Mechanism and effect of thrombospondin-4 polymorphisms on neutrophil function

Recent high-throughput genotyping has sought to identify single nucleotide polymorphisms (SNP) associated with familial coronary artery disease and myocardial infarction (MI).^1-3  One such study detected a highly statistical association between SNPs in 3 of the 5 members of the thrombospondin (TSP) family and MI.¹  The disease-associated TSP-4 SNP, a proline at position 387 (P387) rather than the predominant alanine (A387), showed very strong association with MI (adjusted odds ratio, 1.89) and occurred at high frequency (34%) in a white population.¹  This association was particularly surprising because no prior data had linked TSP-4 with MI; indeed, very little was known about the function of TSP-4 in vivo. Nevertheless, the association of TSP-4 (P387) with MI has been corroborated in several independent studies^4-7  and strongly supports the importance of TSP-4 in cardiovascular pathology.

The thrombospondin family is composed of 5 extracellular matrix glycoproteins. TSP-1 and TSP-2 are the best characterized members of the family and are implicated in diverse responses, including platelet aggregation, inflammatory processes, and regulation of angiogenesis (reviewed in Adams,⁸  Adams and Lawler,⁹  and Bornstein et al¹⁰ ). TSP-4 mRNA was first identified in heart and skeletal muscles,¹¹  and cellular expression of its mRNA and/or protein has been demonstrated in endothelial and smooth muscle cells from brain and heart.¹²  Of the limited information on its functions, TSP-4 has been shown to support myoblast adhesion¹³  and neurite outgrowth¹⁴  and to interact with certain matrix proteins.¹⁵  We have reported that the TSP-4 variants exert distinct effects on endothelial cells with the proatherogenic TSP-4 variant, (P387), failing to support endothelial cell adhesion and proliferation.¹²  However, not only endothelium but also leukocytes are crucial to the inflammatory response underlying atherosclerosis. In early stages of the disease, monocytes and neutrophils infiltrate the intima and generate reactive oxygen species (ROS), which provoke oxidative damage to vascular cells.^16-18  Indeed, epidemiologic studies indicate that peripheral blood leukocytes, particularly neutrophils, correlate with coronary atherosclerosis and MI.¹⁹  In addition, neutrophils are activated in patients with unstable angina and MI and accumulate in eroded atherosclerotic plaques and sites of endothelial cell denudation immediately following bypass grafting.^20-22  In view of these findings, we sought to characterize and compare interactions of TSP-4 and its variants with human neutrophils.

Inhibitors of kinases were purchased from Tocris (Ellisville, MO). Antiphosphotyrosine (4G10) and anti-GST (glutathione S transferase) monoclonal antibodies (mAbs) were from Upstate Biotechnology (Lake Placid, NY), PathScan TM Multiplex Western Cocktail II, all anti-p38 mitogen–activated protein kinase (MAPK) Abs were from Cell Signaling Technology (Beverly, MA). Anti-α_L mAb was from Dako (Carpinteria, CA). The mAbs to α_M (44a), β₂ (IB4), and to major histocompatibility complex 1 (MHC-1; W6/32) were from American Type Culture Collection (Rockville, MD). mAbs against other integrins were from Chemicon International (Temecula, CA). Goat anti–human TSP-4 was from Santa Cruz Biotechnology (Santa Cruz, CA). Neutrophil inhibitory factor (NIF) was provided by Corvas International (San Diego, CA). The Cyquant-Cell Proliferation Kit and Amplex Red Hydrogen Peroxide/Peroxidase Assay Kit were from Molecular Probes (Eugene, OR). Quantikine Immunoassay Kits for human interleukin 8 (IL-8), tumor necrosis factor α (TNF-α), and macrophage inflammatory protein 1α (MIP-1α) were from R & D Systems (Minneapolis, MN).

Full-length TSP-4 variants, TSP-4 (A387) and TSP-4 (P387), were purified from the supernatants of human embryonic kidney 293 (HEK293) cells stably expressing the proteins as described.¹²  TSP-4 variant fragments, which harbor residue 387 and consist of 3 type 2 (epidermal growth factor [EGF]–like) repeats (3EGF–TSP-4; amino acid residues 325-491) were expressed as secreted glutathione S transferase (GST)–fusion proteins in HEK293 cells and purified by affinity chromatography on a column of anti-GST.^12,23  Purification of all TSP-4 variants was performed in endotoxinfree conditions, and endotoxin levels were undetectable (< 0.1 EU/mL using QCL-1000 Limulus Amebocyte Lysate Assay; Bio-Whittaker, Walkersville, MD). Protein preparations were monitored for purity by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and Western blotting as described.¹²  Both variants were properly oligomerized and did not contain aggregates.¹²  In addition, protein bands detected in SDS-PAGE were analyzed by mass spectrometry and were identified as TSP-4. Representative protein preparations were also subjected to the amino acid analysis, and the relative ratios of amino acids confirmed the purity of the TSP-4 preparations.

The TSP-4 peptides containing the SNPs and the predicted Ca²⁺-binding sites (TDIDECRNGACWPN and TDIDECRNGPCWPN) were synthesized in the Cleveland Clinic Biotechnology Core. Characterization of the peptides was done by amino acid analysis (W. K. Warren Medical Research, Oklahoma City, OK).

Granulocytes were isolated from human peripheral blood of healthy volunteers, collected under informed consent, drawn into sterile acid-citratedextrose (1/7 vol 145 mM sodium citrate, pH 4.6, and 2% dextrose) as described.²⁴  Approval for the study was obtained from the Cleveland Clinic Foundation Institutional Review Board in accordance with federal regulations. Informed consent was provided according to the Declaration of Helsinki.

HEK293 cells expressing the recombinant integrins α_Mβ₂, α_Lβ₂, chimeric α_Lβ₂, in which the I-domain of the α_L subunit was replaced by the I-domain of α_M (α_MI-α_Lβ₂), and mock (empty vector-transfected) were developed and maintained as previously described.^25,26 

The adhesion of resting or phorbol 12-myristate 13-acetate (PMA)–stimulated (1 nM PMA for 20 minutes at 37°C) polymorphonuclear leukocytes (PMNs) and HEK293 cells to 96-well tissue culture–nontreated plates (Falcon; Becton Dickinson, San Diego, CA) coated with full-length and 3EGF–TSP-4 variants (0-160 nM in phosphate-buffered saline) was performed as described.^24,27  In inhibition experiments, cells were pretreated with selected antibodies or other reagents for 20 minutes at 22°C. In all adhesion experiments with 3EGF–TSP-4 variants, control wells were coated with GST at the same molar concentrations as the 3EGF–TSP-4, and the data are presented on subtraction of background adhesion to GST, which did not exceed 15% of cell adhesion to the entire 3EGF–TSP-4. The number of adherent cells was quantified using the Cyquant Cell Proliferation assay kit as described.^24,27  Alternatively, cells were processed for H₂O₂ release, cytokines secretion, or Western blot analysis.

Cell migration was measured as described,^24,27  using Costar 24-transwell plates with 3-μm pore polycarbonate filters (Corning, Corning, NY) with full-length TSP-4 (0-350 nM) immobilized on their lower surface. Cells (2 × 10⁵/well) were added to the upper chamber and incubated in a humidified incubator at 37°C and 5% CO₂ for 6 hours. For inhibition experiments, the cells were pretreated with the function-blocking mAbs (20 μg/mL) for 20 minutes at 22°C.

PMA-stimulated PMNs were allowed to adhere to immobilized TSP-4 or the 3EGF–TSP-4 variants (0-300 nM) for up to 4 hours at 37°C in Hanks balanced saline solution containing 1 mM Mg²⁺, 1 mM Ca²⁺, and 0.1% bovine serum albumin (BSA). H₂O₂ was measured in the supernatants of the adherent PMNs using the Amplex Red Hydrogen Peroxide/Peroxidase Assay Kit.

Lysates of adherent PMNs were prepared,²⁸  and equal amounts of protein were analyzed by Western blotting using selected antibodies and developed with enhanced chemiluminescence (Pierce Chemicals, Rockford, IL). Tyrosine phosphorylation of proteins in Western blots was quantified by laser scanning densitometry using Photoshop (Adobe Systems, San Jose, CA) and the National Institutes of Health (NIH) Image software (Research Services Branch, National Institutes of Health, Bethesda, MD).

Human IL-8, TNF-α, and MIP-1α secreted by PMNs adherent to the full-length or 3EGF–TSP-4 variants (100 nM) were measured in supernatants by enzyme-linked immunosorbent assay (ELISA) using Quantikine Immunoassay Kits (R & D Systems).

PMA-stimulated human PMNs were seeded onto Lab-Tek Cell Culture second-generation (CC2) Chamber Slides (Nalge Nunc International, Rochester, NY) previously coated with full-length or 3EGF–TSP-4 variants (160 nM, 2 hours, 37°C) at 5 × 10⁵ cells/well in Dulbecco minimal essential medium (DMEM) F-12 media and incubated for 45 minutes (for α_M staining) or 30 minutes (for phospho-p38 MAPK staining) at 37°C, 5% CO₂. Nonadherent cells were removed by washing, and adherent cells were fixed with 4% paraformaldehyde for 20 minutes at 22°C and stained with biotinylated mAb to α_M subunit (clone 44a) or anti–phospho-p38 MAPK for 30 minutes at 22°C followed by incubation with Alexa 488–conjugated avidin (1:500) or Alexa 488–conjugated goat anti–rabbit immunoglobulin G (IgG) for 30 minutes at 22°C (Molecular Probes). The slides were mounted using Vectashield mounting medium (Vector Laboratories, Burlingame, CA) and observed by fluorescence microscopy. All images were analyzed using a Leica DMR microscope equipped with 5 ×/0.12 NA, 10 ×/0.4 NA, 20 ×/0.5 NA, or 40 ×/0.7 NA objective lenses with numeric apertures ranging from 0.12 to 0.7 (Leica Microsystems, Wetzlar, Germany). Images were photographed with a Qimaging Retiga ExiFas camera (Qimaging, Burnaby, BC, Canada) using Image Pro 5.1 software (Media Cybernetics, Silver Spring, MD). All images were processed with Adobe Photoshop 5.0 (Adobe Systems, San Jose, CA).

α_MI-domain was expressed in Escherichia coli as GST fusion protein,^25,26  and the GST tag was removed by treatment of glutathione Sepharose-bound GST-α_MI-domain with factor Xa (50 U) for 4 hours at 22°C. Microtiter wells (Corning Costar, Cambridge, MA) were coated with GST-free α_MI-domain at 50 μg/mL for 20 hours at 4°C and blocked with 3% BSA for 1 hour at 37°C. 3EGF–TSP-4 variants containing the GST tag (0-1.5 μM) were added to the wells in the presence of 1 mM MgCl₂/1 mM CaCl₂ and incubated for 4 hours at 37°C. Plates were washed with Tris (tris(hydroxymethyl)aminomethane)–buffered saline (TBS)/1 mM MgCl₂/1 mM CaCl₂, and bound 3EGF–TSP-4 variants were detected by an ELISA using anti-GST mAb (Upstate Biotechnology), alkaline phosphatase (AP)–conjugated secondary antibody and para-nitrophenylphosphate (Pierce Chemicals) as the substrate, and absorbance at 405 nm (A₄₀₅) was measured. When the interaction of full-length TSP-4 variants with α_MI-domain was examined, the bound TSP-4 was detected with goat anti–TSP-4 (Santa Cruz Biotechnology). In binding isotherm studies, input concentrations of the GST–3EGF–TSP-4 variants required for half-maximal binding to immobilized α_MI-domain were estimated using the Sigma Plot software (SPSS, Chicago, IL) in which the data were fitted to the one-site binding equation.

The data are expressed as means ± SEMs. To determine the significance of differences between 2 groups, Student t tests were performed using the SPSS; P value less than .05 was considered significant.

We evaluated the capacity of the TSP-4 variants to support the adhesion and migration of isolated human PMNs. Both TSP-4 variants failed to support adhesion of nonstimulated PMNs (Figure 1A). However, when the PMNs were stimulated with PMA, the cells adhered in a dose-dependent manner. The extent of adhesion of the stimulated PMNs at the various concentrations of the TSP-4 variants was not significantly different. Similar results were obtained in cell migration assays. PMA-stimulated PMNs, but not resting cells, migrated toward both TSP-4 variants; no difference was observed in the extent of PMN migration induced by the 2 TSP-4 variants (Figure 1B).

In view of the role of integrins in recognition of other TSP family members and their importance in PMN adhesion to numerous substrates, we investigated the role of these receptors in TSP-4 recognition. Function-blocking mAbs against various integrin β subunits were tested as inhibitors of adhesion of PMA-stimulated PMN adhesion to the TSP-4 variants (Figure 2A). The various mAbs tested had very similar effects on adhesion to both TSP-4 variants. mAb IB4, an anti-β₂ blocking mAb, completely inhibited adhesion. Additionally, a blocking mAb to the β₃ subunit (only α_Vβ₃ is expressed by PMNs), reduced PMN adhesion to both variants by approximately 50%. Blocking mAbs to β₁ and β₅ subunits and a control mAb to unrelated surface antigen, MHC-1, had no effect.

Similar experiments were performed in cell migration assays (Figure 2B). When PMA-stimulated PMNs were preincubated with the mAb to the β₂ integrin subunit (IB4), their migration to both TSP-4 variants was completely blocked. In contrast to the adhesive response, a blocking mAb to the β₁ integrin subunit attenuated PMN migration, but the mAb to α_Vβ₃ had no effect. mAbs to β₅ integrins or the control mAb had no effect on PMN migration.

Known PMN functions dependent on the β₂ integrins are mediated primarily by α_Mβ₂ (macrophage antigen 1 [Mac-1]), α_Lβ₂ (lymphocyte function antigen 1 [LFA-1]), and α_Xβ₂ (p150, 95). We sought to identify which of these β₂ integrins is responsible for TSP-4 recognition. Function-blocking mAbs to the α_M (clone 44a) and to β₂ subunit (clone IB4) completely suppressed PMN adhesion to both TSP-4 variants, whereas blocking mAbs against α_L and α_X failed to reduce PMN binding (Figure 2C, left) NIF, a high-affinity α_Mβ₂-specific ligand, blocks recognition of many α_Mβ₂ ligands, and it attenuated PMN adhesion to both TSP-4 variants in a dose-dependent manner (Figure 2C, right). Inhibition by NIF was complete at a concentration of approximately 10 nM. Complete inhibition by mAb 44a and NIF also was observed in cell migration assays (data not shown).

To verify the role of α_Mβ₂ in recognition of TSP-4, we assessed the ability of HEK293 cell lines expressing β₂ integrins to adhere to both TSP-4 variants. HEK293 cells expressing α_Mβ₂ adhered well to both TSP-4 variants, whereas control (vector-transfected) cells or cells expressing α_Lβ₂ exhibited negligible adhesion (Figure 2D). The interaction of most protein ligands of α_Mβ₂ depends on a region of approximately 200 amino acids, the α_MI-domain, in the α_M subunit. We have previously described a chimeric receptor in which the α_LI-domain in α_Lβ₂ was replaced with the α_MI-domain.²⁶  HEK293 cells expressing this chimeric receptor acquired the capacity to adhere to the TSP-4 variants as efficiently as the α_Mβ₂ cells (Figure 2D), indicating that the α_MI-domain is a major recognition site for TSP-4. Indeed, recombinant-purified α_MI-domain bound both TSP-4 variants in a dose-dependent, saturable manner (Figure 2E). Using the Sigma Plot software, half-maximal binding of TSP-4 (A387) and TSP-4 (P387) to α_MI-domain occurred at input TSP-4 concentrations of 320 ± 46.4 and 102 ± 19.3 nM (means ± SEMs; n = 5), respectively. Taken together, these results indicate that TSP-4 is an adhesive substrate of α_Mβ₂ and recognized by the α_MI-domain and that the TSP-4 variants are equivalent in their support of PMN adhesion.

Although PMNs adhered equally to the TSP-4 variants, cells adherent to the pathogenic TSP-4 (P387) variant were significantly more spread and showed higher expression and more clusters of the α_Mβ₂ integrin than did cells adherent to TSP-4 (A387). The PMNs adherent to TSP-4 (P387) were 2-fold larger (243 ± 8.9 μm²/cell) (Figure 3B, left) and 25% brighter (69 ± 4.7 RFU [relative fluorescent unit] staining with mAb 44a) (Figure 3B, right) as compared with those adherent to TSP-4 (A387) (127 ± 7.45 μm²/cell and 56 ± 4.3 RFU, respectively). Thus, despite the similarities in α_Mβ₂-mediated adhesion to the TSP-4 variants, on engagement, PMNs responded differently to the 2 substrates.

The inflammatory hypothesis of atherosclerosis emphasizes the key roles of leukocyte products in pathogenesis.²⁹  Because oxidative stress is a hallmark of inflammation, we measured the release of H₂O₂, as a representative ROS, from PMNs adherent to the TSP-4 variants. When stimulated with PMA, the 2 TSP-4 variants induced H₂O₂ release in a dose-dependent manner (Figure 4). However, the extent of H₂O₂ production was not the same; TSP-4 (P387) induced approximately 4-fold more H₂O₂ release than TSP-4 (A387). The concentrations of TSP-4 (P387) and TSP-4 (A387) giving a half-maximal response (EC₅₀) was 4.7 ± 0.4 nM and 21.3 ± 1.8 nM, respectively, in 4 experiments performed with PMNs isolated from 4 blood donors, indicating that the P387 variant is a more potent enhancer of the oxidative burst than the A387 variant. An oxidative burst was not observed in nonstimulated PMNs in the presence of either TSP-4 variants, likely because of a lack of adhesion. The differences in oxidative burst triggered by the TSP-4 variants were observed at low PMA concentrations (1-4 nM). When higher concentrations of PMA were added, PMNs released more H₂O₂; the maximal oxidative burst (17 nM H₂O₂/10⁵ cells) at 40 nM PMA was similar on both TSP-4 variants. Next, we examined the production of cytokines from PMNs on exposure to the TSP-4 variants. Secretion of IL-8, TNFα, and MIP-1 was measured. Of these, IL-8 secretion was differentially regulated by the 2 TSP-4 variants, a 2.5-fold increase in its secretion from PMNs adherent to the TSP-4 (P387) variant as compared with TSP-4 (A387) was observed (Table 1). Thus, the TSP-4 (P387) variant enhances proinflammatory functions of human neutrophils.

Because the differences in the functional responses of PMN to the TSP-4 variants were α_Mβ₂ mediated, we compared a downstream signaling event, tyrosine phosphorylation, from integrin engagement. PMNs were allowed to adhere to the 2 TSP-4 variants or to PVP. Equivalent amounts of proteins from cell lysates were subjected to SDS-PAGE and Western blotting with an antiphosphotyrosine mAb (Figure 5A). Phosphorylation of the 48- to 55-kDa proteins was observed in the lysates obtained from cells adherent to the TSP-4 variants at 30 minutes but not to PVP. More importantly, phosphorylation of these proteins was much more robust in PMNs adherent to TSP-4 (P387) than to TSP-4 (A387). Additionally, only TSP-4 (P387), but not TSP-4 (A387) or control PVP, induced prominent phosphorylation of proteins at 36 to 42 kDa. TSP-4 (P387) also triggered more intense phosphorylation of the 28- to 30-kDa proteins as compared with TSP-4 (A387) and PVP. These distinctions in tyrosine phosphorylation were observed with the TSP-4 variants at early time points (5-30 minutes). By 60 minutes, substantial dephosphorylation of the proteins had occurred on all substrates (not shown).

Antibodies to specific phosphoproteins were used to identify the target proteins that were differentially affected by the TSP-4 variants. Adhesion of PMNs to either TSP-4 variant did not induce phosphorylation of Akt (protein kinase B) or p53. Among the 3 MAPK pathways, p38MAPK and stress-activated protein kinase (SAPK)/JNK were activated by TSP-4 (P387), but p44/42MAPK (extracellular signal–regulated kinase 1 and 2 [ERK 1/2]) was not phosphorylated. p38MAPK accounted for 1 of the 36- to 42-kDa phosphoproteins (Figure 5B). This kinase was activated exclusively by TSP-4 (P387) and not by TSP-4 (A387) at 5 and 30 minutes. The lower panel of the figure indicates equal loading of total p38MAPK in each sample. These data were confirmed by immunostaining of adherent neutrophils with a mAb to phosphorylated p38MAPK (Figure 5C). PMNs adherent to TSP-4 (P387) were positive, whereas cells adherent to TSP-4 (A387) or PVP were not. HSP27 and STAT1 were more extensively phosphorylated on PMN adhesion to TSP-4 (P387) (Figure 5D). These latter proteins were phosphorylated at the 15- but not at the 5-minute time point, which suggests that their activation occurs downstream of the early activation of p38MAPK.³⁰  In addition, SAPKs and JNK were more robustly phosphorylated on 15 minutes of adhesion to TSP-4 (P387) as compared with TSP-4 (A387) (Figure 5D). Thus, the proatherogenic variant TSP-4 (P387) induces activation of signaling pathways that respond to environmental stresses and inflammatory cytokines (reviewed in Kyriakis and Avruch³¹  and Dong et al³² ).

PMNs were pretreated with various inhibitors of key kinases: protein kinase A (KT5720), p38MAPK (SB202190), mitogen-activated protein kinase kinase 1 (MEK-1) (PD 98059), phosphatidylinositol 3 (PI3) kinase (Wortmannin, LY294002), and then activation of p38MAPK and H₂O₂ release were measured on adhesion to TSP-4 (P387). When PMNs were pretreated with inhibitor of MAPKK (PD98059), which is upstream of extracellular-regulated kinases 1/2, a very modest effect on p38MAPK phosphorylation and H₂O₂ production was observed (Figure 6A-B). In contrast, a p38MAPK inhibitor (SB202190) blocked its activation (Figure 6A) and completely suppressed H₂O₂ release (Figure 6B), indicating that this kinase is critical for TSP-4 (P387)–dependent H₂O₂ generation. PKA inhibition also abrogated these 2 events, consistent with data indicating that PKA activation occurs upstream of p38MAPK.³³ 

Next, we sought to determine which of the 2 integrins, α_Mβ₂ or α_Vβ₃, implicated in TSP-4 (P387) adhesion (see Figure 2A) was critical for observed responses. Function-blocking mAbs to α_M and β₂ subunits completely suppressed p38MAPK phosphorylation, whereas mAbs inhibiting TSP-4 interaction with α_Vβ₃ failed to reduce phosphorylation of p38MAPK (Figure 6C). In addition, RGD peptides, which block ligand interaction with α_Vβ₃, did not inhibit p38MAPK activation. The lower panel of Figure 6C shows equal loading of total p38MAPK despite quantitative differences in adhesion of untreated and mAb-treated cells. In addition, function-blocking mAbs to α_Mβ₂ and NIF significantly attenuated (by 80%-90%) H₂O₂ production, whereas blocking mAbs to α_Vβ₃, β₅, and β₁ integrins did not (Figure 6D). These results indicate that, although PMNs adhere to TSP-4 (P387) via α_Mβ₂ and partially α_Vβ₃, it is integrin α_Mβ₂ that mediates TSP-4 (P387)–dependent p38MAPK activation and H₂O₂ production.

Thrombospondins consist of multiple domains, which fold independently, and TSP fragments have been used extensively to identify various TSP receptors.^8-10,15  Thus, we tested the differential functions of TSP-4 variant fragments approximately equal to 50 kDa (residues 325-491), which harbor the 387 SNP and consist of 3 type 2 (EGF-like) repeats²³  for recognition by α_Mβ₂. Both 3EGF–TSP-4 variant fragments, like the full-length TSP-4 variants, supported robust and equal adhesion of stimulated PMNs (Figure 7A), which was completely suppressed by function-blocking mAbs to α_M (clone 44a), to β₂ (IB4) integrin subunit, and by NIF (Figure 7B). These results indicate that α_Mβ₂ recognizes the 3 EGF-like domains of TSP-4. To corroborate these data, we analyzed the direct binding of both 3EGF–TSP-4 variants to purified α_MI-domain. As shown in Figure 7C, both TSP-4 variant fragments bound saturably to the α_MI-domain, and half-maximal binding of 3EGF–TSP-4 (A387) and 3EGF–TSP-4 (P387) occurred at 300 ± 40 and 130 ± 16 nM concentrations, respectively (means ± SEMs, n = 4). This difference in apparent affinity was statistically significant (Student t test, P < .01). In addition, the 3EGF–TSP-4 (P387) variant induced more robust release of H₂O₂ (EC₅₀ value, ≈ 4.1 ± 0.4 nM, similar to full-length protein; Figure 7D), secretion of IL-8 (Table 1), and phosphorylation of p38MAPK (Figure 7E) than the 3EGF–TSP-4 (A387) variant. Thus, the α_Mβ₂ binding site resides within the 3 EGF-like domains of TSP-4, and the observed distinct PMN responses to the SNP variants are also attributed to this region. However, small TSP-4 peptides containing the SNPs (TDIDECRNGACWPN and TDIDECRNGPCWPN) did not support adhesion of stimulated PMNs, did not inhibit PMN adhesion to the full-length TSP-4 variants or to the 3EGF TSP-4 fragments, and did not reduce the oxidative burst triggered by the (P387) variant (data not shown).

In the present study, we demonstrate a previously unrecognized interaction between human leukocytes and TSP-4 and show that the MI risk variant, TSP-4 (P387), and the predominant form, TSP-4 (A387), induce distinct biologic responses in these cells. We also show that α_Mβ₂ is a receptor for TSP-4, the first identified receptor for this TSP family member. Although both TSP-4 variants supported PMN adhesion and migration by engaging α_Mβ₂, the disease-associated variant induced more robust prothrombotic and proinflammatory responses as evidenced by greater release of H₂O₂ and IL-8. These differences are the consequence of activation of distinct signaling pathways by the variants.

Our identification of α_Mβ₂ as a TSP-4 receptor is based on several lines of evidence: (1) function-blocking mAbs directed to either α_M or β₂ subunit completely inhibited PMN adhesion to TSP-4; (2) a high-affinity and specific ligand of α_Mβ₂, NIF, also blocked PMN adhesion to TSP-4^34,35 ; (3) HEK293 cells expressing α_Mβ₂, but not α_Lβ₂ or vector-transfected control cells, responded to TSP-4; (4) a chimeric receptor in which the α_MI-domain was placed into the α_Lβ₂ backbone acquired the capacity to recognize TSP-4; and (5) TSP-4 directly interacted with recombinant α_MI-domain. This latter observation, together with the specificities of NIF^34,35  and mAb 44a³⁶  for the α_MI-domain, indicates that this region of the α_M subunit plays a crucial role in TSP-4 recognition. Thus, TSP-4 appears to represent an additional ligand of this promiscuous receptor and its I-domain, and its recognition supports both adhesion and migration of PMNs. Although PMN adhesion to TSP-4 was partially mediated by α_Vβ₃, the observed kinase activation and biologic responses (H₂O₂ production) depended on engagement of α_Mβ₂, legitimizing its classification as a genuine TSP-4 receptor. To date, interactions of TSP-1 with human PMNs have been reported, but the involvement of α_Mβ₂ in recognition remains uncertain. Nathan et al³⁷  demonstrated that PMNs isolated from patients with leukocyte adhesion deficiency (LAD), who lack β₂ integrins, as well as normal PMNs treated with blocking mAbs to the integrin, exhibited diminished adherence to TSP-1, indicating involvement of the β₂ integrins in TSP-1 binding, whereas Suchard et al³⁸  and Suchard and Boxer³⁹  reported that PMNs from other patients with LAD still adhered to TSP-1. The latest study identified 2 TSP-1 regions involved in PMN recognition: F16-G33 in its heparin binding domain and A784-N823 within its type-3 repeats.⁴⁰  Our data with the EGF-3 fragments implicate the type-2 repeats of TSP-4 in recognition by α_Mβ₂. Hence, there is not a common site that mediates recognition of both TSP-1 and TSP-4 by α_Mβ₂.

We observed that PMNs increased H₂O₂ production by 4-fold in response to TSP-4 (P387). Consistent with the observations made with TSP-1 by Nathan,^41,42  ROS production was critically dependent on PMN adhesion and only occurred to a very modest extent with cells exposed to ligand in suspension. Our results also are in agreement with other studies indicating that production of ROS by macrophages is not only dependent on α_Mβ₂⁴³  but also on activation of p38MAPK.^44-46  Although exogenous oxidants can promote p38MAPK activation,⁴⁷  the time course of events indicated that p38MAPK phosphorylation (5-30 minutes) preceded H₂O₂ production (first detected at 3 hours, data not shown). Of the inflammatory cytokines measured, TSP-4 P387 enhanced production of IL-8. It is likely that IL-8 secretion also is β₂ integrin dependent in view of the data demonstrating that α_Mβ₂-dependent PMN adhesion to fibrinogen resulted in IL-8 production by de novo synthesis and that PMNs isolated from β₂-deficient mice are unable to produce MIP-2, a homologue of human IL-8.^48,49 

Analyses of the tyrosine protein phosphorylation patterns on Western blots of lysates of PMNs adherent to TSP-4 provided direct evidence that the variants induce distinct cellular responses. The TSP-4 P387 variant preferentially activated several kinases. Among these, p38MAPK was selectively activated by engagement of TSP-4 (P387) by α_Mβ₂; its phosphorylation was inhibited by mAbs to α_Mβ₂. Subsequently, active p38 MAPK mediated ROS release by adherent PMNs. In addition, we may speculate that p38 MAPK supported enhanced IL-8 secretion in view of the findings that this kinase promotes stabilization of IL-8 mRNA.⁵⁰  Activation of p38 MAPK by (P387) TSP-4 is particularly important, because p38 MAPK is a highly proinflammatory kinase, being critically involved in cell migration, regulation of apoptosis and survival,⁵¹  oxidative burst,⁵²  and production of IL-6 and TNFα in PMNs and other leukocytes.^31,32,53  Thus, p38 MAPK might be an attractive therapeutic target for protection against cardiovascular disease in individuals carrying the P387 variant.⁵⁴  TSP-4 P387 induced activation of other MAPK family members, heat shock protein (HSP) and c-Jun NH2-terminal kinase (JNK). Both are activated by a variety of environmental stresses and regulate cell apoptosis, production of inflammatory cytokines such as TNFα, IL-12, and IL-1 (reviewed in Kyriakis and Avruch,³¹  Dong et al,³²  and Rane et al⁵⁵ ). Thus, TSP-4 (P387), in contrast to the TSP-4 A387 variant, preferentially activates stress-related signaling events in human PMNs, contributing to augmented inflammation.

Interestingly, the TSP-4 (P387) variant significantly enhanced PMN spreading as well as α_Mβ₂ expression and clustering. These differences in PMN spreading are consistent with increased p38MAPK activation, which controls spreading of macrophages⁵⁶  and would more likely lead to distinct PMN responses because H₂O₂ production depends on PMN spreading.⁵⁷ 

The mechanism by which TSP-4 variants induce distinct effects on PMNs is a particularly important issue for future investigations. A387P polymorphism in TSP-4 resides in the third type-2 repeat (EGF-like domain). We did observe that a recombinant fragment of TSP-4 of approximately 50 kDa (residues 325-491) composed of 3 type 2 (EGF-like) repeats²³  (3EGF–TSP-4) (Figure 7) retained their differential effects on cell signaling, IL-8, and ROS production. However, short TSP-4 peptides containing the SNPs did not support PMN adhesion and did not inhibit any of tested PMN responses induced by the full-length or shorter TSP-4 variant fragments. Either the short peptide cannot present the α_Mβ₂ recognition site in an appropriate conformation, or the SNP substitutions exert broader conformational changes, which influence the α_Mβ₂ recognition site residing elsewhere within the 3 EGF-like repeats. From homology to other EGF-related domains, it is predicted that the P387 substitution would lead to formation of a unique Ca²⁺ binding site that would not be present in the A387 variant or with most other substitutions at this position.^58,23  Alterations in Ca²⁺ binding, either through local or global conformational changes, is know to exert profound effects on the structure and function of other TSP family members.^59,60  Such postulated conformational differences within EGF-like domains could account for the higher apparent affinity of the P387 variant for the α_MI-domain as compared with the A387 variant, which, in turn, could explain the differences in the functional responses elicited by the 2 variants even though both interact with α_Mβ₂. Thus, type-2 repeats in TSPs are important integrin binding sites because they are recognized not only by α_Mβ₂ (in TSP-4) but also by the β₁ integrins (in TSP-1).⁶¹ 

Because TSP-4 (P387) promotes endothelial cell detachment,¹²  PMNs would migrate and adhere to it and produce ROS and IL-8, which are extremely important in the genesis of cardiovascular diseases such as atherosclerosis, thrombosis, and heart failure (reviewed in Husemann et al⁴³  and Cai and Harrison⁶² ). Thus, TSP-4 P387 creates a microenvironment rich in proatherogenic stimuli and ripe for lesion development.

We thank Dr J. Lawler for the gift of TSP-4 cDNA and Dr L. Zhang for generation of the chimeric α_MI-α_Lβ₂ receptor.