Rotavirus Infection Induces the Phosphorylation of eIF2{alpha} but Prevents the Formation of Stress Granules

Early during the infection process, rotavirus causes the shutoffof cell protein synthesis, with the nonstructural viral proteinNSP3 playing a vital role in the phenomenon. In this work, wehave found that the translation initiation factor 2 ${alpha}$ (eIF2 ${alpha}$ ) ininfected cells becomes phosphorylated early after virus infectionand remains in this state throughout the virus replication cycle,leading to a further inhibition of cell protein synthesis. Underthese restrictive conditions, however, the viral proteins andsome cellular proteins are efficiently translated. The phosphorylationof eIF2 ${alpha}$ was shown to depend on the synthesis of three viralproteins, VP2, NSP2, and NSP5, since in cells in which the expressionof any of these three proteins was knocked down by RNA interference,the translation factor was not phosphorylated. The modificationof this factor is, however, not needed for the replication ofthe virus, since mutant cells that produce a nonphosphorylatableeIF2 ${alpha}$ sustained virus replication as efficiently as wild-typecells. In uninfected cells, the phosphorylation of eIF2 ${alpha}$ inducesthe formation of stress granules, aggregates of stalled translationcomplexes that prevent the translation of mRNAs. In rotavirus-infectedcells, even though eIF2 ${alpha}$ is phosphorylated these granules arenot formed, suggesting that the virus prevents the assemblyof these structures to allow the translation of its mRNAs. Underthese conditions, some of the cellular proteins that form partof these structures were found to change their intracellularlocalization, with some of them having dramatic changes, likethe poly(A) binding protein, which relocates from the cytoplasmto the nucleus in infected cells, a relocation that dependson the viral protein NSP3.

INTRODUCTION

Top

INTRODUCTION

While every step of the translation process is amenable to regulation,under most circumstances mRNA translation primarily is regulatedat the level of initiation (8). The translation of eukaryoticmRNAs involves the recognition and recruitment of mRNAs by thetranslation initiation machinery and the assembly of the 80Sribosome on the mRNA; this process is mediated by the eukaryoticinitiation factors (eIFs). Translation initiation is a complexprocess that begins with the recognition of the cap nucleotidestructure (m7GpppN) at the 5' end of mRNAs by the cap bindingprotein eIF4E, which is part of the cap binding complex eIF4F.This complex is composed of eIF4E, eIF4A (which is an ATP-dependentRNA helicase), and the scaffolding protein eIF4G. eIF4F functionsas a cap-dependent RNA helicase that promotes the associationof the mRNA with the 40S ribosomal subunit. The binding of Met-tRNAto the 40S ribosomal subunit is mediated by a ternary complexcomposed of eIF2-GTP-Met-tRNA (15). The release of eIFs is assistedby eIF5, which facilitates the hydrolysis of GTP carried outby eIF2. The GDP on eIF2 is exchanged for GTP by eIF2B in aregulated manner that is essential for ensuing rounds of initiation(29).

Many types of stresses reduce global translation by triggeringthe phosphorylation of the ${alpha}$ subunit of eIF2 (eIF2 ${alpha}$ ) at residueSer51. This phosphorylation inhibits the exchange of GDP forGTP catalyzed by eIF2B, which then is sequestered in a complexwith eIF2. Since the cellular level of eIF2B is 10 to 20 timeslower than the level of eIF2, even small changes in the phosphorylationof eIF2 ${alpha}$ have a drastic effect on protein translation. Four proteinkinases are known to phosphorylate Ser51 in eIF2 ${alpha}$ : the heme-regulatedinhibitor kinase; the protein kinase RNA (PKR), activated bydouble-stranded RNA (dsRNA); PKR-like endoplasmic reticulum(ER) kinase (PERK), which is activated in response to ER stress;and the general control nonderepressible-2 kinase (reviewedin reference 33). These kinases serve to arrest translationin various conditions that threaten cell survival, such as viralinfection, nutrient deprivation, and misfolded proteins.

A recently characterized phenomenon associated with the arrestin protein synthesis is the formation of stress granules (SGs),which are cytoplasmic aggregates of stalled translational preinitiationcomplexes that accumulate during stress. It has been proposedthat SGs are sites of mRNA triage at which mRNAs are monitoredfor integrity and composition and then are routed to sites ofreinitiation, degradation, or storage (2, 3, 6, 18, 23). MessengerRNAs within SGs are not degraded, making them available forrapid reinitiation in cells that recover from stress. The relatedRNA binding proteins TIA-1 and TIAR are robust markers of thesecytoplasmic foci (19); these proteins are present in the nucleusof most cells, and in response to a given stress they move tothe cytoplasm, where they rapidly aggregate to form the SGs(reviewed in references 2 and 3). The SGs are formed by small,but not large, ribosomal subunits (18, 20), translation factorslike eIF3, eIF4G, and eIF4E, the poly(A) binding protein (PABP),and proteins that regulate mRNA function, such as HuR, G3BP,and SMN, among others (13, 16, 18, 19, 39).

Rotaviruses are the leading etiologic agents of severe diarrhealdisease in infants and young children, being responsible foran estimated 600,000 annual deaths globally, which places asignificant economic burden on the global health care system(28). These viruses have a genome composed of 11 segments ofdsRNA enclosed in a capsid formed by three concentric layersof protein. During or shortly after cell entry, the infectingvirus uncoats, losing the two surface proteins and yieldinga double-layered particle that is transcriptionally active.The viral mRNAs contain 5'-methylated cap structures but lackthe poly(A) tail characteristic of most cellular mRNAs. Instead,rotavirus mRNAs have at their 3' end a consensus sequence (UGACC)that is conserved in all 11 viral genes (31). The viral transcriptsdirect the synthesis of six structural (VP1 to VP4, VP6, andVP7) and six nonstructural (NSP1 to NSP6) proteins (10). Oncea critical mass of viral proteins are synthesized, they accumulateinto discrete, large cytoplasmic inclusions, termed viroplasms,in which the replication of the virus genome and the assemblyof transcriptionally active progeny double-layered particlesare believed to take place (11, 37). The nonstructural proteinsNSP2 and NSP5 have been shown to be essential for viroplasmformation, such that in the absence of either one of them noviroplasms are formed and the viral replication cycle is blocked(22, 37).

Early in the infection process rotavirus takes over the hosttranslation machinery, causing a shutoff of cell protein synthesis.In this work, we report that the translation initiation factoreIF2 ${alpha}$ becomes phosphorylated early after virus infection andstays phosphorylated throughout the replicative cycle of thevirus. Under these conditions the viral proteins are efficientlytranslated, while most of the cell protein synthesis is arrested.We also found that in infected cells, SGs are not formed, althoughsome of the proteins known to form part of these structureschange their intracellular localizations.

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

Cells, viruses, and antibodies. The rhesus monkey epithelial cell line MA104 was grown in advancedEagle's minimal essential medium (MEM) (Invitrogen) supplementedwith 2% fetal bovine serum (FBS). Wild-type (wt) mouse embryonicfibroblasts (MEFs) and MEFs expressing S51A mutant eIF2 ${alpha}$ (S51A)were obtained from N. Sonenberg, McGill University, Montreal,Canada. These cells were grown in high-glucose Dulbecco's modifiedEagle medium (DMEM) supplemented with 10% heat-inactivated FBSand 1x nonessential amino acids. Rhesus rotavirus strain RRVwas obtained from H. B. Greenberg, Stanford University, Stanford,CA, and was propagated in MA104 cells as described previously(27). Rotavirus lysates were activated with trypsin (10 µg/ml;GibcoBRL) for 30 min at 37°C. Antibodies used include mousemonoclonal antibodies against TIA-1 (generously provided byPaul Anderson, Boston, MA); VP2 (MAb 3A8), provided by H. B.Greenberg, Stanford University, Stanford, CA; PABP and eIF4E(Santa Cruz Biotechnology, Santa Cruz, CA); and ribosomal proteinS6 (Cell Signaling). Rabbit polyclonal sera to NSP2, NSP3, andNSP5 have been described previously (14); polyclonal antibodiesto purified RRV triple-layered particles were produced in rabbitsas described previously (21), and rabbit anti-phosphorylated-eIF2 ${alpha}$ (anti-phospho-eIF2 ${alpha}$ ) was purchased from Cell Signaling. The secondaryantibodies used were goat anti-mouse coupled to Alexa 568 or488 and goat anti-rabbit coupled to Alexa 488 or 568 (MolecularProbes, Eugene, OR).

siRNA transfection. Duplex short interfering RNAs (siRNAs) were obtained from DharmaconResearch (Lafayette, CO), and the target sequences used foreach gene are shown in Table 1. The transfection of siRNAs intoMA104 cells was performed as previously described (21). Briefly,the siRNAs were transfected using Lipofectamine (Invitrogen).The transfection mixture was added to confluent cell monolayersand incubated for 8 h at 37°C. After this incubation, themixture was removed and the cells were kept in MEM with 0.7%FBS for 48 h at 37°C prior to virus infection.

View this table:
[in this window]
[in a new window]

TABLE 1. Sequence of the siRNAs derived from the RRV genome used in this work

Radiolabeling of proteins. For protein labeling, cells grown in 48-well plates were transfectedwith the indicated siRNAs and infected with rotavirus strainRRV at a multiplicity of infection (MOI) of 3. At the indicatedtimes, the medium was replaced by MEM without methionine, supplementedwith 25 µCi/ml of Easy Tag Express-[³⁵S] labeling mix(Dupont NEN), and incubated for different periods of time at37°C, as indicated. The samples were resolved by sodiumdodecyl sulfate-10% polyacrylamide gel electrophoresis (SDS-10%PAGE), adjusting the amount of protein loaded in each lane byCoomassie blue staining, and then examined by autoradiography.

Immunoblot analysis. Cells were lysed for 15 min at 4°C in lysis buffer containing25 mM NaF, 1 mM sodium orthovanadate, 50 mM Tris (pH 7.4), 100mM KCl, 1 mM EDTA, 1 mM dithiothreitol, 10% glycerol, and 1%Triton X-100, supplemented with a protease inhibitor cocktail(Complete, EDTA-free; Roche). The lysates were centrifuged for5 min at 10,000 x g, and the supernatants were collected. Sampleswere diluted in Laemmli sample buffer, denatured by boilingfor 5 min, subjected to SDS-10% PAGE, and transferred to ImmobilonNC (Millipore) membranes. The membranes were blocked with phosphate-bufferedsaline (PBS) containing 5% nonfat dry milk and incubated withthe indicated antibodies. Polyclonal antibodies to eIF2 ${alpha}$ andphospho-eIF2 ${alpha}$ (Ser51) were used as recommended by the manufacturer(Cell Signaling), and viral structural proteins were detectedusing antibodies against triple-layered particles or using monospecificantibodies against the nonstructural proteins. The bound antibodieswere developed by incubation with a peroxidase-labeled secondaryantibody and the Western Lightning system (Perkin Elmer).

Immunofluorescence. Cells on coverslips were fixed with 2% paraformaldehyde for20 min at room temperature, and the coverslips were washed fourtimes with PBS with 50 mM ammonium chloride. The fixed cellswere permeabilized by incubation in 0.5% Triton X-100 in PBSwith 50 mM ammonium chloride and 1% bovine serum albumin (BSA)and then were blocked by incubation with 1% BSA, 50 mM NH₄Clin PBS at 4°C overnight. The coverslips were incubated withprimary antibodies for 1 h at room temperature, followed byan incubation with the corresponding secondary antibody for1 h at room temperature. Coverslips were mounted on glass slidesusing Fluoprep (BioMérieux). The slides were analyzedwith an E600 epifluorescence microscope coupled to a DXM1200digital still camera (Nikon).

RESULTS

Top

RESULTS

Rotavirus infection induces the phosphorylation of eIF2 ${alpha}$ . To establish whether rotavirus infection triggers the phosphorylationof eIF2 ${alpha}$ , we determined the phosphorylation state of this translationfactor in infected cells by immunoblot analysis using antibodiesspecific for either total eIF2 ${alpha}$ or the phosphorylated form ofthis protein. As a positive control, we used drug treatmentsknown to induce cellular stress, like thapsigargin, which inducesER stress by depleting ER Ca²⁺ stores and activates PERK, andsodium arsenite, which induces oxidative stress and promotesthe activation of heme-regulated inhibitor kinase (not shown)(17, 24, 32). Cells either were left uninfected or infectedwith rotavirus strain RRV, and 5 h postinfection (hpi) the cellswere treated with thapsigargin for 30 min and then metabolicallylabeled for 30 min (maintaining the drug during the labelingperiod). The pattern of viral proteins was analyzed by SDS-PAGEand autoradiography, and the total and phosphorylated formsof eIF2 ${alpha}$ were detected by Western blotting (Fig. 1). In mock-infectedcells, thapsigargin caused a severe shutoff of cellular proteinsynthesis that correlated with the highly phosphorylated stateof eIF2 ${alpha}$ in these cells (Fig. 1). In cells infected with rotaviruseither in the presence or absence of the drug, eIF2 ${alpha}$ became phosphorylatedas much as it did in uninfected thapsigargin-treated cells.Under these conditions, the synthesis of viral proteins wasvery efficient (Fig. 1). Interestingly, in infected cells theinhibition of the cell protein synthesis was somewhat lessened(Fig. 1).

View larger version (72K):
[in this window]
[in a new window]

FIG. 1. eIF2 ${alpha}$ is phosphorylated in rotavirus-infected cells. MA104 cells were mock infected or infected with rotavirus strain RRV at an MOI of 3. Five hours postinfection, the cells were left unstressed or were stressed with 400 nm of thapsigargin (Thapsi) for 30 min, and the cells then were metabolically labeled with 25 µCi/ml Easy Tag Express-[³⁵S] for 30 min (maintaining the drug during the labeling period). The cells were lysed, and equivalent amounts of total protein were subjected to SDS-10% PAGE and detected by autoradiography (top panels) or transferred to nitrocellulose and probed with the indicated specific antisera (eIF2 ${alpha}$ , total eIF2 ${alpha}$ sera; eIF2 ${alpha}$ ^P, phosphorylation-specific eIF2 ${alpha}$ sera) (bottom panels). The asterisk indicates the migration of the unglycosylated form of NSP4 that appeared when the cells were treated with thapsigargin.

eIF2 ${alpha}$ remains phosphorylated throughout the rotavirus replication cycle. In general, cellular stress causes transient changes in thephosphorylation status of eIF2 ${alpha}$ . Amino acid starvation and ERstress cause a transient phosphorylation of eIF2 ${alpha}$ during thefirst hour of treatment, whereas dsRNA causes a sustained phosphorylationof this initiator factor (12). To determine if this was thecase for rotavirus infection, we analyzed by Western blottingthe protein synthesis pattern and the phosphorylation stateof eIF2 ${alpha}$ at different times postinfection; as a control, cellswere treated with thapsigargin for different times. Using thephosphospecific antibody, we found that when the cells werestressed with thapsigargin, there was a rapid transient increasein phospho-eIF2 ${alpha}$ that was maximal at 1 h posttreatment (hpt)and returned to basal levels 2 h later (3 hpt); the synthesisof cellular proteins was severely inhibited while the translationfactor was phosphorylated, and it was restored when the factorwas no longer phosphorylated (Fig. 2A). In contrast, in cellsinfected with rotavirus, the phosphorylation of eIF2 ${alpha}$ was apparentby 2 hpi, and it was sustained up to 10 hpi (Fig. 2B). Nevertheless,the synthesis of viral proteins was very efficient under theseconditions, indicating that the viral mRNAs, as well as someselected cellular mRNAs, can be translated even though eIF2 ${alpha}$ is phosphorylated.

View larger version (82K):
[in this window]
[in a new window]

FIG. 2. Phosphorylation of eIF2 ${alpha}$ is maintained during the replicative cycle of rotavirus. (A) MA104 cells were treated with 400 nM of thapsigargin (Thapsi) for 1 h and then were harvested at the indicated times posttreatment (hpt); 30 min before being harvested, the cells were metabolically labeled with 25 µCi/ml of Easy Tag Express-[³⁵S] (maintaining the drug during the labeling period). (B) Cells were infected (virus) or mock infected (M), and 30 min before the indicated times postinfection (hpi) the cells were metabolically labeled with 25 µCi/ml of Easy Tag Express-[³⁵S]. The labeled proteins were resolved by SDS-10% PAGE and detected by autoradiography. The same samples were analyzed by Western blotting using antibodies against phospho-eIF2 ${alpha}$ (eIF2 ${alpha}$ ^P) or total eIF2 ${alpha}$ (eIF2 ${alpha}$ ). The ratio of total eIF2 ${alpha}$ to phospho-eIF2 ${alpha}$ (eIF2 ${alpha}$ /eIF2 ${alpha}$ ^P) is shown at the bottom of the Western blots.

eIF2 ${alpha}$ is not phosphorylated when VP2, NSP2, and NSP5 are knocked down. To determine if individual rotaviral proteins were involvedin inducing/maintaining the phosphorylation status of eIF2 ${alpha}$ ,we silenced the expression of each rotavirus gene by RNA interference(RNAi). In addition to the phosphorylation of eIF2 ${alpha}$ , we alsoevaluated the effect of knocking down each protein on the synthesisof viral and cellular proteins. The cells were transfected withchemically synthesized siRNAs that target each one of the 11RRV genes or with an siRNA complementary to the firefly luciferasegene as an irrelevant control. The effectiveness of each siRNAinitially was confirmed by Western blotting or semiquantitativereverse transcription-PCR (7, 21, 22 and data not shown). Inthese assays, MA104 cells were transfected with each siRNA,infected with RRV 48 h posttransfection, and metabolically labeledfor 30 min at 5.5 hpi. The synthesis of viral and cellular proteinswas assessed by SDS-PAGE and autoradiography (Fig. 3). In thegel shown in Fig. 3, with the exception of NSP1 and NSP5, whichare not detectable under the PAGE conditions used, the knockdownof every viral protein was apparent when the corresponding siRNAwas used. In general, silencing the expression of the VP1, VP3,VP4, VP7, NSP1, NSP3, and NSP4 genes did not significantly affectthe amount of total viral protein synthesized, whereas the synthesisof cellular proteins was shut off, except when NSP3 was silenced,since this protein is involved in the inhibition of the cellularprotein synthesis (26, 30). In contrast, silencing the expressionof the VP2, VP6, NSP2, and NSP5 genes caused a general reductionin the synthesis of the viral proteins, which correlated inall cases with an increased synthesis of cellular proteins.

View larger version (68K):
[in this window]
[in a new window]

FIG. 3. VP2, NSP2, and NSP5 are involved in eIF2 ${alpha}$ phosphorylation. Cells were transfected with the indicated siRNA as described in Materials and Methods. (A) Forty-eight hours posttransfection, the cells were mock infected (M) or infected with RRV at an MOI of 3, and 5.5 hpi they were radiolabeled for 30 min with 25 µCi/ml of Easy Tag Express-[³⁵S] and then lysed. Mock-infected cells were not transfected. The labeled proteins were resolved by SDS-10% PAGE and detected by autoradiography (A) or by immunoblot analysis of phospho-eIF2 ${alpha}$ (eIF2 ${alpha}$ ^P) or total eIF2 ${alpha}$ (eIF2 ${alpha}$ ) (B). Dots at the left of each lane indicate the knocked-down protein. Irr, irrelevant (control) siRNA.

By Western blot analysis we found that eIF2 ${alpha}$ was phosphorylatedin RRV-infected cells transfected with all siRNAs, with theexception of those cells that received the siRNAs targetingthe synthesis of VP2, NSP2, and NSP5, in which the phosphorylationof the translation factor was abolished (Fig. 3B). In the absenceof VP6, eIF2 ${alpha}$ was phosphorylated, although to a lower level thanthat found in control cells transfected with the irrelevantsiRNA. These results suggest that these proteins are directlyor indirectly involved in inducing or maintaining the phosphorylationof eIF2 ${alpha}$ . Since the phosphorylation status of eIF2 ${alpha}$ at a giventime is the result of the equilibrium between the activitiesof the kinase(s)/phosphatase(s) involved, the phosphorylationstate of eIF2 ${alpha}$ was determined at different times postinfectionin cells in which either VP2, NSP2, or NSP5 was silenced. ByWestern blotting we found that this factor was not phosphorylatedbetween 2 and 7 hpi in cells transfected with the siRNA to NSP2or NSP5 (Fig. 4A). However, in cells in which VP2 was knockeddown, the phosphorylation of eIF2 ${alpha}$ followed the kinetics observedin control cells transfected with an irrelevant siRNA for upto 4 hpi, and after that time the equilibrium was clearly displacedin favor of the unphosphorylated state of the factor (Fig. 4A).These results indicate that the mechanism through which VP2favors the phosphorylation state of eIF2 ${alpha}$ is different from thatemployed by NSP2 and NSP5. Figure 4B shows the Western blotanalysis of the presence of these viral proteins to corroboratethe effectiveness of the siRNA-induced knockdowns.

View larger version (21K):
[in this window]
[in a new window]

FIG. 4. Time course of eIF2 ${alpha}$ phosphorylation in VP2, NSP2, or NSP5 knockdown. Cells were transfected with the indicated siRNAs or with an siRNA control (Irr). (A) At 48 hpt, the cells were infected or left uninfected (M) and harvested at the indicated times postinfection (hpi). (B) Cell extracts were analyzed by Western blotting using antibodies against phospho-eIF2 ${alpha}$ (eIF2 ${alpha}$ ^P) or against VP2, NSP2, or NSP5. (C) The number of viroplasms decreases in VP2 knockdowns. Cells transfected with the siRNA to VP2 or with an irrelevant siRNA (Irr) were infected 48 hpt with RRV at an MOI of 3, and at the indicated times postinfection the cells were fixed and immunostained with a polyclonal rabbit antibody to NSP5 as the primary antibody, followed by incubation with goat anti-rabbit immunoglobulin G coupled to Alexa 568 as the secondary antibody. The average numbers of viroplasms per cell (± standard deviations) are shown; 100 cells per condition were counted.

The number of viroplasms decreases throughout the infection process when VP2 is knocked down. NSP2 and NSP5 are required and sufficient for the formationof viroplasms (11). Accordingly, silencing the expression ofeither of these two proteins prevented the formation of theseviral structures (22, 37, and data not shown). On the otherhand, silencing the expression of VP2 results in the formationof fewer viroplasms (see below). Based on this observation andon the fact that in the absence of VP2 the phosphorylation ofeIF2 ${alpha}$ is not blocked until 5 hpi, we determined if there wasa correlation between the phosphorylation status of eIF2 ${alpha}$ andthe formation of viroplasms. For this, cells were transfectedwith the siRNA to VP2 or an irrelevant control; 48 h posttransfectionthe cells were infected with strain RRV, and at the indicatedtimes postinfection the cells were fixed and immunostained withan anti-NSP5 antibody and the number of viroplasms per cellwas counted (Fig. 4C). Interestingly, in cells transfected withthe VP2 siRNA, the number of viroplasms decreased as the infectionproceeded compared to that of cells transfected with irrelevantsiRNA, and this decrement in the number of viroplasms correlatedwith the time at which eIF2 ${alpha}$ appeared to be dephosphorylated(Fig. 4A), suggesting that the phosphorylation of eIF2 ${alpha}$ was relatedto viroplasm formation.

Phosphorylation of eIF2 ${alpha}$ is required for neither viroplasm formation nor rotavirus replication. To establish whether eIF2 ${alpha}$ phosphorylation was required for theformation of viroplasms and ultimately for the replication ofthe virus, we took advantage of a well-characterized MEF cellline that produces a mutant form of eIF2 ${alpha}$ in which serine 51was changed to a nonphosphorylatable alanine (S51A) (35). Mutantor wt MEFs were infected with rotavirus, and at 5.5 hpi theywere metabolically labeled with ³⁵S for 30 min, the synthesisof cellular and viral proteins was analyzed by autoradiography,and the phosphorylation of eIF2 ${alpha}$ was detected by Western blotting(Fig. 5A). As expected, upon rotavirus infection eIF2 ${alpha}$ was phosphorylatedin the wt-infected MEFs but not in the mutant S51A MEFs. Theviral proteins, however, were efficiently synthesized in bothwt and mutant cells, while the shutdown of cellular proteinsynthesis was somewhat less pronounced in the mutant MEFs (Fig.5A). Of interest, RRV replicated better in the mutant MEFs thanin wt cells (the titer was 1.8 ± 0.2 times higher formutant MEFs), as judged from the viral progeny produced in thesecells. Also, when the formation of viroplasms was characterizedfor both wt and mutant MEFs, we found that these viral structuresformed in both cell lines, although in the mutant cells theyappeared to be less abundant (Fig. 5C). Taken together, theseresults indicate that, at least in MEFs, the phosphorylationof eIF2 ${alpha}$ is required for neither the synthesis of rotavirus proteinsnor the formation of viroplasms and production of infectiousviral progeny. Our results also suggest that, in addition tothe known role of rotavirus NSP3 (26, 30, 31), the phosphorylationstatus of eIF2 ${alpha}$ also plays a role in the severe shutoff of cellularprotein synthesis observed during rotavirus infection. To determinethe relative contribution of these two factors to the inhibitionof cellular mRNA translation, we silenced the expression ofNSP3 in wt and S51A mutant MEFs and evaluated the synthesisof viral and cellular proteins by PAGE and autoradiography (Fig.5A). When NSP3 was knocked down in wt MEFs, the synthesis ofcellular proteins was partially recovered, while its restorationwas complete in NSP3-silenced mutant MEFs. These results indicatethat both NSP3 and the phosphorylation of eIF2 ${alpha}$ contribute tothe severe shutoff of cell protein synthesis observed upon rotavirusinfection. Interestingly, the synthesis of viral protein appearednot to be affected by the absence of NSP3 in MEFs, as we havealready reported for MA104 cells (26).

View larger version (50K):
[in this window]
[in a new window]

FIG. 5. Phosphorylation of eIF2 ${alpha}$ is not necessary for rotavirus replication. wt and mutant (S51A) MEFs were infected with rotavirus strain RRV at an MOI of 1 (+) or were mock infected (–), and 5.5 hpi the cells were metabolically labeled with 25 µCi/ml Easy Tag Express-[³⁵S] for 30 min. Cells were lysed, and equivalent amounts of total protein were subjected to SDS-10% PAGE and autoradiography (top gel). siIrr, irrelevant (control) siRNA. siNSP3, siRNA directed to NSP3. (B) The same samples were transferred to nitrocellulose and analyzed by immunoblotting using antibodies against total eIF2 ${alpha}$ (eIF2 ${alpha}$ ) or phospho-eIF2 ${alpha}$ (eIF2 ${alpha}$ ^P) or directed to NSP3 (NSP3). (C) wt and mutant (S51A) MEFs were infected with rotavirus strain RRV at an MOI of 3, and 8 hpi the cells were fixed and immunostained with a polyclonal rabbit antibody to NSP5 as the primary antibody, followed by incubation with goat-anti-rabbit immunoglobulin G coupled to Alexa 568 as the secondary antibody (in green). The cell nuclei were stained with 4',6'-diamidino-2-phenylindole (in blue).

Rotavirus infection prevents the formation of SGs and causes the relocalization of some cellular proteins. It has been described that when eIF2 ${alpha}$ is phosphorylated, stalled48S preinitiation complexes are sequestered into punctate cytoplasmicaggregates known as SGs (2, 3, 19, 20). Since eIF2 ${alpha}$ becomes phosphorylatedduring rotavirus infection, we looked for SG formation in infectedcells. Some of the cellular proteins that have been found tobe present in SGs are TIA-1, eIF4E, PABP, and the ribosomalprotein S6 (18-20); thus, we used antibodies to these proteinsto immunostain MA104 cells infected with rotavirus or cellsthat had been subjected to stress by the addition of sodiumarsenite to detect bona fide SGs (19). In uninfected, nontreatedMA104 cells, PABP and the ribosomal protein S6 have a diffusecytoplasmic distribution that slightly concentrates in the perinuclearregion, while eIF4E has a cytoplasmic distribution that concentratesin punctate structures; the TIA-1 protein preferentially localizedto the nuclei (Fig. 6A), as has been reported for other celllines (19). When the cells were treated with sodium arseniteas a positive control, a fraction of PABP, S6, eIF4E, and TIA-1appeared in SGs (Fig. 6A), indicating that these structuresare indeed formed in MA104 cells; treatment of cells with thapsigarginor heat shock also resulted in the formation of these structures(not shown). To determine if, during rotavirus infection, SGsare formed, the cells were infected and analyzed by immunofluorescence6 hpi; infected cells were detected with an antibody to NSP2,a nonstructural viral protein that stains viroplasms. We foundthat the TIA-1 protein exited the nucleus of rotavirus-infectedcells, but it did not localize to coarse granular structureslike those observed in arsenite-treated cells; rather, it localizedas a fine punctate signal distributed homogeneously throughoutthe cytoplasm. In the case of eIF4E, its punctate cytoplasmiclocalization was still evident, although the number of spotswas reduced compared to that of control noninfected cells, andmost of the protein now was uniformly distributed in the cytoplasm.The distribution of ribosomal protein S6 became more uniformin infected cells, losing its apparent concentration aroundthe nuclei, and surprisingly, the distribution of PABP changeddramatically, from the cytoplasm to the nuclei of the cells(Fig. 6). These results suggest that upon rotavirus infection,the formation of SGs is prevented and at least some of the proteinstypical of SGs are redistributed in infected cells.

View larger version (42K):
[in this window]
[in a new window]

FIG. 6. SGs are not formed during rotavirus infection. MA104 cells were (A) untreated (– virus) or treated with 500 µM sodium arsenite for 1 h (Ars), (B) infected with rotavirus RRV at an MOI of 3 for 6 h (+ virus), or (C) infected at an MOI of 0.5 and treated with arsenite for 1 h at 5 hpi. Cells were fixed and processed for immunofluorescence using the indicated antibodies, as described in Materials and Methods. ${alpha}$ -NSP2, antibody to NSP2; ${alpha}$ -TIA-1, antibody to TIA-1; ${alpha}$ -S6, antibody to S6; ${alpha}$ -eIF4E, antibody to eIF4E; DAPI, 4',6'-diamidino-2-phenylindole.

To determine if the infection of rotavirus indeed prevents theformation of SGs, a monolayer of MA104 cells was infected withrotavirus strain RRV at an MOI of 0.5, such that only a fractionof the cells in the monolayer were infected, and 6 hpi the cellswere treated with sodium arsenite to induce the formation ofSGs. By immunofluorescence we found that the cells that wereinfected, as evidenced by the presence of viroplasms, did nothave SGs, and TIA-1 appeared diffuse in the cytoplasm, whileSGs were present only in uninfected cells (not stained withthe anti-NSP2 serum) (Fig. 6C). The same results were obtainedwhen the cells were stained with antibodies directed to eIF4Eand the ribosomal protein S6 (data not shown). These resultsconfirm that the formation of SGs is prevented in rotavirus-infectedcells, despite the fact that eIF2 ${alpha}$ is mostly phosphorylated underthese conditions.

Relocalization of PABP to the nuclei in infected cells depends on the time course of infection and the presence of NSP3. To determine which factors were involved in the translocationof PABP to the nucleus of the cell, we did a time course infectionin which the cells were infected with RRV for different periodsof time and then fixed and immunostained for PABP and NSP2.Figure 7 shows that PABP kept its cytoplasmic localization upto 3 hpi, and after that time it appeared to relocalize to thecell's nuclei, such that by 7 hpi most cells had a relocalizedPABP. To determine the viral product(s) involved in the redistributionof PABP, we silenced the expression of each viral gene and analyzedthe cells at 6 hpi with antibodies to PABP and to NSP2 (notshown). The relocalization of PABP was not prevented by silencingthe expression of any rotavirus gene with the exception of thegene encoding NSP3, for which the translocation of PABP to thenucleus was completely abolished (Fig. 8). Interestingly, therelocalization of PABP seems to be independent of the phosphorylationof eIF2 ${alpha}$ , since it localized in the nucleus of infected cellseven when VP2 (not shown), NSP2, or NSP5 (Fig. 8) was silenced.These results also indicate that the relocalization of PABPto the nucleus requires only a limited amount of NSP3, sincethe amount of NSP3 and of the other viral proteins synthesizedin the cells is very low when VP2, NSP2, and particularly NSP5were silenced.

View larger version (35K):
[in this window]
[in a new window]

FIG. 7. PABP relocalizes to the nucleus of the cells infected with rotavirus. MA104 cells were infected at an MOI of 3, and at the indicated times postinfection the cells were fixed and processed for immunofluorescence using anti-PABP and anti-NSP2 antibodies ( ${alpha}$ -PABP and ${alpha}$ -NSP2, respectively), as indicated.

View larger version (53K):
[in this window]
[in a new window]

FIG. 8. Translocation of PABP to the nucleus depends on the presence of NSP3 and not on the phosphorylation of eIF2 ${alpha}$ . Cells were transfected with the indicated siRNAs, and 48 hpt the cells were infected (+ virus) or mock infected (– virus) with RRV at an MOI of 3. Six hours postinfection the cells were fixed and processed for immunofluorescence, as described in Materials and Methods, using the indicated anti-PABP and anti-NSP2 antibodies ( ${alpha}$ -PABP and ${alpha}$ -NSP2, respectively). Irr, irrelevant (control) siRNA.

DISCUSSION

Top

DISCUSSION

As obligate intracellular parasites, viruses depend on the celltranslation machinery for the production of their proteins.Thus, the replication of a virus requires that viral mRNAs competesuccessfully with cellular mRNAs for the host translation apparatus.Viruses have developed remarkable strategies to ensure the efficienttranslation of their mRNAs while simultaneously inhibiting cellularprotein synthesis. The two main regulatory steps for the controlof polypeptide chain initiation are the activity of eIF2 andthe formation of the eIF4F complex, both of which have beenshown to be targets of control by viruses (reviewed in references4 and 36). In this work, we found that during rotavirus infectionthe initiation factor eIF2 ${alpha}$ becomes phosphorylated and staysin this state throughout the replication cycle of the virus.However, under these conditions the synthesis of viral, andsome cellular, proteins was not inhibited. Moreover, when rotavirus-infectedcells were treated with thapsigargin, which in uninfected cellsinduces the phosphorylation of eIF2 ${alpha}$ and temporarily completelyabolishes the synthesis of cellular proteins (Fig. 1), the synthesisof viral and cellular polypeptides was comparable to that observedin cells infected in the absence of the drug (Fig. 1). Thus,the activity of eIF2 that is lost upon phosphorylation of the ${alpha}$ subunit of the factor is somehow compensated for or no longerrequired by infected cells. It also is clear that the mechanismthat the virus uses to translate its mRNAs in the presence ofphospho-eIF2 ${alpha}$ also is used by some cellular mRNAs. The totalcellular protein synthesis is nonetheless limited in infectedcells, most probably due to the activity of the nonstructuralvirus protein NSP3, which is known to prevent the efficienttranslation of poly(A)-containing mRNAs (26, 30). Experimentsare under way in our laboratory to characterize the cellularmRNAs that are translated under the restrictive conditions imposedby the virus infection.

It has been shown recently that the phosphorylation of eIF2 ${alpha}$ ,as well as the inhibition of translation factors that preventthe 40S subunit and its associated factors from forming a competent80S particle, results in the formation of SGs (2, 3, 6, 18,23). Of interest, we found that these structures did not formin rotavirus-infected cells, even though eIF2 ${alpha}$ was phosphorylated.A likely scenario to explain these results is that upon rotavirusinfection the host cell triggers the phosphorylation of eIF2 ${alpha}$ to arrest protein synthesis and prevent the translation of viralmRNAs, with the ensuing formation of SGs, aggregates of stalledtranslation complexes that prevent the translation of mRNAs;the formation of these structures during rotavirus infectionwould inhibit the translation of the viral mRNAs. We hypothesizethat as a response to this host response, rotavirus has developedcounterattack measures, one of which could be to prevent theformation of SGs to allow the efficient translation of the viralproteins. This possibility is supported by the observation thatthere is more cellular protein synthesis in cells infected withrotavirus, either treated with drugs known to cause cellularstress or left untreated, than when uninfected cells were treatedwith these drugs, suggesting that the window that the virusopens to translate its mRNAs also is useful for some cellularmRNAs. The mechanism by which the virus prevents the formationof SGs in the cells is not known, but its characterization wouldallow us to learn about the cellular signals that trigger theassembly and disassembly of these structures.

The kinase responsible for the phosphorylation of eIF2 ${alpha}$ in rotavirus-infectedcells is not known, although two obvious candidates are PKR,induced by the presence of dsRNA, or PERK, which can be inducedby the accumulation of viral proteins in the ER (41). When theexpression of each viral gene was silenced through RNAi, wefound that eIF2 ${alpha}$ was not phosphorylated in the absence of VP2,NSP2, or NSP5, conditions under which there was little synthesisof viral proteins. It could be argued that the trigger thatinduces the phosphorylation of eIF2 ${alpha}$ is the accumulation of viralprotein; however, the fact that in VP2-silenced cells eIF2 ${alpha}$ becamephosphorylated during the first 4 hpi, when the synthesis ofviral proteins is greatly inhibited, suggests that the mechanismof phosphorylation is not, or at least is not only, relatedto the accumulation of viral proteins. Although the input genomicdsRNA also could be a triggering factor, it was the same inall silencing experiments and, thus, cannot explain the differencesobserved. Furthermore, the amount of viral RNA transcribed andreplicated when each rotavirus gene was silenced does not correlatewith the phosphorylation status of eIF2 ${alpha}$ (C. Ayala-Breton, H.Arias, C. F. Arias, and S. Lopez, unpublished data). Thus, theexplanation for the lack of phosphorylation of eIF2 ${alpha}$ in the absenceof VP2, NSP2, or NSP5 is not simple, and it needs to be furtherinvestigated.

We also found in this work that, during rotavirus infection,several cellular proteins known to accumulate in SGs upon stresschange their intracellular location even though they do notaggregate in these structures. The most drastic change observedwas that of PABP, which usually is a cytoplasmic protein that,under stress conditions, accumulates in SGs, but in rotavirus-infectedcells it relocalizes to the nucleus, which was found to be dependenton the presence of NSP3. It has been reported previously thatPABP shuttles between the nucleus and the cytoplasm in cellsin which either the transcription of cellular mRNAs was inhibitedor the protein was overexpressed, and it was proposed that theexport pathway of this protein is saturable (1). Based on ourobservations, an alternative explanation is that there is apool of cytoplasmic PABP that is retained through its interactionwith eIF4G; when this interaction does not occur, either becauseit is outcompeted by NSP3 or there is an excess of protein dueto its overexpression (1), the unbound, excess cytoplasmic poolof PABP is shuttled back to the nucleus. Experiments are beingcarried out in our laboratory to test this hypothesis.

The observation that the synthesis of viral protein and theyield of viral progeny is not affected and might be even higherin MEFs that cannot phosphorylate eIF2 ${alpha}$ compared to those ofwt MEFs suggests that the phosphorylation of eIF2 ${alpha}$ is not requiredfor rotavirus replication, similar to what has been reportedrecently for the mouse hepatitis coronavirus (34) and in contrastto what has been observed for reovirus infection, which hasbeen shown to induce a cellular stress response as well as thephosphorylation of eIF2 ${alpha}$ , which favors the replication of theseviruses (38). Other viruses known to induce the shutoff of cellprotein synthesis and the phosphorylation of eIF2 ${alpha}$ are SemlikiForest virus (25), Sindbis virus (40), and herpes simplex virustype 1 (9), but in these cases, as well as in the case of murinehepatitis virus and reovirus but unlike the case of rotavirus,there is a robust formation of SGs in the infected cells. Thus,the mechanism by which different viruses deal with the hostresponse to fight the virus invasion is diverse, and its characterizationshould allow us to learn about the signals that trigger theformation of SGs and the host stress response induced by a virusinfection.

ACKNOWLEDGMENTS

We are grateful to Rafaela Espinosa for her technical assistancewith cell culture, to Nahum Sonenberg and Mauro Costa-Mattiolyfor their help in acquiring the S57A MEFs, and to Paul Gaytanand Eugenio Lopez for their support with the synthesis of oligonucleotides.

This work was supported by grant 55005515 from the Howard HughesMedical Institute. H.M. is a recipient of a scholarship fromDGEP/UNAM and CONACYT.

FOOTNOTES

* Corresponding author. Mailing address: Instituto de Biotecnología, UNAM, Avenida Universidad 2001, Colonia Chamilpa, Cuernavaca, Morelos 62210, Mexico. Phone: (52) (777) 3291615. Fax: (52) (777) 3172388. E-mail: susana{at}ibt.unam.mx

Published ahead of print on 21 November 2007.

REFERENCES

Top