Report

Vaccinia Virus-Induced Cell Motility Requires F11L-Mediated Inhibition of RhoA Signaling

See allHide authors and affiliations

Science  20 Jan 2006:
Vol. 311, Issue 5759, pp. 377-381
DOI: 10.1126/science.1122411

Abstract

RhoA signaling plays a critical role in many cellular processes, including cell migration. Here we show that the vaccinia F11L protein interacts directly with RhoA, inhibiting its signaling by blocking the interaction with its downstream effectors Rho-associated kinase (ROCK) and mDia. RNA interference–mediated depletion of F11L during infection resulted in an absence of vaccinia-induced cell motility and inhibition of viral morphogenesis. Disruption of the RhoA binding site in F11L, which resembles that of ROCK, led to an identical phenotype. Thus, inhibition of RhoA signaling is required for both vaccinia morphogenesis and virus-induced cell motility.

The spatial and temporal regulation of cell adhesion and motility is essential during development and throughout the lifetime of multicellular organisms (1, 2). Deregulation of these two fundamental cellular processes frequently occurs during pathological situations such as tumor cell metastasis (3, 4). Dramatic changes in cell migration and adhesion, as well as loss of contact inhibition, are also observed during many viral infections, including that of vaccinia virus (5, 6). In contrast to the wild-type Western-Reserve (WR) virus, we found that modified vaccinia Ankara (MVA), a highly attenuated virus strain containing multiple mutations and deletions in its genome (79), did not induce cell migration or cellular projections in the absence of serum (Fig. 1, Movies S1 to S3) (10).

Fig. 1.

F11L is required for vaccinia-induced cell motility. (A) Phase contrast movie stills from wound-healing assays reveal that BS-C-1 cells do not migrate into the wound site in the absence of serum (top). In contrast, infection with the wild-type Western Reserve virus (WR, middle), but not modified vaccinia Ankara virus (MVA, bottom), induces BS-C-1 cell migration. The time after the initial wound (hours) is indicated. (B) Enlargement of the WR-infected cell highlighted in (A) (box, 10h) reveals that migrating cells fail to retract their trailing edge and extend multiple lamellipodia. (C) Immunofluorescence analysis reveals that WR and MVA infection results in a loss of actin stress fibers but that only WR induces the projection phenotype. Expression of GFP-F11L induced the projection phenotype in MVA-infected cells. Bar, 10 μm.

To determine the gene (or genes) absent in MVA that are responsible for vaccinia-induced cell motility, we examined the ability of a series of recombinant MVA virus strains, which contained portions of DNA from its parental strain (9), to induce the projection phenotype. Only MVA recombinants rescued with cosmid 44 induced a projection phenotype (fig. S1), suggesting that the genomic region covered by cosmid 44 contained a gene (or genes) required for vaccinia-induced cell motility. Of the five genes disrupted or deleted in this genomic region, only one, F11L, or its green fluorescent protein (GFP)–tagged variant promoted the projection phenotype when expressed in MVA-infected cells (Fig. 1C, fig. S2) (10).

When expressed in uninfected cells, GFP-F11L induced a loss of actin stress fibers reminiscent of the dominant-negative RhoA-N19 (Fig. 2A). Furthermore, F11L inhibited stress fiber induction by coexpressed constitutively active Rho-V14 (Fig. 2A), suggesting that F11L might inhibit signaling by RhoA. To examine whether F11L interacted with RhoA or acted downstream of the guanosine triphosphatase (GTPase), we performed pull-down assays from cells coexpressing GFP-tagged RhoGTPases and F11L tagged with glutathione-S-transferase (GST) (10). RhoA but not Cdc42 or Rac copurified with F11L from cell extracts (Fig. 2B). Moreover, F11L interacted with both constitutively active RhoA-V14 and dominant-negative RhoA-N19 (Fig. 2B). Endogenous F11L and RhoA were also complexed with each other in vivo in infected cells (Fig. 2C). Moreover, a direct interaction between F11L and RhoA could be demonstrated with GST and His-tagged proteins produced in Escherichia coli.

Fig. 2.

F11L binds directly to RhoA. (A) Expression of GFP-F11L or GFP-RhoA-N19 induces a loss of actin stress fibers in BS-C-1 cells. Coexpression of GFP-F11L with GFP-RhoA-V14 blocks the ability of RhoA to stimulate actin stress fiber formation. Bars, 10 μm. (B) Immunoblot analysis reveals that GST-F11L copurified from HeLa cell extracts with GFP-Rho but not GFP-Rac or GFP-Cdc42. GST-F11L also interacted with GFP-RhoA-V14 and GFP-RhoA-N19. The supernatant (Sup) and bound (B/D) samples, together with the respective RhoGTPase or RhoA mutant, are indicated. (C) Immunoblot analysis of coimmunoprecipitation experiments revealed that F11L is complexed with RhoA in infected cells. (D) (Top) Immunoblot analysis of in vitro pull-down assays reveals that alanine substitution of conserved residues shared between F11L and ROCK substantially reduced the binding of F11L to RhoA. (Bottom) Double alanine substitutions in F11L resulted in a complete loss of binding. (E) Immunoblot analysis of in vitro pull-down assays reveals that F11L does not interact with Rho-V14/E40L.

Secondary-structure predictions suggest that F11L adopts a globular structure with an N-terminal domain (residues 1 to 222) rich in β sheet and a C-terminal helical domain (residues 223 to 345) containing at least one contiguous predicted helix longer than 50 residues. We found that the C-terminal helical domain of F11L but not the N-terminal region was able to interact directly with RhoA (fig. S3). Closer examination of the C-terminal region of F11L revealed that residues 299 to 312 share a limited homology to residues 998 to 1010 of Rho-associated kinase (ROCK) (fig. S3). This region of ROCK is known to mediate its interaction with RhoA (11). Substitution of four of the five identical residues with alanine substantially weakened the interaction between F11L and RhoA (Fig. 2D) (10). Combinations of double mutations resulted in a complete lack of binding (Fig. 2D). Thus, F11L appears to interact with RhoA in a manner similar to that of ROCK. Consistent with this hypothesis, the RhoA-V14E40L mutant that is deficient in ROCK binding (12) could not interact with F11L (Fig. 2E). In contrast to the wild-type protein, GFP-tagged F11L mutants deficient in Rho binding were also unable to stimulate the projection phenotype when expressed in MVA-infected cells (fig. S4). Thus, F11L inhibits the ability of the GTPase to signal during infection by binding directly to RhoA.

The failure of the rear of infected cells to detach and the formation of multiple lamellipodia at the front are very reminiscent of the effects of inhibition of ROCK in a variety of migrating cells (1316). We examined whether the interaction of F11L with RhoA-V14 affected the ability of the GTPase to bind ROCK. F11L inhibited the ability of RhoA-V14 to retain ROCK from HeLa cell extracts (Fig. 3A) (10). Treatment of MVA-infected cells with the ROCK inhibitor Y-27632 (16), although inducing the projection phenotype, was unable to stimulate cell migration (Fig. 3B), suggesting that modulation of additional signaling pathways are also required for vaccinia-induced cell motility. Consistent with this notion, F11L also inhibited the ability of RhoA to interact with mDia (Fig. 3A). To analyze whether F11L does indeed modulate RhoA signaling, we examined the level of GTP-bound RhoA in WR-infected cells in the absence of F11L. Transfection of small interfering RNA (siRNA) oligonucleotides against F11L before vaccinia infection markedly suppressed expression of the protein without affecting the levels of other viral proteins (Fig. 3C) (10). A reduction in the level of F11L expression resulted in an increase of GTP-bound RhoA in infected cell extracts (Fig. 3D). Concomitantly, the phosphorylation level of the Rho effector ezrin was increased and virus-induced cell motility was lost (Fig. 3, E and F; Movie S4). Thus, F11L inhibits the ability of RhoA to signal to its downstream effectors during vaccinia virus infection.

Fig. 3.

F11L blocks RhoA signaling. (A) Immunoblot analysis reveals that a complex of GST-F11L and His-Rho-V14 (F11L/V14), in contrast to His-Rho-V14 alone, does not bind ROCK or mDia from HeLa cell extracts. The supernatant (Sup) and bound (B/D) material, together with the respective affinity resin of protein produced in E. coli, is indicated. (B) Inhibition of ROCK with Y-27632 resulted in the formation of cellular projections in MVA-infected (A27L-positive, green) but not in uninfected cells. Bar, 10 μm. (C) Immunoblot analysis reveals that F11L expression can be efficiently inhibited during WR infection with siRNA oligonucleotides (lanes 1 to 6) directed against F11L. Controls are represented by scr4 (scrambled oligonucleotide 4), GFP (unrelated oligonucleotide), and WR (mock-transfected). The loss of F11L does not affect viral protein expression (H5R or A36R). (D) Immunoblot analysis of rhotekin pull-down assays reveals that the level of GTP-Rho (arrowhead) is higher in the absence of F11L. (E) Loss of F11L also results in increased phosphorylation of ezrin (arrowhead). (F) Movie stills from a wound-healing assay of RNAi-treated WR-infected cells reveals that loss of F11L resulted in an absence of virus-induced motility. Transfected cells also contain a rhodamine-labeled control oligonucleotide (red) for identification purposes. The arrowhead indicates a single nontransfected cell, migrating into the wound. The time after the initial wound (hours) is indicated.

The sequence of F11L is strongly conserved within the genome of orthopoxviruses, suggesting that the protein may play an important role in the vaccinia virus life cycle. Indeed F11L, which is expressed from as early as 2 hours after infection, is associated with perinuclear viral factories (fig. S5). To elucidate the function of F11L in the vaccinia life cycle, we set out to delete the F11L gene in the WR genome by homologous recombination. However, in agreement with Kato et al. (17), we were unable to isolate recombinants lacking the F11L gene when using this strategy, suggesting that it is essential for the production of infectious virus in culture. To explore this possibility, we analyzed the effects of siRNA-mediated ablation of F11L expression on the vaccinia life cycle. Loss of F11L expression resulted in an absence of assembled virus particles as judged by the diffuse cytoplasmic localization of mature virus particle markers such as A27L and B5R (Fig. 4A). Analysis of siRNA-treated cells by electron microscopy (10) confirmed that the absence of F11L resulted in a defect in morphogenesis, with a substantial loss of assembled mature virus particles and an accumulation of aberrant immature virions (IVs) (Fig. 4B). An accumulation of IVs is also observed during the abortive life cycle of MVA in HeLa cells (18, 19).

Fig. 4.

Virus morphogenesis requires inhibition of RhoA signaling. (A) The absence of F11L resulted in disruption of perinuclear sites of virus assembly (arrowheads in WR and siRNA-scr4) and loss of assembled virus particles as judged by diffuse cytoplasmic localization of A27L and B5R (highlighted in the insets). Bar, 10 μm. (B) (Top) Epon sections of siRNA-treated HeLa cells infected with WR 8 hours after infection revealed that an absence of F11L resulted in loss of assembled intracellular mature virus particles (arrowheads). Immature virions were present in both samples, although their characteristic smooth curved profiles were often distorted (boxed areas are enlarged, bottom) in the absence of F11L. Bar, 1 μm. (C) Reinfection assays demonstrated that a loss of F11L expression resulted in a large reduction in the number of infectious particles released into the medium from WR-infected cells. Error bars represent the standard deviation of the mean from three independent experiments. (D) Immunoblot analysis of siRNA6-treated WR-infected cells expressing the indicated F11L rescue clone. Expression of the wild-type or L299/T302 mutant is inhibited by the siRNA6 oligonucleotide. In contrast, F11L resist, T302/L308, and V305/K307 are expressed because they contain point mutants that make them resistant to silencing by siRNA6 oligonucleotide. (E) Reinfection assays demonstrated that infectious virus particles are produced in siRNA6-treated WR-infected HeLa cells expressing F11L resist. In contrast, T302/L308 and V305/K307 mutants, which cannot bind RhoA, do not rescue virus morphogenesis in the absence of endogenous F11L expression. Error bars represent the standard deviation of the mean from three independent experiments.

To examine whether the loss of F11L expression reduced production of infectious virus particles, we performed reinfection experiments using the tissue culture medium from siRNA-transfected cells (10). We observed up to a ∼14-fold reduction in the number of infectious virus particles released into the tissue culture medium when compared with untreated cells (Fig. 4C). To analyze if the interaction of F11L with RhoA is required for viral morphogenesis, we examined whether expression of RNAi-resistant F11L mutants deficient in RhoA binding would rescue production of infectious virus in siRNA-transfected cells. In contrast to the wild-type protein, expression of F11L mutants deficient in RhoA binding were unable to rescue production of infectious virus in siRNA-transfected cells (Fig. 4, D and E). Thus, F11L-mediated inhibition of RhoA signaling is required for both vaccinia morphogenesis and infection-induced cell motility. Our observations may also explain, at least in part, why MVA, which lacks functional F11L, is unable to replicate in most cell types in culture (79). They also offer molecular insights into the mechanism by which vaccinia induces cell migration (5, 6). One can envisage that migration of an infected cell will increase the efficiency of virus spread because extracellular virus particles associated with the plasma membrane, which are responsible for direct cell to cell spread, will come into contact with more neighboring noninfected cells than they would if the infected cell were static.

Supporting Online Material

www.sciencemag.org/cgi/content/full/311/5759/377/DC1

Materials and Methods

Figs. S1 to S6

Movies S1 to S4

References

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article