Defective Endoplasmic Reticulum-resident Membrane Protein CLN6 Affects Lysosomal Degradation of Endocytosed Arylsulfatase A*

Variant late infantile neuronal ceroid lipofuscinosis, a lysosomal storage disorder characterized by progressive mental deterioration and blindness, is caused by mutations in a polytopic membrane protein (CLN6) with unknown intracellular localization and function. In this study, transient transfection of BHK21 cells with CLN6 cDNA and immunoblot analysis using peptide-specific CLN6 antibodies demonstrated the expression of a ∼27-kDa protein that does not undergo proteolytic processing. Cross-linking experiments revealed the presence of CLN6 dimers. Using double immunofluorescence microscopy, epitope-tagged CLN6 was shown to be retained in the endoplasmic reticulum (ER) with no colocalization with the cis-Golgi or lysosomal markers. The translocation into the ER and proper folding were confirmed by the N-linked glycosylation of a mutant CLN6 polypeptide. Pulse-chase labeling of fibroblasts from CLN6 patients and from sheep (OCL6) and mouse (nclf) models of the disease followed by immunoprecipitation of cathepsin D indicated that neither the synthesis, sorting nor the proteolytic processing of this lysosomal enzyme was affected in CLN6-defective cells. However, the degradation of the endocytosed index protein arylsulfatase A was strongly reduced in all of the mutant CLN6 cell lines compared with controls. These data suggest that defects in the ER-resident CLN6 protein lead to lysosomal dysfunctions, which may result in lysosomal accumulation of storage material.

Neuronal ceroid lipofuscinoses (NCLs), 1 also collectively called Batten disease, are a group of autosomal recessively inherited neurodegenerative diseases that affect both children and adults. The clinical features include loss of vision, seizures, and mental and motor deterioration. Depending on the affected gene, the clinical signs become visible at various developmental stages (1). Currently, eight forms of human NCLs (CLN1-8) can be differentiated for which six underlying genes have been identified (2). CLN1 and CLN2 encode for soluble lysosomal enzymes, palmitoyl protein thioesterase 1 and tripeptidylpeptidase 1, respectively (3)(4)(5). CLN3, CLN5, CLN6, and CLN8 represent transmembrane proteins of still unknown function (6 -10). Although the nature of these CLN genes differs, defects in most of them lead to massive storage of autofluorescent material in lysosomes that has been identified as the subunit c of the mitochondrial ATP synthase (2,11,12).
Newly synthesized palmitoyl protein thioesterase 1 and tripeptidylpeptidase 1 are transported to the lysosome in a Man-6-P-dependent manner (3,5). Binding to Man-6-P-specific receptors, MPR46 and MPR300, allows the segregation from the secretory pathway and transport from the trans-Golginetwork to the endosomal compartment. Because of the low pH, the receptor-ligand complexes dissociate and the enzymes are delivered to the lysosome. Soluble enzymes that escape binding to the MPR in the trans-Golgi network are secreted but can be partially endocytosed after binding to MPR localized at the cell surface and transported to the lysosome (13). CLN3, CLN5, and CLN8 have been localized in lysosomes and the endoplasmic reticulum (ER)/ER-Golgi-intermediate compartment, respectively (14 - 16). The subcellular localization of CLN6 is not known. The transport of lysosomal membrane proteins requires tyrosine-or dileucine-based sorting signals (17). Recently, lysosomal targeting motifs have been identified in CLN3 protein (18), whereas the localization of CLN8 in the ER is mediated by a KKXX-ER-retrieval signal (16).
The CLN6 gene encodes a highly conserved 311 amino acid protein with six to seven predicted membrane-spanning domains. The deduced amino acid sequence contains no putative N-glycosylation sites or classical lysosomal sorting signals. Furthermore, no sequence homologies to other proteins have been described (9,10).
Various mutations of the CLN6 gene have been identified, leading to the variant late infantile form of NCL (9, 10, 19 -21). Naturally occurring NCLs in South Hampshire and Merino sheep (OCL6) and nclf mice have been localized to CLN6 (22)(23)(24). The severe and progressive neurodegeneration of the cerebral cortex is accompanied by astrocytosis and elevated expression of the radical scavenger protein Mn-SOD (24,25). Furthermore, differences in the fatty acid profiles of brain phosphatidylethanolamine have been observed in OCL6-affected sheep (26).
In this study, we expressed human CLN6 in BHK21 cells to investigate its subcellular localization. Furthermore, the transport of lysosomal enzymes via the biosynthetic and endocytic pathway was examined in fibroblasts from CLN6 patients, OCL6, and nclf mice. The data indicate that CLN6 is an ERresident protein.
The mutation p.Ile153Ser was introduced by PCR using the Quick-Change site-directed mutagenesis kit (Stratagene) and the primers Ile153Serfor 5Ј-GAGAACCCCAGCATCAAGAAT-3Ј and Ile153Serrev 5Ј-ATTCTTGATGCTGGGGTTCTC-3Ј. All of the clones were sequenced to rule out PCR-introduced mutations and to verify reading frames of fusion proteins.
Antibodies-Polyclonal antibodies against mouse CtsD and human CtsD purified from mouse liver and human placenta and against human MPR300 were raised in rabbits (28 -30). The polyclonal anti-Mycantibody was obtained from Santa Cruz Biotechnology (Ors, Santa Cruz, CA). The monoclonal anti-Myc-antibody, a gift from J. Blanz (ZMNH, Hamburg, Germany), was produced in hybridoma cell lines purchased from LGC Promochem (Teddington, United Kingdom), and the monoclonal antibody against protein-disulfide isomerase (PDI) was purchased from Stressgene Biotechnologies Corp. (Victoria, British Columbia, Canada). The polyclonal anti-CLN6 antibody was raised in rabbits against the peptide sequences 155-168 and 281-293 (Eurogentech, Seraing, Belgium) and purified by sequential affinity chromatography on a glutathione S-transferase and a glutathione S-transferase-CLN6-(137-181) fusion protein matrix. Anti-mouse IgG-Cy3 and anti-rabbit IgG-fluorescein isothiocyanate conjugates were obtained from Sigma, and the peroxidase-conjugated goat-anti-rabbit IgG was from Jackson Immu-noResearch Laboratories (West Grove, PA).
Isolation and Culture of Mouse Fibroblasts-CLN6-defective mice (B6.Cg-nclf) were obtained from The Jackson Laboratory (Bar Harbor, ME). Age-matched C57BL/6 mice served as controls. The animals were maintained and killed according to institutional guidelines in animal facilities of the University Hospital Hamburg-Eppendorf. Fibroblasts were prepared from lung biopsies of 3-month-old mice as described previously (31).
Cell Cultures and Cell Transfection-Skin fibroblasts from affected South Hampshire and age-matched control sheep as well as those from CLN6 patients were the same as those described previously (25). The cells were cultured in Dulbecco's modified Eagle's medium (Invitrogen) containing 10% fetal calf serum (PAA Laboratories, Linz, Austria) and 1% penicillin/streptomycin in a humidified atmosphere containing 5% CO 2 , 95% O 2 at 37°C. BHK21 cells were cultured under the same conditions in DMEM/fetal calf serum and transiently transfected as described previously (32). Control cells were transfected with pcDNA3.1/Myc-His(Ϫ) A vector alone.
Metabolic Labelings and Immunoprecipitation-Cells were labeled with [ 35 S]methionine (150 Ci/ml) for 1 h followed by a chase for the indicated time. CtsD was immunoprecipitated from cell extracts and media as described previously (33) and analyzed by SDS-PAGE and fluorography.
Cell Surface Biotinylation and MPR300 Immunoprecipitation-Cell surface proteins were biotinylated with sulfo-NHS-SS-biotin at 4°C as described previously (34). After washing, the cells were incubated in the presence or absence of 50 mM MESNA in 50 mM Tris buffer, pH 8.6, containing 100 mM NaCl, 1 mM MgCl 2 , and 0.1 mM CaCl 2 followed by immunoprecipitation of MPR300 from cell extracts as described previously (29).
Ligand Internalization-Internalization of [ 125 I]ASA was examined as described previously (33). The internalized [ 125 I]transferrin was determined after 3 h of incubation at 37°C. Cells were washed, and cell surface-bound [ 125 I]transferrin was displaced by an acidic wash (35). Cells were solubilized and analyzed by SDS-PAGE and autoradiography.
SDS-PAGE and Western Blotting-Solubilized proteins from transfected and non-transfected cells were separated by SDS-PAGE, transferred to nitrocellulose membrane, and examined for CLN6-and Myc immunoreactivity as described previously (25).
Cross-link Experiments-CLN6-overexpressing BHK21 cells were permeabilized with 0.25% saponin in 50 mM MES buffer, pH 6.5, containing 150 mM NaCl and 0.5% bovine serum albumin for 40 min at 4°C and then cross-linked with BS 3 and DSS at the indicated concentrations. The proteins were analyzed by Western blotting (36, 37).
Other Methods-Protein concentrations were determined using the Bradford protein assay with bovine serum albumin as a standard. Proteins (0.5-1 g) were iodinated with IODO-GEN as described previously (38). For deglycosylation experiments, total BHK21 extracts were solubilized and incubated in the presence or absence of 1 milliunit of peptide N-glycosidase F (PNGase F) or 5 milliunits of endoglycosidase H (endo H, Roche Applied Science) according to the manufacturer's protocol. The samples were separated by SDS-PAGE followed by Western blot analysis. Double immunofluorescence microscopy was performed with transfected BHK21 cells as described previously (25) using anti-Myc (1:50), anti-PDI (1:300), or anti-mouse CtsD antibodies (1:50) as indicated. ␤-Hexosaminidase and ASA activities were determined as described previously (39,40).

Expression of Wild Type and
Myc-tagged CLN6 -Wild type and C-terminally Myc-tagged CLN6 were transiently expressed in BHK21 cells. Non-transfected and pcDNA3.1 vectortransfected BHK21 cells were used as controls. Western blot analysis using the peptide-specific anti-CLN6 antibody revealed a single prominent immunoreactive band of ϳ27 kDa in cell extracts of wild type CLN6-transfected cells (Fig. 1). No specific cross-reacting material was found in cells transfected with the vector only. An immunoreactive band of ϳ30 kDa was observed in cells expressing the CLN6Myc construct. After stripping of the nitrocellulose membrane and reprobing with anti-Myc antibodies, only the ϳ30-kDa Myc-tagged CLN6 was detectable.
CLN6 Is an ER-resident Protein-The deduced amino acid sequence and the predicted protein structure of CLN6 revealed neither potential N-glycosylation site sequences (NX(S/T), where X ϭ any amino acid) nor classical lysosomal sorting motifs. On the other hand, the N-terminal RRR sequence (amino acids residues 5-7) has been detected in the ER type II membrane protein glucosidase I (41), whereas the N-terminal segment of CLN6 encoded by exon 1 (amino acids residues 1-28) has been suggested to function as a putative targeting signal for mitochondria (10). To investigate its subcellular localization, an N-glycosylation site was introduced in the predicted second luminal loop. Ile residue 153 was substituted for a Ser (p.Ile153Ser) by site-directed mutagenesis of the CLN6 cDNA changing the sequence Asn-X-Ile to Asn-X-Ser. When the p.Ile153Ser was expressed in BHK21 cells, Western blot analysis revealed an additional band of ϳ33 kDa, which disappeared after deglycosylation with PNGase F or endo H (Fig.  2A). These data demonstrate that the nascent CLN6 polypeptide had been translocated to the lumen of the ER and became accessible to the glycosylation machinery. This was confirmed  by double immunofluorescence microscopy. The CLN6Myc protein colocalized with the ER marker protein PDI but not with the lysosomal marker enzyme cathepsin D (Fig. 2B, CtsD).
CLN6 Forms Dimers-The capability of CLN6 to form oligomeric structures in membranes was examined by cross-linkage experiments. Semi-intact BHK21 cells overexpressing CLN6 were cross-linked with the non-cleavable reagents BS 3 or DSS. Immunoblot analysis showed that the amounts of a 60-kDa immunoreactive band had increased after cross-linkage in a concentration-dependent manner, whereas the intensity of the 30-kDa CLN6 band decreased (Fig. 3). These results indicate that CLN6 partially exists as a dimer in vivo.
Defective CLN6 Mice Do Not Impair Sorting and Processing of Newly Synthesized Cathepsin D-To examine whether CLN6 is critical for the transport and processing of Man-6-P-containing lysosomal proteins, the secretion and proteolytic modifications of newly synthesized CtsD was determined in fibroblasts from CLN6 patients, OCL6 sheep, nclf mice, and the corresponding control cells. In human fibroblasts, CtsD is synthesized as a 53-kDa precursor form (28). A small fraction of the CtsD precursor (in this study Ն3% of total) is secreted (Fig. 4A). The majority of the precursor protein binds to MPRs in the trans-Golgi network and is transported to prelysosomal/endosomal compartments where it is processed to the 47-kDa intermediate form within 2 h after synthesis (42). The intermediate CtsD forms are delivered to lysosomes and are processed to the mature 31-and 14-kDa forms within 6 h after synthesis (Fig.  4A). When South Hampshire sheep fibroblasts were labeled for 1 h with [ 35 S]methionine, a 45-kDa polypeptide was immunoprecipitated with the anti-human CtsD antibody (Fig. 4B). After a chase period of 6 h, only 13 and 16% of the total CtsD were recovered as a 30-kDa mature form from cells and as a 48-kDa form from the media, respectively (Fig. 4B). After metabolic labeling of primary mouse fibroblasts followed by immunoprecipitation with the anti-mouse CtsD antibody, a predominant 45-kDa polypeptide was observed (32). Small amounts (9% of total synthesized CtsD) were detected as a 51-48-kDa band in the media (Fig. 4C), suggesting a variable extent of glycosylation. These results show that the sorting and processing of newly synthesized CtsD was not affected in fibroblasts from a CLN6 patient, from OCL6 sheep, or from nclf mice. Additionally, the specific activities of two other lysosomal enzymes, ␤-hexosaminidase and ASA, were comparable in control and human, sheep, and mouse CLN6-defective cells.
Accumulation of Endocytosed Arylsulfatase A in CLN6-defective Fibroblasts-To examine whether defective CLN6 affect the endocytic pathway to the lysosome, fibroblasts from CLN6 patients, OCL6 sheep, nclf mice, and the respective controls were incubated with [ 125 I]ASA for 16 h. Because the MPR46 does not function in endocytosis of lysosomal enzymes (43), the uptake of ASA is mediated by the MPR300. The ASA uptake is specific as shown by the complete inhibition in the presence of an excess of Man-6-P (Fig. 5). The densitometric evaluation of autoradiographs of several experiments (n ϭ 3-4) showed that the average amount of [ 125 I]ASA detectable in human CLN6, OCL6, and nclf cells is 2-, 51-, and 2-fold higher, respectively, than in control cells. This effect is specific for the MPR300mediated uptake because the amounts of internalized [ 125 I]transferrin are comparable in control and CLN6-defective cells (Fig. 5B).
The after endocytosis by cell surface biotinylation followed by immunoprecipitation of MPR300 and Western blot analysis (Fig.  6). Even in these cells, the amounts of MPR300 at the cell surface of control and OCL6 fibroblasts were similar, indicating that the increased amounts of intracellular [ 125 I]ASA in OCL6 cells is not the result of increased concentrations of MPR300 at the cell surface. The specificity of the procedure was shown by the effect of MESNA, which cleaved the biotin label linkage.
To examine whether the degradation of endocytosed [ 125 I]ASA was inhibited in CLN6-defective fibroblasts, cells were incubated with [ 125 I]ASA for 3 h at 37°C. After removal of the radioactive media, the cells were either harvested or chased for a further 20 h. The solubilized cells were analyzed by SDS-PAGE and autoradiography (Fig. 7). Densitometry of the autoradiograph revealed that the amount of intracellular [ 125 I]ASA detected in cells from CLN6 patients was 1.5-fold more than in control cells after 3 h of endocytosis. During the following chase period of 20 h, the intracellular [ 125 I]ASA remaining in control cells was reduced to 32%, whereas 96% remained in fibroblasts from CLN6 patients. These data indicate that the degradation of endocytosed [ 125 I]ASA is strongly inhibited in CLN6-defective cells.

DISCUSSION
In the present study, we have examined the subcellular localization of the CLN6 protein. Mutations in the CLN6 gene cause the accumulation of storage material in lysosomes and the clinical manifestation of the variant late infantile neuronal ceroid lipofuscinosis. Here we report that CLN6 is an ERresident protein capable of forming dimers. Furthermore, neither the synthesis, the sorting, nor the transport of newly synthesized CtsD, a lysosomal protease, was impaired in CLN6-defective cells, whereas the degradation of an exogenous protein, ASA was reduced.
Several lines of evidence imply the translocation and localization of the CLN6 protein to the ER. First, the CLN6 antigenic sites were found to colocalize with PDI, an established ER marker, but not with the cis-Golgi marker GM130 or with lysosomes stained for CtsD. Secondly, the CLN6 protein does not undergo proteolytic processing steps. Thirdly, although it contains no potential N-glycosylation sites, the replacement of the residue isoleucine 153 by serine located in the second proposed luminal loop (amino acid residues 137-179) resulted in the addition of an oligosaccharide chain that could be removed by PNGase F treatment. The mutant CLN6 was also completely sensitive to endo H removing the high mannose ER-type oligosaccharide chains but not complex Golgi-type oligosaccharides. These data strongly support the concept of CLN6 residency in the ER.
ER residency of membrane proteins is achieved by retention and/or retrieval mechanisms requiring signal structures and targeting motifs within the proteins. Thus, ER-resident, type I transmembrane proteins contain a dileucine motif in the Cterminal cytoplasmic domain, whereas an arginine-containing signal structure functions as targeting signal in some type II membrane proteins (41,44,45). Recent studies have revealed additional ER retention signals such as RXR (R ϭ arginine, X ϭ a large neutral or positively charged residue) (46 -48). Furthermore, the oligomerization of proteins mediated by transmembrane and/or luminal domains forming large homodimeric or heteromeric protein complexes might prevent package into transport vesicles and export from the ER (44,49,50). The signal structures responsible for the retention of the polytopic CLN6 in the ER are not known. Both the N-and the C-terminal domain harbor potential RRR and RKK retention motifs. However, the replacement of these motifs by lysine and leucine residues, respectively, did not alter the ER localization of the mutant CLN6. 2 It is also possible that the dimerization reported in this study or the assembly of CLN6 in oligomeric complexes prevents its export from the ER.
The function of CLN6 has not been defined. However, the comparable rates of synthesis, sorting, and proteolytic processing of CtsD in CLN6-defective and control cells as well as similar specific activities of two lysosomal enzymes, ␤-hexosaminidase and ASA, suggest that CLN6 does not limit transport of at least three enzymes to the lysosome. On the other hand, this study shows that the capability to degrade exogenous arylsulfatase A is strikingly impaired in CLN6-defective cells. Whereas the half-life of the newly synthesized ASA has been reported to be Ͼ40 days (51), the internalized human ASA is rapidly degraded (Ref. 52 and the present study). It is probable that the acidic pH-dependent octamerization cannot be completed by endocytosed ASA resulting in increased lysosomal instability (53). Taking the endocytosed ASA as an index for the delayed degradation in CLN6-defective cells, the results indicate that either specific lysosomal proteases are absent and/or inactive or other lysosomal parameters such as pH or the presence of cofactors are affected. It is of interest that the incubation of N1E.115 neuroblastoma cells with amyloid-␤ protein fragments, the suspected neurotoxic mediator in Alzheimer's disease, led to an increased expression of acetylcholinesterase because of slowed lysosomal degradation. The data suggest that the lysosomal acidification is impaired (54).
Lysosomal dysfunction attributed to a defective ER-resident membrane protein is not without precedent. CLN8 is another polytopic ER membrane protein defective in progressive epilepsy with mental retardation belonging to the group of NCL disorders (16). It remains to be investigated whether the reduced capability for lysosomal degradation of proteins demonstrated in this study is specific for CLN6 or common in both CLN6-and CLN8-defective cells.