Murine Cytomegalovirus Inhibits Interferon γ–induced Antigen Presentation to CD4 T Cells by Macrophages Via Regulation of Expression of Major Histocompatibility Complex Class II–associated Genes

4–12-wk-old 129Ev/Sv and IFNαβ receptor–deficient (IFNR-α/βR^−/−) mice (32) were housed in a Biosafety Level 2 facility at Washington University in accordance with all Federal and University policies. Sentinel animals were negative for adventitious mouse pathogens by serology. Virus stocks were grown and diluted in low endotoxin (<0.025 ng/ml) DME containing 10% FCS (DME 10%; reference 3). BMMφs were prepared as previously described (23) and were at least 95% F4/80 positive (data not shown). BMMφ cultures were mock infected or infected with MCMV or UV-inactivated MCMV, at a multiplicity of infection (MOI) of 5.0 unless otherwise stated, for 1 h at 37°C in 5% CO₂ with rocking every 5 min. After infection, cells were treated with IFN-γ (100 IU/ml; Genentech, San Francisco, CA) or medium alone for various periods of time while incubating at 37°C in 5% CO₂. Cells were harvested by scraping, then fixed, and analyzed for IA^b expression by flow cytometry. BMMφ viability was determined by trypan blue exclusion.

MCMV (American Type Culture Collection [ATCC] No. VR-194, Lot 10) was grown, inactivated by UV irradiation, and titered by plaque assay in BALB/ 3T12-3 fibroblasts (ATCC, CCL 164; reference 3). Recombinant MCMV expressing bacterial β-galactosidase (RM427; reference 25) was a gift of Dr. Edward Mocarski (Stanford University, Stanford, CA). 3T12 cells as well as MCMV stocks were negative for mycoplasma using the Mycoplasma TC test kit (Gen-Probe, San Diego, CA).

Flow cytometry was performed as previously described (3, 23) on an Epics flow cytometer (Coulter, Miami, FL) using XL analysis software (Coulter) or WinMDI 2.0 (Joseph Trotter, San Diego, CA). IA^b high and low BMMφ were separated by fluorescent activated cell sorting on a FACS^® Vantage fluorescent activated cell sorter (Becton Dickinson, San Jose, CA). β-galactosidase expression in sorted cells was detected by 5-bromo-4-chloro-3 indolyl-β-d-galactopyranoside staining (33).

Wild-type 129 BMMφs were either mock infected or infected with MCMV for 48 h. Supernatants were ultracentrifuged for 30 min at 100,000 g to remove free virus as confirmed by plaque assay, and then placed on naive cultures of wild-type or IFN-α/βR^−/− BMMφs at 2 ml/plate in 60-mm dishes (Sarstedt, Newton, NC) for 1 h. In additional groups, MCMV was added to the cultures at an MOI of 5.0. After 1 h of incubation, IFN-γ (100 IU/ml) was added and culture volumes were raised to 3 ml with fresh medium. Cultures were incubated at 37°C for 48 h and were assayed for IA^b expression.

BMMφs were either mock infected or infected with MCMV for 1 h at 37°C. For STAT1α activation, cells were harvested at 1, 24, and 48 h after infection. 10⁶ cells were suspended in endotoxin free PBS (Sigma Chemical Co., St. Louis, MO) containing 10% FCS, and were treated with IFN-γ at concentrations ranging from 0.5 to 1,000 IU/ml for 10 min. Nuclear extracts were prepared and assayed by electrophoretic mobility shift assay (EMSA) against a ³²P-labeled oligonucleotide probe derived from the IFN-γ activating sequence of the FcγRI promoter (34). Anti-STAT1 super-shift assays were performed by incubation of 2 μg of STAT1 (p91) specific antiserum (Santa Cruz Biotechnology, Santa Cruz, CA) with the nuclear extract/ probe mixture for 45 min at 25°C before acrylamide gel electrophoresis. For analysis of IRF-1 DNA-binding activity, BMMφ were removed from medium containing L cell conditional medium (23) for 15 h before infection with MCMV for 1 h followed by treatment with IFN-γ (100 IU/ml) for an additional 4 h. Nuclear extracts were generated from 10⁶ cells (34) and assayed for IRF-1 DNA binding activity by EMSA using a ³²P-labeled oligonucleotide containing the IFN response factor (IRF)-E of the murine (2′-5′) oligoadenylate synthase promoter (35). Super-shift assays for IRF-1 were performed by adding 2 μg of anti–murine IRF-1 polyclonal antisera (Santa Cruz Biotechnology) to mixtures of extract and probe and incubating at 25°C for 45 min before analysis.

Total cellular RNA was harvested from BMMφ cultures using RNAzol (Tel-Test Inc., Friendswood, TX) and analyzed by Northern blot hybridization (36). PCR-derived probes used for these studies included a 300-bp fragment of the murine IA^b beta chain (5′-ccggaattcgccagtgcctccagaggtgacagtgtatc; 3′-cgcggatccccatgccacagaaacaggtctcaggag), a rat β-actin fragment (5′-tatggagaagatttggcacc; 3′-gtccagacgcaggatggcat; gift of Dr. J. Milbrandt, Washington University) and a 226-bp fragment of mouse Ii p33 cDNA (5′-agctctgtacaccggtgtctctgtcc; 3′-gacattggacgcatcagcaagggag; gift of Dr. E. Unanue, Washington University, St. Louis, MO). The following cDNAs were also used to generate probes: murine IRF-1 (37), rat cyclophilin (38), murine CD45 (provided by Dr. M. Thomas, Washington University), TAP1 (39), TAP2 (40), H-2Ma, and H-2Mb (41) (the latter four gifts of Dr. J. Monaco, University of Cincinnati, Cincinnati, OH). All probes were radiolabeled using the Megaprime DNA Labeling System (Amersham Corp., Arlington Heights, IL). Loading of mRNA was normalized against 28S ribosomal RNA (42) or against cyclophilin or β-actin mRNA levels. Northern blot hybridizations were quantitated on a Phosphorimager (Molecular Dynamics, Sunnyvale, CA). Fold reduction of IFN-γ–induced mRNA levels by MCMV was computed as [(IFN-γ–stimulated signal in uninfected cells) − (unstimulated signal in uninfected cells)]/[(IFN-γ–stimulated signal in infected cells) − (unstimulated signal in infected cells)], after signal was normalized to 28S rRNA or cyclophilin levels.

BMMφs were mock infected or infected with MCMV and treated with IFN-γ (100 IU/ml) for 24 h. Cells were then harvested, counted, and incubated in 1% paraformeldahyde in RPMI 10% FCS for 15 min at 25°C. The fixed cells were washed vigorously, plated at 10³, 5 × 10³, 10⁴, or 5 x 10⁴ cells/well, and incubated with the T cell hybridoma B11 (IA^b-restricted, β-galactosidase peptide 429-441–specific, provided by Dr. Paul Allen, Washington University) in the presence of β-galactosidase peptide 429-441, control peptide, or medium alone for 24 h. Supernatants were then assayed for IL-2 by proliferation of the IL-2–dependent cell line CTLL2 as measured by [³H]thymidine incorporation (43). No IL-2 production above background was observed using a control peptide derived from the MCMV MCK-1 open reading frame (VVLVVSTVADLREPC; reference 44) at 10, 1, or 0.1 μM (data not shown).

We examined IFN-γ–stimulated presentation of peptide antigen to CD4 T cells by primary BMMφs after infection with MCMV. Primary cells were used because previous studies of MCMV regulation of MHC class I expression had revealed differences between cell lines and primary cells (15). Mφs were either mock infected or infected with MCMV, treated with IFN-γ for 24 h, fixed, and tested for the ability to present peptide antigen (Fig. 1,A). IFN-γ treatment of Mφs resulted in effective presentation of β-galactosidase peptide (429–441) to the T cell hybridoma B11, whereas MCMV infection abolished IFN-γ– induced antigen presentation. The antigen presenting cell, rather than the T cell hybridoma, was impaired by virus since (a) Mφs were fixed before incubation with T cells, and (b) addition of fixed MCMV infected Mφs did not efficiently inhibit the capacity of uninfected Mφs to present peptide antigen (Fig. 1,A). Cell surface IA^b levels were evaluated by flow cytometry, since decreased IA^b expression might explain defective antigen presentation. MCMV-infected Mφs failed to respond to IFN-γ stimulation by upregulation of cell surface IA^b (Fig. 1 B).

MCMV and HCMV infection at a low MOI impair IFN-γ–induced MHC class II expression via induction of IFN-α/β (20, 23). To evaluate MCMV's effects on Mφ responsiveness to IFN-γ independently of IFN-α/β, all experiments were performed at a high MOI in Mφs derived from mice carrying a null mutation in the IFN-α/β receptor (IFN-α/βR^−/−; reference 32). These Mφs do not respond to IFN-α/β stimulation, but do respond to stimulation with IFN-γ as measured by induction of IA^b expression (Fig. 1,B and Fig. 2; references 23, 32). Infection of IFN-α/βR^−/− Mφs with increasing doses of MCMV resulted in a progressive decrease in the percentage of cells expressing IFN-γ–induced IA^b (Fig. 2). Cell viability in the MCMV-infected Mφ cultures was >95% at 72 h after infection. Live virus was required for inhibition of IA^b expression, since UV-inactivated MCMV, even at a MOI of 6.0, had no effect on Mφ IA^b expression (Fig. 2).

MCMV-mediated inhibition of IFN-γ– induced IA^b expression was dependent on viral dose (Fig. 2), which suggested a role for direct infection. To address this possibility, IFN-α/βR^−/− Mφs were infected with RM427 (25), a recombinant MCMV expressing β-galactosidase under the MCMV ie1/ie2 promoter/enhancer at an MOI of 0.6, and treated with IFN-γ. Cells were fixed 48 h after infection, stained for IA^b expression, and separated into IA^b high and low populations by fluorescent activated cell sorting (Fig. 3,A). IA^b low Mφs were enriched for β-galatosidase positive (infected) cells (Fig. 3, A and B), whereas the IA^b high cells were only 0.5–2.5% positive for β-galactosidase. This demonstrated that IA^b expression is low in MCMV-infected BMMφs. However, over several experiments, β-galactosidase was detected in <100% of IA^b low cells, raising the possibility that a soluble mediator might contribute to low class II expression in β-galactosidase–negative cells.

To determine whether soluble factors played a role in MCMV's inhibition of IA^b induction in IFN-α/βR^−/− Mφs, supernatant transfer studies were performed. Supernatants from wild-type 129 Mφs infected with MCMV or mock infected for 48 h were ultracentrifuged to remove free virus and placed on naive Mφ cultures derived from either wild-type 129 or IFN-α/βR^−/− mice. IFN-γ was added to these secondary cultures, and after 48 h, IA^b expression was evaluated (Fig. 4). As expected from our previous work (23), supernatant from MCMV-infected Mφs inhibited IFN-γ–induced IA^b expression on wild-type Mφs, but not on IFN-α/βR^−/− Mφs (Fig. 4), confirming that MCMV-induced IFN-α/β can inhibit IFN-γ–induced MHC class II expression. However, there was no evidence for a soluble mediator affecting IA^b expression in MCMV-infected IFN-α/βR^−/− Mφ cultures (Fig. 4). Therefore, direct infection of the Mφ, rather than a soluble mediator, is likely to explain MCMV's effects on MHC IA^b expression.

IFN-γ induction of MHC class II expression occurs at the level of gene transcription (45), and studies in HCMV-infected endothelial cells have shown that HCMV infection regulates MHC class II mRNA levels (21). We therefore evaluated MCMV's effect on mRNA levels of an IFN-γ–induced MHC class II gene, IA^b. Infection with MCMV, but not UV-inactivated MCMV, decreased IFN-γ–induced accumulation of IA^b mRNA 6–8-fold at 24 h and 5–>10-fold at 48 h after infection (Fig. 5).

Although a 6–8-fold reduction of IFN-γ–induced IA^b mRNA accumulation within MCMV-infected cells explained a portion of the effects of MCMV infection on antigen presentation (Fig. 1,A), it did not plausibly account for the at least 30-fold loss of cell surface expression of IA^b (Figs. 1,B, 2, 3, 4). Therefore, we examined the effect of MCMV on expression of another IFN-γ–inducible gene, Ii, which is essential for efficient cell surface expression of MHC class II (46, 47). Northern blot hybridization showed that IFN-γ–induced Ii transcript levels are reduced 5–>10-fold at 24 h and >10-fold at 48 h in MCMV-infected BMMφ cultures. The combined effect of MCMV infection on IFN-γ–induced IA^b and Ii mRNA accumulation plausibly explains most or all of the decreased MHC class II expression on MCMV-infected Mφs.

To address the possibility that MCMV's reduction of IA^b and Ii mRNA levels was due to a general effect on host mRNA, levels of several constitutive host mRNAs were evaluated. MCMV infection for up to 48 h did not significantly decrease levels of β-actin (Fig. 5, A and B), cyclophilin (see Fig. 8), or CD45 (data not shown) mRNAs as measured against a 28S RNA loading control. In contrast, treatment of BMMφs with actinomycin D for 5–10 h revealed diminished β-actin, cyclophilin, and CD45 mRNA levels compared to untreated cells (data not shown), demonstrating that the half lives of these mRNAs were short enough for us to detect significant MCMV induced alteration in their stability. Thus, inhibition of IFN-γ–induced IA^b and Ii mRNA accumulation is not solely due to a general effect on mRNA levels by MCMV.

IFN-γ initiates a cascade of events that leads to MHC class II mRNA accumulation, any one of which could, a priori, be altered by MCMV infection. We examined the effect of MCMV on an essential early event in IFN-γ receptor signaling, activation and nuclear translocation of the latent cytoplasmic transcription factor STAT1α. Activation of STAT1α is essential for IFN-γ induction of MHC class II expression (48). Mφs were either mock infected or infected with MCMV for 1, 24, or 48 h before stimulation with IFN-γ, preparation of nuclear extracts, and evaluation of STAT1α binding activity by EMSA (Fig. 6). The presence of STAT1α in the activated complex was confirmed by super-shift with antibody against STAT1α (Fig. 6), whereas antibody to IRF-1 did not result in retardation of the complex (data not shown). IFN-γ treatment resulted in comparable levels of STAT1α DNA binding activity in both mock- and MCMV-infected Mφs, at 1, 24 (Fig. 6), and 48 (data not shown) h after infection. STAT1α activation was also equivalent in mock- and MCMV-infected Mφs after treatment with IFN-γ at doses as low as 0.5 IU/ml (data not shown). We conclude that the IFN-γ signaling pathway remains intact through STAT1α activation in MCMV-infected BMMφs.

To directly assess the effects of MCMV on the transactivating function of STAT1, we examined IFN-γ induction of IRF-1 mRNA and protein. IRF-1 is rapidly induced at the transcriptional level after IFN-γ stimulation in a STAT1α-dependent manner (48), with maximal mRNA expression by 2 h of cytokine treatment (49). Mφs were mock infected or infected with MCMV at an MOI of 5.0 for 3 h and treated with IFN-γ (100 IU/ml) for the final 2 h of infection. Northern blot analysis showed that induction of IRF-1 mRNA accumulation by IFN-γ in MCMV-infected Mφs was 70–90% that of uninfected cells (Fig. 7,A). To assess MCMV effects on expression of IRF-1 protein, we assayed nuclear extracts of Mφs that were mock or MCMV infected for a total of 5 h, with IFN-γ treatment during the final 4 h for binding to the IRF-E element of the murine (2′-5′) oligoadenylate synthase promoter (35) by EMSA. IFN-γ treatment induced IRF-1 DNA binding activity in MCMV-infected cells (Fig. 7 B) as confirmed by super-shift of the complex after incubation with anti–IRF-1, but not anti-STAT1α, antiserum. IFN-γ–induced IRF-1 DNA binding activity was decreased by at most 50% in MCMV-infected Mφs compared to mock-infected Mφs. IRF-1 DNA binding activity could also be induced by IFN-γ in Mφs that had been infected with MCMV for 24 h (data not shown).

To assess the specificity of the block of IFN-γ signaling caused by MCMV infection, we examined levels of several IFN-γ–induced transcripts in addition to IA^b, Ii, and IRF-1. IFN-α/βR^−/− Mφs were infected with MCMV or UV-inactivated virus, or mock infected for 24 or 48 h in the presence or absence of IFN-γ, and mRNA levels were analyzed by Northern blot hybridization. IFN-γ induction of H-2Ma and H-2Mb expression was significantly inhibited by MCMV at 24 h (Fig. 8), whereas IFN-γ–induced TAP1 mRNA levels were decreased to a lesser extent. However, TAP2 mRNA levels were not significantly decreased by MCMV infection (Fig. 8). Northern blot data from 48-h infections were similar to that of 24-h infections (data not shown). The lack of effect on IFN-γ induction of TAP2 transcript levels by MCMV demonstrated that MCMV selectively affects expression of a subset of IFN-γ– inducible genes.

CD4 T cells are required for clearance of salivary gland MCMV infection (11), yet infectious virus is found in the salivary gland for many weeks after acute infection, suggesting that MCMV antagonizes CD4 T cell function in some manner. It has also been shown that MCMV impairs CD4 T cell priming in vivo (19). These observations led us to examine the effects of MCMV on presentation of antigen to CD4 T cells and induction of genes associated with MHC class II cell surface expression.

MCMV inhibited IFN-γ induction of IA^b and Ii gene expression in IFN-α/βR^−/− BMMφs, resulting in defective antigen presentation to CD4 T cells. Inhibition of IA^b expression required direct infection of the Mφs with live virus. Early IFN-γ signaling events including STAT1α activation and IRF-1 induction remained largely intact in MCMV-infected cells. Further analysis showed that MCMV infection altered the induction of multiple genes encoding proteins important for MHC class II expression and/or presentation of peptide antigen to CD4 T cells, while another IFN-γ–inducible MHC gene was largely unaffected.

MCMV impaired the ability of IFN-γ to induce expression of mRNAs encoding multiple genes important for antigen presentation to CD4 T cells. These genes included the MHC class II allele IA^b, as well as the Ii p33 subunit and the peptide-loading protein subunits H-2Ma and H-2Mb. Although the 6–8-fold reduction of IFN-γ–induced IA^b mRNA levels observed in the setting of MCMV infection undoubtedly contributed to loss of cell surface expression of IA^b (e.g., Fig. 1 B), it was possible that additional factors accounted for the near complete inhibition of IA^b expression. Ii is required for maximal cell surface expression of MHC class II proteins (46, 47), and induction of Ii by IFN-γ was significantly inhibited in MCMV-infected cells. Decreased expression H-2Ma and H-2Mb might also contribute to impaired MHC class II peptide presentation, since these proteins are necessary for optimal loading of peptide onto MHC class II molecules (50–52). However, loss of H-2Ma and H-2Mb expression probably did not contribute to decreased detection of cell surface IA^b, as the antibody 25-9-17S detects IA^b loaded with the class II Ii-associated peptide, CLIP (50–52).

Three lymphocyte classes (CD4 and CD8 T cells, and NK cells) are important for resistance to CMV during acute infection. CD8 T cells promote MCMV clearance and mediate protective immunity (9), and are also important for control of HCMV (53). As a countermeasure, both MCMV and HCMV inhibit antigen presentation to CD8 T cells by limiting MHC class I protein–dependent antigen presentation (12–16). HCMV and MCMV prevent decreased expression of MHC class I on infected cells from triggering NK cells via expression of cell surface MHC class I homologs (17, 18). CD4 T cells and IFN-γ are important for clearance of virus from the salivary gland (1, 11), but priming of CD4 T cells is inhibited in MCMV-infected mice (19). The findings in this paper delineate a potential mechanism by which MCMV might impair CD4 T cell activation and avoid recognition by this third important class of lymphocytes.

We believe that viral impairment of IFN-γ–induced antigen presentation by Mφs in particular has important implications for acute and chronic MCMV pathogenesis. Mφs are involved in dissemination of MCMV and HCMV (25– 27) and yet they are activated by IFN-γ during acute MCMV infection (3, 23). IFN-γ inhibits MCMV growth in Mφs and other cells (3, 54, 55). How then does the virus disseminate in a cell activated by an anti-viral cytokine? We propose that signaling by IFN-γ is inhibited in Mφs if the virus is able to infect them and express protein before activation by IFN-γ. This would allow the virus to take advantage of the mobility of the Mφs for dissemination, while minimizing the effects of IFN-γ on viral growth and enhancement of immune induction. Our hypothesis is supported by the observation that Mφ-like cells that disseminate MCMV are MHC class II negative (25), whereas the majority of Mφs responding to MCMV in inflammatory exudates from immunocompetent mice express high levels of MHC class II and are uninfected (3, 23). Furthermore, we have preliminary data for a role of IFN-γ in controlling chronic MCMV infection (Presti, R., J. Pollock, and H.W. Virgin, unpublished data). Since one site of MCMV and HCMV latency is cells of the monocyte/Mφ lineage, interference with IFN-γ signaling in latently infected cells may permit viral escape from IFN-γ's antiviral effects, allowing reactivation and subsequent growth.

We found that MCMV infection inhibits expression of specific IFN-γ–induced genes. Other viruses such as adenovirus have also been shown to alter IFN-mediated gene induction. Adenovirus protein E1A may block IFN-α signaling by inhibition of STAT2-dependent transactivation involving p300/CBP (56). In our studies, MCMV inhibited IFN-γ signaling distal to STAT1α activation since induction of STAT1α DNA binding activity was normal in infected cells, and mRNA levels of IRF-1, a gene transactivated by STAT1α were also near normal early after IFN-γ treatment. Constitutive and inducible expression of MHC class II, H-2Ma, H-2Mb, and Ii are dependent on the class II transactivator (CIITA) (57–60). MCMV significantly inhibits induced expression of all of these genes. Therefore, one may speculate that a mechanism of viral action is the specific impairment of expression or function of CIITA.

Although there is a formal possibility that MCMV exerts its effect at the level of mRNA stability, we doubt this for two reasons. First, we saw little evidence of a general effect of MCMV on mRNA stability in the presence or absence of IFN-γ over multiple experiments based on comparisons of cyclophilin, β-actin, and CD45 mRNA levels. Second, the effects of MCMV are specific to certain MHC genes. Thus, if MCMV infection interferes with IFN-γ induction of MHC genes by destabilizing mRNAs, one would have to argue that the effects are specific for certain transcripts (e.g., IA^b and Ii but not TAP2, β-actin, or cyclophilin).

We have documented a novel mechanism of immune avoidance by CMV, selective blockade of IFN-γ induction of genes involved in antigen presentation to CD4 T cells. Further characterization of MCMV's block on IFN-γ signaling, and identification of the viral genes involved will likely enhance understanding of CMV pathogenesis and the antiviral functions of the immune system.

This work was supported by a grant to H.W. Virgin IV from the National Institute of Allergy and Infectious Diseases (RO1 AI-39616). H.W. Virgin was additionally supported by the Monsanto/Searle Biomedical Agreement. M. Connick was supported by National Institutes of Health (NIH) training grant AI-01763. M.T. Heise was supported by NIH training grant ST32 AI-07163.