{beta}1 Integrin Mediates Internalization of Mammalian Reovirus

Reovirus infection is initiated by interactions between theattachment protein ${sigma}$ 1 and cell surface carbohydrate and junctionaladhesion molecule A (JAM-A). Expression of a JAM-A mutant lackinga cytoplasmic tail in nonpermissive cells conferred full susceptibilityto reovirus infection, suggesting that cell surface moleculesother than JAM-A mediate viral internalization following attachment.The presence of integrin-binding sequences in reovirus outercapsid protein ${lambda}$ 2, which serves as the structural base for ${sigma}$ 1,suggests that integrins mediate reovirus endocytosis. A ß1integrin-specific antibody, but not antibodies specific forother integrin subunits, inhibited reovirus infection of HeLacells. Expression of a ß1 integrin cDNA, along witha cDNA encoding JAM-A, in nonpermissive chicken embryo fibroblastsconferred susceptibility to reovirus infection. Infectivityof reovirus was significantly reduced in ß1-deficientmouse embryonic stem cells in comparison to isogenic cells expressingß1. However, reovirus bound equivalently to cellsthat differed in levels of ß1 expression, suggestingthat ß1 integrins are involved in a postattachmententry step. Concordantly, uptake of reovirus virions into ß1-deficientcells was substantially diminished in comparison to viral uptakeinto ß1-expressing cells. These data provide evidencethat ß1 integrin facilitates reovirus internalizationand suggest that viral entry occurs by interactions of reovirusvirions with independent attachment and entry receptors on thecell surface.

INTRODUCTION

Top

Introduction

Viral attachment and cell entry are key determinants of targetcell selection in the infected host and thus play importantroles in pathogenesis. Many viruses interact with multiple cellsurface molecules to mediate the processes of attachment andinternalization (68). For example, human immunodeficiency virususes CD4 to bind the cell surface and chemokine receptors tofacilitate the conformational alterations in envelope glycoproteinsthat culminate in fusion of the viral envelope and cell membrane(35). Receptors that serve as initial binding sites have beenidentified for many viruses (25). However, little is known aboutthe postattachment events that lead to nonenveloped virus internalization,in particular those that mediate virus uptake into the endocyticpathway.

Mammalian reoviruses are large, nonenveloped, double-strandedRNA-containing viruses that infect a variety of mammalian species.Following attachment to target cells, reoviruses are internalizedby receptor-mediated endocytosis (3, 13, 14, 72), which is mostlylikely to be clathrin dependent (31). Proteolytic disassemblyin endosomes leads to removal of outer capsid protein ${sigma}$ 3 andcleavage of outer capsid protein µ1 (3, 14, 27, 72). Theresultant disassembly intermediate formed by these events, theinfectious subvirion particle (ISVP), is capable of penetratingendosomal membranes in a µ1-dependent manner to releasethe transcriptionally active viral core particle into the cytoplasm(17, 18, 58), where viral replication takes place. Cellulardeterminants of reovirus receptor-mediated internalization followingattachment and preceding uncoating are poorly defined.

We previously identified junctional adhesion molecule A (JAM-A)as a serotype-independent receptor for reovirus (5, 16, 34).JAM-A is a type 1 transmembrane protein expressed in a varietyof cell types, including polarized endothelial and epithelialcells and circulating leukocytes (52, 55, 81). JAM-A interactswith several scaffolding proteins and cytoplasmic adaptor molecules(6, 28, 29) and is hypothesized to play an important role inmaintaining the barrier function of epithelial junctions (48,52, 55, 60). JAM-A is phosphorylated during platelet activationand required for mitogen-activated protein kinase activationfollowing treatment of endothelial cells with basic fibroblastgrowth factor (57). These data indicate that JAM-A is intimatelyassociated with cytoskeletal and signaling machinery, whichraises the possibility that reovirus binding to JAM-A mediatescytoskeletal rearrangement or signaling events to facilitatevirus internalization.

The attachment mechanisms of reovirus and adenovirus are remarkablysimilar (70, 71). The trimeric attachment proteins of both viruses, ${sigma}$ 1 and fiber, respectively, are structural homologues and foldusing a highly unusual triple ß-spiral motif (10,20, 83). The globular head domains of these molecules are formedfrom eight-stranded ß-barrels with identical interstrandconnectivity (70). The receptors for ${sigma}$ 1 and fiber, JAM-A (5)and coxsackievirus and adenovirus receptor (CAR) (7), respectively,are two-domain, immunoglobulin superfamily proteins that formhomodimers using analogous molecular surfaces (71). Also, bothJAM-A and CAR localize to tight junctions in polarized epithelialcells (23, 52, 55, 60). Remarkably, reovirus and adenovirusengage their respective receptors by thermodynamically favoreddisruption of receptor homodimers (34, 53).

Despite mediating high-affinity attachment of adenovirus tocells, engagement of CAR does not permit efficient adenovirusinternalization. Instead, adenovirus entry is enhanced by high-avidityinteractions of the viral penton base complex with integrins,including ${alpha}$ vß3 and ${alpha}$ vß5 (80). Integrins areheterodimeric cell surface molecules that consist of ${alpha}$ and ßsubunits (43). Integrins function to mediate cellular adhesionto the extracellular matrix, regulate cellular trafficking,and transduce both outside-in and inside-out signaling events(42). In addition to adenovirus, several other pathogenic microorganismshave usurped the adhesion and signaling properties of integrinsto bind or enter host cells (1, 8, 32, 39-41, 44, 49).

To define the molecular basis of reovirus internalization, wefirst tested the capacity of a JAM-A mutant lacking a cytoplasmictail to support reovirus attachment and infection. We foundthat while JAM-A is necessary for efficient attachment to cells,the JAM-A cytoplasmic tail is not required for reovirus infection.Given the mechanistic conservation of reovirus and adenovirusattachment strategies and the observation that reovirus outercapsid protein ${lambda}$ 2 contains the conserved integrin-binding sequencesArg-Gly-Asp (RGD) and Lys-Gly-Glu (KGE), we tested the roleof integrins in reovirus internalization. We found that infectionby reovirus virions is inhibited by antibodies specific forß1 integrin. In addition, cells deficient in ß1integrin have a diminished susceptibility to reovirus infectiondue to a postattachment block to viral entry. Together, thesedata indicate that, following attachment to JAM-A, ß1integrin facilitates internalization of reovirus into cells.Our findings further demonstrate that two seemingly unrelatedviruses utilize distinct cellular molecules to mediate attachmentand internalization in a remarkably similar manner.

MATERIALS AND METHODS

Top

Materials and Methods

Cells, viruses, and antibodies. Spinner-adapted murine L929 (L) cells were grown in either suspensionor monolayer cultures in Joklik's modified Eagle's minimal essentialmedium (Irvine Scientific, Santa Ana, CA) supplemented to contain5% fetal bovine serum, 2 mM L-glutamine, 100 U of penicillinper ml, 100 U of streptomycin per ml, and 0.25 mg amphotericinper ml (Gibco Invitrogen Corp., Grand Island, NY). Chinese hamsterovary (CHO) cells were maintained in Ham's F12 medium (IrvineScientific) supplemented to contain 10% fetal bovine serum,100 U of penicillin per ml, and 100 U of streptomycin per ml.HeLa cells were maintained in Dulbecco's modified Eagle's medium(Gibco Invitrogen Corp.) and supplemented as described for CHOcells. Primary cultures of chicken embryo fibroblasts (CEFs)were obtained from Paul Spearman (Vanderbilt University) andmaintained in medium 199 with Earle's salts and 2.2 mg sodiumbicarbonate per ml (Gibco Invitrogen Corp.) supplemented tocontain 5% fetal bovine serum, 10% tryptose phosphate broth,1% chicken serum (Gibco Invitrogen Corp.), and antibiotics asdescribed for CHO cells. GD25 and GD25ß1A cells wereobtained from Deane Mosher (University of Wisconsin, Madison)(78) and maintained as described for HeLa cells. Medium forGD25ß1A cells was supplemented to contain 10 µgof puromycin (Sigma-Aldrich, St. Louis, MO) per ml to maintainß1 integrin expression.

Reovirus strains type 1 Lang (T1L) and type 3 Dearing (T3D)are laboratory stocks. Working stocks of virus were preparedby plaque purification and passage in L cells (75). Purifiedvirions were generated from second-passage L-cell lysate virusstocks. Virus was purified from infected cell lysates by Freonextraction and CsCl gradient centrifugation as described (36).Bands corresponding to the density of reovirus particles (1.36g/cm³) were collected and dialyzed against virion storage buffer(150 mM NaCl, 15 mM MgCl₂, 10 mM Tris-HCl [pH 7.4]). Reovirusparticle concentration was determined by the equivalence of1 unit of optical density at 260 nm to 2.1 x 10¹² particles(69).

Viral infectivity titers were determined by either plaque assay(75) or fluorescent focus assay (4). ISVPs were generated bytreatment of 2 x 10¹¹ virion particles per ml with 2 mg of ${alpha}$ -chymotrypsin(Sigma-Aldrich) per ml in a volume of 100 µl virion storagebuffer at 37°C for 30 min (2). Reactions were terminatedby the addition of phenylmethylsulfonyl fluoride to a finalconcentration of 1.0 mM. Purified T1L virions in carbonate-bicarbonatebuffer (Sigma-Aldrich) were fluoresceinated by incubation with50 µg fluorescein isothiocyanate (FITC) (Pierce, Rockford,IL) per ml at room temperature for 1 h (38). Excess FITC wasremoved by exhaustive dialysis against phosphate-buffered saline(PBS).

Immunoglobulin G (IgG) fractions of rabbit antisera raised againstT1L and T3D (79) were purified by protein A-Sepharose as previouslydescribed (4). Fluorescently conjugated secondary Alexa antibodieswere obtained from Molecular Probes (Invitrogen, San Diego,CA). The human JAM-A (hJAM-A)-specific monoclonal antibody (MAb)J10.4 and control mouse ascites were provided by Charles Parkos(Emory University School of Medicine) (52), and the murine JAM-A(mJAM-A)-specific MAb H202-106-7-4 was provided by Beat Imof(University of Geneva). The human ${alpha}$ 2-specific MAb AA10 (IgM)(8) and human ß1-specific MAb DE9 (IgG1) (8) wereused as diluted ascites. Human integrin-specific MAbs MAB1980( ${alpha}$ v), MAB1973 ( ${alpha}$ 1), MAB2057 ( ${alpha}$ 3), MAB1378 ( ${alpha}$ 6), MAB1976 ( ${alpha}$ vß3),and MAB1961Z ( ${alpha}$ vß5) were purchased from Chemicon International(Temecula, CA). Antibody BIIG2 ( ${alpha}$ 5) (Developmental HybridomaStudies Bank, University of Iowa, Iowa City, IA) was providedby John Williams (Vanderbilt University). Function-blockinghuman ${alpha}$ 2-specific MAb 6F1 was provided by Richard Bankert (StateUniversity of New York at Buffalo). Function-blocking murineß1 MAb CD29 (IgM) and hamster IgM isotype controlwere purchased from BD Biosciences Pharmingen (San Jose, CA).Murine ß1-specific MAb MAB1997 (Chemicon) and humanß1-specific MAb MAB2253Z (Chemicon) were used to assessexpression of ß1 integrin on GD25 and GD25ß1Acells and HeLa cells, respectively, by flow cytometry. ICAM-1-specificMAb was purchased from Santa Cruz Biotechnology (Santa Cruz,CA). The antibodies used for flow cytometric analysis of HeLacells are shown in Table 1.

View this table:
[in this window]
[in a new window]
TABLE 1. Surface expression of integrins on HeLa cells

Sequence analysis. The sequences of the reovirus ${lambda}$ 2-encoding L2 gene from strainsT1L (NC_004259), type 2 Jones (T2J) (NC_004260), T3D (NC_004275),T1Neth85 (AF378004), T2SV59 (AF378006), T3C9 (AF378007), T3C18(AF378008), T3C87 (AF378009), and T3C93 (AF378010) were alignedusing the protein sequence alignment algorithm in MacVector,version 8.0.1 (Accelrys, San Diego, CA).

Plasmid constructs. Human JAM-A was subcloned into expression plasmid pcDNA3.1+(Invitrogen) (34). Truncation mutant JAM-A- ${Delta}$ CT was generatedby PCR using full-length JAM-A cDNA as the template. Amino acids1 to 260 ( ${Delta}$ 261-299) were cloned and appended with a stop codonusing T7 primer and 5'-TACGGGATCCTCAGGCAAACCAGATGCC-3' as theforward and reverse primers, respectively. The gene-specificprimer encompasses nucleotides 981 to 995 of the JAM-A cDNA.The PCR product was digested with BamHI (recognition site underlinedin the reverse primer sequence) and subcloned into the complementaryrestriction sites of pcDNA3.1+. Fidelity of cloning was confirmedby automated sequencing. Plasmid constructs encoding murineintegrin ${alpha}$ v (33) and ${alpha}$ 2 (30) were previously described. A cDNAencoding murine ß1 integrin cloned into the EcoRIsite of pGEM1 was obtained from Richard Hynes (MassachusettsInstitute of Technology) (D. W. DeSimone, V. Patel, and R. O.Hynes, unpublished). Integrin cDNAs were subcloned into theexpression plasmid pcDNA3.1+.

Transient transfection of CHOs and CEFs. Monolayers of cells in a 24-well plate (Costar, Cambridge, MA)were transfected with empty vector or plasmids encoding receptorconstructs by using Lipofectamine Plus reagent (Invitrogen).Cells were incubated for 24 h to allow receptor expression priorto adsorption with either reovirus virions or ISVPs for infectivitystudies.

Flow cytometric analysis. Surface expression of integrin subunits on HeLa cells was determinedby flow cytometry. Cells were detached from plates by usingPBS-EDTA (20 mM EDTA). Cells were washed and centrifuged at2,000 x g to form a pellet, resuspended with integrin-specificor control antibodies in PBS- bovine serum albumin (BSA) (Sigma-Aldrich)(5% BSA), and incubated at 4°C for 1 h. Cells were washedtwice and incubated with an appropriate secondary antibody conjugatedto R-phycoerythrin (BD Biosciences Pharmingen) at 4°C for1 h. Cells were washed, resuspended in PBS, and analyzed byflow cytometry. Results were analyzed using the Windows MultipleDocument 2.8 flow cytometry application (The Scripps ResearchInstitute, La Jolla, CA).

The mean fluorescence intensity was measured for an averageof 14,000 gated events for cells treated with control or integrin-specificantibodies. Events were gated relative to cells stained withan appropriate secondary antibody conjugated to phycoerythrin.Reovirus binding to GD25 and GD25ß1A cells was analyzedby adsorbing cells with 2 x 10¹¹ FITC-labeled particles of strainT1L at 4°C for 1 h. Cells were washed and analyzed by flowcytometry.

Fluorescent focus assay of viral infection. Cells were plated in 24-well or 96-well plates (Costar) andadsorbed with virus at various multiplicities of infection (MOIs)at either 4°C or room temperature for 30 to 60 min. Inoculawere removed, cells were washed, and complete medium was added.Infected cells were incubated at 37°C for 16 to 24 h toallow a single cycle of viral replication. Cells were fixedwith methanol at –20°C for at least 30 min. Fixedcells were incubated with PBS-BSA (5% BSA) for at least 15 min,followed by incubation with reovirus-specific polyclonal antiserum(1:500) in PBS-Triton X-100 (0.5% Triton X-100) at room temperaturefor 1 h. Cells were washed twice and incubated with an Alexa488- or 546-labeled anti-rabbit IgG (1:1,000) in PBS-TritonX-100 (0.5% Triton X-100) at room temperature for 1 h.

Cells were washed twice and visualized by indirect immunofluorescenceat a magnification of 20x using an Axiovert 200 fluorescencemicroscope (Carl Zeiss, New York, NY). Infected cells (fluorescentfocus units [FFU]) were identified by diffuse cytoplasmic fluorescencestaining that was excluded from the nucleus. Reovirus-infectedcells were quantified by counting random fields of view of equivalentlyconfluent monolayers for three to five fields of view for triplicatewells or by counting the entire well for triplicate wells (4).

Confocal imaging of reovirus internalization. GD25 and GD25ß1A cells were plated on coverslips in24-well plates. Cells were chilled at 4°C for 45 min priorto infection, washed with PBS, adsorbed with 8 x 10⁵ particlesper cell of T1L virions in gelatin saline, and returned to 4°Cfor 1 h. The MOI used was the minimum number of particles requiredto detect signal by confocal immunofluorescence microscopy atearly time points postinfection. Cells were either washed andfixed or nonadherent reovirus was aspirated and replaced withwarm Dulbecco's modified Eagle's medium and returned to 37°C.At 10-min intervals, cells were washed with PBS and fixed with4% formaldehyde for 20 min. Excess formaldehyde was quenchedwith an equal amount of 0.1 M glycine, followed by washing withPBS. Cells were treated with 1% Triton X-100 for 5 min and incubatedwith PBS-BGT (PBS, 0.5% BSA, 0.1% glycine, and 0.05% Tween 20)for 10 min. Cells were incubated with reovirus-specific polyclonalantiserum (1:500) in PBS-BGT for 1 h and washed with PBS-BGT.Cells were stained with donkey anti-rabbit immunoglobulin conjugatedto Alexa Fluor 488 (Molecular Probes) (1:500) to visualize reovirus,phalloidin conjugated to Alexa Fluor 546 (Molecular Probes)(1:100) to visualize actin, and TO-PRO 3 conjugated to AlexaFluor 642 (Molecular Probes) (1:1,000) to visualize DNA. Cellswere incubated for 1 h with secondary antibodies and fluorescentprobes in PBS-BGT and washed with PBS-BGT. Coverslips were removedfrom wells and placed on slides using Prolong Anti-Fade mountingmedium (Molecular Probes). Images were captured on a Zeiss LSM510 laser-scanning confocal microscope using LSM 510 software.

Virus internalization was quantified by enumerating fluorescentparticles localized at the cell periphery and particles internalizedinto the cytoplasm to determine the total number of fluorescentparticles per cell. Ten cells were analyzed for each time point.The number of internalized particles was measured as a percentageof the total number of particles per cell.

Statistical analysis. Means of triplicate samples were compared by using an unpairedStudents' t test (Microsoft Excel, Redmond, WA). P values of<0.05 were considered statistically significant.

RESULTS

Top

Results

The JAM-A cytoplasmic tail is dispensable for reovirus infection. JAM-A is a serotype-independent reovirus receptor with a cytoplasmictail known to interact with a variety of proteins (6, 28, 29).To determine whether the JAM-A cytoplasmic tail is requiredfor reovirus entry, we generated a JAM-A cytoplasmic tail deletionmutant (JAM-A- ${Delta}$ CT) and tested its capacity to support reovirusinfection following transfection of CHO cells. CHO cells donot express detectable levels of JAM-A (55, 60) and are poorlypermissive for reovirus infection (34). Cells were transientlytransfected with plasmids encoding full-length JAM-A, JAM-A- ${Delta}$ CT,or empty vector as a control. Equivalent cell surface expressionof transfected constructs was confirmed by flow cytometry (datanot shown).

The capacity of reovirus to infect CHO cells following transfectionwith the JAM-A constructs was tested using reovirus fluorescentfocus assays. Following transient transfection of CHO cellswith empty vector, JAM-A, or JAM-A- ${Delta}$ CT, cells were adsorbed withreovirus strains T1L and T3D and scored for infection by indirectimmunofluorescence at 20 h postinfection (Fig. 1). Expressionof either full-length or truncated JAM-A was sufficient to allowreovirus infection of CHO cells, permitting viral protein productionof both type 1 and type 3 reovirus strains. These results indicatethat the JAM-A cytoplasmic tail is not required for efficientreovirus attachment and infection.

View larger version (13K):
[in this window]
[in a new window]
FIG. 1. JAM-A cytoplasmic tail is not required for reovirus infection. CHO cells were transiently transfected with empty vector or plasmids encoding JAM-A or JAM-A- ${Delta}$ CT. Following incubation for 24 h to permit receptor expression, cells were adsorbed with reovirus strains T1L (A) or T3D (B) at an MOI of 0.1 FFU per cell at room temperature for 1 h. Cells were washed with PBS, incubated in complete medium at 37°C for 20 h, and stained by indirect immunofluorescence. Infected cells were quantified by counting cells exhibiting cytoplasmic staining in entire wells for triplicate experiments. The results are expressed as the mean FFU per well for triplicate samples. Error bars indicate standard deviations. CHO cells support a low level of infection by type 3 reovirus in the absence of JAM-A, likely attributable to the expression of sialic acid (16, 34).

Reovirus outer capsid proteins contain integrin-binding sequences.Structural and functional analyses indicate that reovirus andadenovirus share remarkably similar mechanisms of attachment(70, 71). To determine whether reovirus outer capsid proteinscontain sequences that could potentially engage integrins, weperformed a search for integrin-binding motifs in the ${sigma}$ 1, ${sigma}$ 3,µ1, and ${lambda}$ 2 proteins, which form the reovirus outer capsid(26). We identified two common integrin-binding motifs, RGDand KGE, in the deduced amino acid sequence of the ${lambda}$ 2 protein(Fig. 2). The RGD motif is conserved in all reovirus strainsfor which sequence information is available (15, 67); the KGEmotif is conserved in all of those strains except T2J (15, 67).The ${lambda}$ 2 protein is a component of the reovirus outer capsid andcore (26). It is structurally arranged as a pentamer at thevirion fivefold axes of symmetry and forms the base for attachmentprotein ${sigma}$ 1 (26, 63). The presence of conserved integrin-bindingmotifs in the reovirus ${lambda}$ 2 protein led us to test whether reovirusutilizes integrins to mediate internalization.

View larger version (72K):
[in this window]
[in a new window]
FIG. 2. Reovirus outer capsid protein ${lambda}$ 2 contains integrin-binding sequences. Alignment of deduced amino acid sequences of the reovirus ${lambda}$ 2 protein for the indicated strains. Amino acid residues are designated by the single-letter code. Amino acid positions are indicated above the first and last letters. Integrin-binding RGD (A) and KGE (B) motifs are highlighted by a black box. Nonconserved sequences are shown in unshaded boxes. CON, consensus sequence.

An antibody specific for ß1 integrin inhibits reovirus infection of HeLa cells. To determine whether integrins are required for reovirus infection,we first used flow cytometry to analyze integrin expressionon the surface of HeLa cells. HeLa cells were incubated withintegrin-specific MAbs and a phycoerythrin-labeled secondaryantibody (Table 1). RGD-binding integrin subunits ${alpha}$ 3, ${alpha}$ 5, ${alpha}$ v, andß1 and KGE-binding integrin subunits ${alpha}$ 1, ${alpha}$ 2, ${alpha}$ 6, andß1 (43) were detected on HeLa cells at levels abovethose in control antibody-treated cells. RGD-binding integrinheterodimer ${alpha}$ vß5 also was detected at levels abovethat of the control, while there was low-level expression of ${alpha}$ vß3. Thus, HeLa cells express both RGD- and KGE-bindingintegrins.

To assess a role for integrins in reovirus replication, we testedantibodies specific for the RGD- and KGE-binding integrins expressedon HeLa cells for the capacity to block reovirus infection.HeLa cells were incubated with integrin-specific and controlantibodies prior to adsorption with reovirus virions. Viralinfection was detected by indirect immunofluorescence (Fig.3A). We found that ß1-specific MAb DE9 resulted ina 50% reduction in infection (P < 0.05), while antibodiesspecific for the other integrin subunits expressed on HeLa cellshad no effect. Control antibodies produced anticipated effects;JAM-A-specific MAb J10.4 inhibited infection, whereas ICAM-specificMAb (data not shown) or control mouse ascites (Fig. 3A) didnot. The effect of MAb DE9 was dose dependent (Fig. 3B), providingfurther evidence that the inhibition of infection was dependenton integrin blockade.

View larger version (21K):
[in this window]
[in a new window]
FIG. 3. ß1 integrin antibody reduces reovirus infection of HeLa cells. HeLa cells were treated with gel saline (GS), control ascites (C), JAM-A-specific MAb J10.4, or antibodies specific for the ${alpha}$ and ß integrins shown (20 µg per ml or as diluted ascites) (A), gel saline, control ascites (Control), ${alpha}$ 2-specific MAb AA10, or ß1-specific MAb DE9 (at the indicated dilutions) (B), antibodies specific for the ${alpha}$ and ß integrins shown in the presence of ß1-specific MAb DE9 (1:10) (C), control ascites, ß1-specific MAb DE9, JAM-A-specific MAb J10.4, or JAM-A-specific MAb J10.4 in combination with ß1-specific MAb DE9 (D), and incubated at room temperature for 1 h. Antibody-treated cells were infected with virions or ISVPs of T1L at an MOI of 0.1 FFU per cell at 4°C for 30 min. Cells were washed with PBS, incubated in complete medium at 37°C for 16 h, and stained by indirect immunofluorescence. Infected cells were quantified by counting cells exhibiting cytoplasmic staining in three fields of view for triplicate samples. The results are expressed as the mean FFU per field for triplicate experiments. Error bars indicate standard deviations. *, P < 0.05 in comparison to the control; **, P < 0.05 in comparison to HeLa cells treated with JAM-A-specific MAb J10.4 alone.

To determine whether particular ${alpha}$ subunits pair with ß1integrin to facilitate reovirus infection, we tested whethertreatment with ${alpha}$ integrin-specific antibodies was capable ofenhancing the inhibitory effect of ß1 integrin-specificMAb DE9 on reovirus infection. We also tested whether antibodiesspecific for other ß integrin subunits expressed onHeLa cells, ß3 and ß5, were capable of infectionblockade. HeLa cells were treated with MAb DE9 in combinationwith other integrin-specific antibodies prior to adsorptionwith reovirus virions (Fig. 3C). While treatment of HeLa cellswith MAb DE9 resulted in a 50% reduction in reovirus infection,none of the other integrin-specific antibodies tested reducedreovirus infection to a greater extent than that resulting fromtreatment with DE9 alone. These results suggest that the integrinepitope bound by reovirus is blocked by ß1-specificMAb DE9 and not by the other MAbs used in these experiments.

JAM-A MAb J10.4 blocks reovirus infection $~$ 90% (Fig. 3A). Todetermine whether the residual level of infection in the presenceof MAb J10.4 was dependent on reovirus interactions with ß1integrin, we treated HeLa cells with JAM-A-specific MAb J10.4in combination with MAb DE9 prior to adsorption with reovirusvirions (Fig. 3D). Treatment of HeLa cells with MAb J10.4 andMAb DE9 completely abrogated reovirus infection, indicatingthat the effect of JAM-A blockade is enhanced when ß1integrin is not available for interactions with reovirus. Treatmentwith MAb DE9 did not significantly inhibit infection by ISVPs(Fig. 3D), suggesting that viral attachment is not affectedby ß1 integrin blockade. Taken together, these resultssupport the conclusion that a ß1-specific antibodyblocks reovirus infection at a step subsequent to attachmentbut prior to uncoating, implicating ß1 integrin inreovirus internalization.

Transient transfection of integrin cDNAs allows reovirus infection of JAM-A-expressing CEFs. Ectopic expression of JAM-A in CEFs rescues infection by reovirusISVPs but not by virions (5), suggesting that these cells exhibita cell-specific block at the entry or uncoating phases of reovirusinfection. To test the capacity of integrins to confer infectionof CEFs by reovirus virions, CEFs were transiently transfectedwith a JAM-A-encoding plasmid in the presence or absence ofmurine ${alpha}$ v, ${alpha}$ 2, or ß1 integrin-encoding plasmids singlyor in ${alpha}$ ß pairs. Transfected cells were infected withreovirus virions or ISVPs, and infection was assessed by indirectimmunofluorescence (Fig. 4). Expression of ß1 integrinpaired with either of the murine ${alpha}$ integrin subunits providedan approximately fourfold enhancement of infection by reovirusvirions in comparison to that in cells transfected with JAM-Aalone. These data suggest that ß1 integrin expressioncomplements a reovirus cell entry defect in CEFs and providefurther support for the involvement of ß1 integrinin reovirus internalization.

View larger version (15K):
[in this window]
[in a new window]
FIG. 4. ß1 integrin expression enhances reovirus infection of CEFs. CEFs were transiently transfected with JAM-A-encoding plasmid alone (vector) or in combination with plasmids encoding the integrin subunits shown. Following 24 h to allow receptor expression, transfected cells were adsorbed with T1L virions or ISVPs at an MOI of 1 FFU per cell at room temperature for 1 h. Cells were washed with PBS, incubated in complete medium at 37°C for 20 h, and stained by indirect immunofluorescence. Infected cells were quantified by counting cells exhibiting cytoplasmic staining in entire wells for duplicate samples. The results are expressed as the mean FFU per well for duplicate experiments. Error bars indicate the range of data. Shown is a representative experiment of three independent experiments performed.

Cells deficient in ß1 integrin have a decreased capacity to support reovirus infection. To further assess a role for ß1 integrin in reovirusinfection, we tested the capacity of reovirus to infect cellsdeficient in the ß1-integrin subunit. GD25 cells aremurine embryonic stem cells derived from ß1-null embryos(78). GD25ß1A cells are GD25 cells that have beenengineered to stably express ß1 integrin and thusserve as an isogenic control for GD25 cells. Flow cytometricanalysis confirmed that while both cells express JAM-A, onlyGD25ß1A cells express ß1 integrin (Fig.5A). GD25 cells (ß1^–/–) and GD25ß1Acells (ß1^+/+) (78) were adsorbed with reovirus virionsor ISVPs, and infection was scored by indirect immunofluorescence(Fig. 5B). In comparison to ß1^+/+ cells, ß1^–/–cells were substantially less susceptible to infection by virions,while infection by ISVPs was equivalent in both cell types.Importantly, preincubation of ß1^+/+ cells with murineß1 integrin-specific MAb CD29 reduced infection inß1^+/+ cells (Fig. 5C), indicating that enhancementof infection is due to expression of ß1 integrin.Therefore, ß1 integrin is required for efficient reovirusinfection.

View larger version (20K):
[in this window]
[in a new window]
FIG. 5. Cells deficient in ß1 integrin are less permissive for reovirus infection. (A) GD25 (ß1^–/–) and GD25ß1A (ß1^+/+) cells were detached from plates with 20 mM EDTA, washed, and incubated with antibodies specific for either murine ß1 integrin or murine JAM-A. Cell surface expression of these molecules was detected by flow cytometry. Data are expressed as fluorescence intensity. GD25 and GD25ß1A cells were untreated (B) or pretreated with ß1-specific MAb CD29 (ß1 Ab) or a hamster isotype-matched control MAb (IgG) at room temperature for 1 h (C), adsorbed with virions or ISVPs of T1L at an MOI of 0.1 FFU per cell, and incubated at 4°C for 30 min. Cells were washed with PBS, incubated in complete medium at 37°C for 20 h, and stained by indirect immunofluorescence. Infected cells were quantified by counting cells exhibiting cytoplasmic staining in five fields of view for duplicate samples. The results are expressed as the mean FFU per field for triplicate experiments. *, P < 0.05 in comparison to the control.

Reovirus binding to ß1^–/– and ß1^+/+ cells is equivalent. Equivalent infection of ß1^–/– and ß1^+/+cells by ISVPs (Fig. 5B) suggests that reovirus is capable ofefficiently binding to both cell types. To directly test thishypothesis, ß1^–/– and ß1^+/+cells were mock treated or incubated with FITC-labeled virionsand binding was assessed by flow cytometry (Fig. 6). In theseexperiments, we found that reovirus binds equivalently to ß1^–/–and ß1^+/+ cells. These data demonstrate a functionfor ß1 integrin in reovirus infection at a step subsequentto viral attachment.

View larger version (17K):
[in this window]
[in a new window]
FIG. 6. Reovirus exhibits equivalent binding to ß1^–/– and ß1^+/+ cells. GD25 (ß1^–/–) and GD25ß1A (ß1^+/+) cells were incubated with either PBS (mock) or 2 x 10¹¹ FITC-labeled T1L virions at 4°C for 1 h and analyzed by flow cytometry to assess reovirus binding to the cell surface. The results are expressed as fluorescence intensity.

ß1 integrin enhances the efficiency of reovirus internalization. To directly assess the role of ß1 integrin in reovirusinternalization, ß1^–/– and ß1^+/+cells were infected at 4°C and then warmed to 37°C overa time course concurrent with reovirus entry (2, 72). At 10-minintervals, cells were fixed, stained for indirect immunofluorescence,and examined by confocal microscopy. Representative confocalmicrographic images of reovirus-infected ß1^–/–and ß1^+/+ cells are shown in Fig. 7. Immediately afterviral adsorption, both cell types exhibited reovirus stainingat the cell periphery. At 10 min postadsorption, some reovirusstaining was observed at the cell periphery, yet intracellularstaining in ß1^+/+ cells was also observed. At 20 and30 min postadsorption, the majority of virions had entered theß1^+/+ cells and had a perinuclear location. In sharpcontrast to the findings made using ß1^+/+ cells, viralentry was markedly delayed in ß1^–/– cells,with the majority of reovirus virions remaining at the cellperiphery throughout the time course. At later time points (30min postadsorption), some virions were present within the cytoplasm,but these were the minority. These findings suggest that expressionof ß1 integrin enhances reovirus entry.

View larger version (59K):
[in this window]
[in a new window]
FIG.7. ß1 integrin enhances reovirus entry into cells. (A) GD25ß1A (ß1^+/+) and (B) GD25 (ß1^–/–) cells were chilled, adsorbed with T1L virions, and incubated at 4°C for 1 h. Nonadherent virus was removed, warm medium was added, and cells were incubated at 37°C for the times shown. Cells were fixed, stained for reovirus (green), actin (red), and DNA (blue), and imaged using confocal immunofluorescence microscopy. Representative digital fluorescence images of the same field are shown in each row.

To quantify reovirus internalization into ß1^–/–and ß1^+/+ cells, we determined the number of internalizedfluorescent particles as a percentage of the total number offluorescent particles per cell at various times postadsorption(Fig. 8). At 0 and 10 min postadsorption, the percentage ofparticles internalized into ß1^–/– andß1^+/+ cells was equivalent, $~$ 10 and $~$ 30%, respectively.However, at 20 and 30 min postadsorption, the percentage ofreovirus particles internalized into ß1^+/+ cells was $~$ 50%, while the percentage of particles internalized into ß1^–/–cells was only $~$ 30% (P < 0.05) (Fig. 8). These data indicatethat ß1 integrin enhances reovirus entry at earlytimes postadsorption, suggesting a direct role for ß1integrin as a reovirus internalization receptor.

View larger version (25K):
[in this window]
[in a new window]
FIG. 8. Quantification of reovirus internalization into ß1^–/– and ß1^+/+ cells. Viral internalization was quantitated by enumerating fluorescent particles localized at the cell periphery and particles internalized into the cytoplasm to determine the total number of fluorescent particles per cell. The results are expressed as mean percent internalization (internalized fluorescent particles/total number of fluorescent particles per cell) for 10 cells for each time point. *, P < 0.05 in comparison to ß1^+/+ cells.

DISCUSSION

Top

Discussion

In this study, we performed experiments to define the moleculardeterminants of reovirus internalization. We show that antibodiesspecific for ß1 integrin inhibit reovirus infectionat a postattachment step. We provide evidence that expressionof ß1 integrin promotes infection by reovirus virionsin cells with a block to viral internalization and that viralentry is substantially diminished in cells deficient in ß1integrin expression. Together, these data provide strong evidencethat ß1 integrin serves as a coreceptor to mediatereovirus internalization. These findings suggest a new modelfor attachment and cell entry of reovirus (Fig. 9). In thismodel, we propose that reovirus initially interacts with cellsvia low-affinity binding to carbohydrate. These interactionsare followed by high-affinity engagement of JAM-A, which positionsthe virus on the cell surface for subsequent interactions withß1 integrin to trigger viral endocytosis.

View larger version (19K):
[in this window]
[in a new window]
FIG. 9. Receptors for reovirus attachment and cell entry. Reovirus initially engages cells by low-affinity interactions with carbohydrate. For type 3 reovirus strains, this carbohydrate is sialic acid. Reovirus-carbohydrate interactions are followed by high-affinity binding to JAM-A, which positions the virus on the cell surface for subsequent interactions with ß1 integrin to trigger viral endocytosis.

Integrins have been identified as attachment and entry receptorsfor several viruses, including echovirus ( ${alpha}$ 2ß1) (8),foot-and-mouth disease virus ( ${alpha}$ vß1, ${alpha}$ vß3,and ${alpha}$ vß6) (9, 44, 45), hantaviruses NY-1 and Sin Nombrevirus (ß3 integrins) (37), Kaposi sarcoma herpesvirus( ${alpha}$ 3ß1) (1), and cytomegalovirus ( ${alpha}$ 2ß1, ${alpha}$ 6ß1,and ${alpha}$ vß3) (32). The Reoviridae family member rotavirusalso engages a variety of integrins for attachment and cellentry. Rotavirus strains RRV, SA11, and Wa bind to the I (inserted)domain of ${alpha}$ 2ß1 integrin via an Asp-Gly-Glu integrin-bindingmotif in the VP4 spike protein to effect viral attachment (39).The interactions of rotavirus outer capsid protein VP7 withintegrins ${alpha}$ xß2 and ${alpha}$ vß3 can mediate viralentry (39, 82). Integrin ${alpha}$ 4ß1 also can serve as a receptorfor rotavirus strain SA11, which contains ${alpha}$ 4ß1 integrin-bindingsequences Leu-Asp-Val in VP7 and Ile-Asp-Ala in VP4 (41). Interestingly,like reovirus, adenovirus engages a specific cell surface protein,CAR, prior to interactions with integrins, which function subsequentto viral attachment to mediate viral endocytosis (50, 80). Therefore,the identification of ß1 integrin as a reovirus internalizationreceptor suggests that the conservation of attachment strategiesused by reovirus and adenovirus (70, 71) extends to mechanismsof internalization.

Although the specific reovirus protein required for integrinbinding is not apparent from our studies, the ${lambda}$ 2 protein is apromising candidate. The ${lambda}$ 2 protein forms a pentameric turretat the virion fivefold symmetry axes and serves as the insertionsite for trimers of attachment protein ${sigma}$ 1 (26). Thus, ${lambda}$ 2 is thereovirus analogue of the adenovirus penton base protein, whichmediates the engagement of integrins by adenovirus (22, 80).Interestingly, ${lambda}$ 2 also contains conserved RGD and KGE motifs(15), the preferred interaction motifs for several ß1integrin heterodimers (43).

Structural information for ${lambda}$ 2 is available in the context ofthe reovirus core but not for the intact virion. In the core,the KGE motif is exposed on the top of the ${lambda}$ 2 turret, where itwould be accessible to a receptor. The RGD motif is also surfaceexposed, but it appears to be less accessible. However, the ${lambda}$ 2 structure in the core may not be identical to that in thevirion, as the protein undergoes major conformational changesduring virion-to-core disassembly (26). Therefore, it is possiblethat both the RGD and KGE motifs are accessible to interactionswith ß1 integrin during engagement of the cell surfaceby the virus.

A human ß1 integrin-specific antibody (DE9) reducedreovirus infection of HeLa cells by 50% (Fig. 3). Similarly,a murine ß1 integrin-specific antibody (CD29) blockedinfection of ß1-expressing mouse embryonic stem cellsby $~$ 50% (Fig. 4). Interestingly, MAb DE9 also blocks infectionof echovirus (8) and cytomegalovirus (32), suggesting that anepitope in ß1 integrin recognized by MAb DE9 may bea preferred binding site for multiple viruses. It is possiblethat the residual level of reovirus infection following ß1integrin antibody treatment is attributable to other internalizationreceptors on the cell surface that may be integrin or nonintegrinmolecules. However, it is noteworthy that treatment of HeLacells with both MAb DE9 and JAM-A-specific MAb J10.4 completelyabolishes reovirus growth (Fig. 3D). This finding suggests thatthe residual infection in J10.4-treated HeLa cells is due toreovirus interactions with ß1 integrin. Thus, it appearsthat blockade of reovirus infection by integrin-specific antibodiesis inefficient because complete inhibition of virus-integrininteractions is not possible if the virus is tightly adheredto the cell surface by JAM-A.

Since antibodies specific for ß3 and ß5integrins did not inhibit reovirus infection, it is likely thatonly ß1 integrin can serve a reovirus internalizationfunction. Antibodies specific for the ${alpha}$ integrin subunits expressedon HeLa cells did not further reduce reovirus infection followingtreatment with a ß1 integrin-specific antibody (Fig.3C). We envision three possible explanations for this result.First, reovirus may directly engage a ligand-binding domainof ß1 integrin. Second, reovirus may utilize ß1integrin when paired with numerous ${alpha}$ subunits that have redundantfunctions. However, treatment of HeLa cells with a ß1integrin-specific antibody and a mixture of antibodies specificfor ${alpha}$ 1, ${alpha}$ 2, ${alpha}$ 3, ${alpha}$ 5, ${alpha}$ 6, and ${alpha}$ v integrins did not diminish reovirusinfection in comparison to cells treated with a ß1integrin-specific antibody alone (data not shown). Third, reovirusmay engage an epitope of an integrin ${alpha}$ subunit that is not recognizedby the antibodies used in our experiments. Further studies arerequired to define the biophysical basis of reovirus-integrininteractions.

JAM-A is required for high-affinity reovirus attachment to numerouscell types (5, 16, 34, 56). However, the JAM-A cytoplasmic tailis not necessary for viral endocytosis (Fig. 1). JAM-A likelytethers the virus to the cell surface to facilitate secondaryinteractions with ß1 integrin (Fig. 9). This modelis analogous to the mechanism of lymphocyte homing, in whichadhesion molecules such as JAM-A provide initial cellular contactsto facilitate subsequent interactions with integrins for diapedesisor signaling (65). An interesting possibility is that JAM-Amay be associated with ß1 integrin on the host cellplasma membrane. If such were the case, initial JAM-A engagementmight facilitate integrin binding, clustering, and viral endocytosis.In support of this hypothesis, JAM-A has been shown to regulateß1 integrin expression and localization (54).

The cytoplasmic domains of integrin subunits are involved ina number of signaling pathways (42). The ß1 integrincytoplasmic domain is linked to cytoskeletal proteins, includingtalin (62) and ${alpha}$ -actinin (59), and signaling molecules, includingpaxillin and focal adhesion kinase (66). In addition, the ß1integrin cytoplasmic domain contains two Asn-Pro-any residue-Tyr(NPXY) motifs (64), which are common sequence motifs in thecytoplasmic domains of many receptors and serve as recognitionsites for the cellular endocytic machinery (21, 24). NPXY motifsinteract with the µ2 subunit of the adaptor protein 2complex (12, 61), which can recruit clathrin and trigger clathrin-mediatedendocytosis (47).

Since clathrin-dependent mechanisms have been implicated inreovirus cell entry (31), it seems plausible that reovirus engagementof ß1 integrin leads to clathrin-mediated endocytosisthrough signaling regulated by the ß1 integrin cytoplasmicdomain. It is noteworthy that Kaposi's sarcoma-related herpesvirusbinding to ${alpha}$ 3ß1 integrin (1) and cytomegalovirus bindingto ß1 integrin (32) activate focal adhesion kinase.In addition, adenovirus engagement of ${alpha}$ v integrins induces activationof phosphoinositide-3-OH kinase, which is required for adenovirusendocytosis (51).

Identification of ß1 integrin as a receptor that triggersreovirus entry raises the possibility that coreceptor bindinginfluences reovirus tropism and disease. Reovirus serotypesdiffer in mechanisms of spread, tropism for cells in the centralnervous system, and disease outcome in the infected host (73).Previous studies using reassortant genetics and comparativesequence analysis demonstrated that these phenotypes segregatemost strongly with viral attachment protein ${sigma}$ 1, suggesting thatreovirus serotypes bind to different receptors (74, 76, 77).However, the ${sigma}$ 1-encoding S1 gene is not the sole determinantof reovirus growth at some sites within the host. For example,the ${lambda}$ 2-encoding L2 gene influences viral growth in the intestine(11) and spread to new hosts (46). Moreover, JAM-A functionsas a receptor for all three reovirus serotypes (16); therefore,JAM-A cannot explain serotype-dependent differences in reoviruspathogenesis. The presence or absence of particular integrinsat distinct physiologic sites may critically influence the courseof reovirus infection. In support of a role for coreceptor utilizationin reovirus growth, reovirus infection can occur in the absenceof ${sigma}$ 1 (19) or JAM-A (5), albeit at greatly reduced efficiency.These findings highlight the complex nature of reovirus attachmentand entry and suggest that reovirus tropism and pathogenesisare not dictated by primary receptor interactions alone. Itis possible that tropism and pathogenesis are determined bythe concerted action of attachment and internalization receptors,perhaps not all of which have been discovered.

ACKNOWLEDGMENTS

We thank Roy Zent for helpful discussions, members of our laboratoryfor review of the manuscript, Kathy Allen and the NashvilleVeterans Affairs Hospital Flow Cytometry Facility for assistancewith flow cytometry, the Vanderbilt Imaging Core for help withconfocal microscopy, and Aaron Derdowski for assistance withthe confocal image analysis. We thank Xuemin Chen and Paul Spearmanfor providing CEFs, Richard Hynes for providing murine integrincDNA constructs, Deane Mosher for providing GD25 and GD25ß1Acells, Charles Parkos for providing hJAM-A-specific MAb J10.4and control ascites, Beat Imof for providing mJAM-A-specificMAb H202-106-7-4, Richard Bankert for providing ${alpha}$ 2-specific MAb6F1, and John Williams for providing antibodies.

This research was supported by Public Health Service awardsAI007281 (M.S.M.), CA09385 (J.C.F.), AI07474 (S.A.K.-B.), andAI32539, the Vanderbilt University Research Council (J.C.F.),and the Elizabeth B. Lamb Center for Pediatric Research. Additionalsupport was provided by Public Health Service awards CA68485for the Vanderbilt Cancer Center and DK20593 for the VanderbiltDiabetes Research and Training Center.

FOOTNOTES

* Corresponding author. Mailing address: Lamb Center for Pediatric Research, D7235 MCN, Vanderbilt University School of Medicine, Nashville, TN 37232. Phone: (615) 343-9943. Fax: (615) 343-9723. E-mail: terry.dermody{at}vanderbilt.edu.

These authors contributed equally to the manuscript.

Present address: Department of Microbiology and Immunology,Yerkes National Primate Research Center, Emory University, Atlanta,GA 30329.

Present address: Department of Microbiology, Mount Sinai Schoolof Medicine, New York, NY 10029.

REFERENCES

Top