Integrins α_vβ₃ and α₅β₁ Mediate Attachment of Lyme Disease Spirochetes to Human Cells

Abstract

Borrelia burgdorferi (sensu lato), the agent of Lyme disease, is able to cause chronic, multisystemic infections in human and animal hosts. Attachment of the spirochete to host cells is likely to be important for the colonization of diverse tissues. The platelet-specific integrin α_IIbβ₃ was previously identified as a receptor for all three species of Lyme disease spirochetes (B. burgdorferi sensu stricto, B. garinii, and B. afzelii). Here we show that B. burgdorferi also recognizes the widely expressed integrins α_vβ₃ and α₅β₁, known as the vitronectin and fibronectin receptors, respectively. Three representatives of each species of Lyme disease spirochete were tested for the ability to bind to purified α_vβ₃ and α₅β₁. All of the strains tested bound to at least one integrin. Binding to one integrin was not always predictive of binding to other integrins, and several different integrin preference profiles were identified. Attachment of the infectious B. burgdorferi strain N40 to purified α_vβ₃ and α₅β₁ was inhibited by RGD peptides and the appropriate receptor-specific antibodies. Binding to α_vβ₃ was also shown by using a transfected cell line that expresses this receptor but not α_IIbβ₃. Attachment of B. burgdorferi N40 to human erythroleukemia cells and to human saphenous vein endothelial cells was mediated by both α₅β₁ and α_vβ₃. Our results show that multiple integrins mediate attachment of Lyme disease spirochetes to host cells.

Lyme disease is the most common arthropod-borne illness in the United States (2, 40) and has been widely reported in Europe and Asia as well. The spirochetal agents of Lyme disease, Borrelia burgdorferi, B. garinii, and B. afzelii, are collectively referred to as B. burgdorferi sensu lato. The clinical manifestations of Lyme disease are complex and can be divided into three stages (27, 40). Localized infection is characterized by a skin rash, erythema migrans, that spreads radially from the site of the tick bite and reflects the migration of the spirochetes through the skin. Early disseminated infection, in which the bacteria invade the vascular system and disseminate to multiple tissues, can include flu-like symptoms, secondary erythema migrans, arthralgia, a variety of neurologic problems, and cardiac manifestations. The third stage, or chronic infection, may occur months to years later, and can affect the skin, joints, and central nervous system. Chronic infection is likely to reflect the establishment of a protected niche in one or more tissues, where the spirochetes persist even in the face of the specific host immune response mounted.

Specific interactions with host tissues and cells are likely to play key roles at each stage of B. burgdorferi infection. In support of this hypothesis, binding to, and migration across, cultured endothelial cell monolayers has been reported (14, 35, 41). B. burgdorferi has also been shown to attach to cultured tick cells (29), to cultured mammalian glial cells, fibroblasts, and epithelial cells (18, 22, 42), and to freshly isolated rodent and human platelets (13, 17).

Binding of B. burgdorferi, B. garinii, and B. afzelii to human platelets is mediated by integrin α_IIbβ₃ (12, 13). Integrins are divalent-cation-dependent, heterodimeric receptors that mediate a variety of cell-cell and cell-extracellular matrix interactions. The specificity of each integrin is determined by the particular combination of α and β polypeptide chains, but the amino acid sequence Arg-Gly-Asp (RGD) is present in several mammalian ligands (24), and synthetic peptides containing this sequence can compete with ligand for receptor occupation. α_IIbβ₃ is the platelet-specific, activation-dependent fibrinogen receptor critical for hemostasis and thrombosis (36). The fact that all of the Borrelia strains that did not bind platelets were noninfectious (12) is consistent with the hypothesis that B. burgdorferi adhesion to platelets via α_IIbβ₃ may be important in the pathogenesis of Lyme disease.

An additional possibility is that the bacterial ligand for α_IIbβ₃ also allows B. burgdorferi to bind to other integrins that are expressed in tissues that are encountered by the organism. Each of the known mammalian ligands for α_IIbβ₃ (fibrinogen, fibronectin, vitronectin, von Willebrand factor, and thrombospondin) also binds to α_vβ₃, the classical vitronectin receptor (24). Fibronectin is also a ligand for several of the β₁-chain integrins, particularly the classical fibronectin receptor α₅β₁ (24). In contrast to α_IIbβ₃, integrins α_vβ₃ and α₅β₁ are widely distributed. For example, integrin α_vβ₃ is expressed by platelets, osteoclasts, smooth muscle cells, some lymphocytes (16, 24, 38), and endothelial cells, where it is thought to play a critical role in angiogenesis (6, 39). Integrin α₅β₁ is found on epithelial and endothelial cells, fibroblasts, lymphocytes, and platelets (39).

In light of the observations that integrin ligands are often recognized by several members of this receptor family and, in particular, the overlapping ligand recognition profiles of α_IIbβ₃, α_vβ₃, and α₅β₁, we explored the possibilities that B. burgdorferi might bind to integrins α_vβ₃ and α₅β₁ and that these interactions could mediate bacterial attachment to the variety of host cells encountered during infection.

MATERIALS AND METHODS

Reagents.

Polyclonal antibodies against the fibronectin and vitronectin receptors were purchased from Telios Pharmaceuticals (San Diego, Calif.). The following monoclonal antibodies (MAb) were from Chemicon (Temecula, Calif.): function-blocking anti-α_vβ₃ MAb LM609 (ascites and purified), blocking anti-α_vβ₅ MAb P1F6, and anti-α_v MAb VnR139, used for immunoblots. The anti-α_IIbβ₃ blocking MAb and the anti-β₃ MAb used for immunoblots were from Immunotech (Westbrook, Maine). Blocking anti-β₁ MAb P4C10 was from Gibco/BRL (Gaithersburg, Md.). Purified anti-α₅β₁ blocking MAb VD1 was a gift from G. Tran Van Nhieu and R. R. Isberg of Tufts University, Boston, Mass. (44). Anti-Lyme spirochete serum was a gift from A. C. Steere (New England Medical Center). Control ascites and the antibiotic G418 (Geneticin) were from Sigma Chemical Co. (St. Louis, Mo.). The synthetic peptides GRGDSP (integrin antagonist) and GRGESP (control) were synthesized at the Tufts Protein Chemistry Facility. The cyclic RGD peptide G4120 (3) was synthesized by Genentech (South San Francisco, Calif.).

Purification of integrins.

Integrin α_IIbβ₃ was purified from human platelets by chromatography over RGD-Sepharose as previously described (13, 38). Integrin α_vβ₃ was purified from human placenta by RGD-Sepharose affinity chromatography essentially as previously described (38). Integrin α₅β₁ was purified by affinity chromatography over an invasin-Sepharose column (31). The buffer used throughout was 25 mM HEPES (pH 7.8)–150 mM NaCl–1 mM MnCl₂–1 mM MgCl₂–0.25 mM CaCl₂ (HBS) containing 10⁻² trypsin-inhibitory units of aprotinin per ml. Octyl-β-d-glucopyranoside (OβDG) was added to a final concentration of 50 mM in HBS to solubilize the integrins. All purification steps were carried out at 4°C; the purified receptors were stored in small aliquots at −70°C. The α_vβ₃ preparation consisted primarily of the α_v and β₃ polypeptides (by gel electrophoresis and immunoblot analyses), but trace amounts of the integrin subunits α_IIb, β₁, and β₅ were also detectable in immunoblots. The α₅β₁ preparation consisted primarily of the α₅ and β₁ subunits, but trace levels of the α_v subunit were also detected by immunoblot analysis. Integrin subunits α₃, α₄, α₆, β₃, and β₅ were not detected.

Bacterial strains and growth conditions.

The Borrelia strains employed in this study were previously characterized in detail for both infectivity and platelet-binding activity (12). The strains designated here as N40, HB19, G39/40, and PKo were cloned derivatives of the original isolates of the same names (12); all others were uncloned. The N40, PBi, PKo, and PBo strains used in this study are infectious; strains HB19, G39/40, PBr, VS102, and VS461 are not infectious (12). Borreliae were cultured at 34°C in MKP medium (32, 37) supplemented with human serum (12, 13), which was previously shown to maximize binding to platelets and to α_IIbβ₃ (12), and stored at −70°C. For each experiment, bacterial stocks were thawed, washed in phosphate-buffered saline (PBS) supplemented with bovine serum albumin (BSA) to 0.2% (wt/vol) (13), and resuspended in HBS supplemented with BSA to 1% (wt/vol) and dextrose to 0.1% (wt/vol) (HBSBD) at a concentration of 2.5 × 10⁷/ml. For some experiments, B. burgdorferi N40 which had been metabolically labeled with [³⁵S]methionine (13) was used in place of the unlabeled bacteria.

Mammalian cell culture.

Cell line 835 was derived from the human embryonic kidney cell line 293 by transfection of the genes encoding the α_v and β₃ integrin subunits (10) and cultured under 7% CO₂ in a mixture of equal parts Dulbecco’s modified Eagle medium (low glucose) and Ham’s F12 nutrient mix with 5% fetal bovine serum, 5% newborn calf serum, and the antibiotic G418 at 400 μg/ml. The human erythroleukemia cell line K562 was cultured in RPMI 1640 medium with 10% fetal bovine serum under 5% CO₂. Human saphenous vein endothelial cells (HSVEC) (28), generously provided by D. W. K. Acheson, A. King, and M. Jacewicz of New England Medical Center, were cultured in gelatinized flasks under 10% CO₂ in medium 199 with 10% fetal bovine serum, 0.1 mg of heparin per ml, and retina-derived endothelial cell growth factor (20). The endothelial character of these cells was confirmed by their rapid release of von Willebrand factor in response to histamine (28). All culture media contained 2 mM glutamine, 100 U of penicillin per ml, and 100 μg of streptomycin per ml.

Precipitation of α_vβ₃ from K562 cell extracts.

Approximately 9 × 10⁷ K562 cells were pelleted from 240 ml of culture and washed twice in 200 volumes of HBS. The final pellet was resuspended in 2 volumes of HBS with aprotinin and phenylmethylsulfonyl fluoride, and OβDG was added to 50 mM. After extraction overnight at 4°C, the sample was clarified twice by centrifugation for 30 min at 14,000 rpm (16,000 × g) in a microcentrifuge at 4°C. Aliquots (250 μl) of the final supernatant were then incubated overnight at 4°C with 50 μl of invasin beads, RGD-Sepharose, or control Sepharose CL-4B beads. The beads were pelleted, washed extensively in HBS–50 mM OβDG, and then heated to 95°C for 5 min in 50 μl of Laemmli sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis sample buffer (30) with no reducing agent. The proteins eluted from the beads were fractionated on a 10% acrylamide gel and transferred to polyvinylidene difluoride membranes under standard conditions (43). The α_v and β₃ subunits were revealed by using standard immunoblot conditions.

Quantitation of Borrelia binding to purified integrins.

Purified integrins were diluted to 1 μg/ml in HBS and dispensed at 50 μl/well into prechilled Linbro 96-well plates (ICN, Irvine, Calif.). Preliminary experiments had determined that receptor coating efficiency was maximal between 0.66 and 1.0 μg/ml; Borrelia binding was also maximal in this range. After incubation overnight at 4°C, the plates were washed once with HBS and then blocked by incubation for 1 h at ambient temperature with HBSBD at 200 μl/well. The blocking buffer was then replaced with HBSBD at 35 μl/well or with the same buffer containing the reagent to be tested. The concentrations of the inhibitory reagents shown here were chosen based on the results of preliminary optimization experiments. Antibodies were used at the concentration at which maximal inhibition of binding to the appropriate receptor was achieved without a significant effect on binding to other integrins. Peptide G4120 was used at the 50% inhibitory concentration (IC₅₀) previously determined in platelet aggregation experiments (3), which was similar to the IC₅₀ for B. burgdorferi interactions with α_IIbβ₃. GRGDSP and GRGESP were used at the minimum concentration required for inhibition of binding to α_vβ₃ by GRGDSP without any significant effect of GRGESP. After incubation for 30 min at room temperature, Borrelia suspensions of 2.5 × 10⁷/ml in HBSBD were added at 15 μl/well. The plates were then centrifuged at 1,200 × g for 10 min and incubated for 1 h at room temperature. Unbound bacteria were removed by washing three times with HBS at 200 μl/well. None of the reagents tested affected either the motility of the bacteria or binding to uncoated wells. Bound, ³⁵S-labeled bacteria were quantitated by adding 1% (wt/vol) SDS to 100 μl/well and transferring the samples to vials for liquid scintillation counting. For quantitation of binding by enzyme-linked immunosorbent assay (ELISA), plates were fixed by the addition of 3% (wt/vol) paraformaldehyde in PBS to 50 μl/well (12). The plates were rinsed with PBS and then blocked with 200 μl (per well) of 5% (wt/vol) nonfat dry milk in PBS (PBSM). Bound borreliae were revealed by incubation with 50 μl (per well) of rabbit anti-Lyme spirochete serum diluted 1:10,000 in PBSM, then with anti-rabbit immunoglobulin G-alkaline phosphatase conjugate diluted 1:10,000 in PBSM, and finally with 1 mg of paranitrophenol-phosphate per ml at 50 μl/well. Optical density was read at 405 nm. Estimates of the percentage of the inoculum bound were made essentially as described previously (12). Dilutions of each bacterial strain were immobilized in microtiter wells containing paraformaldehyde to establish a standard curve of inoculated bacteria for each strain. The ELISA signal obtained for each strain in the integrin-binding assays was then compared to the standard curve for that strain.

Quantitation of Borrelia binding to immobilized mammalian cells.

To promote cell adhesion, sterile Linbro 96-well plates were coated with a 1-μg/ml concentration of a fusion protein containing the integrin-binding domain of the Yersinia pseudotuberculosis invasin protein (MBP-Inv479) (34). Cell line 835 was plated at a density of 0.125 cm²/well (estimated from growth in flasks with a defined area) and incubated under standard growth conditions for 2 days. The medium was then replaced with HBSBD at 35 μl/well, with or without the reagent to be tested, at concentrations chosen as described above. The remainder of the assay protocol was exactly as described above for the purified integrins. Integrity of the cell monolayers was verified at the start and end of each assay. None of the reagents used had any apparent effect on the morphology of either the borreliae or the cultured cells. Statistical analyses were performed by using the two-tailed t test.

Quantitation of Borrelia binding to mammalian cells in suspension.

HSVEC layers were washed twice in PBS and then lifted with PBS supplemented with 10 mM EDTA. The suspension was diluted into 10 volumes of HBSBD, and the cells were pelleted by gentle centrifugation at room temperature. The nonadherent K562 cells were pelleted from the growth medium. Both cell types were resuspended in HBSBD, adjusted to 5 × 10⁵/ml, and then dispensed at 100 μl/tube. MAb (10 μg/ml) were incubated with the cells for 30 min at room temperature prior to the addition of 2.5 × 10^{6 35}S-labeled borreliae. After further incubation for 1 h, the suspension was diluted by addition of 1 ml of HBS and the cells together with bound spirochetes were pelleted by centrifugation for 5 min at 1,800 rpm in an Eppendorf 5415C microcentrifuge. The pellet was resuspended in 1 ml of HBS and then recentrifuged as described above. The final pellet was resuspended in HBS and transferred to vials for liquid scintillation counting. Statistical analyses were performed by using the two-tailed t test.

RESULTS

Integrins α_vβ₃ and α₅β₁ were each purified by affinity chromatography and tested for the ability to bind B. burgdorferi. An infectious clone of strain N40 which binds efficiently to α_IIbβ₃ (13) also bound to microtiter wells coated with the α_vβ₃ or α₅β₁ preparation (Fig. 1). The percentage of inoculated N40 spirochetes bound was highest for immobilized α_IIbβ₃ and lowest for immobilized α₅β₁. This receptor preference profile (i.e., α_IIbβ₃>α_vβ₃>α₅β₁) was highly reproducible and independent of the method of quantitation (Fig. 1).

FIG. 1.

Attachment of B. burgdorferi, B. garinii, and B. afzelii to purified integrins. Bacteria in suspension were added to microtiter wells coated with purified integrins and centrifuged to facilitate bacterial-integrin contact. The plates were incubated for 1 h at room temperature and then washed to remove unbound bacteria. ³⁵S-labeled bacteria were quantitated by liquid scintillation counting. Unlabeled B. burgdorferi, B. garinii, or B. afzelii cells were quantitated by ELISA using a polyclonal rabbit antiserum directed against Lyme disease spirochetes. Binding to uncoated wells was subtracted from that of all others to give the receptor-specific signals displayed. Shown are the means plus the standard deviations of four replicates; the data are representative of multiple determinations. The relative efficiency of binding to each of the three receptors by a given strain was determined in parallel, but the percentage of unlabeled bacteria bound was not quantitated, so comparisons of binding efficiency between strains cannot be made. OD₄₀₅; optical density at 405 nm.

B. burgdorferi (sensu stricto), B. garinii, and B. afzelii were previously shown to bind to α_IIbβ₃ (12). To determine whether binding to multiple integrins is common to Borrelia species that cause Lyme disease, three representatives of each species were tested for binding to α_IIbβ₃, α_vβ₃, and α₅β₁ immobilized in microtiter wells (or to control uncoated wells). Binding to each of the three receptors by all nine strains was quantitated by ELISA. The efficiency of binding to one of the integrins, α_vβ₃, was estimated for six of the strains (see Materials and Methods) and found to range from 10% (for G39/40) to 83% (for PBi), indicating that an easily detectable ELISA signal corresponded to binding of a significant percentage of the bacteria. The nature of ELISA-based quantitation made strain-to-strain comparisons difficult, but by assaying binding by each Borrelia isolate to the three integrins in parallel and in several independent experiments, a highly reproducible preference profile was revealed for each strain (Fig. 1). For example, the noninfectious, high-passage clone of B. burgdorferi HB19 bound most efficiently to α_vβ₃ and least efficiently to α_IIbβ₃, while a third strain, B. burgdorferi G39/40, clone A6 (19), bound most efficiently to α₅β₁ (Fig. 1). Isolates of B. garinii and B. afzelii also bound to the three integrins with highly reproducible preference patterns. Binding to all three integrins by the four infectious strains (N40, PBi, PBo, and PKo) (12) was easily detectable. The observation that some of the noninfectious strains, such as HB19 and G39/40, showed virtually no attachment to some of the integrins suggests that these assays measure specific integrin binding, an assertion supported by inhibition studies (see below and Fig. 2).

FIG. 2.

Specific binding of B. burgdorferi to purified integrins. Purified integrins immobilized in microtiter wells were incubated for 30 min at room temperature with the indicated reagent. Radiolabeled strain N40 or unlabeled strain HB19 was added, and the bacteria were centrifuged, incubated, and quantitated as described in the legend to Fig. 1. Relative binding efficiency is defined as the degree of binding in the presence of each reagent divided by the degree of binding in the absence of any inhibitor. EDTA was used at 10 mM, GRGESP and GRGDSP linear peptides were used at 45 μM, and the cyclic RGD (cRGD) peptide G4120 was used at 150 nM. Purified MAb were used at 10 μg/ml, and ascites was used at a 1:1,000 dilution. See Materials and Methods for the source of each MAb. Shown are the means plus the standard deviations of four replicates. ND, not determined.

Binding of B. burgdorferi N40 to all three integrin preparations was inhibited by synthetic peptides containing the Arg-Gly-Asp (RGD) sequence (Fig. 2). The relative potency of these RGD peptides, however, varied between integrins. Binding to α_IIbβ₃ was significantly inhibited by G4120, a cyclic RGD peptide that binds this receptor with high affinity (3), while linear RGD and RGE peptides had little effect (Fig. 2). This result is consistent with the previously observed IC₅₀ of cyclic peptide G4120 versus linear RGD peptides in the inhibition of platelet aggregation and in the inhibition of binding of purified α_IIbβ₃ to fibrinogen (3). Binding to α_vβ₃ was inhibited most efficiently by G4120, but the linear RGD peptide also had a significant effect. In contrast, binding to α₅β₁ was inhibited most efficiently by the linear RGD peptide, which is consistent with the previous observation that G4120 does not inhibit α₅β₁ interaction with fibronectin (5a). Therefore, the RGD peptide inhibition characteristics differed for each integrin preparation; this suggests that the Borrelia interaction with each integrin is specific and unique. Binding to all receptors was inhibited by EDTA, which chelates the divalent cations required for integrin function.

To further evaluate the specificity of N40 binding to the major polypeptides in each receptor preparation, we tested the effects of various anti-integrin MAb on binding to each of the receptors. As shown previously (13), bacterial binding to purified α_IIbβ₃ was inhibited by the MAb directed against this integrin but not by MAb directed against α_vβ₃ or α₅β₁. N40 binding to α₅β₁ was inhibited by MAb directed against the β₁ chain and against the α₅β₁ complex but not by the anti-α_IIbβ₃ MAb (Fig. 2). Binding to the α_vβ₃ preparation was completely abrogated by a function-blocking MAb that specifically recognizes this receptor complex (9) (Fig. 2) and by an α_v-specific MAb (data not shown). Binding was not affected by anti-α_vβ₅, anti-β₁, or anti-α₅β₁ MAb. The anti-α_IIbβ₃ MAb, however, partially inhibited the attachment of N40 to α_vβ₃. It is possible that the effect of the anti-α_IIbβ₃ antibody on N40 attachment to α_vβ₃ reflects a low-level cross-reactivity of the antibody between the two receptor complexes. It is unlikely, however, that N40 binding to α_vβ₃ is simply due to the trace amount of α_IIbβ₃ present in the preparation, because the α_v-specific and α_vβ₃-specific MAb completely blocked binding to the α_vβ₃ preparation but had no effect on N40 binding to α_IIbβ₃ (Fig. 2 and data not shown).

Additional evidence for specific recognition of α_vβ₃ by B. burgdorferi comes from the analysis of strain HB19, which does not recognize α_IIbβ₃ (Fig. 1 and reference 13) but did bind to the α_vβ₃ preparation. As predicted, attachment of HB19 to α_vβ₃ was inhibited by EDTA, RGD (but not RGE) peptides, and the anti-α_vβ₃ MAb but not by any other anti-integrin MAb, including anti-α_IIbβ₃ MAb (Fig. 2). In addition to showing a clear functional difference between our purified α_vβ₃ and α_IIbβ₃ preparations, these results substantiate our finding that the integrin preference profile can vary from strain to strain and that integrin recognition by Borrelia spirochetes is specific.

To obtain independent evidence that α_vβ₃, as well as α_IIbβ₃, is recognized by B. burgdorferi, we analyzed attachment to cells that express α_vβ₃ but not α_IIbβ₃. Cell line 835 is derived from the human embryonic kidney line 293 by transfection with the genes encoding α_v and β₃ and expresses functional α_vβ₃ on the cell surface (10). In contrast, untransfected 293 cells do not express any β₃-chain integrin (5). Analysis of integrin-mediated binding of B. burgdorferi to intact cells is complicated by the fact that this bacterium also recognizes proteoglycans (21, 25, 32, 33). Attachment of B. burgdorferi N40 and HB19 to 835 cells was therefore evaluated in the presence of MAb directed against several integrins, either alone or in combination with platelet factor 4, which binds certain glycosaminoglycans and blocks B. burgdorferi attachment to diverse cell types (32, 33). Binding of both B. burgdorferi strains to 835 cells was partially inhibited by preincubation of cells with either platelet factor 4 (P ≤ 0.002, platelet factor 4 versus the control, for both N40 and HB19) or the anti-α_vβ₃ MAb (P ≤ 0.02, antibody versus the control, for both N40 and HB19) (Fig. 3), suggesting that bacterial binding was mediated by both integrin α_vβ₃ and proteoglycans. Other anti-integrin MAb had no significant effect. It should be noted, however, that because the cells were plated on invasin, the β₁-chain integrin receptors for invasin may be unavailable for recognition by either the anti-α₅β₁ MAb or B. burgdorferi. The component of binding that was not blocked by the anti-α_vβ₃ MAb alone was apparently mediated by proteoglycan, because the combination of the anti-α_vβ₃ MAb and platelet factor 4 reduced the attachment of both N40 and HB19 to background levels (Fig. 3). None of the anti-integrin MAb affected bacterial binding to 293 cells, either alone or in the presence of platelet factor 4 (data not shown), demonstrating the specificity of the anti-α_vβ₃ MAb. The results of the experiments using 835 cells provide genetic evidence that α_vβ₃, specifically, mediates attachment of B. burgdorferi to human cells.

FIG. 3.

Binding of B. burgdorferi to integrin α_vβ₃ expressed on transfected cells. 835, a cell line that expresses α_vβ₃ as a result of transfection of the genes encoding the α_v and β₃ subunits, was grown to confluence in invasin-coated microtiter wells and incubated with the reagent indicated for 30 min at room temperature. Radiolabeled strain N40 (filled bars) or HB19 (shaded bars) was then added, and the remainder of the assay was performed as described in the legends to Fig. 1 and 2. Platelet factor 4 (PF4) was used at 5 μg/ml. Purified MAb were used at 10 μg/ml and ascites was used at a 1:1,000 dilution. Relative binding efficiency is defined as the degree of binding in the presence of each reagent divided by the degree of binding in the absence of any inhibitor. Shown are the means plus the standard deviations of four replicates.

To determine whether B. burgdorferi binds to integrin α₅β₁ on intact mammalian cells and whether α_vβ₃ and α₅β₁ can participate in the attachment of B. burgdorferi to cells which had not been genetically modified, two additional cell types were employed. α₅β₁ is the only β₁-chain integrin expressed by the human erythroleukemia cell line K562 (23). This cell line also expresses α_vβ₃, as determined by identification of the receptors isolated by RGD Sepharose precipitation of octylglucoside extracts of K562 cells (Fig. 4A). HSVEC were also used to test the roles of α₅β₁ and α_vβ₃ in B. burgdorferi binding, as both of these integrins are expressed by endothelial cells (39). Binding of B. burgdorferi N40 to both cell types was partially inhibited by an anti-α₅β₁ function-blocking MAb, but no reproducible effect was seen with the blocking anti-α_vβ₃ MAb (Fig. 4B). A combination of the anti-α₅β₁ and anti-α_vβ₃ MAb, however, caused significant inhibition of B. burgdorferi attachment to both cell types (P < 0.0006, control versus antibody mixture for K562 cells; P < 0.04, control versus antibody mixture for HSVEC). The residual binding seen in both cell types may reflect B. burgdorferi recognition of other integrin receptor complexes or proteoglycans, although in pilot experiments, platelet factor 4 had no effect on B. burgdorferi attachment to K562 cells (data not shown). No effect was seen with an anti-α_IIb MAb, consistent with the fact that the expression of this integrin subunit is limited to platelets.

FIG. 4.

Binding of B. burgdorferi to both α_vβ₃ and α₅β₁ on human cells. (A) RGD-Sepharose precipitation of α_vβ₃ from K562 cells. Octyglucoside extracts of K562 cells were incubated with invasin-coated beads, control Sepharose beads, or RGD-Sepharose beads. The beads were washed, and bound proteins were fractionated by SDS-polyacrylamide gel electrophoresis under nonreducing conditions and then transferred to a polyvinylidene difluoride membrane. Lanes containing electrophoresis markers (New England Biolabs) and purified α_vβ₃ were stained with Coomassie blue. Lanes containing proteins precipitated from K562 extracts were probed with anti-α_v or anti-β₃ MAb or with control mouse serum by using standard immunoblot protocols. Bound antibody was revealed by using an anti-mouse immunoglobulin G-horseradish peroxidase conjugate and a chemiluminescent substrate. All primary antibodies were used at a 1:500 dilution, and the enzyme conjugate was diluted 1:5,000. The values to the left are molecular masses in kilodaltons. (B) Inhibition of B. burgdorferi attachment to HSVEC and K562 cells by anti-α₅β₁ and anti-α_vβ₃ MAb. Cells in suspension were incubated with the MAb shown (each at 10 μg/ml) for 30 min at room temperature. Radiolabeled B. burgdorferi N40 was then added, and the incubation was continued for 1 h (see Materials and Methods). After washing, bound spirochetes were quantitated by liquid scintillation counting. Shown are the means plus the standard deviations of four replicates after background subtraction.

DISCUSSION

Interaction with a variety of host tissues and cells is likely to play a key role in each of the stages of Borrelia infection. For example, binding to mammalian cells might be involved in the apparent tropism of B. burgdorferi for particular tissues and in bacterial dissemination and persistence. In support of this hypothesis, a number of laboratories have demonstrated that B. burgdorferi binds to a wide variety of cell types in vitro. Only recently, however, has progress toward the identification of the specific molecules involved been made (12, 13, 21, 25, 32, 33). In the current study, we have shown that B. burgdorferi recognizes not only α_IIbβ₃ but also integrins α_vβ₃ and α₅β₁, the vitronectin and fibronectin receptors, respectively. Nine strains representing B. burgdorferi, B. garinii, and B. afzelii were analyzed; all of the strains demonstrated binding to at least one integrin. Interestingly, binding to one integrin was not always predictive of binding to other integrins, and several different but highly reproducible integrin preference profiles were identified in this collection of nine strains. This indicates that the lack of recognition of a particular integrin by a given strain is not simply due to a generalized decrease in gene expression or in protein export. The distinct integrin recognition profiles of the different Borrelia strains, together with the specificity of inhibition of attachment by the various integrin antagonists tested here, demonstrate that Borrelia interactions with α_IIbβ₃, α_vβ₃, and α₅β₁ are specific.

The participation of α_vβ₃ and α₅β₁ in the attachment of B. burgdorferi to human cells was also demonstrated. Binding of B. burgdorferi to HSVEC and K562 cells was partially inhibited by a blocking anti-α₅β₁ antibody. Significant inhibition of B. burgdorferi attachment, however, was achieved only in the presence of both the anti-α_vβ₃ and anti-α₅β₁ function-blocking MAb. The involvement of multiple integrins has previously been demonstrated for adenovirus internalization: a mixture of anti-α_vβ₃ and anti-α_vβ₅ antibodies decreased viral infectivity, while neither antibody alone had any effect (45). Thus, the interaction of B. burgdorferi with integrins on intact cells may be complex. Blocking of both α_vβ₃ and α₅β₁ did not result in complete inhibition of bacterial attachment to either K562 cells or HSVEC, raising the possibility that other integrins or proteoglycans also contribute to attachment. For example, heparin/heparan sulfate proteoglycans participate in spirochetal adhesion to bovine capillary endothelial cells (33). However, platelet factor 4, which blocks binding of B. burgdorferi to multiple cell types (32, 33), had no effect on attachment to K562 cells (data not shown).

The observation that even noninfectious strains displayed integrin-binding activity demonstrates that integrin binding alone is not sufficient for infectivity. This is not a surprising result, given the multifactorial nature of bacterial virulence. Nevertheless, the four infectious strains analyzed in this study bound to all three receptor preparations, consistent with the hypothesis that the ability to bind to multiple integrins is important during infection. Given the widespread expression of α₅β₁ and α_vβ₃, the specific host cells that are targets of integrin binding by B. burgdorferi in vivo are not known. Binding to integrins expressed in target tissues such as heart or joint could promote colonization. In this regard, it is especially intriguing that B. burgdorferi attachment to endothelial cells is inhibited most efficiently by a mixture of antibodies directed against α_vβ₃ and α₅β₁. The spirochete must interact with these cells during transit between perivascular tissues and the bloodstream, and endothelial damage is commonly observed in Lyme disease (1, 4, 26). B. burgdorferi has been shown to bind cultured endothelial cells (35), to cross endothelial cell monolayers (14, 41), and to promote the transmigration of leukocytes across these monolayers by inducing the expression of adhesion molecules and chemoattractants (8, 15). In pilot experiments, we observed no reproducible effect of any anti-integrin antibody on B. burgdorferi attachment to adherent endothelial cells, in contrast to our results obtained by using cells in suspension. This is consistent with the previous observation that integrins α_vβ₃ and α₅β₁ expressed by cultured cells are largely localized to zones of adhesion to the extracellular matrix (9). The cellular localization of α_vβ₃ and α₅β₁ on endothelial cells in vivo remains unclear, but antagonists of α_vβ₃ administered intravenously inhibit neovascularization in the chick chorioallantoic membrane model (7), raising the possibility that this receptor is also available to the spirochete during infection. The use of cultured cells in suspension allowed us to identify at least two integrins that might be involved in the interaction of B. burgdorferi with the vasculature in vivo.

The initiation of tissue colonization by B. burgdorferi potentially involves a multitude of interactions with a variety of host cell molecules. It is clear from this study that a variety of integrins are potential receptors for Lyme disease spirochetes. It will be interesting to determine whether additional integrin families (e.g., the β₂-chain receptors) can also mediate Borrelia attachment to host cells. A recent study showed that B. burgdorferi binds to integrin α_Mβ₂ (also termed CR3 or Mac-1) after opsonization with complement (11). In contrast, B. burgdorferi attachment to platelets and to purified α_IIbβ₃ did not appear to depend on serum components that might be sequestered by the bacteria during in vitro cultivation (13). Furthermore, given that all of the bacterial strains used in this study were cultured in the same medium, the strain-to-strain variations in integrin preference profiles shown here suggest that one or more bacterial integrin ligands are expressed by the different Borrelia strains. One possibility is that β₁ and β₃ chain integrins are recognized by the same bacterial protein, with variations in the amino acid sequences accounting for the observed differences in integrin-binding activity. Alternatively, the products of multiple distinct genes might be required for attachment to different integrins. Resolution of this question, as well as analyses of Borrelia integrin-binding properties in either mammalian or arthropod hosts, awaits the identification and cloning of the molecule(s) involved.