Detection and processing of peripheral myelin protein PMP22 in cultured Schwann cells.

Peripheral myelin protein, 22 kDa (PMP22), is a myelin molecule associated with Schwann cells in peripheral nerves (Snipes, G. J., Suter, U., Welcher, A. A., and Shooter, E. M. (1992) J. Cell Biol. 117, 225-238). Mutations affecting the PMP22 gene have been implicated in the trembler mutation in mice (Suter, U., Welcher, A. A., Ozcelik, T., Snipes, G. J., Kosaras, B., Francke, U., Billings-Gagliardi, S., Sidman, R. L., and Shooter, E. M. (1992) Nature 356, 241-244; Suter, U., Moskow, J. J., Welcher, A. A., Snipes, G. J., Kosaras, B., Sidman, R. L., Buchberg, A. M., and Shooter, E. M. (1992) Proc. Natl. Acad. Sci. U. S. A. 89, 4382-4386) and Charcot-Marie-Tooth Disease in humans (Patel, P. I., Roa, B. B., Welcher, A. A., Schoener-Scott, R., Trask, B. J., Pentao, L., Snipes, G. J., Garcia, C. A., Francke, U., Shooter, E. M., Lupski, J. R., and Suter, U. (1992) Nature genet. 1, 159-165). In this report, we have studied PMP22 production in cultured rat Schwann cells. Schwann cells contain a 1.8-kilobase mRNA transcript coding for PMP22, and its production is up-regulated in vitro by forskolin. Metabolic labeling combined with immunoprecipitation methods using antibodies raised against synthetic peptides of PMP22 reveal that Schwann cells generate the protein from an 18-kDa precursor form which is post-translationally modified by N-linked glycosylation. A second molecule (molecular mass, 48 kDa) that reacted with PMP22 antibodies was also detected in Schwann cells but is not related chemically to PMP22 as determined by pulse-chase labeling. Metabolic labeling of rat sciatic nerve and Western blot analyses of purified rat sciatic nerve myelin reveal that deglycosylation of PMP22 gives rise to an 18-kDa protein similar in size to that in Schwann cells. These results indicate that cultured Schwann cells may provide a good model in which to investigate the production and function of PMP22 and to establish the cellular basis for the protein's involvement in inherited peripheral neuropathies.

Scott, R., Trask, B. J., Pentao, L., Snipes, G. J., Garcia, C. A., Francke, U., Shooter, E. M., Lupski, J. R.,  Nature genet. 1,[159][160][161][162][163][164][165]. In this report, we have studied PMP22 production in cultured rat Schwann cells. Schwann cells contain a 1.8-kilobase mRNA transcript coding for PMP22, and its production is up-regulated in vitro by forskolin. Metabolic labeling combined with immunoprecipitation methods using antibodies raised against synthetic peptides of PMP22 reveal that Schwann cells generate the protein from an 18-kDa precursor form which is post-translationally modified by N-linked glycosylation. A second molecule (molecular mass, 48 kDa) that reacted with PMP22 antibodies was also detected in Schwann cells but is not related chemically to PMP22 as determined by pulse-chase labeling. Metabolic labeling of rat sciatic nerve and Western blot analyses of purified rat sciatic nerve myelin reveal that deglycosylation of PMP22 gives rise to an 18-kDa protein similar in size  : 514-398-1903;Fax: 514-398-8248. to that in Schwann cells. These results indicate that cultured Schwann cells may provide a good model in which to investigate the production and function of PMP22 and to establish the cellular basis for the protein's involvement in inherited peripheral neuropathies.
Peripheral nerve myelin is formed by the ensheathment and compaction of Schwann cell processes around competent axons (Morel1 et al., 1989). Several major structural myelin proteins have been identified in compact myelin, including protein zero (Po),' myelin basic protein (MBP), and myelinassociated glycoprotein. Recently, an additional member of this list, termed peripheral myelin protein, 22 kDa (PMP22), has been identified Snipes et al., 1992).
The protein is associated structurally with Schwann cell membranes and, like other proteins of compact myelin, is produced when myelin is formed . Levels of PMP22 mRNA and protein in peripheral nerves decline when myelin is disrupted as a result of nerve crush injuries and increase when axons regenerate Welcher et al., 1991;Spreyer et al., 1991;Snipes et al., 1992).
The biological role of PMP22 is not known but the protein appears to play an essential role in peripheral nerve function.
The trembler (Tr) mouse mutant carries an autosomal dominant point mutation in the PMP22 gene (Suter et al., 1992a) and a second point mutation affecting PMP22 has been found in the allelic trembler-J (Tr') mouse (Suter et al., 1992b). Tr and T# mice show limb paralysis, tremor, transient seizures, as well as severe peripheral nervous system-specific hypomyelination and continuous Schwann cell proliferation. In humans, the gene coding for PMP22 has been directly implicated in the most common form (1 in 2,500 individuals; Skre, 1974) of inherited human peripheral neuropathy, Charcot-Marie-Tooth disease type 1A (CMTlA; Patel et al., 1992;Valentijn et al., 1992;Timmerman et al., 1992;Matsunami et al., 1992).
The majority of CMTlA cases are associated with a chromosomal duplication involving -1.5 megabases on the short arm of human chromosome 17 (Lupski et al., 1991;Raeymaekers et al., 1991;Hoogendijk et al., 1991). The PMP22 gene maps to the CMTlA duplication region, but is not interrupted The abbreviations used are: Po, protein zero; CMTlA, Charcot-Marie-Tooth disease type 1A; DMEM, Dulbecco's modified Eagle's medium; FCS, fetal calf serum; GGF, crude glial growth factor; HBS, Hepes-buffered saline; MBP, myelin basic proteins; PMP22, peripheral myelin protein, 22 kilodaltons; PBS, phosphate-buffered saline; PIPES, 1,4-piperazinediethanesulfonic acid; PAGE, polyacrylamide gel electrophoresis; Tr, trembler; TrJ, trembler-J. (Patel et al., 1992), suggesting that a gene dosage effect involving PMP22 might be responsible for the demyelinating neuropathy that occurs in CMTlA patients (Patel et al., 1992). Understanding how these two putative mutational mechanisms, point mutations in T r and Tr" and gene duplication in CMTlA, lead to peripheral nervous system myelin deficiencies will require an understanding of the biological function(s) of PMP22. As a first step toward this goal, we have sought to establish an appropriate i n vitro system in which we can study the biology of PMP22.
In this study, we demonstrate that PMP22 is produced by cultured Schwann cells and that its production, like that of other myelin proteins, can be regulated by forskolin which elevates intracellular CAMP concentrations. Using metabolic labeling and immunoprecipitation methods, we have determined that cultured Schwann cells produce PMP22 in an Nlinked glycosylated form that arises from an 18-kDa precursor which can also be detected in sciatic nerve. In the course of these studies, we also detected in cultured Schwann cells, but not in sciatic nerve, a second protein (molecular mass, 48 kDa) that is distinct from but immunologically related to PMP22 in that it can be precipitated with an antipeptide antibody. Our data suggest that cultured Schwann cells may be a useful model for studying the production, regulation, and biological functions of PMP22, information that will be essential for understanding its role in peripheral nerve disease.
Antibodies-Antibodies raised in rabbits against synthetic peptides containing amino acid residues 27-42 (peptide 1) and 117-132 (peptide 2) of rat PMP22 were prepared as described previously . These peptides are predicted from hydrophobicity plots and surface probability and secondary structure predictions to be located on the extracellular portion of this membrane-associated protein. Peptide 1 contains a consensus sequence for N-linked glycosylation at Asn-41.
Immunocytochemistry-Schwann cells grown on poly-L-lysinecoated glass coverslips in 35-mm dishes were rinsed with phosphatebuffered saline (PBS) and fixed in modified Bouin's solution (2% paraformaldehyde, 5% picric acid, 5% sucrose in 0.2 M phosphate buffer) for 20 min at room temperature. Cells were preincubated 20 min in Hepes-buffered saline (HBS) containing 10% FCS to reduce nonspecific antibody binding and treated with antiserum to peptide 2 (diluted 1:lOO in HBS/FCS) or with rabbit IgG to bovine SI00 W O O ; Dakopatts) for 1 h at room temperature. Cells were washed three times with HBS (5 min each), incubated for 1 h at room temperature with fluorescein-conjugated donkey anti-rabbit IgG (Jackson ImmunoResearch Laboratories, Inc.; diluted 1:100 in HBS containing 10% donkey serum), washed three times in HBS, mounted in a Tris-buffered glycerol mountant containing N-propyl gallate (Sigma), examined on a Zeiss Axioskop, and photographed on TMAX film (Kodak). Some cultures were treated with peptide 2 antiserum that had been preincubated 1 h with high concentration (1 mg/ml) of peptide 2 to ensure staining specificity.
Metabolic Labeling of Schwann Cells-Nearly confluent cultures of Schwann cells grown on 100-mm plates were washed twice with prewarmed methionine-free DMEM, incubated with the same medium for 20 min at 37 'C followed by prewarmed methionine-free DMEM (2.5 ml) containing 10% dialyzed FCS, 10 p~ forskolin, and 0.4 mCi/ml [35S]methionine (1100 Ci/mmol; ICN). Medium was removed after 4 h and the cells washed twice with PBS. In pulsechase studies, cells were incubated in medium containing [=S]methionine for 30 min, washed twice with prewarmed DMEM not containing isotope, and incubated in fresh complete medium (DMEM, 10% FCS, 100 pg\ml GGF, and 10 p~ forskolin) for periods up to 5 h. In some experiments, cells were preincubated in medium containing 10 pg/ml tunicamycin for 1.5 h prior to receiving isotope and incubated in the same tunicamycin concentration during the labeling period.
Metabolic Labeling of Sciatic Nerves-Ten-day-old Sprague-Dawley rats, supplied by the farm of the University of Alberta (Edmonton, Alberta, Canada), were sacrificed and sciatic nerves of four littermates pooled, immersed in ice-cold DMEM, cut into small pieces, and incubated at 37 "C for 3 h in DMEM containing 10% dialyzed FCS and 0.4 mCi/ml [35S]methionine. In some experiments, nerve pieces were preincubated with medium containing 25 pg/ml tunicamycin for 1.5 h, and the same concentration of the drug was used in the labeling medium. Following the labeling period, the tissue was centrifuged, the pellet was washed once with cold PBS, and frozen at -80 "C before further processing.
Electrophoresis-Samples from metabolic labeling studies were analyzed by 13-22% gradient SDS-polyacrylamide gel electrophoresis (SDS-PAGE). Gels were fixed in 40% methanol and 10% acetic acid and treated with ENHANCE (Du Pont-New England Nuclear) according to the manufacturers instructions. Dried gels were exposed to Kodak XAR film at -80 "C and developed on an X-Omat film processor.
RNase Protection Analysis-Total RNA was isolated from nearly confluent Schwann cell cultures and neonatal rat sciatic nerves using the guanidine hydrochloride method (Chomczynski and Sacchi, 1987). Total RNA (2.5 pg) was combined with approximately IO6 cpm of 32P-labeled cRNA probe in 30 p1 of hybridization buffer containing 80% formamide, 40 mM PIPES, pH 6.4,0.4 M NaCl, and 1 mM EDTA, heated to 95 "C for 15 min, and transferred to a 55 "C water bath for 16 h. Samples were supplemented with 350 pl of RNase digestion buffer (10 mM Tris-HC1, pH 7.5, 300 mM NaC1, and 5 mM EDTA) containing 40 pg/ml RNase A (Boehringer Mannheim) and 1200 units/ml RNase T1 (Bethesda Research Laboratories) and incubated for 30 min at 30 "C. SDS and proteinase K were added to final concentrations of 0.4% and 100 pg/ml, respectively and tubes were incubated at 37 "C for 20 min. The samples were phenol/chloroformextracted, precipitated with isopropanol in the presence of 40 pg of carrier yeast tRNA, and analyzed on denaturing acrylamide gels (6% acrylamide, 8 M urea, 1 X TBE (Tris-borate-EDTA buffer)).
Hybridization Probes-For RNase protection assays, a 356-base pair BamHI/HincII fragment containing nucleotides 143-499 of the rat PMP22 cDNA  was subcloned into the pGEM-3 vector (Promega), and radiolabeled antisense cRNA probes were transcribed with SP6 polymerase (Bethesda Research Laboratories) and ["PICTP (800 Ci/mmol; Du Pont-New England Nuclear) as described previously (Melton et al., 1984). Template DNA was removed with 5 units RNase-free DNase (Boehringer Mannheim) in the presence of 20 units RNasin (Promega) in a total volume of 25 pl. Reaction mixtures were extracted once with phenol/chloroform and ethanol-precipitated. The probe was electrophoresed on a denaturing acrylamide gel (6% acrylamide, 8 M urea, 1 X TBE) and exposed to XAR film for 1 min; the full-length RNA band was identified and excised, using the film as a template, and eluted in 400 pl of elution buffer (2 M ammonium acetate, 1% SDS, 50 pg/ml yeast tRNA) for 2 h at 37 "C. RNA was ethanol-precipitated and resuspended in 100 p1 of hybridization buffer. For Northern blot analysis, cDNA probes were labeled with [3ZP]dCTP to a specific activity of >lo9 cpmlpg of DNA using the random hexamer procedure (Feinberg and Vogelstein, 1984).
Western Blot Analysis-Myelin membranes were isolated by sucrose gradient ultracentrifugation from 6-week-old rats as described by Smith and Perret (1986) except that the 10.5% sucrose homogenization buffer contained 1 mM phenylmethylsulfonyl fluoride and 4 mM o-phenanthroline. Twenty micrograms of protein were applied to 12% polyacrylamide minigels, electrophoresed, and transferred to Immobilon membranes and detected as described (Snipes et al., 19921, except that 1% normal goat serum was added to all blocking and antibody dilution buffers. The membranes were probed with antipeptide 2 antiserum (1:2000) for 2 h followed by a peroxidase-conjugated goat anti-rabbit IgG (1:10,000; Sigma) for 1 h. The reaction product was visualized by peroxidase-catalyzed chemiluminescence (enhanced chemiluminescence; Amersham) according to the manufacturer's instructions.

Detection of PMP22 in Cultured Schwann Cells i n Vitro-
Schwann cells grown in medium containing forskolin produce a 1.8-kilobase transcript coding for PMP22 (Fig. IC) identical in size to that detected in sciatic nerve Snipes et al., 1992). Cultured Schwann cells produce PMP22 mRNA at considerably lower levels than neonatal rat sciatic nerves (5-10% as determined by densitometry), suggesting that in situ Schwann cells produce more PMP22 than in culture. Removal of forskolin from the culture medium results in a down-regulation of PMP22 mRNA (Fig. lA, lane c) which is reversed when forskolin is added back to medium (Fig. lA,   lanes d and e). This result is in full agreement with previous studies, repeated here (Fig. 1B), showing that forskolin increases 5-10-fold Schwann cell production of mRNA coding for PO (Lemke and Chao, 1988).
PMP22 protein is also produced by cultured Schwann cells and, like PMP22 mRNA, its synthesis is forskolin-regulated. Fig. 2 shows the results of studies in which we metabolically labeled Schwann cells with [35S]methionine and immunoprecipitated cell lysates with a mixture of antisera raised against two synthetic peptides from PMP22 (peptide 1, amino acids 27-42; peptide 2, amino acids 117-132) representing regions predicted to be exposed on the protein's surface (Fig. 3). In these experiments, a molecule (Fig. 2, middle arrow) was immunoprecipitated that electrophoresed on SDS-PAGE as a doublet with an apparent molecular mass of 22 kDa (Fig. 2, lanes b, d, and f). This doublet (PMP22) was not detected in lysates treated with antisera preincubated with excess amounts of peptides 1 and 2 (Fig. 2, lanes a, c, and e ) or in samples immunoprecipitated with preimmune serum (data not shown). Only trace amounts of the protein were evident in Schwann cells grown 3 days in the absence of forskolin (Fig. 2, compare lanes b and d ) . PMP22 levels elevate when forskolin is added back to the culture medium (Fig. 2,   lane f). were specifically immunoprecipitated with the anti-PMP22 antisera mixture. Unlike PMP22, however, levels of this 48-kDa molecule were unaffected by forskolin (Fig. 2,   lanes b, d, and f).  figure. Subsequent studies (see below) suggest that this 18-kDa molecule is the unglycosylated precursor form of PMP22.

Mature PMP22 Protein Is Synthesized from an 18-kDa Precursor and Has a Short Half-life in Cultured Schwann
Cells-A minor 18-kDa protein was occasionally observed in the immunoprecipitation experiments (Figs. 2 and 4). Pulsechase experiments were performed in order to determine the identity of this protein as a possible PMP22 precursor and to measure the half-life of PMP22 and 48-kDa protein in cultured Schwann cells. Schwann cells were metabolically labeled with [35S]methionine for 30 min, chased up to 5 h with unlabeled methionine, and immunoprecipitated with a mixture of antisera to peptides 1 and 2. Fig. 5 shows that the 18-kDa protein is rapidly converted into mature PMP22. The 18-kDa protein also has a size consistent with those observed in earlier i n uitro translation experiments (Manfioletti et al., 1990) and with the molecular weight of the primary structure of PMP22 predicted from its cDNA Spreyer et al., 1991). Taken together, these results suggest that the 18-kDa molecule is a core protein which is posttranslationally modified to give rise to PMP22. Approximately 50% of the PMP22 is turned over within 30-60 min, whereas 48-kDa protein remained unchanged for up to 5 h. This result suggests that PMP22 and the 48-kDa molecule are not chemically related and do not have a precursorproduct relationship.
To examine this possibility further and also to investigate the role of glycosylation in the generation of PMP22, we repeated the metabolic labeling studies using Schwann cells incubated in medium containing tunicamycin which inhibits N-linked glycosylation. Fig. 6 shows that tunicamycin-treated Schwann cells, immunoprecipitated with antibodies to peptide 1 (left panel) or peptide 2 (right panel), contain an 18-kDa protein but not the 22-kDa PMP22. This result suggests that Schwann cell-derived PMP22 is normally glycosylated. It should also be recalled that antibodies to peptide 1 do not recognize processed PMP22 (Fig. 4) but do immunoprecipitate small amounts of the 18-kDa protein from tunicamycintreated cultured Schwann cells. Since peptide 1 represents an amino acid segment of PMP22 that contains a consensus site for N-linked glycosylation (Fig. 3) but which is not glycosylated in the synthetic peptide, it seems reasonable that antibody to peptide 1 may recognize PMP22 protein in the absence but not presence of N-linked carbohydrates in cultured Schwann cells.
Immunocytochemical Localization of PMP22 in Cultured Schwann Cells-We carried out immunocytochemical studies to determine the localization of PMP22 in cultured Schwann cells that are not elaborating myelin. Fixed Schwann cells were immunostained with the PMP22-specific antipeptide 2 antiserum in order to avoid the possible specificity problems of anti-PMP22 peptide 1 antiserum in uitro. Fig. 7 shows that antibodies to peptide 2 detect PMP22 bound to cultured Results show that the half-life of PMP22 is approximately 30 min, but no turnover was evident over the 5-h chase period for the 48-kDa molecule. An 18-kDa protein that was specifically immunoprecipitated was evident before, but not after, chase, a finding consistent with it being a precursor form of PMP22.

blP IP blP IP blP IP blP IP blP IP
Schwann cells with heaviest staining evident on cells flattened onto the culture substratum. Staining was distributed throughout the cytoplasm and was not specifically localized on the cell membrane which might be expected for a myelinbased protein. A similar staining pattern has been described for PO in cultured Schwann cells (Morgan et al., 1991).

PMP22 Is a Myelin Glycoprotein with an 18-kDa Polypeptide Core in
Vivo-We questioned whether the metabolic labeling and immunoprecipitation results we obtained with PMP22 in isolated Schwann cells would be similar or different from those obtained with sciatic nerve in vivo. Accordingly, we carried out short term metabolic labeling experiments followed by immunoprecipitations on sciatic nerve segments maintained for 3 h in tissue culture medium in the presence or absence of tunicamycin. Fig. 8 shows that in nerve samples labeled ex vivo in the absence of tunicamycin, antibodies to peptides 1 and 2 immunoprecipitated PMP22 which migrated on SDS-PAGE as a doublet with mobility identical to that of PMP22 in cultured Schwann cells. It should be recalled that antibody to peptide 1 did not recognize mature PMP22 in Schwann cell extracts (Fig. 4). In nerve segments incubated in medium containing tunicamycin, the 18-kDa protein was immunoprecipitated with both antibodies, as it was from extracts of cultured Schwann cells (compare Figs. 6 and 8). The antipeptide 1 antiserum precipitated the 18-kDa species more completely than the 22-kDa species in sciatic nerve ex vivo.
PMP22 was also analyzed in myelin purified from sciatic nerves of young adult rats. Fig. 9 is a Western blot replica of isolated sciatic nerve myelin that had been treated with shows that the mobility of the 22-kDa molecule following glycanase treatment is shifted to the 18-kDa form. These results are identical to those obtained with tunicamycin in metabolic labeling studies of isolated Schwann cells (Fig. 6) and sciatic nerve segments (Fig. 8) and suggest that PMP22 in sciatic nerve myelin, like that in cultured Schwann cells, has a core size of 18 kDa and is glycosylated to a form with an apparent molecular size of 22 kDa. Antiserum against peptide 1 gave the same results as antiserum against peptide 2, and no 48-kDa protein was detected by Western blot in sciatic nerve extracts (data not shown).

DISCUSSION
Elucidating the function of the PMP22 protein is important for understanding the mechanism by which both a duplication containing the PMP22 gene in humans and two point mutations in the PMP22 gene in mice may give rise to peripheral neuropathies. Toward this goal, we have begun to analyze PMP22 production in cultured Schwann cells in order to establish a suitable model system for studying the regulation and function of this protein. In this study, we have identified both similarities and differences between PMP22 expression in cultured Schwann cells, in intact peripheral nerve, and in purified myelin from young adult nerves.

FIG. I . Expression of PMP22 protein detected by immunocytochemistry in cultured Schwann cells.
Cultured Schwann cells from neonatal rat sciatic nerve were fixed and exposed to antibodies to peptide 2 of PMP22 (a). The bottom leftpanel ( e ) shows Schwann cells exposed to antibodies to SlOO protein. To show the specificity of PMP22 staining, the peptide-specific PMP22 antibody was preblocked with the respective peptide ( c ) . Note that flattened cells (arrow) are more highly positive for PMP22 staining than rounded cells. Corresponding phase-contrast images of the stained cells are shown in the panels on the right side of the figure (b, d, and  f ) . Bar, 25 pM.
The data show that cultured Schwann cells contain a 1.8kilobase mRNA transcript coding for PMP22 identical in size to that in sciatic nerve and that they also produce PMP22 protein in amounts that are easily detected by metabolic labeling methods using radiolabeled methionine. Schwann cell-derived PMP22 has a molecular weight of 22,000, as judged by SDS-PAGE, which is similar to that of PMP22 in sciatic nerve segments labeled metabolically ex vivo and in isolated peripheral nerve myelin studied by Western blot analysis. In some studies, PMP22 migrated as a doublet (see Figs. 2, 4, 6, and 8) that appeared to arise from post-translational changes in glycosylation since in Figs. 6 and 8, the molecule migrated as an 18-kDa band following removal of the N-linked carbohydrates. It is clear that PMP22 from Schwann cells, intact sciatic nerve, and in purified myelin can be altered by agents that affect N-linked glycosylation. Metabolic labeling of Schwann cells and of sciatic nerve segments in medium containing tunicamycin yields an 18-kDa protein which is recognized by antibodies to PMP22 peptides. N-Glycanase treatment of PMP22 in purified sciatic nerve myelin produces the same 18-kDa protein. Pulse-chase studies carried out on Schwann cells suggest that the 18-kDa molecule is a short-lived precursor (half-life under 30 min) that is glycosylated to yield mature PMP22. This result is in full agreement with previous cDNA studies which predict that the core protein of PMP22 has a molecular weight of 18,000 Spreyer et al., 1991). In the absence of tunicamycin, the protein is synthesized in Schwann cells and in sciatic nerve segments with an apparent molecular weight of 22,000, the difference apparently being due to glycosylation. Taken together, these results support the notion that cultured Schwann cells produce a form of PMP22 similar to that in peripheral nerves.
Differences between PMP22 in intact nerve and in cultured Schwann cells were also detected, however. PMP22 expression is markedly lower in cultured Schwann cells than in intact nerves, even when production is stimulated by forskolin. Also, the turnover rate of PMP22 is rapid in cultured Schwann cells with a half-life of 30-60 min. Although the half-life of PMP22 in uiuo has not been directly measured, it seems likely that its turnover rate in peripheral nerves is much slower as is the case for other protein constituents of compact myelin (Davison, 1961;Gould, 1977). Also, in peripheral nerves of aged animals, PMP22 mRNA levels are markedly diminished relative to levels of the protein, suggesting that once synthesized, PMP22 is not rapidly replaced .
We also detected some antigenic differences between PMP22 synthesized by myelinating Schwann cells within explanted sciatic nerve segments and by passaged nonmyelinating Schwann cells in uitro. Antibody 1 raised against a synthetic peptide of PMP22, which contains a consensus site for N-linked glycosylation, failed to recognize PMP22 produced by cultured Schwann cells but did recognize the protein synthesized by Schwann cells in segments of sciatic nerve. This result suggests that subtle differences may exist in the nature of the carbohydrate moieties in PMP22 from the two sources. Antibody to peptide 1 recognized the 18-kDa unglycosylated form of the molecule synthesized by cultured cells and nerve segments in the presence of tunicamycin, suggesting that the antibody recognizes epitopes against which it was raised but not when the consensus site is glycosylated by cultured Schwann cells.
Antibody to peptide 1, but not peptide 2, specifically immunoprecipitated large amounts of a 48-kDa protein from  (Lees and Brostoff, 1984). Lune 2, an identical sample transferred to nitrocellulose and immunostained with antibodies to anti-PMP22 peptide 2 (1:2000 each) using the indirect peroxidase method developed using a chemilumi-  (B, lanes 3 and 4 ) . Results of these studies show that PMP22 in sciatic nerve, like in cultured Schwann cells, is a glycosylated protein that can be converted by deglycosylation into an 18-kDa core protein.
cultured Schwann cells (Fig. 2) that appeared to be distinct from PMP22 by several criteria. Pulse-chase labeling methods showed different rates of protein turnover with no evidence of a precursor-product relationship between the two (Fig. 5).
Also, unlike PMP22, production of the 48-kDa molecule was not affected by forskolin. The identity of this 48-kDa molecule is unknown but it may be a protein that shares an epitope with PMP22 but otherwise is chemically unrelated. It should also be emphasized that this 48-kDa molecule was not evident in metabolically labeled sciatic nerve so the molecule may be produced only by mitogen-expanded Schwann cells.
We also determined by immunocytochemistry that PMP22 is localized predominantly in the cytoplasmic compartment of cultured Schwann cells in vitro in contrast to its localization in myelin membranes of intact nerves in uiuo. Although this result might reflect different cellular compartmentalization of the PMP22 protein, it could also reflect differences in PMP22 function between nonmyelinating Schwann cells in vitro and myelinating cells in uiuo. This issue warrants further attention since it might be critical for the correct interpretation of functional studies of myelin proteins in vitro using either expanded Schwann cells or transfected cell lines, given the possible significance of PMP22 glycosylation as outlined below. Similarly, it will be of interest to determine if the predominantly cytoplasmic localization of PMP22 in Schwann cells is related to its increased turnover rate in the absence of myelin formation. Increased amounts of PMP22 in the cytoplasmic Schwann cell processes could be consistent with a "premyelin" localization or an additional function for PMP22, possibly as a growth regulator, as discussed below.
The finding that PMP22 mRNA and protein are expressed in Schwann Cells in Schwann cell cultures a t considerably lower levels than in Schwann cells in vivo is in agreement with previous studies, indicating an important regulatory influence of the axon on the expression of a variety of Schwann cell genes. Rapid down-regulation of myelin-related genes in the absence of axons occurs in Schwann cells after nerve section in uiuo or in Schwann cell cultures (Brockes et al., 1980;Mirsky et al., 1980;LeBlanc et al., 1987;Trapp et al., 1988;Toma et al., 1992;Lemke and Chao, 1988;Mitchell et al., 1990;Morrison et al., 1991). Conversely, axon-associated signals that promote protein synthesis in Schwann cells can partially and under specific circumstances be mimicked by agents that elevate intracellular cAMP levels (Baron-Van Evercooren et al., 1986;Sobue et al., 1986;Lemke and Chao, 1988;Shuman et al., 1988;Mirsky et al., 1990;Morgan et al., 1991). Forskolin and cAMP analogues, for instance, induce re-expression of Schwann cell-specific glycolipids, MBP, Po and PI70~ glycoprotein in cultured neonatal Schwann cells that fail to express these molecules once removed from axonal contact. Similarly, we report here that expression of PMP22 mRNA and protein is strongly up-regulated in cultured Schwann cells by the adenylate cyclase activator forskolin. Forskolin cannot, however, completely mimic the effect of axons on Schwann cells, since, as seen for other myelin proteins, levels of PMP22 mRNA and protein in forskolin-treated primary Schwann cells are significantly lower than in intact nerves.
PMP22 mRNA is chemically identical to growth arrestspecific-3 mRNA (gas3; Welcher et al., 1991), a transcript that has been associated with cellular growth arrest during the Go phase of the cell cycle in cultured fibroblasts Manfioletti et al., 1990). Given that PMP22 is upregulated by forskolin in passaged Schwann cells in vitro, elevated cAMP levels may also up-regulate PMP22 expression in Schwann cell precursors in uiuo, leading to cellular growth arrest and the differentiation of these cells into the myelinating Schwann cell phenotype. This regulatory mechanism might be reflected in the neurological mouse mutants Tr and TI.' which carry point mutations in the PMP22 gene (Suter et al., 1992a(Suter et al., , 1992b. These defects in PMP22 are most likely responsible for the Tr and TI.' phenotype which is characterized by abnormal continuous Schwann cell proliferation throughout life and severe myelin deficiencies. It should be emphasized, however, that the available data are also consistent with the interpretation that PMP22 is not involved in Schwann cell growth regulation, but rather plays regulatory and functional roles which are related to the other protein components of compact peripheral nervous system myelin. Conceivably, PMP22 may serve two functions in Schwann cell development: as a regulator of Schwann cell division in the initial phase of peripheral nervous system development and as a structural component of myelin during maturation of the peripheral nervous system. Such alternating roles have also been suggested for other myelin proteins, based on embryonic expression patterns of DM-20 (Timsit et al., 1992;Ikenaka et dl., 1992) and the effects of specific mutations in proteolipidprotein (PLP;Schneider et al., 1992), the potential central nervous system counterpart of PMP22 .
Previous studies have suggested that PMP22 is a myelin glycoprotein (Kitamura et al., 1981;Welcher et al., 1991), and in this study we confirm that PMP22 in myelin is N-glycosylated. Analyzed on SDS gels, PMP22 in purified myelin migrates as multiple bands, suggesting heterogeneous posttranslational modifications. This finding is not unexpected for a myelin protein. For example, Po is glycosylated (Everly et al., 1973;Wood and Dawson, 1974;Smith and Sternberger, 1982), sulfated (Matthieu et al., 1975), phosphorylated (Singh andSpritz, 1976;Wiggins and Morell, 1980) and acylated (Agrawal et al., 1983). Several lines of evidence in this study (reviewed above) indicate that PMP22 in cultured Schwann cells, sciatic nerve, and purified myelin is glycosylated, but the identity of the side chains has not been clarified. Furthermore, the nature of the side chains may not be identical in PMP22 from the three sources since antibody to peptide 1, which is a segment of PMP22 containing a consensus sequence for N-linked glycosylation, recognized PMP22 in metabolically labeled segments of sciatic nerve but not in cultured Schwann cells. Identifying the side chains of PMP22 will be of considerable interest, since it seems likely, based on the localization of PMP22 in compact myelin and the nature and consequences of the PMP22 mutations in Tr and Tf' mice, that PMP22 plays some role in the compaction of the myelin sheath. N-Linked glycosylated groups on Po have been implicated previously as a major determinant in the adhesive properties of this molecule (Filbin and Tennekoon, 1991;Yazaki et al., 1992).
In conclusion, results in this study show that PMP22 is expressed in expanded Schwann cell cultures, its production is regulated by forskolin, and the molecule is produced in a glycosylated form that appears to be generated from an 18-kDa precursor. Furthermore, a similar form of the protein is produced within sciatic nerve segments metabolically labeled ex vivo, and like other myelin proteins, PMP22 is a major glycoprotein constituent of isolated rat myelin. Our studies provide the basis for future in vitro experiments aimed at understanding the functional role of PMP22 in the myelination and growth regulation of normal and diseased peripheral nerves.