Promotion of fibroblast adhesion by triple-helical peptide models of type I collagen-derived sequences.

The dissection of the activities mediated by type I collagen requires an approach by which the influence of triple-helical conformation can be evaluated. The αβ and αβ integrin binding sites within type I collagen are dependent upon triple-helical conformation and contained within residues 124-822 from α1(I). Seven α1(I)-derived triple-helical peptides (THPs) were synthesized based on charge clustering (α1(I)256-270, α1(I)385-396, α1(I)406-417, α1(I)415-423, α1(I)448-456, α1(I)496-507, and α1(I)526-537). Three additional THPs were synthesized (α1(I)85-96, α1(I)433-441, and α1(I)772-786) based on previously described or proposed activities (Kleinman, H. K., McGoodwin, E. B., Martin, G. R., Klebe, R. J., Fietzek, P. P., and Wooley, D. E.(1978) J. Biol. Chem. 253, 5642-5646; Staatz, W. D., Fok, K. F., Zutter, M. M., Adams, S. P., Rodriguez, B. A., and Santoro, S. A.(1991) J. Biol. Chem. 266, 7363-7367; San Antonio, J. D., Lander, A. D., Karnovsky, M. J., and Slayter, H. S.(1994) J. Cell Biol. 125, 1179-1188). Of the ten THPs, α1(I)772-786 THP had the greatest activity, with half-maximal normal dermal fibroblast adhesion occurring at a peptide concentration of 1.6 μM. Triple-helicity was essential for activity of this sequence, as the non-triple-helical peptide analog (α1(I)772-786 SSP) exhibited considerably lower levels of cell adhesion promotion even at peptide concentrations as high as 100 μM. Within the sequence itself, residues 784-786 (Gly-Leu-Hyp) were important for cellular recognition, as the α1(I)772-783 THP had greatly reduced cell adhesion activity compared with α1(I)772-786 THP. Preliminary studies indicate that the β integrin subunit mediates fibroblast adhesion to α1(I)772-786 THP. The identification of fibroblast integrin binding sites within type I collagen may have important implications for understanding collagen metabolism.

helical conformation composed of three chains with a repeating Gly-X-Y sequence motif. In addition to providing the structure of connective tissue, collagens can mediate intracellular communication. Cell-collagen interactions play a role in a number of processes including cell migration (1,2), collagen catabolism (3,4), and the aggregation of platelets (5)(6)(7)(8)(9)(10)(11)(12)(13). One approach for developing novel therapies for diseases linked to ECM interactions, such as tumor cell metastasis, atherosclerosis, inflammation, and thrombosis, or to better understand normal processes such as wound healing, is to identify cellular recognition sites within collagen molecules and dissect the structure-activity relationship for receptor-ligand binding (14).
Several distinct sequences derived from type I collagen CB fragments have been identified as cell adhesion sites. The adhesion of Chinese hamster ovary cells to type I collagen is inhibited by the 757-791 sequence located within the ␣1(I)CB7 fragment (17). A peptide incorporating residues ␣1(I)769 -783 supports human fibroblast adhesion and migration and inhibits human fibroblast and human T-lymphocyte attachment to type I collagen (18,19). By using short synthetic peptides, an ␣ 2 ␤ 1 integrin binding site could be identified containing the minimal sequence Asp-Gly-Glu-Ala (20). This sequence corresponded to residues 435-438 within the ␣1(I)CB3 fragment. The ␣1(I)434 -438 peptide inhibits adhesion of platelets and breast carcinoma cells to type I collagen in a concentration-dependent manner (20) and, not surprisingly, was not effective at inhibiting ␣ 2 ␤ 1 -mediated chondrosarcoma cell adhesion to type II collagen (21).
In the present study, we have examined potential cellular recognition sites within type I collagen and studied the significance of triple-helical conformation. Sequences were derived from ␣1(I)CB fragments that possess integrin binding sites. To utilize relatively short sequences (Յ15 residues) and yet ensure triple-helical conformation under biological assay conditions, we applied a methodology developed specifically for the assembly of collagen-model, triple-helical peptides (THPs) (34,35). A total of 11 THPs have been synthesized. Cellular recognition was studied by assaying normal human dermal fibroblast adhesion to these THPs. The influence of triple-helicity was examined by comparing the activity of a SSP and THP containing the same collagen-derived sequence. The potential involvement of integrins in mediating cell adhesion to a specific THP was evaluated by screening monoclonal antibodies (mAbs) against integrin subunits as inhibitors of cell adhesion assays.
Preparation of (N-Tris(Fmoc-Ahx)-Lys-Lys)-Tyr(tBu)-Gly-Sasrin Resin-0.97 g of Fmoc-Tyr(tBu) (2.1 mmol), 1.96 g of Fmoc-Lys(Dde) (2.1 mmol), and 1.96 g of Fmoc-Lys(Dde) (2.1 mmol) were sequentially coupled to 1.0 g of Fmoc-Gly-Sasrin TM resin (substitution level ϭ 0.7 mmol) in a shaker. Fmoc-protected amino acids were preactivated with 0.32 g of HOBt (2.1 mmol) and 0.33 ml of DIPCDI (2.1 mmol) in 15 ml of DMF at room temperature for 15 min. The preactivated amino acid solution was then added together with 0.30 ml of DIEA (2.1 mmol) to the resin, and coupling proceeded for 1 h at room temperature. Fmoc removal was by two treatments with 20 ml of piperidine-DMF (1:4) for 3 and 10 min. After coupling of the second Fmoc-Lys(Dde) and following Fmoc removal, cleavage of Dde protecting groups was achieved with 15 ml of hydrazine-DMF (1:49) for 2 h. 2.23 g of Fmoc-Ahx (6.3 mmol) was then coupled for 45 min to the free N ␣ -and N ⑀ -amino groups after preactivation with 0.97 g of HOBt (6.3 mmol) and 0.99 ml of DIPCDI (6.3 mmol) and following addition of 0.89 ml of DIEA (6.3 mmol). The peptide-resin was washed three times with DMF. After each coupling and deprotection step, the completion of the reaction was examined by the ninhydrin test (37).
After assembly, branched peptide-resins were washed several times with DMF followed by dichloromethane. A small quantity of the branched peptides was deprotected and cleaved with water-TFA (1:19) for 1 h. After precipitation and washing twice in methyl t-butyl ether, the branched peptides were lyophilized to a colorless powder. The The large-scale synthesis of peptide ␣1(I)772-786 THP was on an Applied Biosystems 431A Peptide Synthesizer. Peptide assembly, including final stepwise incorporation of individual Fmoc-Gly, Fmoc-Pro, and Fmoc-Hyp(tBu) residues, was performed by Fmoc solid-phase methodology as described (34) with several modifications. Coupling utilized 0.45 M HBTU, 0.50 M HOBt, and 0.95 M DIEA in DMF for 1 h, while Fmoc removal was achieved with 0.1 M HOBt in piperidine-1methyl-2-pyrrolidinone (1:4) for 24 min and 6 min. The following cleavage procedure was applied for the large-scale synthesis of ␣1(I)772-786 THP and was modified for other THPs according to their side-chain protection (39). 203 mg of the peptide-resin was Fmoc-deprotected with DBU-piperidine-DMF (1:1:48). Side-chain deprotection and cleavage was by treatment with water/thioanisole/TFA (1:1:18). The crude ␣1(I)772-786 THP was precipitated with methyl t-butyl ether, lyophilized, and 70 mg (from a total of 100 mg) was dissolved in 2.0 ml of water.
Preparative RP-HPLC purification was performed on a Rainin Au-toPrep System with a Vydac 218TP152022 C 18 column (15-20 m particle size, 300 Å pore size, 250 ϫ 22 mm) at a flow rate of 5.0 ml/min. The elution gradient was 0 -100% B in 100 min where A was 0.1% TFA in water and B was 0.1% TFA in acetonitrile. Detection was at 229 nm. A second purification was performed on the same system with a semipreparative Vydac 219TP510 diphenyl column (5 m particle size, 300 Å pore size, 250 ϫ 10 mm) at a flow rate of 2 ml/min. The elution gradient was 0 -70% B in 70 min with the described eluents. The purification of ␣1(I)772-786 THP was described recently (40); yield was 18.1 mg (15.2% of theoretical).
Peptide Analysis-Analytical RP-HPLC was performed on a Hewlett-Packard 1090 Liquid Chromatograph equipped with a Vydac 219TP54 diphenyl column (5 m particle size, 300 Å pore size, 250 ϫ 4.6 mm). The flow rate was 1.0 ml/min. Eluants were 0.1% TFA in water (A) and 0.1% TFA in acetonitrile (B). The elution gradient was 0 -50% B in 50 min with a flow of 1 ml/min. Detection was at 229 nm.
Edman degradation sequence analysis was performed on an Applied Biosystems 477A Protein Sequencer/120A Analyzer as described (42,43) for "embedded" (noncovalent) sequencing. FAB mass spectra were obtained on a VG 7070-HF mass spectrometer and ES mass spectra on a Sciex API III double quadrupole mass spectrometer. Laser desorption mass spectrometry was performed on the Kratos Kompact MALDI matrix-assisted laser desorption time-of-flight mass spectrometer. Circular dichroism (CD) spectroscopy was performed on a Jasco 710 spectropolarimeter using a 0.01-cm cell. Thermal transition curves were obtained by recording the molar ellipticity ([]) in the range of 10 -80°C at ϭ 225 nm.
Cell Adhesion Assays-Adhesion assays were as described previously (44) with some alterations. Normal human dermal fibroblasts (NHDF) (Clonetics, San Diego, CA) were cultured on fibroblast growth medium from the same company. Cell attachment assays were performed in 96-well polystyrene Immulon 1 microtiter plates (Dynatech Laboratories Inc., Chantilly, VA). Peptides were dissolved in PBS and absorbed onto wells overnight at 37°C in a humidified incubator with 5% (v/v) CO 2 . Cells were labeled overnight with 200 Ci/ml 35 S (DuPont NEN). Nonspecific binding sites were blocked with 2 mg/ml bovine serum albumin (Pentax, Miles Laboratories, Naperville, IL) in PBS for 2 h at 37°C. NHDF cells were detached with trypsin/EDTA solution and, after release of the cells, the flask was rinsed quickly with HEPES-buffered saline solution and trypsin neutralizing solution (all from Clonetics). The cells were resuspended to 5 ϫ 10 4 /ml in fibroblast growth medium. Aliquots of 75 l of the cell suspension were added to the plate wells and allowed to adhere for 30 min at 37°C. For adhesion assays, wells were washed 3 times with 2 mg/ml bovine serum albumin in Ca 2ϩ /Mg 2ϩ PBS. 150 l of Microscint-40 (Packard Instrument Co.) were added and the plates were read on a Beckman LS 6500 scintillation counter (Beckman Instruments). Adhesion percentages were based on total counts of radioactivity added to each well.
Inhibition of NHDFs was monitored in the presence of soluble synthetic peptides or mAbs generated against the ␣ 1 , ␣ 2 , ␣ 3 , ␣ 5 , and ␤ 1 integrin subunits to determine the cell surface receptor recognizing the ␣1(I)772-786 THP. Competition of cell adhesion was performed on surfaces coated with 5 g/ml type I collagen (half-maximal cell adhesion) or with 10 M ␣1(I)772-786 THP using methods described previously (44,45). Cells were preincubated for 20 or 30 min at 37°C with various concentrations of potential inhibitory peptides or antibodies. The cells were then, in the continued presence of the potential inhibitor, added to the wells and allowed to adhere for 30 min at 37°C. The cells remained viable in the inhibitory peptide-cell solution based upon exclusion of trypan blue dye.

Synthesis and Cellular Activities of Crude Triple-helical Type I Collagen Model
Peptides-Four criteria were used for the identification of potential active sequences from collagen ␣1(I). First, sequences would be from CB fragments that contained integrin binding sites (␣1(I)CB3, ␣1(I)CB7, and ␣1(I)CB8)). Second, sequences containing charge clusters were considered, as several active sequences from type IV collagen have a clustering of charged residues (35, 44, 46 -48). Third, sequences described previously to have binding activities, such as ␣1(I)85-96 (49), ␣1(I)433-441 (20), and ␣1(I)769 -783 (17) were utilized. Fourth, sequences were 9 -15 residues in length, as prior studies on type IV collagen had shown this length to be sufficient to establish cellular activities (35,44). For initial screening purposes, 10 sequences from the ␣1(I) chain (50 -53) were examined (Table I). THPs containing these sequences were assembled using a methodology for the solid-phase synthesis of branched triple-helical peptides (34,35). Three nascent peptide chains were carboxyl-terminally linked through one N ␣ -amino and two N ⑀ -amino groups of Lys and stabilized by 6 repeats of Gly-Pro-Hyp. Edman degradation sequence analysis indicated that each of the 10 crude THPs contained at least 50% of the desired product.
The 10 homotrimeric THPs and a generic THP (designated GPP*) containing only eight repeats of Gly-Pro-Hyp were examined for promotion of NHDF adhesion. At a peptide concentration of 0.1 mg/ml, only the THP incorporating ␣1(I)771-786 showed substantial (ϳ50%) cell adhesion activity (Fig. 1). None of the other crude THPs showed Ͼ10% cell adhesion activity (Fig. 1). The high adhesion of NHDFs to the ␣1(I)772-786 THP was thus examined further.
Synthesis and Characterization of ␣1(I)772-786 THP-The ␣1(I)772-786 THP was resynthesized on a larger scale to allow for extensive structural and biological characterization. Several modifications of our previously described procedure (34,35) were used. The final 6 repeats of Gly-Pro-Hyp, which had previously been coupled manually using Fmoc-Gly-Pro-Hyp (34, 35), were added stepwise (as Fmoc-Gly, Fmoc-Pro, and FIG. 1. Promotion of NHDF adhesion by type I collagen, GPP*, or crude THPs incorporating type I collagen sequences. Substrate concentrations were 0.1 mg/ml for peptides and 0.02 mg/ml for type I collagen. Cells were allowed to adhere to peptide-or proteincoated Immulon 1 plates for 30 min at 37°C. All assays were repeated in triplicate. Conditions are given under "Experimental Procedures."

TABLE I Type I collagen sequences incorporated into triple-helical peptides
Sequences were based primarily on the human ␣1(I) gene (51)(52)(53). The chicken and bovine ␣1(I) protein sequences (50) were used to fill in post-translational modification of Pro to Hyp in the Y positions where applicable.
To study the significance of collagen triple-helical structure on cellular recognition, we compared the NHDF adhesion-promoting ability of the ␣1(I)772-786 THP to the ␣1(I)772-786 SSP (Fig. 6). NHDFs showed a profound preference for binding and adhesion to the THP compared with the SSP. Half-maximal fibroblast cellular adhesion occurred at a THP concentration of 1.6 M, while less than 10% cell adhesion was seen for the SSP up to a concentration of 10 M. At a peptide concentration of 100 M, cell adhesion to the SSP was only ϳ40% of the level promoted by the THP (data not shown). There was no NHDF adhesion to the generic THP containing 8 repeats of Gly-Pro-Hyp (GPP*) (Fig. 6).
Inhibition of Cell Adhesion to ␣1(I)772-786 THP by Antiintegrin mAbs-The inhibition of NHDF cell adhesion to ␣1(I)772-786 THP was compared for different anti-integrin mAbs (Table II). Initial experiments compared inhibition of adhesion by normal mouse IgG and the anti-␤ 1 and anti-␣ 2 mAbs at varying THP coating concentrations. Inhibition of adhesion was seen for the anti-␤ 1 mAb at [THP] ϭ 2.0, 5.1, and 10 M. The anti-␣ 2 mAb results were somewhat inconclusive. Studies at [THP] ϭ 10 M suggested that the anti-␣ 1 and anti-␣ 3 may inhibit adhesion to the ␣1(I)772-786 THP and that the anti-␣ 5 integrin subunit mAb was not an inhibitor.
Inhibition of Cell Adhesion to Type I Collagen-To determine the ability of ␣1(I)772-786 SSP and ␣1(I)772-786 THP to inhibit cell adhesion to type I collagen, NHDFs were incubated with increasing peptide concentrations in the range of 0.5-10 M. Neither the ␣1(I)772-786 SSP nor the ␣1(I)772-786 THP inhibited NHDF adhesion to type I collagen in a concentrationdependent fashion (data not shown). In contrast to the SSP and THP, type I collagen inhibited cell adhesion in a concentrationdependent fashion (data not shown).

DISCUSSION
The dissection of the various biological activities mediated by type I collagen requires an approach by which the influence of triple-helical conformation can be evaluated. More specifically, the mechanisms of collagen catabolism may require two types of cellular recognition sites, those that are dependent upon triple-helical conformation and those that are revealed upon denaturation of the triple-helix. Our initial interest is in iden-tifying sites that are recognized in triple-helical conformation. The ␣ 1 ␤ 1 and ␣ 2 ␤ 1 integrin binding sites are dependent upon triple-helical conformation (56) and contained within ␣1(I)CB3, ␣1(I)CB7, and ␣1(I)CB8 (encompassing residues 124 -822 from ␣1(I)) (2,12,13,15,16). From the 124 -822-residue region, we selected sequences that contained "clusters" of charged residues. Charged residues are often found in clusters in type I collagen (57), and cellular activities have been ascribed previously to collagen-derived synthetic peptides that have clustered charged residues (35, 44, 46 -48). Seven THPs were synthesized based on charge clustering (␣1 (I (17,20,49).
The 10 THPs were screened for cell adhesion activity without prior purification of the peptides. Edman degradation sequence analysis indicated that each crude THP contained a substantial amount of the desired peptide. Thus, the crude THPs were adequate for screening purposes. Of the 10 THPs, ␣1(I)772-786 THP had the greatest cell adhesion-promoting activity. The other 9 THPs exhibited low levels of activity (Ͻ10%), similar to the generic triple-helical peptide GPP*. Although the other THPs may represent active sequences, only the ␣1(I)772-786 THP was pursued in this study.
The large scale synthesis and purification of the ␣1(I)772-  786 THP proceeded as described for other THPs (34), with two important modifications. First, the (Gly-Pro-Hyp) 6 region of the THP was assembled stepwise using individual Fmoc-amino acids, not Fmoc-Gly-Pro-Hyp tripeptide blocks. Stepwise assembly had not been possible previously using Fmoc-Hyp, but was successful with the Fmoc-Hyp(tBu) derivative used here. We believe that tBu side-chain protection of Hyp minimizes interstrand hydrogen bonding. Interstrand hydrogen bonding can be detrimental for efficient peptide assembly (for a recent review, see Ref. 58). Second, a recently developed two-step RP-HPLC method was used for the purification of ␣1(I)772-786 THP (40) which also allowed for the isolation of the deletion peptide ␣1(I)772-783 THP. The ␣1(I)772-786 THP was highly active, with half-maximal cell adhesion occurring at a peptide concentration of 1.6 M. Triple helicity was essential for activity of this sequence, as the non-triple-helical peptide analog (␣1(I)772-786 SSP) exhibited considerably lower levels (Յ40%) of cell adhesion even at peptide concentrations as high as 100 M. The triple-helical dependence for cell binding to the ␣1(I)772-786 sequence is even more pronounced than for the ␣1(IV)1263-1277 sequence described previously (35). Within the ␣1(I)772-786 sequence itself, residues 784 -786 (Gly-Leu-Hyp) were important for cellular recognition, as the ␣1(I)772-783 THP had greatly reduced cell adhesion activity compared with ␣1(I)772-786 THP.
Adhesion of NHDF to the ␣1(I)772-786 THP is inhibited by an anti-␤ 1 integrin subunit mAb. It thus appears that an integrin mediates NHDF binding to ␣1(I)772-786 THP. Our preliminary results were inconclusive, suggesting the ␣ 1 ␤ 1 , ␣ 2 ␤ 1 , or ␣ 3 ␤ 1 integrin may be involved. It has been shown that fibroblasts use the ␣ 2 ␤ 1 integrin for collagen, but not laminin, binding (59). Binding of the ␣ 2 ␤ 1 integrin to the ␣1(I)CB7 fragment (residues 552-822) is conformationally dependent (13), consistent with the conformationally dependent binding of NHDF to ␣1(I)772-786 THP. Alternatively, fibronectin has been proposed as a "bridge" for ovarian cell binding to the ␣1(I)757-791 sequence (17). MMP-1 cleavage of the 775-776 bonds or mutation of ␣1(I) Gln 774 and Ala 777 to Pro dramatically alters fibronectin binding to type I collagen (17,60). The ␣ 5 ␤ 1 can mediate chondrosarcoma cell binding to denatured type II collagen via a fibronectin bridge (21), and thus an integrin may serve a similar function for cell binding via a fibronectin bridge to the ␣1(I)772-786 region of type I collagen. Further investigations are ongoing to definitively determine the integrin(s) utilized for cellular recognition of the ␣1(I)772-786 THP.
One curious result is the ability of the ␣1(I)772-786 THP to promote cell adhesion in a concentration-dependent fashion, but not inhibit cell adhesion to type I collagen. In retrospect, it would have been somewhat surprising if the ␣1(I)772-786 THP did inhibit NHDF binding to type I collagen due to the multiple integrin binding sites within type I collagen. It is also possible that there are interactions between the THP and type I collagen. We have previously demonstrated aggregation of THPs (35), but have not examined the association of THPs and collagen.
Fibroblast interaction with collagen has tremendous implications for understanding the regulation of collagen metabolism and hence processes such as wound healing. Degradation of collagen may proceed (i) intracellularly following phagocytosis or (ii) extracellularly by MMPs (3,4,61). Fibroblasts both phagocytize type I collagen (3,4) and produce MMP-1 (62). For fibroblasts, the two mechanisms of collagen catabolism may be inversely correlated (4). For example, interleukin 1␣ inhibits phagocytosis and enhances pro-MMP-1 release, while transforming growth factor-␤ has the opposite effect (4). There is evidence that suggests that the ␣1(I)772-786 sequence mediates both proposed mechanisms of collagen turnover. We have demonstrated that fibroblasts bind to the triple-helical ␣1(I)772-786 sequence. Internalization of type I collagen by fibroblasts is reduced after collagen is cleaved at the 775-776 bonds (3). Thus, fibroblast phagocytosis of type I collagen appears to include at least part of the 772-786 region. Several members of the MMP family (MMP-1, MMP-2, and MMP-8) hydrolyze the triple-helical region of type I collagen at position 775 in the collagen chains (27,63). Thus, extracellular degradation of type I collagen occurs within the 772-786 region. This region may also regulate MMP production. Prior studies have shown that a SSP incorporating residues ␣1(I)769 -783 supports human fibroblast adhesion and induces the production of MMP-1 (18,19,64). Although the induction mechanism is unknown, it may be related to ␣ 2 ␤ 1 -mediated binding to type I collagen which results in tyrosine phosphorylation of pp125 FAK (65) and induction of MMP-1 mRNA levels (66). If the ␣ 2 ␤ 1 integrin is indeed the cell surface adhesion molecule that binds ␣1(I)772-786 THP, we would be able to study a discreet cell signaling mechanism that influences collagen metabolism. Also, our ␣1(I)772-786 THP may have even greater activity for promoting cell signaling and MMP production than the ␣1(I)769 -783 SSP, as cell adhesion to triple-helical collagen results in considerable up-regulation of protein synthesis compared with denatured collagen (67).
We view the studies presented here as an encouraging start to understanding the variety of biological activities mediated by type I collagen. As stated by Tuckwell et al. (21), "crucially, the demonstration of conformation dependence suggests that linear peptides may be unsuitable (for studying) cell-collagen interactions and implies that more sophisticated methods may be necessary for future studies." Our triple-helical peptide approach appears to be a logical one for the identification of conformationally dependent collagen-mediated functions.