Platelet α₂β₁ integrin activation: contribution of ligand internalization and the α₂-cytoplasmic domain

Collagen is the major matrix protein of the blood vessel wall, and its exposure in ruptured atherosclerotic plaques activates platelets via the platelet collagen receptors, α₂β₁ and glycoprotein (GP) VI, thereby contributing to thrombosis.¹  The importance of the α₂β₁ integrin in thrombosis was initially recognized from studies of patients who lack α₂β₁ expression or who have antibodies against this receptor, because these patients have a mild bleeding disorder and their platelets exhibit defective aggregation in response to collagen.²  In recent studies it was determined that an α₂ gene polymorphism is associated with several cardiovascular disorders as reviewed by Kunicki,³  including myocardial infarction,⁴  vascular mortality in women,⁵  and stroke in younger patients.⁶ 

Although numerous studies have focused on the role of α₂β₁ in the process of collagen-stimulated platelet activation, only recently has any attention been given to the platelet-mediated activation of α₂β₁. Intracellular signals provoke changes in the integrin extracellular domain, resulting in integrin activation, as manifested by an increase in affinity of the integrin for its ligands, an important control mechanism for integrin function.⁷  This regulation is potentially critical for both normal hemostasis and thrombosis. For example, Wen et al⁸  showed that gangliosides from atherosclerotic plaques enhance platelet adhesion to immobilized soluble collagen via the α₂β₁ integrin. In addition, recent studies by Jung and Moroi and Moroi et al have provided evidence for conversion of α₂β₁ to an active conformation on agonist-stimulated platelets, as defined by the ability of these platelets to bind increased amounts of ¹²⁵I-labeled soluble type III collagen.^9-12 

However, many questions remain concerning the process and mechanisms of activation of α₂β₁. For example, α_IIbβ₃ on activated platelets¹³  and α₂β₁ on other cells are known to internalize ligands, and this process may increase the apparent extent of collagen binding to stimulated platelets, independent of significant α₂β₁ activation per se. Thus, the relative contribution of internalization versus activation of α₂β₁ to the binding of soluble collagen to activated platelets is unknown. Also unknown are the roles of the α₂ and β₁ cytoplasmic domains in the process of α₂β₁ activation on platelets. Studies of the α_IIbβ₃ integrin have shown that the cytoplasmic domains of each subunit play important roles in the activation process, particularly, the GFFKR motif, which is highly conserved in all known integrin α subunits.

Here we provide evidence that agonist stimulation of platelets activates the collagen-binding capacity of the α₂β₁ integrin and induces significant collagen internalization. These events occur with minimal changes in α₂β₁ expression on the cell surface. Furthermore, we provide evidence that the α₂ cytoplasmic tail, especially the KLGFFKR domain, plays an important role in α₂β₁ activation. Therefore, these data further clarify properties of the activation and regulation of α₂β₁ on platelets.

Soluble fluorescein isothiocyanate (FITC)–labeled bovine type I collagen (1 mg/mL in 0.01 M acetic acid, which lacks the telopeptides that are necessary for cross-linking) and unlabled bovine type I collagen (4 mg/mL in 0.01 M acetic acid, also lacking telopeptides) were purchased from Chondrex (Redmond, WA). Mouse anti-human α₂β₁ integrin monoclonal antibody (mAb; clone BHA 2.1) was from Chemicon International (Temecula, CA). Hamster antimouse α₂ mAb, control IgG_κ and R-phycoerythrin–conjugated mouse antihamster IgG were obtained from Pharmingen International (San Diego, CA). Bovine serum albumin (BSA; Pentex fraction V) was from Serologicals Proteins (Kankakee, IL). Adenosine 5′-diphosphate (ADP), heparin, apyrase, prostaglandin I₂ (PGI₂), and indomethacin were purchased from Sigma (St Louis, MO). Trypan blue was purchased from EM Diagnostic Systems (Gibbstown, NJ). Convulxin was a kind gift from Dr Thomas J. Kunicki (Scripps Research Institute, La Jolla, CA). Thrombin receptor activation peptide (TRAP) and cytoplasmic domain peptides were synthesized by solid-phase techniques, purified by high-performance liquid chromatography, and analyzed by the University of North Carolina-Chapel Hill Protein Chemistry Laboratory.

Cell-permeable chimeric peptides were designed by placing a hydrophobic delivery sequence N-terminal to the human α₂ or β₁ integrin cytoplasmic tail sequence (Table 1). The hydrophobic delivery sequence (VTVLALGALAGVGVG), is derived from the signal peptide sequence of the human β₃ integrin.¹⁴  Peptides without the cell signal sequence or peptides with a scrambled cytoplasmic tail sequence were designed as negative controls. Mutant α₂ cytoplasmic tail peptides with either a deletion or truncation after the GFFKR motif were also synthesized. Peptides synthesized and purified as in the previous paragraph were further purified by exchange of trifluoroacetic acid for HCl. Peptide amino acid content was confirmed by amino acid analysis.

Approval for this study was obtained from the University of North Carolina-Chapel Hill institutional review board, with informed consent provided according to the Declaration of Helsinki. Whole blood was drawn from healthy volunteers by venipuncture and collected into 0.14 volume of acid-citrate-dextrose anticoagulant. Platelet-rich plasma (PRP) was isolated by centrifugation using a Beckman GP tabletop centrifuge at 200g for 20 minutes at room temperature. PRP was treated with PGI₂ (5 ng/mL) and centrifuged at 800g for 10 minutes. Platelet pellets were resuspended in modified Tyrode buffer (5 mM HEPES [N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid], pH 6.5, 135 mM NaCl, 2.7 mM KCl, 11.9 mM NaHCO₃, 0.42 mM NaH₂PO₄, 1 mM MgCl₂, 1 mg/mL dextrose, 0.1% BSA) and incubated at 37°C with 50 U/mL heparin and 1 U/mL apyrase for 20 minutes and supplemented with 5 ng/mL PGI₂ immediately prior to centrifugation at 800g for 10 minutes. In most experiments the resulting pellet was resuspended in Tyrode buffer, pH 6.5, containing 0.5 U/mL apyrase and incubated for 15 minutes at 37°C. PGI₂ (5 ng/mL) was added prior to centrifugation (800g, 10 minutes). Washed platelet pellets were resuspended at 2 × 10⁸/mL in pH 7.4 Tyrode buffer containing 1% BSA and 2 mM MgCl₂, and incubated at 37°C for 30 to 60 minutes before use in experiments. The platelet concentration was determined with a Thrombo-counter (Coulter, Hialeah, FL) or Z1 Coulter particle counter (Coulter, Miami, FL). All experiments were performed with human platelets except where the use of mouse platelets is indicated. Murine PRP was isolated as described.¹⁵  Briefly, murine blood was obtained by cardiac puncture and collected into a 0.1 volume of 3.8% sodium citrate. Murine PRP was isolated by centrifugation at 90g for 10 minutes at room temperature.

Peptides with or without the signal delivery sequence were labeled by FITC isomer I on Celite 10% (Calbiochem, San Diego, CA) at a 1:2 molar ratio in 100 mM NaHCO₃ buffer (pH 9.0) for 30 minutes at room temperature. Excessive FITC dye was removed by dialysis overnight in NaHCO₃ buffer followed by phosphate-buffered saline (PBS) for 8 hours at 4°C. Platelets were incubated at room temperature for 20 minutes with 120 μM FITC-labeled peptide at a final concentration of 1 × 10⁸ platelets/mL. The platelet suspension was applied to coverslips that were preincubated with poly l-lysine (Sigma). Adhesion was conducted in the dark at room temperature for 20 minutes and coverslips were washed once with HEPES-PBS (pH 7.4). Platelets were fixed with 1% paraformaldehyde (PFA) for 15 minutes at room temperature then washed with NH₄Cl₂, followed by ice-cold 1% BSA in PBS. Coverslips were mounted on slides with FluorSave Reagent (Calbiochem) and the margins sealed with nail polish. Confocal images were obtained on an Olympus IX 70 confocal microscope.

Equal volumes of platelet suspensions were usually added last at a final concentration of 1 × 10⁸/mL to polypropylene tubes containing 25 to 50 μL soluble FITC-labeled collagen (25-50 μg/mL final concentration) and various agonists in Tyrode buffer (pH 7.4, containing 2 mM MgCl₂ and 1% BSA). Incubations were performed at room temperature for 30 minutes and stopped by adding a 10 × volume of ice-cold PBS. Tubes were kept on ice in the dark and were evaluated by flow cytometry within 2 hours.

Peptides were dissolved in PBS and diluted with Tyrode buffer (pH 7.4). Platelets (1 × 10⁸/mL) were incubated with 200 μM peptide at room temperature for 30 minutes and treated with agonist or control buffer for 2 to 5 minutes before fixation with 0.7% PFA (1 hour at room temperature). Fixed platelets were washed once with PBS, centrifuged, and the platelet pellet was resuspended in Tyrode buffer (pH 7.4, containing 1% BSA, 2 mM MgCl₂), and incubated with FITC-labeled collagen (10-25 μg/mL) for 30 minutes at room temperature. The reaction was stopped as described for analysis by flow cytometry. Most experiments were performed with triplicate or duplicate samples.

Cells were individually analyzed on a FACStar Plus flow cytometer (Becton Dickinson, San Jose, CA). Association of FITC-collagen or FITC-labeled peptides with platelets was monitored in the FL1 channel on the gated population of platelets from forward-scatter, side-scatter histograms. Data were collected as fluorescence mean values or as percentages of the positive population. The mean value of the fluorescence intensity of basal binding (ie, FITC-collagen binding to resting platelets) as measured by flow cytometry was normalized to 100 arbitrary units. All mean values of other data points within each experiment were normalized to arbitrary units relative to basal binding. Therefore, basal binding from all experiments is equal to 100. The Student t test was used in statistical analyses.

It was recently reported that collagen binding to the α₂β₁ integrin on platelets increases on agonist stimulation, as detected by ¹²⁵I-labeled soluble type III collagen.^9,16  Using FITC-labeled soluble type I collagen and flow cytometry, we obtained similar results with live platelets. Washed human platelets at room temperature bound significantly more FITC-collagen on TRAP stimulation compared with resting platelets (Figure 1A). Increased binding of FITC-collagen was also observed with other agonists, including ADP (Figure 1B), thrombin, and the GPVI-specific agonist, convulxin (data not shown). The increased binding was inhibited by 10 μg/mL α₂β₁ blocking mAb, but not by a control IgG (Figure 1A), and by 5 mM EDTA (ethylenediaminetetraacetic acid; Figure 1B). These results indicate that the agonist-stimulated FITC-collagen binding is dependent on both α₂β₁ integrin and divalent cation. Similar results were also observed with TRAP- or ADP-stimulated human or mouse PRP (data not shown).

Integrins play a role in fibrinogen internalization in platelets,¹³  collagen internalization in fibroblasts,^17,18  and phagocytosis in macrophages.¹⁹  We therefore asked whether the increased collagen binding was caused by an activation of α₂β₁, an increase in collagen internalization, or both. Trypan blue, which interacts with FITC dye and quenches its fluorescence emission, was used to distinguish internalized from surface-bound FITC-collagen. Trypan blue is expected to quench only surface-bound FITC-collagen molecules on live platelets because it is actively excluded from live cells.^20-23  However, trypan blue will quench intracellular fluorescence in dead cells. In live, non-TRAP–activated, control platelets incubated with FITC-collagen, trypan blue reduced the FITC fluorescence by about 15%, indicating that a portion of collagen was on the surface but that the rest was internalized (Figure 2). Upon TRAP activation, a more than 2.5-fold increase in FITC fluorescence was observed within 30 minutes and about 22% of this fluorescence was quenched by trypan blue, indicating that about 78% of the FITC-collagen was protected from trypan blue quenching and therefore likely to be internalized. The amount of internalized FITC-collagen was significantly decreased in platelets pretreated with an α₂β₁ function–blocking mAb, indicating that collagen internalization is at least partly α₂β₁ dependent (data not shown). To further confirm that internalization was occurring, we predicted that trypan blue should substantially quench FITC fluorescence if trypan blue was not actively excluded from the cell, thereby allowing access to internalized FITC-collagen. Indeed, when resting or TRAP-activated platelets were incubated with FITC-collagen as described, but subsequently fixed with PFA before addition of trypan blue, thereby abolishing its active extrusion, FITC fluorescence was significantly quenched by about 56% in control platelets and 80% in TRAP-activated platelets (Figure 2). These data indicate that a substantial proportion of FITC-collagen becomes internalized by live, agonist-stimulated platelets.

To determine whether agonist-induced increases in collagen binding can occur in the absence of collagen/receptor internalization, platelets were first activated, fixed with PFA in this case to prevent internalization, washed, and incubated with FITC-collagen. We found that increased FITC-collagen binding still occurred to TRAP-stimulated, fixed platelets relative to unstimulated, fixed platelets (Figure 3A). To determine whether agonist-induced increases in FITC-collagen binding were specific, TRAP-activated, fixed platelets were incubated with FITC-collagen with or without 100-fold excess unlabeled collagen (Figure 3A). The excess unlabeled collagen significantly inhibited the agonist-induced FITC-collagen binding to below basal levels (P < .05). Similar results were seen with thrombin or ADP-stimulated murine or human fixed platelets (data not shown), indicating that agonist-induced collagen binding in the absence of receptor-mediated internalization is specific.

To determine whether the increased collagen binding that occurs in the absence of receptor recycling is mediated by α₂β₁, ADP-stimulated, fixed platelets were incubated with an α₂-specific blocking mAb before FITC-collagen addition. This mAb, but not an isotype-matched control IgG, prevented an increase in FITC-collagen binding (Figure 3B; P < .05), indicating that the increased FITC-collagen binding is α₂β₁ dependent. This experiment was performed with murine instead of human platelets because we were unable to identify a blocking antihuman α₂β₁ mAb that retained its functional activity to fixed human platelets. These results are also consistent with an internalization-independent component of α₂β₁-mediated collagen binding.

To determine whether the increased collagen binding to platelets on agonist stimulation was simply caused by increased numbers of α₂β₁ integrin molecules on the activated platelet surface, we evaluated expression of α₂β₁ on resting and agonist-stimulated, fixed platelets with a mAb against α₂β₁ not known to be conformationally sensitive. By flow cytometry, there was no significant difference in α₂β₁ staining intensity between resting or TRAP-activated, fixed human platelets (Figure 4). Similar results were also obtained with ADP as the agonist, with either human or murine platelets (data not shown). Taken together, these results demonstrate that there is a component of agonist-induced collagen binding to platelets through α₂β₁ that is independent of both internalization and increased numbers of available α₂β₁ molecules, suggesting an activation of α₂β₁. Therefore, in most subsequent experiments we fixed resting or activated platelets first, before adding FITC-collagen to measure α₂β₁ activation only.

We next examined the potential roles of the α₂ and β₁ cytoplasmic tails in the activation of this integrin by introducing integrin cytoplasmic tail peptides into platelets. We constructed permeable, chimeric peptides with a hydrophobic delivery signal sequence¹⁴  N-terminal to the integrin tail (Table 1). This hydrophobic signal sequence was previously shown to deliver β₃ cytoplasmic peptides into human erythroleukemic cells and fibroblasts.²⁴  Delivery was confirmed in 2 ways, first by evaluating live platelets incubated with FITC-labeled signal (Sig)–β₁ peptide or β₁ peptide without the signal sequence using flow cytometry with trypan blue quenching, and second, by confocal microscopy. The FITC-labeled chimeric peptide appears to enter platelets because it was protected from trypan blue quenching (Figure 5A), whereas the control peptide (β₁ peptide without the signal sequence) exhibited only an extremely low level of background fluorescence (data not shown). As a positive control for quenching in this experiment, platelets were incubated with an α₂β₁ mAb on ice, followed by an FITC-labeled secondary goat antimouse IgG on ice for 30 minutes before flow cytometry. We found extensive quenching of the FITC-conjugated secondary antibody when internalization was inhibited at 4°C. Confocal immunofluorescence microscopy further confirmed that FITC-labeled chimeric peptides were able to fluorescently label platelets, whereas the control peptide lacking the signal sequence did not (Figure 5B). We obtained similar results in platelets incubated with FITC-labeled Sig-α₂ as observed by light microscopy (data not shown).

To determine how the α₂ or β₁ cytoplasmic tails affected collagen binding to platelets, resting platelets were incubated with the native α₂ sequence (Sig-α₂) chimeric peptide followed by fixation and FITC-collagen addition. The Sig-α₂ peptide significantly increased collagen binding to these nonagonist-treated platelets (P < .05; Figure 6), whereas an α₂ cytoplasmic peptide without the signal sequence (data not shown) or a Sig-α₂ scrambled chimeric peptide had no effect (P > .1; Figure 6). In addition, platelets pretreated with the α₂ chimeric peptide before TRAP addition, exhibited only a slight, but not statistically significant, increase in collagen binding above collagen binding observed with TRAP alone, perhaps because this binding was approaching a maximum. A scrambled peptide had no apparent effect on TRAP-induced collagen binding. Thus, these results suggest that the α₂ tail may activate α₂β₁ in resting platelets. In contrast, the β₁ chimeric peptide had no effect on collagen binding to resting or activated platelets (data not shown).

We next asked which domain of the α₂ tail is required for the increase in collagen binding. Because the GFFKR motif is important in α_IIbβ₃ activation and mutation of this conserved region alters the affinity of α_IIbβ₃ for ligands,^25,26  we designed 2 mutant permeable peptides, one lacking GFFKR, called Sig-α₂ without GFFKR, and another containing GFFKR but truncated C-terminal to it, called Sig-KLGFFKR (Table 1). We found that when resting platelets were treated with Sig-KLGFFKR and subsequently fixed before FITC-collagen addition, even more collagen binding occurred than with the full-length α₂ tail (P < .05; Figure 7A). However, Sig-α₂ without GFFKR was unable to induce collagen binding beyond basal levels (P > .1).

A previous study showed that a lipid-modified peptide corresponding to the cytoplasmic region of integrin α_IIb induced platelet activation and aggregation in part via thromboxane A₂ synthesis.²⁷  We therefore asked whether the chimeric Sig-KLGFFKR peptide induced collagen binding to platelets by activating platelets via signal transduction events. Various platelet inhibitors including apyrase, which scavenges released ADP, indomethacin, which inhibits cyclooxygenase and therefore thromboxane A₂ production, and PGI₂, which increases cyclic adenosine monophosphate (cAMP) formation, were added to live platelets before incubation with Sig-KLGFFKR, fixation, and FITC-collagen addition. These inhibitors had no significant effect on Sig-KLGFFKR–induced collagen binding to platelets (Figure 7B), suggesting that this peptide may affect α₂β₁ directly to increase platelet binding to collagen.

To determine whether the Sig-KLGFFKR peptide induces collagen binding specifically through the α₂β₁ integrin, we preincubated live murine platelets with an α₂-blocking mAb or control IgG before incubation with the peptide and FITC-collagen. Collagen binding induced by Sig-KLGFFKR was partly inhibited by the α₂-blocking mAb (P < .05), but not by control IgG (P > .1; Figure 7C). Similar results were seen with live human platelets (data not shown). These results suggest that at least 40% of the Sig-KLGFFKR–induced collagen binding occurs through α₂β₁, but that other collagen binding sites may also contribute, unlike thrombin stimulation of platelets, which appeared to increase collagen binding largely through α₂β₁ (Figure 1A). To determine if the most abundant platelet integrin, α_IIbβ₃, also contributed to this increased collagen binding, platelets were preincubated with 1B5, a function-blocking mAb against the murine α_IIbβ₃ complex. However, this mAb had no effect on Sig-KLGFFKR–induced collagen binding (data not shown).

To test whether the Sig-KLGFFKR peptide also induces collagen internalization into platelets, we preincubated live platelets with this peptide or control buffer. We then added FITC-collagen for various times and measured internalization as protection from trypan blue quenching. These time-course experiments indicated that the majority of Sig-KLGFFKR–induced FITC-collagen binding to platelets was extracellular and quenched by trypan blue (Figure 8). However, when platelets were treated simultaneously with Sig-KLGFFKR and ADP, most of the FITC-collagen was protected from trypan blue quenching and therefore likely internalized (data not shown).

Although activation of the α_IIbβ₃ integrin on platelets has been extensively studied, the activation properties of other platelet integrins are not well understood. Another important integrin on platelets is the collagen receptor, α₂β₁. This integrin is critical for collagen-induced platelet activation and platelet adhesion to collagen, because human platelets lacking this integrin or blockage of this integrin with function-blocking mAbs results in a loss of collagen-mediated platelet activation and adhesion to collagen.^2,28-30  However, agonist-induced activation of α₂β₁ on platelets has not been extensively studied. Intriguing reports from Jung and Moroi^9,16  provide evidence for α₂β₁ integrin activation on platelets as defined by an enhanced binding capacity of α₂β₁ on live, agonist-stimulated platelets for the soluble ligand ¹²⁵I-type III collagen. In the present study, we further analyzed some fundamental properties of α₂β₁ integrin activation, as measured with soluble FITC-labeled type I collagen and flow cytometry. The FITC-labeled type I collagen used in our study was native bovine collagen type I that lacks the telopeptide, which is essential for collagen molecules to cross-link into fibrillar collagen. We found that type I collagen binding to platelets increases with agonist stimulation in a divalent cation- and α₂β₁-dependent manner. Collagen internalization accounts for a significant part of this increase; however, specific collagen binding still occurs under conditions known to block internalization and is not dependent on increased α₂β₁ expression on the platelet surface. These results suggest that a significant component of soluble FITC-collagen binding to platelets is due to activation of α₂β₁, independent of internalization or increases in α₂β₁ expression levels in the platelet surface. Our results also indicate that increased collagen binding to α₂β₁ may be mediated by the GFFKR motif of the α₂ cytoplasmic tail.

It is important to explore the contribution of collagen internalization to α₂β₁-mediated collagen binding because the increased association of soluble ligand with platelets may not be exclusively due to an increased binding capacity of α₂β₁. Because most studies on integrin-ligand interactions are performed in live platelets at room temperature or 37°C, it is possible that the ligand is internalized through an agonist-induced increase in the rate of integrin-mediated receptor recycling. In fact, we previously found that fibrinogen is rapidly taken up via α_IIbβ₃ into activated platelets.¹³  This suggests that the increase in ligand binding to activated platelets could result from a few activated integrin molecules that exhibit an accelerated uptake of ligand, or from activation of a large proportion of integrin molecules that either remain on the surface or are internalized. Ligand-binding studies per se do not necessarily distinguish between these possibilities. By distinguishing internalized from surface-bound FITC-collagen using trypan blue quenching, we demonstrate that internalization contributes to a significant proportion of the increased association of FITC-collagen with platelets upon agonist stimulation (Figure 2), and that internalization is inhibited by an α₂β₁ function-blocking mAb (data not shown). This is consistent with a previous study of human fibroblasts showing that internalization of type I collagen–coated beads is α₂β₁ dependent and that the α₂ integrin subunit is rapidly recycled or resynthesized following a phagocytic load.³¹  The increased association of soluble type I collagen to activated platelets is decreased by an α₂β₁-blocking mAb, which is in agreement with the findings that soluble collagen interacts largely through the α₂β₁ integrin and minimally through GPVI.^9,32  Interestingly, it was found in a previous study that fibrillar collagen could be internalized by platelets, although the receptor mediating internalization was not described.³³  Because GPVI is a platelet collagen receptor that is most responsible for fibrillar collagen binding to platelets,³²  it would be interesting to determine whether fibrillar collagen internalization occurs in platelets via this receptor. One functional consequence of collagen internalization by platelets could be lysosomal degradation, because this is an important mechanism for collagen clearance in other cells.^34,35  The dynamic regulation of collagen internalization by agonist stimulation suggests its potential significance in wound healing, inflammation, and vascular wall remodeling.

It is most likely that the α₂β₁ integrin is activated by agonist stimulation via inside-out signaling, because FITC-collagen binding to agonist stimulated, fixed platelets was still increased independent of internalization, and there was no significant change in α₂β₁ expression on the platelet surface before or after agonist stimulation. Experiments by Jung and Moroi suggest that the α₂β₁ integrin becomes activated when platelets are flowed over collagen, resulting in a firm adhesion to the immobilized collagen.¹⁰  The activation of α₂β₁ by agonist stimulation supports a modified 2-step, 2-site model in which collagen binds initially to α₂β₁ or GPVI or both, and signals from GPVI increase the affinity of α₂β₁ for collagen.³⁶  Mapping the signaling pathways from different agonist receptors such as GPVI or the thrombin or ADP receptor to α₂β₁ is an important next step in understanding the mechanism of platelet adhesion to collagen.

In addition to strengthening an adhesion to collagen, the activation of α₂β₁ on platelets may also contribute to the extracellular collagen degradation pathway by matrix metalloproteinases (MMPs). Upon activation, platelets release pro-MMP-1 and -2 as well as other MMPs^37,38  and MMP-1 activity increases both intracellularly and extracellularly.³⁹  Recent studies demonstrated that pro-MMP-1 complexes with α₂β₁ on keratinocytes and platelets and that α₂β₁ can bind the pro or active form of this protease via the I domain of α₂ and the linker and hemopexin motifs of MMP-1.^40,41  The possibility exists that activation of α₂β₁ on platelets may also regulate MMP activity or release, which is important in inflammation, matrix remodeling, and metastasis.

To further understand how α₂β₁ activation is regulated, we explored the roles of the α₂ and β₁ cytoplasmic tails in this process. Although the roles of integrin cytoplasmic domains in α_IIbβ₃ activation have been explored, only a few studies have examined the roles of these domains in α₂β₁-mediated processes. For example, several groups have shown that the α subunit cytoplasmic domain contributes to the regulation of adhesion, morphology, and motility mediated by α₂β₁ in K562, T47D, and RD cells.^42-44  Because platelets are anucleate and difficult to genetically manipulate, we introduced excess integrin cytoplasmic tails as cell-permeable chimeric peptides. Previous work from Hawiger and Liu et al has established that cell-permeable chimeric peptides of various β₃ cytoplasmic sequences complexed to a hydrophobic delivery sequence enter human erythroleukemic cells or human fibroblasts and regulate the function of α_IIbβ₃.^14,24  Using this approach, we successfully delivered the α₂ and β₁ cytoplasmic tails into platelets. Although we found no effect of the β₁ tail, excess α₂ cytoplasmic peptide increased FITC-collagen binding in resting but not TRAP-activated platelets. Our data are in agreement with Stephens et al,²⁷  who observed that introduction of a portion of the α_IIb cytoplasmic domain increased fibrinogen binding to platelets via α_IIbβ₃. However, our data differ from Vinogradova and coworkers, where myristoylated α_IIb peptides were found to inhibit fibrinogen binding to platelets upon ADP or thrombin stimulation.⁴⁵  Possibilities that may contribute to this difference are that our study involved a different integrin and different peptide modification, which may affect peptide location and function in cells.

We also performed experiments using live platelets preincubated with the full-length α₂ cytoplasmic tail (Sig-α₂), as well as the Sig-α₂ scramble, and an α₂ peptide lacking the signal sequence, before a 30-minute incubation with FITC-collagen. However, we did not find a significant difference in FITC-collagen binding between these peptides (data not shown). Although we believe that Sig-α₂ does activate collagen binding in live platelets, we speculate that these results are more difficult to observe in live platelets, because FITC-collagen itself is a platelet agonist with or without the Sig-α₂ peptide. Therefore, in live, unfixed platelets, any increase in FITC-collagen binding by the Sig-α₂ peptide may be masked by the effects of agonist stimulation by FITC-collagen. This is one of the reasons that agonist-stimulated, fixed platelets were used in many of our studies, such that FITC-collagen would serve only as a readout for active α₂β₁ and not as an agonist.

To further localize the portion of the integrin tail responsible for α₂β₁ activation, we introduced mutant peptides into platelets and found that an excess of peptide containing only the membrane-proximal KLGFFKR motif of α₂ induced substantial collagen binding to resting platelets. The membrane proximal region is highly conserved among α subunits. These results are consistent with previous studies of the α_IIbβ₃ integrin, which showed that a lipid-soluble peptide, palmitoyl-KVGFFKR, induces platelet α_IIbβ₃ activation. However, activation in that case was partly dependent on thromboxane formation,²⁷  whereas we observed that α₂β₁ integrin activation was independent of multiple intracellular signaling events. Our results are also in agreement with another study showing that deletion of most of the α_IIb cytoplasmic tail downstream of KVG, resulted in the expression of constitutively active α_IIbβ₃ in CHO cells,²⁵  suggesting that the distal cytoplasmic tail of α_IIb (FFKRNRPPLEEDDEEGE) exerts a negative regulatory function and locks α_IIbβ₃ in a low-affinity state. Consistent with the importance of the GFFKR motif, we also found that the Sig-α₂ cytoplasmic domain peptide lacking this motif did not affect FITC-collagen binding in either resting or TRAP-activated platelets. Interestingly, collagen binding induced by the Sig-KLGFFKR peptide was not completely blocked by an α₂β₁ function–blocking mAb, suggesting either the contribution of other activated integrin or nonintegrin receptors, or a peptide-induced conformational change in α₂β₁ such that it cannot be blocked by this mAb. However, this peptide-induced collagen binding was not affected by an α_IIbβ₃ function–blocking mAb, suggesting a lack of contribution of this particular integrin to the non–α₂β₁-mediated portion of this increased binding (data not shown).

We noted that while Sig-KLGFFKR also induced a slight increase in collagen internalization as determined by trypan blue quenching, the majority of Sig-KLGFFKR–induced FITC-collagen binding remained on the surface of resting platelets. Interestingly, internalization induced by Sig-KLGFFKR was greatly increased in ADP-activated platelets (data not shown). This enhanced effect may be due to a synergy between the peptide and ADP, that is, a combination of a Sig-KLGFFKR–induced increase in collagen binding and an ADP-induced acceleration of both internalization and collagen binding. These results suggest that this peptide may directly activate the α₂β₁ integrin.

The activation of α₂β₁ on resting platelets by Sig-KLGFFKR may be explained in 2 ways. One possibility is that the KLGFFKR sequence may compete for a molecule bound to α₂ that is necessary to maintain α₂β₁ in a resting conformation. Removal of this molecule may result in integrin activation. This may explain why excess Sig-β₁ peptide did not alter FITC-collagen binding in our study. Although the KLGFFKR peptide has been shown to interact with the 60-kD protein, calreticulin,⁴⁶  calreticulin binding to α₂β₁ is induced by phorbol myristate acetate, suggesting that its binding to α₂β₁ is more likely to stabilize a high-affinity state rather than maintain a resting conformation of this integrin.⁴⁷  The identities of other molecules that bind to the α₂ subunit are currently unknown but their discovery will lead to a better understanding of the regulation of α₂β₁ activation.

A second explanation of our results is that the α₂ and β₁ subunits may interact to maintain α₂β₁ in a resting conformation. The KLGFFKR motif may disrupt the resting conformation by competing for an α₂-binding site on β₁, thereby activating this integrin, consistent with previous studies of α_IIbβ₃⁴⁸  or α_Lβ₂⁴⁹  in which the conserved GFFKR motif in the integrin α cytoplasmic domain helps to maintain the integrin in a low-affinity, resting state. The reason that the full-length Sig-α₂ peptide was less active than Sig-KLGFFKR may be due to decreased accessibility of this peptide to the sites of α₂ and β₁ interaction or adoption of a secondary conformation that masks the KLGFFKR motif. However, this model is inconsistent with the inability of Sig-β₁ to activate α₂β₁, suggesting that the first model may be more likely_.

In summary, our results suggest that the α₂β₁ integrin plays a role in both platelet-collagen binding and collagen internalization in agonist-stimulated platelets. Activation of α₂β₁ most likely occurs because increased collagen binding to α₂β₁ was observed under conditions where internalization was prevented and because no significant increase in α₂β₁ surface expression was observed on activated platelets. We have further determined that the proximal KLGFFKR motif on the α₂ cytoplasmic tail may play an important role in modulating the activation state of α₂β₁ on platelets. Our data provide further information for understanding the function and regulation of α₂β₁ on platelets and potentially on other cell types such as fibroblasts, osteoblasts, and macrophages where this integrin mediates cell migration, invasion, and phagocytosis.

We thank Russ Henry for peptide synthesis, purification, and analysis, and Weiping Yuan for help with mouse platelet experiments.