|Home | About | Journals | Submit | Contact Us | Français|
Collagen fibers expose distinct domains allowing for specific interactions with other extracellular matrix proteins and cells. To investigate putative collagen domains that govern integrin αVβ3-mediated cellular interactions with native collagen fibers we took advantage of the streptococcal protein CNE that bound native fibrillar collagens. CNE specifically inhibited αVβ3-dependent cell-mediated collagen gel contraction, PDGF BB-induced and αVβ3-mediated adhesion of cells, and binding of fibronectin to native collagen. Using a Toolkit composed of overlapping, 27-residue triple helical segments of collagen type II, two CNE-binding sites present in peptides II-1 and II-44 were identified. These peptides lack the major binding site for collagen-binding β1 integrins, defined by the peptide GFOGER. Peptide II-44 corresponds to a region of collagen known to bind collagenases, discoidin domain receptor 2, SPARC (osteonectin), and fibronectin. In addition to binding fibronectin, peptide II-44 but not II-1 inhibited αVβ3-mediated collagen gel contraction and, when immobilized on plastic, supported adhesion of cells. Reduction of fibronectin expression by siRNA reduced PDGF BB-induced αVβ3-mediated contraction. Reconstitution of collagen types I and II gels in the presence of CNE reduced collagen fibril diameters and fibril melting temperatures. Our data indicate that contraction proceeded through an indirect mechanism involving binding of cell-produced fibronectin to the collagen fibers. Furthermore, our data show that cell-mediated collagen gel contraction does not directly depend on the process of fibril formation.
Collagen fibrils packed in the quarter-staggered fashion expose specific domains that specifically interact with other molecules or molecular assemblies of the interstitial matrix, or with cells (reviewed in Ref. 1). These domains, summarized by Sweeney et al. (2), are reflected in specific binding between, on the one hand, the constituent tropocollagen monomers and, on the other, collagen receptors and other extracellular matrix (ECM)2 components. The introduction of Toolkits of defined synthetic triple helical peptides covering the Col1 domains of collagen types II and III, including hydroxylated proline residues, has enabled the identification of collagenous motifs that interact with other ECM proteins and cells (3).
Interstitial fluid pressure (IFP) plays an important role in control of tissue fluid homeostasis (4). Lowering of IFP occurs during acute inflammation or anaphylaxis and contributes to formation of edema (5). Cell-mediated collagen gel contraction has been used as an in vitro model for studying control of IFP in loose connective tissues (6,–8), but also for wound contraction (9). Several substances that stimulate collagen gel contraction in vitro also increase IFP in vivo and, conversely, substances that inhibit contraction lower IFP. Cell-mediated collagen gel contraction can be mediated by collagen-binding β1 integrins (9,–12) and the collagen-binding integrin α2β1 is of particular importance for control of IFP in rat dermis during homeostasis (13). Contraction by cells lacking collagen-binding β1 integrins, e.g. cells from the murine myoblast cell line C2C12, is induced by PDGF-BB and uses the αVβ3 integrin to contract collagen gels (14,–16). This is paralleled in vivo by the observation that PDGF-BB and insulin normalize IFP that has been lowered as a result of mast cell degranulation by a process dependent on β3 integrins (16, 17). Available data suggest that collagen-binding β1 integrins are involved in control of IFP and thereby fluid volume during homeostasis, whereas integrin αVβ3-mediated contractions are involved in IFP control during inflammatory processes.
Streptococcus equi subspecies equi (S. equi) causes a serious and highly contagious disease in the upper respiratory tract of horses. Cells of S. equi grown in vitro express collagen-binding activity, and a collagen-binding protein called CNE, displaying typical features of a cell surface-anchored protein, has previously been isolated and characterized (18). We have previously shown that among a set of ECM-binding proteins from S. equi the collagen- and fibronectin-binding protein FNE modulates collagen gel contraction (8). FNE, which is secreted, stimulates collagen gel contraction and normalizes IFP lowered as a result of anaphylaxis. The mechanism by which FNE stimulates collagen gel contraction involves binding of fibronectin to collagen fibers and subsequent adhesion of cells to the complex by a mechanism dependent on the integrin αVβ3 (8). As determined by rotary shadowing, FNE binds collagen type I at a region located around 120 nm from the C terminus and therefore presents a high affinity, indirect binding site for fibronectin on the collagen fiber at a domain of collagen that is not known to interact with fibronectin (8). In the current study we have explored the potential to use the collagen-binding streptococcal protein CNE in combination with Toolkits to delineate domains in collagen fibers and molecular mechanisms that are operative in αVβ3-mediated contraction.
The murine myoblasts C2C12 were provided by Dr. Anna Starzinski-Powitz (Goethe-Universitaet, Frankfurt am Main, Germany). These cells lack expression of collagen-binding integrins but express the β1-integrin subunit. C2C12 cells stably expressing human α2-integrin subunit have been described before (19). Human diploid AG1518 skin fibroblasts (Genetic Mutant Cell Repository, Camden, NJ) were used between passages 18 and 24. Cells were propagated in DMEM with Glutamax (Invitrogen) supplemented with 10% fetal bovine serum (FBS) (PAA Laboratories GmbH, Pasching, Austria) and 50 μg/ml of gentamycin (Invitrogen) at 37 °C and 5% CO2. Preparation of recombinant protein CNE has previously been described (18). Recombinant PDGF-BB was purchased from Invitrogen. The specific αV-integrin inhibitor cyclo-(Arg-Gly-Asp-Phe-Val) was obtained from Bachem (Bubendorf, Switzerland). Bovine dermal collagen type I (Purecol, 3 mg/ml) was from Inamed (Fremont, CA). Pepsin-solubilized calf nasal collagen type II, rat skin collagen type III, and Engelbreth-Holm-Swarm-sarcoma collagen type IV were produced according to established protocols (20,–22). Human plasma fibronectin and vitronectin were purified as described (23, 24). Rabbit anti-PDGFR-β and rabbit anti-phosphotyrosine IgG were from Santa Cruz (Santa Cruz Biotechnology, CA). The rabbit anti-integrin β1 IgG has been described elsewhere (25). Anti-mouse integrin β3-subunit (HMβ3), anti-mouse integrin α5-subunit (MFR5), and FITC-labeled anti-hamster IgGs were from BD Pharmingen (San Jose, CA). FITC-labeled anti-rabbit IgG and anti-rat IgG were from Vector Laboratories (Burlingame, CA). 1-Hydroxysuccinimide biotin ester was purchased from Sigma and PD-10 columns from GE Healthcare. Avidin-conjugated alkaline-phosphatase and streptavidin-conjugated horseradish peroxidase (HRP) were from Vector Laboratories and AnaSpec, respectively. Tetramethylbenzidine substrate for HRP was from Pierce. Receptor tyrosine kinase inhibitor AG1296 and MMP inhibitor GM6001 were from Merck (Merck, NJ), and both were used at 10 μm. Recombinant human pro-MMP-1 and aminophenyl mercury acetate were from Sigma (Sigma). Coomassie Brilliant Blue G-250 was from Roth (Karlsruhe, Germany). The anti-human fibronectin polyclonal IgG has been described elsewhere (10). siRNA directed against murine fibronectin was from Sigma-Proligo and the transfection agent N-TER Nanoparticles was from Sigma.
Microtiter plates were coated with fibronectin (10 μg/ml), native or denatured (56 °C for 30 min), collagen types I, II, III, or IV (50 μg/ml) or collagen type II triple helical peptides (10 μg/ml) and incubated overnight at 4 °C or, to prevent renaturation of collagens, at 56 °C. Plates were incubated overnight with 2% BSA at 4 or 56 °C to prevent unspecific binding. Proteins were biotinylated with 1-hydroxysuccinimide biotin ester overnight at 4 °C, followed by a single desalting step with PD-10 columns in PBS with 0.1% BSA and 0.02% azide as preservative. Biotinylated proteins were diluted in 0.5% BSA in PBS and incubated in the coated wells for 2 h at 37 °C. After three washes with PBS supplemented with 0.05% Tween 20 (PBS-Tween), plates were incubated with avidin-conjugated alkaline phosphatase (1:500) for 2 h at room temperature. Plates were washed three times with PBS-Tween and developed with p-nitrophenyl phosphatase substrate (Sigma) (0.6 mg/ml in ethanolamine solution, pH 9.8) at 37 °C until A405 was between 0.1 and 1.0 absorbance units.
The triple helical collagen type II 27-amino acid peptide library has been described previously (26). Microtiter plates were coated with triple helical peptides at 10 μg/ml in a volume of 100 μl/well. Plates were blocked with 10% BSA in Tris-buffered saline (TBS) for 1 h at room temperature. Plates were washed three times with TBS substituted with 0.1% Tween 20 (TBST). Wells were incubated with 75 nm CNE-biotin (assuming 70% recovery from PD-10 column) in 100 μl of PBS without CaCl2 and MgCl2 (pH 7.4) for 1 h at room temperature. Wells were washed three times with TBST before incubation with streptavidin-HRP (1:5000 in TBS) for 1 h at room temperature. Wells were washed 3 times and developed by adding 100 μl of tetramethylbenzidine substrate. Adding 100 μl of 2.5 m H2SO4 stopped development and wells were measured at A450.
Collagen gel contraction was performed and quantified as described elsewhere (10). Briefly, 96-well plates were blocked in 2% BSA and a collagen solution was prepared from 2× DMEM, HEPES, and collagen types I or II. One part cell suspension (106 cells/ml) was mixed with 9 parts collagen solution. When indicated, recombinant proteins, antibodies, and/or inhibitors were added to the cell-collagen solution. Cell-collagen gels (100 μl) were allowed to form and were subsequently detached by ejection of 100 μl of DMEM in the absence or presence of 40 ng/ml of PDGF-BB into the wells. The relaxed, free-floating gels were further incubated at 37 °C and gel diameters were measured microscopically at the indicated time points.
Recombinant human pro-MMP-1 was activated with 1 mm aminophenyl mercury acetate in TCNB buffer (50 mm Tris, 10 mm CaCl2, 150 mm NaCl, 0.05% Brij-35, pH 7.5). 10 μg of triple helical collagen type I was preincubated in the presence or absence of 2.2 μg of CNE (dissolved in PBS with 0.5% BSA) for 1 h at 37 °C (equimolar ratio). Preincubated collagen was incubated for the indicated times with 25 ng of activated MMP-1. Digestion reactions were terminated by adding 4× SDS sample buffer. Protein samples were separated with SDS-PAGE and gels were stained with Coomassie Brilliant Blue to detect banding patterns.
24-Well plates were coated overnight at 4 °C, or, to prevent re-naturation of collagen, at 56 °C with native or heat-denatured collagen type I (50 μg/ml), vitronectin (10 μg/ml), or Toolkit II peptides (10 μg/ml) in Buffer 3 (140 mm NaCl, 4.7 mm KCl, 0.65 mm MgSO4, 1.2 mm CaCl2, 10 mm HEPES, pH 7.4). Plates were washed three times in the same buffer and cells were diluted to 125,000 cells/ml. 50,000 cells were allowed to adhere at 37 °C and where indicated PDGF-BB and/or CNE were added together with the cells at concentrations of 20 ng/ml and 350 nm, respectively. Non-adherent cells were removed and the wells were gently washed twice with pre-warmed Buffer 3. The relative amount of adhered cells was quantified using a hexosaminidase assay, as described previously (27).
Fibrillogenesis of pepsin-extracted bovine skin collagen type I and nasal cartilage collagen type II was monitored by change in turbidity at 400 nm at 4-min intervals. Fibrillogenesis was monitored for 720 min. Four times concentrated buffer (80 mm HEPES, 0.6 m NaCl, pH 7.4), CNE, and 0.012 m NaOH, in a volume equal to neutralize the collagen solution, were mixed, and water was added to a final volume of 238 ml. Twelve ml of a solution of collagen types I or II, purified after pepsin digestion in either 0.012 m HCl or 0.1 m acetic acid was added, yielding a final collagen concentration of 144 μg/ml. Collagen was added just prior to initiation of absorbance readings. The cuvettes were placed in a Beckman DU640 scanning spectrophotometer with a temperature controlled six-place cuvette chamber equilibrated to 37 °C.
Scanning electron microscopy (EM) was performed on collagen gels with or without cells prepared as described above. Gels were dehydrated, critical-point dried, gold-sputtered, and analyzed in a PHILIPS 515 electron microscope. Fibril diameter was quantified with ImageJ software (NIH). Differential scanning calorimetry measurements were performed on collagen fibrils formed in vitro, in the presence or absence of CNE, for 5 h in 37 °C in PBS (molar ratio collagen:CNE was 10:1). The differential scanning calorimetry thermograms were recorded in VP-DSC (MicroCal), at a scan rate of 0.5 °C/min, and medium feedback. Each thermogram was corrected by subtraction of a linear baseline based on a blank buffer sample, and normalized for collagen concentration.
C2C12 cells were seeded in 24-well plates at 90,000 cells per well and grown for 24 h in antibiotic-free DMEM supplemented with 10% FBS. Cells were transfected with a final concentration of 20 nm siRNA directed against murine fibronectin mRNA (PubMed accession number NM_010233) or with control siRNA that has no binding interaction with any known mRNA. A second control consisted of cells that were exposed only to the transfection agent, N-TERTM Nanoparticles, which were used according to the manufacturer's instructions. Cells were harvested at 24 h and fibronectin protein levels were assessed by separating equal amounts of cleared cell lysates with SDS-PAGE and Western blotting with rabbit anti-fibronectin polyclonal IgG. Transfected cells were used in collagen gel contraction assays.
C2C12 cells were seeded in 6-well plates at a density of 500,000/well. Cells were allowed to spread and then serum starved in DMEM with 0.1% FBS for 12–18 h. Cells were pre-treated with 350 nm CNE in DMEM for 2 h and subsequently stimulated with PDGF-BB (20 ng/ml) for 10 min in the presence of 350 nm fresh CNE. Wells were washed twice in ice-cold PBS and lysed on ice in solubilization buffer (50 mm Tris, 150 mm NaCl, 2 mm EGTA, 1 mm Na3VO4, 1% Nonidet P-40, 0.25% sodium deoxycholate and protease inhibitors). Supernatants were pre-cleared with normal rabbit IgG for 1 h at 4 °C. PDGFRβ was immunoprecipitated with 5–10 μg of rabbit anti-PDGFR-β for 1.5 h. Proteins were separated on 7.5% polyacrylamide gels, transferred to nitrocellulose, and blocked in 5% BSA overnight at 4 °C. Membranes were probed with rabbit anti-phosphotyrosine IgG (1:1000) and HRP-labeled donkey anti-rabbit IgG (1:5000), and protein bands were visualized with luminol.
Cells were trypsinized and washed twice with PBS. 500,000 cells were resuspended in 50 μl of primary antibody (10 μg/ml) diluted in 0.5% BSA in PBS together with 10 μg/ml of normal IgG of the same origin as the secondary antibody and incubated on ice for 1 h. Cells were washed two times in cold PBS and resuspended in 50 μl of secondary antibody (diluted 1:50 in 0.5% BSA in PBS) followed by a 30-min incubation on ice. After washing, cells were resuspended in 0.5% BSA in PBS and cell-bound antibodies were detected in a BD Biosciences FACS scan.
C2C12 cells lack collagen-binding β1 integrins but express other β1 integrins such as the fibronectin-binding integrin α5β1. PDGF-BB induces integrin αVβ3-dependent collagen gel contraction by these cells (8). In agreement with these earlier reports, C2C12 cells contracted collagen gels only after stimulation by PDGF-BB (Fig. 1, upper panel). C2C12 cells with forced expression of α2β1, after transfecting the cells with full-length human α2-integrin subunit (C2C12-α2), efficiently contracted collagen lattices, even in the absence of external stimuli (Fig. 1, lower panel). During initial experiments with a panel of recombinant streptococcal proteins (data not shown), only protein CNE was found to inhibit αVβ3-mediated collagen gel contraction. At a final concentration of 350 nm, CNE inhibited PDGF BB-induced contraction of C2C12 cells by an average of 86 ± 5% (Fig. 1, middle panel), but had no inhibitory effect on contraction mediated by C2C12-α2 cells (Fig. 1, lower panel). AG1518 fibroblasts normally contract collagen lattices by using collagen-binding β1 integrins but not αVβ3 (supplemental Fig. S1). The monoclonal anti-human β1 integrin antibody M13, which blocks β1 integrin function, inhibited AG1518 human fibroblast-mediated collagen gel contraction, an effect that could be overcome by addition of PDGF-BB. This effect of PDGF-BB was in turn dependent on αVβ3 integrin, because the contraction was blocked by a cyclic RGD peptide, which specifically inhibits αVβ3 integrin function at the concentrations used here. Similarly, in the presence of M13, the effect of PDGF-BB was abolished by addition of CNE. However, addition of CNE had no effect on control contraction in the absence of M13 (supplemental Fig. S1, A and B). When taken together, our data show that CNE specifically inhibited αVβ3 integrin-mediated collagen gel contraction but had no effect on collagen-binding β1 integrin-mediated contraction.
Recombinant streptococcal CNE bound to native interstitial collagen types I, II, and III but not to collagen type IV in solid phase assays (Fig. 2A). In these assays, collagens were coated at neutral pH and 37 °C allowing for fibril formation in the plates. Denaturation of the collagens by heating to 56 °C at neutral pH reduced CNE affinity below the detection limit (Fig. 2B). Average avidities for the binding of CNE to the various collagen fibrils were estimated to ~125 nm for collagen type I, ~50 nm for collagen type II, and ~100 nm for collagen type III based on data from the solid phase experiments. These findings show that CNE only bound to native triple helical collagen chains.
Because CNE bound native collagens, it could possibly interfere with the formation of collagen fibrils, thereby changing the biomechanical properties of collagen gels and in such a manner hamper the contractibility of the collagen matix. We tested the ability of CNE to inhibit fibril formation of collagen types I and II by monitoring change in turbidity. CNE effectively inhibited fibril formation in a dose-dependent fashion (Fig. 3, A and B). Furthermore, we tested the ability of CNE to reduce fibril diameter of collagen type I and II fibers reconstituted in vitro. As revealed by scanning electron microscopy (Fig. 3, C–F), CNE reduced the average fibril diameter from 130 to 90 nm (Fig. 3, G and H). Because collagen fibril formation is influenced by CNE in vitro, we also analyzed its effect on collagen denaturation. After incubating collagen types I or II with CNE and allowing fibrils to form in vitro, the samples were run in a differential scanning calorimeter to determine the collagen denaturation curves. During denaturation, two melting peaks were produced; the early peak corresponds to denaturing free collagen monomers (triple helices), and the later peak due to denaturation of the collagen fibrils. Addition of CNE did not affect denaturation of free collagen monomers at 40 °C but lowered the melting point of fibrils by about 5 °C from 50 to 45 °C for collagen type I (Fig. 3I) and about 2.5 °C from 50 to 47.5 °C for collagen type II (Fig. 3J). These findings demonstrate that CNE could modulate the biomechanical properties of collagen gels.
We investigated the possibility that PDGF-BB acted on MMPs that cleave a specific site in the collagen triple helix, thereby relaxing the triple helix that can result in exposure of nearby RGD sequences in a conformation that could be recognized by αVβ3. In such a scenario, CNE could exert its effects by inhibiting MMP-induced exposure of RGD sequences and subsequent abrogation of contraction. Therefore, we first tested whether CNE could inhibit enzymatic digestion of collagen by MMPs. Using zymography, we found no inhibitory effect of CNE on MMP-1 collagenase activity at the CNE to collagen ratio used here, 1:1 (Fig. 4A). To further test whether MMP activity affected PDGF BB-induced αVβ3-mediated contraction we made use of the MMP inhibitor GM6001 that selectively blocks activities of MMP-1, -2, -3, -8, and -9 or MMP inhibitor III that selectively blocks activities of MMP-1, -2, -3, -7, and -13. Neither of the inhibitors blocked PDGF BB-induced αVβ3 integrin-mediated contraction (Fig. 4B and data not shown). However, GM6001 potentiated the PDGF BB-induced αVβ3 integrin-mediated contraction, suggesting that MMPs have an inhibitory rather than stimulatory effect on collagen gel contraction.
We investigated the possibility that CNE exerts its inhibitory action on αVβ3–mediated collagen gel contraction through interference with the function of integrins, e.g. by blocking ligand-binding sites on these cell adhesion receptors. C2C12 and C2C12-α2 cells were seeded on plastic dishes coated with vitronectin, heat-denatured or native collagen type I in the presence or absence of CNE. C2C12 cells adhered to heat-denatured collagen and vitronectin, but not significantly to native collagen type I within a 30-min time frame. Addition of CNE at a concentration of 350 nm had no inhibitory effect on adhesion to heat-denatured collagen or vitronectin (Fig. 5A, upper panel). These findings demonstrate that CNE did not inhibit the function of αVβ3. As expected, C2C12-α2 cells adhered effectively to native collagen type I, as well as to vitronectin and heat-denatured collagen (Fig. 5A, lower panel). Adhesion of C2C12-α2 cells to any of the tested ligands was not affected by the presence of CNE (Fig. 5A, lower panel). Upon longer incubation of C2C12 cells on native collagen type I (>2 h), cells started to adhere (Fig. 5B). Interestingly and in analogy to collagen gel contraction, PDGF-BB significantly stimulated the latter process and this stimulatory effect was reduced to control levels when CNE, at a concentration of 350 nm, was present in the incubation medium (Fig. 5B). These data show that CNE did not interfere with the functionality of the investigated integrin but with a PDGF BB-stimulated process that possibly involves de novo protein synthesis.
So far, we reported on the effects of CNE on modulation of collagen gel contraction brought about by integrin αVβ3 and modulation of biomechanical properties of collagen gels. To study the underlying mechanisms of these two distinct actions it was important to identify binding sites in collagen for CNE. For this purpose we took advantage of Toolkit II, comprising the full-length human collagen type II sequence, consisting of synthetic triple helix peptides 27 amino acids in length (26). The last nine amino acids of each peptide overlap with the first nine of the next peptide so that the middle nine amino acids are unique sequences. Both ends of each peptide comprise GPP pentamers (GPP5) to induce triple helical folding of the insert. In this approach 56 peptides were created that comprise Toolkit II. First we established that C2C12-mediated contraction of collagen type II gels shared characteristics with contraction of collagen type I gels. PDGF-BB-induced contraction of collagen type II gels and this effect was abolished by CNE (Fig. 6A). CNE bound 2 different peptides, peptides II-1 and II-44 from Toolkit II with a signal to noise ratio above 3 (Fig. 6B). Several II-44 variant peptides (amino acid substitutions and shorter variants of the peptide) bound CNE, even with distinct and non-overlapping sequences (data not shown), suggesting either two different binding sites within peptide II-44, or a lack of requirement for all the amino acids in the sequence. This means that binding would be dependent on either specific structure- and/or charge-dependent features of the amino acid sequence in triple helical collagens. To further establish whether peptides II-1 and/or II-44 contain domains crucial for C2C12 cell binding, adhesion assays were performed on plates coated with peptide II-1, II-44, or control peptide GPP10. C2C12 cells bound effectively to peptide II-44, whereas binding to peptide II-1 was of similar low magnitude as binding to GPP10 (Fig. 6C). Because contraction of collagen type I and II gels were similarly modulated by PDGF-BB and CNE, and because the α1(I) collagen chain of collagen type I has a significant amino acid sequence homology with type II, we reasoned that the CNE-binding sites in collagen type I might be translated from the Toolkit II peptides. The bovine α1(II) collagen is, respectively, 96 and 100% identical with human α1(II) collagen in the amino acid sequences that encompass Toolkit peptides II-1 and II-44. The bovine collagen type I that has been used in this study has a high amino acid sequence homology with human collagen type II. Indeed, the sequence homology between bovine α1(I) collagen and human α1(II) collagen corresponding to peptides II-1 and II-44 are 85 and 93%, respectively (Fig. 6D), suggesting that it is likely that these sequences in collagen type I also mediate binding to CNE.
So far we have demonstrated two effects of CNE, namely inhibition of αVβ3-mediated PDGF BB-induced collagen gel contraction and inhibition of fibrillogenesis of collagen. We addressed the issue whether these two activities could be attributed to two distinct collagen peptides that were recognized by CNE. Soluble peptide II-44 affected collagen gel contraction such that it inhibited contraction at a dose of 50 μg/ml, whereas peptide II-1 had no effect at the same dose (Fig. 7). Conversely, whereas peptide II-1 inhibited fibrillogenesis measured by change in turbidity of dilute collagen type I or II solutions incubated at 37 °C, peptide II-44 slightly stimulated fibrillogenesis (data not shown). These findings suggest that the collagen region defined by peptide II-44, and not peptide II-1, is involved in αVβ3-mediated PDGF BB-induced contraction, whereas this region is not involved in fibrillogenesis.
Previously, our laboratories have identified several proteins that bind peptide II-44, including DDR2, SPARC (osteonectin) (26, 28), and fibronectin.3 Peptide II-44 also contains the MMP cleavage site. Together, these data indicate that the collagen locus defined by peptide II-44 contains a broad-specificity binding region. Furthermore, PDGF-BB is known to induce increased synthesis of several ECM proteins including fibronectin (8). Therefore we asked whether fibronectin was involved in PDGF BB-induced αVβ3-mediated collagen gel contraction. A requirement for fibronectin in this in vitro system would imply that CNE interferes with binding between fibronectin and collagen. Binding of human plasma fibronectin and CNE to Toolkit peptides II-1 and II-44 was investigated in solid phase assays (Fig. 8). As expected CNE bound both peptides, but fibronectin only bound peptide II-44 (Fig. 8A). Furthermore, CNE inhibited binding of biotin-labeled fibronectin to immobilized collagen type I in a dose-dependent manner (Fig. 8B). In addition, biotin-labeled CNE was unable to bind to immobilized fibronectin (Fig. 8C). Together these data suggest that the two proteins compete for the same binding site or closely neighboring binding sites.
The potential role of fibronectin in PDGF BB-induced adhesion to native collagen type I by C2C12 cells was investigated using an anti-fibronectin IgG that binds and blocks adhesive processes mediated by fibronectin (10). This IgG effectively blocked PDGF BB-induced adhesion suggesting a role for fibronectin in αVβ3-mediated PDGF BB-induced adhesion to native collagen (Fig. 9A). Note that the control and PDGF-BB values in Fig. 9A are identical to those shown in Fig. 5B. To investigate the requirement of endogenously produced fibronectin for PDGF BB-induced C2C12 cell-mediated collagen gel contraction we took the approach to down-regulate expression of fibronectin in these cells by siRNA. siRNA directed against fibronectin knocked down protein levels of fibronectin in C2C12 cells to around 35% as compared with a scrambled oligonucleotide (siRNA-) or transfection agent alone (N-TER) (Fig. 9B). siRNA-transfected cells were put in collagen gels and gel diameters were measured after 24 h. Fibronectin siRNA inhibited PDGF BB-induced contraction by C2C12 cells, whereas siRNA- and N-TER still allowed for reduction of gel diameter after PDGF-BB stimulation (Fig. 9C). These results demonstrate that fibronectin is required for efficient PDGF BB-induced αVβ3-mediated collagen gel contraction.
We have investigated cell-mediated integrin αVβ3-dependent collagen gel contraction and adhesion using the collagen-binding protein CNE from S. equi subspecies equi. This bacterial cell-surface protein bound native fibrillar collagen types I, II, and III with high affinities (apparent Kd values ranging from 50 to 125 nm) but not denatured collagens. Based on the finding that CNE specifically inhibited αVβ3 integrin-mediated contraction and adhesion to native collagen, as well as fibrillogenesis, we reasoned that identification of the site(s) in collagen to which CNE binds might offer new insight into these processes. In our efforts to detect the CNE-binding sites we made use of the previously described collagen type II Toolkit (26, 29). CNE bound effectively to two peptides from Toolkit II and to a few additional peptides but with low signal to noise ratio. The two high affinity peptides were located at the N terminus of the triple helical part of collagen type II, i.e. peptide II-1, and three quarters toward the C terminus, i.e. peptide II-44.
The inhibitory effect of CNE on integrin αVβ3-mediated collagen gel contraction could potentially have been due to one or more of several possibilities. Thus, the effect could have been due to the fact that CNE binds and impairs function of αVβ3, this possibility could be ruled out by the finding that CNE had no effect on adhesion of cells to vitronectin, a process that is strictly dependent on αVβ3. Furthermore, because αVβ3-mediated contraction had to be induced by PDGF-BB, the inhibitory effect could have been due to that CNE negatively affected ligand binding or activation of the PDGF receptors. The effect of PDGF-BB on αVβ3-mediated contraction required activation of PDGF receptors because the tyrosine kinase inhibitor AG1296 blocked PDGF BB-induced contraction by C2C12 cells (supplemental Fig. S2A). However, CNE at a concentration that inhibited PDGF BB-induced contraction had no effect on PDGF β-receptor phosphorylation (supplemental Fig. S2B). PDGF BB-induced αVβ3 integrin-mediated contraction was furthermore, not likely dependent on changes in cell surface expression of β1, β3, or α5 integrin subunits because stimulation of C2C12 cells with PDGF-BB did not show differences in expression levels of these integrins (supplemental Fig. S2, C and D). The possibility that the effect of CNE on αVβ3-mediated collagen gel contraction was restricted to C2C12 cells or to cells lacking collagen-binding β1 integrins seems less likely based on collagen gel contraction experiments using AG1518 fibroblasts that effectively contract collagen gels also in the absence of exogenous stimulators and that utilize collagen-binding β1 integrins for this contraction. Monoclonal anti-β1 integrin IgG inhibited contraction mediated by human AG1518 diploid fibroblasts, an effect that could be overcome by PDGF-BB. This effect could in turn be abolished either by CNE or a cyclic peptide that blocks αVβ3-mediated cell interactions (supplemental Fig. S1B). CNE had, however, no effect on contraction by AG1518 cells in the absence of anti-β1 integrin IgG (supplemental Fig. S1A). Together, our data show that CNE specifically inhibits αVβ3-mediated contraction by binding to the native collagen fibers. The fact that peptide II-44 but not peptide II-1 supported adhesion of C2C12 cells and that soluble peptide II-44 but not peptide II-1 inhibited αVβ3 integrin-mediated contraction of collagen type I gels suggests that this site, recognized by CNE, constitutes a major recognition site in collagen for αVβ3 integrin-mediated cell interactions.
Previous studies using Toolkits have identified several proteins that bind to peptide II-44, including DDR2 and SPARC (osteonectin) (26, 28). MMPs are also known to bind and cleave collagen at this position, within the first few residues of peptide II-44. In the present studies we have presented evidence that argues against involvement of MMPs in PDGF BB-induced αVβ3-mediated contraction. In fact, inhibition of MMPs rather stimulated contraction, potentially by protecting the collagen fibers from cleavage during the contraction process, suggesting that contraction is optimally executed when collagen fibers are non-cleaved. Furthermore, our data speak against the idea that αVβ3-directed adhesion depended on exposed RGD sequences. This is based on the fact that peptide II-44 lacks RGD sequences. It remains possible that GM6001 exerts its effect here by displacing MMPs from collagen, allowing enhanced fibronectin binding and in consequence, greater gel contraction. We can, however, not exclude a role for DDRs in αVβ3-directed contraction. It is possible that DDR1 or -2 participates in activation of αVβ3 on C2C12 cells; however, data showing that DDR receptors do not activate β1-integrins on C2C12 cells have been reported (30). This speaks against the idea of a general mechanism for contraction based on activation of integrins by DDRs.
The observation that CNE inhibited fibrillogenesis of pepsin-solubilized collagen opens the possibility that αVβ3-mediated collagen gel contraction depends on the fibrillar status of the collagen matrix. The data presented herein speak, however, against this possibility. First and most importantly, contraction mediated by α2β1 occurred equally well in the presence of CNE as in its absence. Second, whereas soluble peptide II-44 inhibited αVβ3-mediated contraction it had little effect on fibril formation. Thus, even though CNE could modulate the biomechanical properties of collagen gels, it specifically inhibited collagen gel contraction mediated by αVβ3 integrins, whereas it did not affect contraction mediated by collagen-binding β1 integrins. Because CNE bound native collagens, this integrin-specific inhibition of contraction suggested that, during the respective contractile processes, α2β1 and αvβ3 integrins would be recruited to different domains on native collagen. Indeed, the key α2β1-binding sites, GLOGER, GFOGER, and GMOGER (31), are located in peptides II-7/8, II-28, and II-31, respectively, whereas αVβ3 is considered to bind the collagen triple helix only indirectly, as described here.
Many studies have reported on the effects of PDGF-BB on cells, including increased cytoskeletal dynamics (32) and synthesis of ECM proteins including fibronectin (8). The αVβ3 integrin-mediated attachment of cells to native collagen was increased by stimulation of the cells with PDGF-BB and proceeded only after a lag phase of around 120 min. This is in sharp contrast to attachment mediated by α2β1 that typically was completed within 30–60 min. Furthermore, α2β1-mediated contraction and adhesion proceeded in the absence of PDGF-BB. Data presented herein indicate that de novo synthesis of fibronectin is required for αVβ3-directed and PDGF-BB-induced contraction. Because PDGF-BB is known to induce production of fibronectin (8), it is thus possible that PDGF-BB induced the synthesis of fibronectin that has either a bridging or adaptive function. Our data strongly suggest that αVβ3 integrin-mediated contraction by C2C12 cells depends on binding of endogenously produced fibronectin to a site in collagen corresponding to peptide II-44. These findings are in agreement with that one fibronectin-binding site in collagen located adjacent to the MMP-binding site and thus to the region encompassing peptide II-44 (33,–35). It is noteworthy that adhesion of cells to plates coated with peptide II-44 occurred effectively also without stimulation of cells by PDGF-BB. This is most likely due to that in the absence of exogenous factors C2C12 cells express low, but sufficient amounts of cell-surface fibronectin that allow for attachment to coated substrates offering high-density binding sites. Previous studies from our laboratory show that suspended C2C12 cells indeed express cell-surface fibronectin (36), in line with such an assumption. In light of this, a low density of fibronectin-binding sites in fibrillar collagen gels could reflect the requirement for PDGF-BB to increase availability of fibronectin in αVβ3-directed contraction. Therefore, we suggest that induction of an increase in fibronectin expression leading to supportive cell-collagen fiber adhesions, together with increased cytoskeletal dynamics brought about by PDGF-BB could possibly be the mechanistic background to the effect of PDGF-BB on C2C12 cell-mediated collagen gel contraction. Fibronectin is present in physiological and pathological connective tissue compartments. Studies are ongoing to determine whether fibronectin is also required during αVβ3-directed normalization of IFP in vivo.
In a previous report we showed that the secreted collagen and fibronectin-binding protein FNE from S. equi has a possible function in blocking inflammatory driven edema formation that is part of the innate immune response. CNE would function in the opposite direction: by blocking αVβ3-dependent normalization of the lowered IFP, CNE would promote long standing edema. It is possible that bacteria can modulate edema response in either direction by expressing and shedding different surface components thereby modeling tissue responses during different phases of the infection.
We thank Viveka Tillgren for expert technical assistance.
*This work was supported by grants from the Swedish Science Council (to K. R.), the Swedish Cancer Foundation (to K. R.), the King Gustaf the V:s 80-årsfond (to K. R.), the Swedish Research Council for Environment, Agricultural Sciences and Spatial planning Grant FORMAS 221-2006-989 (to B. G.), Swedish Horse Board Grant H0747197 (to B. G.), and the Wellcome Trust and Medical Research Council supported synthesis of the Toolkits (to R. W. F.).
3D. Bihan, unpublished data.
2The abbreviations used are: