(Received for publication, November 27, 1996, and in revised form, February 21, 1997)
From the Departments of Dermatology and ¶ Dental Oral Biology,
Northwestern University Medical School, Chicago, Illinois 60611 and
the Division of Dermatology, Palo Alto Veterans Affair
Health System, Stanford University School of Medicine,
Stanford, California 94305
Type VII collagen, the major component of anchoring fibrils, consists of a central collagenous triple-helical domain flanked by two noncollagenous domains, NC1 and NC2. The NC1 domain contains multiple submodules with homology to known adhesive molecules including fibronectin type III-like repeats and the A domain of von Willebrand factor. In this study, we produced the entire NC1 domain of human type VII collagen in the stably transfected human kidney 293 cell clones and purified large quantities of the recombinant NC1 protein from serum-free culture media. The recombinant NC1 formed interchain disulfide-bonded dimers and trimers and was N-linked glycosylated. Tunicamycin inhibited the cellular secretion of NC1, suggesting that N-linked glycosylation may play a role in NC1 secretion. The recombinant NC1 was indistinguishable from the authentic NC1 obtained from human amnions or WISH cells with respect to N-linked sugar content, electrophoretic mobility, rotary shadow imaging, and binding affinity to type IV collagen. Purified recombinant NC1, like authentic NC1, also bound specifically to fibronectin, collagen type I, and a laminin 5/6 complex. Both monomeric and trimeric forms of NC1 exhibited equal affinity for these extracellular matrix components, suggesting that the individual arms of NC1 can function independently. The multiple interactions of NC1 with other extracellular matrix components may support epidermal-dermal adhesion.
Type VII collagen, a genetically distinct member of the collagen family, resides within the basement membrane zone (BMZ)1 beneath stratified squamous epithelium (1, 2). Type VII collagen is a major component of anchoring fibrils, attachment structures within the basement membrane between the epidermis and dermis of human skin (3, 4). In inherited forms of dystrophic epidermolysis bullosa (DEB), anchoring fibrils are diminutive and/or reduced in number (5-7). Epidermolysis bullosa acquisita (EBA), an acquired autoimmune form of epidermolysis bullosa, is characterized by circulating and tissue-bound IgG autoantibodies against type VII collagen (8, 9). Ultrastructural studies have demonstrated a dramatic paucity of anchoring fibrils within the dermal-epidermal junction of patients with EBA (10). These observations suggest that type VII collagen plays an important role in maintaining epidermal-dermal adherence. Type VII collagen has been cloned, and a genetic linkage has been established between the inherited forms of DEB and type VII collagen (11-14).
Type VII collagen is composed of three identical chains, each
consisting of a 145-kDa central collagenous triple-helical segment
characterized by repeating Gly-X-Y amino acid sequences, flanked by a large 145-kDa amino-terminal, noncollagenous domain (NC1),
and a smaller 34-kDa carboxyl-terminal noncollagenous domain (NC2) (4,
15). In the extracellular space, the individual type VII collagen
molecules form antiparallel tail-to-tail dimers with a small
carboxyl-terminal overlap, and a portion of the NC2 domain is
proteolytically removed (16, 17). The antiparallel dimers then
aggregate laterally in a nonstaggered manner to form anchoring fibrils
that contain large globular NC1 domains pointing outwards (4). Sequence
analysis of the NC1 domain revealed the presence of multiple submodules
with homology to adhesive proteins (12, 18). These submodules may play
an important role in the attachment of the NC1 domain to other BMZ
structures within the lamina densa and sublamina densa zone. The mature
NC1 polypeptide contains 1,235 amino acids and is characterized by the
presence of a segment with 40% homology to cartilage matrix protein,
nine consecutive fibronectin type III-like repeats (FNIII), a segment
with homology to the A domain of von Willebrand factor (VWF-A), a
potential cell attachment site RGD motif, and a segment rich in
cysteine and proline at the carboxyl-terminal end of NC1. The NC1
domain also contains three consensus sequences, NX(S/T), for
N-linked glycosylation. Other extracellular matrix (ECM)
proteins have domains with homology to the FNIII and VWF-A domains.
Studies with these ECM proteins have demonstrated that these motifs are capable of binding to different types of collagen and may mediate significant protein-protein interactions (19-21).
The NC1 domain of type VII collagen has been implicated in a variety of cellular functions. Solid phase binding studies suggest that it interacts with type IV collagen, a major component of the lamina densa and anchoring plaques (22). It has been shown that autoantibodies in the serum of EBA patients recognize epitopes within the NC1 domain of type VII collagen, particularly within the FNIII and VWF-A submodules (23, 24). Therefore, these regions may play a direct role in adhesion of type VII collagen to other BMZ and matrix proteins. The exact function of the NC1 domain, however, is not known. The lack of knowledge about NC1 is at least partly due to the difficulty in obtaining sufficient quantity of purified NC1 by biochemical means, because type VII collagen represents less than 0.001% of the total collagen extracted from human skin (25). Furthermore, a major problem in studying the biological functions of type VII collagen has been the necessity of using denaturing or proteolytic conditions to release type VII collagen from supramolecular complexes with other extracellular macromolecules during purification procedures. In the current study, we used an efficient eukaryotic expression system to obtain large quantities of recombinant NC1 in conditioned medium from human kidney 293 cells. The recombinant NC1 was found to resemble authentic NC1 with regard to structural and functional properties. Our functional studies on the binding of NC1 to several ECM proteins, including type IV collagen, fibronectin, a complex of laminin 5 and laminin 6 (laminin 5/6), and type I collagen provide a molecular basis for the role of NC1 in epidermal-dermal adherence of human skin and in anchoring fibril assembly. This study also sets the stage for producing and characterizing mutants of NC1 for further dissection of the various structural and functional domains.
The fragment
corresponding to the 5 end of human type VII collagen cDNA was
generated by reverse transcriptase-PCR amplification using human
amniotic epithelial cells (WISH) cDNA as template and
oligonucleotides MC1: GGATCCCTCGGACCTGCCAAGGCCC AC: and L3: GAATTCAGGTCGGGTCACAGGCACGC as primers based on the published sequences (12). The cycling conditions were 94 °C for 7 min, followed by 40 cycles of 95 °C, 45 s; 57 °C, 1 min; 72 °C, 2 min; and a 7-min extension at 72 °C. The PCR product was subcloned into a TA
vector (Invitrogen, San Diego, CA), and DNA sequence analysis was
performed and compared with the published sequence to confirm its
identity (12). The NC1 domain of type VII collagen cDNA (3.8 kb),
which encodes the entire open reading frame including the signal
peptide at the 5
end as well as a PCR-generated in frame stop codon at
the 3
end, was prepared from the 5
end fragment and three cDNA
clones previously described (23). The construct utilized the
overlapping internal restriction sites including ApaLI,
XhoI, and AvrII from each cDNA fragment. The
subsequent constructs were made in M13 mp18, TA vector and pGEX vector
(Pharmacia Biotech Inc.), as modified by Dr. George Giudice, Medical
College of Wisconsin (23). The correct ligation and in frame insertion of various DNA fragments were confirmed by DNA sequence analysis. The
entire NC1 domain construct was ligated into the
HindIII/XbaI-restricted eukaryotic expression
vector pRC/CMV (Invitrogen, San Diego, CA) which was designated as
CMV/NC1 as shown in Fig. 1.
Transfection of Cells, Northern Hybridization
The human embryonic kidney cell line 293 (ATCC, Rockville, MD) was routinely cultured in Dulbecco's modified essential medium/Ham's F12 (1:1) supplemented with 10% fetal bovine serum. The CMV/NC1 construct was transfected into 293 cells using Lipofectin (1:1 (w/w) liposome formulation of the cationic lipid N-[1-(2,3-dioleoyloxy)propyl]-N,N,N-trimethylammonium chloride, and dioleoyl phosphatidylethanolamine (Life Technologies, Inc.)). Cells were seeded in 100-mm dishes at 106 cells per dish and allowed to recover overnight, then transfected using a ratio of 70 µl of Lipofectin to 25 µg of DNA with incubation for 6 h in serum-free medium. The medium was changed to one containing serum, and the cells were cultured for an additional 48 h. The transfected cells were then removed by trypsinization and plated on 150-mm dishes and cultured in the presence of 500 µg of G418/ml medium for 2-3 weeks. Individual clones resistant to G418 were isolated with a cloning cylinder, removed by trypsinization, and grown in G418-containing medium.
For Northern blot analysis, total RNA was extracted using guanidine isothiocyanate/CsCl2 density gradient centrifugation, as described previously (26). The total RNA was size-fractionated and transferred to nylon membranes (GeneScreen Plus, DuPont NEN) and hybridized with a 32P-labeled (random-primer labeling kit, BMB, Indianapolis, IN) 2.4-kb human type VII collagen cDNA probe containing FNIII submodules within the NC1 domain.
Protein Purification and AnalysisFor immunoblot analysis, clonal cell lines resistant to G418 were grown to confluence, the medium was changed to serum-free medium, and the cultures were maintained for an additional 24 h. The media were collected, concentrated 10-fold by Centricon-10 (Amicon, Beverly, MA), and subjected to 6% SDS-PAGE. Proteins were then electrotransferred onto a nitrocellulose membrane as described (27). The presence of recombinant NC1 was detected with a mouse monoclonal antibody, LH 7.2 (1:200) (Sigma), against the NC1 domain of type VII collagen followed by a horseradish peroxidase-conjugated goat anti-mouse IgG (1:400) and enhanced chemiluminescence detection reagent (Amersham Corp., UK).
For large scale purification of recombinant NC1, serum-free media were equilibrated to 5 mM EDTA, 50 µM phenylmethylsulfonyl fluoride, and 50 µM N-ethylmaleimide, and precipitated with 300 mg/ml ammonium sulfate at 4 °C overnight with stirring (4). Precipitated proteins were collected by centrifuging at 13,000 rpm for 1 h, resuspended and dialyzed in Buffer A containing 65 mM NaCl, 25 mM Tris-HCl, pH 7.8, and 1 mM EDTA, and passed over a Q-Sepharose column (Pharmacia) equilibrated in the same buffer. Proteins that did not bind to the column were further purified on a Superdex 200 (HR 10/30. Pharmacia) equilibrated in 25 mM Tris-HCl, pH 7.8, 1 mM EDTA, and 200 mM NaCl.
Preparation of Authentic NC1 Domain of Type VII CollagenProteins secreted into the medium of WISH cells were precipitated with ammonium sulfate as described (4). After centrifugation, the pellet was resuspended in 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 10 mM CaCl2, and 5 mM N-ethylmaleimide and dialyzed overnight against the same buffer. Next, 50 units of bacterial collagenase (clostridiopeptidase A type III, Sigma) were added. The enzymatic digestion was allowed to proceed overnight at 37 °C. The digestion product was subjected to immunoblot analysis as described above or deglycosylation as described below.
Purification of the authentic NC1 domain of type VII collagen from collagenase digests of human amnion was accomplished as described previously (28) using affinity chromatography with coupled monoclonal antibodies NP185 and NP32. Purified authentic NC1 was used for ECM binding studies as described below.
Tunicamycin TreatmentConfluent 293 cells were cultured at 37 °C for 24 h with 1 µg/ml tunicamycin diluted from a 500 µg/ml tunicamycin stock solution in 0.01 N NaOH (Sigma) in serum containing medium. Cells were then incubated for an additional 24 h in serum-free medium. The media were collected, concentrated 10-fold by Centricon-10, and subjected to 6% SDS-PAGE and immunoblot analysis.
Deglycosylation of NC1Purified recombinant NC1 (100 ng) or authentic NC1 was denatured by heating at 100 °C in the presence of 1% SDS for 5 min and treated with 0.5 unit of peptide N-glycosidase F (Sigma) in 50 mM sodium phosphate, pH 7.4, 10 mM EDTA, and 1% Nonidet P-40 for 15 h at 37 °C and then analyzed by 6% SDS-PAGE and immunoblot analysis.
Protein Binding AssayBinding of soluble NC1 to immobilized ligands followed by a colorimetric enzyme-linked antibody reaction was performed as described previously (29). Multiwell plates (96 wells, Dynatch Laboratory Inc., Alexandria, VA) were coated overnight with ECM proteins (4 µg) in 100 mM carbonate buffer, pH 9.3. The wells were then blocked with 1% bovine serum albumin in phosphate-buffered saline, 0.05% Tween 20 (PBST). Coated wells were subsequently incubated with purified recombinant or authentic NC1 at concentrations between 2.5 to 40 µg/ml overnight at 4 °C. The binding of NC1 to ECM proteins was detected with a mixture of monoclonal anti-type VII collagen antibodies clone 185 (Life Technologies, Inc.) and LH 7.2 (Sigma) at a dilution of 1:300 followed by incubation with alkaline phosphatase-conjugated goat anti-mouse IgG (1:400) (Organon Teknika-Cappel, Durham, NC). The development of a colorimetric reaction using p-nitrophenyl phosphate as a substrate (Bio-Rad) was measured by reading the absorbance of the product at 405 nm (Labsystems Multiskan Multisoft, Finland).
The ligands used for binding included collagen types I, IV, and V (Collaborative Biomedical Product, Bedford, MA) and pepsinized type IV collagen (Life Technologies, Inc.). Laminin 1 was prepared from the EHS tumor as described (29). Fibronectin was prepared from human plasma as described previously (30) or purchased from Life Technologies, Inc. Purified fragments corresponding to the 7 S and NC1 domains of type IV collagen were the generous gift of Dr. Rupert Timpl, Max Planck Institute, Martinsried, Germany. Laminin 5/6 was purified from collagenase-solubilized human amniotic membranes by antibody affinity chromatography using mAb K140-Sepharose as described (31).
Electron microscopy of proteins visualized by rotary shadowing followed
an established protocol (32). Thirty µl of recombinant NC1 protein
dissolved in 0.2 M ammonium bicarbonate at a concentration of 1 mg/ml were mixed with 3 ml of glycerol buffer (30% 0.2 M ammonium bicarbonate and 70% glycerol) to provide a
final concentration of 10 µg/ml protein. One hundred µl of solution
were then placed onto a freshly separated mica sheet and sprayed using
the sandwich technique. After drying, the samples were rotary shadowed
in a Blazers 400 at an angle of 7° with 0.26 nm of platinum added to the surface. The samples were then coated with carbon at an angle of
90° for 13 s at 100 mA. The resulting carbon layer was then
floated off into a dish filled with double distilled H2O and picked up with 300 mesh copper grids. Samples were then examined in
a JEOL 1200 EX transmission electron microscope at 60 kV.
The entire cDNA sequence coding for the NC1 domain of type VII collagen, including its own signal peptide sequence, was cloned into the eukaryotic expression vector pRC/CMV as shown in Fig. 1 and used to transfect human 293 cells. A total of 18 individual G418-resistant cell clones were isolated and screened for secretion of NC1 in their conditioned media by immunoblot analysis with a monoclonal antibody directed against NC1. Production of the 145-kDa recombinant protein was detected in eight clonal cell lines (data not shown).
Northern hybridization of the NC1 stably transfected 293 cell clones
demonstrated abundant amounts of the 4-kb exogenous mRNA which is
the size expected from the cDNA construct (Fig. 2).
No hybridization, even after long exposure, was detected in the
parental 293 cell, indicating very little or no production of
endogenous type VII collagen in these cells.
Purification and Characterization of Structural Properties of Recombinant NC1
The recombinant NC1 was purified from serum-free
culture medium that contained 15-20 µg/ml NC1. In serum-free culture
medium, the recombinant product could be easily identified as a major 145-kDa band that was lacking in the parental 293 cells (Fig. 3A, compare lane 1 and
2). This band accounted for 20-30% of total secreted
proteins. The ammonium sulfate precipitate of the conditioned medium
proteins was then passed over a Q-Sepharose column. Most of the NC1
protein came out in the unbound fraction with a purity of approximately
90% (Fig. 3A, lane 3). NC1 was further purified to >95% electrophoretic homogeneity by molecular sieve Superdex-200 chromatography (Fig. 3A, lane 4). The purity of
NC1 was demonstrated by the appearance of a single band on a
silver-stained SDS-PAGE slab gel (lane 5). The final
purified yield was 5-10 mg/liter culture medium.
The purified recombinant NC1 and authentic NC1 prepared from the collagenase digestion of WISH cell-derived type VII collagen showed identical molecular mass by immunoblot analysis (Fig. 3B). A similar result was obtained when recombinant NC1 was compared with authentic NC1 purified from human amnions (data not shown). When the purified recombinant NC1 was subjected to immunoblot analysis under nonreducing conditions, it appeared mainly as monomers of 145-kDa, dimers of about 290-kDa, and trimers of about 450-kDa (Fig. 3C, lane 2). Under reducing conditions, however, all the dimers and trimers converted into monomers (Fig. 3C, lane 1). Quantitation of these species of NC1 using densitometry analysis showed approximately 60% as monomers, 30% as dimers, and 10% as trimers. The purified recombinant NC1 was also subjected to amino acid composition analysis and shown to be within the limits of analytical error of that predicted from the cDNA sequence (data not shown).
Rotary shadowing of the recombinant NC1 revealed abundant amounts of
what appeared as single arm NC1 species (monomers) (Fig. 4A). Substantial amounts of NC1 also appeared
to be configured as two arms (dimers) (Fig. 4B). Finally,
a minority of the NC1 was configured into a structure whose image
closely resembled the globular amino termini of type VII collagen
consisting of three identical chains held by disulfide bonds into
trident-like structures (trimers) (Fig. 4C). It appears that
the individual arms of NC1 are extended rod-like shaped structures
under physiological conditions. Similar species of single-, two-, and
three-armed structures have also been previously reported from rotary
shadowing analysis of NC1 partially purified from human amnion
(28).
Demonstration of N-Linked Glycosylation of the NC1 Domain of Type VII Collagen
Sequence analysis of the NC1 domain revealed three
potential sites for N-linked glycosylation (12). To examine
whether or not NC1 is indeed N-linked glycosylated,
confluent NC1 stably transfected 293 cells were subjected to
tunicamycin treatment. As shown in Fig. 5A, a
lower molecular weight NC1 species was produced in the cells treated
with tunicamycin in comparison with the untreated cells, indicating
that the recombinant NC1 is N-linked glycosylated.
Interestingly, as shown in Fig. 5B, lanes 1 and 2, tunicamycin treatment also significantly inhibited the
secretion of the recombinant NC1 into the medium. Accordingly, larger
amounts of NC1 accumulated inside the cells, as shown in Fig. 5B,
lanes 3 and 4. Fig. 5A is a Coomassie
Blue-stained slab gel of the proteins used in the subsequent Western
blots (Fig. 5B) and demonstrates that the presence of
tunicamycin in the 293 cell cultures (lane 2) selectively
inhibits glycosylation and secretion of the 145-kDa NC1 band, whereas
the other secreted proteins are not affected (compare lanes
1 and 2). This result suggests that total protein synthesis and secretion are not affected in these cells by the presence
of tunicamycin. The tunicamycin-treated 293 cells were morphologically
identical to the untreated 293 cells, and the cellular viability, as
assessed by trypan blue dye exclusion, was identical with or without
the presence of tunicamycin (data not shown).
Deglycosylation of the purified recombinant NC1 with peptide N-glycosidase F converted it to the lower molecular weight species, confirming that NC1 contains N-linked sugar chains (Fig. 5C, lanes 1 and 2). Furthermore, when authentic NC1 purified from WISH cells (Fig. 5C, lanes 3 and 4) was subjected to the same enzyme treatment and analysis, the size difference between the deglycosylated and glycosylated forms of authentic NC1 was identical to recombinant NC1. These results suggest that authentic NC1 and recombinant NC1 have similar amounts of N-linked sugar content.
Interaction of NC1 with ECM ProteinsSequence analysis of the
NC1 revealed the presence of multiple submodules with homology to
adhesive proteins, suggesting that the NC1 domain may mediate
interactions with lamina densa and dermal ECM proteins (12, 18).
Previous solid phase interaction studies indicated that the NC1 domain
of type VII collagen interacts with type IV collagen (22). Therefore,
we examined the ability of purified recombinant NC1 to bind to
immobilized ECM components including type IV collagen (before and after
pepsinization), the NC1 domain of type IV collagen, the 7 S domain of
type IV collagen, type I collagen, type V collagen, laminin-1, the
laminin 5/6 complex, and fibronectin using an enzyme-linked
immunoabsorbent assay (29). To compare functionality between authentic
and recombinant NC1, NC1 purified from amnion was also included in the
binding assay. As shown in Fig. 6, NC1 binds to native
type IV collagen and the NC1 domain of type IV collagen but does not
bind to pepsinized type IV collagen or the 7 S domain of type IV
collagen. These results were consistent with that of a previous study
which suggested that the noncollagenous domain of type IV collagen is
required for this interaction (22). Moreover, we found that both
recombinant NC1 and authentic NC1 also had affinity for fibronectin,
the laminin 5/6 complex, and type I collagen. In contrast, little or no
binding occurred with laminin-1 or type V collagen. No significant
difference between the authentic NC1 and recombinant NC1 in terms of
binding affinity for various ECM was noted. Furthermore, as shown in
Fig. 7, the binding of the NC1 to collagens types I and
IV, the laminin 5/6 complex, and fibronectin was specific,
dose-dependent, and saturable.
Previously, rotary shadowing of the native NC1 showed trident-like
structures in which each arm of the structure corresponded to a single
extended chain (28). These data suggest that individual arms of NC1
may independently interact with ECM proteins. To examine whether
individual NC1 arms could exhibit ECM binding activity, recombinant NC1
was fractionated further by Superdex-200 chromatography to isolate
fractions of pure monomers and trimers (Fig.
8A). Using smaller fraction volumes, further
fractionation of peak I showed that the 450-kDa trimers came out first,
followed by a mixture of trimers and dimers (290-kDa), and the last
fractions of peak I contained only dimers. Fig. 8B shows a
representative immunoblot of these fractions from peaks I and II under
reducing (+) or nonreducing (
) conditions. Peak I yielded two
electrophoretic protein bands of approximately 450- and 290-kDa under
nonreducing conditions which could be separated from each other. They
all converted into monomers under reducing conditions. Peak II
contained primarily monomeric NC1 species. When the monomeric and
trimeric NC1 species were compared for binding affinity to ECM
components, as shown in Fig. 9, the binding affinities
to fibronectin, the laminin 5/6 complex, and collagen types I and IV
were comparable. Because the monomeric and trimeric species of NC1 bind
ECM components equally, this indicates that each arm of NC1 may
function independently.
The efficient production and secretion of the NC1 domain of human type VII collagen was achieved by using an eukaryotic expression vector in a human cell line which showed no production of endogenous type VII collagen. The recombinant NC1 was secreted in large quantities into the culture medium (15-20 mg/liter), which allowed purification of the protein in milligram quantities. The purified recombinant NC1 was identical or similar to authentic NC1 purified from WISH cells or human amnion with regard to molecular mass (145-kDa), immunochemical properties, N-linked glycosylation, and binding to fibronectin, the laminin 5/6 complex, and collagen types I and IV.
Fig. 3 shows that recombinant NC1 is capable of forming
disulfide-bonded trimers under nonreducing conditions. This finding is
consistent with the observation in a previous biochemical study using
partially purified NC1 which suggested that each arm of NC1 is
contributed by a single chain peptide held by interchain disulfide
bonds into a trident-like shape (28). In the previous study, however, a
significant amount of the partially purified NC1 appeared to exist as a
trimeric species, whereas in the study reported here, less than 10% of
purified recombinant NC1 was in the form of trimers. It is possible
that in the case of tissue or cell culture extraction, the folding of
homotrimeric forms of type VII collagen
chains with intact helical
domains and the formation of stable interchain bonds between NC1
domains have already occurred in situ. Therefore, the
pre-existing trimeric NC1 species are likely to remain intact following
the extraction and purification procedure. In contrast, in the case of
eukaryotic synthesis of recombinant NC1, triple helical domains are
absent. All trimer formation must occur de novo and without
the assistance of adjacent helical domains. It is likely that the
assembly of NC1 chains into stable, disulfide-bonded trimers is
dependent to some degree on the existence of adjacent triple helical
domains. Studies with other collagens are consistent with this concept. For example, it has been shown previously that interchain, disulfide bonding of the amino-terminal propeptides for type III collagen and the
carboxyl-terminal propeptides for type XII collagen require the
formation of a stable triple helix (33, 34). Furthermore, with regard
to basement membrane type IV collagen, it has been proposed that
lateral aggregation via the type IV collagen triple helical domain is
essential for proper interchain disulfide linkage between the weakly
interacting carboxyl-terminal NC1 domains (35).
The calculated size of NC1 from the cDNA sequence is 133,802 Da, which is smaller than the size of 145-kDa estimated from its migration by SDS-PAGE using authentic or recombinant NC1 (2). The tunicamycin inhibition and N-glycosidase F treatments demonstrated that the NC1 expressed in the 293 cells is N-glycosylated. The NC1 isolated from WISH cells also contains the N-linked sugar chains. In concordance with these data, we previously demonstrated by metabisulfite-fuchsin staining of SDS-PAGE slab gels that NC1 extracted from the BMZ of human skin contained carbohydrate (36). Since type VII collagen cDNA reveals three putative N-linked glycosylation sites that are all localized within the NC1 domain, we believe that the N-linked glycosylation sites within NC1 account for all of the carbohydrate content of type VII collagen. The fact that there is no size difference between the deglycosylated and glycosylated forms of NC1 derived from WISH cells or stably transfected 293 cells suggests that a similar degree of glycosylation occurs within both authentic and recombinant NC1. Whether authentic and recombinant NC1 are glycosylated at exactly the same sites is not yet known.
With regard to the potential function of NC1 glycosylation, the tunicamycin inhibition of core N-linked oligosaccharide addition decreases the cellular secretion of unglycosylated NC1. Tunicamycin has no effect on the overall secretion of other protein bands as shown in Fig. 5A, suggesting that glycosylation may play an important role in promoting secretion of NC1 from the cells into the extracellular space. Numerous studies have shown that N-linked oligosaccharide attachment is required for proper folding and oligomerization of many secreted and cell surface glycoproteins (37). Failure to achieve a native folded structure can severely retard intracellular transport and lead to degradation of the retained proteins (38). In this study, when we blocked NC1 glycosylation with tunicamycin, we found higher levels of accumulated unglycosylated NC1 inside the cells. It is likely that the effect of tunicamycin is mainly on NC1 secretion rather than on its stability. In accordance with these data, it was reported that tunicamycin retards cellular secretion of procollagen from human fibroblasts (39).
The NC1 domains of type VII collagen are believed to interact with
components of the basement membrane and, therefore, to mediate the
attachment of the basement membrane to the dermis. Burgeson et
al. (22) showed that NC1 isolated from human amnion bound to type
IV collagen but that the binding was abolished by pepsin digestion of
type IV collagen. As shown here in this study, our recombinant NC1
behaves in the same fashion. However, our current study differs from
the report by Burgeson (22) in that our recombinant NC1 binds to type I
collagen. As shown in Fig. 6, both recombinant NC1 and authentic NC1
exhibited similar binding affinities to type I collagen. This different
result is quite revealing. In the previous study by Burgeson (22) where
type I collagen binding was not detected, the antibody NP185 was used to detect NC1. This antibody specifically binds near the VWF-A homology
region (40). The VWF-A motif has been shown to mediate interactions
between type I collagen and other matrix proteins, such as the 3
chain of type VI collagen and cartilage matrix protein (19-21). In
contrast, in this study, we used a mixture of LH 7.2 and NP185
antibodies. It is possible that the type I collagen-NC1 interaction may
have masked the NC1-NP185 interaction epitope. The type I collagen-NC1
binding could only be detected with the LH 7.2 antibody but not with
the NP185 antibody (data not shown). This observation supports the
possibility that the type I collagen binding site on NC1 co-localizes
with the NP185 antibody recognition site.
Another potential function of the NC1 domain of type VII collagen is binding to anchoring filaments within the skin BMZ. Laminin 5 (or nicein/kalinin) is a component of anchoring filaments within the lower lamina lucida (41). Anchoring filaments are thread-like structures that bridge the hemidesmosomes to the lamina densa of the dermal-epidermal junction (42, 43). Previous studies suggested that anchoring fibrils below the lamina densa associated with hemidesmosomes in basal keratinocyte through putative interactions between the NC1 domain of type VII collagen and components of anchoring filaments (44, 45). As shown in Fig. 6, both the recombinant NC1 and authentic NC1 are capable of binding to the laminin 5/6 complex. Since our studies used a complex of laminin 5 and laminin 6, it still remains to be established whether the NC1 binding affinity detected was contributed by laminin 5, laminin 6, or both.
Both recombinant NC1 and authentic NC1 specifically bound to
fibronectin. We have previously mapped the fibronectin-binding site
within the type VII collagen chain to a unique site within the
triple helical collagenous domain immediately flanking the NC2 domain
(46). In that study, the recombinant protein was produced from
Escherichia coli as a glutathione
S-transferase-fusion protein and was not
post-translationally modified. In the present work, the recombinant or
authentic NC1 was produced from eukaryotic cells allowing
N-linked glycosylation, disulfide bond formation, and proper
conformational folding. Thus, it is possible that the type VII
collagen-fibronectin interaction may involve multiple sites, some of
which require post-translational modifications (such as the NC1 site)
and some of which can occur under denaturing conditions, such as the
site within the collagenous helical domain.
We also performed affinity experiments using monomeric or trimeric forms of recombinant NC1. Our results indicate that monomeric NC1 is comparable in potency to trimeric NC1 in binding to fibronectin, the laminin 5/6 complex, and type I and IV collagens. Both previous and present rotary shadowing studies of authentic and recombinant NC1 demonstrate that individual arms are most often independent and extended and do not interact with each other to form a single globular domain but rather form trident structures. Biochemical and biophysical studies indicate that the individual arms of NC1 have an intrinsic rod-like character in solution, strongly supporting the argument that each arm maintains an independent conformation (28). Since the arms of the NC1 domains are identical, they must be, therefore, individually capable of interacting with BMZ and ECM components, potentially allowing trivalent and multivalent interactions of a single type VII collagen molecule with several macromolecules.
The biological implications of the interactions between type VII collagen (NC1 domain) and fibronectin, and type I collagen are unknown. One possibility is that these interactions are important for the adherence of the cutaneous basement membrane to the papillary dermis. Fibronectin is known to self-aggregate, and this property, along with its affinity for type VII collagen (47), may play a role in the assembly of type VII collagen molecules into antiparallel dimers and/or cross-aggregation of these dimers into anchoring fibrils. In the context of collagen type I binding, anchoring fibrils have been proposed to be capable of entrapping other components of the dermal connective tissue, such as the broad, banded collagen fibrils consisting primarily of types I and III collagens (44, 48). The binding of type VII collagen to type I collagen via NC1 may play a role in securing the cutaneous BMZ to the dermis by physical entrapment of connective tissue elements.
The expression system used here allows for the production of native, post-translationally modified NC1 in quantities sufficient for comprehensive functional and structural studies. The specific binding of NC1 to the complex of laminin 5/6, fibronectin, and type I and type IV collagens suggests that the NC1 domain could cross-link to proteins within the lamina densa, anchoring plaques, and anchoring filaments. These interactions would contribute to the integrity of the interactions between the lamina lucida, the lamina densa, and the papillary dermis. We are now in position to produce various site-directed and deletional mutants to help elucidate the precise role of VWF-A and FNIII within the NC1 domain in the context of interactions with other BMZ and ECM proteins. These studies may facilitate the interpretation of the molecular defects within NC1 that characterize DEB and correlate genetic mutations with clinical phenotypes.
We are grateful for the expert technical assistance of Tom Dahl for the rotary shadowing analysis. We thank Dr. Wei Li, Ben May Cancer Institute, University of Chicago for reading the manuscript.