures were isolated from newborn mice and grown for 8 -10 days before shaking off OPC growing on top of the astrocyte monolayer. OPC were differentiated in SATO medium, and the number of MBP-positive cells (as a marker for mature oligodendrocytes) was determined between 1 and 4 days of differentiation (Fig. 7A). The percentage of MBP-positive cells in transgenic cultures was not significantly different from wild-type cultures (Fig. 7B). Thus, in vitro differentiation of oligodendrocytes was not inhibited or delayed in the presence of PSA. As shown in Fig.  7C, wild-type (wt) mature MBP-positive oligodendrocytes from wild-type mice were PSA-negative. In contrast, oligodendrocytes from transgenic mice were PSA-positive (Fig.  7C, tg). PSA was concentrated around the cell bodies but was also found in the membranous extensions of oligodendrocytes. Transgenic oligodendrocytes displayed a less mature morphology, with fewer processes (Fig. 7C) and less extensive membrane sheaths (Fig. 7A). Co-staining for PSA and NCAM revealed partial co-localization of PSA and NCAM in the membrane extensions of oligodendrocytes (Fig. 8A). However, in the perinuclear region, which showed strong PSA staining, the NCAM signal was very low, and PSA and NCAM did not co-localize, suggesting that other proteins beside NCAM carried PSA in the transgenic oligodendrocytes (Fig. 8B). Since ST8SiaIV is highly overexpressed and exhibits autopolysialylation activity, it is likely that the perinuclear PSA staining is due to autopolysialylated ST8SiaIV. We cannot, however, rule out the possibility that other glycoproteins were modified by the overexpressed ST8SiaIV. The distributions of NCAM (Fig. 8A) and L-MAG (Fig. 8C) in the membrane extensions were not significantly altered in the transgenic cells as compared with wild-type controls.
Ultrastructural Analysis of Myelin in ST8SiaIV-transgenic Mice-Although myelin content was reduced, no obvious severe behavioral deficits were observed in young mice up to an age of about 4 months. Interestingly, however, we observed FIGURE 5. ST8SiaIV overexpression did not reduce sialylation of glycoproteins and glycosphingolipids. A, membranes (membr.) and total homogenates (homog.) of the forebrain (30 g of protein/lane) of wild-type and transgenic mice (8 months of age) were examined by Western blotting with the sialic acid binding lectin from T. mobilensis. B, lectin blot of myelin samples (100 g/lane) from 3-week-old transgenic (lines tg246 and tg281) and wildtype mice. Sphingolipids were isolated from 8-month-old forebrain (C) or purified myelin (D) and analyzed by TLC using chloroform, methanol, 0.22% CaCl 2 (60:35:8) as a solvent system. The TLC shown in the upper part of D was developed in chloroform/methanol/water (70: 30:4). Lipids corresponding to 1 mg (wet weight) of brain tissue and 100 g of myelin, respectively, were loaded onto HPTLC plates. Gangliosides (GM 1 , GD 1a , GD 1b , and GT 1b ), galactosylceramide (GalC), sulfatide, and sphingomyelin (SM) from bovine brain served as standards. PE and PL, phosphatidylethanolamine and other phospholipids, respectively. progressive neuromotor coordination deficits and hind limb paralysis in a minor fraction (about 5%) of older transgenic mice of the line tg246. The reason for this variation might be the variable genetic background. We examined the central nervous system of adult (6-month-old) transgenic mice by electron microscopy for myelin deficits (Fig. 9). Structural abnormalities like redundant myelination (Fig. 9D), axonal degeneration (Fig. 9B), and vacuolar degeneration (Fig. 9E) in the optic nerve (Fig. 9) and the spinal cord (data not shown) were observed in both transgenic lines but not in wild-type mice. These abnormalities were observed in all transgenic mice; however, mice of the tg246 line displayed a more severe phenotype. Furthermore, we found indications for demyelination in the transgenic line tg246 (Fig. 9C). Demyelination in these mice was also observed in the spinal cord (data not shown). Since a severe demyelination was only observed in one line, we cannot formally exclude the possi-bility of a transgene insertion artifact. The presence of neurological symptoms (see above) in line tg246, however, correlated well with the higher transgene expression level in this line. Furthermore, this line also revealed a more pronounced effect of the transgene expression on oligodendrocyte differentiation and myelin formation (see above). Thus, the absence of the severe phenotype in the line tg281 might be explained by the lower transgene expression level.

DISCUSSION
Oligodendrocyte differentiation and myelination of axons in the central nervous system are controlled by several positive and negative signals (36). Negative signals are important to prevent the premature differentiation of OPC and myelination before they reach their final destination. Examples for negative regulators of myelination are the sphingolipid sulfatide (37) and LINGO-1 (38). Another potential negative regulator of myelination is PSA.
PSA is required for the migration of OPC as shown using endoN treatment or OPC isolated from NCAM-deficient mice (11). More recent observations indicate that PSA is also required for the PDGF-dependent migration of OPC (12). Upon differentiation of OPC into oligodendrocyte precursors and thereafter mature myelin-forming oligodendrocytes, PSA is down-regulated (8,9). Down-regulation is mainly caused by reduced ST8SiaIV mRNA synthesis (39). Differentiating oligodendrocytes up-regulate PLP. Accordingly, due to the use of the  PLP promoter in the transgenic construct, down-regulation of PSA was prevented in PLP-ST8SiaIV transgenic mice and most NCAM-120, the predominant glycosylphosphatidylinositollinked NCAM isoform in oligodendrocytes, was synthesized as polysialylated protein.
Although NCAM is the major PSA-carrying glycoprotein, other proteins can be polysialylated (e.g. the polysialyltransferases themselves) (40,41). In the oligodendrocyte culture, we observed strong PSA signals in the perinuclear region of ST8SiaIV-transgenic cells. Whether these proteins are the overexpressed ST8SiaIV and/or other proteins is currently unknown. Overexpression of ST8SiaIV, however, did not lead to a general reduction in the sialylation level of other glycoproteins or glycolipids. The ST8SiaIV overexpression also did not affect the distribution of NCAM or L-MAG in the oligodendrocyte culture. Furthermore, the composition of purified myelin was not affected in the transgenic mice. These observations suggest that the transgene expression did not significantly influence localization of myelin membrane components. We cannot, however, rule out the possibility that attachment of PSA to non-NCAM proteins that were detected in cultured oligodendrocytes is in part responsible for the observed phenotypes. This could be examined by overexpressing ST8SiaIV in NCAM-deficient mice.
When PSA down-regulation was abolished, the number of PLP-positive oligodendrocytes in the forebrain was significantly reduced. This observation was unexpected, because one should expect that only PLP-expressing cells express the transgene, which is under the control of the PLP promoter. Our interpretation of this observation is that a low activity of the PLP promoter of the transgene is sufficient (due to the large copy number of the transgene) to result in the synthesis of PSA-NCAM that inhibits further differentiation and thereby further up-regulation of PLP expression. The initial low activity of the endogenous PLP promoter might result in PLP expression levels that are below the detection limit of the in situ hybridization.
Franceschini et al. (42) showed that ST8SiaIV-overexpressing neural precursor cells transplanted into the brain of MBP-deficient dysmyelinating shiverer mice showed delayed myelination. When myelination finally occurred, this was accompanied by down-regulation of PSA (42). PSA down-regulation most likely occurred at the posttranscriptional level. Since PSA down-regulation was more pronounced in vivo than in vitro, Franceschini et al. (42) concluded that signaling from the axons might induce the down-regulation of PSA synthesis. In contrast to this study, we did not observe down-regulation of PSA during myelination in ST8SiaIV-transgenic mice. The reason for these contrary observations is presently not clear, but it might be due to the different expression systems used (mouse versus human; EGFP fusion protein versus nontagged protein, retrovirus versus pronucleus injection). Nevertheless, our results show that myelin formation does not necessarily require loss of PSA from the oligodendrocyte membrane, since myelin was formed in the transgenic mice, and PSA-NCAM partially co-localized with MBP and was present in purified myelin. Brains were, however, hypomyelinated, and myelin formed appeared to be less stable. Furthermore, PSA affected maintenance of myelin and axon-supporting function of oligodendrocytes, as indicated by redundant myelination and axonal degeneration. The observed demyelination in mice of one transgenic line raises the possibility that the presence of larger amounts of PSA in myelin can interfere with myelin maintenance, although at present, we cannot rule out the possibility of a transgene insertion artifact.
PSA-negative mice have recently been generated by deleting both polysialyltransferases, ST8SiaIV and ST8SiaII (43). In these mice, the density of some myelinated fiber tracks was significantly decreased. However, it was not clear whether this phenotype results from inefficient myelination or reduced axon numbers. The damage of the corpus callosum and other fiber tracts in ST8SiaIV/ST8SiaII double knock-out mice appeared to be secondary to the ventricular dilation (43). Since the phenotype of these PSA-negative mice could be rescued by simultaneously deleting NCAM, it seems that an important role of PSA is to block NCAM function during development (43). If the effect of PSA overexpression on the number of mature (PLPpositive) oligodendrocytes and myelin formation would be due to block NCAM function, NCAM-deficient mice should phenocopy ST8SiaIV-overexpressing mice in this respect. Deficits in myelin formation of the central nervous system of NCAMdeficient mice, however, have not been reported, and myelination was normal in peripheral nerves (44). Furthermore, at least in the rostral migratory stream, the number of oligodendrocytes was found to be increased (45). Taken together, these data suggest that inhibition of myelin formation and myelin instabilities in ST8SiaIV-transgenic mice is not due to a blockade of NCAM function by PSA. How could the presence of PSA inhibit myelin formation? Removal of PSA with endoN enhances not only NCAM homophilic interactions but also adhesion between other cell surface molecules, like cadherins, integrins, or L1 (46,47). According to the proposed "pull/push" model (48), the degree of specific interactions between adhesion molecules and the amount of (nonspecific) repulsive steric forces provided by PSA determine the strength of adhesion and the distance between opposing cells. Johnson et al. (49) showed that the magnitude of repulsion mediated by PSA correlated directly with the reduced homophilic NCAM and N-cadherin interaction. It is thus possible that oligodendrocyte processes covered with PSA are less efficient in wrapping around axons. If the PSA effect is solely due to the "pull/push" mechanism, PSA should achieve similar effects, irrespective of its presence on oligodendrocytes or the axonal membrane. In line with this, Meyer-Franke et al. (50) showed that the removal of PSA from axons in neuron/oligodendrocyte co-cultures strongly increased the alignment of oligodendrocytes with the axonal membranes. Similarly, PSA on axonal membranes has been implied as an inhibitor of myelination (16,17).
Modulation of signal transduction pathways by PSA has been demonstrated recently. Endoneuraminidase treatment of neuronal cultures significantly affected their brain-derived neurotrophic factor-dependent survival, and it was shown that this was due to an inhibition of brain-derived neurotrophic factor signaling after PSA removal (51). Seidenfaden et al. (52) recently showed that endoN treatment of neuroblastoma cells affects NCAM-mediated signaling through the mitogen-activated protein kinase pathway. Thus, the PSA modification of NCAM has a strong effect on signal transduction in different systems. ST8SiaIV-transgenic oligodendrocytes displayed a less mature morphology, with fewer processes and less extensive membrane sheaths. Similar observations have been made by Franceschini et al. (42). We therefore hypothesize that the presence of PSA on NCAM-120 in oligodendrocytes interferes with signaling pathways required for efficient process extension and wrapping of myelin sheaths around axons. One possible candidate that might be affected by polysialylation is Fyn kinase, which is essential for oligodendrocyte differentiation (53,54) and up-regulation of MBP gene expression (55) and which moreover co-localizes with NCAM-120 in detergentresistant membranes (56). In preliminary experiments, we did not observe changes in the tyrosine phosphorylation in purified myelin in transgenic mice. 4 However, it is possible that an inhibition of signal transduction via Fyn kinase or others is only transient during the initiation of myelination.
In addition to its inhibitory effect on oligodendrocyte differentiation (13) and myelin formation, the continuous synthesis of PSA-NCAM also affected myelin in the adult animal. Structural abnormalities of myelin (redundant myelination) and vacuolar degeneration were observed. These observations show that PSA down-regulation is not only necessary for efficient differentiation of oligodendrocytes and myelin formation but that its presence in myelin also interferes with the stable maintenance of the myelin sheath and axonal support by oligodendrocytes.