Abstract
Immunolocalization studies demonstrate that Type VI collagen forms a flexible network that interweaves among collagen fibrils in the dermis of skin as well as in other loose connective tissues. Although binding of Type VI collagen with other matrix components has been suggested, no structural evidence supporting these studies has been reported. In this study, we demonstrate that Type VI microfilaments consistently cross-banded collagen fibrils near the “d” band, indicating that the interaction of Type VI collagen with banded fibrils is not passive. This “d” band is also the location of the binding domain of decorin to banded fibrils, suggesting that decorin mediates the interaction of Type VI microfilaments with banded fibers. Examination of the architecture of the Type VI network in a decorin nullizygous mouse demonstrates a continuance of this specific interaction, indicating that the association is not entirely dependent on the presence of decorin. At least one other component, whose identity is uncertain, persists near the “d” band, which may also serve to mediate the attachment of Type VI collagen to collagen fibrils.
Keywords
I
The banded collagen fibrils of skin are heteropolymers composed primarily of Type I collagen, with lesser amounts of collagen Types III and V. Type III collagen is limited to the outside of the fibrils and may participate in the regulation of fibril diameter. The bulk of Type V collagen is within the interior of the fibril, though a portion of the Type V molecule may protrude from the fibril surface (for review see Keene et al. 1997).
Decorin (Krusius and Ruoslahti 1986; Fisher et al. 1989; Scott 1996; Weber et al. 1996) is a member of an expanding family of small leucine-rich proteoglycans (SLRP) whose hallmark is their ability to interact with extracellular and cell surface proteins, thereby regulating matrix assembly and cell growth (Iozzo and Murdoch 1996; Iozzo 1997). Decorin is known to bind various types of collagen, including Types I, II, III, V, and VI (Vogel et al. 1984; Bidanset et al. 1992; Ramamurthy et al. 1996), and to retard collagen fibrillogenesis in vitro. Decorin is essential for the orderly assembly of fibrillar collagen during development, as demonstrated by mice that harbor a targeted disruption of both decorin alleles (Danielson et al. 1997). These mice are viable but manifest a skin fragility phenotype. In the dermis of the Dcn -/- mouse, many of the individual banded collagen fibrils are twisted and of variable diameter along their lengths. Similar collagen fibril profiles are seen in human disorders that lead to fragile skin phenotypes, suggesting that the perturbance of normal banded collagen fibrillogenesis is sufficient to cause fragile skin.
Immunocytochemical studies in fetal and neonate cornea suggest an association of decorin with Type VI collagen microfilaments (Takahashi et al. 1993). The interaction between Type VI and decorin is unaffected by chondroitinase in solid-phase binding assays (Bidanset et al. 1992). Chondroitinase has been shown to disrupt the Type VI network in cornea (Nakamura et al. 1997), indicating that the carbohydrate portion of another proteoglycan may be important to the stability of Type VI collagen.
In this study we sought to determine if there is a consistent site of interaction between Type VI collagen microfilaments and banded collagen fibers, and to establish whether or not the presence of decorin is required for this biological interaction. We show that Type VI microfilaments interact near the “d” band of both human and mouse dermal collagen fibrils. Surprisingly, the association of Type VI microfilaments with collagen fibrils persists in decorin-deficient animals, suggesting that additional components, likely to be other members of the SLRP gene family, also mediate an attachment of Type VI collagen to banded collagen fibrils.
Materials and Methods
Antibodies
The rabbit PAb 1909 recognizing human decorin was purchased from Chemicon (Temecula, CA). The rabbit PAb LF113 recognizing the amino terminal peptide region of mouse decorin was a generous gift from Dr. Larry Fisher (Fisher et al. 1995). The mouse MAb 3C4 recognizing human Type VI collagen and also the PAb 279 rabbit antibody recognizing mouse Type VI collagen were generous gifts from Dr. Eva Engvall.
Immunocytochemistry
En bloc immunolabeling of tissues was carried out using a previously described protocol (Sakai and Keene 1994), with some modification. Freshly obtained human neonatal foreskin and abdominal skin samples from both a decorin nullizygous mouse and a wild-type littermate (Danielson et al. 1997) were cut perpendicularly to the epithelium into strips of 0.2–1 mm and then incubated in primary antibody diluted 1:5 in PBS, pH 7.4, overnight at 4C. After a 6-hr wash in PBS at 4C, the samples were incubated in appropriate 5-nm colloidal gold conjugate (Amersham; Poole, UK) diluted 1:3 in BSA buffer (20 mM Tris-HCl, 0.9% NaCl, 1 mg/ml BSA), pH 8.0, overnight at 4C, followed by an extensive rinse in PBS as above. The samples were then rinsed briefly in 0.1 M sodium cacodylate buffer, pH 7.4, fixed in cacodylate-buffered 1.5% formaldehyde/1.5% glutaraldehyde containing 6000 ppm ruthenium hexammine trichloride (RHT), rinsed in cacodylate buffer containing 6000 ppm RHT, postfixed in 1% OsO4 containing 6000 ppm RHT, rinsed in ruthenium-free cacodylate buffer, dehydrated in a graded series of ethanol to 100%, washed in propylene oxide, and embedded in Spurr's epoxy.
Results
Labeling of human skin with MAb 3C4 (specific for Type VI collagen) resulted in the decoration of Type VI microfilaments at 100-nm intervals (Figure 1C, inset), consistent with the length of the Type VI collagen molecule and the known periodicity of Type VI microfilaments (Keene et al. 1988; Kuo et al. 1995). The addition of RHT to the fixatives after the immunocytochemical protocol (see Materials and Methods) resulted in staining of the Type VI microfilaments between labeled regions (Figure 1C, inset). The Type VI microfilaments are so fine in structure that they are virtually invisible unless labeled with antibody or stained with ruthenium. With the use of ruthenium in the absence of antibody, Type VI microfilaments can be recognized by a thin region, representing the collagen helix, followed by a thickened region, representing globular domains, present at 100-nm intervals (Figures 1A-1C, open arrows). Staining with RHT also results in the formation of a dense knob on the surface of the banded collagen fibrils near the “d” band (Figures 1A-1C), which is the same area to which antibody to decorin (PAb 1909) localizes (Figures 1A-1C, closed arrows). Careful examination reveals that when Type VI microfilaments cross banded collagen fibrils, the intersection invariably includes the region near the “d” band of collagen fibrils (Figures 1A-1C, circles) and is therefore not a random binding location.
Normal human skin. Decorin antibody directs gold particles to a region near the “d” band of collagen fibrils (closed arrows,
To gain further insight into the nature of the moiety interacting near the “d” band with Type VI collagen and considering that decorin may be implicated in performance of this function, we analyzed the interaction of Type VI microfilaments and collagen fibrils in the dermis of both wild-type and decorin-deficient mice. The architecture of the Type VI collagen network in mouse, as determined by immunocytochemical staining, closely resembles that reported in human skin (data not shown). En bloc immunolabeling of normal mouse skin with antibody specific for decorin (LF 113) results in periodic labeling of the fibers to a region near the “d” band (Figures 2A and 2C). Also consistent with human skin are the presence of knobs on the surface of the fibrils, near the “d” band, after exposure to RHT-containing fixatives (Figure 2B, arrowheads). As in human tissue, RHT-stained Type VI microfilaments are consistently seen to intersect banded collagen fibrils near the “d” band (Figure 2C, open arrow). As expected, exposure of decorin-deficient skin (Dcn -/-) to the decorin-specific antibody results in a complete lack of secondary gold labeling (Figure 2D). However, the presence of small knobs on the surface of the fibrils after RHT staining persists in the Dcn -/- mouse (Figure 2D, arrowheads), suggesting that another component is present near the “d” band in addition to decorin. Interestingly, fields could still be commonly found in which Type VI collagen microfilaments interacted at the “d” band of banded collagen in the decorin nullizygous mouse (Figure 2E, arrows). Collectively, we interpret these data as an indication that decorin is not the only component associated with collagen fibrils that mediate an interaction with Type VI collagen.
Mouse skin. In normal mice, MAb LF 113 to decorin localizes to a region near the “d” band (
Discussion
Type VI collagen microfilaments are composed of three different polypeptide chains, which form short triple helical regions and large carboxyl and amino terminal domains. Monomers align anti-parallel to form dimers, and dimers align with their ends in register to form covalently crosslinked tetramers. Tetramers associate via the globular domains to form microfilaments (Furthmayer 1983; Kuo et al. 1989, 1995). When reacted with MAb, microfilaments composed of Type VI collagen demonstrate a unique 100-nm periodicity in the electron microscope (Bruns et al. 1986; Keene et al. 1988). A recognizable periodic pattern is also seen after staining with RHT, whereby a thickened area represents globular domains and the thinner region represents helical portions. This pattern has been reported to be specific for Type VI collagen and is comparable to negative stain and rotary shadowed images (Keene et al. 1988; Keene et al. 1997). Type VI collagen microfilaments are believed to contribute to the organization and mechanical properties of the dermis by loosely enmeshing matrix components such as blood vessels, nerves, collagen fibrils, and cells while still allowing flexibility (Keene et al. 1988).
A mechanism for anchoring the Type VI collagen network to collagen fibrils has not been previously demonstrated in tissue. Significant in this study is the identification of a definitive site at which Type VI collagen microfibrils interact with the surface of banded fibrils, indicating that the role of Type VI in tissue integrity is more than passive. The location of this interactive site is near the “d” band of the collagen fibrils, which is demonstrated here and elsewhere to be the site of decorin binding.
Solid-phase binding assays have demonstrated an affinity between decorin and Type VI collagen (Bidanset et al. 1992), further suggesting that decorin mediates an attachment of Type VI microfilaments to banded collagen fibrils. It appears likely that the loss of potential decorin binding domains for Type VI collagen in the Dcn -/- mouse may contribute to the frail skin phenotype. However, we were surprised that a specific interaction of Type VI collagen and banded fibers persisted in the Dcn -/- mouse. Also surprising was the presence of small ruthenium-positive knobs in the Dcn -/- mouse, present at the same location as larger decorin-containing knobs on the surface of fibers in the wild-type mouse. This indicates that at least one other component, in addition to decorin, shares the site of Type VI interaction. This other component may substitute for decorin, as in the Dcn -/- mouse, during development and in tissue homeostasis. That this component is a proteoglycan is suggested by the mechanism of RHT staining, causing small proteoglycans to be retained in a precipitated form similar to that seen here (Hunziker et al. 1983). Given that chondroitinase, which cleaves carbohydrate groups in proteoglycans, disrupts the Type VI matrix in cornea (Nakamura et al. 1997) and that chondroitinase has been shown not to affect the binding of decorin to Type VI in solid-phase binding assays, this further evidence suggests that the unknown component is an SLRP. We conclude that the role of mediating an interaction between Type VI microfilaments and the surface of banded fibers does not uniquely belong to decorin.
Other small, leucine-rich proteoglycans are known to be present within the matrix including fibromodulin, biglycan, and lumican (Heinegård and Oldberg 1989, 1993). Among these, biglycan has been shown to interact with Type VI collagen. Some evidence places biglycan free in the matrix with no affinity for banded collagen fibrils (Brown and Vogel 1989; Roughley and Lee 1994), whereas other evidence suggests not only that biglycan and Type I collagen interact but that the binding domains of biglycan and decorin are very close (Schönherr et al. 1995). Fibromodulin has been shown to bind to Type I collagen and to regulate fibrillogenesis (Hedbom and Heinegård 1989), but is has been reported to bind a triple helical site separate from decorin (Hedbom and Heinegård 1993). Further studies are needed to define the identity of matrix components, in addition to decorin, that contribute to the anchoring of the Type VI collagen network in skin. We are continuing our efforts in this area to produce and characterize a mouse deficient in both decorin and biglycan. If successful, we hope to define the role of fibromodulin in anchoring the network of Type VI microfilaments.
Footnotes
Acknowledgements
Supported in part by grants from the Shriners Hospital for Children (DRK, CCR) and by National Institutes of Health grants RO1 CA39481 and RO1 CA47282 (RVI).
Electron microscopy facilities were provided in part by the Fred Meyer and R. Blaine Bramble Charitable Trust Foundations. We are deeply grateful to E. Engvall and L. Fisher for providing precious immunological reagents.