STAT2 Acts as a Host Range Determinant for Species-Specific Paramyxovirus Interferon Antagonism and Simian Virus 5 Replication

Cells and viruses.

NIH 3T3 cells (gift from Stuart Aaronson, Mt. Sinai) and STAT1-deficient U3A cells (16, 18) were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% Cosmic calf serum (HyClone). STAT2-expressing NIH 3T3 clones were engineered by transfection of NIH 3T3 cells with pEFHA-STAT2 (22) and a vector encoding puromycin resistance (pBABE-puro [17]) and were selected in 2 μg of puromycin/ml, and clones were screened by STAT2 immunoblotting (C20; Santa Cruz Biotechnology).

SV5 (strain W3A, derived from a genetically defined recombinant virus system [10, 14]) and HPIV2 (Greer strain) were propagated and titered in simian CV1 cells. Plaque assays were performed on CV1 cells by using an overlay containing 0.5% agar with DMEM and 10 mM HEPES (pH 7.2). Cells were fixed at 4 to 6 days postinfection (p.i.) with 3.7% formaldehyde, and plaques were visualized after being stained with 0.1% crystal violet in 20% ethanol as described elsewhere (22).

IFN-α/β-responsive transcription assays.

For luciferase assays, cells were transfected with Superfect reagent (Qiagen) according to the manufacturer's method with a LacZ plasmid as a control for transfection efficiency, a reporter gene, and either an empty vector or cDNA expression plasmids. The IFN-α/β-responsive reporter gene contained five copies of the ISG54 ISRE upstream of the TATA box and the firefly luciferase gene open reading frame. After 24 h, the transfection medium was replaced with fresh medium or medium supplemented with 1,000 U of recombinant human IFN-α (Hoffman LaRoche)/ml or 1,000 U of recombinant murine IFN-β (PBL Biomedical)/ml. Cells were harvested 6 to 12 h later in luciferase assay lysis buffer, and luciferase activity was measured according to the manufacturer's protocol (Promega). Values for luciferase activity were normalized to β-galactosidase activity. In all cases, average values of triplicate experiments are shown. Expression vectors for human STAT1, V proteins, human STAT2, and human STAT2 fragments have been described elsewhere (12, 21, 22). The murine STAT1 cDNA expression vector was provided by David Levy (New York University). The murine STAT2 cDNA expression vector was provided by Christian Schindler (Columbia University).

Infection, cell extraction, and immunoblotting.

For examination of STAT degradation, cells were infected with SV5 or HPIV2 at a multiplicity of infection (MOI) of 10 PFU/cell or mock infected and prepared for analysis at 16 h p.i. Whole-cell detergent extracts were prepared as described previously (21). Equal amounts of total protein (20 μg) were separated by sodium dodecyl sulfate-7% polyacrylamide gel electrophoresis (SDS-7% PAGE) and transferred to nitrocellulose filters. Immunoblotting was performed with antiserum specific for human STAT2 (C20; Santa Cruz Biotechnology), antiserum recognizing murine STAT1α and -β (E23; Santa Cruz Biotechnology), and antiserum specific for murine STAT2 (provided by Christian Schindler, Columbia University) (23), and filters were processed for chemiluminescence detection. For detection of viral nucleocapsid proteins, antiserum raised against HPIV2-infected cells that recognizes both HPIV2 and SV5 nucleocapsid proteins was used (Whittaker Biochemicals; gift from Anne Moscona, Mt. Sinai) (22).

For examination of SV5 growth, cells were washed with serum-free medium and inoculated with SV5 at the MOIs indicated in the legend for Fig. 4. After a 2-h adsorption period, cells were washed and medium was replaced with DMEM plus 2% serum and, in some cases, IFN-α/β. After being harvested, supernatants were clarified by centrifugation and filtration through a 0.2-μm-pore-size filter and adjusted to 0.75% bovine serum albumin, and virus titer was determined by plaque assays on monolayers of CV1 cells. For longer-term infections, assays were based on previous studies of SV5 persistence (28, 29). Cells were infected, and the inoculum was replaced at 2 h p.i. with low-serum medium with or without supplemental IFN-α/β. Virus was allowed to replicate for 48 h, and then the medium was harvested and replaced with low-serum medium with or without IFN-α/β. A second harvest was made 48 h later, and the chronically infected cell monolayer was washed and trypsinized for passage to larger dishes to allow for unhindered cell division in serum-containing medium lacking IFN-α/β. The supernatant of these cultures was harvested 7 days later (11 days p.i.).

Murine STAT1 is susceptible to SV5 V protein in human cells.

A simple explanation for the failure of SV5 to antagonize murine IFN-α/β responses is that the small number of sequence differences between mouse and human STAT1 proteins create an intrinsic defect in V protein-mediated recognition or destruction. To test the ability of the mouse STAT1 to function as a V protein target, STAT1-deficient human U3A cells were used as the host for an IFN-α/β response antagonism assay. Due to their STAT1 deficiency, U3A cells cannot induce transcription of a transfected ISRE-luciferase reporter gene in response to IFN-α/β stimulation, but complementation of the cells by transfection with a human STAT1 expression vector restores a robust IFN-α/β-stimulated luciferase response (Fig. 1). This response is completely eliminated by cotransfection with an SV5 V protein expression plasmid. When a murine STAT1 cDNA is transfected, similar complementation of IFN-α/β-stimulated reporter gene activity is observed. Coexpression of SV5 V with murine STAT1 completely suppressed the IFN-α/β response. These results clearly demonstrate that murine and human STAT1 proteins are equally susceptible to V-dependent IFN-α/β suppression, indicating that the two STAT1 orthologues are not intrinsically different with respect to SV5 sensitivity.

Human STAT2 overcomes barriers to IFN-α/β antagonism.

The successful antagonism by SV5 V of murine STAT1-dependent IFN-α/β responses in a human cell line suggests that mouse cells are inherently deficient in either a targeting or enzymatic component of the STAT degradation machinery. In conjunction with the observation that both STAT1 and STAT2 must be present in human cells to create a permissive environment for selective STAT1 degradation by SV5 (22), it seemed reasonable to hypothesize that sequence differences between human and murine STAT2 might account for the functional deficiency for SV5 IFN-α/β antagonism in mice. To test this hypothesis, we assessed human STAT2 for the ability to complement defective STAT degradation in an IFN-α/β-responsive murine system. NIH 3T3 fibroblasts respond well to stimulation with murine IFN-β: there is a robust stimulation of an ISRE-dependent luciferase reporter gene after 6 h of treatment with mouse IFN-β (Fig. 2A). This IFN-β-stimulated transcription is insensitive to expression of the SV5 V protein, consistent with earlier reports of species specificity (3, 28). To examine the generality of this effect for STAT-destroying paramyxoviruses, the HPIV2 V protein was also tested in the murine system. Expression of the HPIV2 V protein also failed to antagonize the murine IFN-α/β response in these cells, a result possibly related to the reported incomplete HPIV2 replication in mouse L929 cells (13). The ability of both viral V proteins to antagonize the murine IFN-α/β response changed dramatically upon expression of human STAT2 in the mouse cell transcription assay. Coexpression of either SV5 or HPIV2 V protein along with human STAT2 in the NIH 3T3 cells resulted in a potent inhibition of IFN-α/β signaling, completely eliminating ISRE-dependent luciferase activity (Fig. 2A). Control experiments with cells coexpressing either human or mouse STAT1 did not inhibit the IFN-α/β-responsive luciferase activity, but mouse STAT2 overexpression resulted in a slight V-dependent reduction of reporter gene activity (Fig. 2A). These results suggest that the STAT2 locus may be an important host range determinant for SV5, as mouse STAT2 is inefficient at mediating IFN-α/β inhibition, while expression of human STAT2 in mouse cells enables efficient V protein-dependent IFN-α/β antagonism.

Previous work demonstrated that defective STAT degradation in STAT2-deficient human cells could be rescued by expression of an amino-terminal STAT2 fragment encompassing amino acids 1 to 578 (22). The ability of human STAT2 fragments to confer V protein susceptibility to the murine cells was also tested. Expression of human STAT2 amino acids 1 to 578 was sufficient to render the murine cells susceptible to IFN-α/β antagonism by both paramyxovirus V proteins, but the isolated DNA binding domain and linker domain (amino acids 316 to 578) did not (Fig. 2B), confirming the important role of the STAT2 N terminus in the paramyxovirus-induced STAT degradation system. This result demonstrates that the crucial sequence differences underlying the species specificity of SV5 IFN-α/β suppression map to the same region of STAT2 that is absolutely required for STAT1 or STAT2 degradation by paramyxoviruses in human cells. The STAT2 fragment comprising amino acids 1 to 578 lacks domains required for IFN-α/β-responsive ISGF3 activation, tyrosine phosphorylation, and oligomerization, indicating that IFN-α/β signaling itself is not required for acquired IFN-α/β antagonism and affirming the mechanistic conclusions drawn from the study of human STAT degradation (22).

Expression of human STAT2 permits STAT degradation.

The transfection assays demonstrate that expression of human STAT2 in mouse cells transfers the ability of SV5 V to antagonize IFN-α/β signaling and suggest that STAT2 plays a role in restriction of the SV5 host range. To determine if human STAT2 expression enables virus-induced STAT protein degradation, an NIH 3T3 clone with stable expression of human STAT2 was isolated and subjected to infection with SV5 and HPIV2 (Fig. 3). Mock-infected- and infected-cell lysates were separated by SDS-PAGE and processed for immunoblotting. Probing with antiserum that recognizes both virus nucleocapsid proteins revealed that SV5 could synthesize viral proteins in the NIH 3T3 cells but HPIV2 could not (Fig. 3A). Importantly, infected-cell lysates from human STAT2-expressing cells revealed that expression of human STAT2 did not alter the level of SV5 protein synthesis (Fig. 3A). Probing with antiserum specific for murine STAT2 demonstrates that its level remained constant regardless of virus infection (Fig. 3B). Consistent with the overexpression results (Fig. 2B), a minor reduction of STAT1 was observed in SV5-infected NIH 3T3 cells after probing with STAT1-specific antiserum, but SV5 efficiently induced complete and specific degradation of STAT1 only in the cells expressing human STAT2 (Fig. 3C). The effect of SV5 infection was limited to STAT1, as the level of human STAT2 in the SV5-infected cells remained intact (Fig. 3D). Remarkably, this experiment reveals that, despite the attenuation of HPIV2, a partial loss of both murine STAT1 and human STAT2, but not of murine STAT2, is achieved by the HPIV2 inoculum when human STAT2 is present. The finding of HPIV2-induced STAT degradation in the absence of significant viral protein synthesis suggests that the virus-associated V protein enters the cell upon virus-cell membrane fusion and is active for inducing degradation of its target (Fig. 3D). This degradation is similar to results obtained with UV-inactivated SV5 (3, 28) and is consistent with the demonstration that V is a virus component present at approximately 350 molecules per virion (calculated for SV5 in reference 24). The HPIV2 inoculum also decreased the abundance of murine STAT1, again only in the cells expressing human STAT2 (Fig. 3C). This is in agreement with the promiscuous actions of HPIV2 observed in STAT2-complemented human STAT2-deficient U6A cells, where both STAT1 and STAT2 degradation was noted (22). No significant difference in the abundance of murine STAT2 was observed for HPIV2 infection. These results indicate that the efficient degradation of STAT proteins requires human STAT2 and that murine STAT2 is not targeted by either virus, even in a degradation-permissive cellular environment. Importantly, all degradation activities were observed only in cells expressing human STAT2, confirming that paramyxovirus-induced STAT degradation can utilize human STAT2 to proceed across species barriers.

Expression of human STAT2 permits efficient SV5 replication in mouse cells.

IFN-α/β antagonism and STAT degradation by SV5 were found to only occur in mouse cells if human STAT2 is expressed. The results suggest that STAT2 may function as a host factor important for the restriction of SV5 replication. To directly examine replication of SV5 in the presence or absence of human STAT2-dependent IFN-α/β antagonism, NIH 3T3 and NIH 3T3/human STAT2 cell lines were subjected to SV5 infection in the presence or absence of exogenous IFN-α/β (Fig. 4). To verify that expression of human STAT2 in the NIH 3T3 cells did not alter the intrinsic ability to mount an antiviral response induced by IFN-α/β, both lines were subjected to a 6 h of IFN-α/β pretreatment followed by infection with SV5 at a high MOI. Regardless of human STAT2 expression, IFN-α/β pretreatment resulted in a dramatic decrease in virus production in a 48-h growth period (Fig. 4A). This result demonstrates that expression of human STAT2 has neither positive nor negative effects on the ability of the mouse cells to mount an antiviral response.

To test the ability of SV5 to inhibit the endogenous IFN-α/β response in the NIH 3T3 cell lines, lower-multiplicity infections were used. Cells were infected with a range of SV5 concentrations, and virus replication was evaluated 48 h later. In all cases, SV5 infectious titers were higher in the presence of human STAT2, indicating that human STAT2 confers a growth advantage for SV5 in mouse cells (Fig. 4B). SV5 infection has been demonstrated to induce IFN-β biosynthesis, but the level of induction is low compared with that for Sendai virus (3). In addition, the endogenous IFN-α/β produced in the NIH 3T3 cell experiments is unlikely to attain optimal antiviral concentrations under the in vitro culture conditions in light of the medium changes required for washing and infecting cells with SV5. To assure high levels of IFN-α/β in our assays as a better imitation of a native infection, where high IFN-α/β concentrations accumulate locally, additional experiments were conducted with supplemental IFN-α/β added to the medium. The growth of SV5 was tested in the presence of IFN-α/β added exogenously at 2 h p.i. (1,000 U/ml, final concentration), coincident with replacement of the virus inoculum. Under these conditions, slightly lower titers were produced by all infections, but SV5 replication was dramatically improved in the presence of human STAT2, giving rise to ∼10-fold differences in SV5 recovery (Fig. 4C). This result indicates that the ability to block IFN-α/β signaling conferred by the presence of human STAT2 provides a clear advantage for SV5 replication in mouse cells.

SV5 growth in murine cells has been characterized as having an early phase, permissive for virus protein synthesis, followed by loss of virus replication by 3 to 4 days after infection. Experiments were conducted to evaluate longer-term effects of acquired IFN-α/β antagonism on SV5 replication using methodology developed for the study of SV5 persistence (28, 29). Cells were infected, and the inoculum was replaced at 2 h p.i. with low-serum medium with or without supplemental IFN-α/β. Virus was allowed to replicate for 48 h, and then the medium was harvested and replaced with low-serum medium with or without IFN-α/β. A second harvest was made 48 h later, and the chronically infected cell monolayer was washed and trypsinized for passage to larger dishes to allow for unhindered cell division in serum-containing medium lacking IFN-α/β. The supernatant of these cultures was harvested 7 days later (11 days p.i.). No significant differences in cellular cytopathic effects or plating efficiency were noted for any of the samples (data not shown). During the first 2 days following infection, only a minor SV5 replication difference between the two cell lines in the absence of exogenously added IFN-α/β was observed. In the presence of exogenous IFN-α/β, SV5 titers were reduced overall, consistent with data in Fig. 4C, but the presence of human STAT2 allowed SV5 to replicate five times more efficiently than in control NIH 3T3 cells under the same conditions (Fig. 4D). During the next 2 days, SV5 replicated much more efficiently in the human STAT2-containing cell lines, resulting in 100-fold increases in virus titers, regardless of IFN-α/β supplementation. Cells were passaged and grown for 7 more days without added IFN-α/β. While SV5 replication was observed in control NIH 3T3 cells (∼10³ PFU/ml), far more efficient replication was maintained for the human STAT2-containing cells regardless of prior IFN-α/β exposure, giving rise to a consistently 10-fold increase in virus yield. These results indicate that the acquisition of IFN-α/β suppression by SV5 in mouse cells containing human STAT2 results in higher viral titers, allowing for more productive replication in both acute- and chronic-infection scenarios.