A common mechanism of PLP/DM20 misfolding causescysteine-mediated endoplasmic reticulum retention inoligodendrocytes and Pelizaeus–Merzbacher diseaseAjit-Singh Dhaunchak* and Klaus-Armin Nave*†‡

*Department of Neurogenetics, Max Planck Institute of Experimental Medicine, D-37075 Gottingen, Germany; and †Hertie Institute of Multiple SclerosisResearch, D-37075 Gottingen, Germany

Edited by Floyd E. Bloom, The Scripps Research Institute, La Jolla, CA, and approved August 21, 2007 (received for review May 26, 2007)

A large number of mutations in the human PLP1 gene lead toabnormal myelination and oligodendrocyte death in Pelizaeus–Merzbacher disease (PMD). Here we show that a major subgroupof PMD mutations that map into the extracellular loop region ofPLP/DM20 leads to the failure of oligodendrocytes to form thecorrect intramolecular disulfide bridges. This leads to abnormalprotein cross-links and endoplasmic reticulum retention and acti-vates the unfolded protein response. Importantly, surface expres-sion of mutant PLP/DM20 can be restored and the unfolded proteinresponse can be reverted by the removal of two cysteines. Thus,covalent protein cross-links emerge as a cause, rather than as aconsequence, of endoplasmic reticulum retention.

cysteine bridge � membrane protein misfolding � proteolipid protein �quality control � dysmyelination

In the nervous system, integral membrane proteins, such asreceptors, channels, or adhesion proteins, have been associ-

ated with a wide spectrum of neurodegenerative disorders. Somemultimeric proteins, including ion channels and connexins, canonly exit the endoplasmic reticulum (ER) when properly assem-bled (1). However, genotype–phenotype correlations remaindifficult. Specifically, substitutions in extracellular loop regionsrarely show obvious relationships to the presumed function ofthe protein. A better understanding of the essential features ofmisfolded proteins that are retained in the ER (2, 3) is requiredto develop therapeutic strategies.

Misfolded proteins in the ER can induce the unfolded proteinresponse (UPR), which includes ER growth and transcriptionalactivation of genes encoding chaperones (4, 5). In mammaliancells, the UPR can also trigger apoptosis (reviewed in refs. 6 and7). Not surprisingly, in many diseases abnormal cell death hasbeen associated with mutant membrane proteins.

We have studied the molecular consequences of missense mu-tations in the PLP1 gene, encoding the major integral membraneprotein of CNS myelin. Numerous PLP1 missense mutations causeER retention and oligodendrocyte death in Pelizaeus–Merzbacherdisease (PMD), whereas null mutations of the same gene are welltolerated and allow myelination (8, 9). (For a comprehensive listof PLP1 mutations, see www.med.wayne.edu/neurology/clinicalprograms/pelizaeus-merzbacher/plp.html.)

PLP and its splice isoform, DM20, are tetraspanins with twoextracellular loop regions, EC1 and EC2, that interact in vivowith the opposing membrane in myelin (10, 11). Both the N andC termini of PLP protrude into the cytosol (Fig. 1A). Thelocation of two disulfide bridges within EC2 is known (C183–C227

and C200–C219) (12).Because many PLP1 missense mutations lead to oligodendrocyte

death (13), it is difficult to dissect the subcellular pathomechanismin vivo. However, overexpression in nonglial cells has demonstratedER retention of mutant PLP/DM20 and its interaction with chap-erones (14–17), providing indirect evidence of misfolding also in theabsence of structural data. For a specific substitution in transmem-brane domain 4 of PLPA242V, it has been demonstrated that

premature oligomerization causes ER retention (18). For themajority of PLP1 mutations, which involve the extracellular loopregion (Fig. 1A), the disease mechanism is unclear.

Results and DiscussionIn the search for an essential feature of membrane proteinmisfolding, we used the oligodendroglial cell line oli-neu (19), inwhich endogeneous proteolipids are practically undetectable byimmunostaining (data not shown). We studied PLP with a mycepitope or fused to enhanced green fluorescent protein (EGFP)separated by a flexible linker (Fig. 1 A). In all these experiments,the wild-type protein PLPWT, including tagged PLPWT isoforms,passed the ‘‘quality control’’ of the ER, exhibited surfaceexpression within 8 h of transfection [see supporting information(SI) Movie 1], and accumulated in LAMP1-positive late endo-somes within 24 h (Fig. 1B, SI Fig. 5A, and SI Movie 2). In contrast,PLP mutants were retained in the ER of oligodendrocytes (Fig.1B and SI Movie 3). Surface expression of wild-type PLP wasconfirmed by live staining (SI Fig. 6A) with conformation-sensitive monoclonal antibody O10 (15).

Cysteine Residues Are Critical for PLP Folding and Surface Expression.We hypothesized that in PMD cysteine residues in EC2 fail toform normal disulfide bridges as an essential feature of prote-olipid misfolding and ER retention. First, we analyzed corre-sponding cysteine-to-serine and cysteine-to-alanine mutants(Fig. 1B and SI Fig. 5). PLP lacking the ‘‘inner’’ bridge C183–C227

was strictly retained in the ER, as indicated by a reticularimmunostaining of cells that also lacked visible processes (Fig.1B Right and SI Fig. 5B). Thus, the membrane-proximal bridgeis essential to passing the quality control system of the ER.Unexpectedly, PLP lacking the ‘‘outer’’ bridge, C200–C219, wasreadily detectable on the cell surface (Fig. 1B Left), very similarto wild-type PLP. These oli-neu cells also exhibited a ‘‘mature’’multipolar morphology with numerous filopodial processes(magnified in Fig. 1B Insets). Surface expression was confirmedby live staining with the monoclonal antibody 3F4 (20) againstan extracellular PLP epitope (Fig. 1 A and data not shown). Asexpected, a quadruple mutant (PLP lacking all four cysteines)was strictly retained in the ER (data not shown).

To biochemically confirm the presence (or absence) of PLP at

Author contributions: A.-S.D. and K.-A.N. designed research; A.-S.D. performed research;A.-S.D. and K.-A.N. analyzed data; and A.-S.D. and K.-A.N. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Abbreviations: ER, endoplasmic reticulum; UPR, unfolded protein response; PMD,Pelizaeus–Merzbacher disease; ME, mercaptoethanol; DPBS, Dulbecco’s PBS; RT, roomtemperature; TBS, Tris-buffered saline.

‡To whom correspondence should be addressed. E-mail: nave@em.mpg.de.

This article contains supporting information online at www.pnas.org/cgi/content/full/0704975104/DC1.

© 2007 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0704975104 PNAS � November 6, 2007 � vol. 104 � no. 45 � 17813–17818

NEU

ROSC

IEN

CE

the cell surface, we transfected COS7 cells and biotinylated allsurface proteins before harvesting and precipitated the markedproteins with streptavidin-conjugated agarose beads. Subse-quent Western blot analysis demonstrated that PLPWT,PLPC200S, and PLPC219S were biotinylated cell membrane pro-teins. In contrast, PLPC183S and PLPC227S were almost undetect-able (Fig. 1C), confirming their intracellular retention. Interest-ingly, although PLP lacking the disulfide bridge C200–C219

reached the surface of transfected cells and accumulated in aLAMP1-positive endosomal compartment (similarly to PLPWT)(SI Fig. 5 A and C–E), it remained a monoclonal antibodyO10-negative misfolded mutant (SI Fig. 6C). The absence of theconformation-sensitive O10 epitope (compare SI Fig. 6 A and C)in surface-expressed PLP cysteine mutants demonstrates that PLPmisfolding and ER retention can, in principle, be uncoupled.

Although the outer disulfide bridge in EC2 (C200–C219) isdispensable for surface expression (Fig. 1B), we found a naturalPMD mutation (21) of one of these cysteines (PLPC219Y) withobvious ER retention (Fig. 2 A and B). Because PLPC219S did notcause ER retention (Fig. 1B), we reasoned that the unpaired C200

would be sterically more exposed when a residue larger thanserine (Y219) caused local misfolding and that the exposed C200

was the real cause of ER retention. Indeed, substituting C200 ina double-mutant protein (PLPC200S,C219Y) completely rescuedthe mutant phenotype (Fig. 2 C and D). Western blot analysisrevealed that PLPC219Y (but not PLPWT or the rescued doublemutant) formed dimers that were sensitive to reducing agents(Fig. 2E). This strongly suggests that the retention of PLPC219Y

is caused by cysteine (C200) oxidation and abnormal cross-links(which we think prevent subsequent oligomerization). In theseexperiments, true dimers on gels could be distinguished fromother adducts by coexpression of PLP–myc with PLP–EGFP andthe emergence of novel hybrid bands (SI Fig. 7, lane 1). Apparently,the inability to form normal cysteine bridges leads to alternativeintermolecular cysteine oxidation products, as was recently hypoth-esized for mutant TNFR1 and collagen X chains (22, 23).

amino acids in EC2 that are substituted in patients with PMD are marked inyellow. Those that have been studied in detail here carry the single-letter codeof the wild-type sequence, labeled in red. (B) In transfected, fixed, andpermeabilized oli-neu cells, mutant PLP–myc (shown in green) colocalizedwith the endogenous chaperone and thiol-disulfide oxidoreductase PDI(shown in red). Only merged images are shown. (Left) Wild-type protein,PLPWT, reached the cell surface, as demonstrated by green fluorescent micro-spikes at the tip of processes (magnified in Insets). (Right) In contrast, PLPMSD

(derived from jimpy-msd mice) failed to reach the cell surface. There are nolabeled microspikes (magnified in Insets), but there is substantial overlapbetween PLPMSD and PDI (shown in yellow). Note also the paucity of cellularprocesses. From two disulfide bridges in EC2, the ‘‘outer’’ one (Cys200–Cys219)is dispensable for folding and cell-surface expression. To test the function ofeach disulfide bridge (see A) for PLP folding and colocalization with PDI(shown in red), single and double cysteine-to-serine substitutions were engi-neered for each disulfide bridge. Replacing one or both cysteines of the outerbridge did not interfere with cell-surface labeling of PLP (shown in green), asindicated by fluorescent microspikes (Left and magnified in Insets). In contrast,replacing one or both cysteines of the membrane-proximal bridge (Cys183–Cys227) led to severe misfolding, as visualized by ER retention and colocaliza-tion of PLP with PDI (shown as yellow overlay in Insets), similar to PLPMSD

(topmost Right). Note also the paucity of cellular processes. (Scale bar: 20 �m.)(C) To obtain independent biochemical evidence that PLPC200S and PLPC219S

(lacking the outer bridge) are cell-surface expressed in COS7 cells, all mem-brane proteins were biotinylated with membrane-impermeable sulfo-NHSbiotin before cell lysis, and proteins were pulled down with streptavidin-conjugated agarose beads. Only myc-epitope labeled PLPWT, PLPC200S, andPLPC219S could be detected on Western blots (the first, third, and fourth lanes).(Left) The absence of actin in these six lanes confirms that only live cells werebiotinylated. Total lysates served as positive controls for transfection andloading. Mock, transfected with plasmids lacking a cDNA insert.

Fig. 1. Topology of proteolipid protein PLP/DM20 and the role of cysteineresidues in subcellular trafficking. (A) Two-dimensional model of PLP (276residues shown as black beads) and its splice isoform, DM20, lacking 35residues (marked in gray) from an intracellular loop. The orientation of fourtransmembrane domains (TM1–TM4) positions both N and C termini in thecytoplasm. Within the second extracellular domain (EC2), the position of fourcysteine residues (marked in green) forming two disulfide bridges (marked inred) is indicated. Also indicated are C-terminal epitope tags used in this study(EGFP or myc) and the approximate positions of extracellular (3F4) and intra-cellular (A431) antibody binding sites common to PLP and DM20. Positions of

17814 � www.pnas.org�cgi�doi�10.1073�pnas.0704975104 Dhaunchak and Nave

Rescue of ER Retention and the UPR. PLPWT derived from primaryoligodendrocytes or transfected oli-neu cells migrated on gelslargely as monomers (26 kDa), under both reducing and non-reducing conditions (Fig. 2F). In contrast, PMD-causing iso-forms (PLPD202N, PLPR204G, and PLPC219Y) formed dimers that,upon the addition of mercaptoethanol (ME), were reduced tomonomers (Fig. 2G). Again, when two cysteines were replaced(Cys200,219Ser), these PMD-causing mutations yielded predom-

inantly monomeric PLP, also under nonreducing conditions (Fig.2G and data not shown). Thus, it is a feature of diverse PMDmutations in EC2 to cause alternative oxidation products andabnormal PLP dimers.

To prove the critical role of cysteines in PMD mutations,independent of the position of the primary substitution in EC2,we generated EGFP-tagged PLP isoforms with the followingPMD-causing mutations: PLPD202N, PLPR204G, PLPV208D,

Fig. 2. Critical cysteine residue in PLP/DM20 for ER retention and protein dimerization. (A–D) oli-neu cells were transfected to express a myc-epitope-labelednatural PMD mutant, PLPC219Y (A and B), or a double mutant, PLPC200S,C219Y (C and D). Only merged images are shown. The PMD mutant was retained as visualized24 h after transfection by lack of processes and colocalization of PLP (shown in green) with the ER marker PDI (shown in red in A). Mutant PLP (shown in green)also showed segregation from the late endosomal/lysosomal marker LAMP1 (shown in red in B and magnified in Inset). In contrast, the double mutant PLPbehaved like a wild-type protein (see Fig. 1B and SI Fig. 5A), with reduced colocalization with PDI and the branched morphology of oli-neu cells (shown in C)and the emergence of green fluorescent microspikes on processes (shown in D and magnified in Inset). PLPC200S,C219Y also overlapped with the endosomal marker(shown as the yellow area in D). Thus, it is not Y219 but an unpaired C200 that emerges as the cause of PMD. (Scale bar: 10 �m.) (E) When oli-neu cell extracts wereanalyzed by semiquantitative Western blotting, mutant PLPC219Y revealed predominantly a dimer band, whereas PLPWT and the ‘‘rescued ’’ PLPC200S,C219Y weremonomeric. In the presence of ME, all forms were monomeric, demonstrating that dimerization of PLPC219Y can be attributed to cysteine oxidation. (F) SDS/PAGEof cellular extracts from primary oligodendrocytes (OL) and oli-neu cells with immunodetection of PLP/DM20. When analyzed under nonreducing conditions,endogenous PLP/DM20 expression in cultured oligodendrocytes (lane 1) led to a small percentage of dimerized PLP and DM20, detected by antibody 3F4. Absenceof dimers in the presence of ME suggests that these are cysteine-mediated cross-links. Transfected oli-neu cells, expressing the slightly larger epitope-taggedPLP–myc (lane 2), also exhibited a small percentage of ME-sensitive PLP dimers, detected with antibody 3F4. Note that antibody 3F4 detects endogenous andinduced PLP/DM20 expression. (G) Detection of mutant PLP as dimers in oli-neu glial cells, transfected to express the natural PMD mutant PLPD202N. By performingWestern blot analysis of nonreducing gels (using antibody 3F4), �50% of total PLP was detectable in the 52-kDa putative dimeric form (lane 1). When PLP wasfurther modified by the substitution of C200 and C219 for serine (see also Fig. 3), such dimer formation was largely prevented (lane 2) with most PLP beingmonomeric, similar to wild-type PLP (compare with F). All dimers were completely absent in the presence of 150 mM ME in the gels (lanes 4 and 5), demonstratingthe involvement of cysteine cross-links. The same results were found for PLPR204G (data not shown). Note that endogenous DM20 (in mock-transfected cells)migrated exclusively as a monomer, even under nonreducing conditions (lanes 3 and 6).

Dhaunchak and Nave PNAS � November 6, 2007 � vol. 104 � no. 45 � 17815

NEU

ROSC

IEN

CE

PLPL209H, and PLPP215S. As expected, all mutant PLP isoformswere strictly retained in the ER of oli-neu cells (Fig. 3A Left).However, when combined with the Cys200,219Ser substitution,these triple-mutant PLP isoforms showed a wild-type-like sur-face expression, as demonstrated by live staining with themonoclonal antibody 3F4 (Fig. 3 A Right and B). Similarly, inCOS7 cells and in primary oligodendrocytes, EGFP-tagged PLPisoforms with a PMD mutation accumulated in the endosomalcompartment when rescued with the substitutions Cys200,219Ser(Fig. 4A and data not shown).

The expression of misfolded PLP in mutant mice induces BIPand CHOP mRNA (24). In agreement with these observations,we found elevated BIP and CHOP mRNA in transfected oli-neucells expressing PLPD202N, which was reduced to control levels byremoval of the two luminal cysteines (Fig. 4B). ER resident Ire1tranduces ER stress to the nucleus by splicing Xbp1 mRNA (25).Spliced Xbp1 mRNA encodes for an activated transcriptionfactor that enhances the transcription of ER chaperone genes.We exclusively detected spliced Xbp1 mRNA in transfectedoli-neu cells expressing PLPD202N. However, this splice productdisappeared upon removal of the luminal cysteines (Fig. 4C).

Taken together, our data reveal that a subset of disease-associated point mutations in PLP1 converge mechanistically byperturbing the formation of an intramolecular disulfide bridge inPLP/DM20 in the lumen of the ER. This disulfide bridge itselfis dispensable for normal PLP/DM20 folding and trafficking.Importantly, it is not the substituted amino acid itself that causesER retention. However, when the unpaired cysteine becomessterically exposed, it engages in intermolecular cross-links (withPLP itself or other proteins). Abnormal PLP adducts fail tooligomerize (i.e., are monoclonal antibody O10-negative) andbecome the primary cause of ER retention and, thus, oligoden-drocyte dysfunction and death in vivo. ER retention of PLP maycause further abnormal protein cross-links.

Many studies have demonstrated a critical role of cysteines inprotein retention that seems similar to our initial finding ofPLPC219Y retention (Fig. 2 A), which provides a plausible causeof PMD. Indeed, unpaired cysteines in secretory monomericproteins, such as immunoglobulins, were proposed as physiolog-ical retention signals (26, 27). Here, we went an important stepfurther by showing for PMD that several noncysteine mutations,which were mechanistically not understood (such as PLPD202N,PLPR204G, and others), do not alter the number of pairedcysteines but can still be explained by the cysteine pathway.

By performing a successful rescue experiment with compen-satory mutations, we could also solve an inherent problem ofearlier studies, namely, to distinguish between cause and effectof protein retention. Theoretically, chaperone-mediated PLPretention could be the primary cause of the cysteine-mediatedcross-links, if we assume that once proteins are retained in theER such cysteine-mediated cross-links occur nonspecifically(14). Alternatively, specific mutations lead to PLP misfoldingand the exposure of unpaired cysteines, which then become the

Fig. 3. PMD-causing PLP mutations can be rescued by the replacement ofcysteines. (A) To distinguish PLP at the cell surface from PLP in intracellularcompartments, oli-neu cells that express various EGFP-tagged mutant PLPisoforms (shown in green) were additionally live-stained for PLP by usingmonoclonal antibody 3F4 (shown in red; see Fig. 1A). Only merged images areshown. The same results were obtained with COS7 cells (data not shown). Weanalyzed various natural PMD-causing mutations that map into EC2 (PLPD202N,PLPR204G, PLPV208D, PLPL209H, and PLPP215S) for cell-surface expression of PLP inthe absence (Left) or the presence (Right) of additional point mutations thatsubstituted C200 and C219 for serine. Remarkably, in the absence of C200 andC219, the PMD-causing mutants were fully rescued from ER retention, and PLPwas localized at the cell surface (labeled in red) and in the late endosomal/

lysosomal compartment (shown in green). In the presence of EC2 cysteines, allPMD mutants were negative for 3F4 immunoreactivity and confined to the ER.Note also the lack of glial processes. This reveals that the ER retention of manynatural PMD-causing PLP mutants, which do not alter cysteine residues,instead require cysteines for ER retention. Grayscale images, shown on theright, better reveal the distribution of double-labeled PLP at the cell surface(3F4) and in intracellular compartments (GFP). (Scale bar: 10 �m.) (B) PLPD202N

trafficking to the cell surface in the presence or absence of C200 and C219. Thepercentage of cells live-stained by antibody 3F4 for surface-expressed PLP (ina population of cells expressing EGFP-tagged mutant PLP) were calculatedfrom transfection experiments (n � 3). The PMD mutation PLPD202N was fullyretained in the ER. Note that, in the absence of C200 and C219, PLPD202N wasrescued from ER retention, because 95% of GFP-positive cells were stained byantibody 3F4.

17816 � www.pnas.org�cgi�doi�10.1073�pnas.0704975104 Dhaunchak and Nave

primary cause of ER retention. Our study is in strong support ofthe second model, because we could release several of the knownmutant PLP isoforms from ER retention by the removal ofcysteines in triple-mutant proteins.

Our model is likely to apply to the subset of PMD casesassociated with single substitutions in EC2, although not all ofthem can be rescued by removal of the dispensable cysteines[such as in L223P (data not shown)]. The mechanism is unlikelyto explain the mutations that map into transmembrane domainsand act by premature oligomerization (18). However, given thatmany other membrane proteins, not only tetraspanins, harborintramolecular disulfide bridges in extracellular loop regions andare sensitive to point mutations in these domains (for example,see ref. 28), we suggest that the mechanism identified here mightbe relevant to a broader spectrum of diseases.

Materials and MethodsMolecular Cloning. Plasmid pPLP-EGFP was generated by fusingthe EGFP to the C terminus of PLP. The product was cloned intovector pEGFP-N1 by using EcoRI/NotI sites. All site-directedmutants were generated by circular amplification of this plasmidwith PFU Turbo DNA polymerase (Stratagene, La Jolla, CA),followed by DpnI digestion (New England Biolabs, Ipswich,MA) and transformation into Escherichia coli. Individual cloneswere analyzed by analytic restriction digestion and DNA se-quence analysis. Detailed cloning protocols and all primersequences are available on request.

DNA Transfection. COS7 cells were washed with PBS, trypsinized,pelleted, and resuspended in DMEM. Plasmids were used at 10�g per 300 �l of cell suspension [4 � 106 cells per milliliter ofDMEM and 10% (vol/vol) FBS] for electroporation with theBio-Rad (Hercules, CA) Gene Pulser (350 V and 450 �F).oli-neu cells and primary oligodendrocytes were either trans-fected with FuGENE 6 (Roche Applied Science, Indianapolis,IN) or by electroporation. For electroporation, cells were de-tached mechanically by pipetting, pelleted, and resuspended inSato medium (DMEM supplemented with transferrin, insulin,putrescine, sodium selenite, progesterone, L-thyroxine, andtriiodo-L-thyronine). Plasmids were used at 10 �g per 400 �l ofcell suspension (5 � 106 cells per milliliter, in Sato medium) forelectroporation (250 V and 960 �F). (Scale bars: 20 �m.)

Protein Biotinylation. For biotinylation of cell membrane proteins,the following steps were performed on water/ice slurry (unlessotherwise stated), and reagents were precooled to 4°C. Twenty-four hours after transfection, cells grown in poly(L-lysine)-coated six-well Falcon plates (Becton Dickinson, Franklin Lakes,NJ) were washed twice with Dulbecco’s PBS (DPBS) (0.7 mMCaCl2/2.6 mM KCl/1.5 mM KH2PO4/0.5 mM MgCl2/136 mMNaCl/8.1 mM Na2HPO4). Cells were then incubated for 30seconds with 0.5 mg/ml N-hydroxysulfosuccinimide biotin (sulfo-NHS biotin) (Pierce, Rockford, IL) in DPBS (prepared freshly).Before lysing the cells in 400 �l of lysis buffer 1 (50 mM Tris�Cl,pH 7.5/140 mM NaCl/1 mM EDTA/1 mM PMSF/1% TritonX-100/0.1% SDS) at room temperature (RT), unbound sulfo-NHS biotin was quenched by washing once with DPBS contain-ing 25 mM lysine monohydrochloride. Lysates were cleared bycentrifugation at 13,000 � g for 20 min at RT and incubated withstreptavidin-conjugated agarose beads for 2 h at RT. Agarosebeads were washed five times with lysis buffer and once with PBSat RT. Beads were finally boiled with 4� lithium dodecyl sulfate(LDS) loading buffer, separated on NuPAGE 4–12% Bis-Trisprecasted gels (Invitrogen, Carlsbad, CA), and immunoblottedfor PLP and actin by following standard procedures.

SDS/PAGE and Western Blot Analysis. Before lysing cells in 1� SDSloading dye [25 mM Tris, pH 6.7/1% SDS, 5% (vol/vol) glycerol/0.005% bromophenol blue] or in lysis buffer 2 (25 mM Tris, pH7.5/150 mM NaCl/1 mM EDTA/1% Triton X-100), free cysteineswere blocked by incubation in 13.3 mM iodoacetamide in DPBS.Samples were separated on 12% (wt/vol) SDS gels by usingBio-Rad Mini-PROTEAN 3 chambers and transferred to PVDFmembranes by using Invitrogen blotting chambers. Membraneswere blocked for at least 1 h at RT with 5% (wt/vol) nonfat drymilk in Tris-buffered saline (TBS) (50 mM Tris base/150 mMNaCl, pH 7.4). Primary antibody (blocking buffer) was appliedfor at least 1 h at RT (or overnight at 4°C). After four washes inbuffer (0.05% Tween 20 in TBS), HRP-conjugated secondaryantibodies were applied for 1 h, followed by four washes withwash buffer. Membranes were exposed by using the EnhancedChemiluminescence Detection kit (PerkinElmer, Wellesley, MA).

RNA Isolation and Quantitative RT-PCR. RNA was isolated fromtransfected cells by using the RNeasy Mini kit (Qiagen, Valencia,CA). Reverse transcription was performed on 1 �g of total RNAby using random nanomers and SuperScript II reverse transcrip-tase (Invitrogen). Real-time PCR was performed according tothe manufacturer’s instructions by using a rapid thermal cyclersystem (LightCycler, Applied Biosystems, Foster City, CA). Amaster mixture containing SYBR green, Taq polymerase, nu-cleotides, and buffer was mixed with primers designed by usingthe Roche Applied Science universal probe library (http://qpcr2.probefinder.com/input.jsp?organism � mouse) and 20 ngof cDNA. Quantitative RT-PCR was performed in triplicatesfrom pooled cDNA obtained from two independent transfec-

Fig. 4. Rescue of PLP trafficking in primary oligodendrocytes and the attenu-ation of the UPR. (A) Confocal analysis of primary oligodendrocytes expressingeither EGFP-tagged PLPD202N�C200,219S (Left) or EGFP-tagged PLPP215S�C200,219S

(Right). Note that both proteins traffic to the cell surface and accumulate in anendosomal compartment (shown at higher magnification in Insets), similar toEGFP-tagged wild-type PLP (data not shown) and similar to their behavior inoli-neucells (seeFig.3). (B)RelativemRNAlevelsof theUPRmarkersBIPandCHOPinPLP-expressingoli-neucells.Forquantitativereal-timePCR,cDNAwasobtained24 h after transfection. Expression of PLPD202N (an ER-retained mutant) resultedin the up-regulation of BIP and CHOP mRNA when compared with cells express-ing wild-type PLP. Note that triple mutant PLP is comparable with wild-type PLP.BIP and CHOP mRNA levels were normalized to PLP expression. (C) Xbp1 mRNAwas spliced in response to ER stress. oli-neu cells expressing ER-retained PLPD202N

induced splicing of a Xbp1 mRNA, whereas cells expressing triple mutant PLP onlyexhibited unspliced Xbp1. Xbp1(U) and Xbp1(S) denote unspliced and splicedmRNA, respectively. Treatment of cells with tunicamycin served as a positivecontrol to trigger the UPR.

Dhaunchak and Nave PNAS � November 6, 2007 � vol. 104 � no. 45 � 17817

NEU

ROSC

IEN

CE

tants. Actin and 18S mRNA were used as internal standards, andexpression levels were normalized to PLP mRNA. As positivecontrol, the UPR was induced by tunicamycin (2 �g�ml�1).

Antibodies. Mouse monoclonal antibodies were anti-c-myc (clone9E10 from Sigma, St. Louis, MO), anti-PLP (3F4, kindly pro-vided by Marjorie B. Lees, Eunice Kennedy Shriver Center,Waltham, MA), and anti-actin (Chemicon, Temecula, CA). Ratmonoclonal antibodies were anti-LAMP1 (clone cd107a fromPharMingen, San Diego, CA) and anti-PLP [O10 (15)]. Poly-clonal rabbit sera were against PDI (Stressgen Assay Designs,Ann Arbor, MI) and PLP (A431, against the C-terminal peptideGRGTKF). For immunofluorescence and Western blot analysis,f luorochrome- and HRP-conjugated secondary antibodies werepurchased from Dianova (Hamburg, Germany). Antibody dilu-tions for immunocytochemistry were as follow: anti-c-myc,1:2,000; 3F4 and O10, 1:50; anti-LAMP1, 1:200; anti-PDI, 1:200;and A431, 1:1,000. Cy2-conjugated secondary antibodies werediluted 1:1,000 and Cy3-conjugated antibodies were diluted1:2,500. For Western analysis, dilutions were as follow: anti-c-myc, 1:1,000; 3F4, 1:50; anti-actin, 1:2,000; and anti-mouse HRP,1:20,000.

Cell Culture. COS7 cells were maintained on untreated tissueculture dishes (Falcon, Becton Dickinson) in DMEM and 10%(vol/vol) FBS. Cells were grown at 37°C in a 5% CO2 atmo-sphere, and medium was changed every third day. For passagingcells, confluent plates were washed once with PBS, followed bya short trypsination with 0.05% trypsin EDTA (Sigma). oli-neucells (kindly provided by J. Trotter, Johannes Gutenberg Uni-versity, Mainz, Germany) were maintained in Sato mediumcontaining 1–5% (vol/vol) horse serum on poly(L-lysine)-coatedtissue culture dishes (Falcon, Becton Dickinson). Cells weregrown at 37°C in a 5% CO2 atmosphere. For passaging, conflu-ent plates were washed once with medium, followed by a shorttrypsination with 0.005% trypsin EDTA.

Immunocytochemistry. Immunostainings were carried out 18–20 hafter transfection. All steps were performed at RT, unless statedotherwise. Cells grown on poly(L-lysine)-coated coverslips werewashed once with TBS [25 mM Tris/136 mM NaCl/2.6 mM KCl,pH 7.5] and fixed for 5 min in 2% (wt/vol) paraformaldehyde/TBS. Cells were then washed twice for 10 min in TBS, perme-abilized with 0.01% saponin in TBS for 10 min, and blocked inblocking buffer [TBS containing 2% (vol/vol) goat serum, 2%(wt/vol) BSA, 0.02% biotin, and 0.1% porcine skin gelatin] forat least 30 min. Primary antibodies diluted in blocking bufferwere applied for at least 1 h at RT or overnight at 4°C. After

three washes in TBS (10 min each), f luorochrome-conjugatedsecondary antibodies were applied for at least 45 min. Afterthree washes with TBS (10 min each), cells were rinsed indistilled water and mounted in Aqua-Poly/Mount (Polysciences,Warrington, PA) on glass slides.

Live staining of cells with antibody 3F4 was performed at 4°Con water/ice slurry (unless stated otherwise). Cells grown onpoly(L-lysine)-coated coverslips (24-well plates) were washedonce with ice-cold DMEM, and primary antibodies diluted inice-cold DMEM were applied directly to cells for 10 min. Cellswere washed twice with DPBS and fixed with 2% (wt/vol)paraformaldehyde for 10 min. Cells were shifted to RT duringfixation. After two additional washes in DPBS, Cy3-conjugatedsecondary antibodies diluted in DMEM were applied for at least30 min. Cells were washed twice in DPBS, rinsed in doublydistilled H2O (ddH2O), and mounted. Live staining of cells withantibody O10 was performed as previously described (15).

Confocal Analysis. Fluorescent images were captured on an LSM510 confocal microscope (Zeiss Microimaging, Jena, Germany)with a 63� oil Plan Apochromat objective with an N.A. of 1.4(Zeiss MicroImaging). For time-lapse live-cell imaging, cover-slips with cells were mounted into a live-cell imaging chamberand observed in a low autofluorescence imaging medium [10mM Hepes-buffered, phenol-red-free DMEM (Gibco, Carlsbad,CA) and 1% horse serum] at 37°C. Temperature was controlledby means of a Tempcontrol 37-2 digital system (PeCon, Erbach,Germany) or a custom-built perfusion system. Images wereacquired at 10-sec intervals for the indicated time periods.

For final analysis, captured laser scanning microscopy (LSM)images were exported as tagged image file (TIF) images. Doc-umentation and processing of TIF images were done withPhotoshop 7.0.1 (Adobe Systems, San Jose, CA). For quantifi-cation of Western blots, only nonsaturated developed Westernblots were used and scanned as 8-bit, 256-level grayscale images[at 1,200 dots per inch (1 inch � 2.54 cm) on an Epson (LongBeach, CA) F-3200 scanner]. Scanned images were first trans-formed to 512 � 512 pixels by using Photoshop 7.0.1 andexported as TIFs. Intensities of individual bands were quantifiedby using Bio-Rad Quantity One software.

We thank A. Helenius, M. Simons, E. Kramer, H. Werner, and membersof the K.-A.N. Laboratory for helpful discussions. We also thank J.Trotter for oli-neu cells and M. Lees for providing antibody 3F4. Ourwork is supported by grants from the Deutsche ForschungsgemeinschaftCentre for Molecular Physiology of the Brain (SFB523), the National MSSociety, the Hertie Institute of Multiple Sclerosis Research, the MyelinProject, and the Del Marmol Foundation.

1. Hurtley SM, Helenius A (1989) Annu Rev Cell Biol 5:277–307.2. Ellgaard L, Helenius A (2003) Nat Rev Mol Cell Biol 4:181–191.3. Ellgaard L, Molinari M, Helenius A (1999) Science 286:1882–1888.4. Rutkowski DT, Kaufman RJ (2004) Trends Cell Biol 14:20–28.5. Patil C, Walter P (2001) Curr Opin Cell Biol 13:349–355.6. Harding HP, Calfon M, Urano F, Novoa I, Ron D (2002) Annu Rev Cell Dev

Biol 18:575–599.7. Federovitch CM, Ron D, Hampton RY (2005) Curr Opin Cell Biol 17:409–414.8. Klugmann M, Schwab MH, Puhlhofer A, Schneider A, Zimmermann F,

Griffiths IR, Nave KA (1997) Neuron 18:59–70.9. Boison D, Stoffel W (1994) Proc Natl Acad Sci USA 91:11709–11713.

10. Stoffel W, Subkowski T, Jander S (1989) Biol Chem Hoppe-Seyler 370:165–176.11. Popot JL, Pham Dinh D, Dautigny A (1991) J Membr Biol 120:233–246.12. Weimbs T, Stoffel W (1992) Biochemistry 31:12289–12296.13. Gow A, Southwood CM, Lazzarini RA (1998) J Cell Biol 140:925–934.14. Swanton E, High S, Woodman P (2003) EMBO J 22:2948–2958.15. Jung M, Sommer I, Schachner M, Nave KA (1996) J Neurosci 16:7920–7929.16. Gow A, Lazzarini RA (1996) Nat Genet 13:422–428.17. Southwood C, Olson K, Wu CY, Gow A (2007) J Neurosci Res 85:471–478.18. Swanton E, Holland A, High S, Woodman P (2005) Proc Natl Acad Sci USA

102:4342–4347.

19. Jung M, Kramer E, Grzenkowski M, Tang K, Blakemore W, Aguzzi A, KhazaieK, Chlichlia K, von Blankenfeld G, Kettenmann H, et al. (1995) Eur J Neurosci7:1245–1265.

20. Greer JM, Dyer CA, Pakaski M, Symonowicz C, Lees MB (1996) NeurochemRes 21:431–440.

21. Osaka H, Kawanishi C, Inoue K, Onishi H, Kobayashi T, Sugiyama N,Kosaka K, Nezu A, Fujii K, Sugita K, et al. (1999) Ann Neurol45:59–64.

22. Lobito AA, Kimberley FC, Muppidi JR, Komarow H, Jackson AJ, Hull KM,Kastner DL, Screaton GR, Siegel RM (2006) Blood 108:1320–1327.

23. Wilson R, Freddi S, Chan D, Cheah KS, Bateman JF (2005) J Biol Chem280:15544–15552.

24. Southwood CM, Garbern J, Jiang W, Gow A (2002) Neuron 36:585–596.25. Yoshida H, Matsui T, Yamamoto A, Okada T, Mori K (2001) Cell

107:881–891.26. Fra AM, Fagioli C, Finazzi D, Sitia R, Alberini CM (1993) EMBO J

12:4755–4761.27. Sitia R, Neuberger M, Alberini C, Bet P, Fra A, Valetti C, Williams G, Milstein

C (1990) Cell 60:781–790.28. Kleopa KA, Scherer SS (2006) Neuromol Med 8:107–122.

17818 � www.pnas.org�cgi�doi�10.1073�pnas.0704975104 Dhaunchak and Nave