Cysteine Residues in the Large Extracellular Loop (EC2) Are Essential for the Function of the Stress-regulated Glycoprotein M6a*

Gpm6a was identified as a stress-responsive gene in the hippocampal formation. This gene is down-regulated in the hippocampus of both socially and physically stressed animals, and this effect can be reversed by antidepressant treatment. Previously we showed that the stress-regulated protein M6a is a key modulator for neurite outgrowth and filopodium/spine formation. In the present work, mutational analysis was used to characterize the action of M6a at the molecular level. We show that four cysteines 162, 174, 192, and 202 within EC2 are functionally crucial sites. The presence of cysteines 162 and 202 is essential for the efficient cell surface expression of the M6a protein. In contrast, cysteines 174 and 192, which form a disulfide bridge as shown by biochemical analysis, are not required for the efficient surface expression of M6a. Their mutation to alanine does not interfere with the localization of M6a to filopodial protrusions in primary hippocampal neurons. The neurons expressing C174A and/or C192A mutants display decreased filopodia number. In non-permeabilized cells, these mutant proteins are not recognized by a function-blocking monoclonal antibody directed to M6a. Moreover, neurons in contact with axons expressing C174A/C192A mutant display significantly lower density of presynaptic clusters over their dendrites. Taken together, this study demonstrates that cysteines in the EC2 domain are critical for the role of M6a in filopodium outgrowth and synaptogenesis.

Glycoprotein M6a is a neuronally expressed member of the proteolipid protein (PLP/DM20) family (1) whose gene has been identified as a stress-responsive gene in the hippocampal formation. In several animal models of chronic stress, expression levels for M6a in hippocampal tissue were found to be diminished by chronic stress exposure and this effect was counteracted by treatment with antidepressants (2)(3). Recently, an association of the M6A gene with the subgroup of schizophrenia patients with high levels of depression has been reported (4). These findings suggest that M6a plays a role in the stress-in-duced hippocampal alterations that are found in psychiatric disorders in general. M6a is prominently expressed in the central nervous system as early as embryonic day 10 and remains detectable in adulthood (5). Originally, it was identified as an antigen reacting with the monoclonal M6 antibody and its role as a modulator of neurite outgrowth was postulated (6). Lagenaur et al. (6) demonstrated that IgG or Fab fragments of M6 antibody interfere with the extension of neurites by cultured cerebellar neurons. A recent study by Zhao et al. (7) shows that M6a expressed in the murine neural retina also regulates neurite extension. The neurite outgrowth of M6a-overexpressing retinal cells was strikingly enhanced, although M6a did not affect differentiation and proliferation.
Even though the precise biological function of M6a still remains unclear, there is a growing body of evidence indicating the importance of M6a in the processes of neural development such as neurite extension and differentiation. For example, a study by Mukobata et al. (8) reported that M6a expression enhances nerve growth factor-primed neurite extension in rat pheochromocytoma PC12 cells. They show that it also induces an increase in the intracellular Ca 2ϩ concentration of PC12 cells and that the anti-M6a antibody efficiently interferes with both nerve growth factor-triggered Ca 2ϩ influx and neurite extension (8). Next, inhibition of mouse M6a expression was found to lead to decreased differentiation of neurons derived from mouse embryonic stem cells (9). Furthermore, it has been demonstrated that M6a plays an important role in neurite/ filopodium outgrowth and synapse formation (10). This study shows that M6a overexpression induces neurite formation and increases filopodia density in hippocampal neurons. M6a knockdown with small interference RNA methodology showed that M6a low expressing neurons display decreased filopodia number and a lower density of synaptophysin clusters. The reduced M6a expression by chronic stress might be directly related to the morphological alterations found in the hippocampus of chronically stressed animals. The mechanism that would explain how M6a regulates neurite/filopodium outgrowth and its involvement in chronic stress response remains unclear.
In this study our goal was to identify the regions within M6a that are critical for the neurite/filopodium outgrowth. To define the putative biologically critical amino acid residues we took advantage of the striking structural similarities that M6a bears to the tetraspanin family of proteins. The functional spec-ificity of tetraspanins is determined by the EC2 3 region. The EC2 is subdivided into a constant region and a variable region. The variable subdomain, which contains nearly all of the known tetraspanin protein-protein interaction sites, is inserted within the conserved subdomain and their relative topology is governed by the occurrence of key disulfides. The central role of the two disulfide bridges in stabilizing the EC2 structure was firmly established. In addition, most EC2s of tetraspanins are glycosylated in one or more potential N-glycosylation sites (11)(12)(13)(14).
Similarly to tetraspanins, the extracellular region of M6a contains cysteines residues that may be functionally crucial sites. In this report, we show that cysteines 174 and 192 within the EC2 are important for the role of the M6a protein in filopodia formation and synaptogenesis.

EXPERIMENTAL PROCEDURES
Site-directed Mutagenesis-All mutations were introduced into GFP-M6a, a mammalian expression plasmid coding for wild type (WT) M6a fused with green fluorescent protein (GFP) that has been described previously (10). The expressed fusion protein is functional and localizes correctly to the plasma membrane of transfected cell lines or neuronal primary cultures. To create the cysteine mutants, cysteine residues at positions 44,46,162,174,192, and 202 were mutated to alanine. The substitution of conserved cysteine residues with alanine rather than serine has been proposed to minimize the effects of the substituting amino acid on proper protein folding and stability (15). To mutate two N-glycosylation sites, asparagines at positions 164 and 208 were substituted by glutamines. Multiple mutations in EC2, where two or more cysteines (Cys 162 , Cys 174 , Cys 192 , and Cys 202 ) or two asparagines (Asn 164 and Asn 208 ) were mutated, were created by conducting serial rounds of sitedirected mutagenesis. The desired mutations were generated by standard PCR mutagenesis techniques using BD Advantage 2 Polymerase Mix (BD Biosciences Clontech). Two overlapping oligonucleotide primers, one of which contains the target mutation (purchased from Sigma) were used to amplify the template DNA and were as follows: C44A forward, 5Ј-T GGC  TGT GGC CAT GAA GCC CTT TCT GGA ACA-3Ј, reverse,  5Ј-AAG GGC TTC ATG GCC ACA GCC AGC GAA CAG  GGC AA-3Ј; C46A forward, 5Ј-T GGC CAT GAA GCC CTT  TCT GGA ACA GTC AA-3Ј, reverse, 5Ј-AGA AAG GGC TTC  ATG GCC AGC GCC ACA GAA CA-3Ј; double C44A/C46A  forward, 5Ј-T GGC CAT GAA GCC CTT TCT GGA ACA  GT-3Ј, reverse, 5Ј-A AAG GGC TTC ATG GCC AGC GCC  AGC GAA CAG GGC AA-3Ј; C162A forward, 5Ј-TC AAT  GTG TGG ACC ATC GCC CGG AAC ACC A-3Ј, reverse,  5Ј-GAT GGT CCA CAC ATT GAA ATA CAT GTA CAC  GGG C-3Ј; C174A forward, 5Ј-TA GTG GAG GGA GCA AAT  CTC GCC TTG GAT CTG-3Ј, reverse, 5Ј-GAG ATT TGC  TCC CTC CAC TAG AGT GGT GTT-3Ј; C192A forward,   5Ј-CA ATT GGA GAG GAA AAG AAA ATT GCT ACT GCC  TCT G-3Ј, reverse, 5Ј-AAT TTT CTT TTC CTC TCC AAT  TGT CAC AAT CCC-3Ј; C202A forward, 5Ј-CT GAG AAC  TTC CTG AGG ATG GCT GAA TCT ACT-3Ј, reverse,  5Ј-CAT CCT CAG GAA GTT CTC AGA GGC AGT ACA-3Ј;  N164Q forward, 5Ј-C AAT GTG TGG ACC ATC TGC CGG  CAA ACC ACT C-3Ј, reverse, 5Ј-CCG GCA GAT GGT CCA  CAC ATT GAA ATA CAT GTA CAC G-3Ј; N208Q forward,  5Ј-G AGG ATG TGT GAA TCT ACT GAG CTG CAG ATG  ACC TTC CAC-3Ј, reverse, 5Ј-CAG CTC AGT AGA TTC  ACA CAT CCT CAG GAA GTT CTC AGA GGC-3Ј. Following temperature cycling, the DpnI endonuclease was used to digest the parental DNA template and to select for mutationcontaining synthesized DNA. The identity of all mutant constructs were verified by DNA sequencing on ABI 377 or ABI 3130 sequencers (Applied Biosystems).
Hippocampal Cultures, Cell Lines, and Plasmid Transfections-Dissociated neuronal cultures were prepared from hippocampi of embryonic day 19 Wistar rats, as described (16). Briefly, tissue was treated with 0.25% trypsin in Hanks' solution for 15 min at 37°C. A single-cell solution was prepared in Eagle's minimal essential medium (MEM) containing 2 mM glutamine, 10 M sodium pyruvate, 100 units/ml penicillin, and 100 g/ml streptomycin (MEM1X) with 10% (v/v) horse serum. Cells were seeded on plates or coverslips coated with 0.1 mg/ml poly-L-lysine hydrobromide (Sigma) and 20 g/l laminin (Invitrogen) at a density of 20,000 cells/cm 2 . After 2 h, medium was changed to MEM/N2 (MEM1X with 1 g/liter ovalbumin and B27 serum-free supplements from Invitrogen). Mouse neuroblastoma N2a and COS-7 cells were cultured in Dulbecco's modified Eagle's medium with 10% (v/v) fetal bovine serum, penicillin, and streptomycin. Neurons or N2a cells growing in 24-well plates were transiently transfected with 2 g of DNA mixed with 1 and 2 l of Lipofectamine 2000 (Invitrogen), respectively, following the manufacturer's instructions (medium was changed 4 -6 h later).
Proteinase K Treatment and Western Blotting-COS-7 cells in 100-mm plates were grown to 90% confluence and transfected with plasmids encoding WT GFP-M6a fusion protein or double C174A/C192A mutant protein (24 g of DNA) using Lipofectamine 2000 (Invitrogen) following the manufacturer's instructions. Twenty-four hours following the transfection, cells were rinsed with PBS and incubated with PBS alone or with proteinase K (PK) at a final concentration 1 mg/ml (Promega) for 30 min at 37°C. At the end of the incubation, dishes were placed on ice and 0.5 mM phenylmethylsulfonyl fluoride was added to halt digestion. Cells were scraped and transferred to Eppendorf tubes. Samples were then rapidly centrifuged, and the supernatants were aspirated. Cell pellets were lysed in a modified RIPA buffer (150 mM NaCl, 50 mM Tris-HCl, 1% deoxycholate sodium salt, 0.1% SDS, pH 8) supplemented with protease inhibitor mixture (Sigma) for 15 min on ice. Proteins were then precipitated by trichloroacetic acid/acetone precipitation at Ϫ20°C overnight. After the precipitation, samples were centrifuged at 11,500 ϫ g for 15 min at 4°C, and the supernatants were discarded. The protein pellets were dissolved in the appropriate volume of rehydration buffer (10 mM dithiothreitol, 20 mM Tris-HCl, pH 6.8, 9 M urea) and the concentration of solubilized proteins measured using a Bradford assay (Bio-Rad Protein Assay). 5ϫ SDS sample buffer was added to each sample. The protein samples were loaded on 10% SDSpolyacrylamide gels (15 g/lane) and transferred to a nitrocellulose membrane by electroblotting. After 2 h of blocking in Tris-buffered saline containing 5% nonfat dried milk, the membranes were incubated with rabbit anti-GFP polyclonal serum (1/750) (Invitrogen). Antigen-antibody complexes were detected using an anti-rabbit horseradish peroxidase-linked secondary antibody (1/10,000) (Dako, Denmark) according to standard Enhanced Chemiluminescence (ECL) Western blotting protocol using Super Signal West Pico chemiluminescent substrate (Pierce).
Cell Surface Biotinylation of Free Cysteines-Cell surface protein were biotinylated with N-biotinylaminoethyl methanethiosulfonate (MTSEA-biotin; Toronto Research Chemicals). MTSEA-biotin is a membrane-impermeable, thiol-reactive agent that reacts only with cysteinyl sulfhydryl groups of free cysteines. Thus only cysteines not in a disulfide bond and in an accessible area of the protein tertiary structure will form a mixed disulfide with this alkylthiosulfonate and can be detected on immunoblots of avidin-biotin pulldown assays. No biotinylation would indicate that all cysteines in the extracellular region are unavailable, either from being disulfide-bonded and/or from residing in an area of the molecule that is inaccessible to MTSEA-biotin. A Cys 3 Ala point mutation of a residue that normally forms a disulfide bond would result in the bond being broken and the liberation of a partner cysteine residue would be detected as an increase in the biotinylation. In case of a double Cys 3 Ala mutation, when the correct pair of cysteines in a disulfide bond had been mutated the MTSEAbiotinylation would be decreased. In contrast, a double mutation corresponding to incorrect partners may result in an increase in biotinylation because of formation of a further free cysteine residue. COS-7 cells in 100-mm plates were grown to 90% confluence and transfected with plasmids encoding GFPtagged WT M6a, PLP, and the indicated M6a mutants (20 g of DNA) using Lipofectamine 2000 (Invitrogen) following the manufacturer's instructions. Twenty-four hours following transfection, cells were rinsed twice with ice-cold PBS and incubated with MTSEA-biotin (at a final concentration 0.5 mM) in PBS for 30 min at 4°C. Unbound MTSEA-biotin in the reaction was quenched by 50 mM NH 4 Cl. Cells were washed twice in 25 mM Tris, 150 mM NaCl, pH 7.2, scraped, and transferred to Eppendorf tubes with 500 l of lysis buffer (10 mM Tris, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, pH 7.2) supplemented with a protease inhibitor mixture (Sigma). Cells were disrupted by sonicating on ice using five 1-s bursts and incubated on ice for 30 min. After centrifugation at 10,000 ϫ g for 2 min at 4°C, 400 l of the clarified supernatant was added to 400 l of NeutrAvidin-agarose slurry (Pierce) pre-equilibrated according to the manufacturer's instructions. Samples were rotated overnight at 4°C to allow binding of neutravidin to biotinylated proteins. Then, the beads were washed three times with washing buffer (10 mM Tris, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, 1% deoxycholate, pH 7.2) with a 2-min centrifugation (1,000 ϫ g) at 4°C between washes. After the final wash, biotinylated proteins were eluted by 2 ϫ 50 l of 1ϫ SDS sample buffer with 100 mM dithiothreitol and boiled for 5 min before SDS-PAGE and Western blotting analyses. The remaining 100 l of clarified cell lysate was used to assess the total amount of the overexpressed GFP-tagged protein in cellular lysate (input control for quantification purposes). Proteins were precipitated by trichloroacetic acid/acetone precipitation as described above, dissolved in 20 l of rehydration buffer, and the concentration of solubilized proteins was measured using the Bradford assay (Bio-Rad protein assay). 5ϫ SDS sample buffer was added to each sample. The protein samples were loaded on 10% SDSpolyacrylamide gels, transferred to a nitrocellulose membrane by electroblotting, and processed for anti-GFP antibody (rabbit anti-GFP polyclonal serum (1/750); Invitrogen) staining as described previously. Immunoblots were quantified by densitometric analyses using ImageJ software.
Quantification and Statistical Data Analysis-For the neuroblastona cell line N2a, the percentage of cells showing filopodial protrusions (visualized by the F-actin marker phalloidin) was calculated for both transfected as well as non-transfected cells from the same coverslip. Transfection with WT M6a and GFP alone was used as a control. On average, 60 -120 cells from three to nine groups for each transfection condition from multiple experiments were analyzed. The ratio of transfected to non-transfected filopodia-bearing cells from the same coverslip was determined. Filopodium density (number of protrusions per 20-m neurite length measured within 50 m from the soma as shown in the Fig. 7A) was quantified in 33-55 neurites of different neurons from 2 to 12 groups for each overexpression culture from multiple experiments. Group means were then analyzed for overall statistical significance by using oneway ANOVA followed by Dunnett's multiple comparison test for post hoc effects versus WT M6a and GFP as controls or by a Student's t test. To quantify the spots of synaptohysin 1 immunoreactivity, regions where transfected axons (identified by green labeling and their lack of MAP2 immunostaining) contacted non-transfected MAP2 positive dendrites were selected. The number of immunopositive synaptophysin 1 puncta was counted in a 40-m neurite length along similar distances out from the soma in 22 neurites of a minimum 11 different neurons per group for overexpression cultures and repeated on three separate culture preparations. Group means were then analyzed for overall statistical significance by using one-way analysis of variance (ANOVA) followed by Tukey's multiple comparison tests. Calculations were performed with the software GraphPad Prism.

Mutation of Cysteine Residues in EC2 of M6a; Cysteine Residues 162 and 202 Are Critical for the Correct M6a
Localization in the Plasma Membrane-To define the part of the M6a molecule or amino acid residues that might be critical for the induction of filopodial protrusion formation, we constructed a series of M6a mutants expected to interfere with the function of the protein. The selection of proposed mutation sites was based on structural similarity of M6a to tetraspanins and numerous mutational analysis data available on the function of this family of proteins. Fig. 1 shows the predicted computational model of M6a displaying M6a structural features. It contains four transmembrane domains, a small extracellular loop (EC1), a very short intracellular loop, and a large extracellular loop (EC2), flanked by relatively short N-and C-terminal cytoplasmic tails. There are four cysteines residues in EC2 domain (Cys 162 , Cys 174 , Cys 192 , and Cys 202 ) that may be functionally important and two cysteines residues in the EC1 domain (Cys 44 and Cys 46 ). The EC2 domain also contains two potential N-glyco-sylation sites (Asn 164 and Asn 208 ). In the cytoplasmic part of the protein, two potential sites for protein kinase C and one for the casein kinase II phosphorylation are present as well as additional juxtamembrane cysteine residues.
The constructs were transfected into the neuroblastoma cell line N2a, and subcellular distribution of WT and mutant M6a proteins was analyzed by immunofluorescence microscopy. Fig. 2A shows non-permeabilized N2a cells transfected with the WT M6a and mutants. Single C162A and 202A mutants and all variants that include these mutations were detected intracellulary and none of them displays cell surface expression that can be clearly seen in control WT M6a-transfected cells ( Fig. 2A, maximized views). Taken together, these results show that Cys 3 Ala mutations of Cys 162 and Cys 202 in the EC2 region disrupt the cell membrane localization of the M6a protein. We suppose that they could interfere with proper protein folding and/or cell surface trafficking.
On the other hand, when N2a cells were transfected with single C174A or C192A or their combination C174A/C192A M6a mutants, we observed intense staining of the plasma membrane ( Fig. 2A). To determine whether the C174A/C192A mutant was exposed on the outside of the cell, we probed its accessibility to digestion with the extracellular PK, a broad spectrum protease widely used to demonstrate the surface exposure of the membrane proteins. If WT M6a or the C174A/ C192A mutant were expressed on the plasma membrane, their extracellular part would be accessible to PK digestion. Intracelullar parts of the proteins would be inaccessible, and thus protected from PK digestion. On Western blot we would be able to detect the fragments undigested by PK. Using anti-GFP antibody we expected to detect a cleaved fragment (ϳ32 kDa) corresponding to the GFP-tagged N-terminal part including the first transmembrane domain after PK treatment. In case the proteins are not cell surface exposed, and thus inaccessible to protease digestion, only full-length GFP fusion proteins would be detected by anti-GFP antibody after PK treatment. Intact COS-7 cells transfected with the double C174A/C192A mutant or WT M6a as a control were treated with PK. Cleaved products were analyzed on immunoblots using anti-GFP antibody. In both cases, PK treatment of cells expressing either WT M6a or the double C174A/C192A mutant yielded protein fragments of ϳ32 kDa (Fig. 2B), which is consistent with the size of the N-terminal GFP-tagged fragment localized inside the cell. This indicates that the mutant protein is accessible to PK digestion, and thus present on the plasma membrane.
Detection of Disulfide Bonds in the Extracellular Region of M6a-To investigate whether cysteine residues within the extracellular region of M6a form disulfide bonds, MTSEA-biotin labeling of cell surface proteins was performed. We used this assay to assess the availability of accessible free cysteines in the extracellular region of WT M6a and M6a cysteine mutants.

EC2 Cysteine Residues Are Essential for M6a Function
Intact COS-7 cells transfected with GFP-tagged WT M6a, and the indicated M6a cysteine mutants, were treated with MTSEA-biotin. GFP-tagged PLP was used as a control because it belongs to the same family of proteins as M6a and the state of all cysteines residues of PLP and its membrane topology are well known (17). Cell surface-biotinylated proteins were pulled down by neutravidin and analyzed on immunoblots using anti-GFP antibody. For quantification purposes we show data as the ratio of biotinylated proteins pulled down by neutravidin to the total in cellular lysates, as measured by densitometric analysis of immunoblots. Fig. 3, A and B, demonstrates that WT M6a as well as PLP are biotinylated, indicating that some cysteines in the extracellular region of WT molecules are free and thus accessible to MTSEA-biotin. A further increase in the biotinylation was detected for C174A and C192A mutants. This indicates that these residues are normally involved in disulfide bond formation in WT M6a. Double C174A/C192A mutant showed a decrease in biotinylation, and thus, less free cysteines available for MTSEA-biotinylation (Fig. 3, A and B). . Immunoblot analysis of cell lysates before (Ϫ) and after (ϩ) PK treatment shows undigested fragments of ϳ32 kDa (arrow) resulting from proteolytic cleavage of cell surface-exposed proteins after (ϩ) PK treatment. This suggests that Cys 174 and Cys 192 correspond to the disulfide pair in WT M6a. Biotinylation of C44A and C46A mutants did not differ from that of WT M6a. The double C44A/C46A mutant showed a decrease in biotinylation compared with WT M6a, although the difference was not statistically significant (Fig. 3C). It was not possible to test for the cell surface MTSEA-biotinylation of Cys 162 and Cys 202 because their mutation to alanine interfered with cell surface expression of the mutant proteins.

Mutation of Cysteines 174 and 192 Prevents Epitope Recognition by a Function Blocking
Anti-M6a mAb-The model for M6a topography predicts that EC2 is positioned on the outside of the cell. A previous study reported that monoclonal anti-M6a antibody directed against the extracellular part of the protein interferes with the extension of neurites by cultured cerebellar neurons (6). Interestingly, mutagenesis studies of tetraspanins demonstrated that EC2 of tetraspanins is a crucial site for specific functional protein-protein interactions (18,19). It contains nearly all of the known tetraspanin protein-protein interaction sites and, in addition, monoclonal antibodies that recognize cell surface epitopes so far exclusively recognize EC2 (19 -21).
Thus, we were interested whether the Cys 174 and Cys 192 mutations in EC2 of M6a would interfere with binding of the anti-M6a antibody that recognizes a surface-exposed epitope and has been reported to affect neurite extension (6). Primary hippocampal neurons, as well as N2a cells, were transfected with the indicated GFP-tagged mutants. Immunostaining was performed in non-permeabilized cells with rat anti-M6a mAb. Wild type GFPtagged M6a was used as a control. We show that the surface-exposed region of WT GFP-tagged M6a is recognized by anti-M6a antibody in hippocampal neurons (Fig. 4). The complete colocalization can be observed in the merge image. In contrast, the cells transfected with C174A, C192A, and double C174A/ C192A mutants do not display any overlap in the merge image. The same results were observed when neuroblastoma N2a cells were transfected (data not shown). We conclude that although C174A and/or C192A express on the plasma membrane (Fig. 2,  A and B), they do not bind to the function blocking anti-M6a antibody. Fig. 1, the extracellular region of M6a contains apart from the EC2 domain also a small extracellular loop (EC1) that could potentially be involved in anti-M6a antibody binding and/or play a role in the induction of filopodial protrusion formation. There are two cysteines residues (Cys 44 and Cys 46 ) in the EC1 domain of M6a that attracted our attention as suitable candidates for mutational analysis. Single C44A and C46A point mutations, as well as the double C44A/C46A point mutation were introduced into GFP-M6a. When N2a cells were transfected with these M6a mutants intense staining of the plasma membrane was observed in all three cases (data not shown). Thus, the single C44A, C46A, and double C44A/C46A mutant proteins are able to localize to the plasma membrane the same way as WT M6a.

Mutation of Cysteine Residues in EC1 Does Not Affect the Cell Membrane Localization of M6a Protein or Its Recognition by a Function Blocking Anti-M6a mAb-As shown
To test whether EC1 of M6a is part of the surface-exposed epitope recognized by the function blocking anti-M6a antibody, primary hippocampal neurons were transfected with the indicated GFP-tagged mutants. Immunostaining was performed in non-permeabilized cells with rat anti-M6a mAb. The results indicate that the surface-exposed regions of C44A, C46A, as well as C44A/C46A GFP-tagged M6a mutants are completely recognized by anti-M6a antibody (Fig. 5). The colocalization can be observed in the merge image. We conclude that EC1 is not the surface-exposed epitope recognized anti-M6a antibody. The mutation of cysteines residues Cys 44 and Cys 46 in EC1 does not interfere with the binding of anti-M6a antibody.
C174A and C192A Mutants Do Not Induce Filopodium Formation in the Neuronal Cell Line as well as in Primary Hip-pocampal Neurons-The effect of C174A and C192A mutations on filopodial protrusion formation was analyzed and the results compared with that obtained with the WT GFP-M6a or GFP alone (10). First, N2a cells were transfected with GFP-tagged mutant proteins, WT protein and GFP alone, and stained with the F-actin marker phalloidin to visualize filopodia. In the Fig. 6A, an example of a WT GFP-M6a positive cell shows localization of the fusion protein at filopodia labeled with phalloidin (maximized views labeled 1). As can be observed, the expression of the exogenous WT M6a promotes filopodium formation in the neuronal cell line. Non-transfected cells do not form filopodia to the same extent as transfected cells (Fig. 6A, red cells, maximized views labeled 2). Next, the number of cells showing filopodial protrusions was quantified and the results are displayed as a ratio of transfected to non-transfected filopodia-bearing cells from the same coverslip. Fig. 6B demonstrates that the number of cells showing filopodial protrusions in cells transfected with C174A, C192A, and double C174A/C192A mutants is significantly lower than in the WT M6a-transfected N2a cells.
To confirm that C174A and C192A mutations of M6a have the same effect in cells that normally express M6a, we next examined the effect of these mutations on filopodium formation in primary hippocampal neurons. Neurons were transfected on day 3 after plating with the indicated M6a mutants (single C174A, C192A, and double C174A/C192A mutant) or with WT M6a and GFP alone as controls. Neurons were stained with the anti-MAP2 antibody to label dendrites and with the F-actin marker phalloidin to visualize filopodia. Filopodium density (number of protrusions per 20-m of neurite length) as shown in the enlarged pictures (Fig. 7A) was quantified. Fig. 7B shows that WT M6a-transfected neurons have a significantly higher number of filopodia than neurons transfected with C174A, C192A, and double C174A/C192A mutants. Thus, it can be concluded that even though localization of the C174A, C192A, and double C174A/C192A M6a mutants does not differ from that of WT M6a, their overexpression does not promote filopodium formation in the neuronal cell line N2a as well as in primary hippocampal neurons.
Effect of C44A and C46A Mutants within the EC1 on Filopodium Formation-To analyze if overexpression of EC1 cysteine mutants would result in a similar alternation of filopodium formation, we examined the effect of C44A and C46A mutations on filopodium formation in the neuroblastoma  NOVEMBER 13, 2009 • VOLUME 284 • NUMBER 46 cell line as well as in primary hippocampal neurons. The neuroblastoma cells N2a were transfected with GFP-tagged mutant proteins, WT protein, and GFP alone, and stained with the F-actin marker phalloidin to visualize filopodia. The number of cells showing filopodial protrusions was quantified and the results are displayed as described above. Fig. 8A shows that the number of cells showing filopodial protrusions in cells transfected with C44A, C46A, and double C44A/C46A mutants does not significantly differ from that in the WT M6a-transfected N2a cells. Although there is a slight decrease in the number of filopodia formed when cells were transfected with C44A, C46A, or C44A/C46A mutants, the differences with WT M6a were not statistically significant.

EC2 Cysteine Residues Are Essential for M6a Function
Next, we examined the effect of these mutations on filopodial protrusion formation in primary hippocampal neurons. Neurons were transfected on day 3 after plating with the indicated M6a mutants or with WT M6a and GFP alone as controls. Filopodium density was quantified (Fig. 8B). The neurons transfected with C44A and C46A mutants do not have a significantly different number of filopodia compared with WT M6a-transfected neurons. But in the case of the double C44A/C46A mutant a significant decrease in filopodia formation was observed.
Thus, although the mutations of the Cys 44 and Cys 46 in EC1 do not prevent binding of anti-M6a antibody they do slightly interfere with filopodium formation. We suggest that EC1 contributes to the optimal functioning of the protein in the induction of filopodial protrusion formation.
The N-Glycosylation Mutants of M6a in EC2 Are Not Affected in Their Function-To gain a more detailed insight into the functional importance of EC2 we were further interested in finding out whether the replacement of another conserved amino acid residue within EC2 different from cysteine would have the same negative effect on filopodium formation or cell membrane localization of the M6a protein. Previous studies have shown that most EC2 tetraspanins are glycosylated in one or more potential N-glycosylation sites. Therefore, among the possible sites of interest we focused on the two potential N-glycosylation sites that are analogously present in EC2 of M6a (Asn 164 and Asn 208 ).
To examine the potential involvement of the two N-glycosylation sites within the EC2 region in the induction of filopodium formation, asparagines 164 and 208 were substituted by glutamines individually or in combination. When N2a cells were transfected with these M6a mutants intense staining of plasma membrane was observed in all three cases (data not shown). Hence, the Asn 3 Gln mutations within EC2 did not disrupt the cell membrane localization of M6a protein.
Next, we analyzed the effect of these Asn 3 Gln mutations on filopodial protrusion formations in neuroblastoma cell lines. The N2a were transfected with GFP-tagged mutant proteins, WT protein, and GFP alone, and stained with the F-actin marker phalloidin. Fig. 9A shows that the number of cells showing filopodial protrusions in cells transfected with N164Q, N208Q, and double N164Q/N208Q mutants does not significantly differ from that in the WT M6a-transfected N2a cells. Thus, we conclude that Asn 3 Gln mutations within the EC2 domain do not interfere with either cell membrane localization of the protein or induction of filopodium formation.
Next, we were interested to see whether the Asn 3 Gln mutations within EC2 would affect anti-M6a antibody epitope recognition. Primary hippocampal neurons were transfected with the indicated GFPtagged mutants and fixed as described above. Subsequently, immunostaining was performed in non-permeabilized cells with rat anti-M6a mAb. We show that surface-exposed regions of N164Q, N208Q, as well as double N164Q/ N208Q GFP-tagged M6a mutants are completely recognized by anti-M6a antibody in hippocampal neurons (Fig. 9B). The complete colocalization can be observed in the merge image. We draw the conclusion that mutation of asparagine residues 164 and 208 within EC2 does not interfere with the binding of the function blocking anti-M6a antibody.
Neurons in Contact with Axons Expressing C174A/C192A Mutant Display Significantly Lower Density of Presynaptic Clusters over Their Dendrites-Synapse formation in the vertebrate central nervous system is a dynamic process, requiring bidirectional communication between pre-and postsynaptic partners. Real time imaging experiments demonstrate that both axonal and dendritic filopodia actively participate in synapse formation (22). In our previous study (10) we have demonstrated that M6a down-regulation using small interfering RNA technology causes not only decreased filopodium density but also reduced synaptic density in hippocampal neurons. Therefore, it seemed interesting to know whether overexpression of the EC2 cysteine C174A/C192A mutant affects synaptic density. It has been shown that synaptic vesicle antigens are concentrated in presynaptic elements at points of contact between axons and appropriate postsynaptic targets in hippocampal cultures (23).
To determine the number of synaptic puncta, primary hippocampal neurons after 9 days in vitro were transfected with the GFP-tagged C174A/C192A mutant, WT M6a, and GFP alone. Neurons were fixed as described previously and co-im-  NOVEMBER 13, 2009 • VOLUME 284 • NUMBER 46 munostained with the antibodies against synaptophysin 1 to visualize synaptic puncta and MAP2, a specific marker for dendrites. The number of synaptophysin 1 fluorescent puncta was counted (Fig. 10) in regions where transfected axons (identified by green labeling and their lack of MAP2 immunostaining) contacted non-transfected MAP2 positive dendrites (Fig. 10A). Fig.  10B shows that dendrites that are in contact with WT M6a expressing axons display significantly higher density of presynaptic clusters compared with the dendrites that are in contact with GFP-transfected axons. On the contrary, overexpression of the C174A/C192A mutant reduces the density of presynaptic clusters over a fixed length of dendrites when compared with WT M6a. Thus, we conclude that Cys 174 and Cys 192 mutations within EC2 of M6a not only affects filopodium formation but also synaptic density in hippocampal neurons.

DISCUSSION
The hippocampus is a particularly sensitive and vulnerable brain region that responds to stress and stress hormones. It is known that chronic stress affects the morphology and synaptic organization of the hippocampus, predominantly within CA3 but also CA1 and dentate gyrus (24 -29). In our previous studies, we demonstrated that chronic stress induces a decrease in hippocampal mRNA levels for the neuronal membrane glycoprotein M6a shown to modulate neurite outgrowth and filopodium/spine formation (10). The reduced M6a expression by chronic stress has been suggested to be related to the morphological alterations found in the hippocampus of chronically stressed animals. To understand the involvement of M6a in chronic stress response it was imperative to characterize the action of M6a at the molecular level. In the present work, we identify amino acid residues within M6a important for the function of the protein. and C202A mutants suggests that these residues may form a disulfide bond essential for proper protein folding and/or cell surface trafficking. The biotinylation of WT M6a further indicates also that free cysteines are present in the extracellular region of M6a. No increase in biotinylation of the C44A and C46A mutants compared with WT M6a suggested that these residues are either not involved in disulfide bond formation or their partners in a putative disulfide bond reside in an inaccessible region of the molecule. But because the double C44A/ C46A mutant showed a decrease in biotinylation compare to WT M6a we suppose that these two EC1 cysteines are free thiols responsible for biotinylation of WT M6a. Our results completely coincide with the data published on PLP from the same family of proteins, which we used here as a control. In the extracellular region of PLP, four conserved cysteines in EC2 are connected by disulfide bonds: Cys 200 -Cys 219 and Cys 183 -Cys 227 . On the other hand, Cys 32 and Cys 34 are free thiols. We suppose that these two cysteines account for the MTSEA-biotinylation of PLP in our experiment. Because of their localization at the border of the first transmembrane domain (TMD) and EC1 (almost buried in the plasma membrane) they are less accessible to MTSEA-biotin than Cys 44 and Cys 46 of M6a, thus resulting in less biotinylation of PLP compared with WT M6a.
The critical role of EC2 of M6a described by the present study is in agreement with studies published on the functional importance of the EC2 of tetraspannins. It has been firmly established that the functional specificity of tetraspanins is determined by the EC2 region. For example, strong CD151 association with integrins requires the Gln 194 -Arg 195 -Asp 196 site in the EC2 loop of CD151. Mutation of the QRD site not only causes loss of integrin association, but also disrupts integrindependent cell spreading (19). An additional example of the functional importance of EC2 in tetraspanins is the human brain-specific tetraspanin TM4SF2/A15 that is associated with mental retardation when inactivated by a point mutation, P172H. Notably, P172H also lies within the variable region of EC2 (30). This is also true in the case of peripherin/rds, a tetraspanin expressed exclusively in retinal rod and cone photoreceptor cells. Mutations of peripherin/rds lead to several retinal diseases in humans. The majority (about 70%) of the peripherin/rds mutations that cause disease are located within EC2 (31).
We speculate that EC2 of M6a may be involved in mediating specific lateral homotypic and/or heterotypic interactions with other partner molecules (transmembrane proteins or soluble ligands) that have to be identified and C174A and C192A mutations may lead to the loss of this intermolecular association. This hypothesis is based on the fact that the role of tetraspanins is mediated by their ability to interact with numerous proteins. Tetraspanins are involved in primary direct interactions with transmembrane proteins, including immunoglobulin superfamily members EWI-2 and EWI-F (32, 33), Claudin-1 (34), epidermal growth factor receptor membrane-bound ligands (35), integrins (36), other tetraspanins forming homodimers (37), as well as intracellular proteins such as syntenin-1 (38). In addition, secondary interactions of tetraspanins include not only multiple distinct tetraspanins, and many other transmembrane proteins (39), but also signaling enzymes such as PKC isoforms (40) and phosphatidylinositol 4-kinase (41). Through the network of homotypic and heterotypic interactions, tetraspanins regulate the spatial juxtaposition of associated proteins on the plasma membrane, which results in coordination of signaling pathways and cross-talk with cytoskeletal structures (11,42,43). The putative protein-protein interaction sites of M6a may map to EC2, and the alteration of EC2 conformation by Cys 3 Ala mutations would thus affect its function supposedly by changing its ability to interact with potential binding partners. The identification of the M6a interactions is thus the next important step that would provide further information about possible signaling pathways or protein-protein interactions through which M6a could act to affect neuronal remodeling.
Our hypothesis that the potential M6a interactions with external ligands and/or specific homo/heterotypic protein-protein interactions could consequently regulate signaling pathways inside the cell is well justified by the previous finding that monoclonal anti-M6a antibody directed against the extracellu- tein in the induction of filopodial protrusion formation. This would be in accordance with M6a structural analogy to tetraspanins. Although there is little known about the structure and function of EC1 in tetraspanins, in one study where EC1 from CD81 was replaced with a short linker sequence, EC1 was shown to be required for optimal expression of EC2, but not for binding to the hepatitis C virus E2 protein (44). Besides, it has been shown that in tetraspanins, monoclonal antibodies that recognize cell surface epitopes so far exclusively recognize EC2.
Another possibility is that Cys 3 Ala mutations within EC2 disrupt lateral interactions indirectly by affecting the proper packing of M6a TMDs. Tetraspanin TMDs have been shown to be involved in crucial intramolecular and intermolecular interactions that are essential for correct intramolecular folding and intermolecular associations with other transmembrane proteins in tetraspanin-enriched domains (45). Notably, a recent study by Wu et al. (46) that identified the -opioid receptor to be associated with M6a demonstrated that TMDs 3 and 4 are required for direct interaction of the -opioid receptor-M6a.
M6a belongs to the proteolipid protein (PLP/DM20) family of myelin proteins (1) that includes glia-specific PLP/DM20, and neuron-specific M6a and M6b detected in both neuronal and glial populations (5,47). Two distinct integrin ␣ subunits mediate the outgrowth of processes in nerve and glial cells. Process extension by glial cells is likely to involve the ␣ V integrin. In contrast, ␣1 integrin, a laminin receptor, is up-regulated by nerve growth factor and other differentiation stimuli and involved in neurite extension by nerve cells (48). PLP/ DM20 has been shown to participate in integrin receptor signaling in oligodendrocytes (49). It interacted directly with the cytoplasmic domain of the ␣ v integrin and this complex appeared important for binding of fibronectin to oligodendrocytes. PLP antibody specific for the extracellular loop domain of PLP inhibited binding of fibronectin to oligodendrocytes (49). Given the high degree of homology between M6a and PLP, it is also possible that M6a is related to integrin signaling.
A possible role of M6a in the formation of synapses has been suggested (7,10). In the mouse hippocampal tissue, M6a mRNA expresses in granule cells of the dentate gyrus and pyramidal neurons in CA1 and CA3. Immunoreactivity for M6a is concentrated in the regions of relatively dense synaptic contact (10). Accordingly, ultrastructural studies have located M6a immunoreactivity in the synapses of the rat cerebellum and in the axon terminals of the rat cerebellar molecular layer (50). Moreover, neuroanatomical data demonstrated that M6a is an axonal component of glutamatergic neurons and that it is localized to distinct sites of the axonal plasma membrane of pyramidal and granule cells in adult rat forebrain and cerebellum (51). In the mouse retina, the expression of M6a parallels that of the presynaptic protein, synaptophysin (7). In our previous study, we found reduced synaptophysin immunoreactivity and lower synaptic cluster density in cultures with decreased M6a levels (10). It has been established that synapse formation by dissociated neurons in culture strongly correlates with focal accumulations of structures that can be labeled with antibodies against synaptic vesicle proteins, such as synaptophysin 1 and synapsin 1 (23). Our present study shows that dendrites that are in contact with EC2 mutant C174A/C192A expressing axons display significantly lower density of presynaptic clusters compared with the dendrites that are in contact with WT M6a-transfected axons.
It has been observed that profound changes in the morphology of the mossy fiber terminals and significant loss of synapses are detected in hippocampal formation of chronically stressed and corticosterone-treated rats. These alterations are partially reversible following rehabilitation from stress or corticosterone treatments (29). Thus, with respect to the apparent enrichment of M6a within mossy fibers (51) and the fact that in hippocampal formation chronic stress reduces M6a mRNA expression (2,3), it appears feasible that M6a may regulate presynaptic plasticity possibly by mediating molecular interactions with specific partner molecules and the EC2 domain with its four cysteine residues playing a crucial role. Consequently, stress-induced changes in M6a expression may modify the molecular interactions with functional consequences for neuronal remodeling.