RhoB is a component of the human cytomegalovirus assembly complex and is required for efficient viral production

RhoB is upregulated upon HCMV infection and localizes at the AC in HCMV-infected cells

In the first series of experiments, we studied the cytoplasmic distribution of RhoB at Early and Late stages of HCMV infection in fixed cells by immunofluorescence microscopy. Infected cells were stained with the viral nuclear protein pUL44 or with the viral proteins, pp65, pUL97, and pUL32 that were used to define the cytoplasmic area that corresponded to the AC at late stages of the infection,^11,50 although at earlier timepoints they localize at the nucleus.^51-53 Specifically, infected fibroblasts with an MOI of 0.5 pfu/cell were fixed and stained for RhoB and each of these viral proteins 2 days post-infection (d.p.i.), prior to AC formation, as well as 3 d.p.i. when the AC was formed. RhoB underwent significant changes in the abundance and localization over the time course studied. Compared to the hardly detectable, scattered, punctuate distribution of RhoB in the cytoplasm of mock infected cells (Fig. 1A), RhoB accumulated more intensely throughout the cytoplasm in HCMV infected cells and progressively localized more strikingly close to the nucleus 2 d.p.i. (Fig. 1B and D), a phenomenon that culminated after 3 days of infection when the AC was formed. At this late stage of infection, RhoB was recruited to the assembly site and remarkably localized at the AC along with the above viral AC markers, demonstrating that RhoB is associated in this viral compartment (Fig. 1C and E). RhoB cytoplasmic translocation to the AC was accompanied by increased RhoB protein levels, as determined by western blot analysis. RhoB gradual upregulation paralleled the time course of AC formation and RhoB protein levels reached maximum levels 4 d.p.i., when the AC was well-formed (Fig. 1F). A positive and a negative cellular AC marker, the trans-Golgi Network marker, TGN-38, and the early endosomal marker, Rab5, respectively, were also employed to further study the subcellular localization of RhoB in infected cells. RhoB clearly showed the same distribution pattern with the AC marker, TGN-38, in HCMV-infected cells (Fig. 1G upper panel). On the contrary, the early endosomal protein, Rab5, exhibited limited subcellular association with the AC-recruited RhoB, as Rab5 was rather scattered in the cytoplasm than localized in the AC, exclusively (Fig. 1G lower panel). The same pattern of subcellular association has been observed among several TGN markers, Rab5 and HCMV AC proteins^6,7,11,54 corroborating the specificity of RhoB localization in the assembly site of the virus.

Live cell imaging of AC formation and RhoB AC recruitment

The results on fixed cells indicating that a considerable proportion of RhoB progressively translocates to the AC, prompted us to study the dynamics of this phenomenon in live infected cells. HFF cells were electroporated with the retroviral vector mRFP-RhoB that expresses RhoB fused to the red fluorescent molecule, mRFP and infected with the recombinant virus UL32-EGFP-HCMV-TB40 expressing the capsid-associated tegument protein pUL32 fused to EGFP. Immunogold labeling of EGFP and subsequent electron microscopy using this recombinant virus has proven that the EGFP signal represents true virion particles and does not originate from EGFP polypeptides while the kinetics and the proper localization of pUL32 by newly synthesized HCMV particles during the late stages of infection using this recombinant virus have been successfully shown in HFF infected cells.⁵¹ However, upon electroporation of HFFs, we observed a slight delay regarding the timeframe of expression of viral proteins in infected cells. Time-lapse microscopy was carried out by sequential photography of infected fibroblasts cultures in an inverted fluorescence microscope. In uninfected cells transiently expressing mRFP-RhoB, the exogenous RhoB was evenly distributed throughout the cytoplasm, slightly more intense around the nucleus (Fig. 2A, a). Images captured from live cells infected with UL32-EGFP-HCMV-TB40 (MOI=5) showed that a subset of both RhoB and pUL32 protein were recruited at the AC, at late stages of the infection (Fig. 2A, b–d). The remainder fraction of pUL32 protein was observed in the nucleus (Fig. 2A, b) while both proteins were also observed all over the cytoplasm of the infected cells (Fig. 2A, b–d).

In HCMV-infected cells, RhoB exhibited a dynamic rearrangement throughout the course of the infection compared to uninfected cells. RhoB cytoplasmic accumulation was first more striking around and adjacent to the nucleus (Fig. 2B, a). The initiation of AC formation became evident by the appearance of a small green fluorescent focus representing the pUL32 viral protein and conceivably newly synthesized pUL32-containing virions, adjacent to the nucleus that progressively increased in size (Fig. 2B, a, see arrow and b–c). Concomitantly, the nascent AC was surrounded by a cytoplasmic accumulation of RhoB that progressively became more compact and shortly, a significant proportion of RhoB was localized along with the pUL32 at the HCMV AC (Fig. 2B, c–e). RhoB distribution at the AC followed the dynamic changes regarding the shape of the AC (Fig. S1A and B). RhoB cytoplasmic localization was initially more striking at the AC (Fig. 2B, c–e), but progressively accumulated at the periphery of the cells at late stages of the infection (Fig. 2B, b–e and Fig. S2). Interestingly, detailed microscopic analysis showed that after the formation of the AC, a proportion of exogenous RhoB seemed to leave the AC and it was found in the cytoplasmic area outside the AC, whereas a fraction of RhoB remained at the AC (Fig. 2C, arrows). EGFP foci representing pUL32-containing newly synthesized virions were also observed in the cytoplasmic area between the AC and the cell membrane at late stages of infection. Viral particle movements at this stage of the infection were extensively presented in a previous study⁵¹ and it would be of interest to correlate RhoB dynamic cytoplasmic distribution and the movement of virions toward cell periphery in a future study.

RhoB localization and actin organization in cellular projections at late stages of HCMV infection

Live cell microscopy experiments revealed a dynamic redistribution of RhoB during the course of infection. Indeed, monitoring single-infected cells for prolonged times illustrated that the initial bright perinuclear accumulations of RhoB were followed by its recruitment to the AC adjacent to the nucleus and the subsequent progressive distribution of these moving puncta through the cytoplasm, reaching at the periphery and strikingly accumulating at cellular projections, at later times of the infection.

Infected fibroblasts with the recombinant HCMV expressing pUL32-EGFP and transiently overexpressing mRFP-RhoB formed remarkable cellular projections at the final stages of infection which were often branched (Fig. 3A). Overexpressed mRFP-RhoB, apart from the AC, was also strikingly localized at the edges of these projections, while pUL32-containing HCMV particles were also found in these formations (Fig. 3A), although after detailed microscopic analysis the spatial association with RhoB was not complete. Live cell analysis showed that the protrusions in the infected cells were not formed due to passive retraction of the cells, but they were actively produced during infection. On the contrary, in live uninfected cells mRFP-RhoB was distributed evenly throughout the cytoplasm and slightly more intense around the nucleus (Fig. 2A, a). This pattern remained unaltered in mock infected cells even at incubation times longer than the timeframe of HCMV infection (Fig. S3), confirming that RhoB specifically translocates to the tips of cellular protrusions due to HCMV at late stages of the infection. The fact that HCMV infection causes accumulation of mRFP-RhoB at cellular projections in live cells was further verified with the wild-type HCMV (data not shown).

Similar localization of endogenous RhoB in cellular projections was observed in UL32-EGFP-HCMV-TB40-infected cells (Fig. 3B). These projections were also found to contain numerous pUL32-EGFP puncta representing HCMV capsids.⁵¹ It was observed that both endogenous RhoB and pUL32-EGFP-containing virions, occasionally accumulated in cellular projections (Fig. 3B, see arrows). In non-confluent cell cultures, these cell projections established intimate contacts with neighboring cells (Fig. 3C, enlargement).

Staining of similarly infected cells with rhodamine-labeled phalloidin, showed that compared to mock infected cells (Fig. 4a), they formed wide cellular projections that contained elongated actin stress fibers along with the releasing pUL32-EGFP virions (Fig. 4b–g). Detailed microscopic analysis revealed that the intense stress fibers were exclusively located at the stripe sides of each cellular projection, whereas the actin cortex was depleted at the projection tips and actin was locally reorganized at the egress sites, forming a blind edge (Fig. 4, enlargements e–g). Thus, the formation of the aforementioned actin-containing cellular projections combined with the dynamic translocation of RhoB and the capsid-associated tegument protein, pUL32, suggests a potential role in the process of enhancing the intercellular and/or intracellular viral spread, a phenomenon that would be very interesting to further study with more detailed experiments.

RhoB localizes at the AC in the absence of late viral proteins from the AC and affects their recruitment to the AC

The spatial-temporal localization of the RhoB GTPase in the HCMV AC raised the question whether RhoB attracts the viral proteins to the AC or whether the viral proteins forming the AC recruit RhoB to the AC. To address this issue, HFFs were either mock infected or infected with HCMV AD169 and either treated with the antiviral drug, ganciclovir (GCV), that inhibits viral DNA synthesis or with the drug solvent. Cells were fixed 4 d.p.i. and stained for the viral AC markers, pp65 or pUL32, as well as for RhoB (Fig. 5). Treatment of mock infected cells with ganciclovir did not affect the localization of endogenous RhoB (Fig. 5b, c, k, l). The immunofluorescence analysis illustrated that despite the failure of recruitment of the above viral proteins and presumably and other viral proteins to the AC and the lack of formation of a typical HCMV AC containing viral proteins in the GCV-treated cells, (Fig. 5g and p), RhoB retained its ability to translocate to the cytoplasmic area, in a spherical structure juxtaposed to the nucleus, apparently corresponding to the AC (Fig. 5h and q). Therefore, in infected cells RhoB translocates to the AC, regardless the presence or absence of the antiviral drug GCV and regardless the presence (Fig. 5, d–f and m–o) or absence (Fig. 5, g–i and p–r) of the viral AC proteins at the AC.

To address the reverse scenario, HFFs were transduced with the doxycycline-inducible lentiviral vector, TRIPZshRhoB that targets RhoB and subsequently infected with the UL32-EGFP-HCMV-TB40 virus. A doxycycline-inducible TRIPZshRNA lentiviral non-targeting (scrambled) vector (named TRIPZshscr) was also used, serving as a negative shRNA control vector. Although the shRNA cassette, inducibly expressed by the lentiviral TRIPZshRhoB vector, effectively downregulated RhoB expression as detected by western blotting in total cell extracts, immunofluorescence microscopy showed that individual cells could still express residual RhoB protein which was localized at the AC in the setting of infection (data not shown). Apparently, the remaining protein after the imperfect RhoB depletion is still capable of being recruited at this viral compartment. When RhoB was knocked down upon the doxycycline-induction of the shRNA expression, we could observe cells in which pUL32-EGFP did not localize at the AC or cells in which the pUL32-EGFP accumulation at the AC was affected 5 d.p.i. (Fig. 6A, upper panel), whereas at the same timeframe, non-induced (−Dox) cells (Fig. 6A, middle panel) or cells expressing the shscr control vector (Fig. 6A, lower panel) exhibited a typical AC, confirming the specificity of the effect in RhoB-silenced cells. That was also the case when similarly lentiviral transduced cells as described above, were stained for pp65 (Fig. 6B) or pUL97 (Fig. 6C), 5 days after HCMV AD169 infection. The above observations provide evidence that RhoB is not only associated with the AC, but even more, it is also an important constituent, contributing to the formation of the viral AC.

Depletion of RhoB results in reduction of progeny virion production

To explore the overall effect of RhoB on the viral yield, we used the above described shRNA approach. HFFs were first transduced either with the doxycycline-inducible lentiviral vector, TRIPZshRhoB, in order to knockdown RhoB or control vectors [empty vector (EV) or shRNA scramble vector (shscr) in the presence or absence of Dox] followed by HCMV AD169 infection. Supernatants from all HCMV-infected cells were harvested 5 days after infection and titrated on fresh HFFs by IE1 fluorescence, combining 2 previously described approaches.^55,56 The percentage of the IE1-positive cells was statistically significantly reduced in the fresh HFFs infected with supernatants of the RhoB-depleted cells, compared to the ones infected with the supernatants of the control non-induced TRIPZshRhoB fibroblasts (Fig. 7A and B). A similar statistically significant decrease in the number of IE1-expressing cells derived from the RhoB-silenced cells was evident when they were compared to the IE1-positive cells that corresponded to the EV or shscr-transduced HCMV-infected cells, either in the presence or absence of doxycycline (Fig. 7B). Therefore, the knockdown of RhoB results in a profound decrease in the viral yield, suggesting a key role of RhoB in HCMV production.

Reduction of the viral yield may be due to a block in virus entry provoked by the knockdown of RhoB. To address this issue, HFFs that had been transduced with the shRNA scramble or with the TRIPZshRhoB vector in the presence or absence of doxycycline, were infected with wild-type HCMV and the cell-associated virus was assessed 6 h.p.i. Real-time PCR analysis revealed that the number of the viral genomes that had infected the cells was comparable, irrespective of RhoB expression levels (Fig. 7C), confirming the efficient viral entry in the RhoB-depleted cells.

Decreased progeny virus was not due to an inhibition in viral protein synthesis either, as indicated by a western blot analysis of the viral proteins, IE1, pUL44 and pUL97. Regardless of the expression levels of endogenous RhoB (normal or knockdown), the expression of the above viral proteins, representing the immediate-early, early and late viral gene expression, respectively, was not affected (Fig. 7D).

Actin reorganization at the AC in HCMV-infected cells

RhoB has been reported to regulate actin fiber structure.^57,58 Normal fibroblasts have a rather impressive actin filament network. Exploring the effect of RhoB depletion on the cytoskeleton by transducing HFFs with the lentiviral vector, TRIPZshRhoB, we observed a robust modification in the alignment of stress fibers, from the parallel-running actin fibers in the normal fibroblasts (Fig. 4a) to a random orientation of the fibers where a subset of bundles acquired a vertical direction to the main axis in the RhoB-depleted cells (Fig. S4A). The organization of actin cytoskeleton remained unaffected in cells transduced with the TRIPZshscr control vector (Fig. S4B), confirming that the pronounced change in the organization of actin filaments is specifically due to the genetic silencing of RhoB.

In contrast to the fine structure of actin cytoskeleton in mock infected human fibroblasts (Fig. 4a), shortly upon HCMV infection, actin fibers are disrupted and this has been reported to enhance the translocation of the incoming viral particles to the host nucleus and hence it is to the benefit of the virus.^44,46,59,60 However, the fate of actin cytoskeleton at late stages of the infection has not been thoroughly studied so far. To this end, HFFs were infected with wild-type HCMV and co-stained for the cytoplasmic protein, RhoB, as well as for actin using rhodamine-phalloidin. Additionally, the large replication compartments formed at the nucleus containing pUL44, served as markers of advanced viral infection. The destructive effect of the virus on the actin cytoskeleton was still apparent 2 days after infection, when the infected cells showed less stress fiber formation compared to their mock-infected counterparts (Fig. 8). However, a progressive reconstitution of actin filaments initiating 3 d.p.i. and sustaining for several days was evident, predominantly at the site of the AC. In particular, newly synthesized actin filaments were formed, initially relatively short and progressively longer and branched, remarkably originating from a juxtanuclear location corresponding to the core of the viral AC. Thus, the formation of the AC seems to induce the reorganization of the actin network within the host cell probably facilitating HCMV assembly and likely the subsequent steps of virion migration and/or egress.

Disassembly of actin fibers does not disperse the AC but reduces the viral yield

Taking into account that RhoB can affect actin structure in fibroblasts and also that actin filaments are reorganized at the AC of infected cells and at cellular projections at late times of infection, we tested whether the reduction in the viral yield upon RhoB knock down may be associated with alterations in actin fibers organization induced by RhoB depletion. HFFs were infected with HCMV AD169 and on day 5 p.i. the cell culture medium was removed, fresh medium was added and the cells were treated for 1 hour with Cytochalasin B, a potent disruptor of actin filaments. Immunofluorescence analysis showed that despite the compromised actin cytoskeleton after the Cytochalasin B treatment, the viral AC was not dispersed, as indicated by the recruitment of either RhoB or pp65 in this characteristic compartment (Fig. 9A, b, c and e). In contrast, the destruction of the microtubule network by nocodazol for 1 hour severely affected the RhoB localization at the AC (Fig. 9A, f), confirming that the integrity of the microtubule network is critical for the maintenance of the AC⁵⁰ and that RhoB is a component of this viral compartment.

Titration of supernatants harvested from the above Cytochalasin B-treated cells was performed by IE1 fluorescence, as descibed above and showed a statistically significant reduction of virus titer compared to the non-treated cells (Fig. 9B), suggesting that although the gross structural integrity of the AC was likely retained, in terms of several of its components, its functionality might have been affected. Reduction of viral titer upon Cytochalasin B treatment implies that actin cytoskeleton could play a role during viral assembly and/or egress. Other mechanisms that remain to be elucidated in future studies may also contribute.

Cells, viruses and small molecules

Human foreskin fibroblasts (HFFs) and HEK 293T cells were cultured in Dulbecco's modified Eagle's medium with high glucose (Biosera, UK) supplemented with 10% fetal bovine serum, 100 U/ml penicillin and 100 μg/ml streptomycin. The wild-type HCMV laboratory strain, AD169, as well as the recombinant TB40 strain, UL32-EGFP-HCMV-TB40⁵¹ (kindly provided by C. Sinzger) were employed. Both viruses were propagated in HFFs and isolated by centrifugation from the supernatants when fibroblasts exhibited excessive cytopathic effects according to standard protocols. Viral pellet was resuspended in cell culture medium and virus aliquots were stored at −80°C until use. Virus titer of viral stocks was assessed by a standard plaque assay method performed on the HFFs. For viral infections, the cells were infected with HCMV at the indicated MOI, were incubated for 2 hours and then the inoculum was removed and replaced by fresh medium. Moreover, in order to compare the viral yield in different experimental conditions, supernatants from HCMV-infected cells were harvested when indicated and titrated on fresh HFFs by IE1 fluorescence, combining 2 previously described approaches.^55,56 Briefly, equal volumes of each supernatant were used to infect fresh HFFs grown on coverslips, and after 24 hours incubation, they were stained for IE1. Cell nuclei were counterstained using DAPI. The percentage of infected cells (IE1-positive cells) was calculated and statistical analysis was performed to compare the viral titers.

Ganciclovir (GCV), purchased from Sigma Chemicals (St. Louis, MO), was dissolved in distilled water. Stock solutions (10mM) were stored at −80°C, while HCMV-infected HFFs were treated with GCV at a final concentration of 30 μM. Cytochalasin B was purchased from Sigma Aldrich (Deisenhofen, Germany) and dissolved in dimethyl sulfoxide at 2 mg/ml. Prior to use, it was added to the culture medium to yield a final concentration of 2 μg/ml. The microtubule-depolymerizing agent, nocodazole (Sigma Aldrich, Deisenhofen, Germany), was also diluted in DMSO and used at a final concentration of 2 μM. HCMV-infected cells were treated with either Cytochalasin B or nocodazole for 1 hour in fresh medium, 5 d.p.i. and subsequently the cells were fixed and stained.

Transient transfection

The retroviral vector, mRFP-ENDO (mRFP-RhoB), expressesing RhoB fused to the red fluorescent molecule, mRFP, was kindly provided by L. Opresco (Pacific Northwest National Lab) and was used for transfection experiments. For live cell imaging, the HFFs were transfected with the mRFP-RhoB vector using the Amaxa Nucleofector kit for primary mammalian fibroblasts according to the manufacturer's instructions (Amaxa Biosystems, Cologne, Germany). Cells were electroporated in suspension, plated in chambers with coverslip quality glass bottoms (Lab-Tek; Nunc) and infected with the recombinant virus, UL32-EGFP-HCMV-ΤΒ40, the following day.

Live cell microscopy

Time-lapse microscopy was performed as previously described.⁸² Briefly, the cells were seeded into 2-well, chambered coverglass units with coverslip quality glass bottoms (Lab-Tek; Nunc) at a density of 2 × 10⁵ cells per well. Live cells were observed under a Leica DMIRE2 inverted fluorescence microscope, equipped with a Leica DFC300 FX digital camera, using a ×63 dry objective lens. To maintain cell viability for prolonged times of the experiments, the microscope stage was enclosed within a controlled environment of constant temperature, CO₂ and humidity. The excitation wavelength was controlled by mercury lamp illumination and a manual filter wheel equipped with filters suitable for EGFP, the monomeric red fluorescent protein and DAPI. The camera image acquisition was controlled by IM50 software (Leica). Particular care was taken in order to avoid overlap between channels by collecting the data from each channel sequentially and routing the emission signals through appropriate band pass filters. Single images or timed image sequences were exported as TIFF files from the IM50 software.

Lentiviral production and transduction

The doxycycline-inducible TRIPZ lentiviral shRNA vector targeting RhoB, TRIPZshRhoB, (Clone ID: V2THS_172671) as well as the TRIPZshRNA lentiviral non-targeting (scrambled) vector, TRIPZshscr, (Clone ID: RHS4743) were purchased from Thermo Scientific-Dharmacon. In these vectors, turboRFP and shRNA are part of a single transcript, allowing the visual marking of shRNA-expressing cells. HEK 293T cells were co-transfected with the vectors TRIPZshRNA (expressing either shRhoB or scrambled shRNA), pCMV/VSV-G (where VSV-G is vesicular stomatitis virus protein G) and pCMV-dR8.2 dvpr. Transfection was carried out using Turbofect transfection reagent (Thermo Scientific). The pLKO.1 lentiviral empty vector (EV) or the pLKO.1scr vector, expressing a scrambled shRNA served as a negative controls. To transduce the HFFs with the packaged viruses, early passage cells were incubated with the lentiviral supernatants in the presence of 5 µg/ml polybrene (Sigma-Aldrich, Deisenhofen, Germany) for 24 hours. Forty-eight hours later, cells were selected with puromycin (2μg/ml). Transduced cells were treated with doxycycline (500ng/ml), when required, to induce shRNA expression.

Antibodies/immunofluorescence/Western Blot analysis

The HCMV immediate-early IE1 and early pUL44 gene products were detected using monoclonal antibodies against IE1 (BS500) and pUL44 (BS510), respectively, as previously described.⁸³ The viral tegument protein pUL32 was detected using the mouse monoclonal antibody (MAb) XP1,⁵¹ the UL97 gene product was detected using the mouse polyclonal antibody #1343,⁸⁴ while the UL99-encoded pp28 mouse monoclonal antibody (10B4-29) was also used (kind gift of T. Shenk, Princeton University).⁸⁵ Antibodies against HCMV pp65 (sc-71229), RhoB (sc-180), TGN38 (sc-271624), Rab5 (sc-46692) and β-actin (sc-8432) serving as a loading control in western blot analysis, were purchased from Santa Cruz (Santa Cruz, CA). Rhodamine (phalloidin R415) and fluorescein phalloidin (F432) were purchased from Molecular Probes (Eugene, OR); a monoclonal antibody against tubulin (DM1A) was purchased from Sigma.

Anti-mouse, as well as anti-rabbit horseradish peroxidase-conjugated secondary antibodies were obtained from Sigma while Alexa 488 and Cy3-conjugated secondary antibodies were purchased from Molecular Probes.

For indirect immunofluorescence experiments, the HFFs were seeded on 13-mm diameter glass coverslips and infected with either HCMV AD169 or with UL32-EGFP-HCMV-TB40 at a confluency of at least 70%. Cells were fixed and stained on the day indicated for each experiment. The conditions for fixation and immunodetection of viral and cellular proteins, were as previously described.⁸⁶ For microtubule staining, cells were fixed in 4% paraformaldehyde in cytoskeleton buffer (1.1 M Na2HPO4, 0.4 M KH2PO4, 137 mM NaCl, 5 mM KCl, 2 mM MgCl2, 2 mM EGTA, 5mM PIPES, 5 mM Glucose, pH 6.1) for 5 min at room temperature.

For Western blotting, extracts from infected or lentiviral transduced cells were prepared in sodium dodecyl sulfate loading buffer, separated on sodium dodecyl sulfate-containing 8–10% polyacrylamide gels, and transferred to nitrocellulose membranes. Western blotting and chemiluminescence detection were performed as previously described.⁸⁷

Real-time PCR

The Nanogen Q-CMV Real-Time PCR kit (Nanogen Advanced Diagnostics,Italy) was used to quantify the HCMV genomes in combination with the ABI 7500 Fast system (ABI). The primers for HCMV are specific for the exon 4 region of the HCMV MIEA gene (major immediate early antigen, HCMV UL123) while in parallel, a region of the human β globin gene is also amplified. The limit of detection of the assay is approximately 0.25log₁₀ copies/ml. All reactions were performed in triplicate, following the manufacturer's instructions.