Characterization of plasminogen as an adhesive ligand for integrins α_Mβ₂ (Mac-1) and α₅β₁ (VLA-5)

Plasminogen (Pg), the zymogen form of the serine proteinase plasmin (Pm), its activators, and inhibitors compose the Pg system, which has been implicated in numerous physiological and pathological processes involving extracellular proteolysis. Broad substrate recognition exhibited by Pm allows it to perform multiple tasks. Pm is the major enzyme responsible for degradation of fibrin clots. In addition, the interaction of Pg and its activators, tissue plasminogen activator (t-PA) and urokinase plasminogen activator (u-PA), with cell surfaces has been proposed to play a pivotal role in assisting cell migration by degrading the extracellular matrix directly or indirectly through activation of metalloproteinases. Numerous processes that involve cell migration and tissue remodeling, such as wound healing, angiogenesis, ovulation, trophoblast invasion, neurite outgrowth, and tumor cell invasion, depend on generation of Pm activity (reviewed by Irigoyen et al¹  and Herren et al² ). Direct evidence for the accessory role of Pg in cell migration was obtained in studies with mice genetically deficient in Pg; leukocyte migration in response to the inflammatory stimulus was significantly diminished.^3,4 

Binding of Pg to fibrin clots or cell surfaces enhances t-PA–induced Pg activation many fold (reviewed by Nieuwenhuizen⁵  and Plow et al⁶ ). The association of Pg with several extracellular matrix (ECM) proteins, including fibronectin, collagen IV, laminin, and fibrin(ogen), also dramatically stimulates its activation,^7,8  suggesting that the ECM proteins may provide alternative surfaces for assembly and regulation of Pg conversion to Pm. The interactions of Pg with the ECM proteins in vitro point to the Pg potential to be deposited within the extravascular interstitial tissues. Pg is not synthesized by other tissues except liver and, therefore, should traverse the endothelial barrier in blood vessels to become associated with the underlying ECM. This might occur during the inflammatory response when plasma proteins leak through the vessels. Deposition of plasma proteins—for example, fibrin(ogen)—is well documented within inflamed tissues.^9,10  Indeed, recent immunohistochemical studies localized Pg in human atherosclerotic coronary arteries in cellular and acellular portions of the plaque.¹¹  The binding of Glu-Pg, the native form of the molecule, and its Lys-Pg derivative, which does not contain the N-terminal 77/78 residues, to cells and the ECM proteins is mediated primarily by low-affinity lysine binding sites (LBSs) present in kringle domains 1, 2, 4, and 5 of Pg.^12-15  The occupation of LBSs induces a large conformational change in Pg and facilitates its activation.¹⁶  LBSs of Pg are responsible for multiple interactions of this protein and are of high capacity and ubiquitous in their nature. A growing list of candidate receptors on the cell surface for soluble Pg and their structural and functional heterogeneity supports the LBS-mediated recognition mechanism. Several proteins with carboxy-terminal lysines (α-enolase, annexin II) have been identified as Pg receptors on the cell surface,^17,18  while other Pg-binding molecules (amphoterin, integrin glycoprotein IIbIIIa [GPIIbIIIa], cytokeratin-8, dipeptidyl peptidase, tissue factor) appear to utilize internal lysines.^11,19-22  The interactions of surface-bound Pg with several proteins involves, in addition to LBSs, some other binding regions,^23,24  suggesting that the alteration of Pg conformation may control recognition.

Because Pg is deposited within the ECM at the sites of vascular injury and can bind multiple cell surface proteins, we have examined whether integrins can engage Pg as an adhesive ligand. Our previous studies with model cells expressing integrin α_Mβ₂ demonstrated that Pg is capable of supporting cell adhesion.²⁵  However, the relationships between Pg/Pm and α_Mβ₂ and/or other integrins on leukocytes that abundantly express this integrin have not been defined. The data presented in this paper show that integrins α_Mβ₂ (CD11b/CD18, Mac-1, CR3) and α₅β₁ (CD49e/CD29, very late antigen-5 [VLA-5]) on leukocytes can bind Glu-Pg and that α₅β₁ is involved in Pg activation.

Human Glu-Pg was purified from fresh human plasma using affinity chromatography on lysine-Sepharose followed by gel filtration as described.²⁶  The N-terminal fragment of Pg (residues 1 to 78) was generated during spontaneous conversion of native Glu-Pg into Lys-Pg due to the traces of active Pm. For that purpose, Glu-Pg (5 to 10 mg/mL) purified without inhibitors was incubated at 4° C for 24 to 26 hours, and the progress of Glu-Pg to Lys-Pg conversion was monitored by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) using 7.5% homogeneous gel (Pharmacia Fast Gel; Pharmacia, Uppsala, Sweden). Its identity was confirmed by N-terminal sequence analysis, and a single EPLDDYVNTQ sequence was determined. Lys-Pg was prepared as described.²⁶  Lys-Pm was from Enzyme Research Laboratories (South Bend, IN). Glu-Pg was labeled with ¹²⁵I using IODO-GEN according to the manufacturer's protocol (Pierce, Rockford, IL). The expression, purification, and characterization of the α_MI-domain as a fusion protein with glutathione S-transferase (GST) have been previously described.²⁷  The free α_MI-domain was prepared by cleavage of the fusion construct with thrombin (Enzyme Research Laboratories, South Bend, IL). Streptokinase, t-PA, N-formyl-methionyl-leucyl-phenylalanine (fMLP), tranexamic acid, and ϵ-aminocaproic acid (ϵ-ACA) were from Sigma (St Louis, MO). The chromogenic plasmin substrate Val-Leu-Lys-p-nitroanilide (S-2251) was from Chromogenix Diapharma Group (Franklin, OH). The peptide P2, corresponding to the γ-chain residues γ377 to γ395 of human fibrinogen (YSMKKTTMKIIPFNRLTIG), was described previously.²⁸  Anti-α_Mβ₂ function-blocking monoclonal antibodies (mAbs) 44a and IB4 recognize the epitopes within the α_MI-domain²⁹  and the β₂ subunit,³⁰  respectively. The mechanism by which mAb IB4 blocks adhesion is not direct and apparently involves the allosteric mechanism.³¹  Monoclonal antibody 1965 (clone JBA1), directed against the β₁ integrin subunit, and polyclonal anti-α₅β₁ antibody 1950 were from Chemicon (Temecula, CA). Polyclonal anti-Pg antibody was from Accurate Chemicals (Westbury, NY), and anti-GST mAb (clone DG122-2A7) was from Upstate Biotechnology (Lake Placid, NY). Monoclonal antibody 8A11, directed against human fibronectin,³²  and mAb 4-2,³³  recognizing human fibrinogen, were described previously.

Human embryonic kidney (HEK) 293 cells expressing α_Mβ₂ and mock-transfected cells were described previously.³⁴  To generate cells that express the “I-less” form of α_Mβ₂, the construct in which the α_M sequence Glu131-Gly321 was deleted from the α_M reading frame was generated by overlap extension mutagenesis.³⁵  Briefly, the 2 pairs of primers were designed, and α_M coding regions 73 to 509 bp and 1083 to 3543 bp were amplified by polymerase chain reaction (PCR) and then fused together in the subsequent PCR reaction. The modified α_M sequence was cloned into pcDNA3.1. HEK 293 cells were stably transfected with a created plasmid and pcDNA3.1-β₂ vector as previously described.²⁵  Erythroleukemic K562 cells were obtained from American Type Culture Collection (ATCC) (Rockville, MD). The cell lines were maintained in Dulbecco modified Eagle medium (DMEM)–F12 (BioWhittaker, Walkersville, MD) supplemented with 10% fetal bovine serum (FBS) and 25 mM HEPES (N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid). U937 monocytoid cells were obtained from ATCC and cultured in RPMI 1640 supplemented with 10% FBS. Neutrophils were isolated under sterile conditions from peripheral blood obtained from consenting volunteers and anticoagulated with acid-citrate-dextrose essentially as described.³⁶ 

The wells of polysterene microtiter plates (Immulon 4HBX, Dynex Technologies, Chantilly, VA) or Costar (Bethesda, MD) were coated with various concentrations of protein ligands for 3 hours at 37° C and postcoated with 0.5% polyvinylpyrrolidone (PVP) for 1 hour at 37° C. The cells were labeled with 10 μM calcein acetoxymethyl ester (calcein-AM) (Molecular Probes, Eugene, OR), and assays were performed as described previously.³⁷  Fluorescence was measured in a fluorescence plate reader (CytoFluorII, Applied Biosystems, Foster City, CA), and the number of adherent cells was determined from a standard curve constructed using the fluorescence of aliquots with a known number of labeled cells.

The binding of Pg to the ECM deposited by fibroblasts was analyzed by immunofluorescent staining. WI38 human lung fibroblasts were obtained from ATCC and were grown in DMEM supplemented with 10% FBS, 2 mM glutamine, and 2 mM sodium pyruvate. Cells were harvested from the flasks with trypsin-EDTA (trypsin–ethylenediaminetetraacetic acid), washed, and cultured on glass coverslips in 12-well plates in DMEM-F12 supplemented with 10% FBS and 2 mM sodium pyruvate for 72 hours. Cells in selected wells were rinsed, and fresh medium supplemented with 0.25 mg/mL fibrinogen was added. A thin film of fibrin that formed on the top of the fibroblast layer after 1 hour was removed, and 0.5 mL of 0.1 mg/mL Pg was added to the wells with and without added fibrinogen. After 1 hour of incubation at 37° C, the media were removed and cells were fixed in 3.7% paraformaldehyde for 20 minutes. After blocking with 3% bovine serum albumin (BSA) in phosphate-buffered saline (PBS) (blocking buffer), primary antibodies directed against human fibronectin, fibrinogen, and Pg diluted in the blocking buffer were added. After 2 hours of incubation, the slides were washed with triethanolamine-buffered saline (TBS) plus 0.05% Tween 20, and the secondary goat antimouse immunoglobulin G (IgG) conjugated to Alexa 594 (1:1500 dilution) or goat antirabbit IgG conjugated to Alexa 488 (1:1500 dilution) was added for 30 minutes. Images of the fluorescent-labeled ECM were obtained using a Leica TCS-NT laser scanning confocal microscope (Leica, Heidelberg, Germany) and HCX Plan Apo × 63 NA 1.4 oil (λ) lenses. Leica's Confocal software was used for image acquisition, and Corel Photo-Paint 8.0 software (Jericho, NY) was used for image processing.

To test the interaction of immobilized Pg or its derivatives with soluble α_MI-domain, 96-well microtiter plates were coated overnight at 4° C with Pg and Pm (100 μL, 5 μg/mL in the presence of 1 mM phenylmethylsulfonyl fluoride [PMSF]) or the N-terminal Pg peptide (100 μL, 20 μM) and postcoated with 0.4% polyvinyl alcohol (PVA). The α_MI-domain (100 μL, 10 μg/mL) as a fusion with GST was added for 1 hour at 25° C, the plates were washed, and anti-GST mAb was added. The bound α_MI-domain was quantitated after the reaction with goat antimouse IgG conjugated with alkaline phosphatase. Nonspecific binding of the α_MI-domain to PVA-coated wells was subtracted.

To measure the binding of soluble Pg to the immobilized α_MI-domain, microtiter plates were coated with free α_MI-domain (100 μL, 5.0 μg/mL) and postcoated with 0.4% PVA. ¹²⁵I-Pg (100 μL per well, 20 μg/mL, 10⁶ cpm) was added with or without the peptide P2 (100 μM), tranexamic acid (5 mM), or ϵ-ACA (50 mM). After 60 minutes of incubation at 25° C, the wells were washed with TBS–Tween 20, the bound Pg was eluted with 200 μLof1%SDS–5 mM dithiothreitol (DTT) at 60° C for 30 minutes, and the radioactivity was measured.

Binding parameters for the interaction of soluble Glu-Pg with the α_MI-domain were measured by using a Biacore 3000 SPR-based biosensor (Biacore, Uppsala, Sweden). The α_MI-domain was coupled with a CM5 chip (Biacore) at a concentration of 1100 response units according to the manufacturer's protocol. Different concentrations of Glu-Pg in HSB-P buffer (Biacore; 10 mM HEPES [pH 7.4], 150 mM NaCl, 0.005% polysorbate 20 adjusted to 1 mM MgCl₂) were flowed over the chip containing the α_MI-domain. All data were corrected for the response obtained using a blank reference flow cell that was activated with 1-ethyl-3-(3-dimethyl aminopropyl) carbodiimide/N-hydroxysuccinimide (EDC/NHS) and then blocked with ethanolamine. Steady-state experiments were performed by injecting the α_MI-domain at 10 μL/min for 8 minutes. The chip surface was regenerated with 2 M NaCl and 50 mM NaOH. Data were analyzed using the BIAevaluation 3.1 program. Nonspecific binding to the blank flow cell was subtracted.

Neutrophils, α_Mβ₂-expressing HEK 293, or U937 cells were washed in Hanks balanced salt solution (HBSS) and resuspended in the same media containing 0.1% BSA. Cells (10⁷/mL) were incubated with ¹²⁵I-Pg (10 μg/mL, 5 × 10⁵ cpm/mL) in the presence or absence of mAb 44a (40 μg/mL), mAb IB4 (30 μg/mL), tranexamic acid (5 mM), and phorbol myristate acetate (PMA) (10 nM) in a total volume 150 μLat 25° C for 30 minutes; 50 μL aliquots were layered over 300 μL of 20% sucrose in HBSS-BSA and centrifuged for 3 minutes in a Beckman microfuge (Beckman Instruments, Fullerton, CA). The tube tips were amputated, and radioactivity was counted in a gamma counter.

A total of 1.5 × 10⁶ U937 monocytoid, α_Mβ₂-expressing, or mock-transfected HEK 293 cells in HBSS were incubated in the solution containing the chromogenic substrate S-2251 (0.7 mM) in the presence or absence of Glu-Pg (7.0 μg/mL), anti-α_M (mAb 44a), anti-α₅β₁ (mAb 1950), anti-β₁ (1965), and t-PA (70 ng/mL) in a total volume of 150 μL at 25° C. Absorbance was measured at 405 nm after 180 minutes, and absorbance of the samples that did not contain cells or Pg was subtracted.

Previous in vitro studies demonstrated that Pg is capable of interacting with ECM proteins, including fibronectin, laminin, collagen IV, and fibrin.^7,8,23,24,38  In 2 recent reports, Pg was found in the atherosclerotic plaque¹¹  and in the zona pellucida of mammalian oocytes.³⁹  However, the pattern of Pg deposition within the ECM has not been investigated. Therefore, we have examined whether Pg localizes with fibronectin and fibrinogen within the ECM produced by live cells using immunofluorescence. WI38 human lung fibroblasts were cultured to confluence in the presence or absence of fibrinogen, and then soluble Glu-Pg was added to cell layers. After several days in culture, fibroblasts develop an extensive extracellular fibronectin matrix (Figure 1A-B). Coculturing of fibrinogen with cells resulted in deposition of fibrin(ogen) into the matrix (Figure 1C). Binding of exogenous Glu-Pg to fibronectin- or fibrinogen-containing matrices was demonstrated by staining the matrix with anti-Pg antibodies (Figure 1E-F). The presence of endogenous Pg was not detected (Figure 1D). The fibrillar pattern of Pg staining was similar to that of fibronectin and fibrinogen, indicating that Glu-Pg binds to these proteins within the ECM and potentially can serve as an adhesive ligand for cell surface integrins.

We have previously demonstrated that integrin α_Mβ₂ can bind Glu-Pg.²⁵  Therefore, adhesion of the α_Mβ₂-expressing HEK 293 cells to immobilized Glu-Pg was examined further. As shown in Figure 2, Pg supported efficient adhesion, which was dependent on the concentration of Pg used. The specificity of the α_Mβ₂-Pg interaction was determined using function-blocking antibodies directed to the α_M (mAb 44a) and β₂ (mAb IB4) subunits. Cell adhesion was reduced by both mAbs (Figure 2, shown for mAb 44a). Monoclonal antibodies inhibited adhesion more effectively at the low Pg concentrations: at 1.5 to 2 μg/mL Pg, mAb 44a inhibited adhesion by about 90% to 75%, whereas about 50% inhibition was achieved at higher coating concentrations of Pg. Specificity of the interaction between α_Mβ₂ and Pg was also confirmed by the finding that at 100 μM, the fibrinogen peptide P2, a specific inhibitor of α_Mβ₂-ligand interactions,^25,28  abolished adhesion to Glu-Pg by 90% (not shown). Mock-transfected cells also adhered to Glu-Pg, but the level of adhesion was about 1.6 times lower than that of the α_Mβ₂-expressing cells, and mAb 44a did not block adhesion. Thus, these results suggest that α_Mβ₂ can mediate adhesion to Pg and that other cell surface molecules may contribute to adhesion (see below).

We next examined whether Pm, the proteolytically active form of Pg, supports α_Mβ₂-mediated adhesion. As shown in Figure 3A, Pm supported a much lower extent of adhesion than Glu-Pg; at 5 μg/mL coating concentration, about 3.5-fold fewer cells adhered to Pm. The low level of cell adhesion mediated by Pm could not be attributed to its proteolytic activity because Pm was inactivated by phenylmethylsulfonyl fluoride (PMSF) prior to immobilization onto the plastic. Because Pm varies from its native zymogen form, Glu-Pg, by the absence of the N-terminal peptide (residues 1 to about 78), we tested cell adhesion to the Lys-Pg form, which also lacks this segment. No cell adhesion to immobilized Lys-Pg was detected at any concentration of coated protein (Figure 3A). In parallel experiments, almost equal signals were obtained using anti-Pg polyclonal antibodies, indicating that similar amounts of Glu-Pg, Lys-Pm, and Lys-Pg were immobilized (not shown). The difference in the adhesion-promoting activity between Glu-Pg and its N-terminally truncated derivatives prompted us to examine the ability of the 78-residue N-terminal Pg domain to support adhesion. When immobilized on microtiter plates, the N-terminal fragment supported efficient adhesion of the α_Mβ₂-expressing cells, and the cell treatment with mAb 44a inhibited adhesion completely (Figure 3B). These data suggested that the N-terminal domain of Pg contains the α_Mβ₂-binding site. However, on a molar basis, about 340-fold higher coating concentrations of the N-terminal fragment were required for maximal adhesion than that of Pg, suggesting that other site(s) in Pg can contribute to its binding to α_Mβ₂.

To probe the structural requirements of α_Mβ₂ binding, the interactions of Glu-Pg, Pm, and the N-terminal fragment with the α_MI-domain were tested. Previous data demonstrated that within α_Mβ₂, the independently folded 24 kDa α_MI-domain is responsible for binding of several ligands.⁴⁰  Initially, we examined the ability of the integrin, in which the α_MI-domain was deleted by genetic manipulation, to interact with Glu-Pg. Adhesion of cells expressing the “I-less” integrin was about 10% of that of wild-type α_Mβ₂-expressing cells, suggesting the requirement of the α_MI-domain in Pg recognition (these experiments were performed in the presence of anti-α₅β₁; not shown). To substantiate the role of the α_MI-domain, Glu-Pg, Lys-Pm, and the N-terminal fragment were immobilized on microtiter plates, and their interactions with soluble α_MI-domain were tested. Each of the proteins bound the α_MI-domain (Figure 4), with Glu-Pg being the most active substrate, and the interactions were inhibited by the fibrinogen peptide P2. Because this peptide binds to the site within the α_MI-domain implicated in binding of several α_Mβ₂ ligands,²⁵  these data suggest that the common ligand-binding site of the α_MI-domain is involved in binding to Pg. However, the additional interactions of the α_MI-domain may account for the incomplete inhibition. Taken together, these observations demonstrate that the α_MI-domain within α_Mβ₂ is the critical structure for Glu-Pg binding.

The expression of cell surface binding sites for Pg was shown previously for many cells, including leukocytes, and was implicated in generating the proteolytic activity required for degradation of the ECM during leukocyte migration (reviewed by Plow et al⁶ ). Therefore, we have investigated whether α_Mβ₂ serves as a cell surface receptor for soluble Pg by measuring the binding of Glu-Pg to the α_Mβ₂-expressing cells in suspension. As shown in Figure 5A, the incubation of cells with ¹²⁵I-Glu-Pg resulted in its binding, but neither anti-α_Mβ₂ mAb inhibited this interaction. Similar results were obtained when the binding of Glu-Pg was tested indirectly by measuring the activity of Pm generated upon the conversion of cell-bound Glu-Pg in the presence of t-PA (not shown). In contrast, the carboxy-terminal lysine analog tranexamic acid, at 5 mM, inhibited the Pg binding completely. This is in agreement with previous findings that the binding of soluble Pg to cell surfaces is mediated by its LBS.⁴¹  Also, when the immobilized α_MI-domain was tested for its ability to bind soluble ¹²⁵I-Glu-Pg, only tranexamic acid and ϵ-ACA inhibited the interaction whereas the P2 peptide was not effective (Figure 5B). To characterize the binding of soluble Glu-Pg with the α_MI-domain further, we have determined the dissociation constant (K_d) for this interaction using surface plasmon resonance (SPR). Figure 5C shows that the binding of Glu-Pg to the α_MI-domain was dose dependent and saturable. The Scatchard plot constructed from the binding data (Figure 5C, inset in the bottom panel) demonstrated the presence of a major class of binding sites with a K_d of 0.3 ± 0.01 μM, which falls within the range reported for the interactions of Pg LBSs with cells (reviewed by Plow et al⁴² ). Taken together, these analyses suggest that soluble Glu-Pg can bind the α_MI-domain entirely through the Pg LBSs and that the specific ligand-binding site of the α_MI-domain is engaged with the immobilized Glu-Pg only. This pattern of binding is consistent with the conclusion that α_Mβ₂ utilizes different mechanisms for binding of the 2 forms of Glu-Pg (immobilized versus soluble). Such differential ligand recognition is a hallmark of many integrins, including α_Mβ₂.^43,44 

The above findings (Figure 2) showed that function-blocking antibodies directed against α_Mβ₂ were not able to inhibit adhesion of the α_Mβ₂-expressing cells to Pg completely. To test whether other integrins are involved in cell adhesion to Pg, mock-transfected HEK 293 cells were tested. The cells efficiently adhered to Glu-Pg but not Pm (Figure 6A). Because HEK 293 cells express several β₁ integrins, with α₅β₁ being the predominant integrin species,³⁴  the effect of function-blocking antibodies against these integrins was tested. Anti-α₅β₁ and anti-β₁ antibodies inhibited adhesion to Glu-Pg by about 50% (shown for anti-α₅β₁ antibody 1950). These antibodies inhibited cell adhesion to fibronectin completely. As a control, anti-α_Mβ₂ blocking reagents (mAb 44a and P2) had no effect on adhesion of HEK 293 cells. To ascribe the contribution of the anti-integrin–resistant component of adhesion to the function of Pg LBSs, cell adhesion was conducted in the presence of tranexamic acid. At 5 mM, tranexamic acid inhibited cell adhesion by 50%, and a combination of anti-α₅β₁ Ab 1950 and tranexamic acid blocked adhesion by 90%. To determine if α₅β₁-mediated adhesion to Glu-Pg occurs with cells other than HEK 293 cells, we tested K562 human erythroleukemic cells. These cells express integrin α₅β₁ and do not express α_Mβ₂. As shown in Figure 6B, Glu-Pg supported efficient adhesion of K562 cells, and adding anti-α₅β₁ antibody 1950 inhibited adhesion by about 75%. In addition, the LBS of Glu-Pg participated in K562 cell adhesion because tranexamic acid alone blocked adhesion by 35%. A combination of anti-α₅β₁ and tranexamic acid decreased adhesion more effectively (95%) than either agent separately.

To extend our findings with transfected cells, we examined whether Glu-Pg and Pm can support adhesion of blood neutrophils. Neutrophils express α₅β₁ and α_Mβ₂, and the latter can be up-regulated further by agonist stimulation. Neutrophils isolated from fresh human blood adhered to both Glu-Pg and Pm, with Glu-Pg being a 2- to 3-fold better substrate. Activation with 100 nM fMLP increased adhesion to Pg and Pm by about 5-fold (not shown). To dissect the contribution of specific integrin-binding sites and LBSs, neutrophil adhesion to Glu-Pg was tested in the presence of tranexamic acid and integrin-specific mAbs. As shown in Figure 7, inhibition of the LBSs decreased adhesion of fMLP-activated neutrophils to Glu-Pg by only 23%, and anti-α_Mβ₂ and anti-α₅β₁ alone produced 66% and 28% inhibition, respectively. The combination of each mAb with tranexamic acid inhibited adhesion more effectively, and the treatment of neutrophils with a mixture of 2 mAbs and tranexamic acid resulted in the almost complete inhibition of adhesion. These results suggest that adhesion of neutrophils to Glu-Pg depends on both integrins α_Mβ₂ and α₅β₁ and that the LBSs of Glu-Pg contribute to adhesion by binding to cell surface proteins. The participation of the 2 integrins and LBSs in adhesion to Glu-Pg was also demonstrable with U937 cells (not shown).

Previous studies demonstrated that Pg binding to cells through the LBSs enhanced its activation by t-PA. To test whether Pg-integrin interactions are accompanied by Pg activation, we have measured the Pm activity generated in the cell suspension containing Glu-Pg in the presence or absence of anti-α_M, anti-α₅β₁, or anti-β₁ function-blocking antibodies. Anti-α₅β₁ and anti-β₁ inhibited Glu-Pg activation in a concentration-dependent manner on the 3 cell lines tested, including the α_Mβ₂-expressing, mock-transfected HEK 293, and U937 monocytoid cells (Figure 8, shown for U937 cells). The maximal inhibition achieved with both antibodies was about 50% to 55%. The binding of Glu-Pg to α₅β₁ on these cells appears to be RGD independent because the RGD peptide at 100 μM did not inhibit Glu-Pg activation. With each cell line, the function-blocking mAb 44a (against α_M) did not affect Glu-Pg activation whereas tranexamic acid decreased the activation by 55% ± 6%. Although addition of t-PA stimulated Glu-Pg activation, it did not change the inhibitory effect of anti-α₅β₁ (not shown). The ability of anti-α₅β₁ antibody to interfere with Glu-Pg activation indicates that in contrast to α_Mβ₂, α₅β₁ can recognize both immobilized and soluble Glu-Pg. Moreover, because generation of the Pm activity was detected in the absence of exogenously added Pg activators, the data suggest that α₅β₁ is involved in Pg activation.

This study was designed to examine the ability of human Pg to support cell adhesion by binding to the cell surface integrin receptors. The main findings are that (1) Glu-Pg can be deposited into the extracellular matrix in which it localizes with the 2 major matrix proteins, fibronectin and fibrinogen; (2) Glu-Pg is capable of supporting cell adhesion through integrins α_Mβ₂ and α₅β₁; (3) the Pg LBSs contribute to adhesion by binding cell surface lysines; and (4) cell-associated Glu-Pg activation is inhibited by anti-α₅β₁ mAb. The ability of Pg to support adhesion has been demonstrated with various cells expressing both α_Mβ₂ and α₅β₁, including neutrophils, cultured U937 monocytoid cells, and genetically engineered cells.

Recent studies of Pg interactions have focused on the binding of soluble Pg to cells and provided a good understanding of events that occur during the assembly of the Pg system on cell surfaces. Similar to other proteins such as fibrinogen, vitronectin, fibronectin, and others, plasminogen can exist in 2 states: as a soluble plasma protein and in association with insoluble extracellular matrix. Previous in vitro studies demonstrated that Pg and its activators, t-PA and u-PA, bind to the components of the ECM and the basement membrane.^7,8,23,24  Pg also can be found within inflamed tissues¹¹  where it was apparently derived from circulation. Although the predominant source of Pg is the liver, recent studies demonstrated a broad expression of Pg mRNA in murine⁴⁵  and several human tissues⁴⁶  and hypothesized that the extrahepatic organs may be local sources of Pg in vivo. Therefore, cell binding to Pg incorporated into the ECM could have been envisioned; however, this potential mechanism of cell interactions with Pg has not been distinguished in previous studies. Accordingly, we have examined the role of Pg in supporting cell adhesion mediated by integrins.

We demonstrated that immobilized Pg supports efficient cell adhesion, and integrins α_Mβ₂ and α₅β₁ contribute to this process. Identification of α_Mβ₂ as a receptor for Pg was based on inhibition of cell adhesion by function-blocking mAbs to this receptor and on the interaction of Pg with the recombinant α_MI-domain. The integrin α_Mβ₂ is expressed mainly on phagocytic leukocytes and mediates important leukocyte functions during the inflammatory responses.⁴⁷  α_Mβ₂ is a multiligand receptor that recognizes a variety of different molecules, including many ECM proteins (reviewed by Yakubenko et al²⁵ ). Several proteases, including neutrophil elastase,⁴⁸  myeloperoxidase,⁴⁹  and catalase,⁴⁴  have been previously demonstrated to bind α_Mβ₂. The present study extends this group of α_Mβ₂ ligands to include Pg. Within α_Mβ₂, the α_MI-domain is primarily responsible for its ligand-binding properties⁴⁰  and, accordingly, Pg falls into the category of ligands that bind this domain. Also, consistent with previous findings of ligand-binding properties of β₂ integrins, the related leukocyte integrin α_Lβ₂ (CD11a/CD18, lymphocyte function-associated antigen-1 [LFA-1]), which has a narrow ligand-binding specificity, does not bind Pg.²⁵ 

We have found a distinction in the way immobilized and soluble Pg engage the α_MI-domain. The interaction between the immobilized Pg and α_Mβ₂ was mediated by a specific ligand-binding site in the α_MI-domain and the complementary site in Pg inasmuch as it was blocked by the specific inhibitor of α_Mβ₂, the peptide P2. In addition, the Pg LBSs contribute to the interaction with the α_MI-domain because lysine analogs tranexamic acid or ϵ-ACA blocked cell adhesion to Pg. In contrast, binding of soluble Pg to immobilized α_MI-domain was mediated only by the Pg LBSs because tranexamic acid and ϵ-ACA completely inhibited this interaction whereas α_Mβ₂ blocking agents were not effective. Tranexamic acid also fully blocked the binding of soluble Pg to cells in suspension, suggesting that even if α_Mβ₂ does participate in Pg binding, this should involve α_Mβ₂ lysines. The proposed molecular interactions of α_Mβ₂ with 2 forms of Pg are depicted in Figure 9. Previous studies demonstrated that stimulation of neutrophils with PMA enhanced their Pg binding rapidly and substantially.⁵⁰  This up-regulation was not dependent upon the proteolytic remodeling of cell surface proteins, leading to the increased availability of the C-terminal lysine residues. Notably, neutrophil surface expression of α_Mβ₂ is up-regulated 3- to 10-fold as a result of its mobilization from internal pools in response to agonist stimulation. However, although α_Mβ₂ may potentially contribute to accumulation of Pg on the cell surface, this appears to involve the Pg LBSs because anti-α_Mβ₂ mAbs did not block the binding of soluble Pg to cells and did not interfere with generation of the Pm activity by cells.

The α_Mβ₂-mediated adhesion to Pg was markedly higher to Glu-Pg than that to Lys-Pg and Lys-Pm. The Lys-Pg form is generated by the cleavage of an about 8 kDa N-terminal fragment from intact Glu-Pg by Pm. The 2 forms of Pg have different conformations,⁵¹  and Lys-Pg is more readily activated by activators.^52,53  When tested with the α_Mβ₂-expressing cells, the N-terminal Pg fragment supported efficient adhesion and also directly bound to the recombinant α_MI-domain. Because the N-terminal fragment does not contain the LBS, its interaction with the receptor appears to be mediated by a specific integrin-binding site. However, the N-terminal domain may not be the only binding site for α_Mβ₂ in Glu-Pg because higher molar concentrations of the N-terminal fragment than those of Glu-Pg were required to support effective adhesion. The difference in adhesion supported by Glu-Pg and Lys-Pg speculatively can be attributed to alterations in the conformation. The precedents of modulation of cell adhesion by the conformational states of proteins are numerous (reviewed by Lishko et al³³ ). For example, thrombospondin supports platelet adhesion when coated onto the surface in the presence of Ca²⁺, whereas it becomes essentially nonadhesive as a calcium-depleted conformer.⁵⁴ 

Integrin α₅β₁ was identified as the second integrin on the surface of HEK 293 cells that contributes to adhesion to Pg. The role of this integrin in adhesion to Glu-Pg was extended with neutrophils, U937 monocytoid cells, and K562 erythroleukemic cells. Integrin α₅β₁ is a receptor for fibronectin⁵⁵  and fibrin(ogen),⁵⁶  and α₅β₁-mediated cell adhesion to these ligands is inhibited by RGD-containing peptides that duplicate sequences in these ligands. In addition, recent studies demonstrated the non-RGD–dependent binding of α₅β₁ to fibrinogen.⁵⁷  Glu-Pg does not contain RGD and, thus, further studies are needed to define the binding specificity of this novel α₅β₁ ligand. In contrast to α_Mβ₂, α₅β₁ appears to engage both immobilized and soluble Glu-Pg (Figure 9) because anti-α₅β₁ but not anti-α_Mβ₂ mAbs inhibited generation of Pm activity by cells. Although the α₅β₁-dependent conversion of Pg to Pm by cultured cells was detected, the mechanism by which α₅β₁ contributes to Pg activation remains to be elucidated. In addition, it is unlikely that this process would occur on the surface of normal nonstimulated circulating leukocytes in the absence of Pg activators.

Previous studies provided evidence of a direct involvement of Pg in migration of phagocytic leukocytes in response to inflammatory stimuli,^3,4  and the Pg system was also implicated in assisting migration of other cells in a variety of physiopathological processes.^1,2  However, the precise mechanism by which Pg participates in cell migration is unknown. The relationships between Pg in the ECM and Pg bound to cell surface receptors, including integrins, is also not clear. Because both cell surface receptors and the ECM proteins can bind Pg, a question arises as to whether migrating leukocytes, which carry large amounts of Pg, are capable of exchanging it with the ECM. The interactions of Pg with cells and the ECM are mediated primarily by weak LBSs and, therefore, it is interesting to speculate that Pg can be carried over from the cell surface onto the ECM and/or in the opposite direction. In this regard, we recently observed that fMLP-activated neutrophils picked ¹²⁵I-Pg from the fibrin/¹²⁵I-Pg substrate to which they adhered in the presence of nonlabeled Pg (V.K.L., unpublished results, 2002).

In summary, the present studies revealed the interaction of Pg with integrins α_Mβ₂ and α₅β₁. The physiological role of this association remains unclear, and further studies should determine the relationships between Pg and integrins in leukocyte adhesion and migration and the role of integrins in Pg activation.