Structural characterization of two variants of fibulin-1 that differ in nidogen affinity

Two C-terminal variants C and D of mouse fibulin-1 were purified from the culture medium of stably transfected human kidney cell clones. They showed, after rotary shadowing, a dumbbell-like structure of about 33 nm in length. Pepsin digestion demonstrated stability of the disulfide-bonded domains 1 (anaphylatoxin-like) and II (multiple EGF-like motifs) but not for domain III which is different in the variants. A close similarity of the variants was observed in immunochemical assays indicating that domain III epitopes are not very antigenic. Binding analysis in solid phase assays demonstrated for variant C a 100-fold stronger binding to the basement membrane protein nidogen than for variant D. Both interactions were sensitive to EDTA. Surface plasmon resonance assays confirmed this difference and showed KD = 60 nM for variant C and KD > 1 microM for variant D. Lower binding activities and smaller differences between both variants were observed for the calcium-dependent binding to fibronectin, laminin-1 and collagen IV. Self aggregation into nest-like oligomers was observed at high concentrations of fibulin-1 which was not sensitive to EDTA.

Recombinant expression and structural and binding properties of alpha 1(VI) and alpha 2(VI) chains of human collagen type VI

Full-length alpha 1(VI) and alpha 2(VI) cDNAs in an eukaryotic expression vector were used to obtain stably transfected human kidney cell clones and to purify these collagen-VI chains in substantial quantities from the culture medium. Both chains appeared mainly as monomers together with some dimers that were disulfide linked through their C-terminal globular domains. Despite sufficient hydroxylation of proline and lysine residues, the chains did not form a triple-helix, as shown by electronmicroscopy, CD spectra and pepsin sensitivity. Digestion of the chains with bacterial collagenase released the N-terminal and C-terminal globular domains, which were identified by their size and partial sequences. They showed a substantial content of alpha-helical conformation and a distinct globular structure after rotary shadowing. Antibodies could be raised that distinguished between the two chains and reacted with the globular domains. The alpha 2(VI) but not the alpha 1(VI) chain showed binding to a heparan sulfate proteoglycan (perlecan), fibronectin and pepsin-solubilized collagen VI. Purified globular domains did not bind these ligands indicating the localization of binding sites within the triple-helical domain. Both chains showed a distinct affinity for heparin but failed to bind to various collagen types.

Protein binding and cell adhesion properties of two laminin isoforms (AmB1eB2e, AmB1sB2e) from human placenta

Two isoforms of laminin were extracted from human placenta by neutral buffer containing EDTA, copurified through several steps and finally separated by Mono Q anion exchange chromatography. One variant consisted of disulphide-linked 340, 230 and 190 kDa subunits, which were identified by immunoblotting as Am, B1e and B2e chains. In the other variant, the B1e chain was replaced by B1s of 180 kDa. After rotary shadowing, both variants showed a similar cross-shaped structure. The nidogen content of these laminins was substoichiometric and variable (3-70%), indicating loss by endogenous proteolysis. Yet both human isoforms were able to bind mouse nidogen with an affinity (Kd approximately 0.5 nM) comparable to that of AeB1eB2e laminin from a mouse tumour. Since the binding site is known to be contributed by a single EGF-like motif of the B2e chain, this demonstrates that activity of this site is independent of chain assembly. Binding activity of both isoforms to collagen IV and the heparan sulphate proteoglycan perlecan was correlated to the nidogen content and could be enhanced by adding nidogen. Binding to heparin was only partial and heparin did not inhibit perlecan binding. This indicated a crucial role for nidogen in mediating the integration of these laminin isoforms into basement membranes. Variant AmB1sB2e showed calcium-dependent binding to fibulin-1, while only a little activity was found for AmB1eB2e. Both isoforms promoted adhesion and spreading of several cell lines. Adhesion could be completely inhibited by antibodies to the integrin beta 1 subunit but not, or only weakly, by antibodies against beta 3, alpha 2, alpha 3, alpha 5 and alpha 6 subunits. No inhibition was observed with an Arg-Gly-Asp-containing peptide.

Similarity in structure between C1q and the collectins as judged by electron microscopy

The collectins are carbohydrate binding proteins which, like C1q, contain collagen-like sequences. The collectins belong to group III of the family of lectins containing C-type carbohydrate recognition domains (CRDs). The structural similarity between the collectins and C1q is clearly demonstrated by electron microscopy in that they all contain multiple polypeptides which are organised into subunits containing triple-helical stalks throughout their collagen-like regions and globular 'heads' in the C-terminal regions. Four, or six, of these structures are associated via distinct, short, N-terminal regions to form the oligomeric molecules seen in the electron microscope. The overall structural similarity between C1q and the collectins, however, does not extend to similarity in amino acid sequences over the C-terminal regions. The C-terminal regions of C1q, unlike those of the collectins, do not contain the conserved residues found in the CRDs present in the C-type lectins. Instead, C1q has a high degree of homology to collagen sequences (Type VIII and X) and this is consistent with the fact that, unlike the collectins, C1q binds to protein motifs in IgG, or IgM, rather than to carbohydrate structures. Also, despite sometimes showing interruptions in their collagen-like regions, the collectins do not always display a 'bend' in their collagen-like 'stalks' similar to that which is seen in C1q. Therefore, C1q may be more closely related to collagens than to the collectins. The collectins can be classed into two distinct group, with MBP and SP-A being hexamers and SP-D, conglutinin and collectin-43 (CL-43) being tetramers, with proteins in the latter group also having significantly larger dimensions with respect to the length of their collagen-like 'stalks'.

Structural similarity between lung surfactant protein D and conglutinin. Two distinct, C-type lectins containing collagen-like sequences

Preparations of bovine lung surfactant D (SP-D) and conglutinin were examined by electron microscopy, gel-filtration and SDS/PAGE. SP-D is composed of non-covalently linked subunits, of 160 kDa, which each contain three, disulphide-linked, 44-kDa polypeptide chains. In the electron microscope a single 160-kDa subunit of SP-D appears as a 45.8 +/- 3-nm-long rod connected to a small globular 'head'. Particles were also seen which correspond to non-covalently linked dimers, trimers and tetramers of the 160-kDa monomer subunit of SP-D. The tetramer structure contains 12 polypeptide chains and is very similar to the electron microscopy images and model reported by Strang et al. [Strang, C. J., Slayter, US., Lachmann, P. J. and Davis, A. E. (1986) Biochem. J. 236, 3811-389] for bovine conglutinin in which four 160-kDa subunits are disulphide-linked to give a molecule of expected molecular mass of 528 kDa. This study confirmed the findings by Strang et al. in the above paper for intact conglutinin and also emphasised that the rod-like structures, of length 37.6 +/- 3.7 nm, seen in the conglutinin subunits were significantly shorter than those in SP-D despite the close similarity in amino acid sequence (79% identify) and chain length between the two proteins. In addition, a truncated form of conglutinin was found in the conglutinin preparations, due to limited proteolysis of the Arg-Ala bond at position 54 in the 44-kDa chains. These truncated conglutinin chains yield a subunit composed of three shortened, non-disulphide-linked, chains and this subunit appears as a monomer with a rod length of 34.2 +/- 2.8 nm in the electron microscope. On gel-filtration, a proportion of the SP-D preparation behaved, as expected, as a molecule with an apparent molecular mass of 600 kDa. The remainder of the SP-D preparation behaved as aggregated material with a molecular mass greater than 900 kDa which yielded no distinct structures in the electron microscope. Intact conglutinin was eluted at a position greater than 900 kDa but yet provided clear electron microscopy images of the tetramer structure described above.(ABSTRACT TRUNCATED AT 400 WORDS)

Structure and binding properties of collagen type XIV isolated from human placenta

Collagen XIV was isolated from neutral salt extracts of human placenta and purified by several chromatographic steps including affinity binding to heparin. The same procedures also led to the purification of a tissue form of fibronectin. Collagen XIV was demonstrated by partial sequence analysis of its Col1 and Col2 domains and by electron microscopy to be a disulphide-linked molecule with a characteristic cross-shape. The individual chains had a size of approximately 210 kD, which was reduced to approximately 180 kD (domain NC3) after treatment with bacterial collagenase. Specific antibodies mainly to NC3 epitopes were obtained by affinity chromatography and used in tissue and cell analyses by immunoblotting and radioimmunoassays. Two sequences from NC3 were identified on fragments obtained after trypsin cleavage. They were identical to cDNA-derived sequences of undulin, a noncollagenous extracellular matrix protein. This suggests that collagen XIV and undulin may be different splice variants from the same gene. Heparin binding was confirmed in ligand assays with a large basement membrane heparan sulphate proteoglycan. This binding could be inhibited by heparin and heparan sulphate but not by chondroitin sulphate. In addition, collagen XIV bound to the triple helical domain of collagen VI. The interactions with heparin sulphate proteoglycan and collagen VI were not shared by the NC3 domain, or by reduced and alkylated collagen XIV. No or only low binding was observed for collagens I-V, pN-collagens I and III, and several noncollagenous matrix proteins, including laminin, recombinant nidogen, BM-40/osteonectin, plasma and tissue fibronectin, vitronectin, and von Willebrand factor. Insignificant activity was also shown in cell attachment assays with nine established cell lines.

Recombinant nidogen consists of three globular domains and mediates binding of laminin to collagen type IV

Recombinant mouse nidogen and two fragments were produced in mammalian cells and purified from culture medium without resorting to denaturing conditions. The truncated products were fragments Nd-I (positions 1-905) comprising the N-terminal globule and rod-like domain and Nd-II corresponding mainly to the C-terminal globule (position 906-1217). Recombinant nidogen was indistinguishable from authentic nidogen obtained by guanidine dissociation from tumor tissue with respect to size, N-terminal sequence, CD spectra and immunochemical properties. They differed in protease stability and shape indicating that the N-terminal domain of the more native, recombinant protein consists of two globules connected by a flexible segment. This established a new model for the shape of nidogen consisting of three globes of variable mass (31-56 kDa) connected by either a rod-like or a thin segment. Recombinant nidogen formed stable complexes (Kd less than or equal to 1 nM) with laminin and collagen IV in binding assays with soluble and immobilized ligands and as shown by electron microscopy. Inhibition assays demonstrated different binding sites on nidogen for both ligands with different specificities. This was confirmed in studies with fragment Nd-I binding to collagen IV and fragment Nd-II binding to laminin fragment P1. In addition, recombinant nidogen but not Nd-I was able to bridge between laminin or P1 and collagen IV. Formation of such ternary complexes implicates a similar role for nidogen in the supramolecular organization of basement membranes.

Molecular characterization of cuticle and interstitial collagens from worms collected at deep sea hydrothermal vents

Two different collagens were isolated and characterized from the body walls of the vestimentiferan tube worm Riftia pachyptila and the annelid Alvinella pompejana, both living around hydrothermal vents at a depth of 2600 m. The acid-soluble cuticle collagens consisted of a long triple helix (2.4 microns for Alvinella, 1.5 microns for Riftia) terminating into a globular domain. Molecular masses of 2600 and 1700 kDa, respectively, were estimated from their dimensions. The two cuticle collagens were also quite different in amino acid composition, in agreement with their different supramolecular organizations within tissues. Interstitial collagens corresponding to cross-striated fibrils underneath the epidermal cells could be solubilized by digestion with pepsin and consisted of a single alpha-chain. They were similar in molecular mass (340 kDa) and length (280 nm) but differed in composition and banding patterns of segment-long-spacing fibrils. This implicates significant sequence differences also in comparison to fibril-forming vertebrate collagens, although all form typical quarter-staggered fibrils. The thermal stability of the worm collagens was, with one exception (interstitial collagen of Riftia), in the range of mammalian and bird collagens (37 to 46 degrees C), and thus distinctly above that of shallow sea water annelids. Yet, their 4-hydroxyproline contents were not directly correlated to this stability. About 20% of Riftia collagen alpha-chain sequence was elucidated by Edman degradation and showed typical Gly-X-Y repeats but only a limited homology (45 to 58% identity) to fibril-forming vertebrate collagens. A single triplet imperfection and the variable hydroxylation of proline in the X position were additional unique features. It suggests that this collagen represents an ancestral form of fibril-forming collagens not directly corresponding to an individual fibril-forming collagen type of vertebrates.

Identification of a major antigenic epitope on CNBr-fragment 11 of type II collagen recognized by murine autoreactive B cells

Immunization of certain strains of mice with native type II collagen (CII) induces both development of arthritis and an antibody response to autologous CII. The autoantibody response in a high-responder strain, the DBA/1 mouse, has been described earlier, and a number of monoclonal antibodies have been characterized for arthritogenicity and autoreactive binding to cartilage in vivo and in vitro. Here we map the antigenic epitope of one of these arthritogenic monoclonal antibodies (CII-C1). It belongs to a group of antibodies recognizing the CNBr fragment alpha 1(II)-CB11 of CII. Using the enzyme-linked immunosorbent assay technique, we show that the antibody reacts only with native, triplehelical CII, but not with other collagens. The antibody is able to stain specifically the CB11 fragment by immunoblotting, suggesting some partial renaturation of the CNBr fragment into triple-helical structures after blotting. The binding site of CII-C1 on CB11 was further focused by rotary shadowing of antibody-labeled CII to a site 89 +/- 8 nm from the amino end of CII, corresponding to the middle of CB11. This location was confirmed by cleavage of CB11 with trypsin, separation of the tryptic peptides by high-performance liquid chromatography and dot-blot analysis of the antigenic peptides with the CII-C1 antibody. Sequencing of the single positive peptide located the antigenic epitope within the sequence GFAGQAGPAGATGAPGRP (residues 316-333). Assuming 0.29 nm per residue, this corresponds to a position within 92-96.5 nm from NH2 terminal end of CII. Apart from glycine residues, which are not exposed on the triple-helical structure, only two amino acid residues (F-x-y-Q) are conserved in CII from different species but are not found in the triple-helix of other collagens except type IV collagen. Therefore, this structure is likely to be of critical importance for the binding of the CII-C1 antibody. Of potential importance is that this structure is also found in certain other arthritogenic proteins such as 65-kDa mycobacterial protein, in CMV and EBV.

Cell adhesion, spreading and neurite stimulation by laminin fragment E8 depends on maintenance of secondary and tertiary structure in its rod and globular domain

The cell adhesion, spreading and neurite-promoting properties of mouse tumor laminin fragment E8, which contains major site(s) responsible for laminin-cell interactions, were probed by proteolytic degradation, denaturation, synthetic peptides and antibody inhibition. Removal of more than half of the N-terminal portion contributing to the rod-like domain did not effect cell attachment or spreading although neurite-promoting activity was reduced. More extensive degradation of the rod or of the globular domains of E8, or separation of the globule from the rod, also resulted in loss of cell spreading activity although weak attachment was found to an A chain subfragment comprising the globular domain and a short piece of the rod. Exposure of E8 to increasing concentrations of dissociating agents produce an apparently reversible denaturation but an irreversible loss of both attachment and neurite-promoting activities, as did reduction and alkylation of disulfide bonds in the globular domain. Although cell adhesion and spreading were blocked by antibodies to an alpha 6 integrin subunit, neurite outgrowth was unaffected, indicating two distinct receptors for these two activities. Furthermore, a synthetic peptide, the sequence of which is found in the vicinity of adhesion and neurite-promoting sites and previously implicated in neurite growth and cell attachment activities, was found to be inactive. These results indicate that the major cell attachment and neurite-promoting sites of laminin are distinct although both require the native conformation of parts of the rod and the terminal globular domain of the long arm of laminin.

Binding of the pentamer/hexamer forms of mannan-binding protein to zymosan activates the proenzyme C1r2C1s2 complex, of the classical pathway of complement, without involvement of C1q

The serum lectin, mannan binding protein (MBP), was isolated in a yield of 40 micrograms/liter from pooled normal human serum by affinity chromatography on mannan-Sepharose, followed by gel-filtration and ion-exchange chromatography and finally by passage down an anti-IgM Sepharose column. A rabbit antiserum was prepared against the purified MBP and an enzyme-linked immunoassay developed that used both the specificity of the polyclonal antibody and the Ca+(+)-dependent carbohydrate binding property of MBP. Assay of the sera from 103 blood-donors showed a wide range of MBP levels, ranging from 0 to 870 micrograms/liter. MBP, after interaction with zymosan, caused efficient activation of a C1r2 125I-C1s2 complex that was prepared by incubation of 125I-C1s2 with serum, from a patient with a complete genetic deficiency of C1q, followed by gel-filtration on Sepharose 6B. The purified MBP is composed of a mixture of trimers, tetramers, pentamers, and hexamers of an approximate 90-kDa structural unit as judged by chromatography, SDS-PAGE and electron microscopy studies. Only the molecules in the pentamer/hexamer fraction, which have a similar overall structure to that of C1q, appeared to cause efficient, zymosan-dependent, activation of C1s within the C1r2C1s2 complex. The pentamer/hexamer form of MBP may therefore play an important role in antibody-independent activation of the C system during the early stages of certain infections.

Binding of the pentamer/hexamer forms of mannan-binding protein to zymosan activates the proenzyme C1r2C1s2 complex, of the classical pathway of complement, without involvement of C1q

The serum lectin, mannan binding protein (MBP), was isolated in a yield of 40 micrograms/liter from pooled normal human serum by affinity chromatography on mannan-Sepharose, followed by gel-filtration and ion-exchange chromatography and finally by passage down an anti-IgM Sepharose column. A rabbit antiserum was prepared against the purified MBP and an enzyme-linked immunoassay developed that used both the specificity of the polyclonal antibody and the Ca+(+)-dependent carbohydrate binding property of MBP. Assay of the sera from 103 blood-donors showed a wide range of MBP levels, ranging from 0 to 870 micrograms/liter. MBP, after interaction with zymosan, caused efficient activation of a C1r2 125I-C1s2 complex that was prepared by incubation of 125I-C1s2 with serum, from a patient with a complete genetic deficiency of C1q, followed by gel-filtration on Sepharose 6B. The purified MBP is composed of a mixture of trimers, tetramers, pentamers, and hexamers of an approximate 90-kDa structural unit as judged by chromatography, SDS-PAGE and electron microscopy studies. Only the molecules in the pentamer/hexamer fraction, which have a similar overall structure to that of C1q, appeared to cause efficient, zymosan-dependent, activation of C1s within the C1r2C1s2 complex. The pentamer/hexamer form of MBP may therefore play an important role in antibody-independent activation of the C system during the early stages of certain infections.

Binding of nidogen and the laminin-nidogen complex to basement membrane collagen type IV

The laminin-nidogen complex and purified nidogen both bind collagen IV but not other collagens, as shown by solid-state ligand-binding and inhibition assays. Laminin purified from the dissociated complex and a variety of laminin proteolytic fragments failed to bind collagen IV. Complexes formed in solution between nidogen or laminin-nidogen and collagen IV were visualized by rotary shadowing which identified one major binding site about 80 nm away from the C-terminus of the collagen triple helix. A second, weaker binding site may exist closer to its N-terminus. Binding sites of nidogen were assigned to its C-terminal globular domain which also possesses laminin-binding structures. A more diverse collagen-IV-binding pattern was observed for the laminin nidogen complex, whereby interactions may involve both nidogen and short-arm structures of laminin.

Extended and globular protein domains in cartilage proteoglycans

Electron microscopy after rotary shadowing and negative staining of the large chondroitin sulphate proteoglycan from rat chondrosarcoma, bovine nasal cartilage and pig laryngeal cartilage demonstrated a unique multidomain structure for the protein core. A main characteristic is a pair of globular domains (diameter 6-8 nm), one of which forms the N-terminal hyaluronate-binding region. They are connected by a 25 nm-long rod-like domain of limited flexibility. This segment is continued by a 280 nm-long polypeptide strand containing most chondroitin sulphate chains (average length 40 nm) in a brush-like array and is terminated by a small C-terminal globular domain. The core protein showed a variable extent of degradation, including the loss of the C-terminal globular domain and sections of variable length of the chondroitin sulphate-bearing strand. The high abundance (30-50%) of the C-terminal domain in some extracted proteoglycan preparations indicated that this structure is present in the cartilage matrix rather than being a precursor-specific segment. It may contain the hepatolectin-like segment deduced from cDNA sequences corresponding to the 3'-end of protein core mRNA [Doege, Fernandez, Hassell, Sasaki & Yamada (1986) J. Biol. Chem. 261, 8108-8111; Sai, Tanaka, Kosher & Tanzer (1986) Proc. Natl. Acad. Sci. 83, 5081-5085; Oldberg, Antonsson & Heinegård (1987) Biochem. J. 243, 255-259].

Domain structure of cartilage proteoglycans revealed by rotary shadowing of intact and fragmented molecules

The rotary-shadowing technique for molecular electron microscopy was used to study cartilage proteoglycan structure. The high resolution of the method allowed demonstration of two distinct globular domains as well as a more strand-like portion in the core protein of large aggregating proteoglycans. Studies of proteoglycan aggregates and fragments showed that the globular domains represent the part of the proteoglycans that binds to the hyaluronic acid, i.e. the hyaluronic acid-binding region juxtapositioned to the keratan sulphate-attachment region. The strand-like portion represents the chondroitin sulphate-attachment region. Low-Mr proteoglycans from cartilage could be seen as a globule connected to one or two side-chain filaments of chondroitin sulphate.

Structure and interactions of heparan sulfate proteoglycans from a mouse tumor basement membrane

Various forms of heparan sulfate proteoglycan were solubilized from the mouse Engelbreth-Holm-Swarm (EHS) sarcoma by extraction with 0.5 M NaCl, collagenase digestion and extraction with 4 M guanidine. They could be separated into high (greater than or equal to 1.65 g/ml) and low (1.38 g/ml) buoyant density variants. The high-density form from the NaCl extract and collagenase digest had Mr = 130000 and So20,W = 4.5 S and contained 4-10% protein, indicating Mr = 5 000-12 000 for the protein core. This proteoglycan exhibited polydispersity as shown by rotary shadowing electron microscopy and ultracentrifugation. An average molecule consisted of four heparan sulfate chains (Mr = 29 000) each with a length of 32 +/- 10 nm. The low-density form (Mr about 400 000) could not be completely purified and contained about 50% protein. As shown by radioimmunoassay, the various proteoglycans shared similar protein cores. Labeling of the tumor in vivo or in vitro demonstrated preferential incorporation of radioactive sulfate in the high-density form. The high-density proteoglycan interacted in affinity chromatography by virtue of its heparan sulfate chains with laminin, fibronectin, the globular domain NC1 and the triple helix of collagen IV. These interactions were abolished at moderate concentrations of NaCl (0.1-0.2 M) and in the presence of heparin, chondroitin sulfate or dextran sulfate. Interactions with the globule NC1 could also be demonstrated by velocity band centrifugation in sucrose gradients and a binding constant of about 10(6) M-1 was derived.

Monoclonal antibodies against chicken type IV and V collagens: electron microscopic mapping of the epitopes after rotary shadowing

The location of the epitopes for monoclonal antibodies against chicken type IV and type V collagens were directly determined in the electron microscope after rotary shadowing of antibody/collagen mixtures. Three monoclonal antibodies against type IV collagen were examined, each one of which was previously demonstrated to be specific for only one of the three pepsin-resistant fragments of the molecule. The three native fragments were designated (F1)2F2, F3, and 7S, and the antibodies that specifically recognize each fragment were called, respectively, IA8 , IIB12 , and ID2 . By electron microscopy, monoclonal antibody IA8 recognized an epitope located in the center of fragment (F1)2F2 and in tetramers of type IV collagen at a distance of 288 nm from the 7S domain, the region of overlap of four type IV molecules. Monoclonal antibody IIB12 , in contrast, recognized an epitope located only 73 nm from the 7S domain. This result therefore provides direct visual evidence that the F3 fragment is located closest to the 7S domain and the order of the fragments must be 7S-F3-(F1)2F2. The epitope for antibody ID2 was located in the overlap region of the 7S domain, and often several antibody molecules were observed to binding to a single 7S domain. The high frequency with which antibody molecules were observed to bind to fragments of type IV collagen suggests that there is a single population of type IV molecules of chain organization [alpha 1(IV)]2 alpha 2(IV), and that four identical molecules must form a tetramer that is joined in an antiparallel manner at the 7S domain. The monoclonal antibodies against type V collagen, called AB12 and DH2 , were both found to recognize epitopes close to one another, the epitopes being located 45-48 nm from one end of the type V collagen molecule. The significance of this result still remains uncertain, but suggests that this site is probably highly immunoreactive. It may also be related to the specific cleavage site of type V collagen by selected metalloproteinases and by alpha-thrombin. This cleavage site is also known to be located close to one end of the type V molecule.

Subunit structure and assembly of the globular domain of basement-membrane collagen type IV

The globular domain of collagen IV was solubilized by collagenase digestion from a mouse tumor, human placenta and bovine aorta and was purified by chromatographic methods. The materials show a unique, mainly non-collagenous amino acid composition and contain small amounts of glucosamine and galactosamine. The globular structures with Mr = 170 000 appear as a hexameric assembly originating from two collagen IV molecules. Subunits of this assembly are two different dimers Da and Db (Mr about 56 000) and monomers (Mr = 28 000). Their N-terminal amino acid sequences start with short triple-helical sequences, which overlap with the C-terminal triple helix of the alpha 1(IV) and alpha 2(IV) chain, demonstrating that the globule originates from the C terminus of collagen IV. Dimers arise from monomers by disulfide cross-linking (form Db) and/or formation of non-reducible cross-links (form Da). Reduction under non-denaturing conditions causes partial dissociation of the globule and of collagen IV dimers, indicating that reducible cross-links are formed between monomers of two different collagen IV molecules. Dissociation of the hexamer into the subunits can be achieved with 8 M urea, sodium dodecyl sulfate or in the pH range 2.5-4. The latter indicates that carboxyl groups are essential for association. Mixtures of the subunits (monomers and dimers) or purified dimers reassemble in neutral buffer into hexamers as shown by ultracentrifugation and electron microscopy. Reconstituted hexamers, however, dissociate in a much broader pH range than the native globules. Circular dichroic spectra indicate that the structure is more completely refolded from acid-treated than from urea-treated material. These data suggest that globules originating from monomers (as existing in single collagen IV molecules) are stabilized by the adjacent triple helix. Covalent cross-link formation stabilizes the globular structure and allows reconstitution in stoichiometric proportions.

Isolation and characterization of a precursor form of M collagen from embryonic chicken cartilage

A disulfide-cross-linked collagen has been extracted with neutral salt solutions from organ cultures of embryonic chick sternal cartilage. This collagen, which we term pM collagen, is presumed to be the native extracellular precursor molecule to disulfide-cross-linked collagen fragments recently described. Cleavage of pM collagen under native conditions with pepsin gives rise to the collagen fragments M1 and M2, which had also been isolated from pepsin extracts of chick hyaline cartilage [K. von der Mark, M. van Menxel & H. Wiedemann (1982) Eur. J. Biochem. 124, 57-62]. Native pM collagen was purified by DEAE-cellulose chromatography and agarose gel filtration. On agarose and following polyacrylamide gel electrophoresis, the unreduced molecule migrates with an apparent Mr of 300 000. Reduction of disulfide bridges produces two subunits with Mr 80 000 (pMa) and 60 000 (pMb) when compared with collagen standards. Cyanogen bromide cleavage of pMa and pMb, excised from dodecyl sulfate gels, resulted in different peptide maps, indicating that both components are genetically distinct polypeptide chains. The occasional appearance of the unreduced pM collagen as a doublet band on dodecyl sulfate gels and the observation that pMa and pMb occur in non-stoichiometric ratios suggests that pMa and pMb form separate native molecules, although their incorporation into a single pM molecule cannot be excluded. Native pM collagen was completely digested with bacterial collagenase, and contained hydroxyproline and proline in a ratio of 1.15:1, indicating the absence of significant non-collagenous domains. Thus it represents, despite several pepsinlabile sites, more likely a largely triplehelical, processed form of collagen rather than a procollagen-like molecule containing globular domains. Processing of pM collagen to M1 and M2 fragments or other intermediate forms was not observed in cartilage organ culture or in chondrocyte cell cultures within 18 h.

The use of monoclonal antibodies to fragments of chicken type IV collagen in structural and localization studies

In previous experiments, three pepsin-resistant fragments of type IV collagen were isolated from chicken gizzards and designated 7S, F3, and (F1)2F2 (Mayne, R., and Zettergren, J. G. (1980) Biochemistry 19, 4065-4072). In the present experiments, a series of monoclonal antibodies to type IV collagen were prepared, each one of which recognized an epitope present in only one of the three fragments. A high molecular weight fraction of type IV collagen (designated 7S + arms (215 nm)) was isolated after agarose gel filtration and characterized by electron microscopy after rotary shadowing and by gel electrophoresis. Analysis of 7S + arms (215 nm) by inhibition enzyme-linked immunosorbent assay demonstrated the presence of the epitopes for 7S and F3 but not for (F1)2F2. This result, therefore, provides additional evidence that the order of the pepsin-resistant fragments of chicken type IV collagen is 7S-F3-(F1)2F2.

Electron-microscopical approach to a structural model of intima collagen

Intima collagen was studied by electron microscopy (rotary shadowing and negative staining) and by analytical ultracentrifugation. It was found that the monomeric unit (Mr 170 000) consists of a 105 nm-long triple helix terminated by a small globular domain (Mr about 30 000) at one end and a large globular domain (Mr about 40 000) at the other end. The monomer was produced by selective reduction of interchain disulphide bridges. Before reduction, dimers, tetramers and larger filamentous structures were found. Dimers are lateral staggered aggregates of two monomers aligned in an anti-parallel fashion. This gives rise to an inner 75 nm-long region of two slightly intertwisted triple helices flanked by the large globular domains. The outer triple-helical segments (length 30 nm) with the small globular domains at their ends emerge at both sides of this structure. Interchain disulphide bridges are probably located in the vicinity of the large domains. Only the outer segments could be degraded by bacterial collagenase. In tetramers the outer segments of two dimers are covalently linked, forming a scissors-like structure. In the fibrous forms several tetramers are assembled end-to-end with an overlap between the outer segments. The molecular masses and sedimentation coefficients were calculated for these various forms from the electron-microscopically observed dimensions and agreed with results obtained by ultracentrifugation. The unique structure of intima collagen suggests that it originates from a microfibrillar component and that it can be considered a unique collagenous protein, for which we propose the designation type VI collagen.