Collagen XII Interacts with Avian Tenascin-X through Its NC3 Domain*

Large oligomeric proteins often contain several binding sites for different molecules and can therefore induce formation of larger protein complexes. Collagen XII, a multidomain protein with a small collagenous region, interacts with fibrillar collagens through its C-terminal region. However, no interactions to other extracellular proteins have been identified involving the non-collagenous N-terminal NC3 domain. To further elucidate the components of protein complexes present close to collagen fibrils, different extracellular matrix proteins were tested for interaction in a solid phase assay. Binding to the NC3 domain of collagen XII was found for the avian homologue of tenascin-X that in humans is linked to Ehlers-Danlos disease. The binding was further characterized by surface plasmon resonance spectroscopy and supported by immunohistochemical co-localization in chick and mouse tissue. On the ultrastructural level, detection of collagen XII and tenascin-X by immunogold labeling confirmed this finding.

The integrity of extracellular matrix is maintained by supramolecular networks assembled by a large variety of matrix macromolecules. Among those is the group of 28 different collagen types so far described in the literature. Collagens are further subdivided into several families reflecting their assemblyforming properties (1). As a common feature, fibril-associated collagens with interrupted triple helices (FACITs) 2 comprise at least two collagenous domains interrupted by non-collagenous domains. Collagen XII is a member of the FACIT subfamily, and its three chains are encoded by a single gene. The two collagenous domains (Col1 and Col2) are interrupted and flanked by three non-collagenous domains (NC1-NC3), of which the large trimeric NC3 domain contains up to 90% of the molecular mass of collagen XII. The N-terminal NC3 part of each polypeptide chain consists of two to four von Willebrand factor type A domains, several fibronectin type III repeats, and a thrombospondin N-terminal domain (2,3). The size of the protein is variable due to two alternative splice sites located at the 5Ј-and 3Ј-ends of the collagen XII mRNA. Especially remarkable is the splicing mechanism that produces mRNA encoding for NC3 domains that differ ϳ100 kDa in mass, denoted as XIIA for the large form and as XIIB for the small form (4,5). The other alternative splicing generates two NC1 domains of similar size, termed -1 or -2, which results in the nomenclature of the four different collagen XII isoforms: XIIA-1, XIIA-2, XIIB-1 and XIIB-2 (for overview of the domain structure please refer to Fig. 2A) (6). The isoforms differ in their histological and developmental distribution: the large forms XIIA-1 and XIIA-2 are preferably expressed during embryonic stages, whereas the expression of the small forms persists in adult tissues (7). Furthermore, the small isoform XIIB-1 predominantly occurs in ligaments and tendons, whereas the XIIB-2 shows a more widespread expression (6). The isoforms also differ in their biochemical properties. The large forms contain an additional heparin binding site in the 7th fibronectin type III domain; in addition, covalently linked glycosaminoglycan chains are attached to them (4,8), whereas alternative splicing in the NC1 domain leads to an additional heparin binding site in the XIIA/ B-1 form (9). Collagen XII also interacts with decorin (10) and is a component of collagen I-containing fibrils (8,11).
Tenascins are a distinct family of extracellular matrix proteins with four members in vertebrates (TN-C, TN-R, TN-W, and TN-X). They all exhibit a similar domain structure with an N-terminal oligomerization domain, a series of epidermal growth factor modules followed by several fibronectin type III domains and a large globular fibrinogen-related domain (for review see Ref. 12). Mammalian tenascin-X can associate with collagen I-containing fibrils through binding to the glycosaminoglycan chains of the small fibril-associated proteoglycan, decorin (13)(14)(15). In addition, cells in culture adhere to tenascin-X through interaction with integrin receptors (16). The first indication for the function of tenascin-X in vivo was derived from the finding that human patients lacking the protein develop a connective tissue disorder, Ehlers-Danlos syndrome (17). Today, the suggestion is widely accepted that tenascin-X is a regulator of collagen deposition in vivo, consistent with a reduced density of collagen fibrils in the skin of tenascin-X null mice (18). However, the molecular mechanism behind these findings still remains elusive (19).
A closely related gene and protein have been identified in the chick (20,21). The avian protein is most homologous to mammalian tenascin-X in its C-terminal region and shares with it a serine/proline-rich domain but has smaller subunits with only one (instead of 18) complete epidermal growth factor repeats and with fibronectin type III modules not related in sequence to any other avian or vertebrate tenascin. At the time of discovery, these considerable differences justified a new name, tenascin-Y, for the avian protein (20 -22). However, recent phylogenetic studies of the tenascin gene family in chordates revealed that chick tenascin-Y belongs to the tenascin-X branch and that it is more closely related to mammalian than to Xenopus tenascin-X. 3 The human tenascin-X gene is located in the human MHCIII locus on chromosome 6q21.3, a locus that, according to comparative mapping information corresponds to chicken chromosome 16 (23). At this location, the chicken tenascin-Xgene is in immediate synteny with the complement C4, TAP1, TAP2, and MHC-II-like genes that represent orthologues of the genes present in the human MHCIII locus next to the human tenascin-X gene, proving a common evolutionary origin for this genomic region. 3 For these reasons, it is justified to abandon the former name tenascin-Y and to call the protein "avian tenascin-X" instead.
For collagen XII, the function as a modulator of tissue biomechanical properties by bridging collagen I-containing fibrils to other extracellular matrix components has been suggested (24,25). Our findings reported here support this concept. Avian tenascin-X interacts with the NC3 domain of avian collagen XII, thereby establishing a mechanical coherence of banded collagen fibrils with their extrafibrillar environment.

EXPERIMENTAL PROCEDURES
Antibodies and Antibody Production-Antibodies against mouse collagen XII were produced as described previously (26). Concentration dependence of the chick collagen XII-tenascin-X interaction determined in a solid phase binding assay and by surface plasmon resonance spectroscopy. A, different concentrations of tenascin-X were incubated with collagen XII or tenascin-C coated onto microtiter plates. Binding was revealed in an enzyme-linked immunosorbent assay-style manner with an affinity-purified polyclonal antibody against chick tenascin-X (KX8). The resulting saturation curve was used for calculation of an apparent K D value. B, chick tenascin-X was coated onto the plates, and different concentrations of soluble chick collagen XII were added. The bound collagen XII was detected with an affinity-purified antibody (522). ⌬E, measured extinction minus blank value. C and D, surface plasmon resonance spectroscopy was performed with tenascin-X (C) or collagen XII (D) as soluble analyte. The amount of interacting analyte (3, 10, 30, and 100 nM) was monitored by measuring the variation in the plasmon resonance angle as function of time and expressed in terms of response units. The background signal has been subtracted from each curve; the curves are shown in ascending order depending on the analyte concentration used. Fittings and overlay plots were done with the Biaevaluation software version 3.2. The continuous black lines represent the fitted curves.
Briefly, cDNA coding for the fibronectin type III domains 14 -18 of the mouse collagen XII was amplified by PCR using primers that introduced a NheI restriction site at the 5Ј-end and a BamHI site at the 3Ј-end (M193, forward, CACGCTAGCA-GAGGACTGTCAAGAAACATCC and M194, reverse, TTG-GGATCCTTAGGTCTGTTCTTTGATGGGGACA). The cDNA was cloned into a modified pPET (EMD Biosciences) vector carrying a His 6 tag with a thrombin cleavage site. Upon transformation with the recombinant plasmid, Escherichia coli cells (Bl21; EMD Biosciences) were induced with 1 mM isopropyl-1thio-␤-D-galactopyranoside and grown for 16 h at 30°C. The cells were harvested (15 min, 5000 ϫ g, 4°C) and resuspended in Tris-buffered saline, pH 8.0, containing 7 M urea. The bacteria were sonicated, followed by removal of insoluble cell debris by centrifugation (30 min, 20000 ϫ g, 4°C). After 2-fold dilu-tion with H 2 O the supernatant was applied to a nickel-chelating Sepharose column (GE Healthcare) and eluted with binding buffer containing 40 -80 mM imidazole. Following removal of urea by dialysis, thrombin cleavage was performed overnight at room temperature (5 mM CaCl 2 , 1 unit/mg thrombin; Sigma-Aldrich), and the cleaved His 6 tag was removed by passing the solution again over a nickel-chelating Sepharose column.
The recombinant protein was used to immunize a rabbit from which the antiserum was purified by affinity chromatography on a column with antigen coupled to CNBractivated Sepharose (GE Healthcare). The specific antibody, termed KR33, was eluted with 150 mM NaCl, 0.1 M triethylamine, pH 11.5, and the eluate neutralized with 1 M Tris-HCl, pH 6.8. Polyclonal rabbit antibody against chick collagen XII (522) (8), polyclonal chick antibody against mouse tenascin-X (KX3), polyclonal rabbit antibody against chick tenascin-X (KX8) (21,27), and polyclonal rabbit antibody against chick tenascin-C (TN474) (28) have been characterized before.
SDS-Polyacrylamide Gel Electrophoresis and Determination of Protein Concentration-SDS-polyacrylamid gel electrophoresis (SDS-PAGE) was performed as described by Laemmli (29). 1.3 g of protein/ lane was separated on 3-10% SDS-PAGE gradient gels under either reducing or non-reducing conditions and visualized with a silver staining kit (Invitrogen) according to the manufacturer's instructions. Protein concentrations were determined using a binicotinic acid assay kit (Uptima) following the manufacturer's instructions.
Solid Phase Binding Assay-The production of recombinant avian tenascin-X (formerly tenascin-Y) has been described previously (20). Tenascin-C and collagen XII were purified from chick embryo fibroblast-conditioned medium by monoclonal antibody affinity chromatography following established procedures (8,30). To calculate the molar concentrations of the proteins utilized for titration measurements, the molecular masses of the monomeric forms of avian tenascin-X (205 kDa), avian tenascin-C (200 kDa), and the mean value of the small and large splice variants of avian collagen XII, estimating a molar ratio of 1:1 (274.5 kDa), were used. Purified proteins were diluted in TBS, pH 7.4, and 5 g/ml (250 ng/well) were coated onto FIGURE 2. Domain structure (A) and SDS-PAGE analysis (B) of the applied proteins. A, a schematic depiction of the domain structures of the large (XIIA) and small (XIIB) variants of chick collagen XII, of mouse tenascin-X, of chick tenascin-X, and of chick tenascin-C. B, purified collagen XII (cCol XII) and tenascin-C (cTN-C) from chick fibroblasts as well as purified recombinant chick tenascin-X (cTN-X) were subjected to SDS-PAGE with (ϩDTT ) or without (ϪDTT ) dithiothreitol prior reduction. The 3-10% gradient gels were silver stained. 96-well plates (Nunc Maxisorb) at 4°C overnight. After washing with TBS, unspecific binding sites were blocked at room temperature with 5% skimmed milk powder in TBS for 2 h. Ligands were diluted in blocking buffer to concentrations from 0.03 to 300 nM for the tenascins or 0.023 to 230 nM for collagen XII and incubated for 1.5 h. For competition experiments the competitor was added to the ligand solution before incubation. After removing excess ligand by washing twice with TBS, bound ligand was fixed with 2.5% (v/v) glutaraldehyde for 10 min. Bound ligands were detected with specific affinity-purified antibodies, rabbit against chick collagen XII (522), chick tenascin-X (KX8), or chick tenascin-C (TN474) followed by swine anti-rabbit horseradish peroxidase-coupled IgG (Dako Cytomation). For enzymatic reaction, wells were incubated with 50 l of 0.25 mM tetramethylbenzidine and 0.005% (v/v) H 2 O 2 in 0.1 M sodium acetate, pH 6.0, for 10 min. The reaction was stopped with 50 ml/well 2.5 M H 2 SO 4 , and the absorbance was measured at 450 nm using a microplate reader (Labsystems Multiscan MS). For analysis, measurements of wells treated equally, except for the addition of ligand, were subtracted as blank values. All buffers contained 2 mM CaCl 2 .
Surface Plasmon Resonance Spectroscopy-Surface plasmon resonance spectroscopy was performed using a Biacore 2000 (BIAcore AB) system. Avian tenascin-X, fulllength avian collagen XII, and the purified collagen XII NC3 domain were coupled in 25 mM sodium acetate, pH 4.7, with a flow rate of 5 l/min to a CM5 chip. The chip was previously activated with N-hydroxysuccinimide and 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride. After coupling the required amount of protein (ϳ1000 response units), unbound reactive groups were saturated with 1 M ethanolamine hydrochloride, pH 8.5. Experiments were carried out using different concentrations (3, 10, 30, and 100 nM) of avian tenascin-X and full-length collagen XII diluted in running buffer (20 mM Hepes, 150 mM NaCl, 2 mM CaCl 2 , 0.005% P20). The analyte was passed over the sensor chip with a constant flow rate of 30 l/min for 120 s, and dissociation was measured over 350 s. Fittings of the data, overlay plots, and calculation of K D values were done with BIAevaluation software 3.2.
Collagenase Digestion-To generate avian collagen XII molecules consisting only of the trimeric NC3 domain, i.e. lacking their collagenous parts as well as the NC1 and NC2 domains, the protein was subjected to collagenase digestion. A solution of 77 nM collagen XII was incubated with 100 units/ml highly purified collagenase (CLSPA; Worthington Biochemicals) in TBS containing 5 mM CaCl 2 and 1 mM 4-(2aminoethyl)benzylsulfonyl fluoride (Roche Applied Science) for 4 h at 37°C. Collagenase-treated collagen XII was immediately used for SDS-PAGE and solid phase binding assay. Alternatively, the NC3 domain was separated from collagenase and digestion fragments by running the sample over a gel filtration column (Superose 12 HR 10/30; GE Healthcare).
Immunohistochemistry-Immunohistochemistry was performed on cryosections of embryonic chick (E14.5) and newborn mouse (P1). The frozen sections were preincubated in ice-cold methanol for 2 min, blocked for 1 h with 5% normal goat serum in phosphate-buffered saline containing 0.2% Tween 20, and incubated with the primary antibodies against collagen XII (KR33 or 522) overnight at 4°C followed by Cy TM 3-conjugated goat anti-rabbit IgG (Dako Cytomation). For co-staining of mouse sections a chick antibody against tenascin-X (KX3) was used followed by Cy TM 2-conjugated donkey anti-chick IgG (Dako Cytomation). For co-staining of chick sections, the 522 antibody was biotinylated using Biotin-X-NHS (Calbiochem) according to the manufacturer's instructions. Co-staining was accomplished using sequential incubation, first with KX8 antibody for detection of tenascin-X followed by incubation with Cy TM 2-conjugated affinity-purified Fab fragment goat anti-rabbit IgG (Dianova), and second with biotinylated 522 antibody followed by Cy TM 3-conjugated anti-biotin antibody (Sigma). Stained sections were analyzed, and pictures were taken with a confocal laser scanning microscope (Leica TCS SL) using two lasers in parallel with the excitation wavelengths 488 nm for Cy TM 2 and 543 nm for Cy TM 3.
Immunoelectron Microscopy-Fragments of native collagen fibrils were isolated from newborn mouse skin, placed on grids, immunostained with KR33, KX3, and colloidal gold-labeled secondary antibodies, and analyzed by transmission electron microscopy as previously described (31). Pre-Embedding immunogold labeling of mouse skin was carried out as previously described (32). 15-day embryonic chick skin was harvested fresh, rinsed briefly in Dulbecco's modification of Eagle's medium (DMEM), and then immersed in antibody diluted 1:5 in DMEM overnight at 4°C. The tissues were rinsed for several hours in DMEM and then immersed in appropriate 10-nm gold secondary conjugate (GAR G10; Amersham Biosciences) overnight at 4°C. Following an extensive rinse in DMEM the tissues were fixed in 1.5% glutaraldehyde/1.5% paraformaldehyde with 0.05% tannic acid, rinsed, and then post-fixed in 1% OsO 4 . Following dehydration in ethanol, the tissues were embedded in Spurr's epoxy and 80-nm-thick sections were contrasted with uranyl acetate and lead citrate and examined using a Philips 410 TEM.
Adult mouse skin was rinsed extensively in DMEM and then immersed overnight in antibody diluted 1:10 in DMEM. Following a 4-h rinse in DMEM, tissues were immersed in appropriate 1-nm gold secondary conjugate (Amersham Biosciences GAR G1 for type XII antibody or Aurion GACh ultrasmall for TN-X) and then rinsed extensively in DMEM followed by a brief rinse in phosphate-buffered saline. Gold particles were then enhanced using the Nanoprobes GEEM gold enhance kit. Briefly, tissue in buffer is chilled on ice, incubated on ice for 15 min in complete enhance solution, and then warmed quickly to 25°C and incubated for 5 min. Tissue is rinsed in ice-cold phosphate-buffered saline and then in 0.1 M cacodylate buffer. After gold enhancement, the tissue is fixed in 1.5% glutaraldehyde/1.5% paraformaldehyde containing 0.05% tannic acid and then dehydrated, embedded and stained, and observed as described above.

RESULTS
Interaction between Collagen XII and Tenascin-X-In a solid phase binding screen for possible interaction partners of avian collagen XII, we discovered avian tenascin-X as a possible candidate. Titration experiments showed saturable binding using either collagen XII or tenascin-X as the soluble ligand. Because of its similar domain structure (33), tenascin-C was used as a negative control (Fig. 1, A and B). To calculate the molar protein concentrations for the titration measurements, the molecular masses of the monomeric forms of avian tenascin-X (205 kDa) and a mean value of the small and large splice variants of avian collagen XII, estimating a molar ratio of 1:1, (274.5 kDa) were used. In the titration experiment, half-maximal saturation FIGURE 4. Tissue distribution in chick embryos of collagen XII and tenascin-X and co-localization of the two proteins. Immunohistochemistry was performed on frozen chick embryo leg sections (E14.5). Co-staining was accomplished using sequential incubations with the antibodies KX8 for detection of chick tenascin-X and 522 for localization of collagen XII. Collagen XII (red) is strongly expressed in the papillary dermis, subcutis, in the connective tissue surrounding blood vessels (A), and in epimysium as well as perimysium and tendon (B, C). Tenascin-X (green) is highly expressed in the reticular dermis and around blood vessels (A), subcutis, epimysium, and perimysium (B, C ). The two proteins co-localize (yellow) at the border between papillary and reticular dermis, in the subcutis, in the connective tissue surrounding blood vessels (A), and in the epimysium and the perimysium (B, C ). Bars, 200 m in panels A and B and 40 m in panel C. fe, feather bud; de, dermis; bv, blood vessels; sc, subcutis; ep, epimysium; pe, perimysium; tn, tendon.
was reached at a concentration of soluble collagen XII of 2.5 Ϯ 0.5 ϫ 10 Ϫ8 M. When tenascin-X was the soluble binding partner, half-maximal saturation was found at 1.5 Ϯ 0.5 ϫ 10 Ϫ8 M, i.e. at a closely comparable concentration. The small discrepancy might originate from different avidities of the oligomeric forms of the two proteins.
To validate the results, surface plasmon resonance spectroscopy was performed and the associations and dissociations of the obtained binding curves were analyzed in a Langmuir 1:1 binding model (Fig. 1, C and D). Apparent K D values for collagen XII as soluble interaction partner of 1.34 ϫ 10 Ϫ8 M and for soluble tenascin-X of 6.44 ϫ 10 Ϫ9 M were calculated, which is in good agreement with the solid phase binding data.
The purity of the isolated proteins was checked by reducing and non-reducing SDS-PAGE (Fig. 2B). Avian tenascin-X occurs in trimeric and hexameric forms (20) containing at least three or six binding sites for collagen XII. Upon reduction the oligomeric forms dissociate into monomers. A similar pattern can be observed for tenascin-C, whose splice variants result in three bands on reducing gels (28,33). Collagen XII can exist as homo-or heterotrimeric combinations of the small and the large splice variants, which represent the four major bands on non-reducing SDS-PAGE. Upon reduction, the polypeptide pattern is complex but entirely explainable (8). The trimeric forms dissociate into large and small monomers and into various non-reducible dimers. Additional bands on reducing SDS-PAGE are due to glycosaminoglycan side chains attached to the large isoform of collagen XII (4).
The NC3 Domain of Collagen XII Contains the Binding Site for Tenascin-X-Collagen XII interacts with collagen I-containing fibrils through its collagenous domain (8). Additionally, two interaction sites for heparin are known, one at the very end of a C-terminal splice variant (6,9) and the other within the 7th fibronectin type III domain of the large form of collagen XII (8). To exclude the heparin binding sites being involved in binding of collagen XII to tenascin-X, competition experiments were performed. Even in 130-fold molar excess, heparin had no effect on the binding of collagen XII to tenascin-X (Fig. 3A). To study the contribution of the collagenous region (consisting of domains NC1, Col1, NC2, and Col2) of collagen XII in the binding process, collagenase digestion was performed. Treatment of collagen XII with highly purified bacterial collagenase simplified the complex pattern in reducing SDS-PAGE to two major bands that represent the NC3 domains of the small and large splice variants of the protein (Fig. 3B). The NC3 domain was purified via gel filtration chromatography, and because the cysteines are not removed the NC3 remains a trimeric molecule. (Fig. 3B, lane 4). In solid phase binding assays, collagenase-treated collagen XII showed a similar interaction with immobilized tenascin-X compared with the intact protein (Fig. 3C). To confirm those results, the purified NC3 domains were measured in surface resonance spectroscopy interaction experiments; from the obtained curves the apparent K D for the binding of soluble tenascin-X to the fragment is 5.90 ϫ 10 Ϫ9 M (Fig. 3D), i.e. very similar to intact collagen XII (cf. Fig. 1D). Taken together these results indicate that the binding site(s) for tenascin-X reside within the NC3 domain of collagen XII.
Partial Co-distribution of Collagen XII with Tenascin-X in Chick Skin and Muscle-The extracellular matrix of skeletal muscle is arranged in three levels: the individual muscle fibers are surrounded by the endomysium, bundles of muscle fibers are encased by the perimysium, and the complete muscle is embedded in the epimysium. It is known that chick tenascin-X is a component of muscle extracellular matrix and is primarily situated in the epimysium and perimysium of developing muscles, whereas tendons are negative for tenascin-X, both at the mRNA and the protein level (20,21). Collagen XII is a component of the skeletal muscle extracellular matrix as well. The protein and the mRNA are located in the epimysium, perimysium, and unlike tenascin-X, also in tendon (8,34). By double immunofluorescence we could show that the two proteins colocalize, seemingly on the same interstitial fibrils, in the epimysium and perimysium of developing chick muscles, whereas tendon is only positive for collagen XII (Fig. 4, B and C).
In the developing chick skin, tenascin-X is located throughout the dermis with increasing concentration in its lower parts and in the connective tissue layers surrounding small blood vessels in the dermis (20,21). In contrast, collagen XII is more concentrated in the upper part of the dermis, in the subcutis, and also in the wall of blood vessels containing smooth muscle cells (4,8). Overlay of immunofluorescence stainings of the two proteins revealed co-localization in the subcutis and in a defined zone of the middle dermis layer (Fig. 4A).
To analyze the less well characterized spatial expression of collagen XII in mouse embryos, we raised the polyclonal antibody KR33, which recognizes both the small and large splice forms of mouse collagen XII. We found that collagen XII is expressed throughout the developing dermis with increasing concentrations in the subepithelial layer. In the developing mouse skeletal muscle, collagen XII is situated in the epimy-sium and perimysium, but the endomysium is only weakly positive for collagen XII (Fig. 5, A-C). Similar to the findings in chick (34), there is a strong expression of collagen XII in tendon, perichondrium, and periosteum (35). Tenascin-X is expressed throughout the dermis of mouse embryos, and the staining decreases toward the epidermis. In mouse skeletal muscle, tenascin-X is present in the connective tissue along the muscle fibers (27). Overlay analysis revealed a co-localization of collagen XII with tenascin-X in the epimysium and perimysium of the developing skeletal muscle (Fig. 5, B and C) and in a broad but defined zone of the reticular dermis (Fig. 5A). Taken together, the spatial organization of collagen XII and tenascin-X in mouse embryos resembles closely the distribution of the two proteins in chick with well defined areas of co-localization.
Ultrastructural Analysis of the Spatial Expression of Collagen XII and Tenascin-X-By immunofluorescence staining, the fibrillar distribution pattern for collagen XII and tenascin-X is clearly visible. Based on ultrastructural analysis, a fibril-associated localization has been suggested for both proteins (8,13). On ultrathin immunogold-labeled sections of the embryonic chick skin, collagen XII and tenascin-X were located directly on interstitial fibrils or in electron-dense material attached to the fibrils in the same area of the dermis (Fig. 6, A and B). We made the same observation for collagen XII and tenascin-X in the mouse dermis (Fig. 6, C and D), whereas, consistent with immunofluorescence analysis, chick tendon showed only a fibril-associated labeling for collagen XII but not for tenascin-X (data not shown).
To further support the co-localization of both proteins in the same ultrastructure, double labeling for collagen XII and tenascin-X was performed on fragments of matrix suprastructures extracted from newborn mouse skin under native conditions. Immunogold labels for both proteins were clearly located together on electron-dense material. A gallery of such mats together with a larger overview of doubly labeled suprastructures is shown in Fig. 7. 223 gold particles corresponding to an immunolabel for collagen XII (18 nm gold particles) were arbitrarily chosen, and the distances to the nearest particle indicating tenascin-X-labeling (12 nm gold particles) were determined. Co-localization was considered to occur if the gold particles could be identified within the same electron-dense structure and if the distance between collagen XII and tenascin-X-gold labels was less than 100 nm. This distance is comparable with the dimensions of the non-collagenous domains of collagen XII. By these criteria, 82% of collagen XII labels were associated with those of tenascin-X. However, tenascin-X labeling also occurred in structures that apparently lacked collagen XII, indicating that tenascin-X also occurred in suprastructures in which collagen XII had no part.

DISCUSSION
Collagen XII belongs to the subfamily of collagens designated as FACITs. It has been proposed that FACITs are surface components of banded collagen fibrils with their C-terminal, collagenous FACIT domains anchored into the fibril body. In this model (25), exemplified by the schematic drawing shown in Fig.  8, their N-terminal domains can reach out into the perifibrillar matrix to interact with other extracellular matrix components and cell surfaces.
Collagen IX, another FACIT, appears to form such bridges between collagen II-containing fibrils and non-collagenous cartilage matrix components, including matrilins and COMP (36 -39). Also, there is evidence that collagen XII too can bind to interstitial fibrils through its collagenous domains (8,11). In addition, a glycosaminoglycan-dependent binding of decorin to collagen XII outside the NC3 domain was shown (10). However, for the large NC3 domain, which comprises up to 90% of the molecular mass, no interaction partners other than heparin have been described (8). This large multidomain part of collagen XII consists of fibronectin type III domains, von Willebrand factor A domains, and a single thrombospondin N-terminal domain. von Willebrand factor type A domains, which are found in a variety of proteins, are known to mediate protein-protein interactions. For example, the von Willebrand factor type A domains present in the integrin ␣1␤1 and ␣2␤1 receptors are responsible for the interaction with fibrillar collagens (40).
The present study describes a novel interaction between collagen XII and tenascin-X, which is saturable and has an apparent K D in the range of 10 nM. This binding is not sensitive to treatment of collagen XII with collagenase, and for the purified NC3 domain similar binding parameters were obtained compared with the fulllength molecule. Therefore the binding site is located within the non-collagenous NC3 domain. Previously, a heparin binding site was located in the 7th fibronectin type III domain of the large form of collagen XII (8). However, the heparin binding site, saturated with soluble heparin, does not influence the binding to tenascin-X and, therefore, does not overlap with the tenascin-X binding region in the NC3 domain.
Tenascin-X also binds to heparin (14,21,27), and binding to collagen I-containing fibrils was demonstrated (41). Furthermore, in a different study this interaction was suggested to be mediated by decorin (15). In addition, evidence for tena- FIGURE 7. Immunoelectronmicroscopic analysis of collagen I-containing fibril fragments extracted from mouse skin. Native fibril fragments were extracted from mouse dermis and doubly labeled for collagen XII and for tenascin-X with specific antibodies followed by colloidal gold-labeled secondary antibodies (particle size 18 nm for collagen XII and 12 nm for tenascin-X). Bar is 150 nm. FIGURE 8. Model for the supramolecular assembly of interstitial fibrils. Collagen XII and tenascin-X act as adaptor molecules to interconnect collagen I-containing fibrils with each other and therefore might modulate the distances between fibrils. This would in turn influence the biomechanical properties of tissues. Collagen XII directly or indirectly interacts with the fibrils, whereas in the case of tenascin-X the interaction is mediated by decorin. The binding of glycosaminoglycans (GAGs) to both proteins suggests the involvement of other potential interaction partners. In addition, tenascin-X is able to bridge the extracellular network to cell surfaces due to the interaction with integrin receptors.
scin-X mediating cell attachment through binding to integrin receptors was presented (16).
Taking published information together with the new interaction studies presented here results in a more comprehensive picture of the protein network associated with fibrils containing collagens I and XII, which presumably contributes to the integrity of the extracellular matrix and facilitates cellular interactions. The components involved in these interactions are schematically visualized in Fig. 8.
Immunofluorescence analysis of tissue sections shows a partial co-localization of collagen XII with tenascin-X. Especially in the reticular dermis and in the epimysium and perimysium of the muscle exact co-staining of both proteins was seen, whereas in other tissues predominantly one of the proteins is present, e.g. collagen XII in the papillary dermis and in tendon. These results were confirmed by ultrastructural analysis, which revealed association of both proteins with banded fibrils in the dermis and demonstrated co-localization on fibrils extracted from mouse dermis. In particular, collagen fibrils are tightly surrounded by a remarkable, electron-dense material, which could be labeled for collagen XII as well as for tenascin-X and might represent complexes of several proteins held together by the numerous interactions as indicated in Fig. 8.
In mice lacking tenascin-X as well as in patients suffering from Ehlers-Danlos Syndrome because of the absence of tenascin-X, it has been shown that there is less collagen I in the skin and that the packing of the fibrils is less tight (18). Therefore, it has been suggested that tenascin-X regulates collagen fibril deposition and spacing through multiple interactions with collagen I or interactions with other proteins that in turn themselves interact with collagen I fibrils (19). In the absence of tenascin-X, as in the aforementioned cases, the biomechanical tissue stability is compromised because crucial interactions between fibrils and the perifibrillar matrix are abrogated. Interestingly, fibroblasts derived from tenascin-X Ϫ/Ϫ mice show a decreased collagen XII mRNA expression level compared with wild-type fibroblasts (42).
Recently, it has been demonstrated that collagen IX, the FACIT prototype, mediates the interactions between fibroblasts and banded collagen fibers through integrins (43). For collagen XII, no direct interaction with cell surface receptors could be shown so far. Because of the integration of collagen XII into protein complexes linking the interstitial fibrils to other extracellular matrix components and cell surfaces, collagen XII might function as a modulator of tissue biomechanical properties (24). It is tempting to speculate that the restricted spatial co-localization of collagen XII and tenascin-X plays a role in generating specific mechanical properties in those tissues through collagen XII acting in concert with tenascin-X as a regulator of fibril deposition and spacing. Further studies are necessary to elucidate the composition and function of such protein complexes.