The Localization of Bullous Pemphigoid Antigen 180 (BP180) in Hemidesmosomes Is Mediated by Its Cytoplasmic Domain and Seems to be Regulated by the β4 Integrin Subunit

The 804G cell line and the African monkey kidney cell line COS-7 were cultured in DMEM (Gibco BRL, Paisley, UK) supplemented with 10% (vol/vol) bovine FCS, 100 U/ml penicillin, and 100 U/ml streptomycin. The 804G cell line derived from a rat bladder tumor has been described previously (40, 41). The cells were grown at 37°C in a humidified, 5% CO₂ atmosphere.

The following antibodies were used. The mouse IgG1 mAb antiFLAG™M2 against the FLAG™ peptide (DYKDDDDK) was purchased (IBI, Eastman Kodak Company, New Haven, CT). The rabbit antiserum J17 directed against the intracellular portion of BP180 was kindly provided by Dr. J.C.R. Jones (Cell and Molecular Biology, Northwestern University Medical School, Chicago, IL) (18, 19). The mouse IgG1 mAb 1A8c directed against the intracellular portion of BP180, and the mouse mAb 121 directed against HD1 were kindly donated by Dr. K. Owaribe (Department of Molecular Biology, Nagoya University, Nagoya, Japan) (14, 38). The mouse mAb 450-10D against the cytoplasmic domain of β4 was kindly supplied by Dr. S.J. Kennel (Oak Ridge National Laboratory, Oak Ridge, TN) (22). The rat mAb GoH3 against an extracellular epitope of human α6 (47), a rabbit antiserum against the cytoplasmic domain of β4 (34), a rabbit antiserum against the COOH-terminal portion of α6A (5) as well as a rabbit antiserum to rat IgG (46) have been described previously. The mAb against vinculin, clone VIN-11-5, was purchased (Sigma Chem. Co., St. Louis, MO). A rabbit antiserum against the COOH-terminal domain of BP230, was kindly provided by Dr. J.R. Stanley (Department of Dermatology, University of Pennsylvania, Philadelphia, PA) (52). Species specific FITC-conjugated goat anti–mouse IgG (Nordic Immunochemicals Laboratory, Tilburg, The Netherlands) and Texas red–conjugated donkey anti–rabbit IgG (Amersham International plc, Buckinghamshire, UK) were purchased.

The BP180 nucleotide and protein sequences are numbered according to Giudice et al. (10) and Hopkinson et al. (19), respectively. Full-length BP180 was obtained from human keratinocyte RNA by reverse transcriptase-PCR with primers based on published sequence of human BP180 (GenBank accession number M91669) (10) using the Riboclone cDNA Synthesis Kit (Promega, Madison, WI), the RNA-PCR Kit (Perkin Elmer, Roche Molecular Systems, Inc., Branchburg, NY), and the PCR Reagent System (GIBCO-BRL). The PCR reaction consisted of 35 cycles of 15 s at 95°C, 30 s at 55°C and 90 s at 72°C. PCR products corresponding to overlapping portions of BP180 were first cloned into pBluescript II SK+ (Stratagene, La Jolla, CA) and ligated together to obtain full-length BP180 using distinct restriction sites: PflmI, StuI, SacII, ClaI and EcoRI (at nucleotide positions 566, 1307, 1694, 2548 and 3619, respectively). A PCR site-directed mutagenesis system was subsequently used to prepare the various expression plasmids (Fig. 1 A). To generate clone A, B, C, and D, a 5′ end primer was used that contained an EcoRI and a NotI site for cloning, sequences for optimal initiation of translation (bold) (24), sequences coding for the FLAG™ peptide (underlined) (IBI, Eastman Kodak Company) for immunodetection, as well as nucleotides corresponding to sequences between 109 and 131 of human BP180 (italic) located immediately downstream from the predicted methionine start site of BP180 (nucleotide 106-109) (18) (5′-GCCGGAATTCGCGGCCGCCGCCATGGACTACAAGGACGACGATGACAAGGATGTAACCAAGAAA - AACAAACG 3′-). The 3′ primer of clone A contained a NotI and a XbaI restriction site and the nucleotides corresponding to sequences between 4579 and 4599 of BP180 (italic) including the stop codon (5′-CGAT GCGGCCGCTCTAGATCACGGCTTGACAGCAATACT 3′-). The 3′ end primer of clone B contained a NotI and a XbaI restriction site, a stop codon, and nucleotides corresponding to sequences 1641 and 1662 of BP 180 (italic) (5′-CGATGCGGCCGC TCTAGATTATATTCTATCCATGCTGTCCCCA 3′-) The 3′ primer used to generate clone C contained a NotI and a XbaI site restriction site, a stop codon, and nucleotides corresponding to sequences between 1480 and 1503 of BP180(italic) (5′-CGATGCGGCCGCTCTAGATTACTTCCACCAGCTGCAGCAGGAGCC 3′-). The 3′ primer used for clone D was designed in order to generate a chimeric protein composed of the cytoplasmic domain of BP180 fused to the membrane localization sequences of K-Ras (KMSKDGKKKKKKSKTKCVIM) (GenBank accession number M54968 and M38506). This primer contained a NotI and a XbaI restriction site, a stop codon, nucleotides encoding the membrane localization sequences of the K-Ras (bold) as well as the sequence between 1479-1503 of BP180 (italic) (5′-CGATGCGGCCGCTCTAGATTACATAATTACACACTTTGTCTTTGACTTCTTTT - TCTTCTTTTTACCATCTTTGCTCATCTTCTTCCACCAGCTGCAGC - AGGAGCC 3′-). The clones E, F, and G, which encoded chimeric proteins with increasing internal truncations of the cytoplasmic portion of BP180 starting from amino acid 244, 201 and 113, respectively, to 402, were generated by partial digestion of clone D with Ecl136 II (nucleotide sites 440, 701, and 830) and complete digestion with StuI (nucleotide 1307). The construct H lacking the sequences encoding amino acids 1 to 36 was generated using a 5′ primer that contained a XbaI site, sequences for optimal initiation of translation (bold) and for the FLAG™ tag (underlined), as well as nucleotides corresponding to sequences between 214 and 230 of BP180 (5′ GCCGTCTAGACGCCATGGACTACAAGGACGACGATGACAAGAGCAATGGCTATGCTAAAACAGC 3′-). Combined alanine substitutions at positions Ser¹⁶⁹, Ser¹⁷⁵, and Ser¹⁸⁰ were introduced in clone B using the overlap extension method for site directed mutagenesis as previously described (5).

The correctness of all clones was verified by sequence analysis, the various constructs were subsequently cloned using the XbaI and/or a NotI restriction site in the eukaryotic expression vector pCI-neo (Promega, Madison, WI). This vector carries a CMV enhancer/promotor for strong constitutive expression, an intron located upstream from the multiple cloning site as well as a SV40 origin for episomal replication in cell lines expressing the SV large T antigen, such as COS-7 cells. Transfection studies have indicated that an intron flanking the cDNA of interest may significantly increase the gene expression level (2, 20).

For generation of the wild-type α6A cDNA, two overlapping cDNA clones, A33 and A84, isolated from a λgt11 human keratinocyte library were used as previously described (5). The full-length α6A was inserted into the HindIII site of the pRc/CMV expression vector (Invitrogen, San Diego, CA) (Fig. 1 B) (5).

For generation of full-length β4A cDNA, a cDNA clone isolated from a λgt11 human keratinocyte library was used that encodes β4A from position 1880 to the 3′ terminus. The λ DNA was digested with SfiI and the resulting 2.2-kb fragment was exchanged for the 2.35-kb SfiI fragment of β4B (35, 37). The cDNA was then ligated into the XbaI site of the pRc/ CMV vector (Invitrogen). The cDNA-construct encoding a truncated β4 molecule, clone β4¹³⁸², was obtained by exchanging a PCR fragment with a wild-type fragment of β4 and will be described in detail elsewhere. After sequence analysis, the β4¹³⁸² cDNA construct was subcloned into pcDNA3 (Invitrogen) (Fig. 1 B). The cDNA plasmid encoding a β4 with combined phenylalanine substitutions of the tyrosine activation motif (30) was kindly provided by Dr. F.G. Giancotti (Department of Pathology, New York University School of Medicine, New York, NY). The construct was assembled into pcDNA3. Construction of the β4/β1 cDNA encoding a chimeric protein, in which the cytoplasmic domain of β4, with the exception of a 19–amino acid stretch immediately close to the transmembrane domain, was replaced with that of β1, has been recently described. The β4/β1 cDNA was inserted into the XbaI site of the pcDNA-1Hyg expression vector (43).

The following cDNA constructs were used as a control for transfection. A full-length murine BP180 cDNA (29), provided by Dr. J. Uitto (Department of Dermatology, Thomas Jefferson University, Philadelphia, PA), was subcloned into the EcoRI site of pcDNA3. A full-length CD31 cDNA, provided by Dr. B. Seed (Massachusetts General Hospital, Boston, MA), was subcloned into the HindIII/NotI sites of pRc/CMV. A fulllength CD8 cDNA, provided by Dr. A. Kelly, Guy's and St. Thomas's Medical and Dental School, London, UK), was subcloned into the HindIII and BamH1 site of pcDNA3.

For transfection, cells were first grown to 50–80% confluency in six-well tissue culture plates (Falcon, Becton Dickinson, Lincoln Park, NJ). The rat bladder carcinoma cell line 804G was transfected using a cationic lipid, Lipofectin^® (1:1-N-[1-(2,3-dioleyloxy) propyl]-n,n,n-trimethylammonium chloride and dioleoyl phosphatidylethanolamine (GIBCO-BRL). The DNA:Lipofectin^® mixture was prepared using serum free medium (OPTIMEM^®, GIBCO-BRL). The final concentration of plasmid DNA and Lipofectin^® in serum free transfection medium was 5 μg/ml and 20 μg/ml, respectively. 1 ml of transfection medium was added to each monolayer that had been previously washed with serum free medium and cells were incubated with the transfection medium for 12–18 h at 37°C with 5% CO₂. The transfection medium was then replaced with normal growth medium and cells were incubated for additional 24–36 h and then assayed for gene expression. COS-7 cells were transfected using the DEAE-dextran method (4) and assayed for gene expression after 36 h.

Cells were lysed with 1% SDS in 25 mM Tris-HCl, pH 7.5, 4 mM EDTA, 100 mM NaCl, 1 mM PMSF, 10 μg/ml leupeptin and 10 μg/ml soybean trypsin inhibitor. Protein concentration in the cell lysates was determined with the BCA protein assay reagent (Pierce, Rockford, IL). Equal amounts of protein were loaded on a 8.5% or 13.5% SDS polyacrylamide gel, separated, and electrophoretically transferred to nitrocellulose sheets for 1 h at 240 mA in 25 mM Tris-HCl, pH 8.3, 192 mM glycine, 20% (vol/vol) methanol and 0.0375% SDS as previously described (53). The filters were incubated in TBST (10 mM Tris-HCl, pH 7.6, and 150 mM Nacl, 0.1% Tween20) containing 2% (wt/vol) BSA and 2% (wt/vol) baby milk powder for 12 h at 4°C. Subsequently the filters were probed with the primary antibody for 90 min, after which they were washed four times for 5 min in TBST. The filters were subsequently incubated with horseradish peroxidase– linked sheep anti–mouse IgG or horseradish peroxidase–linked goat anti– rabbit IgG (Amersham International plc), diluted to 1:5,000 and 1:2,000, respectively, for 1 h and then washed with TBST. Proteins were visualized using enhanced chemiluminescence (Amersham International plc).

Cells grown on glass coverslips in six-well tissue culture plates were fixed with 1% formaldehyde for 10 min and permeabilized with 0.5% Triton X-100 for 5 min. After rinsing with PBS and blocking with 2% (wt/vol) BSA in PBS for 30 min, the cells were incubated with primary antibody for 30 min at 37°C, then washed three times with PBS. The cells were stained with fluorescein- or Texas red–labeled anti–mouse IgG, anti–rat IgG or rabbit IgG for 30 min at 37°C. The coverslips were subsequently washed, mounted with Vectashield (Vector, Burlingame, CA), and viewed under an MRC-600 confocal scanning laser microscope (Bio-Rad Laboratories, Richmond, CA).

Transfected COS-7 cells were washed twice with PBS and incubated with DMEM without methionine and cysteine (ICN Biomedicals Inc., Costa Mesa, CA) for 1 h at 37°C. Cells were then labeled with 100 μCi/ml [³⁵S]methionine/cysteine (Amersham International plc) for 4 h, then washed and lysed with 1% NP-40 in 25 mM Tris-HCl, pH 7.5, 4 mM EDTA, 100 mM NaCl, 1 mM PMSF, 10 μg/ml leupeptin, and 10 μg/ml soybean trypsin inhibitor. Cells were also lysed with 1% Triton X-100, 2 mM CaCl₂, 1 mM PMSF, 10 μg/ml leupeptin, and 10 μg/ml soybean trypsin inhibitor or 10 mM CHAPS in HBSS. The lysates were clarified by centrifugation at 10,000 g and precleared with protein A–Sepharose CL-4B (Pharmacia LKB Biotechnology Inc., Uppsala, Sweden). Samples of precleared lysates were immunoprecipitated with antibodies previously bound to protein A–Sepharose or to protein A–Sepharose to which rabbit anti– mouse IgG was attached. The immunoprecipitates were subsequently analyzed by SDS-PAGE.

The role of specific sequences of BP180 in the localization of the protein in HD was investigated by introducing a series of cDNA-constructs into the epithelial cell line 804G. The 804G cells, which have previously been used to study the assembly and formation of HD, contain the structural components of HD, i.e., proteins that may interact with BP180, including the α6β4 integrin, BP230, and HD1 (40, 41, 48). As illustrated in Fig. 1,A, we have generated cDNAconstructs encoding full-length (clone A) or two truncated forms of BP180, that consist of either the entire cytoplasmic domain, the transmembrane region and a segment of 30 amino acids located adjacent to the transmembrane region (clone B) or of the cytoplasmic domain alone (clone C). Clone B was used because a previous study had indicated that the protein encoded by this construct contains the minimal sequence of the ectodomain of BP180 essential for incorporation into HD (18). To further investigate the contribution of distinct cytoplasmic domains in the recruitment of BP180 to HD, we generated cDNA clones encoding chimeric proteins, in which the membrane localization sequence of K-Ras (28, 39) had been linked to various parts of the cytoplasmic region of BP180 (clones D to H). To identify proteins encoded by the transfected plasmids a 24-bp sequence encoding the FLAG™ peptide was added at the 5′ end of the cDNAs. In addition to DNA sequencing of the various constructs, the proteins encoded by the clones were analyzed by in vitro transcription/translation followed by SDS-PAGE and found to be of the predicted mol wt (not shown). Finally, to assess whether the transiently transfected 804G cells expressed the appropriately sized recombinant BP180 molecules, we performed immunoblot analysis of cell extracts (Fig. 2) using mAbs directed against the FLAG™ peptide as well as against the cytoplasmic domain of BP180. In transfected cells, the apparent mol wt of the recombinant proteins were close to the sizes predicted on the basis of the corresponding cDNA sequence, i.e., 151.6 kD, 55.7 kD, 49.4 kD, 51.7 kD, 35.3 kD, 30.9 kD, 21.4 kD, and 47.9 for clones A, B, C, D, E, F, G, and H, respectively. The size of the protein encoded by clone A, corresponding to full-length human BP180, was slightly smaller than that of endogenous rat BP180, possibly due to insufficient posttranslational processing of this protein and/or a species difference. In extracts from COS-7 cells transfected with cDNAs for either human or mouse full-length BP180, the mol wt of the expressed proteins were similar (not shown).

As shown by confocal laser immunofluorescence microscopy, the hemidesmosomal proteins of 804G cells are concentrated at sites of cell-substrate contact in structures appearing as dots and patches arranged in a characteristic “Swiss cheese” pattern (40, 41, 48). To investigate the distribution of the recombinant BP180 molecules, transiently transfected cells were subjected to double immunofluorescence microscopy. The mAb anti-FLAG™M2 directed against the FLAG™ peptide was used to distinguish the recombinant, tagged proteins from endogenous wild-type BP180. When 804G cells were transfected with clones A and B, mutant proteins were found at the basal side of the cells, colocalized with α6 (Fig. 3) and β4, and with BP230 (not shown) producing the typical Swiss cheese–like staining pattern. However, the clone C encoded protein, which lacks the transmembrane and extracellular domains, was not only localized in HD-like structures, but was also diffusely distributed in the cytoplasm (Fig. 3, E and F). The localization of this mutant protein was more obvious in cells with low transgene expression levels, as estimated by fluorescent staining intensities. These results indicate that the full-length (clone A) as well as the two truncated forms of BP180 (clones B and C), that lack either the whole collagenous extracellular domain or both the transmembrane and extracellular domain, are normally incorporated into HD of 804G cells. The peculiar localization of clone C encoded protein in both HD and the cytoplasm suggests that not only the cytoplasmic domain but also the transmembrane region and/or portions of the extracellular domain are required for efficient recruitment of the protein into HD; or that the amount of the recombinant protein is too large to be completely incorporated into HD. The second possibility is supported by the observation that in cells with high levels of the mutant proteins encoded by clone A or B, these proteins were not found uniquely in HD, but also localized diffusely over the cell surface (not shown).

To further define the function of distinct cytoplasmic domains, cDNA-constructs encoding the plasma membrane targeting sequence of K-Ras fused to various cytoplasmic parts of BP180, were expressed in 804G cells (clones D to H, Fig. 4). The K-Ras membrane localization signal, which consists of a CAAX motif (in which A means aliphatic -, and X refers to any amino acid) and a lysine-rich polybasic domain, targets proteins with high efficiency to the inner surface of the plasma membrane (28, 39). We assumed that because of the presence of the the K-ras sequence, the truncated BP180 proteins would become localized at the plasma membrane where interactions with hemidesmosomal proteins take place. Transient expression of clone D, which contained the sequence for the entire cytoplasmic domain of BP180, resulted in the incorporation of the chimeric protein in HD-like structures, where it was codistributed with endogenous α6 (Fig. 4, A and B) and β4, and with BP230 (not shown). Computer generated z-sections demonstrated that the mutant protein was concentrated at the basal cell surface (Fig. 4). Recombinant proteins encoded by clones E and F with increasing internal deletions of 159 and 202 amino acids in the cytoplasmic domain of BP180, showed also the ability to be correctly localized in HD (Fig. 4, C and E). However, in many cells these mutants were also distributed at the apicolateral cell membrane, suggesting that the amount of the recombinant proteins was too large to be completely incorporated into HD or, alternatively, that their incorporation into HDs was partially defective. These recombinant proteins lacked the degenerate set of four 24-26 residue tandem repeats (10). In contrast, the protein encoded by clone G, in which, compared to clone F, there is an additional deletion of 88 amino acids towards the NH₂ terminus, was never found to be codistributed with α6, but remained diffusely distributed at the cell surface (Fig. 4, G and H). Finally, the same diffuse cell membrane localization was observed with the chimeric protein encoded by clone H lacking the first 36 amino acids at the NH₂ terminus (Fig. 4, I and J). This short segment has previously been reported to be required for the localization of a truncated BP180 protein at the ventral surface of 804G cells (18). These results show that the chimeric protein consisting of a 265–amino acid stretch of the cytoplasmic tail of BP180 (clone F) contains sequences sufficient for the localization of the protein in HD. Furthermore, deletion of specific segments located at the NH₂ terminus (clone H) and within the central portion of the cytoplasmic domain of BP180 (clone G) completely prevents the recruitment of mutant forms of BP180 into HD.

A recent study has suggested that phosphorylation at serine residues of BP180 may regulate its localization into HD (23). Analysis of clone F encoded protein, which contains the minimal sequences required for the targeting of the protein to HD, reveals the presence of several potential phosphorylation sites (Thr²⁸, Ser⁴¹, Thr¹⁵⁷, Ser¹⁶⁹, Ser¹⁸⁰, and Thr¹⁹¹) as well as a recognition sequence for p34^cdc2 kinase (Ser¹⁷⁵) (19). Notably, the deletion of three serine residues (Ser¹⁶⁹, Ser¹⁷⁵, and Ser¹⁸⁰) located in the central portion of the cytoplasmic domain of the BP180 recombinant protein encoded by clone G, prevents its incorporation into HD. To examine the role of these serine residues, we mutated the recombinant protein encoded by clone D by substituting the three Ser¹⁶⁹, Ser¹⁷⁵, and Ser¹⁸⁰ by alanines. As shown in Fig. 5, by immunofluorescence, this mutated BP180 molecule was distributed in a Swiss cheese–like pattern, which indicates that this recombinant protein was correctly localized in HD. Thus, phosphorylation of the above Ser residues does not seem to be required for the localization of BP180 in HD.

Previous studies have shown that the α6β4 integrin plays an essential role in the assembly of HD (8, 26, 30, 31, 55). We next investigated the effect of the α6β4 integrin on the subcellular distribution of wild-type or mutant forms of BP180 by expressing cDNAs for these different components in COS-7 cells. COS-7 cells express HD1, the α3β1 integrin, and a small amount of α6β4, but lack BP230, BP180, and α6β4 (37) (see Fig. 9,A). In COS-7 cells, vinculin (Fig. 6,B) and β1 are localized in punctate and streaky arrays at sites of cell-substrate contact which represent focal adhesions, while a similar localization of α6 was observed in only a few cells. In contrast, HD1 is found throughout the cytoplasm in a dense and delicate cytoskeletal network, similar to that previously described for plectin (Fig. 6 D) (37, 57).

In COS-7 cells transfected with clone A encoding wildtype BP180, the recombinant molecule remained diffusely distributed in the cytoplasm, where it was localized in granules and tubular structures, particularly prominent in perinuclear regions (not shown). Because the recombinant protein was synthesized but not expressed on the cell surface, as shown by immunoprecipitation (Fig. 9) and FACS analysis (not shown), the full-length BP180 was probably retained in the ER and/or Golgi apparatus. This was most likely due to incorrect folding and/or impaired trimer formation of this collagenous protein (16). In contrast, the mutant proteins encoded by clones B or D were properly expressed at the plasma membrane. Both proteins were distributed predominantly at the apicolateral cell surface. By double immunofluorescence microscopy, the localization of these mutant forms of BP180 was clearly different from that of vinculin (Fig. 6, A and B), β1, and HD1 (Fig. 6, C and D).

Transfection of COS-7 cells with cDNAs for α6A and β4A resulted in the expression of the α6β4 integrin at sites of cell-substrate contact and, concomitantly, in the redistribution of HD1 from the cytoskeleton into junctional complexes containing α6β4 (37) (Fig. 7, B and D, and Fig. 8,B). When COS-7 cells were transfected with cDNAs for α6A and a β4/β1 chimera encoding the extracellular domain of β4 fused to the cytoplasmic domain of β1, α6 was associated with β4/β1 as assessed by immunoprecipitation (not shown) and was found together with vinculin in focal contacts (see Fig. 8 F) (43). In these cells, HD1 remained diffusely distributed in a pattern similar to that observed in untransfected COS-7 cells.

When clone B or D cDNA was cotransfected with α6A and β4A cDNAs, the mutant BP180 proteins were concentrated at the basal cell surface and were codistributed with the α6β4 integrin (Figs. 7 and 8, A and B) and HD1 (Fig. 7, C and D), although in some cells staining of the apicolateral cell surface was still observed, in particular of the clone D encoded protein (Fig. 7, A and C). In contrast, after identical transfections with the β4/β1 chimera instead of β4, the two mutant BP180 proteins were not codistributed with α6β4/β1 in focal adhesions, but remained diffusely distributed at the cell surface (Fig. 8, E and F). These results show that the BP180 mutants encoded by clones B and D contain sequences necessary and sufficient for the codistribution of BP180 with α6β4 and HD1, and that the β4 cytoplasmic domain is involved in determining the subcellular distribution of the BP180 recombinants and HD1. The colocalization of BP180, α6β4, and HD1 along the basal aspect of the transfected cells is unlikely to be coincidental. First, BP180 recombinants were not concentrated at the basal cell side in the absence of α6β4. Second, when cells were cotransfected with cDNAs for either CD8 or CD31, these two transmembrane proteins were distributed diffusely at the cell membrane and did not codistribute with α6β4 (not shown).

Finally, COS-7 cells were also cotransfected with clone G or H with more extensive deletions: in this case, the staining of the plasma membrane was too intense, due to the high level of expression, to assess the colocalization with α6β4.

A recent study has provided evidence for an interaction between BP180 and α6 (18). To determine the ability of mutant forms of BP180 to associate directly with α6 and β4, we performed immunoprecipitation experiments using extracts from radiolabeled COS-7 cells that were transiently transfected with either of the clones A, B, or D together with cDNAs encoding α6A and β4. The SDS-PAGE analysis of these immunoprecipitates is shown in Fig. 9,A. In all cases, the mol wt of the detected proteins corresponded with the sizes predicted on the basis of the cDNA sequence. Neither α6 nor β4 were precipitated in association with BP180 recombinant polypeptides encoded by clones A, B, or D by the mAb anti-FLAG™M2. Furthermore, the recombinant forms of BP180 encoded by these clones could not be coimmunoprecipitated with either α6 or β4 by an anti-α6 or anti-β4 antiserum, respectively. Identical results were obtained when COS-7 cells were transfected with clones A, B, or D together with cDNA for either α6A or β4 (Fig. 9 B). When the samples immunoprecipitated with the anti-α6 or anti-β4 antiserum were subjected to immunoblotting with the mAb anti-FLAG™M2, this mAb did not show any reactivity with the various precipitates. Moreover, no coprecipitation of recombinant forms of BP180 with either α6A or β4A was observed when cells were lysed with different detergents, such as Triton X-100 or CHAPS (not shown). In contrast, immunoprecipitation with either a rabbit anti-α6 or a rabbit anti-β4 antiserum of lysates from cells transfected with cDNAs for the α6A and β4 yielded, as expected, a heterodimer complex consisting of α6 and β4. Together, these data do not provide evidence for a physical association between mutant forms of BP180 and the α6β4 integrin in transfected COS-7 cells.

As an alternative approach to identify domains involved in the association of BP180 and α6β4, we generated a cDNA construct encoding a β4 molecule, β4¹³⁸², with a truncated cytoplasmic domain, which contains the first pair of FNIII and a segment of 64 amino acids at the NH₂ terminus of the connecting segment, but which lacks the tyrosine activation motif (TAM) (30) of the connecting segment as well as the third and fourth FNIII. In COS-7 cells transfected with cDNAs encoding α6A and this truncated β4, the β4 mutant protein was distributed together with α6 and HD1 along the basal cell surface in a pattern indistinguishable from that observed in cells transfected with wild-type α6A and β4A, as assessed by double immunofluorescence staining (Fig. 7,F). Strikingly, in COS-7 cells expressing α6 and β4¹³⁸² the mutant forms of BP180 encoded by clone B (not shown) and D were not colocalized with α6β4 and HD1 in these junctional complexes, but in most cells remained diffusely distributed at the apicolateral cell surface (Fig. 7, E and F). Since recent studies (30, 31) have demonstrated that phosphorylation of the TAM of β4 is required for the incorporation of α6β4 into HD and for their assembly, we then investigated if the TAM of β4 plays a role in the subcellular distribution of BP180 by using a β4 subunit carrying a mutated TAM. In COS-7 cells transfected with cDNAs for α6A and a β4 subunit with combined phenylalanine substitutions at the TAM, the colocalization potential of the clone B encoded protein with the mutated β4 was impaired as compared to that of wildtype β4. In most cells expressing α6 and the mutated β4, clone B encoded protein was diffusely distributed over the plasma membrane (Fig. 8, C and D), and in only a few transfected cells it was concentrated at the basal cell surface together with α6β4 (not shown). Notably, in cells expressing α6 and β4 with a mutated TAM, HD1 was found to be codistributed with α6β4 in a pattern indistinguishable from that observed with α6 associated with wild-type β4 (not shown). Taken together, these results indicate, first, that sequences contained in the COOH-terminal half of the cytoplasmic tail of β4 are involved in determining the subcellular distribution of BP180; second, that mutations of the β4 TAM interfere with the efficient colocalization of BP180 with α6β4; finally, that colocalization of HD1 with α6β4 is, in contrast to that of BP180, mediated by sequences within the NH₂-terminal half of the cytoplasmic domain of β4.

We have studied the contribution of various segments of BP180 to its incorporation into HD by transfecting 804G cells with cDNA-constructs encoding wild-type, deletion mutant forms of BP180, and chimeric proteins, in which the membrane targeting sequence of K-Ras had been fused to segments of the cytoplasmic domain of BP180. The results indicate that the cytoplasmic domain of BP180 contains sequences sufficient for the incorporation of mutant forms of BP180 into HD-like structures in 804G cells. The localization of BP180 in HD appears to be mediated by a region that spans 265 amino acids (clone F) comprising two important domains, that are located at the NH₂ terminus and within the central portion of the BP180 cytoplasmic domain.

Based on transient transfections of 804G cells, it was recently reported that a 36–amino acid stretch located at the NH₂ terminus determines the localization of BP180 at the basal side of the cell, whereas the extracellular segment close to the membrane-spanning domain is required for its recruitment into HD (18). Our results confirm the functional importance of the NH₂-terminal domain of BP180. However, our observation that a chimeric protein composed of the membrane targeting signal of K-Ras and a cytoplasmic tail of BP180 lacking this NH₂-terminal region (clone H) is not only localized at the apical, but also often at the basolateral cell surface, indicates that this segment is not uniquely involved in sorting the protein to the basal cell surface, but rather is required for targeting BP180 to HD. The membrane targeting signal of K-Ras may be responsible for the diffuse localization of the protein (28, 39), but this possibility seems less likely, because recombinant molecules encoded by clones B or D, that lack or contain the K-Ras membrane localization sequence, respectively, were frequently similarly distributed. Alternatively, it is conceivable that the localization of the mutant protein is dynamic and dependent on its relative expression level in transfectants.

The results of the subcellular localization of chimeric proteins with deletions of increasing size within the BP180 cytoplasmic domain indicate the existence of an additional, as yet unrecognized, functionally important region. Whereas a truncated BP180 molecule consisting of a 265– amino acid stretch was localized in HD (clone F), a recombinant protein with an additional internal deletion of 88 amino acids towards the NH₂ terminus (clone G) was not.

What are the mechanisms by which the truncations of the cytoplasmic domain of BP180 interfere with its incorporation in HD? First, the deleted regions may contain the sequences responsible for the interaction of BP180 with elements of the hemidesmosomal cytoskeleton. Second, conformational changes in protein architecture may be responsible for the failure of clone G and clone F encoded proteins to be incorporated into HD. PHD secondary structural analysis (42) predicts that the protein encoded by clone F, containing the minimal sequences for recruitment into HD, has several β-sheet regions as well as one α–helical domain, which are absent in the recombinant protein encoded by clone G. Moreover, the functionally important domain at the NH₂ terminus of BP180 contains a β-sheet. The above two regions at the NH₂ terminus and within the central portion of the cytoplasmic domain may thus be involved in the proper folding of BP180, which may be essential for its assembly in HD. Finally, the truncations may interfere with the transmission of an intracellular signal required to render BP180 competent to interact with other components of HD. Phosphorylation pathways appear to critically regulate the assembly of HD (9). Inhibition of the phosphorylation of the β4 integrin subunit was found to prevent localization of the α6β4 integrin in HD (9). In addition, a recent study suggests that a protein kinase C is involved in the phosphorylation of BP180 at serine residues, which affects its localization at the basal cell membrane of squamous carcinoma cells (23). Although combined alanine substitutions at three serine residues located in the central portion of the cytoplasmic domain of BP180 did not affect its localization in HD, it is conceivable that the truncations in clone G and H encoded proteins G might interfere with signaling pathways that control the association of BP180 with HD.

Our results demonstrate that removal of the extracellular and transmembrane regions of BP180 or their substitution by the membrane localization sequence of K-Ras does not abolish the localization of BP180 in HD of 804G cells. This was unexpected, because in a previous study, deletion of a segment of 27 amino acids located immediately after the transmembrane domain prevented the incorporation of the BP180 protein into HD, suggesting that this region is required for targeting (18). However, in this latter study (18) it has not been ascertained whether the truncated protein was correctly inserted in the plasma membrane. BP180 is a type II transmembrane protein, whose insertion in the membrane depends on both the “signal-anchor” sequence of the transmembrane domain and the number and type of flanking charged amino acids (15). Therefore, deletion of the region proximal to the transmembrane domain may have a deleterious effect on the translocation and insertion of the truncated protein into the plasma membrane, preventing its interaction with the hemidesmosomal cytoskeleton. Since our transfection studies only assess the ability of mutant proteins to be incorporated, together with endogenous components, into HD (48), our findings do not exclude the possibility that the ectodomain of BP180 is also involved in the localization of the protein in HD as well as in their nucleation and formation by binding to an extracellular ligand(s).

To analyze the potential of BP180 to interact with α6β4, we have performed transfection experiments using COS-7 cells. In COS-7 cells, transfected with cDNAs for the α6A and β4 integrin subunits, expression of the α6β4 integrin results in the generation of distinct structures at sites of cell-substrate contact (37). These junctional complexes, containing α6β4 and HD1, may correspond to the recently described type II HD, which are found in some cell types and which lack the two hemidesmosomal components BP230 and BP180 present in HD (54). When COS-7 cells were cotransfected with clones B or D, the recombinant BP180 molecules were present in the same complexes as α6β4. These findings confirm the previously reported colocalization of a mutant form of BP180 (identical to that encoded by clone B) with α6β4 in FG cells which, like COS-7 cells, lack endogenous BP180 and BP230 (18). However, the observation that codistribution also occurred with the mutant form of BP180 lacking the entire extracellular region (clone D), suggests that the cytoplasmic domain of BP180 contains sufficient elements for its recruitment into complexes containing α6β4 and HD1.

Our mutagenesis experiments indicate that the COOHterminal half of the β4 subunit contains sequences required for the colocalization of mutant forms of BP180 with α6β4 in transfected COS-7 cells. This region of β4 comprises the third and fourth FNIII as well as part of the connecting segment including the TAM, that consists of two tyrosine residues at position 1422 and 1440 of the β4 subunit (30). Recent studies have demonstrated that phosphorylation of this TAM is required for the incorporation of α6β4 into HD as well as their assembly (30). Combined phenylalanine substitutions of the TAM impaired the capacity of BP180 to codistribute with α6β4, suggesting that the TAM is involved in transducing signals that critically regulate the subcellular distribution of BP180 in cos-7 cells. However, the observation that in some cells the expression of a mutated β4 TAM did not completely suppress this codistribution suggests that additional factors also contribute. Since type III fibronectin homology modules have been implicated in protein–protein interactions (3, 48), it is conceivable that the third and fourth FNIII within the COOH-terminal half of the cytoplasmic domain of β4 also coordinate the localization of recombinant BP180 molecules with the α6β4 integrin by providing a binding site for interaction. However, coimmunoprecipitation experiments failed to provide evidence for a direct association between mutant BP180 proteins and α6β4, but a weak association between these proteins which is lost after cell lysis cannot be ruled out.

The results of our immunofluorescence analysis indicate that neither the elimination of the COOH-terminal half of β4 nor the mutation of the β4 TAM prevent the subcellular redistribution of HD1 with α6β4, which occurs after transfection of COS-7 cells with cDNAs for α6 and the mutated β4 molecules. The colocalization of HD1 with α6β4 may thus be mediated by distinct sequences in the NH₂-terminal half of the cytoplasmic domain of β4. Recent in vitro binding and transfection studies using β4 recombinant proteins have provided evidence that HD1 associates with the cytoplasmic tail of β4 (37). Based on our findings in COS-7 cells, it is tempting to speculate that α6β4 and HD1 form a complex that serves as a core for the assembly of HD to which BP180 and BP230 aggregate in a later step, possibly regulated by TAM-dependent signals. Clearly, further experiments are needed to elucidate the mechanisms controlling the association of the α6β4 integrin with the other hemidesmosomal cytoskeletal molecules.

It is difficult to reconcile some of our findings with a previous report stating that a mutant form of BP180, identical to that encoded by clone B, coprecipitated with α6 from HT1080 cells transfected with this mutant, which contain α6β1 but not α6β4. This interaction appears to rely on the extracellular segment of BP180 close to the transmembrane domain (18). It is possible that the reported association between α6 and BP180 only occurs in some cell types. In fact, in cultured keratinocytes derived from a patient with junctional EB associated with pyloric atresia, who was completely deficient for β4, α6 was associated with β1 and was localized in focal contacts, whereas BP180 was present in HD (36). This latter observation is consistent with our results in transfected COS-7 cells, in which BP180 recombinants were not codistributed with α6, associated with β4/β1, in focal contacts. Alternatively, coprecipitation of α6 with a recombinant form of BP180 may have been due to incomplete solubilization of BP180 and/or aggregation of molecules after cell lysis.

In conclusion, the studies reported here show that the cytoplasmic tail of BP180 contains sequences critical and sufficient for mediating the localization of BP180 in HD. Upon expression of both the α6 and β4 integrin subunits in cells that do not form HD, mutant forms of BP180 are codistributed with α6β4 and HD1 in complexes along the basal cell surface. In these complexes, the localization of recombinant BP180 molecules with α6β4 appears to be regulated by sequences within the COOH-terminal half of the β4 tail, including the third and fourth FNIII and part of the connecting segment containing the TAM. These observations provide new insights relevant for the understanding of the molecular interactions and regulatory elements involved in the organization and assembly of HD.

This work was supported by grants from the Netherlands Organization for Scientific Research (NWO 902-11-060), the Dutch Cancer Society (NKI 91-260), and from the Swiss National Foundation for Scientific Research (L. Borradori).

BP180

bullous pemphigoid antigen 180

BP230

bullous pemphigoid antigen 230

EB

epidermolysis bullosa

FNIII

type III fibronectin repeat

HD

hemidesmosomes

IF

intermediate filament

TAM

tyrosine activation motif