Journal of Neuroscience Research 29:141-148 (1991)
Differential Expression of MAG Isoforms During Development L. Pedraza, A.B. Frey, B.L. Hempstead, D.R. Colman, and J.L. Salzer Departments of Cell Biology (L.P., A.B.F., J.L.S.) and Neurology (J.L.S.), New York University Medical School, New York; Division of Hematology, Cornell Medical Center, New York (B.L.H.); Department of Cell Biology, Columbia University Medical Center, New York (D.R.C.), New York The myelin-associated glycoproteins (MAG) mediate the cell interactions of oligodendrocytes and Schwann cells with axons that are myelinated. MAG exists in two developmentally regulated isoforms: large MAG (L-MAG) and small MAG (S-MAG). In this paper, we have studied the tissue-specific and developmentally regulated alternative splicing of these isoforms using monospecific antibodies that recognize epitopes common to both isoforms or that are present only on L-MAG. In the central nervous system (CNS), LMAG is the major form synthesized early in development, and it persists as a significant proportion of the MAG present in the adult. In the peripheral nervous system (PNS), L-MAG is expressed at modest levels during development; it is virtually absent in the adult. Thus, the expression of L-MAG is not limited to the CNS, as was formerly believed, suggesting that it plays a common role during the early stages of myelin formation by both oligodendrocytes and Schwann cells. In both the CNS and PNS, S-MAG is the predominant isoform in the adult. A higher-molecular-weight form of MAG is present in the PNS at low abundance, that is developmentally regulated, and appears to be a glycosylation variant. An analysis of the carbohydrate residues on MAG demonstrates that it contains both N-linked and 0-linked sugars that could be modulated during development. These results suggest a possible mechanism for the regulation of MAG function during myelinogenesis via the expression of alternative isoforms and carbohydrate modifications. Key words: MAG, myelin-associated glycoprotein, alternative splicing, immunoblotting, 0-linked, Nlinked
(Sternberger et al., 1979; Trapp and Quarles, 1982) and its ability to promote binding to axons in vitro (Poltorak et al., 1987; Sadoul et al., 1990). In agreement with its proposed function as a cell adhesion molecule, MAG is known to be a member of the immunoglobulin gene superfamily (Arquint et al., 1987; Lai et al., 1987; Salzer et al., 1987), a family of proteins that frequently function in tissue-specific cell interactions (Williams and Barclay, 1988; Salzer and Colman, 1989). MAG has recently been shown to promote the ensheathment and to enhance the outgrowth of large-caliber fibers in vitro (Johnson et al., 1989; Owens et a]., 1990). These results suggest that MAG has a role in the sorting and segregation of the subset of nerve fibers destined to become myelinated. Two isoforms of MAG have been identified (Frail and Braun, 1984) that differ in the sequences and size of their cytoplasmic domains (Lai et al., 1987; Salzer et al., 1987). These proteins, designated large MAG (L-MAG) and small MAG (S-MAG), correspond to core polypeptides of 67 and 62 kD, respectively (Pedraza et al., 1990), but comigrate as a broad 100-kD band as a result of extensive glycosylation. They are translated from an alternatively spliced transcript encoded by a single gene (Barton et al., 1987; D’Eustachio et al., 1988; Lai et al., 1987). The messenger RNA (mRNA) for these isoforms are differentially expressed during development and between the central nervous system (CNS) and peripheral nervous system (PNS) (Frail et al., 1985; Lai et al., 1987; Tropak et al., 1988). L-MAG is expressed early in development, during the rapid phase of myelination, whereas S-MAG is expressed later in development and has been thought to be the only isoform synthesized in the PNS. These findings have suggested that these proteins have distinct functions at different stages of myelination and, potentially, in oligodendrocytes vs. Schwann cells. If so, the regulation of the alternative splicing of
INTRODUCTION
with its presence in the periaxonal glial membrane 0 1991 Wiley-Liss, Inc.
y i r k , NY 10016
142
Pedraza et al.
the two isoforms of MAG could represent a mechanism for regulating the function of this protein during myelinogenesis. To date, the level of expression of the MAG isoforms has been inferred from the abundance of their corresponding mRNA, rather than by direct protein determinations. While the regulation of myelin protein abundance is believed to be largely transcriptional (Lemke, 1988), post-transcriptional mechanisms are also believed to be important (Brunden and Poduslo, 1987) and could differentially affect the accumulation of the MAG proteins. Thus, while small amounts of L-MAG mRNA have been detected in the PNS (Frail et al., 1985; Tropak et al., 1988), the protein has not (Agrawal et al., 1990; Noronha et al., 1989). This apparent discrepancy could reflect the transient expression of the L-MAG protein during peripheral nerve development, or alternatively post-transcriptional mechanisms that limit the translation of the L-MAG mRNA or the stability of the L-MAG protein. We have therefore directly analyzed the levels of L-MAG and total MAG protein in the CNS and PNS using monoclonal antibodies of defined specificity. We have found that L-MAG is a normal component of peripheral myelin during development, suggesting that it has a common function during the formation of myelin by Schwann cells and oligodendrocytes. Unlike its expression in the CNS, the L-MAG isoform is a minor isoform in the PNS at all stages of myelination, accounting for only 10% of total MAG at its peak. We have also found evidence for a developmentally regulated glycosylation variant of MAG that is present in the PNS but not in the CNS. Analysis of the carbohydrate residues on MAG demonstrated that it contains both O-linked and N-linked oligosaccharides that could be modulated during development. These results are considered with respect to the potential regulation of MAG function by expression of alternative isoforms and carbohydrate modifications.
sonicated in 2% sodium dodecyl sulfate (SDS), and boiled for 2 min. In each case, insoluble materials werepelleted at 13,OOOg for 10 min and discarded. Total cell lysates from 10' transfected cells expressing each of the insoforms of MAG were prepared as previously described (Pedraza et al., 1990). Aliquots (50 pg of protein) of tissue supernatents or membranes were fractionated on 8 -1 6% SDS-polyacrylamide gel electrophoresis (PAGE) gradient gels and blotted onto nitrocellulose (Towbin et al., 1979). Blotted samples were probed with anti-MAG monoclonal antibodies (I: 1,OOO), washed and incubated with a rabbit antimouse antibody at 1:500 (Cappel). MAG reactive bands were visualized by incubating the blots with [1251]proteinA (ICN Biochemicals) at 2 X lo5 c p d m l , washing, and then exposing the blots for autoradiography . Optical densitometry was performed with a Hoefer (GS300) scanning densitometer on the autoradiographs, to quantitate the level of MAG expression. In other studies using synthetic peptides, the peptides were dissolved in 1 N NaOH, and the solution was neutralized. Peptide, 1 pg, was spotted onto nitrocellulose, air dried, and fixed with a solution of 25% isopropyl alcohol and 10% acetic acid. Immunodetection with anti-MAG antibodies was performed as described above.
MATERIALS AND METHODS Antibodies The generation of antipeptide antibodies that recognize an epitope common to both isoforms or an epitope specific to L-MAG has previously been described (Pedraza et al., 1990). Ascites containing the HNK-l or the antihuman MAG monoclonal antibodies GEN S10 and S3 (Nobile-Orazio et al., 1984) were generous gifts of Dr. N. Latov.
RESULTS Characterization of Antibodies Specific for L-MAG We have determined the reactivity of an anti-MAG monoclonal antibody GEN S10, a mouse monoclonal generated against human MAG (Nobile-Orazio et al., 1984). GEN S10, like GEN S1, specifically recognizes CNS but not PNS MAG (Nobile-Orazio et al., 1984), suggesting that both antibodies specifically recognize LMAG but not S-MAG. To confirm this specificity, we have compared the reactivity of GEN S10 with the individual isoforms of MAG expressed in transfected fibroblast lines (Pedraza et al., 1990) (Fig. 1A) by Western blotting. As a control antibody, we have used GEN S3, a mouse monoclonal (Nobile-Orazio et al., 1984) that
Sample Preparation and Immunoblotting Procedures Sciatic nerves or brainstems were removed from postnatal rats, cut into approximately 3-mm fragments,
Enzymatic and Chemical Deglycosylation of MAG MAG was immunoprecipitated with an antipeptide antibody that recognizes both isoforms (Pedraza et al., 1990). Enzymatic deglycosylation was then performed using O-glycanase or endoglycosidase F (both from Boehringer Mannheim) at pH 6.5, following the manufacturer's instructions. Pretreatment of MAG with neuraminidase prior to O-glycanase did not alter the results. Chemical deglycosylation was done on total lysates of rat brainstem or sciatic nerve using trifluoromethanesulfonic acid as described (Horvath et a]., 1989).
Developmental Expression of MAG Isoforms
IA
18
143
trast, and as expected, an antipeptide antibody raised against the COOH terminal 21 amino acids of L-MAG only recognizes this peptide (lane b).
GEN SIODevelopmental Expression of the L-MAG Protein We have used the anti-MAG monoclonal antibodies to determine the developmental expression of LMAG and total MAG in the CNS and PNS. In quantita0 tive comparisons, selected ascites of GEN S3 and S10 were found to react equally well (+3%), on immunoblots, to equivalent amounts of L-MAG from transfected cells; the radioactivity was proportional to the quantity of MAG blotted (data not shown). Thus, the amount of total MAG (L-MAG plus S-MAG) vs. L-MAG could be comat! c d pared directly using GEN S3 and GEN S10, respectively . We have obtained essentially identical results to Fig. 1 . A. Reactivity of anti-MAG monoclonal antibodies. those shown here using antipeptide antibodies specific Total cell lysates from transfected Schwann cells expressing isoforms of S-MAG (lanes a, c) or L-MAG (lanes b, d) were for L-MAG or react with both isoforms (data not shown). The temporal expression of MAG in the CNS and blotted after SDS-PAGE. The anti-MAG monoclonals GEN S3 (lanes a, b) and GEN SlO (lanes c, d) were used as the the PNS (from postnatal day 1 through the adult) is primary antibodies. Immunoblots were then incubated with a shown in Figures 2 and 3, respectively. In the brainstem, secondary antibody followed by [12511proteinA and visualized MAG is first detectable at low levels at postnatal day 6. by antoradiography . B. Reactivity of L-MAG antibodies with Substantial levels of MAG appear at postnatal day 12, synthetic peptides. Two synthetic peptides, corresponding to peak at about 1 month, and remain relatively stable nonoverlapping sequences of the COOH terminus of L-MAG, thereafter (Fig. 2A). The stable levels of total MAG were spotted onto nitrocellulose. One peptide (amino acids protein were somewhat surprising, as MAG mRNA lev583-603) was spotted in lane a and the other (amino acids els are known to decline substantially after day 25 606-626) in lane b. GEN SlO and an antipeptide antibody (Salzer et al., 1987). Consistent with previous studies of (antiP2) directed against the COOH terminus, were reacted mRNA expression (Lai et al., 1987; Tropak et al., 1988), with the blot. the L-MAG protein is synthesized at the earliest stages of myelination in the CNS (Fig. 2B). At days 6-12, the L-MAG isoform accounts for virtually all the MAG was recently shown (Pedraza et al., 1990) to react with present in the brainstem (see Fig. 4A). Expression of an epitope of the second Ig-like domain of MAG com- L-MAG peaks at postnatal day 2.5 and thereafter demon to both S- and L-MAG (Fig. IA, lanes a and b, clines. It is also apparent that the size of MAG declines respectively). By contrast, GEN S10 reacts only with during development, shifting from 110 kD initially to L-MAG (compare lanes c and d) and therefore recog- 105 kD in the adult (Fig. 2A). However, no shift in the nizes an epitope present at the COOH terminus specific molecular weight of L-MAG is evident during developto this isoform (the terminal 53 amino acids). The reac- ment (Fig. 2B). Thus, the decrease in the size of MAG tivity of GEN S10 for L-MAG is of interest because this during CNS development appears to reflect the transition antibody does not react with human PNS MAG or from L-MAG to S-MAG rather than a change in the dMAG (see Discussion). amount of glycosylation of MAG. There are 53 amino acids at the COOH terminus of In the PNS, MAG is expressed immediately after L-MAG that are unique to this isoform (amino acids birth, accumulates until approximately postnatal day 9, 574-626) (Salzer et al., 1987). To further define the and remains stable thereafter (Fig. 3A). L-MAG is sequence recognized by GEN S10, we determined its readily detected beginning on postnatal day 3 , is most reactivity with two synthetic peptides corresponding to abundant at days 6-9, and persists at low but detectable amino acids 583-603 or 606-626 (the COOH terminus) levels in the adult (Fig. 3B). Proportionately, L-MAG of the published L-MAG sequence (Lai et al., 1987; expression is maximal at days 6-9, as previously noted Salzer et al., 1987). As shown in Figure lB, we found for its mRNA expression (Tropak et al., 1988), a period that GEN S 10 specifically reacts with a synthetic peptide corresponding to the time of most rapid myelination in corresponding to amino acids 583-603 (lane a), but not the rat sciatic nerve (Wood and Engel, 1976). At its with amino acid sequence 606-626 (lane b). By con- relative peak (day 6), L-MAG accounts for an estimated
anti P2
-
-
144
Pedraza et al.
2A
I
3 6 9 M 15 19 2 5 4 0 A
Fig. 2 . Temporal expression of MAG in the CNS. Samples of brainstem from postnatal rats from the ages indicated were blotted and probed with GEN S3 (A), which recognizes both isoforms of MAG, and GEN S10 (B), which is specific for L-MAG. Blots were then incubated with a secondary antibody
28
I 3 6 9
and ['2sI]protein A. In both cases, autoradiographs were exposed overnight. Ages of animals in postnatal days are indicated (A = adult). Small amounts of MAG were seen at days 6 and 9 on longer exposures than that shown here.
3B
3A
I
3
6
9 12 15 19 2540A
12 15 19 2 5 4 0 A
I
3 6 9 12 1 5 1 9 2 5 4 0 A
Fig. 3. Temporal expression of MAG in the PNS. Rat sciatic nerve preparations from postnatal rats of the indicated ages were analyzed by SDS-PAGE and immunoblotting. Total MAG was visualized by GEN S3 (A), and L-MAG was visu-
alized by GEN S10 (B). Exposure times of the autoradiographs were overnight (A) and 5 days (B). The arrow in B indicates the position of the higher-molecular-weight form of PNS MAG.
10% of the MAG in sciatic nerve. Thus, L-MAG is a normal component of PNS myelin and, as in the CNS, it is expressed at times of rapid myelination. In sharp contrast to the CNS however, L-MAG is always a minor isoform, and in the adult comprises < 1 % of the total MAG. These results are presented graphically in Figure 4. By subtracting the amount of L-MAG from the total MAG, we have estimated the level of S-MAG expressed in the brainstem and in the sciatic nerve during development. In the CNS, S-MAG is first present at about day 15; levels of S-MAG probably gradually increase thereafter, reaching about 80% of the total MAG present in the adult brainstem. In parallel, the amount of L-MAG present in the brainstem declines, even though total MAG levels are quite stable. In the PNS, S-MAG expression and total MAG expression are closely correlated at all time points. The declining levels of L-MAG, particularly evident in the CNS after day 40 (Fig. 4A), suggest that this isoform of MAG turns over during late stages of development.
Expression of a Developmentally Regulated Variant of MAG in the PNS Also apparent (Fig. 3B, arrow) is an apparently discrete, higher-molecular-weight form of MAG of 110 kD. This band is not a cross-reactive protein of similar molecular weight, as it has been detected by four different anti-MAG antibodies: two antipeptide antibodies and two monoclonal antibodies. This MAG band becomes maximal at postnatal days 15-19 in the sciatic nerve and persists at comparable levels at all subsequent time points, including the adult. It is present on the L-MAG polypeptide backbone and, on the basis of its abundance, may be present on the S-MAG polypeptide as well. Interestingly, the small amount of L-MAG detectable in the adult corresponds exclusively to this higher-molecular-weight form. Immunoprecipitation of L-MAG from the adult demonstrated that it migrates at a somewhat higher molecular weight than does L-MAG from early postnatal time points (data not shown). We were interested in whether this higher-molecu-
Developmental Expression of MAG Isoforms 3000-
145
A S-MAG
2500 -
--c L-MAG
2000 1500-
1000-
500
-
-
25001
B
2000 -
-w-
S-MAG
--t
L-MAG
1 500 -
1000 500 -,
0
0
5
10
15
20
25
30
35
40
A
Fig. 4. Temporal and tissue-specific expression of MAG isoforms. Autoradiographs, such as those shown in Figures 2 and 3 , were quantitated by densitometry, and the values from two separate experiments (including correction for length of exposure) were averaged. Total MAG and L-MAG were measured
directly. S-MAG values shown are calculated by subtracting L-MAG values from total MAG. (A) Expression of MAG isoforms in the brainstem. (B) Expression of MAG in the sciatic nerve. Age of the sample is shown below (A = adult); units at the left are arbitrary densitometric values.
lar-weight MAG species might represent an alternatively spliced isoform of MAG specific to the PNS or whether it is a glycosylation variant. We chemically deglycosylated MAG from day 15 sciatic nerve prior to blotting (Fig. 5 ) . After chemical deglycosylation, PNS MAG migrates as a single band with a molecular weight of -65 kD (Fig. 5 , lane d), consistent with the size of the SMAG polypeptide, which predominates at this time point. Similarly, chemically deglycosylated L-MAG (visualized with the S 10 antibody) from sciatic nerves of different ages migrates with a uniform molecular weight of -70 kD (data not shown). These results strongly suggest that the higher-molecular-weight band present in the PNS is a glycosylation variant and not a novel alternatively spliced isoform of MAG. Because the HNK-1 epitope is a carbohydrate epitope known to be present on a proportion of MAG molecules in some animal species (Kruse et al., 1984;
McGany et al., 1983), we examined whether the highermolecular-weight band present in the PNS reacted with this antibody. However, neither the upper nor lower MAG bands reacted with HNK-1 (data not shown).
MAG Contains 0-Linked and N-Linked Oligosaccharides As part of an analysis of the nature of the glycosyl residues present on PNS MAG, including the nature of the higher-molecular-weight MAG species, we investigated the oligosaccharides present on MAG. Consistent with our earlier studies (Salzer et al., 1987), we found that most of thexarbohydrate residues on MAG are Nlinked, as evidenced by the sharp decrease in the size of the protein after endoglycosidase F treatment (Fig. 5, lane c). In addition, MAG from both the CNS (data not shown) and PNS (Fig. 5 , lane b) contains 0-linked sugars that are specifically removed by 0-glycanase. The
146
Pedraza et al.
kD
I0571 a
b
C
d
Fig. 5. MAG contains 0- and N-linked oligosaccharides. Immunoprecipitates of MAG from sciatic nerve (postnatal day 15) were either blotted directly (lane a) or treated with O-glycanase (lane b) or endoglycosidase F (lane c ) . The results of chemical deglycosylation are shown in lane d. molecular weight of the O-linked sugars present on MAG corresponds to -5 kD.Consistent with the presence of both O-linked and N-linked oligosaccharides, after chemical deglycosylation (Fig. 5, lane d), a procedure that removes all carbohydrate residues, MAG migrates faster than after deglycosylation with either enzyme alone. Studies to determine whether the proportion of O-linked and N-linked sugars change during development in the PNS are in progress.
DISCUSSION The MAG are cell adhesion molecules on glial cells that mediate binding to certain axons. Like other members of the immunoglobulin gene superfamily (Williams and Barclay, 1988; Salzer and Colman, 1989) and most myelin-specific proteins (Campagnoni, 1988; Sutcliffe, 1987), MAG exists in alternatively spliced isoforms that are developmentally regulated. The functional significance of the distinct cytoplasmic domains of L- and SMAG is not yet known. Potential functions for these cytoplasmic segments may be to regulate the binding affinity(ies) of the ectodomain of MAG differentially or, alternatively, interact with distinct constituents of the cytoplasmic compartment of the glial cell. To further clarify the temporal and tissue specific expression of the individual MAG isoforms, we have compared their expression during development in the PNS and CNS using monospecific antibodies. These
studies have shown that L-MAG is a normal component of the forming myelin sheath in both the CNS and PNS and is therefore likely to have a common function in both cell types. These results confirm the recent observation (Owens et al., 1990), using immunofluorescence, that L-MAG is transiently expressed in Schwann cells forming myelin in vitro. At its peak, L-MAG constitutes only about 10% of the total MAG protein in the PNS. Thus, while these studies demonstrate for the first time the expression of L-MAG by Schwann cells in vivo, they are in agreement with earlier studies (Frail et al., 1985; Tropak et al., 1988) that reported that L-MAG is the major isoform at early stages of CNS myelination and that SMAG is the major isoform of PNS and mature CNS myelin. The studies reported here also indicate that LMAG turns over during maturation of the myelin sheath. We have also noted the expression of a highermolecular-weight variant of MAG that is present in the PNS at later developmental stages. The loss of this higher-molecular-weight band after chemical deglycosylation suggests that it represents additional glycosylation on a proportion of MAG molecules rather than a novel alternatively spliced isoform of MAG in the PNS. Thus, MAG, like many other cell adhesion molecules identified in the nervous system (Schachner, 1989), appears to be variably glycosylated during development, being composed of a heterogeneous population of molecules with differing amounts of carbohydrate residues, both O-linked and N-linked oligosaccharides. In other proteins, these different carbohydrate moieties may confer either novel adhesive characteristics on the proteins or modulate existing binding affinities (Salzer and Colman, 1989). In the PNS, but not the CNS, MAG is present in the Schmidt Lanterman incisures and paranodal loops (Trapp et al., 1989). Whether this glycosylation variant of PNS MAG has a unique function in these additional sites will require further investigation. In the studies reported here, we have analyzed the expression of MAG by populations of cells: Schwann cells in the sciatic nerve and oligodendrocytes in the brainstem. It is not yet known whether the two isoforms of MAG are expressed simultaneously within a single cell, as would seem likely, or are expressed separately by discrete populations of myelinating glial cells. The availability of polyclonal and monoclonal antibodies specific to L-MAG, and the development of antibodies specific to the S-MAG isoform (Fujita et al., 1990) should facilitate studies focused on whether the two isoforms are coexpressed by a single cell and if so, whether they are present in the same sites within that cell. One incidental finding of this study was that a monoclonal antibody, generated against human MAG, was found to be specific for the L-MAG isoform. In view of previous studies demonstrating that this antibody only
Developmental Expression of MAG Isoforms
reacts with human CNS but not with human PNS myelin (Nobile-Orazio et al., 1984), the tissue-specific pattern of isoform expression shown here for rats is likely to be similar in humans. Whether the temporal pattern of expression may be different in humans is not yet known. In addition, this L-MAG specific monoclonal antibody was reported not to bind to dMAG (Nobile-Orazio et al., 1984). dMAG is a degradation product of MAG that results from the action of a neutral protease present in myelin (Sato et al., 1984). It is present at elevated concentrations in the plaques of patients with multiple sclerosis (Moller et al., 1987). The demonstration that GEN S10 reacts with a cytoplasmic epitope of L-MAG, but not with dMAG, supports previous studies suggesting that dMAG results from the loss of a cytoplasmic segment of MAG (Quarles, 1989). The potentially distinct functions of the two isoforms of MAG have not been defined and hence the significance of the precise tissue and temporal pattern of their expression is as yet unknown. A possible function for L-MAG was, however, suggested by recent studies in which overexpression of this protein led to the accelerated segregation of nerve fibers destined to be myelinated (Owens et al., 1990). These results suggest that L-MAG may have a specific function in the initial wrapping of the nerve fiber and is therefore more abundantly expressed at the onset of myelination. In this model, the high level expression of L-MAG in the CNS at the onset of myelination could reflect the simultaneous ensheathment of multiple nerve fibers by a single oligodendrocyte (Remahl and Hildebrand, 1990), unlike the Schwann cell, which myelinates a single axon. Regulation of the abundance of L-MAG and of S-MAG by alternative splicing could therefore represent an important mechanism for regulating the initial axonal-glial interactions of myelination.
ACKNOWLEDGMENTS We thank N. Latov for the generous gift of the GEN S3, GEN S10, and HNK-1 antibodies; R. Milner and J. G. Sutcliffe for the P6 synthetic peptide; and J . Culkin and H. Plesken for assistance in the preparation of the illustrations. This work was supported by a Muscular Dystrophy Association fellowship to L. P., an Irma T. Hirschl Career Scientist Award to J. L. S . , and by NIH grants NS27680 to D. R. C. and NS26001 to J. L. S . REFERENCES Agrawal, HC, Noronha AB, Agrawal D, Quarles RH (1990): The myelin-associated glycoprotein is phosphorylated in the peripheral nervous system. Biochem Biophys Res Commun 169:953958. Arquint MJ, Roder J, Chia LS, Down J , Wilkinson D, Bayley H,
147
Braun P, Dunn R (1987): Molecular cloning and primary structure of myelin-associated glycoprotein. Proc Natl Acad Sci USA 84:600-604. Barton DE, Arquint M, Roder J, Dunn R, Francke U (1987): The myelin-associated glycoprotein gene: Mapping to human chromosome 19 and mouse chromosome 7 and expression in quivering mice. Genomics 1:107-112. Brunden KR, Poduslo JF (1987): Lysosomal delivery of the major myelin glycoprotein in the absence of myelin assembly: Posttranslational regulation of the level of expression by Schwann cells. J Cell Biol. 104:661-669. Campagnoni AT (1988): Molecular biology of myelin proteins from the central nervous system. J Neurochem 51:l-14. D’Eustachio P, Colman DR, Salzer JL (1988): The chromosomal location of the gene that encodes the myelin-associated glycoproteins. J Neurochem 50589-593. Frail DE, Braun PE (1984): Two developmentally regulated messenger RNAs differing in their coding regions may exist for the myelin-associated glycoprotein. J Biol Cheni 259: 1485714862. Frail DE, Webster HD, Braun PE (1985): Developmental expression of the myelin-associated glycoprotein in the peripheral nervous system is different from that in the central nervous system. J Neurochem 45: 1308-1 3 10. Fujita N, Sat0 S, Ishiguro H, Inuzuka T, Baba H, Kurihara T, Takahashi Y, Miyatake T (1990): The large isofotm of myelinassociated glycoprotein is scarcely expressed in the Quaking mouse brain. J Neurochem 55:1056-1059. Horvath E, Edwards AM, Bell JC, Braun PE (1989): Chemical degtycosylation on a micro-scale of membrane glycoproteins with retention of phosphoryl-protein linkages. J Neurosci Res 24: 398-401. Johnson PW, Abramow-Newerly W, Seilheimer B, Sadoul R, Tropak MB, Arquint M, Dunn RJ, Schachner M , Roder JD (1989): Recombinant myelin-associated glycoprotein confers neural adhesion and neurite outgrowth function. Neuron 3:377-385 Kruse J, Mailhammer R, Wernecke H, Faissner A, Sommer I , Goridis C, Schachner M ( 1984): Neural cell adhesion molecules and myelin-associated glycoprotein share a common carbohydrate moiety recognized by monoclonal antibodies L2 and HNK-I . Nature 31 1:153-155. Lai C, Brow MA, Nave K-A, Noronha AB, Quarles RH, Bloom FE, Milner RJ, Sutcliffe JG (1987): Two forms of IB236imyelin associated glycoprotein, a cell adhesion molecule for postnatal development, are produced by alternate splicing. Proc Natl Acad Sci USA 84:4337-4341. Lemke G (1988): Unwrapping the genes of myelin. Neuron 1 5 3 5 543. McGarry RC, Helfand SL, Quarles RH, Roder JC (1983): Recognition of myelin-associated glycoprotein by the monoclonal antibody HNK-I. Nature (Lond) 306:376-378. Moller JR, Yanagisawa K, Brady RO, Tourtellotte WW, Quarles RH (1987): Myelin-associated glycoprotein in multiple sclerosis lesions: A quantitative and qualitative analysis. Ann Neurol 22: 469-414. Nobile-Orazio E, Hays AP, Latov N, Perman G, Golier J , Shy ME, Freddo L (1984): Specificity of mouse and human monoclonal antibodies to myelin-associated glycoprotein. Neurology (NY) 34: 1336-1342. Noronha AB, Hammer JA, Lai C, Kiel M, Milner RJ, Sutcliffe JG, Quarles RH (1 989): Myelin-associated glycoprotein (MAG) and rat brain-specific 1B236 protein: Mapping of epitopes and demonstration of immunological identity. J Mol Neurosci I : 159-170.
148
Pedraza et al.
Owens GC, Boyd CJ, Bunge RP, Salzer JL (1990): Expression of recombinant myelin-associated glycoprotein in primary Schwann cells promotes the initial investment of axons by myelinating Schwann cells. J Cell Biol 111:1171-1182. Pedraza L, Owens GC, Green LAD, Salzer JL (1990): The myelinassociated glycoproteins: Membrane disposition, evidence of a novel disulfide linkage between immunoglobulin-like domains, and post-translational palmitylation. J Cell Biol 1 1 1:26512661. Poltorak M, Sadoul R, Keilhauer G, Landa C, Fahrig T, Schachner M (1987): Myelin-associated glycoprotein, a member of the L21 HNK-I family of neural cell adhesion molecules, is involved in neuron-oligodendrocyte and oligodendrocyte-oligodendrocyte interactions. J Cell Biol 105:1893-1899. Quarks RH (1983/1984): Myelin-associated glycoprotein in development and disease. Dev Neurosci 6:285-303. Quarks RH (1 989): Myelin-associated glycoprotein in demyelinating disorders. Crit Rev Neurobiol 5:l-28. Remahl S, Hildebrand C (1990): Relation between axons and oligodendroglial cells during initial myelination. I. The glial unit. J Neurocytol 19:3 13-328. Sadoul R, Fahrig T, Bartig U , Schachner M (1990): Binding properties of liposomes containing the myelin-associated glycoprotein MAG to neural cell cultures. J Neurosci Res 25:1-13. Salzer JL, Colman DR (1989): Mechanisms of cell adhesion in the nervous system: Role of the immunoglobulin gene superfamily. Dev Neurosci 11:377-390. Salzer JL, Holmes WP, Colman DR (1987): The amino acid sequences of the myelin-associated glycoproteins: Homology to the immunoglobulin gene superfamily . J Cell Biol 104:957965.
Sato S, Yanagisawa K, Miyatake T (1984): Conversion of myelinassociated glycoprotein (MAG) to a smaller derivative by calcium activated neutral protease (CANP)-like enzyme in myelin and inhibition by E-64 analogue. Neurochem Res 9:629-635. Schachner M (1989): Families of neural adhesion molecules. Ciba Found Symp 145: 156-169. Sternberger NH, Quarks RH, Itoyama Y , Webster HD (1979): Myelin-associated glycoprotein demonstrated immunocytochemically in myelin and myelin-forming cells of developing rat. Proc Natl Acad Sci USA 76:1510-1514. Sutcliffe JG (1987): The genes for myelin. TIG 3:73-76. Towbin H, Staehelin T, Gordon J (1979): Electrophoretic transfer of proteins from polyarcylamide gels to nitrocellulose sheets: Procedure and some applications. Proc Natl Acad Sci USA 76: 4350-4354. Trapp BD, Quarks RH (1982): Presence of the myelin-associated glycoprotein correlates with alterations in the periodicity of peripheral myelin. J Cell Biol 92877-882, Trdpp BD, Andrews SB, Cootauco C, Quarles R (1989): The myelinassociated glycoprotein is enriched in multivesicular bodies and periaxonal membranes of actively myelinating oligodendrocytes. J Cell Biol 109:2417-2426. Tropak MB, Johnson PW, Dunn RJ, Roder JC (1988): Differential splicing of MAG transcripts during CNS and PNS development. Brain Res 464:143-155. Williams AF, Barclay AN (1988): The immunoglobulin superfamily-Domains for cell surface recognition. Annu Rev Immunol 6:38 1-405. Wood JG, Engel EL (1976): Peripheral nerve glycoproteins and myelin fine structure during development of rat sciatic nerve. J Neurochem S:605-6 15.
Differential Expression of MAG Isoforms During Development L. Pedraza, A.B. Frey, B.L. Hempstead, D.R. Colman, and J.L. Salzer Departments of Cell Biology (L.P., A.B.F., J.L.S.) and Neurology (J.L.S.), New York University Medical School, New York; Division of Hematology, Cornell Medical Center, New York (B.L.H.); Department of Cell Biology, Columbia University Medical Center, New York (D.R.C.), New York The myelin-associated glycoproteins (MAG) mediate the cell interactions of oligodendrocytes and Schwann cells with axons that are myelinated. MAG exists in two developmentally regulated isoforms: large MAG (L-MAG) and small MAG (S-MAG). In this paper, we have studied the tissue-specific and developmentally regulated alternative splicing of these isoforms using monospecific antibodies that recognize epitopes common to both isoforms or that are present only on L-MAG. In the central nervous system (CNS), LMAG is the major form synthesized early in development, and it persists as a significant proportion of the MAG present in the adult. In the peripheral nervous system (PNS), L-MAG is expressed at modest levels during development; it is virtually absent in the adult. Thus, the expression of L-MAG is not limited to the CNS, as was formerly believed, suggesting that it plays a common role during the early stages of myelin formation by both oligodendrocytes and Schwann cells. In both the CNS and PNS, S-MAG is the predominant isoform in the adult. A higher-molecular-weight form of MAG is present in the PNS at low abundance, that is developmentally regulated, and appears to be a glycosylation variant. An analysis of the carbohydrate residues on MAG demonstrates that it contains both N-linked and 0-linked sugars that could be modulated during development. These results suggest a possible mechanism for the regulation of MAG function during myelinogenesis via the expression of alternative isoforms and carbohydrate modifications. Key words: MAG, myelin-associated glycoprotein, alternative splicing, immunoblotting, 0-linked, Nlinked
(Sternberger et al., 1979; Trapp and Quarles, 1982) and its ability to promote binding to axons in vitro (Poltorak et al., 1987; Sadoul et al., 1990). In agreement with its proposed function as a cell adhesion molecule, MAG is known to be a member of the immunoglobulin gene superfamily (Arquint et al., 1987; Lai et al., 1987; Salzer et al., 1987), a family of proteins that frequently function in tissue-specific cell interactions (Williams and Barclay, 1988; Salzer and Colman, 1989). MAG has recently been shown to promote the ensheathment and to enhance the outgrowth of large-caliber fibers in vitro (Johnson et al., 1989; Owens et a]., 1990). These results suggest that MAG has a role in the sorting and segregation of the subset of nerve fibers destined to become myelinated. Two isoforms of MAG have been identified (Frail and Braun, 1984) that differ in the sequences and size of their cytoplasmic domains (Lai et al., 1987; Salzer et al., 1987). These proteins, designated large MAG (L-MAG) and small MAG (S-MAG), correspond to core polypeptides of 67 and 62 kD, respectively (Pedraza et al., 1990), but comigrate as a broad 100-kD band as a result of extensive glycosylation. They are translated from an alternatively spliced transcript encoded by a single gene (Barton et al., 1987; D’Eustachio et al., 1988; Lai et al., 1987). The messenger RNA (mRNA) for these isoforms are differentially expressed during development and between the central nervous system (CNS) and peripheral nervous system (PNS) (Frail et al., 1985; Lai et al., 1987; Tropak et al., 1988). L-MAG is expressed early in development, during the rapid phase of myelination, whereas S-MAG is expressed later in development and has been thought to be the only isoform synthesized in the PNS. These findings have suggested that these proteins have distinct functions at different stages of myelination and, potentially, in oligodendrocytes vs. Schwann cells. If so, the regulation of the alternative splicing of
INTRODUCTION
with its presence in the periaxonal glial membrane 0 1991 Wiley-Liss, Inc.
y i r k , NY 10016
142
Pedraza et al.
the two isoforms of MAG could represent a mechanism for regulating the function of this protein during myelinogenesis. To date, the level of expression of the MAG isoforms has been inferred from the abundance of their corresponding mRNA, rather than by direct protein determinations. While the regulation of myelin protein abundance is believed to be largely transcriptional (Lemke, 1988), post-transcriptional mechanisms are also believed to be important (Brunden and Poduslo, 1987) and could differentially affect the accumulation of the MAG proteins. Thus, while small amounts of L-MAG mRNA have been detected in the PNS (Frail et al., 1985; Tropak et al., 1988), the protein has not (Agrawal et al., 1990; Noronha et al., 1989). This apparent discrepancy could reflect the transient expression of the L-MAG protein during peripheral nerve development, or alternatively post-transcriptional mechanisms that limit the translation of the L-MAG mRNA or the stability of the L-MAG protein. We have therefore directly analyzed the levels of L-MAG and total MAG protein in the CNS and PNS using monoclonal antibodies of defined specificity. We have found that L-MAG is a normal component of peripheral myelin during development, suggesting that it has a common function during the formation of myelin by Schwann cells and oligodendrocytes. Unlike its expression in the CNS, the L-MAG isoform is a minor isoform in the PNS at all stages of myelination, accounting for only 10% of total MAG at its peak. We have also found evidence for a developmentally regulated glycosylation variant of MAG that is present in the PNS but not in the CNS. Analysis of the carbohydrate residues on MAG demonstrated that it contains both O-linked and N-linked oligosaccharides that could be modulated during development. These results are considered with respect to the potential regulation of MAG function by expression of alternative isoforms and carbohydrate modifications.
sonicated in 2% sodium dodecyl sulfate (SDS), and boiled for 2 min. In each case, insoluble materials werepelleted at 13,OOOg for 10 min and discarded. Total cell lysates from 10' transfected cells expressing each of the insoforms of MAG were prepared as previously described (Pedraza et al., 1990). Aliquots (50 pg of protein) of tissue supernatents or membranes were fractionated on 8 -1 6% SDS-polyacrylamide gel electrophoresis (PAGE) gradient gels and blotted onto nitrocellulose (Towbin et al., 1979). Blotted samples were probed with anti-MAG monoclonal antibodies (I: 1,OOO), washed and incubated with a rabbit antimouse antibody at 1:500 (Cappel). MAG reactive bands were visualized by incubating the blots with [1251]proteinA (ICN Biochemicals) at 2 X lo5 c p d m l , washing, and then exposing the blots for autoradiography . Optical densitometry was performed with a Hoefer (GS300) scanning densitometer on the autoradiographs, to quantitate the level of MAG expression. In other studies using synthetic peptides, the peptides were dissolved in 1 N NaOH, and the solution was neutralized. Peptide, 1 pg, was spotted onto nitrocellulose, air dried, and fixed with a solution of 25% isopropyl alcohol and 10% acetic acid. Immunodetection with anti-MAG antibodies was performed as described above.
MATERIALS AND METHODS Antibodies The generation of antipeptide antibodies that recognize an epitope common to both isoforms or an epitope specific to L-MAG has previously been described (Pedraza et al., 1990). Ascites containing the HNK-l or the antihuman MAG monoclonal antibodies GEN S10 and S3 (Nobile-Orazio et al., 1984) were generous gifts of Dr. N. Latov.
RESULTS Characterization of Antibodies Specific for L-MAG We have determined the reactivity of an anti-MAG monoclonal antibody GEN S10, a mouse monoclonal generated against human MAG (Nobile-Orazio et al., 1984). GEN S10, like GEN S1, specifically recognizes CNS but not PNS MAG (Nobile-Orazio et al., 1984), suggesting that both antibodies specifically recognize LMAG but not S-MAG. To confirm this specificity, we have compared the reactivity of GEN S10 with the individual isoforms of MAG expressed in transfected fibroblast lines (Pedraza et al., 1990) (Fig. 1A) by Western blotting. As a control antibody, we have used GEN S3, a mouse monoclonal (Nobile-Orazio et al., 1984) that
Sample Preparation and Immunoblotting Procedures Sciatic nerves or brainstems were removed from postnatal rats, cut into approximately 3-mm fragments,
Enzymatic and Chemical Deglycosylation of MAG MAG was immunoprecipitated with an antipeptide antibody that recognizes both isoforms (Pedraza et al., 1990). Enzymatic deglycosylation was then performed using O-glycanase or endoglycosidase F (both from Boehringer Mannheim) at pH 6.5, following the manufacturer's instructions. Pretreatment of MAG with neuraminidase prior to O-glycanase did not alter the results. Chemical deglycosylation was done on total lysates of rat brainstem or sciatic nerve using trifluoromethanesulfonic acid as described (Horvath et a]., 1989).
Developmental Expression of MAG Isoforms
IA
18
143
trast, and as expected, an antipeptide antibody raised against the COOH terminal 21 amino acids of L-MAG only recognizes this peptide (lane b).
GEN SIODevelopmental Expression of the L-MAG Protein We have used the anti-MAG monoclonal antibodies to determine the developmental expression of LMAG and total MAG in the CNS and PNS. In quantita0 tive comparisons, selected ascites of GEN S3 and S10 were found to react equally well (+3%), on immunoblots, to equivalent amounts of L-MAG from transfected cells; the radioactivity was proportional to the quantity of MAG blotted (data not shown). Thus, the amount of total MAG (L-MAG plus S-MAG) vs. L-MAG could be comat! c d pared directly using GEN S3 and GEN S10, respectively . We have obtained essentially identical results to Fig. 1 . A. Reactivity of anti-MAG monoclonal antibodies. those shown here using antipeptide antibodies specific Total cell lysates from transfected Schwann cells expressing isoforms of S-MAG (lanes a, c) or L-MAG (lanes b, d) were for L-MAG or react with both isoforms (data not shown). The temporal expression of MAG in the CNS and blotted after SDS-PAGE. The anti-MAG monoclonals GEN S3 (lanes a, b) and GEN SlO (lanes c, d) were used as the the PNS (from postnatal day 1 through the adult) is primary antibodies. Immunoblots were then incubated with a shown in Figures 2 and 3, respectively. In the brainstem, secondary antibody followed by [12511proteinA and visualized MAG is first detectable at low levels at postnatal day 6. by antoradiography . B. Reactivity of L-MAG antibodies with Substantial levels of MAG appear at postnatal day 12, synthetic peptides. Two synthetic peptides, corresponding to peak at about 1 month, and remain relatively stable nonoverlapping sequences of the COOH terminus of L-MAG, thereafter (Fig. 2A). The stable levels of total MAG were spotted onto nitrocellulose. One peptide (amino acids protein were somewhat surprising, as MAG mRNA lev583-603) was spotted in lane a and the other (amino acids els are known to decline substantially after day 25 606-626) in lane b. GEN SlO and an antipeptide antibody (Salzer et al., 1987). Consistent with previous studies of (antiP2) directed against the COOH terminus, were reacted mRNA expression (Lai et al., 1987; Tropak et al., 1988), with the blot. the L-MAG protein is synthesized at the earliest stages of myelination in the CNS (Fig. 2B). At days 6-12, the L-MAG isoform accounts for virtually all the MAG was recently shown (Pedraza et al., 1990) to react with present in the brainstem (see Fig. 4A). Expression of an epitope of the second Ig-like domain of MAG com- L-MAG peaks at postnatal day 2.5 and thereafter demon to both S- and L-MAG (Fig. IA, lanes a and b, clines. It is also apparent that the size of MAG declines respectively). By contrast, GEN S10 reacts only with during development, shifting from 110 kD initially to L-MAG (compare lanes c and d) and therefore recog- 105 kD in the adult (Fig. 2A). However, no shift in the nizes an epitope present at the COOH terminus specific molecular weight of L-MAG is evident during developto this isoform (the terminal 53 amino acids). The reac- ment (Fig. 2B). Thus, the decrease in the size of MAG tivity of GEN S10 for L-MAG is of interest because this during CNS development appears to reflect the transition antibody does not react with human PNS MAG or from L-MAG to S-MAG rather than a change in the dMAG (see Discussion). amount of glycosylation of MAG. There are 53 amino acids at the COOH terminus of In the PNS, MAG is expressed immediately after L-MAG that are unique to this isoform (amino acids birth, accumulates until approximately postnatal day 9, 574-626) (Salzer et al., 1987). To further define the and remains stable thereafter (Fig. 3A). L-MAG is sequence recognized by GEN S10, we determined its readily detected beginning on postnatal day 3 , is most reactivity with two synthetic peptides corresponding to abundant at days 6-9, and persists at low but detectable amino acids 583-603 or 606-626 (the COOH terminus) levels in the adult (Fig. 3B). Proportionately, L-MAG of the published L-MAG sequence (Lai et al., 1987; expression is maximal at days 6-9, as previously noted Salzer et al., 1987). As shown in Figure lB, we found for its mRNA expression (Tropak et al., 1988), a period that GEN S 10 specifically reacts with a synthetic peptide corresponding to the time of most rapid myelination in corresponding to amino acids 583-603 (lane a), but not the rat sciatic nerve (Wood and Engel, 1976). At its with amino acid sequence 606-626 (lane b). By con- relative peak (day 6), L-MAG accounts for an estimated
anti P2
-
-
144
Pedraza et al.
2A
I
3 6 9 M 15 19 2 5 4 0 A
Fig. 2 . Temporal expression of MAG in the CNS. Samples of brainstem from postnatal rats from the ages indicated were blotted and probed with GEN S3 (A), which recognizes both isoforms of MAG, and GEN S10 (B), which is specific for L-MAG. Blots were then incubated with a secondary antibody
28
I 3 6 9
and ['2sI]protein A. In both cases, autoradiographs were exposed overnight. Ages of animals in postnatal days are indicated (A = adult). Small amounts of MAG were seen at days 6 and 9 on longer exposures than that shown here.
3B
3A
I
3
6
9 12 15 19 2540A
12 15 19 2 5 4 0 A
I
3 6 9 12 1 5 1 9 2 5 4 0 A
Fig. 3. Temporal expression of MAG in the PNS. Rat sciatic nerve preparations from postnatal rats of the indicated ages were analyzed by SDS-PAGE and immunoblotting. Total MAG was visualized by GEN S3 (A), and L-MAG was visu-
alized by GEN S10 (B). Exposure times of the autoradiographs were overnight (A) and 5 days (B). The arrow in B indicates the position of the higher-molecular-weight form of PNS MAG.
10% of the MAG in sciatic nerve. Thus, L-MAG is a normal component of PNS myelin and, as in the CNS, it is expressed at times of rapid myelination. In sharp contrast to the CNS however, L-MAG is always a minor isoform, and in the adult comprises < 1 % of the total MAG. These results are presented graphically in Figure 4. By subtracting the amount of L-MAG from the total MAG, we have estimated the level of S-MAG expressed in the brainstem and in the sciatic nerve during development. In the CNS, S-MAG is first present at about day 15; levels of S-MAG probably gradually increase thereafter, reaching about 80% of the total MAG present in the adult brainstem. In parallel, the amount of L-MAG present in the brainstem declines, even though total MAG levels are quite stable. In the PNS, S-MAG expression and total MAG expression are closely correlated at all time points. The declining levels of L-MAG, particularly evident in the CNS after day 40 (Fig. 4A), suggest that this isoform of MAG turns over during late stages of development.
Expression of a Developmentally Regulated Variant of MAG in the PNS Also apparent (Fig. 3B, arrow) is an apparently discrete, higher-molecular-weight form of MAG of 110 kD. This band is not a cross-reactive protein of similar molecular weight, as it has been detected by four different anti-MAG antibodies: two antipeptide antibodies and two monoclonal antibodies. This MAG band becomes maximal at postnatal days 15-19 in the sciatic nerve and persists at comparable levels at all subsequent time points, including the adult. It is present on the L-MAG polypeptide backbone and, on the basis of its abundance, may be present on the S-MAG polypeptide as well. Interestingly, the small amount of L-MAG detectable in the adult corresponds exclusively to this higher-molecular-weight form. Immunoprecipitation of L-MAG from the adult demonstrated that it migrates at a somewhat higher molecular weight than does L-MAG from early postnatal time points (data not shown). We were interested in whether this higher-molecu-
Developmental Expression of MAG Isoforms 3000-
145
A S-MAG
2500 -
--c L-MAG
2000 1500-
1000-
500
-
-
25001
B
2000 -
-w-
S-MAG
--t
L-MAG
1 500 -
1000 500 -,
0
0
5
10
15
20
25
30
35
40
A
Fig. 4. Temporal and tissue-specific expression of MAG isoforms. Autoradiographs, such as those shown in Figures 2 and 3 , were quantitated by densitometry, and the values from two separate experiments (including correction for length of exposure) were averaged. Total MAG and L-MAG were measured
directly. S-MAG values shown are calculated by subtracting L-MAG values from total MAG. (A) Expression of MAG isoforms in the brainstem. (B) Expression of MAG in the sciatic nerve. Age of the sample is shown below (A = adult); units at the left are arbitrary densitometric values.
lar-weight MAG species might represent an alternatively spliced isoform of MAG specific to the PNS or whether it is a glycosylation variant. We chemically deglycosylated MAG from day 15 sciatic nerve prior to blotting (Fig. 5 ) . After chemical deglycosylation, PNS MAG migrates as a single band with a molecular weight of -65 kD (Fig. 5 , lane d), consistent with the size of the SMAG polypeptide, which predominates at this time point. Similarly, chemically deglycosylated L-MAG (visualized with the S 10 antibody) from sciatic nerves of different ages migrates with a uniform molecular weight of -70 kD (data not shown). These results strongly suggest that the higher-molecular-weight band present in the PNS is a glycosylation variant and not a novel alternatively spliced isoform of MAG. Because the HNK-1 epitope is a carbohydrate epitope known to be present on a proportion of MAG molecules in some animal species (Kruse et al., 1984;
McGany et al., 1983), we examined whether the highermolecular-weight band present in the PNS reacted with this antibody. However, neither the upper nor lower MAG bands reacted with HNK-1 (data not shown).
MAG Contains 0-Linked and N-Linked Oligosaccharides As part of an analysis of the nature of the glycosyl residues present on PNS MAG, including the nature of the higher-molecular-weight MAG species, we investigated the oligosaccharides present on MAG. Consistent with our earlier studies (Salzer et al., 1987), we found that most of thexarbohydrate residues on MAG are Nlinked, as evidenced by the sharp decrease in the size of the protein after endoglycosidase F treatment (Fig. 5, lane c). In addition, MAG from both the CNS (data not shown) and PNS (Fig. 5 , lane b) contains 0-linked sugars that are specifically removed by 0-glycanase. The
146
Pedraza et al.
kD
I0571 a
b
C
d
Fig. 5. MAG contains 0- and N-linked oligosaccharides. Immunoprecipitates of MAG from sciatic nerve (postnatal day 15) were either blotted directly (lane a) or treated with O-glycanase (lane b) or endoglycosidase F (lane c ) . The results of chemical deglycosylation are shown in lane d. molecular weight of the O-linked sugars present on MAG corresponds to -5 kD.Consistent with the presence of both O-linked and N-linked oligosaccharides, after chemical deglycosylation (Fig. 5, lane d), a procedure that removes all carbohydrate residues, MAG migrates faster than after deglycosylation with either enzyme alone. Studies to determine whether the proportion of O-linked and N-linked sugars change during development in the PNS are in progress.
DISCUSSION The MAG are cell adhesion molecules on glial cells that mediate binding to certain axons. Like other members of the immunoglobulin gene superfamily (Williams and Barclay, 1988; Salzer and Colman, 1989) and most myelin-specific proteins (Campagnoni, 1988; Sutcliffe, 1987), MAG exists in alternatively spliced isoforms that are developmentally regulated. The functional significance of the distinct cytoplasmic domains of L- and SMAG is not yet known. Potential functions for these cytoplasmic segments may be to regulate the binding affinity(ies) of the ectodomain of MAG differentially or, alternatively, interact with distinct constituents of the cytoplasmic compartment of the glial cell. To further clarify the temporal and tissue specific expression of the individual MAG isoforms, we have compared their expression during development in the PNS and CNS using monospecific antibodies. These
studies have shown that L-MAG is a normal component of the forming myelin sheath in both the CNS and PNS and is therefore likely to have a common function in both cell types. These results confirm the recent observation (Owens et al., 1990), using immunofluorescence, that L-MAG is transiently expressed in Schwann cells forming myelin in vitro. At its peak, L-MAG constitutes only about 10% of the total MAG protein in the PNS. Thus, while these studies demonstrate for the first time the expression of L-MAG by Schwann cells in vivo, they are in agreement with earlier studies (Frail et al., 1985; Tropak et al., 1988) that reported that L-MAG is the major isoform at early stages of CNS myelination and that SMAG is the major isoform of PNS and mature CNS myelin. The studies reported here also indicate that LMAG turns over during maturation of the myelin sheath. We have also noted the expression of a highermolecular-weight variant of MAG that is present in the PNS at later developmental stages. The loss of this higher-molecular-weight band after chemical deglycosylation suggests that it represents additional glycosylation on a proportion of MAG molecules rather than a novel alternatively spliced isoform of MAG in the PNS. Thus, MAG, like many other cell adhesion molecules identified in the nervous system (Schachner, 1989), appears to be variably glycosylated during development, being composed of a heterogeneous population of molecules with differing amounts of carbohydrate residues, both O-linked and N-linked oligosaccharides. In other proteins, these different carbohydrate moieties may confer either novel adhesive characteristics on the proteins or modulate existing binding affinities (Salzer and Colman, 1989). In the PNS, but not the CNS, MAG is present in the Schmidt Lanterman incisures and paranodal loops (Trapp et al., 1989). Whether this glycosylation variant of PNS MAG has a unique function in these additional sites will require further investigation. In the studies reported here, we have analyzed the expression of MAG by populations of cells: Schwann cells in the sciatic nerve and oligodendrocytes in the brainstem. It is not yet known whether the two isoforms of MAG are expressed simultaneously within a single cell, as would seem likely, or are expressed separately by discrete populations of myelinating glial cells. The availability of polyclonal and monoclonal antibodies specific to L-MAG, and the development of antibodies specific to the S-MAG isoform (Fujita et al., 1990) should facilitate studies focused on whether the two isoforms are coexpressed by a single cell and if so, whether they are present in the same sites within that cell. One incidental finding of this study was that a monoclonal antibody, generated against human MAG, was found to be specific for the L-MAG isoform. In view of previous studies demonstrating that this antibody only
Developmental Expression of MAG Isoforms
reacts with human CNS but not with human PNS myelin (Nobile-Orazio et al., 1984), the tissue-specific pattern of isoform expression shown here for rats is likely to be similar in humans. Whether the temporal pattern of expression may be different in humans is not yet known. In addition, this L-MAG specific monoclonal antibody was reported not to bind to dMAG (Nobile-Orazio et al., 1984). dMAG is a degradation product of MAG that results from the action of a neutral protease present in myelin (Sato et al., 1984). It is present at elevated concentrations in the plaques of patients with multiple sclerosis (Moller et al., 1987). The demonstration that GEN S10 reacts with a cytoplasmic epitope of L-MAG, but not with dMAG, supports previous studies suggesting that dMAG results from the loss of a cytoplasmic segment of MAG (Quarles, 1989). The potentially distinct functions of the two isoforms of MAG have not been defined and hence the significance of the precise tissue and temporal pattern of their expression is as yet unknown. A possible function for L-MAG was, however, suggested by recent studies in which overexpression of this protein led to the accelerated segregation of nerve fibers destined to be myelinated (Owens et al., 1990). These results suggest that L-MAG may have a specific function in the initial wrapping of the nerve fiber and is therefore more abundantly expressed at the onset of myelination. In this model, the high level expression of L-MAG in the CNS at the onset of myelination could reflect the simultaneous ensheathment of multiple nerve fibers by a single oligodendrocyte (Remahl and Hildebrand, 1990), unlike the Schwann cell, which myelinates a single axon. Regulation of the abundance of L-MAG and of S-MAG by alternative splicing could therefore represent an important mechanism for regulating the initial axonal-glial interactions of myelination.
ACKNOWLEDGMENTS We thank N. Latov for the generous gift of the GEN S3, GEN S10, and HNK-1 antibodies; R. Milner and J. G. Sutcliffe for the P6 synthetic peptide; and J . Culkin and H. Plesken for assistance in the preparation of the illustrations. This work was supported by a Muscular Dystrophy Association fellowship to L. P., an Irma T. Hirschl Career Scientist Award to J. L. S . , and by NIH grants NS27680 to D. R. C. and NS26001 to J. L. S . REFERENCES Agrawal, HC, Noronha AB, Agrawal D, Quarles RH (1990): The myelin-associated glycoprotein is phosphorylated in the peripheral nervous system. Biochem Biophys Res Commun 169:953958. Arquint MJ, Roder J, Chia LS, Down J , Wilkinson D, Bayley H,
147
Braun P, Dunn R (1987): Molecular cloning and primary structure of myelin-associated glycoprotein. Proc Natl Acad Sci USA 84:600-604. Barton DE, Arquint M, Roder J, Dunn R, Francke U (1987): The myelin-associated glycoprotein gene: Mapping to human chromosome 19 and mouse chromosome 7 and expression in quivering mice. Genomics 1:107-112. Brunden KR, Poduslo JF (1987): Lysosomal delivery of the major myelin glycoprotein in the absence of myelin assembly: Posttranslational regulation of the level of expression by Schwann cells. J Cell Biol. 104:661-669. Campagnoni AT (1988): Molecular biology of myelin proteins from the central nervous system. J Neurochem 51:l-14. D’Eustachio P, Colman DR, Salzer JL (1988): The chromosomal location of the gene that encodes the myelin-associated glycoproteins. J Neurochem 50589-593. Frail DE, Braun PE (1984): Two developmentally regulated messenger RNAs differing in their coding regions may exist for the myelin-associated glycoprotein. J Biol Cheni 259: 1485714862. Frail DE, Webster HD, Braun PE (1985): Developmental expression of the myelin-associated glycoprotein in the peripheral nervous system is different from that in the central nervous system. J Neurochem 45: 1308-1 3 10. Fujita N, Sat0 S, Ishiguro H, Inuzuka T, Baba H, Kurihara T, Takahashi Y, Miyatake T (1990): The large isofotm of myelinassociated glycoprotein is scarcely expressed in the Quaking mouse brain. J Neurochem 55:1056-1059. Horvath E, Edwards AM, Bell JC, Braun PE (1989): Chemical degtycosylation on a micro-scale of membrane glycoproteins with retention of phosphoryl-protein linkages. J Neurosci Res 24: 398-401. Johnson PW, Abramow-Newerly W, Seilheimer B, Sadoul R, Tropak MB, Arquint M, Dunn RJ, Schachner M , Roder JD (1989): Recombinant myelin-associated glycoprotein confers neural adhesion and neurite outgrowth function. Neuron 3:377-385 Kruse J, Mailhammer R, Wernecke H, Faissner A, Sommer I , Goridis C, Schachner M ( 1984): Neural cell adhesion molecules and myelin-associated glycoprotein share a common carbohydrate moiety recognized by monoclonal antibodies L2 and HNK-I . Nature 31 1:153-155. Lai C, Brow MA, Nave K-A, Noronha AB, Quarles RH, Bloom FE, Milner RJ, Sutcliffe JG (1987): Two forms of IB236imyelin associated glycoprotein, a cell adhesion molecule for postnatal development, are produced by alternate splicing. Proc Natl Acad Sci USA 84:4337-4341. Lemke G (1988): Unwrapping the genes of myelin. Neuron 1 5 3 5 543. McGarry RC, Helfand SL, Quarles RH, Roder JC (1983): Recognition of myelin-associated glycoprotein by the monoclonal antibody HNK-I. Nature (Lond) 306:376-378. Moller JR, Yanagisawa K, Brady RO, Tourtellotte WW, Quarles RH (1987): Myelin-associated glycoprotein in multiple sclerosis lesions: A quantitative and qualitative analysis. Ann Neurol 22: 469-414. Nobile-Orazio E, Hays AP, Latov N, Perman G, Golier J , Shy ME, Freddo L (1984): Specificity of mouse and human monoclonal antibodies to myelin-associated glycoprotein. Neurology (NY) 34: 1336-1342. Noronha AB, Hammer JA, Lai C, Kiel M, Milner RJ, Sutcliffe JG, Quarles RH (1 989): Myelin-associated glycoprotein (MAG) and rat brain-specific 1B236 protein: Mapping of epitopes and demonstration of immunological identity. J Mol Neurosci I : 159-170.
148
Pedraza et al.
Owens GC, Boyd CJ, Bunge RP, Salzer JL (1990): Expression of recombinant myelin-associated glycoprotein in primary Schwann cells promotes the initial investment of axons by myelinating Schwann cells. J Cell Biol 111:1171-1182. Pedraza L, Owens GC, Green LAD, Salzer JL (1990): The myelinassociated glycoproteins: Membrane disposition, evidence of a novel disulfide linkage between immunoglobulin-like domains, and post-translational palmitylation. J Cell Biol 1 1 1:26512661. Poltorak M, Sadoul R, Keilhauer G, Landa C, Fahrig T, Schachner M (1987): Myelin-associated glycoprotein, a member of the L21 HNK-I family of neural cell adhesion molecules, is involved in neuron-oligodendrocyte and oligodendrocyte-oligodendrocyte interactions. J Cell Biol 105:1893-1899. Quarks RH (1983/1984): Myelin-associated glycoprotein in development and disease. Dev Neurosci 6:285-303. Quarks RH (1 989): Myelin-associated glycoprotein in demyelinating disorders. Crit Rev Neurobiol 5:l-28. Remahl S, Hildebrand C (1990): Relation between axons and oligodendroglial cells during initial myelination. I. The glial unit. J Neurocytol 19:3 13-328. Sadoul R, Fahrig T, Bartig U , Schachner M (1990): Binding properties of liposomes containing the myelin-associated glycoprotein MAG to neural cell cultures. J Neurosci Res 25:1-13. Salzer JL, Colman DR (1989): Mechanisms of cell adhesion in the nervous system: Role of the immunoglobulin gene superfamily. Dev Neurosci 11:377-390. Salzer JL, Holmes WP, Colman DR (1987): The amino acid sequences of the myelin-associated glycoproteins: Homology to the immunoglobulin gene superfamily . J Cell Biol 104:957965.
Sato S, Yanagisawa K, Miyatake T (1984): Conversion of myelinassociated glycoprotein (MAG) to a smaller derivative by calcium activated neutral protease (CANP)-like enzyme in myelin and inhibition by E-64 analogue. Neurochem Res 9:629-635. Schachner M (1989): Families of neural adhesion molecules. Ciba Found Symp 145: 156-169. Sternberger NH, Quarks RH, Itoyama Y , Webster HD (1979): Myelin-associated glycoprotein demonstrated immunocytochemically in myelin and myelin-forming cells of developing rat. Proc Natl Acad Sci USA 76:1510-1514. Sutcliffe JG (1987): The genes for myelin. TIG 3:73-76. Towbin H, Staehelin T, Gordon J (1979): Electrophoretic transfer of proteins from polyarcylamide gels to nitrocellulose sheets: Procedure and some applications. Proc Natl Acad Sci USA 76: 4350-4354. Trapp BD, Quarks RH (1982): Presence of the myelin-associated glycoprotein correlates with alterations in the periodicity of peripheral myelin. J Cell Biol 92877-882, Trdpp BD, Andrews SB, Cootauco C, Quarles R (1989): The myelinassociated glycoprotein is enriched in multivesicular bodies and periaxonal membranes of actively myelinating oligodendrocytes. J Cell Biol 109:2417-2426. Tropak MB, Johnson PW, Dunn RJ, Roder JC (1988): Differential splicing of MAG transcripts during CNS and PNS development. Brain Res 464:143-155. Williams AF, Barclay AN (1988): The immunoglobulin superfamily-Domains for cell surface recognition. Annu Rev Immunol 6:38 1-405. Wood JG, Engel EL (1976): Peripheral nerve glycoproteins and myelin fine structure during development of rat sciatic nerve. J Neurochem S:605-6 15.
