Integrin α_Vβ₃ Binds to the RGD Motif of Glycoprotein B of Kaposi's Sarcoma-Associated Herpesvirus and Functions as an RGD-Dependent Entry Receptor

Abstract

Kaposi's sarcoma-associated herpesvirus (KSHV) envelope-associated glycoprotein B (gB) is involved in the initial steps of binding to host cells during KSHV infection. gB contains an RGD motif reported to bind the integrin α₃β₁ during virus entry. Although the ligand specificity of α₃β₁ has been controversial, current literature indicates that α₃β₁ ligand recognition is independent of RGD. We compared α₃β₁ to the RGD-binding integrin, α_Vβ₃, for binding to envelope-associated gB and a gB(RGD) peptide. Adhesion assays demonstrated that β₃-CHO cells overexpressing α_Vβ₃ specifically bound gB(RGD), whereas α₃-CHO cells overexpressing α₃β₁ did not. Function-blocking antibodies to α_Vβ₃ inhibited the adhesion of HT1080 fibrosarcoma cells to gB(RGD), while antibodies to α₃β₁ did not. Using affinity-purified integrins and confocal microscopy, α_Vβ₃ bound to gB(RGD) and KSHV virions, demonstrating direct receptor-ligand interactions. Specific α_Vβ₃ antagonists, including cyclic and dicyclic RGD peptides and α_Vβ₃ function-blocking antibodies, inhibited KSHV infection by 70 to 80%. Keratinocytes from α₃-null mice lacking α₃β₁ were fully competent for infection by KSHV, and reconstitution of α₃β₁ function by transfection with α₃ cDNA reduced KSHV infectivity from 74% to 55%. Additional inhibitory effects of α₃β₁ on the cell surface expression of α_Vβ₃ and on α_Vβ₃-mediated adhesion of α₃-CHO cells overexpressing α₃β₁ were detected, consistent with previous reports of transdominant inhibition of α_Vβ₃ function by α₃β₁. These observations may explain previous reports of an inhibition of KSHV infection by soluble α₃β₁. Our studies demonstrate that α_Vβ₃ is a cellular receptor mediating both the cell adhesion and entry of KSHV into target cells through binding the virion-associated gB(RGD).

Kaposi's sarcoma-associated herpesvirus (KSHV)/human herpesvirus 8 was originally detected in KS lesions (12) and has been implicated in the etiology of KS and AIDS-associated pleural effusion lymphoma and multicentric Castleman's disease (52). KS is a multifocal lesion characterized by intense angiogenesis, proliferation, and infiltration by inflammatory cells (18). KSHV is detected in a number of cell types present in KS lesions, including the characteristic KS spindle cells, infiltrating B cells, and neovascular endothelial cells (4). KSHV infection is latent in these cells, with only a small percentage of cells undergoing active virus replication. In vitro, KSHV has a broad tropism and infects endothelial, epithelial, and mesenchymal cells (5, 47).

Similar to other herpesviruses, KSHV's infection of susceptible cells occurs through at least two separate binding events. The initial binding to heparan sulfate on cell-surface proteoglycans is mediated by the virion envelope-associated glycoproteins, K8.1 (63) and glycoprotein B (gB) (3). Although heparan-sulfate binding is reversible and not essential for virus entry, it is thought to localize virus on the cell surface, enhancing opportunities for direct interactions between virion glycoproteins and specific cellular entry receptors (3, 59). A number of entry receptor candidates have been reported for KSHV. The cystine transporter, xCT, was identified as a KSHV fusion-entry receptor by functional cDNA selection (27). DC-SIGN was shown to function as a receptor for KSHV on myeloid dendritic cells and macrophages (46). The α₃β₁ cellular integrin has also been identified as a potential entry receptor for KSHV (2), and integrin signaling has been implicated in the KSHV infection process and facilitates uptake into the host cell, intracellular trafficking of the viral capsid, and establishment of the latent infection (36, 39, 40, 55). These findings are of particular interest since the integrin receptor family is exploited by a number of other viruses as an entry pathway for infection.

While the KSHV virion proteins that interact with the xCT and DC-SIGN receptors have not yet been identified, the KSHV envelope-associated gB was reported to be the viral ligand recognized by the α₃β₁ integrin receptor (2). gB is highly conserved within members of the herpesvirus family and plays an important role in the infection of susceptible host cells by different herpesviruses (42). Sequence analysis of KSHV gB identified an N-terminal Arg-Gly-Asp (RGD) tripeptide motif, also found in extracellular matrix proteins that interact with members of the integrin receptor family (2, 49, 50). Integrin receptors consist of noncovalently associated αβ heterodimers, and ligand specificity is determined by different combinations of α and β monomers. The diverse integrin family is divided based on ligand specificity into subfamilies which include integrin receptors specific for RGD (group I), laminin (group II), collagen (group III), and leukocytes (group IV) (24). The presence of the RGD motif in the KSHV gB suggested that it might be a ligand of a group I RGD-binding integrin, such as α_Vβ₃, α_Vβ₅, or α₅β₁; however, gB was reported to interact with the laminin-binding integrin α₃β₁ (2). Although α₃β₁ was once regarded as a promiscuous receptor with weak affinity for RGD and other ligands, more-recent studies have shown that it binds with high affinity to laminin 5, laminin 10/11, and bacterial invasin in an RGD-independent manner (11, 14, 29, 30, 37, 41, 69). These findings raise questions regarding the direct involvement of α₃β₁ in RGD-mediated binding to KSHV gB and the role of α₃β₁ in RGD-mediated entry of KSHV into target cells.

We have used a variety of approaches to study the binding specificity of the RGD motif of KSHV gB and the role of RGD-binding integrins in KSHV infection. We have analyzed integrin binding to the native KSHV virion envelope-associated gB and to a peptide derived from the N-terminal RGD motif of KSHV gB [gB(RGD)] using cell adhesion and affinity binding studies. We have also used peptide antagonists and function-blocking antibodies to identify the integrin responsible for RGD-mediated binding and entry of KSHV into target cells. Our studies show that the group I integrin, α_Vβ₃, specifically interacts with the RGD motif of KSHV gB and functions as a receptor in cell adhesion and KSHV entry during infection. We were unable to detect any specific interactions between the α₃β₁ integrin and gB or any of the other envelope-associated glycoproteins present on the KSHV virion. We observed an inhibition of KSHV infectivity and a downregulation of α_Vβ₃ expression and function in cells overexpressing α₃β₁. These effects correlate with the transdominant inhibition of α_Vβ₃ function by α₃β₁ and other integrins containing the β₁ subunit seen by others (7, 22, 38) and may explain previous reports of inhibitory effects of soluble α₃β₁ on KSHV infectivity (2, 39, 55). Our results are consistent with the results of studies showing the importance of integrin-mediated entry and integrin-associated signaling during KSHV infection but suggest that the integrin receptor directly interacting with the RGD motif of KSHV gB is α_Vβ₃ integrin.

MATERIALS AND METHODS

Cell lines.

The human HT1080 fibrosarcoma cell line was a gift of W. G. Carter. African green monkey Vero cells were from the American Type Culture Collection. The Chinese hamster ovary (CHO) parental cell line, pBJ, and human integrin-expressing transfected cell lines α₃-CHO (60) and β₃-CHO (61) which overexpress α₃β₁ and α_Vβ₃ integrins, respectively, were a kind gift from Y. Takada. Cells were maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. The CHO cell medium also contained G418 (700 μg/ml). BCBL-1 cells latently infected with KSHV (48) were cultured in RPMI complete medium with 10% fetal bovine serum, 100 U/ml penicillin, 100 μg/ml streptomycin, 1.0 mM HEPES, and 0.01% 2-mercaptoethanol at 37°C. The mouse keratinocyte cell lines and culture medium have been described previously (16). The cell lines include MK-116 (MK^+/+), derived from wild-type keratinocytes; MK-546 (MK^−/−), derived from α₃-null keratinocytes; and MK-546-hα₃A (MK^−/−,α3), derived from MK-546 cells stably transfected with human α₃A integrin.

Immunological reagents.

The following integrin-specific mouse monoclonal antibodies (MAbs) were purchased from Chemicon: anti-α₃ (clone ASC-6, clone P1B5, and clone 29A3 [for blotting]), anti-α_Vβ₃ (clone LM609), anti-β₃ (clone B3A), anti-α₅β₁ (clone JBS5), and anti-α_Vβ₅ (clone P1F6). Other antibody reagents included mouse anti-K8.1A (clone 2A3) and rat anti-latency-associated nuclear antigen (LANA) (clone LN53) (Advanced Biotechnologies), mouse antibiotin (Molecular Probes), goat antibiotin-horseradish peroxidase (HRP) (Sigma), goat anti-mouse immunoglobulin G (IgG)-HRP, goat anti-mouse IgG-alkaline phosphatase (alk-phos), goat anti-rat IgG-HRP (BioSource), normal goat serum, nonimmune-mouse IgG, and mouse antibiotin-HRP (Jackson).

Peptides and purified integrin receptors.

Biotinylated peptides corresponding to the 12-amino-acid sequences of the RGD motif within the predicted N terminus of the mature KSHV gB and an RGE motif with a single-amino-acid mutation were synthesized by Res Gen (Huntsville, AL): gB(RGD), AHSRGDTFQTSSGCG, and gB(RGE), AHSRGETFQTSSGCG. The C-terminal GCG amino acids were added for coupling purposes. The α_Vβ₃ antagonists and peptide controls were cyclo(Arg-Gly-Asp-D-Phe-Lys) (cRGD), dimeric H-Glu[cyclo(Arg-Gly-Asp-D-Phe-Lys)]₂ (dcRGD), cyclo(Arg-Ala-Asp-D-Phe-Lys) (cRAD), and linear GRGDS and GRGES peptides (Peptides International). Affinity-purified human α_Vβ₃ (cc1020) and α₃β₁ (cc1092) integrin receptors in an octyl-beta-d-glucopyranoside formulation were purchased from Chemicon.

Flow cytometry.

The expression levels of different integrins present on the CHO or HT1080 cells were determined by using fixed-cell suspensions incubated with anti-integrin antibodies and fluorescein isothiocyanate-labeled goat anti-mouse IgG and analyzed on a FACScan instrument. The data are given as the mean fluorescence intensities (MFIs) of the stained cell populations.

Adhesion assay.

NeutrAvidin plates were coated with biotinylated gB(RGD) or gB(RGE) peptides at different concentrations. Maleimide plates (Pierce) were coated with laminin (20 μg/ml). The plates were blocked with phosphate-buffered saline-2% bovine serum albumin. HT1080 or CHO cell suspensions (2 × 10⁴ cells/well) were incubated with immobilized peptides at 25°C for 1 h to allow cell adhesion. In some experiments, the cells were cooled to 4°C and pretreated for 1 h at 4°C with various function-blocking anti-integrin antibodies prior to being plated on peptide (10 μg/ml). Following cell adhesion at 25°C, the plates were inverted to remove nonadherent cells and then incubated at 37°C for 30 min to allow adherent cells to spread. The adherent cells were fixed, stained with 0.5% toluidine blue, and counted manually. The experiments were performed in triplicate or quadruplicate, and the means and standard errors of the cell numbers were determined.

Virus purification.

KSHV virions were purified from filtered culture supernatant (180 ml) of tetradecanoyl phorbol acetate (TPA)-treated BCBL-1 cells (20 ng/ml; 6 days) by centrifugation onto a cushion of 50% Opti-Prep (iodixanol) (19). An aliquot of the concentrated virus was labeled with 10 μg/ml NHS-PEO4-biotin (Pierce) and further purified by centrifugation on a 20%, 25%, 30%, and 40% Opti-Prep step gradient. Virus was quantitated by real-time quantitative PCR. A single DNA peak was obtained at the 20%-to-25% Opti-Prep interface (fractions 11 and 12; see Fig. 6A). The peak fractions were pooled and represent a 450-fold concentration of the original BCBL-1 supernatant. Virions were characterized by immobilization on poly-l-lysine-coated slides and staining with anti-K8.1 or antibiotin antibodies using goat anti-IgG-HRP and 488 and 594 tyramide signal amplification, respectively (Molecular Probes).

FIG. 6.

Gradient purification and characterization of KSHV virions for binding and infection studies. Infectious KSHV virions were purified from culture medium of TPA-induced BCBL cells by double-step gradient centrifugation in Opti-Prep. (A) Virus-containing fractions were identified by real-time quantitative PCR analysis for KSHV genomic DNA. (B, C) Biotinylated virions showed discrete staining and colocalization (arrows) of antibody reactivity to biotin (B) and KSHV envelope-associated glycoprotein K8.1 (C). Bar, 5 μm.

Rhesus rhadinovirus (RRV), a close homolog of KSHV (15), was used as a gB(RGD)-negative control virus. RRV, a gift of M. Axthelm and S. Wong, was propagated as described previously (9). RRV virions were concentrated, biotinylated, and gradient purified as described above for KSHV. The infectivity of the purified RRV was confirmed by using rhesus primary fetal fibroblasts (data not shown).

KSHV infection.

MK cell lines and Vero cells were seeded at 2.5 × 10³ cells per well 24 h prior to infection. The cell cultures were infected with different dilutions of gradient-purified KSHV virions for 3 h at 37°C, washed, incubated for an additional 24 h, fixed, and stained for KSHV LANA using the LN53 anti-LANA antibody essentially as described previously (9). The nuclei were stained with TOPRO (Molecular Probes). The percentage of cells expressing nuclear LANA was determined manually using confocal images.

In some experiments, the ability of KSHV to infect HT1080 in the presence of specific α_Vβ₃ antagonists was determined. HT1080 cells (3 × 10⁴) were chilled on ice and pretreated with peptide (1 mM) or anti-α_Vβ₃ antibody (LM609; 5 μg) for 1 h at 4°C. Gradient-purified KSHV titrated to yield an infection in 75% of the target cells (3 μl, approximately equivalent to 1 ml of BCBL-1 supernatant) was added and incubated for 1 h at 4°C. The cells were washed, cultured for 24 h, and fixed. The percentages of cells expressing LANA and the relative amounts of nuclear fluorescence were determined by using confocal imaging with the LN53 antibody. The histogram function within the Zeiss LSM software was used to measure the absolute frequency and intensity of fluorescent pixels within the nuclear LANA dots. The fluorescence intensities ranged from 0 (no fluorescence) to 255 (maximum fluorescence), and the lower threshold was set at 100 to eliminate nonspecific fluorescence.

Magnetic bead assays.

NeutrAvidin (Pierce) was coupled to Dynabeads with carboxylic acid using the EDC/NHS method (Dynal Biotech). The biotinylated gB(RGD) and gB(RGE) peptides (approximately 0.34 mg peptide/2.2 mg beads) or biotinylated gradient-purified KSHV or RRV virions were immobilized on the NeutrAvidin beads (50 μl virions/2 mg beads). To monitor coupling, the KSHV-coated beads were stained with anti-K8.1 antibody, while the RRV-coated beads and gB peptide-coated beads were stained with antibiotin antibody, and immunofluorescence was visualized by confocal microscopy.

(i) Affinity purification of integrins bound to immobilized peptide.

β₃-CHO or α₃-CHO cells (5 × 10⁶) were solubilized with lysis buffer (100 mM octyl-beta-glucoside, 25 mM Tris HCl, 150 mM NaCl, 1.0 mM MgCl₂, pH 7.4, containing 2 mM Pefabloc [Roche]) for 1 h at 4°C. The cell lysates were incubated with peptide-coated beads, and bound proteins were eluted with 20 mM EDTA, resolved on 10% NuPAGE gels (Invitrogen), and transferred to polyvinylidene difluoride membranes. The membranes were probed with either anti-β₃ (MAb 2023) or anti-α₃ (MAb 2290) antibody followed by alk-phos-labeled goat anti-IgG. Western blue alk-phos substrate (Promega) was used for the final enzymatic development.

(ii) Affinity purification of integrins bound to immobilized virions.

KSHV virion-coated beads or d-biotin-coated control beads were incubated with octylglucoside lysates of β₃-CHO, α₃-CHO, or HT1080 cells. The beads were immobilized on poly-l-lysine-coated slides, fixed, and blocked with 10% normal goat serum. Bound integrins were detected by confocal immunofluorescence microscopy using anti-β₃ (MAb 2023) or anti-α₃ (clone P1B5) antibody followed by goat anti-IgG-HRP with TSA 488 development. The Zeiss LSM software histogram function was used to determine the sum of the integrin fluorescence pixels (intensity range, 100 to 255) associated with individual beads. The mean sum ± standard deviation (SD) values for 10 beads per group were determined. The fluorescence pixels that were specific for integrin bound to KSHV were quantitated by subtracting the mean control bead values from the mean values for KSHV-bound integrin.

(iii) Binding of purified integrin receptor to immobilized peptides and purified virions.

Dynabeads coated with either gB(RGD) or gB(RGE) peptides or KSHV or RRV virions were incubated with 0.7 μg of affinity-purified α_Vβ₃ or α₃β₁ integrin receptors in phosphate-buffered saline containing 1 mM MgCl₂. The beads were washed and fixed, and bound integrins were detected by confocal immunofluorescence microscopy using the anti-β₃ (clone B3A) or anti-α₃ (clone P1B5) antibodies as described above.

RESULTS

The RGD motif in the KSHV envelope-associated gB has sequence similarity to extracellular matrix and viral ligands recognized by the α_Vβ₃ integrin.

Sequence analysis of KSHV gB identified an Arg-Gly-Asp (RGD) motif near the N terminus of the predicted mature protein (49, 50). RGD motifs are recognized by a number of different integrin receptors, and the amino acids flanking RGD motifs contribute to receptor specificity (44). We compared the KSHV RGD motif and flanking region (Fig. 1A) to the RGD motifs and flanking regions of several integrin-binding extracellular matrix (Fig. 1C) and viral proteins (Fig. 1B) for clues to potential KSHV receptor specificity. Visual inspection of the sequences flanking the core RGD motif identified similarities in amino acid composition both upstream and downstream of the RGD motif (Fig. 1A). Alanine (A) residues upstream and a proline (P) residue downstream of the RGD motif were conserved in several RGD-containing ligands. An important conservation of the structurally similar serine (S) and threonine (T) residues flanking both sides of the RGD motif was noted between the KSHV sequence and other ligands. Similar to the sequences of vitronectin and the adenovirus 2 penton base, the KSHV gB contained a phenylalanine (F) residue 2 amino acids downstream of the RGD motif. Overall, the KSHV gB and adenovirus penton base proteins were the most similar, with 6 of 7 identical amino acids within the RGD motif and downstream flanking sequences. The RGD domain of the penton base protein has been shown to interact with the integrin heterodimers α_Vβ₃ and α_Vβ₅ (67), suggesting that the KSHV RGD motif might also bind to these or other integrins containing the α_V integrin subunit.

FIG. 1.

The RGD motif of KSHV gB has sequence similarity to ligands of the α_Vβ₃ integrin. The RGD motif of KSHV gB (A) was aligned with the RGD motifs of viral (B) and extracellular matrix (C) ligands of cellular integrin receptors. The RGD motif is boxed, and conserved residues, including conservative (S)erine-(T)hreonine changes, are highlighted. The sequence from which the KSHV gB(RGD) peptide is derived is indicated. ADV-2, adenovirus 2 (penton base) (NCBI accession no. NP_040521); FMDV-C VP1, foot-and mouth-disease virus serotype C1 viral protein 1 (NCBI accession no. JC1329); FMDV-O VP1, (foot-and-mouth disease virus serotype O viral protein 1 (NCBI accession no. P03305); HPEV1 VP1, human parechovirus 1 viral protein 1 (NCBI accession no. Q66578). Vitronectin, NCBI accession no. P04004; fibrinogen, NCBI accession no. NP_000499; fibronectin, NCBI accession no. AAD00019.

The glycoprotein B RGD motif is a ligand for cell surface α_Vβ₃ integrin.

Previously, the KSHV gB N-terminal domain was shown to promote RGD-dependent fibroblast adhesion; however, the receptor mediating gB binding was not identified (64). We therefore evaluated the ability of the α₃β₁ integrin, a proposed KSHV entry receptor (2), to mediate RGD-dependent cell adhesion to KSHV gB and compared it to that of α_Vβ₃, one of the integrins predicted to bind the RGD motif of KSHV gB in our sequence alignment study. The α₃-CHO cell line, which overexpresses human-hamster chimeric α₃β₁ integrin (60), was compared to the β₃-CHO cell line, which overexpresses a chimeric α_Vβ₃ integrin (61), for the ability to adhere to gB(RGD), a peptide derived from the N-terminal domain of the KSHV gB (see Fig. 1). Both chimeric integrins have been previously shown to be functional (60, 61). The pBJ-CHO cell line, which lacks an integrin insert, was used as a control. Initial analysis of the CHO cell lines by flow cytometry showed increased surface expression of the appropriate recombinant human integrins (Fig. 2A). To assess adhesion, tissue culture plates were coated with gB(RGD) peptide. Control experiments utilized a gB(RGE) peptide containing a single-amino-acid substitution within the RGD motif. The micrographs in Fig. 2 show that β₃-CHO cells (Fig. 2C) adhered significantly more to gB(RGD) than did the α₃-CHO cells (Fig. 2D) or the pBJ-CHO control cells (Fig. 2E). Adherent β₃-CHO cells showed hallmarks of integrin activation and signaling with a spread morphology and the formation of vinculin-containing focal contacts (data not shown). The adhesion of the β₃-CHO cells was specific since no adhesion to the gB(RGE) control peptide was detected (data not shown). Although the α₃-CHO cells failed to show significant adhesion to gB(RGD), cell adhesion and spreading was observed when they were plated on a known α₃β₁ ligand, such as laminin (Fig. 2F).

FIG. 2.

The α_Vβ₃ integrin mediates cell adhesion to KSHV gB(RGD). (A) Fluorescence-activated cell sorter analysis of integrin levels in the pBJ parental CHO cells compared to the levels in β₃-CHO and α₃-CHO recombinant cells expressing chimeric human-hamster α_Vβ₃ and α₃β₁ integrins, respectively, using anti-α_Vβ₃ (LM609) and anti-α₃ (ASC-6) integrin antibodies. (B) Fluorescence-activated cell sorter analysis of α_Vβ₃ levels in pBJ-CHO, β₃-CHO, and α₃-CHO cells. (C to F) Cell adhesion to immobilized gB(RGD) peptide (10 μg/ml) by β₃-CHO (C), α₃-CHO (D), and the parental pBJ-CHO (E) cells and to immobilized laminin by α₃-CHO cells (F) is shown. Bar, 20 μm. (G) Cell adhesion to increasing concentrations of gB(RGD) peptide. The data represent the mean numbers ± standard errors of adherent cells in quadruplicate wells. Control experiments showed no adhesion to the gB(RGE) peptide with a single-amino-acid mutation in the RGD domain (data not shown).

Cell adhesion was dependent on the amount of immobilized gB(RGD) peptide. At 0.5 μg/ml, there were >500 adherent β₃-CHO cells and no adherent α₃-CHO or pBJ-CHO control cells (Fig. 2G). Maximal β₃-CHO adhesion was observed at 1 μg/ml of gB(RGD). At the highest concentration of gB(RGD) tested, (2 μg/ml), there were a small number of adherent α₃-CHO and pBJ-CHO cells (5% and 13% of the adherent β₃-CHO cells, respectively). To determine whether the binding of the α₃-CHO and pBJ-CHO cells could be due to endogenous hamster α_Vβ₃, the level of α_Vβ₃ expression for each CHO cell line was determined. As shown in Fig. 2B, the pBJ-CHO control cell line expressed low levels of hamster α_Vβ₃ (MFI, 15) compared to the level in β₃-CHO cells (MFI, 54). The α₃-CHO cell line displayed even less α_Vβ₃ integrin (MFI, 6). Thus, the levels of endogenous hamster α_Vβ₃ expression in the different CHO cell lines correlate with their ability to adhere to gB(RGD) peptide. This suggests that the adhesion of the α₃-CHO and pBJ-CHO cells seen at high concentrations of gB(RGD) peptide is mediated by the endogenous hamster α_Vβ₃.

To examine the direct receptor-ligand interactions between α_Vβ₃ and gB(RGD), gB(RGD) peptide-conjugated magnetic beads were used for affinity binding assays. Biotinylated gB(RGD) or gB(RGE) peptides bound to NeutrAvidin magnetic beads were incubated with octyl-beta-glucoside lysates of β₃-CHO or α₃-CHO cells in the presence of 1 mM MgCl₂. The β₃-CHO and α₃-CHO cell lysates contained significant amounts of β₃ (105 kDa) and α₃ (160 kDa) integrins, respectively (Fig. 3, lanes 1 and 4, respectively). The beads were washed, and bound proteins were eluted by using EDTA. RGD interactions are stabilized by divalent cations, and dissociation with metal chelators, such as EDTA, is diagnostic of RGD-mediated binding (56-58). The eluted proteins were analyzed by immunoblotting using antibodies specific for either the α₃ or β₃ integrin. The α_Vβ₃ integrin was detected in the EDTA eluates from the gB(RGD) beads (lane 2), whereas no α₃β₁ integrin was detected (lane 5). Neither integrin was detected in the eluates from the gB(RGE) control beads (Fig. 3, lanes 3 and 6). These studies demonstrate a specific interaction between α_Vβ₃ and the RGD motif of KSHV gB.

FIG. 3.

The RGD motif of KSHV gB specifically binds the α_Vβ₃ integrin. Magnetic beads coated with the KSHV gB(RGD) peptide (lanes 2, 5) or gB(RGE) control peptide (lanes 3, 6) were incubated with lysates from either the β₃-CHO (lanes 2, 3) or α₃-CHO (lanes 5, 6) cells. Bound proteins were eluted with EDTA and characterized by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunoblot analysis using anti-β₃ antibody (MAb 2023) to detect α_Vβ₃ (lanes 1 to 3) and anti-α₃ antibody (MAb 2290) to detect α₃β₁ (lanes 4 to 6). Integrin levels in the initial β₃-CHO (lane 1) and α₃-CHO (lane 4) cell lysates are shown.

KSHV gB(RGD)-mediated cell adhesion is blocked by α_Vβ₃ antibodies.

To show that the chimeric human/hamster integrin receptors overexpressed in CHO cells did not have an altered affinity for ligand, we also performed adhesion assays using the human fibrosarcoma cell line HT1080, which expresses human integrin receptors. In these studies, we evaluated the ability of integrin function-blocking antibodies to inhibit the adhesion of HT1080 cells to the gB(RGD) peptide. HT1080 cells were chosen due to their wide variety of expressed integrin receptors (65). The surface integrin expression on HT1080 cells was first determined by flow cytometry using a panel of antibodies to different α and β integrin subunits and receptor dimers. In all cases, single-peak histograms were observed, and the percentage of HT1080 cells staining with each antibody was high (87 to 99%) (Fig. 4A). Significant levels of the laminin-binding integrin α₃β₁ and the RGD-binding integrins α_Vβ₃, α_Vβ₅, and α₅β₁ were detected; however, the levels of α_Vβ₃ were comparatively low, as has been previously reported (70).

FIG. 4.

α_Vβ₃ function-blocking antibodies inhibit gB(RGD)-mediated HT1080 cell adhesion. (A) Flow cytometry analysis of cell surface integrin levels on HT1080 human fibrosarcoma cells using anti-α₃ (ASC-6), anti-α_Vβ₅ (P1F6), anti-α₅β₁ (JBS5), or anti-α_Vβ₃ (LM609) antibodies. (B) HT1080 cell adhesion to immobilized gB(RGD) peptide (10 μg/ml) in the presence of different concentrations of the function-blocking integrin antibodies used for the experiments whose results are shown in panel A. Adherent cells were quantitated manually. The data represent the mean cell numbers ± standard errors of the results for triplicate wells. Standard errors smaller than the plot symbol are not shown.

HT1080 cell suspensions were pretreated at 4°C with function-blocking antibodies to different integrin receptors prior to being plated on immobilized gB(RGD), as described in Materials and Methods. Antibodies specific for α_Vβ₃ blocked more than 95% of cell adhesion to gB(RGD) at concentrations as low as 0.4 μg/ml (Fig. 4B). These results confirmed the role of α_Vβ₃ in cell adhesion to gB(RGD) seen with the transfected β3-CHO cells, described above. In contrast, function-blocking antibodies to the fibronectin receptor α₅β₁ and the laminin receptor α₃β₁ were unable to block the adhesion of HT1080 cells to gB(RGD). In fact, these antibody treatments increased cell adhesion to 110% and 125%, respectively, of that observed without antibody pretreatment. Antibodies to the RGD-binding integrin, α_Vβ₅, inhibited 30% of adhesion at the highest antibody concentration. These results indicate that the RGD motif of KSHV gB has a strong specificity for integrins containing the α_V integrin subunit, with the strongest reactivity to α_Vβ₃, the vitronectin receptor.

Purified α_Vβ₃ receptor binds directly to gB(RGD) and intact KSHV virions.

To demonstrate a direct interaction between the α_Vβ₃ integrin and KSHV gB, we evaluated the binding of purified soluble α_Vβ₃ integrin receptor to both gB(RGD) peptide and KSHV virion-coated beads. The results were compared to the binding of purified α₃β₁ integrin which has been previously reported to inhibit KSHV infection of fibroblasts (2). Initial experiments were performed to determine the binding specificities of the commercial preparations of affinity-purified α_Vβ₃ and α₃β₁ receptors to gB(RGD) peptide-coated beads. NeutrAvidin beads coated with biotinylated gB(RGD) were incubated with either purified α_Vβ₃ or α₃β₁ heterodimers. Bound receptors were fixed, and the beads were immobilized on poly-l-lysine-coated slides. Confocal immunofluorescence microscopy was used to detect bound receptors by reacting the beads with antibodies to either α_Vβ₃ or α₃β_1. As shown in Fig. 5, α_Vβ₃ receptor was detected on the surface of the gB(RGD)-coated beads (Fig. 5A), but not on the control gB(RGE)-coated beads (Fig. 5B). In contrast, the purified α₃β₁ receptor failed to bind to either the gB(RGD)- or gB(RGE)-coated beads (Fig. 5C, D). These results are consistent with the binding data obtained by using whole-cell lysates that are shown in Fig. 3.

FIG. 5.

Purified α_Vβ₃ receptor specifically binds the KSHV gB(RGD) peptide. Biotinylated gB(RGD) and control gB(RGE) peptides were coupled to magnetic beads and incubated with commercial preparations of affinity-purified functional α_Vβ₃ or α₃β₁ integrin receptors (R). Bound α_Vβ₃ or α₃β₁ integrins were detected by using anti-β₃ (A, B) or anti-α₃ (α₃β₁) (C, D) antibodies and confocal microscopy (green fluorescence).

In order to examine the binding of purified integrins to native envelope-associated gB present on intact virions, KSHV virions were purified from TPA-induced BCBL-1 cell cultures by density gradient centrifugation in Opti-Prep (iodixanol) and biotinylated, as described in Materials and Methods. Real-time PCR analysis of KSHV DNA in the gradient fractions showed a single viral peak at the 25%-to-30% Opti-Prep interface (Fig. 6A). Biotinylated virions in the peak fractions were immobilized on poly-l-lysine-coated slides and characterized by confocal immunofluorescence microscopy using antibodies to biotin and to the KSHV envelope-associated K8.1 glycoprotein. Discrete particles were detected which showed a high degree of colocalization of biotin (Fig. 6B) and K8.1 (Fig. 6C) staining, consistent with the presence of intact, enveloped KSHV virions suitable for receptor binding studies.

The gradient-purified biotinylated KSHV virions were coupled to NeutrAvidin-coated beads and fixed. As a binding control, NeutrAvidin beads were coupled to gradient-purified preparations of biotinylated virions of RRV, a macaque herpesvirus that is closely related to KSHV (15). Sequence analysis of the RRV genome has identified a close homolog of the KSHV gB (53) which lacks a putative integrin-binding RGD motif. The relative amounts and distribution of virions coating the beads were determined by confocal immunofluorescence microscopy using the anti-K8.1A antibody for KSHV and the antibiotin antibody for RRV. Highly concentrated KSHV virions were detected at the bead surface, with some localized enhanced fluorescence concentrations possibly indicating areas of KSHV aggregation (Fig. 7A). The surface density of biotinylated RRV was similar to that of KSHV, but the uniform immunofluorescence pattern suggested fewer surface RRV aggregates (Fig. 7B). The KSHV and RRV virion-coupled beads were incubated with affinity-purified α_Vβ₃ and α₃β₁ integrin receptors, and bound receptor was detected by confocal immunofluorescence using antibodies to β₃ and α₃ integrins. The antibodies to β₃ reacted strongly with the KSHV-coated beads, demonstrating the presence of bound α_Vβ₃ receptor (Fig. 7C). The pattern of fluorescence was similar to that seen with the anti-K8.1 antibody, indicating that the α_Vβ₃ integrin receptor was bound to the KSHV virions. No binding of the affinity-purified α₃β₁ integrin to the KSHV-coated bead surface was detected (Fig. 7E). The binding of purified α_Vβ₃ to the KSHV virion was specific since no binding was detected to the control macaque RRV virions which lack the gB RGD motif (Fig. 7D). The purified α₃β₁ integrin receptor was also unable to bind to the RRV virions (Fig. 7F).

FIG. 7.

Purified α_Vβ₃ receptor specifically binds to KSHV virions. Gradient-purified KSHV and RRV virions which had been biotinylated and coupled to magnetic beads were reacted with anti-K8.1 (A) or antibiotin (B) antibodies, respectively, to monitor virus-bead coupling by confocal microscopy. −, no receptor. The virion-coated beads were incubated with commercial preparations of affinity-purified α_Vβ₃ integrin (C, D) or α₃β₁ integrin (E, F), and bound integrin was detected by confocal microscopy using either anti-β₃ (B3A) (C, D) or anti-α₃ (ASC-6) (E, F) antibodies. Arrows show localized enhanced fluorescence concentrations possibly indicating areas of KSHV aggregation. Bound antibodies were detected as green fluorescence.

Soluble α_Vβ₃ integrin in detergent-extracted cell lysates specifically binds KSHV virions.

To determine if α_Vβ₃ associates with KSHV in the presence of other cellular proteins within whole-cell lysates, gradient-purified biotinylated KSHV virions were coupled to beads, and conjugated virus was detected by confocal immunofluorescence using antibody to the K8.1 envelope glycoprotein (Fig. 8E). The KSHV-coupled beads were incubated with octyl-beta-glucoside lysates of β₃-CHO or α₃-CHO cells. d-Biotin-conjugated beads were used as a control. Bound integrins were detected by immunofluorescence microscopy. High levels of the β₃ integrin from the β₃-CHO lysate bound to the KSHV beads (Fig. 8A). No binding to the control beads was detected (Fig. 8B). The α₃ integrin from lysates of the α₃-CHO cells was also detected on the KSHV beads, but at a lower level than in the β₃ lysates (Fig. 8C). However, similar amounts of α₃ also bound to the control beads containing no virus, suggesting that the binding was nonspecific. Quantitation of the bead-associated fluorescence revealed a significant and specific reaction between α_Vβ₃ in cell lysates and the purified KSHV virions (Fig. 8F, G). This binding is similar to that seen with the purified α_Vβ₃ receptor, described above. No significant association was detected between the KSHV virions and the α₃β₁ integrin even in the context of other cellular proteins in the cell lysate (Fig. 8F, G). These studies were repeated using lysates of human HT1080 cells which express significant levels of functional α_Vβ₃ and α₃β₁ integrins with the same results (data not shown).

FIG. 8.

Soluble α_Vβ₃ from cell lysates specifically binds to KSHV virions. Lysates from cells overexpressing either α_Vβ₃ (β₃-CHO) or α₃β₁ (α₃-CHO) were incubated with gradient-purified KSHV virions coupled to magnetic beads (A, C) or control beads (B, D) with no virus. Bound α_Vβ₃ and α₃β₁ were detected by using anti-β₃ (B3A) (A, B) or anti-α₃ (ASC-6) (C, D) antibodies, respectively. The presence of bead-bound KSHV was demonstrated by anti-K8.1 antibody immunofluorescence (E). Individual pixels of bead fluorescence (intensity range, 100 to 255) were quantitated, and the mean (n = 10) sum ± standard deviation of pixels was calculated (F). Specific binding to KSHV was determined by subtracting the control bead fluorescence (G). No significant binding of α₃β₁ was detected.

KSHV cell entry is inhibited by α_Vβ₃ antagonists.

Since our data showed a strong reactivity between α_Vβ₃ and KSHV virions, we performed infection-blocking studies to determine whether the α_Vβ₃ integrin functions as a KSHV entry receptor. Cyclic RGD peptides are selective α_Vβ₃ antagonists and block α_Vβ₃ integrin binding at concentrations 80-fold lower than their linear counterparts (21, 43). Furthermore, changing the ligand valency by converting monomeric peptide to dimeric increases the affinity for α_Vβ₃ (34). Therefore, we tested the ability of the cyclic RGD peptides to block KSHV infection mediated by α_Vβ₃. cRGD and dcRGD were chosen since these peptides have high affinities for α_Vβ₃ and low affinities for the α_Vβ₅, α_IIbβ₃, and α₅β₁ integrins (25, 28). The inhibitory activities of the cyclic peptides were compared to the activity of the linear GRGDS peptide derived from the fibronectin sequence. The linear peptide GRGES and the cyclic peptide cRAD, which both contained single-amino-acid alterations in the RGD motif, were used as controls.

Adherent HT1080 cells were trypsinized, chilled (4°C) to prevent receptor internalization, and pretreated with peptide before infection with gradient-purified KSHV, as described in Materials and Methods. Latent infection was assayed at 24 h by confocal immunofluorescence detection of KSHV LANA. In the absence of inhibitor, approximately 75% of the HT1080 cells were infected with KSHV (Fig. 9A). Cells treated with the control peptides, cRAD and GRGES, showed only a small decrease in the number of LANA-expressing cells. In contrast, the α_Vβ₃-specific cRGD peptide and the linear GRGDS peptide both inhibited approximately 78% of KSHV infection, while the dcRGD peptide inhibited approximately 64% (Fig. 9A).

FIG. 9.

KSHV infection is inhibited by α_Vβ₃-specific ligands and function-blocking antibodies. Infection-blocking studies were performed by pretreating HT1080 cells (4°C) with RGD-containing peptide ligands, including the linear GRGDS peptide and the α_Vβ₃-specific cyclic (cRGD) and dicyclic (dcRGD) peptides or the α_Vβ₃ function-blocking antibody (LM609). Controls included the linear GRGES and cyclic cRAD peptides and normal mouse IgG. Gradient-purified KSHV virions titrated to yield an infection in 75% of the cells were incubated with the cells for 1 h (4°C) in the presence of peptide or antibody. Unbound KSHV was removed by washing, and the percentage of cells expressing KSHV LANA was determined 24 h later in multiple fields by confocal microscopy (A). The standard deviation is shown. Representative micrographs showing nuclear LANA expression in control cRAD-treated (B) and cRGD-treated (C) HT1080 cells infected with KSHV are shown. Arrows indicate representative dots of LANA fluorescence. LANA fluorescence was quantitated, and the frequency and intensity (range, 100 to 255) of fluorescent pixels within LANA dots in individual nuclei (n = 30) were plotted (D).

Multiple large dots of LANA fluorescence were detected within the nuclei of cells infected in the absence of inhibitor (data not shown) and cells infected in the presence of the control cRAD peptide (Fig. 9B). A noticeable reduction in the number of LANA dots and the level of LANA expression in each dot was observed in the few cells infected in the presence of the α_Vβ₃ cRGD antagonist (Fig. 9C). To quantitate changes in LANA expression, the fluorescence intensities of the LANA dots in 30 LANA-positive nuclei from the both cRAD- and cRGD-treated cells were compared. The frequency of fluorescent pixels and the intensity of each pixel (spanning a range of intensities from weakly fluorescent (100) to the brightest pixels (255) were determined for each dot. As shown in Fig. 9D, LANA expression was inhibited by treatment with the cRGD peptide across the intensity range. More importantly, there was an approximate fivefold reduction in the frequency of the brightest LANA pixels (intensity of 255).

To further substantiate the role of the α_Vβ₃ integrin in KSHV entry, the function-blocking anti-α_Vβ₃ MAb LM609 was tested for its ability to inhibit KSHV infection. As shown in Fig. 9A, the LM609 antibody was a potent inhibitor of KSHV infection and inhibited infection levels by 75%, equivalent to the results obtained with the cRGD and GRGDS peptides. No inhibition was detected with the normal IgG control antibodies. These studies and those with the α_Vβ₃ peptide antagonists demonstrate that specific blocking of the RGD-mediated interaction with the α_Vβ₃ receptor can inhibit KSHV infection and establishment of latency. The similar levels of inhibition seen with the cyclic and linear RGD peptides strongly implicate α_Vβ₃ as a major KSHV entry receptor.

KSHV infection of mouse keratinocytes is independent of α₃β₁ integrin.

Although our studies failed to confirm the previously reported interaction between KSHV and the α₃β₁ integrin, the possibility remained that α₃β₁ could facilitate KSHV entry by synergizing with α_Vβ₃ or other receptors, such as the xCT heavy chain which associates with integrins (35). We therefore compared the ability of KSHV to infect mouse keratinocytes derived from (i) wild-type mice (MK^+/+), (ii) homozygous α₃-null mice which lack functional cell surface α₃β₁ integrin receptors (MK^−/−), or (iii) α₃-null MK cells engineered to produce recombinant α₃β₁ (MK^−/−,α3) (16). The α₃-null and wild-type mouse keratinocytes express equivalent amounts of other integrins, including α_V (16). The mouse keratinocytes were infected with different amounts of gradient-purified KSHV, and infection was quantitated 24 h later by immunofluorescence detection of nuclear LANA expression. The Vero cell line, which is highly infectible by KSHV, was used as a positive infection control. The virus infection curves for both the wild-type MK^+/+ and the α₃-null MK^−/− cells were very similar, although at all points the infection of the α₃-null MK^−/− cells was higher than that of the wild-type cells, attaining 74% at the highest virus dose tested (Fig. 10). By comparison, this virus dose yielded >95% infection of the Vero cells. In contrast, MK^−/−,α3 cells which overexpress the α₃β₁ integrin showed decreased infection levels, attaining only 55% infection of the cell culture at the highest virus dose tested. These results, which were corroborated in three independent experiments, indicate that KSHV infection of mouse keratinocytes is independent of α₃β₁. In fact, the reduced level of infection of the α₃-null mouse keratinocytes transfected with recombinant α₃A integrin cDNA suggested that the overexpression of α₃β₁ negatively affected KSHV entry and establishment of a latent infection.

FIG. 10.

KSHV infection of mouse keratinocytes is independent of α₃β₁ integrin. Wild-type mouse keratinocytes (MK^+/+), keratinocytes derived from α₃-null mice which lack α₃β₁ integrins (MK^−/−), or α₃-null cells transfected with human α₃A integrin (MK^−/−,α3) to reconstitute α₃β₁ expression were infected with different amounts of gradient-purified KSHV. Twenty-four hours postinfection, the percentage of infected cells was determined by quantitating cells expressing KSHV LANA using confocal microscopy. The KSHV-susceptible Vero cell line was used as a positive control. Cells analyzed: α₃-null MK^−/− (n = 353); MK^+/+ (n = 532); MK^−/−,α3 (n = 451); Vero (n = 293).

DISCUSSION

This study demonstrates that the α_Vβ₃ integrin is a cellular receptor mediating both the cell adhesion and entry of KSHV into target cells through binding to the RGD motif of the virion-associated gB. This conclusion is supported by multiple lines of evidence. First, analysis of the RGD motif of the KSHV gB revealed strong sequence similarities to analogous motifs in known ligands of the α_Vβ₃ integrin receptor. Second, the gB(RGD) peptide, derived from the KSHV gB sequence, promoted RGD-dependent cell adhesion that was specifically blocked by an α_Vβ₃ function-blocking MAb. Function-blocking antibodies to α₅β₁or α₃β₁ integrins were unable to block adhesion. Antibodies to α_Vβ₅ partially blocked adhesion, supporting the role of the α_V integrin subunit in gB(RGD)-integrin interactions. Third, the overexpression of α_Vβ₃ integrin receptors strongly increased the adherence of β₃-CHO cells to gB(RGD) over that seen with the parental pBJ-CHO cells or α₃-CHO cells overexpressing the α₃β₁ integrin. Fourth, immunoblot analysis revealed that the gB(RGD) peptide specifically bound to α_Vβ₃ receptors in octyl-beta-glucoside lysates of β₃-CHO cells in a divalent cation-dependent manner. Fifth, confocal microscopy demonstrated a specific binding between affinity-purified α_Vβ₃ receptors and both gB(RGD) peptide and gradient-purified KSHV virions. Finally, KSHV infection of HT1080 fibrosarcoma cells was specifically blocked by α_Vβ₃-specific cyclic RGD peptides and a specific α_Vβ₃ function-blocking antibody.

The α_Vβ₃ integrin is a cell surface receptor that is widely expressed in proliferating endothelial cells, arterial smooth muscle cells, and certain populations of leukocytes and tumor cells (23). Integrin α_Vβ₃ mediates the adhesion of cells to a number of extracellular matrix proteins containing RGD motifs, including vitronectin and fibronectin, and is responsible for mediating cell-cell interactions through binding to cell-associated glycoproteins containing RGD motifs. The α_Vβ₃ integrin has been implicated in cellular processes ranging from wound healing to tumor angiogenesis and tumor progression. In addition, α_Vβ₃ has been coopted by a number of pathogens, including foot-and-mouth disease virus, coxsackievirus, hantavirus, and adenovirus (45), as a receptor for the entry of target cells during the infection process. Our finding that α_Vβ₃ functions as an entry receptor for KSHV further expands the number of viruses that utilize this receptor for entry. In KS tumors, α_Vβ₃ is highly expressed on KS tumor spindle cells and in neovascular endothelial cells and B cells that infiltrate KS tumors (8, 26, 51, 62). These cells are also consistently infected with KSHV in KS lesions, suggesting a central role for α_Vβ₃ in KSHV infection and in the pathogenesis of KS.

The ability of the integrin α_Vβ₃ to bind RGD-containing proteins has been exhaustively reported in the literature. Our finding that α_Vβ₃ binds to the RGD motif of KSHV gB confirms this specificity. Ligand-binding studies have shown that α_Vβ₃ and the closely related α_Vβ₅ interact specifically with peptides containing the consensus sequence RGDXF, where X is S or T (33). Panning studies of phage libraries displaying cyclic peptides demonstrated that both the α_Vβ₃ and α_Vβ₅ integrins selectively bound the peptide sequence SRGDTF, which constitutes the core of the KSHV RGD motif (33). In contrast, the α₅β₁ and α_IIbβ₃ integrins, which also bind RGD motifs, did not select peptides with sequences similar to that of the KSHV RGD motif. These panning studies correlate with the results of our antibody-blocking studies which showed that antibodies to α_Vβ₃ and α_Vβ₅ could significantly block binding to the KSHV gB(RGD) peptide, whereas antibodies to α₅β₁ could not. These data strongly support the hypothesis that the RGD motif of KSHV gB specifically interacts with integrins containing the α_V integrin subunit, especially α_Vβ₃.

A previous study has shown that de novo infection of naïve cells with KSHV results in a Poisson distribution of viral episomes within the infected cells (1). Furthermore, the number and summed immunofluorescence of LANA dots in the infected cell nuclei were directly proportional to the amount of intracellular viral DNA. Our results showed that blocking KSHV infection with α_Vβ₃ antagonists not only decreased the number of infected cells but also decreased the amount of LANA fluorescence in those cells that were infected. This suggests that blocking viral entry can reduce the number of virions that enter an infected cell and decrease the number of LANA dots and associated immunofluorescence. Studies are ongoing to confirm this.

A previous study reported that α₃β₁ interacts with KSHV gB and functions as a cellular entry receptor for KSHV (2). It was suggested that this interaction may be RGD mediated. However, recent studies have shown that α₃β₁ binding to its high-affinity ligands is RGD independent (11, 14, 20, 29, 30, 37, 41, 66, 69). Other studies using purified recombinant α₃β₁ detected no binding to RGD-containing proteins, such as fibronectin (17), while transfection studies demonstrated that the α₃β₁ integrin is not sufficient for cell attachment to matrices containing RGD motifs (14, 66). Furthermore, the characterized α₃β₁ recognition motifs NLR (37) and PPFLMLLKGSTR (31) are not present in the KSHV gB sequence. Our studies revealed no interaction between α₃β₁ and the RGD motif of KSHV gB, and α₃β₁ function-blocking antibodies failed to inhibit cell adhesion to the gB(RGD) peptide. Although α₃-CHO cells overexpressing α₃β₁ showed enhanced adhesion to a laminin substrate, no adhesion to the gB(RGD) peptide greater that detected with the parental pBJ-CHO cells was observed. Affinity purification studies revealed no binding between the gB(RGD) peptide and solubilized α₃β₁ receptors from octyl-beta-glucoside lysates of α₃-CHO cells. In addition, the gB(RGD) peptide and the native envelope-associated gB present on gradient-purified KSHV virions were unable to bind to commercial preparations of affinity-purified α₃β₁ receptors. Infection studies revealed that α₃-null mouse keratinocytes lacking α₃β₁ integrin were competent for infection by KSHV, and the expression of α₃β₁ in either the parental mouse keratinocytes or α₃-null mouse keratinocytes transfected with human α₃ decreased infection levels.

While our study failed to detect a direct interaction between α₃β₁ and the RGD motif of KSHV gB, it did not directly address a potential secondary role for α₃β₁ in KSHV binding, entry, or downstream signaling. Studies have shown colocalization of α_Vβ₃ and α₃β₁ integrin at cell surface focal adhesions due to shared interactions with other molecules, such as actin, focal adhesion kinase, vinculin, α-actinin, and paxillin which form the adhesion plaque (10, 13, 68). It has been suggested that the α₃β₁ integrin may mediate secondary responses after other integrins, such as α_Vβ₃, bind their specific ligands (for a review, see reference 6). These secondary responses could include “integrin cross talk” and the modulation of downstream signaling or the recruitment of cytoskeleton and other integrins into membrane microdomains involved in KSHV entry. The data from our anti-α₃ blocking studies and α₃-CHO adhesion studies corroborate the results of others who have shown that α₃β₁ can modulate α_Vβ₃ function (7, 22). Our infection studies revealed that α₃-null mouse keratinocytes, which are completely lacking in α₃β₁, are competent for KSHV infection, bringing into question the importance of a secondary response, at least in mouse keratinocytes. However, it is possible that α₃β₁ plays a role in KSHV entry in other cell types, since our studies were based on different experimental approaches and target cells than used previously (2).

Several lines of evidence from our study confirm the transdominant inhibitory effect of α₃β₁ and other integrins containing the β₁ subunit on the receptor function of α_Vβ₃ that has been seen by others (7, 22, 32, 38). First, the antibody blocking of α₃β₁ function in our cell adhesion assays led to an increased level of α_Vβ₃-mediated adhesion of HT1080 cells to the gB(RGD) peptide. Second, the overexpression of α₃ integrin in the α₃-CHO cells resulted in a 60% decrease in the surface expression of α_Vβ₃ and a 62% decrease in α_Vβ₃-dependent adhesion of the α₃-CHO cells to the gB(RGD) peptide compared to those of the pBJ parental CHO cells. Third, the overexpression of α₃ integrin in the rescued α₃-null mouse keratinocytes resulted in decreased levels of KSHV infection compared to the levels in both α₃-null mouse keratinocytes and wild-type mouse keratinocytes. The inhibitory effect of α₃β₁ on α_Vβ₃ function may explain the ability of soluble α₃β₁ to inhibit KSHV infection, as reported previously (2, 39, 55).

Several previous studies have implicated integrin binding and the downstream activation of integrin signaling pathways in the KSHV infection process (36, 40, 54, 64). Our data showing that α_Vβ₃ functions as a major integrin receptor mediating RGD-dependent interactions with the KSHV gB during viral entry are compatible with the results of these studies.