INTRODUCTION

Cell interaction with collagens is mediated by specific cell-surface receptors, important among which are the integrins ₁₁ and ₂₁ (1). Integrins recognize discrete amino acid sequences, the best known of which, perhaps, is the RGD¹ sequence, which serves as a cell-binding site in several matrix proteins and is recognized by a number of integrins, including the ₁-integrins ₃₁, ₄₁, ₅₁, and _v1, and the ₃ integrins _IIb3 and _v3 (1-4). In collagen IV, a triple-helical-dependent ₁₁-binding site has been described, involving Arg⁴⁶¹ of the 2(IV) chain and Asp⁴⁶¹ in the 1(IV) chain (5).

₂₁ is a platelet receptor that may be important in the activation of platelets by collagen in hemostasis, an event that may be expressed pathologically as thrombosis (6-8). Fragmentation studies have indicated the presence in collagen I of several platelet ₂₁-binding sites whose recognition is dependent on collagen triple-helical conformation (9). On the basis of inhibition of platelet adhesion to this collagen by short, linear (non-triple-helical) peptides, an ₂₁-binding site has been assigned to the sequence DGEA, which corresponds to residues 435-438 of the 1(I) chain and is found in the CNBr-derived fragment 1(I)CB3 (10).² However, others have observed no inhibition of ₂₁-mediated cell adhesion to collagen by DGEA-containing peptides (9, 11-15). Platelet adhesion to 1(I)CB3 is ₂₁-dependent (9, 16), but the fragment exhibits little platelet aggregatory activity (17). In contrast, the highly structurally homologous collagen III fragment 1(III)CB4 not only supports ₂₁-mediated adhesion, as reported here, but is also highly aggregatory (17). This suggests that recognition of ₂₁ may be insufficient for platelet activation and that recognition of additional reactive sequences in collagen by other receptors may be required. Pertinent to this, we have recently described potent platelet activation by simple collagen-like synthetic peptides comprising a GPP* repeat sequence, whose activity is totally ₂₁-independent (18). The DGEA sequence reported to serve as an ₂₁ site in 1(I)CB3 is not present in 1(III)CB4. To define more precisely the primary structural requirements of collagen for platelet reactivity and to identify the ₂₁ site in 1(III)CB4, we have synthesized this fragment as an overlapping series of triple-helical peptides. Here we describe studies on the ability of these peptides to support platelet adhesion and activation, and the adhesion of HT 1080 cells, which also express the integrin ₂₁. A preliminary account of some of this work has been given (14).

EXPERIMENTAL PROCEDURES

Collagens I and III were purified from bovine skin, following limited pepsin digestion, and the collagen type III-derived fragment 1(III)CB4 isolated from the purified parent collagen, as described previously (9, 17). Bovine tendon collagen (type I) fibers, dialyzed and diluted using 0.01 M acetic acid (17), were a gift from Ethicon Inc., Somerville, NJ, and were used as a standard platelet aggregatory agent.

The anti-(human integrin ₂-subunit) mAb 6F1 (19) was a generous gift from Dr. B. S. Coller, Mount Sinai Hospital, New York, NY. Anti-(human integrin ₂-subunit) mAb, clone A2-IIE10, was purchased from TCS Biologicals Ltd., Botolph Claydon, Bucks., United Kingdom (UK), and the anti-(human integrin ₁-subunit) mAb 13 from Becton Dickinson UK Ltd., Oxford, UK.

Platelet adhesion was measured in Falcon 1008 35-mm Petri dishes using ⁵¹Cr-labeled gel-filtered human platelets as described (18). When testing mAbs for inhibitory activity, platelets were preincubated with antibody for 15 min. The effect of mAbs on platelet adhesion was tested by one-way ANOVA (analysis of variance), either within experiments or using the mean values from at least three separate experiments.

Platelet aggregation was measured turbidimetrically using human citrated platelet-rich plasma as previously (18).

HT 1080 cells, from the European Collection of Animal Cell Cultures, Porton Down, Wilts., UK, were maintained in Eagle's minimal essential medium containing 15% fetal bovine serum, 2 mM glutamine, 100 IU/ml penicillin, 100 µg/ml streptomycin, and 2.5 µg/ml amphotericin. Cells were harvested with trypsin/EDTA, suspended in Eagle's minimal essential medium containing 20% fetal bovine serum, washed four times with Dulbecco's phosphate-buffered saline solution (Ca²⁺- and Mg²⁺-free) and finally suspended in adhesion buffer (Tris-buffered saline solution) containing 1 mM Mg²⁺. Immulon 2 multiwell plates were coated with collagen or peptide, normally at 10 µg/ml, for 1 h at 20 °C. Cell suspension (0.1 ml, 3 × 10⁴ cells) was added to each well and adhesion measured after 90 min at 20 or 37 °C, as required. Adhesion was determined using a Coulter Counter (model ZF) to count unattached cells. Cells were preincubated with antibody, when testing for inhibition, for 15 min. The significance of any inhibition of adhesion by mAbs was tested using the same statistical analysis as for platelets.

Peptides were synthesized as homotrimers covalently bonded at the C terminus, as shown in Fig. 1A, essentially as described by Fields and colleagues (20, 21), and employing standard Fmoc chemistry. The initial synthesis of a branched peptide structure on the resin was undertaken manually. For each synthesis, 0.25 g of Fmoc-peptide amide linker-polyethylene glycol-polystyrene resin (PerSeptive Biosystems, Hertford, UK; 0.2 mmol/g), preswollen in DMF, were loaded into a 10-mm diameter column and washed with DMF. Subsequent DMF washes, at 3.3 ml/min, were for 10 min prior to deprotection and for 20 min after deprotection. Fmoc amino acids were double-coupled using O-(7-azabenzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate/N,N-diisopropylethylamine chemistry (22) for 60 min employing a 4-fold excess of reagents. Fmoc groups were removed with a mixture of 2% 1,8-diazabicyclo(5.4.0)undec-7-ene and 2% piperidine in DMF for 30 min. The Dde protecting group was removed with 2% hydrazine hydrate in DMF for 30 min. Coupling and deprotection at each step was monitored using the appropriate color tests.

The resin was initially coupled to Fmoc-6-aminohexanoic acid-OH to introduce a spacer arm. The trimeric frame was then generated by first coupling Fmoc-Lys(Dde)-OH, followed by Fmoc-Lys(Fmoc)-OH. Removal of both Fmoc groups and the Dde group produced three amino groups to each of which was attached an 6-aminohexanoic acid spacer arm, providing three foci for subsequent peptide synthesis. Amino acid analysis at this stage verified the structure of the branched peptide resin.

The remainder of the synthesis was carried out automatically on a PerSeptive Biosystems 9050 Plus Pepsynthesizer using O-(7-azabenzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate chemistry and deprotection with 20% piperidine in DMF. The completed peptide was cleaved from the resin with 88% trifluoroacetic acid, 5% phenol, 5% water and 2% triisopropylsilane. Peptides were purified using reverse-phase chromatography (23) and the composition verified by amino acid analysis.

In all, seven overlapping peptides, designated 1-7, based on the sequence of the fragment 1(III)CB4, were synthesized. The triple-helical stability of each peptide was assessed by polarimetry as described previously (18).

Peptides were cross-linked with 3-(2-pyridyldithio)propionic acid N-hydroxysuccinimide ester as before (18).

RESULTS

The sequence of each of the seven peptides based on the known sequence of bovine 1(III)CB4 (24) is shown in Fig. 1B. Each sequence was constructed as a homotrimer covalently bound at the C terminus, as indicated in Fig. 1A, to promote correct alignment of the three chains as they adopt a triple-helical conformation, necessary for the expression of platelet reactivity (25). Initial studies indicated that each sequence as a covalently bonded trimer was only able to form a triple-helical structure at relatively low temperatures. Additional GPP* triplets were therefore added at each end, as indicated in Fig. 1B, to allow the adoption of a triple helix stable at 20 °C, the minimum temperature at which cell reactivity of the peptides could be measured. A GCP* triplet was also included at each end to allow cross-linking to produce a polymer, since collagen quaternary structure is also essential for the expression of its aggregatory activity (25). Melting temperatures (T_m1/2) were as follows: peptide 1, 33 °C; 2, 25 °C; 3, 25 °C; 4, 34 °C; 5, 26 °C; 6, 27 °C; 7, 26 °C. A representative melting curve is shown in Fig. 2.

Following cross-linking, all seven peptides exhibited platelet aggregatory activity at 20 °C. The minimum concentration (µg/ml) required for activity was as follows: peptide 1, 20; peptide 2, 30; peptide 3, 30; peptide 4, 1; peptide 5, 500; peptide 6, 0.5; and peptide 7, 5. This represents a 1000-fold difference in activity between the most and the least reactive. In accord with the melting temperatures and the need for a collagen triple-helical conformation for platelet reactivity (25), none of the peptides exhibited aggregatory activity at 37 °C. In accord with the requirement for a quaternary structure for collagen to express aggregatory activity (25), none of the uncross-linked peptides showed activity. For any given peptide, aggregatory activity recorded from one cross-linked sample to another was relatively consistent. Thus peptides 1-3 were always of moderate activity, whereas peptide 5 showed low activity. Peptide 6 was highly active (more so than collagen fibers, which were active at 2 µg/ml, and the parent collagen type III polymerized by random cross-linking, active at 10-20 µg/ml; Ref. 17) and was of comparable activity to the collagen type III fragment 1(III)CB4 (randomly cross-linked) from which it is derived (17). Peptide 7 showed activity intermediate between peptides 1-3 and peptide 6. For reasons unknown, the activity of peptide 4 was variable. Usually, the peptide showed activity comparable to or greater than collagen fibers, but occasionally the peptide showed no activity even when tested at concentrations up to 250 µg/ml. Platelet aggregation by peptide 6 is shown by way of example in Fig. 3.

9

7

Mg²⁺-dependent platelet adhesion to monomeric collagen type I immobilized on plastic and to the type I-derived fragment 1(I)CB3 at 20 °C is strongly inhibited by anti-integrin ₂₁ mAbs (9, 16, 19). Adhesion to both monomeric collagen III (26) and (as observed in this study) the type III-derived fragment 1(III)CB4, which is structurally highly homologous with 1(I)CB3, is also Mg²⁺-dependent and equally well inhibited by anti-integrin ₂₁ mAbs (data not shown). To identify the ₂₁ recognition site(s) in 1(III)CB4, we examined adhesion to seven 1(III)CB4-based peptides. Good adhesion occurred at 20 °C to all the peptides in the presence of Mg²⁺ and was strongly decreased in the presence of EDTA (Table I). In accord with the need for a triple-helical conformation for adhesion (9), no adhesion to peptides occurred at 37 °C (data not shown). mAbs against the integrin ₂₁ failed to inhibit adhesion except in the case of peptide 6, adhesion to which was reduced by a maximum of around 30%. Typical inhibition data are shown in Table II. Analysis of pooled data from Table II and other experiments showed that anti-₂ mAbs reduced the adhesion of platelets to monomeric collagen (type I) from 44 ± 1.5% to 2.3 ± 0.5% (p < 0.001) and to peptide 6 from 44 ± 1.4% to 34 ± 0.9% (p < 0.01). The anti-₂ mAbs exerted no significant effect on platelet adhesion to any other peptide. Higher concentrations of mAb beyond those utilized here with peptide 6 (as shown in Table II) gave the same results (data not shown).

Table I. Divalent cation-dependent platelet adhesion to peptides 1-7 Adhesion was measured at 60 min to collagen I (as the monomer isolated from bovine skin) or peptide adsorbed to Falcon 1008 Petri dishes (at 10 µg/ml for 1 h at 20 °C). Adhesion at 20 °C was in the presence of 2 mM Mg2+ or 2 mM EDTA as indicated. Results are expressed as the mean ± S.E. of three determinations. Exp. a-c, three representative experiments. Adhesion Mg2+ EDTA % Exp. a   Collagen 50  ± 1.3 0   Peptide 1 42  ± 3.8 8  ± 2.0   Peptide 2 46  ± 4.0 8  ± 0.9   Peptide 3 43  ± 2.8 6  ± 0.5   Peptide 4 51  ± 0.7 4  ± 0.1 Exp. b   Collagen 44  ± 1.5 1  ± 0.2   Peptide 5 28  ± 1.0 3  ± 0.3   Peptide 6 31  ± 1.8 4  ± 0.1 Exp. c   Collagen 49  ± 3.6 0   Peptide 5 42  ± 1.0 13  ± 0.1   Peptide 6 42  ± 0.6 3  ± 1.0   Peptide 7 53  ± 0.4 11  ± 0.4

Table II. 21-dependent platelet adhesion to peptide 6  Adhesion in the presence of 2 mM Mg2+ was measured as described in the legend to Table I. Anti-2 mAb 6F1 was used at 2 µg/ml and the anti-2 mAb A2-IIE10 at 5 µg/ml. Exp. a-f, six representative experiments. Adhesion Control + mAb 6F1 % Exp. a   Collagen 41  ± 0.9 1  ± 0.2   Peptide 1 40  ± 1.5 42  ± 0.6   Peptide 2 39  ± 2.0 38  ± 2.2 Exp. b   Collagen 44  ± 0.2 6  ± 0.1   Peptide 3 36  ± 0.1 36  ± 0.7   Peptide 4 40  ± 1.7 38  ± 1.9 Exp. c   Collagen 45  ± 0.5 1  ± 0.1   Peptide 5 43  ± 0.9 44  ± 1.4   Peptide 6 46  ± 2.5 34  ± 3.4 Exp. d   Collagen 41  ± 2.1 1  ± 0.1   Peptide 6 44  ± 0.5 32  ± 2.5 Exp. e   Collagen 47  ± 0.5 7  ± 0.6   Peptide 4 51  ± 1.9 50  ± 1.6   Peptide 7 50  ± 0.6 49  ± 1.0 Control + mAb A2-IIE10 Exp. f   Collagen 58  ± 0.5 5  ± 1.0   Peptide 1 56  ± 1.5 57  ± 0.7   Peptide 3 49  ± 1.2 47  ± 2.0   Peptide 5 45  ± 3.0 46  ± 0.4   Peptide 6 46  ± 1.0 35  ± 1.0   Peptide 7 57  ± 1.0 59  ± 1.2

To verify the presence of an ₂₁ recognition sequence in peptide 6, we also examined the adhesion of HT 1080 cells, which are known to adhere to collagen (type I) via ₂₁ (11, 15, 27-29). We found good adhesion (up to 80% at 37 °C at 90 min) to both collagens I and III, that was Mg²⁺-dependent and was inhibited by anti-integrin ₂₁ mAbs. Inhibition by antibody was generally 90%, but on occasion a residual adhesion, up to around 30% of the total, remained (results not shown), as reported by others (11). Adhesion to the type III CNBr-derived fragment 1(III)CB4 at 20 °C occurred at about 75% of that to the parent collagen, was similarly Mg²⁺-dependent, and was as effectively blocked by anti-₂₁ mAbs (data not shown). No significant adhesion occurred to any of the seven synthetic triple-helical peptides except peptide 6. As can be seen from the data presented in Table III, adhesion occurred to this peptide although some variation in level was noted between experiments. Irrespective of the actual level, adhesion was strongly inhibited by anti-₂ mAbs, confirming the presence of an ₂₁ recognition site in peptide 6. 

Table III. The adhesion of HT 1080 cells to collagen and peptide 6  The adhesion of HT 1080 cells to either monomeric collagen type I or peptide 6 was measured at 20 °C, as described under "Experimental Procedures," in the presence or absence of anti-2 mAbs (6F1, 2 µg/ml, or A2-IIE10, 5 µg/ml). Data shown are the means of at least four determinations, each performed in triplicate. Significance was determined using ANOVA; *, p < 0.001; **, p < 0.01.  Adhesion (mean ± S.E.) Control + mAb % Collagen 74  ± 2.8 9.1  ± 2.4* Peptide 6 32  ± 14.4 1.5  ± 0.3**

DISCUSSION

The ability of all seven peptides to cause platelet aggregation is consistent with our proposal that the collagen triple helix possesses an intrinsic platelet reactivity (18). The large variation in aggregatory activity between peptides further emphasizes the important influence of the primary sequence on this activity. We have previously argued that the intrinsic activity of the helix will be modified by the presence of stimulatory or inhibitory sequences within the collagen primary structure (18). The high activity of peptide 4 is consistent with the presence of the sequence GKP*GEP*GPKGEA, which we and others have previously proposed may serve as a platelet activation signal in collagen III (17, 30-32). Likewise, the potency of peptide 6 is consistent with the evidence presented here that this peptide contains an ₂₁ recognition sequence. Our data lend support to the view (33, 34) that collagen-platelet interaction is a two-step process involving an initial recognition of adhesive sequences, such as those recognized by ₂₁, as a primary interaction that may be especially important to retain platelets under flow conditions (7, 35). Subsequent engagement with a second receptor induces signaling leading to platelet activation (aggregation). We consider that this signaling receptor recognizes the collagen triple helix. In support of this, it has been found that the platelet-reactive (GPP*)₁₀-based peptides (18) induce in platelets signaling events identical to those evoked by collagen (36-39). Platelet activation in response to recognition of the triple helix may be enhanced by the presence within the helix of specific "activation" sequences such as the putative activation sequence in peptide 4.

All seven peptides were able to support conformation-dependent platelet adhesion under static conditions at a level comparable to that occurring to collagen. Adhesion to triple-helical monomeric collagen is Mg²⁺-dependent and mediated by the integrin ₂₁ (7-10, 16, 19). Adhesion to the peptides was also Mg²⁺-dependent, but only peptide 6 revealed any dependence on integrin ₂₁ for adhesion, indicating the existence of an additional divalent-cation dependent mechanism of adhesion that is not mediated by ₂₁. We have previously noted that polymeric collagen, i.e. collagen fibers, and the highly reactive synthetic (GPP*)₁₀-based peptides, which spontaneously form micropolymers, support substantial adhesion by a divalent cation-independent mechanism not involving ₂₁ (9, 18). At least three mechanisms of platelet adhesion under static conditions can therefore be discerned: two cation-dependent, of which only one is ₂₁-dependent; and a third, which is both cation- and ₂₁-independent.

Partial inhibition of platelet adhesion to peptide 6 by anti-₂₁ mAbs indicated the presence of an ₂₁ recognition sequence in this peptide, which was confirmed with HT 1080 cells. In contrast to platelets, these cells appear only to utilize ₂₁-dependent adhesion since, of the seven peptides, only peptide 6 was able to support significant adhesion and this was inhibitable with anti-integrin ₂₁ mAbs, confirming the presence of an ₂₁ recognition site in this peptide. The reason for the variation between experiments in the level of adhesion to this peptide is not known, but may suggest that presentation of the ₂₁ binding site in the correct conformation is crucial for these cells, and that the conformation may be affected or obscured in some way during the coating procedure.

From a comparison of the primary structure of peptide 6 with those of the two adjacent inactive peptides 5 and 7 (see Fig. 1), it is possible to deduce that GGPP*GPR, equivalent to residues 522-528 of the 1(III) collagen triple-helical chain (24), is the only sequence unique to peptide 6 and must represent the minimum structure involved in the recognition of integrin ₂₁. In support of the importance of this sequence in collagen-platelet interaction, it has been found that a mAb directed against human collagen III (40) and able to inhibit collagen III-induced platelet aggregation recognizes an epitope corresponding to residues 520-528 of the type III molecule.³ Integrin recognition sites generally contain a negatively charged residue, normally Asp (3). Conceivably the Glu residue in peptide 6 at residue position 515 or, perhaps less likely, that at position 537 may fulfill this requirement.