Immunolocalization of type VIII collagen in vascular tissue

Type VIII collagen was first detected in the culture medium of aortic endothelial cells. Subsequently its synthesis by a number of other cell lines was demonstrated. Its presence in vascular tissue is reported here. It is a component of sheep aorta and carotid artery but could not be demonstrated in the jugular vein. It is principally localized in the subendothelial region but this can only be demonstrated after pretreatment of the tissue with proteases. Thus type VIII collagen is a constituent of blood vessels.     

Biosynthetic and structural properties of endothelial cell type VIII collagen

A highly unusual endothelial cell collagen (Sage, H., Pritzl, P., and Bornstein, P., (1980) Biochemistry 19, 5747-5755) has been characterized in greater detail. Pulse-chase experiments with bovine aortic endothelial cells revealed two nondisulfide-bonded collagens, of apparent chain Mr = 177,000 and 125,000, with an estimated synthesis and secretion time of 75 min. Stepwise, quantitative processing to stable lower molecular weight forms as described for type I procollagen was not observed. Endothelial collagen was secreted over a temperature range of 24-37 degrees C and, prior to heat denaturation, did not display affinity for a gelatin-binding fragment of fibronectin coupled to Sepharose. The presence of a pepsin-resistant domain (Mr = 50,000) in both the soluble and cell layer-associated forms of this protein was shown by ion exchange chromatography and sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Endothelial collagen was cleaved by vertebrate collagenase into several discrete fragments that differed in molecular weight from the characteristic alpha A and alpha B fragments generated from the interstitial collagens. Nontriple helical domains corresponding to the NH2- and COOH-terminal propeptides of other procollagen types were not found after incubation of endothelial collagen with bacterial collagenase. Additional evidence for the lack of extended noncollagenous sequences was provided by studies with mast cell proteases, which convert native procollagen to collagen but are unreactive toward native interstitial collagens. Endothelial collagen was not cleaved by these enzymes at 37 degrees C, but, as observed for interstitial collagen alpha chains, required prior heating at elevated temperatures for cleavage to occur. In view of this unique set of structural characteristics, and a distribution that is not restricted to the endothelium, we have designated this protein as type VIII collagen.     

Cleavage of type VIII collagen by human neutrophil elastase.

In this report, the susceptibility of type VIII collagen to human neutrophil elastase is compared to other extracellular matrix components. Type X collagen is degraded to specific fragments at a substrate to enzyme ratio of 5:1 after 20 h at room temperature, but type VIII collagen is almost completely degraded after only 4 h incubation at a substrate to enzyme ratio of 50:1 and partly degraded after only 15 min. Laminin, merosin and types I, III, IV and V collagen exhibit no susceptibility to neutrophil elastase under the latter conditions, while fibronectin is degraded.     

The cloning and sequencing of alpha 1(VIII) collagen cDNAs demonstrate that type VIII collagen is a short chain collagen and contains triple-helical and carboxyl-terminal non-triple-helical domains similar to those of type X collagen

We have isolated two overlapping cDNA clones covering 2425 base pairs encoding a short type VIII collagen chain synthesized by rabbit corneal endothelial cells. The cDNAs encode an open reading frame of 744 amino acid residues containing a triple-helical domain of 454 residues flanked by 117- and 173-residue amino and carboxyl non-triple-helical domains (called NC2 and NC1, respectively). Based on the identity between the DNA-derived amino acid sequence and the amino acid sequence of a type VIII collagen CNBr peptide obtained from rabbit corneal Descemet's membrane, we conclude that the cDNAs code for a type VIII collagen chain. We give this chain the designation alpha 1(VIII). The alpha 1(VIII) triple-helical domain contains eight imperfections in the Gly-X-Y repeated structure with Gly-X instead of a full triplet. The length of the triple-helical domain and number and relative locations of these imperfections are remarkably similar to those of chicken alpha 1(X) collagen. The amino acid sequence of the carboxyl three-quarters of the NC1 domain has high sequence similarity to that of alpha 1(X) collagen. These data suggest that the triple-helix coding portions and carboxyl three-quarters of the NC1 domains of the alpha 1(VIII) and alpha 1(X) genes have a common evolutionary origin.     

Characteristics of cellular immune responses to collagen type I or collagen type II

We have examined the murine cell-mediated immune (CMI) response to collagens type I (CI) and type II (CII) as measured by in vivo delayed-type hypersensitivity responses. We have verified the histopathology and kinetics of the cell-mediated immune responses. Predominant cell-mediated responses were obtained 7, 10, or 14 days following immunization. A presumed antibody-mediated reaction was observed at later times (e.g., greater than 21 days following immunization). The CMI responses to the collagens show a strain-dependent relationship. For CI, the CMI response profile shows H-2b greater than or equal to H-2k = H-2q much greater than H-2d. For bovine CII, the response profile is H-2d greater than H-2b = H-2k = H-2q; the chick CII response profile is H-2q = H-2k greater than H-2b = H-2d, and in limited testing, only the H-2q strain could generate murine CII-specific cell-mediated immune responses. The CII-specific CMI response is cross-reactive with CII from several species of animals, but not with CI. Further, the collagen-specific CMI response can be elicited with certain cyanogen-bromide fragments of bovine CII. Finally, our study also demonstrates that there is a non-H-2-linked locus(i) involved in the development of CII-induced arthritis.     

Human macrophages synthesize type VIII collagen in vitro and in the atherosclerotic plaque.

Type VIII collagen is a short-chain collagen that is present in increased amounts in atherosclerotic lesions. Although the physiological function of this matrix protein is unclear, recent data suggest an important role in tissue remodeling. Type VIII collagen in the atherosclerotic lesion is mainly derived from smooth muscle cells. We now show that macrophages in the atherosclerotic vessel wall and monocytes in adjacent mural thrombi also express type VIII collagen. We demonstrated this using a novel combined fluorescence technique that simultaneously stains, within the same tissue section, specific RNAs by in situ hybridization and proteins by indirect immunofluorescence. In culture, human monocyte/macrophages expressed type VIII collagen at all time points from 1 h to 3 wk after isolation. Western blotting and immunoprecipitation also revealed secretion of type VIII collagen into the medium of 14-day-old macrophages. Because this is the first report of secretion of a collagen by macrophages, we tested the effect of lipopolysaccharide (LPS) and interferon gamma, substances that stimulate macrophages to secrete lytic enzymes, on macrophage expression of type VIII collagen. LPS and interferon gamma decreased expression of type VIII collagen. By contrast, secretion of matrix metalloproteinase 1 (MMP 1) was increased, indicating a switch from a collagen-producing to a degradative phenotype. Double in situ hybridization studies of expression of type VIII collagen and MMP 1 in human coronary arteries showed that in regions important for plaque stability, the ratio of MMP 1 RNA to macrophage type VIII collagen RNA varies widely, indicating that the transition from one phenotype to the other that we observed in vitro may also occur in vivo.     

Characterization of the collagen in the hexagonal lattice of Descemet's membrane: its relation to type VIII collagen

To investigate the nature of the hexagonal lattice structure in Descemet's membrane, monoclonal antibodies were raised against a homogenate of bovine Descemet's membranes. They were screened by immunofluorescence microscopy to obtain antibodies that label Descement's membrane. Some monoclonal antibodies labeled both Descemet's membrane and fine filaments within the stroma. In electron microscopy, with immunogold labeling on a critical point dried specimen, the antibodies labeled the hexagonal lattices and long-spacing structures produced by the bovine corneal endothelial cells in culture; 6A2 antibodies labeled the nodes of the lattice and 9H3 antibodies labeled the sides of the lattice. These antibodies also labeled the hexagonal lattice of Descemet's membrane in situ in ultrathin frozen sectioning. In immunofluorescence, these antibodies stained the sclera, choroid, and optic nerve sheath and its septum. They also labeled the dura mater of the spinal cord, and the perichondrium of the tracheal cartilage. In immunoblotting, the antibodies recognized 64-kD collagenous peptides both in tissue culture and in Descemet's membrane in vivo. They also recognized 50-kD pepsin-resistant fragments from Descemet's membranes that are related to type VIII collagen. However, they did not react either in immunoblotting or in immunoprecipitation with medium of subconfluent cultures from which type VIII collagen had been obtained. The results are discussed with reference to the nature of type VIII collagen, which is currently under dispute. This lattice collagen may be a member of a novel class of long-spacing fibrils.     

In vivo glycosylation of dermal and tendon type I collagen

Recent studies show that native collagen fibers in the extracellular space can be subject to nonenzymatic glycosylation and that the extent of such glycosylation increases in clinical hyperglycemia and aging. In the present study, a comparison was made on the extent of glycosylation in rat tail tendon and in the soluble and insoluble fractions of collagen separated from rat skin after in vivo labeling with [14C]glucose. It was observed that nonenzymatic glycosylation occurred maximally in the salt-soluble fraction as measured by the level of ketoamine linked hexose. 14C radioactivity incorporation as well as the number of free amino groups was also increased in this fraction. However, the amounts of O-glycosidically linked sugars did not show much variation between the soluble and insoluble fractions. These findings could be correlated to the enhanced metabolic turnover of newly synthesized collagen in diabetics.     

Expression of type VIII collagen during morphogenesis of the chicken and mouse heart.

The expression of type VIII collagen is restricted, in adult mammals, to specialized extracellular matrices and to a select subset of blood vessels. We have examined the distribution of type VIII collagen in sequential stages of mouse and chicken embryos and found a temporal and spatially restricted pattern of expression during cardiogenesis. Type VIII collagen was first detected by immunocytochemistry on Day 11 in the developing mouse embryo and at stage 19 in the chicken embryo. The distribution of this protein was rapidly modulated during cardiac morphogenesis. Initially (Day 11 in the mouse embryo), type VIII collagen was associated with cardiac myoblasts. From Days 15 to 18, the immunoreactive component was progressively diminished in the myocardium; however, this collagen was observed in the subendocardial layer of the atrioventricular canal and later in the cardiac jelly (or the myocardial basement membrane, an area associated with the formation of cardiac valves). On Day 17, type VIII collagen was also detected in the subendothelium (intima) and tunica media of large vessels. Neonatal and adult hearts contained low to undetectable levels of type VIII collagen. The presence of type VIII collagen was confirmed by immunoblot analysis of heart extracts at different stages of development. A major 185-kDa component, as well as polypeptides of 68 and 15 kDa, reacted with anti-type VIII collagen IgG. Exposure of heart extracts to hyaluronidase or reducing agent eliminated immunoreactivity of the 185-kDa component but not that of the 68- and 15-kDa polypeptides. Type VIII collagen therefore appears to be associated with a hyaluronidase-sensitive component of the extracellular matrix during a temporally restricted stage of embryonic cardiogenesis. The contribution of this collagen to cardiac morphogenesis might reside, in part, in its ability to influence the differentiation of the myocardium and formation of the cardiac valves.     

Astrocytes express type VIII collagen during the repair process of brain cold injury

We recognized for the first time upregulation of type VIII collagen gene expression during the repair process in the mouse brain cold injury model. Immunohistochemical staining showed that type VIII collagen expression was around the necrotic region, where reactive astrocytes are frequently observed. Cultured astrocytes demonstrated a high expression of type VIII collagen genes. TGF-beta1 enhanced the expression of both alpha1(VIII) and alpha2(VIII) genes by astrocytes in culture. Further, we tested selected biological activities of type VIII collagen, compared with those of type I, IV, and V collagens and fibronectin. Astrocytes adhered to type VIII collagen via receptors requiring metal ions. Astrocyte migration on type VIII collagen was more stimulated than that observed on the other ECM molecules. These data indicate that type VIII collagen plays an important role in glial scar formation during the repair process by astrocytes.     

Partial characterization of type X collagen from bovine growth-plate cartilage. Evidence that type X collagen is processed in vivo

Sequential extraction of bovine growth-plate cartilage with 4 M guanidinium chloride and pepsin was used to identify the intact and pepsinized forms respectively of type X collagen. This collagen occurs predominantly as the processed [alpha 1(X)]3 form in vivo, although the procollagen [pro alpha 1(X)]3 form can also be detected. The bovine pro alpha 1(X) and alpha 1(X) chains have Mr values identical to the corresponding chick species (Mr 59,000 and 49,000). However, the pepsinized alpha 1(X)p chains (Mr 47,000) are larger than those of the chick (Mr 45,000), and the bovine collagen type X is further distinguished by being disulphide-bonded within the triple-helical domain.     

Distribution of non-enzymatically bound glucose in in vivo and in vitro glycosylated type I collagen molecules

Non-enzymatic glycosylation of collagen occurs both in vivo during diabetes and in vitro after incubation with glucose. Glycosylated collagen exhibits altered physicochemical and biological properties which could explain some of the complications of diabetes. To provide a mechanistic explanation of this modification the localization of bound glucose was investigated using NaB[3H]H4 reduction and CNBr cleavage. Glucose fixation is distributed mainly on the alpha 1CB6 peptide after in vitro glycosylation whereas this distribution occurs less specifically during diabetes. It is concluded that fibrillogenesis alteration of in vitro glycosylated collagen is related to glucose fixation on free epsilon NH2 sites normally implied in intermolecular interactions.     

Type VIII collagen: heterotrimeric chain association

Two chains, alpha1(VIII) and alpha2(VIII), have been described for type VIII collagen. Early work suggested that these chains were present in a 2:1 ratio, although recent work has shown that homotrimers can form and predominate in some tissues. In order to address the question of whether the alpha1(VIII) and alpha2(VIII) chains could co-polymerise we made a shortened alpha1(VIII) chain and expressed this with full length alpha2(VIII) chain in an in vitro translation system supplemented with semi-permeabilised cells. Heterotrimers containing either two or one alpha2(VIII) were evident. Interestingly, a point mutation in the NC1 domain of the alpha1(VIII) chain abrogated trimer formation. In addition we were able to demonstrate chain association of the alpha1(X) chain of type X collagen with the shortened alpha1(VIII) chain. Variations in chain association were seen when altered ratios of message were used. These results demonstrate the importance of the NC1 domain in chain association and suggest that gene expression regulates the composition and function of type VIII collagen by varying chain composition.     

The collagenous protein with elastin crosslinks from Descemet's membrane is not related to type VIII collagen

A collagen-like insoluble protein containing the elastin cross-links (desmosine and isodesmosine) has been isolated from Descemet's membrane. Recently type VIII collagen (endothelial collagen) has been shown to be a major constituent of this membrane. Biochemical studies suggest that these two proteins are unrelated. The cyanogen bromide peptide maps show negligible similarity. Antiserum raised against oxalic acid digests of elastin (alpha-elastin) did not react against an oxalic acid digests of type VIII collagen but did show some reaction against the cross-linked preparation. Immunofluorescent localization has demonstrated the presence of type VIII collagen in trachea but a desmosine cross-linked collagen could not be isolated from this tissue.     

Electroconductivity of collagen in vitro and in vivo

Electrical conductivity was measured on thermally reconstituted collagen fibers in vitro and on isolated rat tail tendon collagen fiber bundles in vivo, The results obtained indicated that collagen per se is not an electroconductor under physiological conditions, but rather a biological insulator.     

Gene and protein expressions of type I collagen are regulated tissue-specifically in rat hyaline cartilages in vivo

The present study was designed to investigate how rat hyaline cartilages at various sites in vivo express the gene and protein of type I collagen using in situ hybridization and immunohistochemistry. The gene of pro alpha 1(I) collagen was expressed by chondrocytes in articular cartilage, and the protein of type I collagen was identified in the cartilage matrix. In contrast, growth plate cartilage expressed the gene of pro alpha 1(I) collagen, but no protein of type I collagen. Neither gene nor protein of type I collagen was expressed in cartilages of trachea and nasal septum. The present study suggested that expression of type I collagen in hyaline cartilages may be regulated tissue-specifically at gene and/or protein levels.