αvβ6- and αvβ8-Integrins Serve As Interchangeable Receptors for HSV gH/gL to Promote Endocytosis and Activation of Membrane Fusion

Herpes simplex virus (HSV) - and herpesviruses in general - encode for a multipartite entry/fusion apparatus. In HSV it consists of the HSV-specific glycoprotein D (gD), and three additional glycoproteins, gH/gL and gB, conserved across the Herpesviridae family and responsible for the execution of fusion. According to the current model, upon receptor binding, gD propagates the activation to gH/gL and to gB in a cascade fashion. Questions remain about how the cascade of activation is controlled and how it is synchronized with virion endocytosis, to avoid premature activation and exhaustion of the glycoproteins. We considered the possibility that such control might be carried out by as yet unknown receptors. Indeed, receptors for HSV gB, but not for gH/gL, have been described. In other members of the Herpesviridae family, such as Epstein-Barr virus, integrin receptors bind gH/gL and trigger conformational changes in the glycoproteins. We report that αvβ6- and αvβ8-integrins serve as receptors for HSV entry into experimental models of keratinocytes and other epithelial and neuronal cells. Evidence rests on loss of function experiments, in which integrins were blocked by antibodies or silenced, and gain of function experiments in which αvβ6-integrin was expressed in integrin-negative cells. αvβ6- and αvβ8-integrins acted independently and are thus interchangeable. Both bind gH/gL with high affinity. The interaction profoundly affects the route of HSV entry and directs the virus to acidic endosomes. In the case of αvβ8, but not αvβ6-integrin, the portal of entry is located at lipid microdomains and requires dynamin 2. Thus, a major role of αvβ6- or αvβ8-integrin in HSV infection appears to be to function as gH/gL receptors and to promote virus endocytosis. We propose that placing the gH/gL activation under the integrin trigger point enables HSV to synchronize virion endocytosis with the cascade of glycoprotein activation that culminates in execution of fusion.


Introduction
The glycoproteins of enveloped virions fulfill three major functions to enable virus entry into target cells; the attachment of virions to cells, a step that partly determines the type of cells that the virus targets, hence the viral tropism; the triggering of fusion, i.e. the activation of the fusion machinery, and the execution of fusion. For a number of viruses, a fourth event occurs between these steps, virion internalization by endocytosis, or macropinocytosis. The domains responsible for all these activities are often localized in one or two glycoproteins; this is the case for example for ortho-, paramyxo-and retroviruses. Virion glycoproteins can be considered ready-to-use machines that need to undergo a transition in conformation from the metastable fusion-inactive to the fusionactive form, in order to induce the merging of the two membranesthat of the virion and that of cell -so that lipids are mixed and fusion is executed [1]. A fundamental aspect of the process is that the steps are sequentially ordered and coordinated, to ensure that the glycoprotein transition takes place only after the virus has attached to the cells. Indeed, a premature activation would irreversibly exhaust the fusogenic potential of the virion glycoproteins, and lead to failure to infect. A key question is therefore how the timing of glycoprotein transition and activation is controlled. Essentially, there are two strategies. Either the glycoprotein transition is dependent on the glycoprotein encounter with the cognate cellular receptor, or on the low pH of the endosomal compartment. These levels of control guarantee that the virion fusion machinery is only active after the virus has attached to cells, or, for those viruses which undergo internalization, after they have been endocytosed and the endosomal pH has been lowered. According to this view, two major functions of cellular receptors are determination of viral tropism and triggering of fusion.
Herpes simplex virus (HSV), and herpesviruses in general, exhibit a high level of complexity since they encode a multipartite entry/ fusion machinery [2,3]. Some of the herpesvirus glycoproteins are species-specific; they play a role in the initial steps of virus entry and interact with receptors that belong to variety of molecular families. A prototypic example is HSV glycoprotein D (gD) that binds alternatively two major receptors, nectin1 and herpesvirus entry mediator (HVEM) [4][5][6]. It functions as a determinant of HSV tropism and as the trigger of fusion capable to activate the downstream glycoproteins gH/gL and gB [7][8][9][10]. Conformational modification of gD following receptor binding was inferred by structural, biochemical and molecular biology approaches [7,8,[11][12][13]. gD encodes a specific domain -the profusion domain -that interacts with, or at least propagates the activation to gH/gL, and thereafter or simultaneously activates gB in a cascade fashion [9,10,14,15]. Species-specific glycoproteins among human herpesiviruses include Epstein Barr virus (EBV) gp42, which binds major histocompatibility complex II to enable virus entry into lymphocytic cells [16,17]; the human cytomegalovirus (HCMV) glycoproteins of the 128-131 locus [18], which confer endothelial and epithelial tropism; the human herpesvirus 6 (HHV-6) gQ1-gQ2 that enable gH/gL interaction with CD134 [19].
In addition to the species-specific glycoproteins, three glycoproteins -the heterodimer gH/gL and gB -constitute the conserved core fusion apparatus across the Herpesviridae family. gB has structural features typical of fusion glycoproteins. In contrast, gH/gL does not resemble any known structure. A number of approaches support the view that gH/gL acts as an intermediate in the activation cascade of HSV entry/fusion glycoproteins. Remarkably, for some herpesviruses, gH/gL or gB serve as major receptor-binding glycoproteins and therefore their activation may be under the direct control of encounter with the receptor. Thus, members of the integrin family serve as receptors for EBV and HCMV gH/gL [20][21][22], and for HHV-8 gB [23]. EphrinA2 receptor serves as a receptor for HHV-8 gH/ gL [24,25]. Receptors for HSV gB and varicella zoster virus (VZV) gB include paired immunoglobulin-like type 2 receptor alpha (PILRa), myelin associated glycoprotein, and non-muscle myosin IIA [26][27][28]; their role in virus entry remains to be fully elucidated.
We have recently discovered two additional functions of cellular receptors for HSV, namely that they serve as routing factors to define the pathway of entry and that they serve as viral sensors capable of initiating the innate response of the cell, and thus coupling virus entry to the innate response. Specifically, HSV enters different cells by different pathways. avb3-integrin is a determinant in the choice of the HSV pathway of entry which routes the HSV receptor nectin1, and consequently HSV, to lipid microdomains [29,30]. In this way, avb3-integrin enables entry of HSV through a pathway dependent on lipid microdomains, dynamin 2 and acidic endosomes. avb3-integrin interacts at low affinity with HSV gH/gL [31,32]. Furthermore, avb3-integrin also binds Toll-like receptor 2 (TLR2), which, in turn, binds gH/ gL [33]. In this way avb3-integrin senses HSV and cooperates with and reinforces the TLR2-dependent response [31]. A signaling cascade is initiated that leads to activation of the transcription factor NF-kB, and to production of intereferon (IFN) a and b, the major innate defenses of the cell against HSV [31].
Here, we investigated whether integrins other than avb3 play a role in HSV entry. We investigated the av group of integrins, which are preferentially expressed in epithelial cells, and bind their ligands through the RGD domain. avb6-integrin is upregulated in epithelial malignancies. We report that avb6and avb8-integrins serve as interchangeable HSV receptors for entry into keratinocytes, other epithelial and neuronal cells. Each integrin acted independently of the other. Both bind gH/gL at high affinity. This interaction profoundly affects the pathway of HSV entry, which takes the way of acidic endosomes. In the case of avb8-integrin, the portal of entry is located at lipid microdomains and requires dynamin 2. In the case of avb6-integrin, the portal of entry is located outside the lipid microdomains and does not need dynamin 2. We propose that the major role of avb6and avb8integrins as gH/gL receptors is to promote HSV endocytosis, and thereby to synchronize virion internalization with the cascade of glycoprotein activation.

Results
A soluble form of HSV gH/gL binds avb6and avb8integrins at high affinity The interaction between HSV gH/gL and integrins was explored using surface plasmon resonance spectroscopy of soluble integrins and soluble gH/gL. A kinetic analysis of integrin binding to increasing concentrations of soluble gH/gL indicated that gH/ gL interacted with both avb6-integrin and with avb8-integrin ( Fig. 1). To obtain rate constants of gH/gL binding and to calculate a dissociation constant (K D ), k obs was plotted against the concentration of gH/gL. The dependence of k obs on gH/gL concentration was described by a linear function suggesting that the interaction has a single step. The intersection point with the Yaxes corresponds to a dissociation rate constant k off , and the slope to an associate rate constant k on . The rate constants and the resulting K D values ( Table 1) are indicative of a high affinity interaction. In contrast, no interaction between integrin avb5 and gH/gL was detected.
Expression of avb6and avb8-integrins in epithelial cell lines and silencing by siRNAs avb6-integrin is expressed in epithelial cells and upregulated in epithelial cancer cells. avb8-integrin is expressed in some types of epithelial cells, as well as in glial and dendritic cells [34,35]. We determined the expression of avb6and of avb8-integrins in 293T cells (epithelial cells transformed by an adenovirus fragment), HeLa, colon carcinoma SW480, keratinocyte HaCaT and in the neuronal SK-N-SH cell lines. By qRT-PCR ( Fig. 2A) b8-integrin was expressed at higher levels than b6-integrin in 293T, HeLa and

Author Summary
In order to infect their hosts and cause disease, viruses must enter their host cells. The human pathogen herpes simplex virus (HSV) -and herpesviruses in general -are equipped with a complex, multipartite entry apparatus, made of four glycoproteins -gD, gH/gL, gB. These glycoproteins must be activated in a timely, coordinated manner. According to the current model, the flux of activation goes from receptor-bound gD, to gH/gL and gB. The premature activation, and hence exhaustion of the glycoproteins must also be prevented. We report on a checkpoint at the gH/gL level. Specifically, avb6and avb8integrins serve as receptors for HSV entry into keratinocytes and other epithelial and neuronal cells. Both bind gH/gL with high affinity. The interaction profoundly affects the pathway of HSV entry, promoting HSV endocytosis into acidic endosomes. For avb8-integrin, the portal of entry is at lipid microdomains and requires dynamin 2. We propose that, by placing the activation of gH/gL under control of an integrin trigger point, HSV can synchronize virion endocytosis with the cascade of activation that culminates in the execution of fusion between the virion envelope and cellular membranes.
avb6and avb8-Integrins As HSV gH/gL Receptors SK-N-SH cells, and to comparably high levels in SW480 and HaCaT cells. The epithelial cell lines were analyzed also by flow cytometry; they exhibited the highest fluorescence intensity with the antibody to avb6-integrin (Fig. 2 B). A direct comparison between avb6and avb8-integrin extent of expression in a same cell line can not be performed by flow cytometry, since reactivity in this assay is strongly influenced by the properties of the antibody being used.
The extent of silencing achieved with specific si-RNAs to b6or to b8-integrin as measured by q-RT-PCR is shown in Fig. 2 C. The extent of silencing was greater than 80% for both integrins in 293T cells. It was around 75 and 50% for b6and b8-integrin in SW480 cells, and about 40% and 90% for b6and b8-integrin in HeLa cells. Silencing was specific as there was no off-target effect in any of the cell line. Thus, when b6-integrin was silenced, b8-integrin mRNA level was not decreased. Similarly, when b8integrin was silenced, b6-integrin mRNA level was not decreased. Silencing was confirmed by flow cytometry analysis in 293T cells and expressed as median fluorescence intensity for b6-integrin MAbs to avb6or avb8-integrin or silencing of b6or b8integrin inhibit HSV-1 infection To define if avb6and avb8-integrin play a role in HSV infection, we tested the effect of function-blocking monoclonal antibodies (MAbs) to avb6-integrin (MAb 2077Z), or avb8integrin (MAb 37E1), on HSV-1 infection. The neutralizing R1.302 MAb to nectin1 was used as a positive control. Cells were preincubated with increasing amounts of MAbs, and infected in the presence of MAbs with the recombinant R8102 which carries a Lac-Z reporter gene. A large body of evidence indicates that the extent of b-galactosidase (b-gal) expression directly reflects the extent of infection [4,5]. Fig. 3A-E shows that MAbs to avb6 or to avb8-integrin inhibited R8102 infection in a dose-dependent manner in all cells. The only exception was MAb to avb8-integrin which failed to inhibit infection in 293T cells, even though these cells exhibited cell surface expression of this integrin. Thus, both avb6 and avb8-integrins play a critical role in HSV infection of epithelial, keratynocytic and neuronal cells.
As a second approach, silencing of b6or of b8-integrin with 50 nM si-RNA inhibited infection in the three lines -HeLa and SW480 (Fig. 4A) and 293T ( Fig. 4 B-E). Thus, two assays concordantly indicate a requirement for avb6and avb8-integrins in HSV infection.
Because all cell lines tested contained both avb6-integrin and avb8-integrin, we asked whether each one of the two integrins is required independently one of the other, or whether they play an interchangeable role. In the latter case, the expectation is that in cells silenced for a single integrin, the extent of inhibition of infection is smaller than the extent of silencing. By contrast, in cells simultaneously silenced for both integrins, the inhibition of infection is expected to be higher than that achieved in cells silenced for a single integrin. Fig. 4B-D shows the extent of silencing as percentage of positive cells; their mean fluorescence intensities is shown in Fig. 2 E, H. Indeed, in 293T cells silenced for either or the other integrin, infection was reduced to a lower extent than expression. When the two integrins were simultaneously silenced, infection dropped to about 10%. In a replicate experiment, R8102 infection in singly or doubly silenced cells was quantified by means of b-gal. Fig. 4 E shows that infection was inhibited by about 75% in double-silenced cells, and by about 50% in singly silenced cells. The results indicate that avb6-integrin and avb8-integrin play each a critical role in HSV infection, each independently of the other. Thus, they appear to act in an interchangeable fashion.

Gain of function: Expression of avb6-integrin in integrinnegative K562 cells increases HSV infection
To confirm the role of avb6-integrin in HSV infection, we performed a gain of function experiment. The myelocytic cell line K562 expresses a very limited number of integrins, predominantly a5b1, and expresses no detectable av [36]. A stable cell line in which about 20% of cells transgenically express avb6-integrin (K562 avb6 ) was obtained from Dr S. Blystone. The wt-K562 and K562 avb6 cells were infected with RLM5 HSV mutant, which carries the GFP moiety and enables flow cytometry quantification of infection. To authenticate the results, infection was carried in the presence or absence of MAb 2077Z to avb6-integrin, MAb L230 to av-subunit, or polyclonal antibody (PAb) R140 to HVEM as a control. Myelocytic cells, including K562 cells, are hardly  avb6and avb8-Integrins As HSV gH/gL Receptors infected by HSV and likely restrict infection at entry and postentry steps [37]. Fig. 5 shows a representative experiment, and reports figures relative to the average percentage infected cells, as determined in four independent experiments. It can be seen that the number of infected cells doubled in avb6-integrin + K562 cells, relative to wt-K562 cells (10.17 versus 5.39%). Extent of infection in avb6-integrin + K562 cells was strongly inhibited when infection took place in the presence of MAb 2077Z to avb6-integrin (3.12%), or MAb L230 to av-subunit (5.2%), or in the presence of anti-HVEM PAb R140 (1.98%). Although the percentage of infected cells is rather low, in agreement with well-known resistance to HSV infection exhibited by this type of cells [37], the results clearly indicate that expression of avb6-integrin enhances HSV infection of K562 cells; furthermore, the single avb6-integrin -in the absence of avb8-integrin -was sufficient to induce the increase in infection. The latter finding reinforces the conclusion drawn from the previous series of experiments that each of the two integrins is sufficient to promote HSV infection.
Silencing of b6or b8-integrin reduces HSV-1 attachment to cells The next series of experiments was designed to shed light on the role played by avb6or avb8-integrin in HSV infection. First, we defined the step(s) in the HSV entry process in which they participate. Cells were exposed to anti-integrin antibodies prior, during and post virus absorption to cells, prior and during virus absorption, or only after virus absorption to cells. The extent of inhibition of infection was then quantified. These experiments could not be performed in 293T cells, as in these cells the anti-avb8-integrin MAb fails to inhibit infection (see, Fig. 3A). As To verify a role for avb6and avb8-integrin in HSV attachment, we measured the effect of integrin silencing on HSV absorption to cells at 4uC. Virus absorption was quantified as a decrease in infectious virions present in the inoculum. Fig. 6 C-E shows that virus absorption was inhibited by about 50% in b6or b8-integrin silenced 293T, HeLa, and SW480 cells, indicating that avb6and avb8-integrins contribute to virus absorption to cells. This contribution is small relative to that exerted by heparan sulphate. In cells where heparan sulphate is removed by treatment with heparitinase, in mutants defective in heparan sulphate biosynthesis, or in cells in which infection is competitively blocked by heparin, the decrease in infection is about hundred fold [38] [39]. Notwithstanding this consideration, we asked whether the decrease in HSV absorption seen in cells silenced for b6or b8-integrins was to be attributed to the decrease in integrins, or reflected an indirect effect of integrin silencing on the extent of cell surface expression of heparan sulphate (HS). We exploited the ability of HSV gB to bind HS, and made use of a soluble form of gB (gBt) to quantify the HS binding sites present on the surface of non-silenced versus silenced cells. The properties of gB t , truncated at aa 730 and tagged with a strep tag (previously named gB 730t-st ) were described [33]. Binding to cell surface was quantified by cell enzyme linked immunosorbent assay (CELISA). Preliminarily, we ascertained that the binding of gB t was competitively inhibited by heparin; in contrast, attachment of a soluble form of gH/gL (gHt/gL) was not inhibited, as expected ( Fig. 6 F) [33]. Fig. 6 G -I shows that there was no significant decrease in gB t binding, following b6or b8-integrin silencing. These results rule out that the decrease in HSV absorption to b6or b8-integrin-silenced cells (Fig. 6 C-E) was due to a silencingmediated decrease in HS binding sites.
avb6and avb8-integrins do not substitute for nectin1, yet they greatly increase extent of infection in cells that express nectin1 at low levels For some herpesviruses, e.g. EBV, integrins suffice as receptors for infection of epithelial cells. Here, we asked whether either avb6or avb8integrin suffices as an HSV receptor for entry, or whether their roles are in addition to that of the gD receptors. We made use of the J cells, which are negative for both gD receptors, hence cannot be infected by HSV unless a receptor is transgenically expressed [5]. J cells express endogenous hamster integrins, likely at low levels, hence they are not suitable to test the effect of integrin silencing, or of antibodies to integrins, since the siRNAs and the antibodies were directed to the human orthologs. They were transfected with plasmids encoding for nectin1, or avb6-integrin, or avb8-integrin (300 ng DNA/well for each plasmid), and then infected with increasing amounts of R8102. Fig. 7 A shows that J cells expressing avb6-integrin alone or avb8integrin alone, in the absence of nectin1, did not enable HSV infection, in contrast to cells expressing nectin1 alone. Thus, neither of the two integrins suffices as HSV receptor in cells negative for gD receptors. Parenthetically, in this assay, in the absence of gD receptors, the candidate gB receptor PILR a [40] did not enable infection.
Next, J cells were transfected with low amounts of nectin1 plasmid, plus avb6-integrinor avb8-integrin plasmids (75 ng DNA/well for nectin1 plasmid, 300 ng DNA/well each for a and for b integrin-subunits). It can be seen from Fig. 7 B that avb6or avb8-integrin each doubled the efficiency of infection attained with nectin1 alone. Thus, avb6and avb8-integrin are not sufficient to substitute for nectin1, but they do increase nectin1dependent infection.
Infection and cell-to-cell fusion mediated by avb6-, but not by avb8-integrin requires the RGD motif in HSV gH av-integrins can interact with their ligands through a RGD motif. Because HSV gH carries an RDG motif at aa 176-178, we asked whether the interaction of avb6or of avb8-integrin with gH occurs through this motif. A mutant form of gH in which the RGD motif was substituted to ADA [41] was used in two functional assays: infection and cell-to-cell fusion. For infection, the DgH HSV mutant ScgHZ [42] was grown in cell lines expressing gH wt or gH ADA . The cell-expressed gH complements the deletion in the virus, and virions pseudotyped with gH wt or gH ADA are generated. J cells transfected with nectin1 plus avb6or avb8-integrin, or with nectin1 alone, were infected with virions carrying gH wt or gH ADA . Fig. 8 A-C shows that infection of J cells expressing nectin1 alone occurred irrespective of whether virions carried gH wt or gH ADA , in agreement with a previous report [32]. In contrast, infection of J cells expressing nectin1 plus avb6integrin was severely impaired when virions carried gH ADA . Infection of J cells expressing nectin1 plus avb8-integrin was only slightly inhibited when virions carried gH ADA .
The cell-to-cell fusion assay mimics virus-to-cell entry in that it requires the same four essential virion glycoproteins (gD, gH/gL, gB), and one of the gD receptors nectin1 or HVEM [43]. The effector cells are transfected with the quartet of gH/gL, gD, gB, plus T7-promoter-driven luciferase [44]. The target cells are    . (A, B) The indicated cells were exposed to MAb 2077Z to avb6-integrin, MAb 37E1 to avb8-integrin, MAb L230 to av-integrin, MAb R1.302 to nectin1, or control IgGs. In the ''Prior-during-post'' treatment cells were exposed to the antibodies from 1 h prior to infection till time of harvest, at 6-8 h after infection. In the ''Prior-during'' treatment cells were exposed to the antibodies from 1 h prior to infection and during virus absorption. In the ''Post'' treatment, cells were exposed to the antibodies from the end of virus absorption till harvesting. Cells were infected with R8102 (3 pfu/cell). Infection was quantified in triplicates as detailed in the legend to Fig. 3, and expressed as percentage relative to cells treated with control IgGs. (C-E) Inhibition of R8102 absorption to integrin-silenced cells. b6or b8-integrins were silenced by siRNAs. Control cells received siRNA control. Cells silenced for 2-3 days were infected with R8102 for 120 min at 4uC. Aliquots of the viral inoculum were withdrawn in duplicates at the indicated times, and immediately titrated in Vero cells. Extent of virus absorption is expressed as percentage of the amount of virus present in the inoculum at 0 time. Bars show SD. (F) The binding of gB t , and not that of gH t /gL, to cells is inhibited by heparin. One-StrEP tagged gB t , gH t /gL or GFP t (2 mM each), the latter as a negative control, were preincubated or not with heparin (5 mg/ml) for 1 h at 4uC, and then added to cells grown in 96-well plates, for 1 h at 4uC. transfected with the required glycoprotein receptor and T7-polymerase. In our assay the target J cells were transfected with gH wt or gH ADA plus the trio of gD, gL, gB. The effector cells were transfected with nectin1 alone, nectin1 plus avb6-integrin, or nectin1 plus avb8-integrin. Fig. 8 D shows that cell-to-cell fusion was increased by 40% or 30% when J cells expressed avb6integrin or avb8-integrin, plus nectin1, relative to cells expressing nectin1 alone. This occurred with the wt allele of gH. When the gH ADA substituted for gH wt there was a dramatic inhibition in the avb6-integrin-enhanced fusion, but not in the avb8-integrinenhanced fusion. The results of the two assays indicate that the gH interaction with avb6-, but not with the avb8-integrin, entails the RGD motif in gH.
avb6-integrin and avb8-integrin enable specific endocytic pathways of HSV entry HSV enters different cells by different pathways. avb3 integrin was identified in our laboratory as a cellular determinant in the choice of the pathway of entry, capable of routing HSV to a Binding was detected by means of HRP-conjugated MAb to the One-StrEP tag and the o-phenylenediamine substrate [41]. Each point represents the average of triplicates. The values obtained with GFP t were considered as background values and subtracted. Bars represent SD. (G-I). Determination of heparin sulphate binding sites by means of gB t to integrin silenced and control-silenced cells. (G-I). 293T (G), HeLa (H) and SW480 (I) cells, controlsilenced or integrn-silenced, were exposed to gB t , pretreated or not heparin. Binding was detect as descibed in panel F. Columns represent the binding of gB t , after subtraction of the values obtained in the presence of heparin. Each column represents the average of triplicates. Bars denote SD. doi:10.1371/journal.ppat.1003806.g006  Cell-to-cell fusion between effector J cells expressing gH wt , or gH ADA plus the trio of gD, gL and gB, and luciferase, and target J cells expressing nectin1 alone, or nectin1 plus avb6-integrin, or nectin1 plus avb8-integrin plus Renilla luciferase. Fusion was quantified by means of a T7 promoter-driven reporter luciferase gene transfected in effector cells, and expressed as percentage relative luciferase units (R.L.U.). 100% is the value obtained in cells expressing nectin1 alone and gH wt plus the trio of gD, gL, gB. Each point represents the average of triplicates. Bars show SD. doi:10.1371/journal.ppat.1003806.g008 pathway dependent on lipid microdomains and dynamin 2, and proceeding to acidic endosomes [30]. Whether integrins other than avb3 are determinants in the choice of HSV entry pathways, and whether different integrins route HSV to different pathways is not known. To address these questions, we analyzed the effects of well known entry inhibitors on HSV infection of J cells expressing nectin1 alone, or nectin1 plus avb6-, avb8-, or avb3-integrins. J cells are suitable to address this question, since, when transfected with a gD receptor they enable a pathway of entry independent of lipid rafts and acidic endosome [45].
The significant inhibitors were bafilomycin A (BFLA), a specific inhibitor of the Na-H pump, hence of endosomal acidification, filipin III, an inhibitor of the lipid microdomain platforms, dynasore, an inhibitor of the dynamin 2 GTPase required to seal the endosomal invaginations and generate endosomes, and wortmannin, an inhibitor of phosphoinositide 3-kinase (PI3K). The results were as follows (Fig. 9). Infection of J cells expressing nectin1 alone was not sensitive to BFLA (Fig. 9 A), as reported [45], and occurs by fusion at plasma membrane or at a neutral pH compartment. Anyone of the three integrins -avb6, avb8 and avb3 -rendered the entry pathway sensitive to BFLA (Fig. 9 A). The wortmannin sensitivity closely mirrored that of BFLA, implying an involvement of PI3K when HSV entry occurs through acidic endosomes, but not at the plasma membrane or in a neutral compartment (Fig. 9 B). Filipin III did not inhibit infection in J cells expressing nectin1 alone ( Fig. 9 C), but did inhibit infection in J cells expressing nectin1 plus avb8-, or avb3integrin (Fig. 9 C). Remarkably, filipin III failed to inhibit infection of J cells expressing nectin1 plus avb6-integrin. The effect of dynasore closely overlapped that of filipin III (Fig. 9 D). The filipin III and dynasore sensitivity upon avb3-integrin expression was in agreement with our previous findings [30]. Cumulatively, the results indicate that both avb6and avb8-integrin greatly impact on the HSV entry pathway. Both switch HSV to acidic endosome entry. avb8-, like avb3-integrin, routes HSV to a pathway dependent on lipid microdomain and dynamin2; in contrast, avb6-integrin routes HSV to a pathway independent of lipid microdomain and dynamin 2.

Discussion
The epithelial av group of integrins, also named as RGD receptors, includes avb3-, avb5-, avb6-, avb8-integrins, in addition to the aIIbb3. We report that: Both avb6and avb8-integrins bind gH/gL at high affinity. The binding is highly selective as avb5-integrin does not bind gH/gL at all. The binding sites of avb6and of avb8-integrins in gH/gL differ; it is the RGD motif for avb6-integrin, but not for avb8-integrin. Consistent with the high affinity binding, both the avb6and avb8-integrin contributed to attachment of virions to cells, a step which so far was ascribed to gC and gB binding to HS, and to gD binding to one its receptors. The contribution of avb6or avb8-integrin to HSV attachment was much lower than that of heparan sulphate, whose absence reduces infection by about 100 fold [38] [39]. Together, the requirement for avb6or avb8-integrin for infection and the high affinity binding are criteria to define avb6and avb8-integrins as HSV receptors. Importantly, as shown also here, avb6and avb8-integrins are preferentially expressed in epithelial cells, and b8-integrin is expressed at high levels in the central nervous system [34,35]. Epithelial cells and central nervous system are major targets of HSV infection in vivo. Further yet, avb6 is up-regulated during tissue remodeling, including that accompanying wound healing [46], a condition that favors keratinocyte infection by HSV [47].
The roles of avb6or avb8-integrin as HSV receptors are in contrast to that of avb3-integrin. The latter binds HSV gH/gL at a 100 fold lower affinity [31], and serves as a routing factor, but not as a bona fide receptor; its silencing modified the route of infection but does not reduce it [31]. (iv) avb6or avb8-integrin did not substitute for the gD receptor nectin1. The role of integrins is therefore additional -and not alternative -to that of gD receptors. Altogether, HSV entry into cells requires both the interaction of gD with one of its receptors, and the gH/ gL interaction with avb6or avb8-integrin, a view which impacts on the current model of HSV entry, as detailed below (v) The most dramatic effects exerted by avb6and avb8integrins were modifications of the pathway of HSV entry. Both integrins routed HSV from a plasma-membrane (or neutral pH compartment) portal of entry to acidic endosomes, and, concomitantly, to PI3K-dependency. Interestingly, the two integrins differed in the specific endocytic pathway they enabled. avb8-integrin enabled an entry pathway dependent on lipid microdomains and dynamin 2. avb6-integrin enabled a pathway of entry independent of lipid microdomains and dynamin 2.
Cumulatively, the novel features to emerge from this study are that HSV entry into epithelial cells requires the gH/gL interaction with one of the two interchangeable receptors, avb6-integrin or avb8-integrin. This interaction profoundly affects the pathway of entry, which takes the way of acidic endosomes. In the case of involvement of avb8-integrin, the portal of entry is located at lipid microdomains and necessitates dynamin 2. In the case of involvement of avb6-integrin, the portal of entry is located outside the lipid microdomains and does not necessitate dynamin 2.
Two questions arise. Why did HSV evolved to employ gH/gL receptors in addition to the gD receptors? And, why did HSV choose integrins as gH/gL receptors? Two characteristics of the entry of HSV, and of herpesviruses in general, are the multipartite nature of the entry apparatus, and the possibility to enter cells through a variety of pathways. The currently accepted model on how the HSV multipartite entry/fusion apparatus mediates fusion of the virion envelope with the cell membranes envisions sequential activation steps in a cascade fashion [7,8,11,12,15]. Even though the details of the interactions are still being defined, there is consensus that, upon gD binding to one of its receptors, conformational changes to gD ensue [11,12], such that gD interacts with, or anyway propagates the activation to gH/gL, and thereafter or simultaneously to gB [9,10,14]. Current results introduce the intervention of the integrin receptors in this process. We propose that, in addition to promoting endocytosis, the interaction of gH/gL with the integrin receptor contributes to induce conformational changes and activation of gH/gL, as seen in EBV [48] (Fig. 10). Consequently, integrins would act as a trigger point in the activation cascade of the HSV fusion glycoproteins. Indeed, by placing the step of gH/gL activation under the double control of receptor-activated gD and of the integrin trigger point, the virus would ensure that the premature activation of gB is prevented, and, importantly, that virion endocytosis is synchronized with the activation of the fusion apparatus. In particular, when entry occurs at the plasma membranes, the receptor-mediated gD activation propagates to gH/gL and gB, possibly in a very brief timescale, perhaps in a matter of fraction of seconds. In the case of HSV entry by endocytosis, there must be a pause in the activation cascade, or a checkpoint that prevents the premature activation of gH/gL and gB, before endocytosis has taken place. An activating receptor for gH -in this case integrins, which are machines specialized in promoting endocytosis [49] -is ideally suited to fulfill this synchronization (Fig. 10). The entry pathways exploited by HSV include low pH or neutral pH endosomes, as well as the plasma membrane, which is also a neutral pH compartment. As a consequence, HSV had to evolve a strategy for control of activation of the fusion glycoproteins independent of acidic pH. The integrin trigger point on gH/gL results in a control on fusion  (1), the interaction of gD with one of its receptors activates gD (2); the activation is propagated to gH/gL; the details of this interaction are still under investigation. Integrins intervene (3): they promote endocytosis, and, most likely, induce or contribute to conformational changes, hence to activation of gH/gL. The gH/gL activation is propagated to gB. Endosomal acidification (4). Fusion execution by gB (5). Color code of the glycoproteins as in [69]. gH/gL and gB are shown in the resolved post-fusion structures. doi:10.1371/journal.ppat.1003806.g010 independent of acidic pH, and thus confers to the virus the flexibility to use alternative pathways in different cells, and expands the range of cells that HSV infects.
Up to now, among herpesviruses, integrins are known to serve as receptors for EBV and HCMV, which engage with integrins through gH/gL, and for HHV-8, which engages with integrins through gB [20][21][22][23]. In EBV, the integrin-gH/gL interaction triggers conformational changes to the glycoproteins, likely activating them [48]. Whether all human herpesviruses need integrins in some way to enter certain cell types is an attractive open possibility. All herpesviruses encode a multipartite fusion apparatus, and enter a number of cells by endocytosis. Importantly, for EBV, HCMV and HHV8, placing the activation of gH/ gL (EBV, HCMV) or gB (HHV-8) under an integrin trigger point may well result in a synchronization of the final steps of activation of the glycoproteins responsible for fusion execution. While many viruses in other families make use of integrin receptors to promote endocytosis [50][51][52], most either lack an envelope, e.g. adenoviruses and reoviruses, or do not encode an entry system as complex as that of a herpesvirus. Their usage of integrins thus intrinsically differs from that of HSV, as entry is less dependent on checkpoints to control an activation cascade.
We recently found that avb3integrin -which functions as a routing factor and drives HSV entry to a pathway dependent on lipid microdomains, dynamin 2 and acidic endosomes [30,29] serves as a sensor of the virus. In cooperation with TLR2, it elicits a branch of the innate response that contributes to NF-kB activation and production of a specific set of cytokines, mainly IFNa and b [31]. The question naturally arises as to whether the engagement of avb6 or avb8-integrin as receptors initiates a similar inimical response to the virus. In turn, this raises the question of what selective advantage is provided to HSV by use of an integrin to control fusion. As noted in earlier studies, HSV has evolved to evade the innate response soon after the onset of viral protein synthesis; in particular, infected cell protein 0 (ICP0) and ICP27 provide the first line of evasion [53][54][55][56]. We conclude that, as highlighted in this work, HSV takes advantage of integrins to enable its endocytic entry and to exert a checkpoint on the cascade of glycoprotein activation. It subsequently counteracts the innate response, when virus entry is completed, by aid of the immediate early proteins.

Cells, viruses and soluble proteins
293T, HeLa, SK-N-SH and J (a derivative of BHK-TK 2 cells lacking any HSV receptor) [5] cells were grown in Dulbecco's modified Eagle's medium (DMEM) containing 5% to 20% fetal bovine serum (FBS). Colon carcinoma SW480, HaCaT and K562 cells were grown in L15, DMEM 4.5% glucose and Iscove's modified Dulbecco's medium, respectively. The growth medium for K562 avb6 (a gift from Dr S. Blystone) contained 750 mg/ml neomycin G418. F6 cells were a stably transformed Vero cell line expressing HSV-1 gH under the control of HSV-1 gD promoter [42]. R8102, a HSV-1 recombinant carrying LacZ under the control of the a27 promoter [5] and RLM5, a HSV-1(F) recombinant expressing green fluorescent protein (GFP) [57] were described. In the gH deletion mutant (DgH HSV) SCgHZ the gH gene was replaced with LacZ gene, the virus was grown and titrated in the complementing F6 cells [42]. 293-B6 AVAP [58] and 293-B8 AVAP cells [59] (a gift of Dr. Stephen Nishimura, University of California at San Francisco), which secrete, respectively, truncated avb6 and avb8 integrin conjugated to alkaline phosphatase (AP), were grown in DMEM (Sigma) supplemented with 10% fetal bovine serum and 1% nonessential amino acids. Insect Sf9 cells were grown in Sf900 II medium (Invitrogen) and infected at a multiplicity of 2 with equal ratios of baculoviruses expressing truncated forms of av and b5 cloned in frame respectively with a fos or jun dimerization domain [60] (a gift of Dr. Glen Nemerow, Scripps Research Institute). Soluble truncated integrins were purified as previously described [48]. Soluble gH t /gL and gB t , carrying One-StrEP-Tag epitope for affinity chromatography purification, were described previously [33,41]. One-StrEP-tagged green fluorescent protein (GFP t ) was provided by IBA GmbH (Göttingen) and used as negative control.

Plasmids
Plasmids encoding HSV-1 gD, gB, gL and gH all under the control of the cytomegalovirus (CMV) promoter were described [61]. gH ADA carries the indicated substitutions in the RGD motif [41]. pCAGT7 contained the T7 RNA polymerase gene under the control of the CAG promoter and pT7EMCVLuc plasmid expressed the firefly luciferase under the control of the T7 promoter [62]. EGFR2ND (epithelial growth factor receptor 2 D), named Erb-2, carriers the extracellular domain and transmembrane sequence of rat HER-2/neu, and is deleted of the tyrosine kinase domain [63]. Plasmids encoding nectin1 [5], Renilla luciferase (Promega), av-, b6and b8-integrin were described [36,64].

Antibodies
MAb 2077Z to the avb6 integrin heterodimer is a functionblocking antibody from Chemicon. MAb 37E1 to avb8 integrin, MAb R1.302 to nectin1 and PAb to HVEM were gifts from Nushimura S.L [65], M. Lopez [5] and G. H. Cohen and R. Eisenberg [66]. MAb L230 is a function-blocking antibody directed to av integrin [67]. For surface plasmon resonance analysis MAb LS-C44264 to placental alkaline phosphatase (LifeSpan BioScience) was used to capture integrins avb6AP, and avb8AP the non-blocking monoclonal antibody LM142 to av integrin (Millipore) was used to capture integrin avb5.

Surface plasmon resonance spectroscopy
The kinetic parameters of the interaction between gH/gL and integrins were measured with a Biacore 2000 instrument (Biacore AB). Antibodies were immobilized on a research-grade CM-5 sensor chip by amino coupling with 1,200 relative units (RU) as a target for immobilization and used to capture soluble integrins. The first flow cell (FC1) was always used as a reference (no antibody immobilized), and the signal from FC1 was automatically subtracted by Biacore software from data obtained from the other three flow cells. Measurements were made at 25uC. Integrins were injected at a flow rate of 10 to 20 ml/min; gH/gL was injected immediately after integrin capture by using the low-dispersal injection setting (kinject) at a flow rate of 50 ml/min. After each run, the sensor chip surface was regenerated by injection of 6 M guanidine chloride in 25 mM HEPES-NaOH, pH 7.2 (50 ml at a flow rate of 100 ml/min), followed by additional washing. The baseline remained stable during 30 to 40 runs. Integrins and gH/ gL were centrifuged before use and diluted in running buffer if necessary (10 mM HEPES-NaOH, pH 7.4, 150 mM NaCl, 0.005% surfactant P20 (Biacore, GE Healthcare). Each sample was degassed before injection.
Curve fitting BIAevaluation 4.1 software was used for the Biacore trace alignments and to zero the baseline. All traces were in full avb6and avb8-Integrins As HSV gH/gL Receptors accordance with a 1:1 Langmuir model. Formation and dissociation sections of Biacore traces fit a single-exponential function which was accepted as a model. Least-squares fitting of data and determinations of standard errors of the fitted parameters were conducted using the program KaleidaGraph (Synergy Software, Reading, PA).
b6and b8-integrin silencing One target plus siRNAb6 or b8 (Dharmacon, ON-TARGET plus, smart pool) were trasfected into 293T, HeLA or SW480 cells, 50 nM for each siRNA, by means of Dharmafect I, according to Dharmacon protocol for adherent cells. Control-silenced cells were transfected with siRNA to E.coli-polA_0054 (CGC GUG AUA UGC GAC GCG AUA AAG) synthesized by IBA Gmbh.
Inhibition of HSV infection by MAbs to avb6 or avb8integrins and in integrin-silenced cells 293T, HeLa, SW480, HaCaT and SK-N-SH cells in 96 wells were preincubated with increasing amounts of MAb R1.302 to nectin1, MAb 2077Z to avb6-integrin, MAb 37E1 to avb8integrin, or with mouse IgGs for 60 min at 37uC. R8102 (3 pfu/ cell) was added to the MAb-containg medium for additional 90 min. Viral inoculum was removed, and cells were overlaid with DMEM containing MAbs for 6-8 h. Control-silenced, b6integrin-silenced or b8-integrin-silenced 293T, HeLA or SW480 cells were infected with R8102 (3 pfu/cell) for 90 min at 37uC. The viral inoculum was removed and cells were overlaid with DMEM 1% FBS for 6-8 h. The extent of infection was assessed through b-galactosidase (b-Gal) activity at 405 nm, by means of onitrophenyl-b-D-galactopyranoside (ONPG), or by in situ staining with 5-bromo-4-chloro-3-indolyl-b-D-galactopyranoside [5]. K562 avb6 were preincubated or not with 40 ug/ml of MAb 2077Z to avb6-integrin, MAb L230 to av-integrin, or PAb R140 to HVEM for 1 h at 37uC. K562 and K562 avb6 cells were infected with RLM5 (20 pfu/cell) in 3 ml for 90 min. The viral inoculum was removed and cells were overlaid with DMEM containing or not the indicated antibodies for 6 h. Control-silenced, b6-integrinsilenced, b8-integrin-silenced or double-silenced 293T cells were infected with RLM5 (3 pfu/cells) for 90 min at 37uC. Viral inoculum was removed; the extent of infection was assessed through EGFP expression in infected cells, 6 h after infection.

Flow cytometry
Expression of avb6 or avb8-integrins in 293T, HeLA and SW480 cells was determined by FACS. In 293T cells (controlsilenced and integrin-silenced) integrin expression was determined 72 h after transfection with siRNActrl, siRNAb6, siRNAb8 or a mixture of both. Cells were washed once with PBS (phosphate buffered saline) and once with PBS containing 10% FBS and allowed to react with MAb 2077Z to avb6-integrin, MAb 37E1 to avb8-integrin heterodimers for 1 h at 4uC. Cells were washed three times with PBS and allowed to react with phycoerythrin conjugated secondary antibody (Becton-Dickinson Pharmingen). Where indicated, cells were infected with RLM5, and infection was quantified through EGFP expression at 6 h after infection. The inhibition of HSV infection by MAb 2077Z to avb6-integrin, MAb L230 to av-integrin and PAb to HVEM in K562 and K562 avb6 cells was quantified through enhanced green fluorescent protein (EGFP) expression from the RLM5. Cytofluorimetric analysis was performed using a FACScalibur cytometer (BD), equipped with an argon laser on a minimum of 10,000 cells per sample, acquired in list mode.

Inhibition of infection by pharmacological inhibitors
The stock solutions of filipin III (2.5 mM), dynasore (100 mM), bafilomycin A (BFLA) (160 mM) and wortmannin (2 M) (all from Sigma Aldrich) in dimethyl sulfoxide were stored at -20uC. Cells were exposed to the inhibitors for 1 h at 37uC and then infected with R8102 (3 pfu/cell) for 90 min in the presence of inhibitors. The viral inoculum was removed, and the cells were overlaid with medium containing inhibitors for 6-8 h. For filipin III and dynasore, cells were preincubated with the compounds at 37uC for 30 min or 60 min, respectively, and infected for 30 min (30 pfu/cell) in the same medium. Viral inoculum was removed; infected cells were overlaid without inhibitor and harvested 6-8 h after infection.
Virions complemented with gH wt or gH ADA 293T cells in T 150 flasks were transfected with gH wt or gH ADA plus gL. After 4 h, cells were infected with a gH2/+ stock of SCgHZ (5 pfu/cell) [42]. Cell-absorbed virions were inactivated by a 1 min rinse with 40 mM sodium citrate-10 mM KCl-135 mM NaCl (pH 3). The monolayers were then rinsed twice with PBS and overlaid with medium containing 1% fetal calf serum. Cells were incubated overnight at 37uC. Extracellular virions were harvested, pelletted by high speed centrifugation, and titrated in F6 cells.
Contribution of avb6or avb8-integrin to HSV infection J cells were transfected with plasmids encoding nectin1 (300 ng DNA/24 well), av plus b6, or av plus b8 integrins (300 ng each/ 24 well) by means of Lipofect 2000 (Life Technologies). Alternatively, J cells were transfected with low amount of nectin1 plasmid (75 ng DNA/24 well), plus or minus av+b6 or av+b8integrins (300 ng DNA/24 well). The total amount of transfected plasmid DNA was made equal (675 ng/24 well) by the addition of Erb-2 plasmid DNA. 48 h after transfection, cells were infected with increasing MOI (2.5-30 pfu/cell) of R8102, or of SCgHZ virions complemented with gH wt or with gH ADA , for 90 min at 37uC. After infection cells were overlaid with DMEM and harvested 16-18 h after infection.

Cell-cell fusion assay
The luciferase-based cell-cell fusion assay was performed as described [44], with small modifications. Effector J cells were transfected with plasmid encoding gL, gD, gB, gH wt or gH ADA . Target J cells were transfected with Renilla luciferase, low amount of nectin1 (75 ng DNA/24 well), plus or minus av+b6, or av+b8 (300 ng DNA for each plasmid/24 well). The total amount of plasmid DNA transfected was made equal by the addition of Erb-2 plasmid DNA. 24 h after transfection, effector and target cells were co-cultured for additional 24 h. Fusion was quantified by means of the T7 promoter-driven reporter luciferase gene; the amount of lysate was quantified by Renilla luciferase. Quantification was performed by means of Dual luciferase report assay (Promega). The extent of fusion was expressed as relative luciferase units (R.L.U.); 100% represents the extent of fusion induced by gH wt in cells expressing nectin1 alone.
Binding of to cell surface by CELISA Cells grown in 96-well were incubated for 1 h at 4uC with One-StrEP tagged gB t , gH t /gL, or GFP t (2 mM) in DMEM containing 5% FBS and 30 mM HEPES [68], washed three times with the same buffer, and further incubated for 1 h at 4uC with HRP (horse radish peroxidase)-conjugated MAb to One-StrEP tag (Strep-Tactin) (IBA GmbH, Göttingen). Following three additional rinsings, cells were reacted with o-phenylenediamine (Sigma-Aldrich) at 0.5 mg/ml; the optical density was read at 490 nm. For heparin inhibition studies glycoproteins were preincubated for 1 h at 4uC with 5 mg/ml heparin (Sigma-Aldrich), prior to addition to cells.