ATF3 regulates MCMV infection in mice by modulating IFN-γ expression in natural killer cells

Abstract

Activating transcription factor 3 (ATF3) is a negative regulator of proinflammatory cytokine expression in macrophages, and ATF3-deficient mice are more susceptible to endotoxic shock. Here, we demonstrate that ATF3 interacts with a cis-regulatory element of the IFN-γ gene in natural killer (NK) cells, and that ATF3null NK cells show increased transcription and secretion of IFN-γ. NK cell-derived IFN-γ has previously been demonstrated to be protective against murine cytomegalovirus (MCMV) infection, and we show here that ATF3null mice exhibit decreased hepatic viral load and reduced liver histopathology upon challenge with MCMV. Reconstitution of NK-deficient mice with ATF3null NK cells more effectively controlled MCMV infection than mice reconstituted with WT cells, indicating that ATF3 acts within NK cells to regulate antiviral responses.

Interferon (IFN)-γ is a pleiotropic cytokine that functions at the innate-adaptive immune interface to promote antibacterial, antitumor, and antiviral responses (1). IFN-γ is produced by natural killer (NK), NKT, CD4⁺, and/or CD8⁺ T cells stimulated with the T helper (Th1)-skewing cytokines IL-12 and/or IL-18, with the cellular source of IFN-γ depending on the specific disease model. Murine cytomegalovirus (MCMV) is a herpes virus that causes a systemic infection, and NK cell production of IFN-γ early during infection is absolutely required for antiviral control (2). IFN-γ has many effects during MCMV infection; in general, it modulates the host response by regulating the production of cytokines and chemokines to recruit and activate effector cells (3), increasing the levels of MHC class I (4), and activating antiviral effector mechanisms (5, 6). A similar role for IFN-γ and NK cells in regulating herpes virus infection in humans has been suggested by genetic deficiency studies (ref. 7 and reviewed in refs. 8 and 9).

Although much is known about the regulation of IFN-γ expression in T cells, less is understood about this process in NK cells. It is known that the transcription factors and epigenetic modifications that regulate expression at the IFN-γ locus are different in T cells and NK cells (10–12). The transcription factors SMAD3 (13), NF-κBp50 (14), Hlx (15), and Runx3 (16) negatively regulate IFN-γ transcription in NK cells. Tight regulation of IFN-γ expression is required to limit pathology; however, the impact of negative regulators of IFN-γ expression in NK cells on disease has not been explored.

Our laboratory previously identified activating transcription factor 3 (ATF3) as an important negative regulator of proinflammatory cytokine gene expression in macrophages by using transcriptional profiling and computational analysis (17). Here, we observe that ATF3 expression is highly induced in the livers of MCMV-infected mice and purified NK cells. ATF3null mice show augmented antiviral responses against MCMV, an infection model where protection absolutely depends on NK-derived IFN-γ (reviewed in ref. 18). NK-deficient mice reconstituted with WT NK or ATF3null NK cells before MCMV infection clearly demonstrated that ATF3 acts within NK cells to regulate MCMV infection. ATF3 interacts directly with a cis-regulatory element of the IFN-γ gene in NK cells, and ATF3null NK cells show increased transcription and secretion of IFN-γ. Taken together, these data strongly suggest that ATF3 regulates anti-MCMV responses by controlling the production of IFN-γ in NK cells.

Results

ATF3null Mice Show Enhanced Protection Against MCMV Infection.

The mouse liver is a major target organ for MCMV replication. We observed that ATF3 expression was increased in the livers of mice infected with a sublethal dose of MCMV for 3 days (Fig. 1A). We evaluated the effectiveness of host antiviral responses by measuring the viral load within the liver and examining the corresponding infection-induced histopathology. ATF3null mice controlled viral replication better than their WT counterparts; we observed a striking decrease in viral load in the livers of ATF3null mice relative to WT mice (Fig. 1B). These findings were corroborated by the observation that ATF3null mice exhibited substantially less MCMV-induced hepatic histopathology (Fig. 1 C and D). In all infected mice, discrete inflammatory foci were distributed multifocally and randomly throughout the liver parenchyma, usually surrounding infected hepatocytes. These cells presented with typical MCMV-induced changes, including cytomegalic cells with characteristic eosinophilic intranuclear inclusion bodies (INIB). In addition, the hepatitis was characterized by variable mononuclear cells, neutrophils, apoptotic cells, and necrotic debris. Cytomegalic hepatocytes with INIB were also found without adjacent inflammatory cells, and there were mild, focally extensive foci of hepatic coagulative necrosis in some sections. Liver sections prepared from infected ATF3null mice had inflammatory foci containing decreased numbers of viral-induced cytomegalic cells and INIB (sites of active viral replication) compared with WT mice (Fig. 1 C and D). The hepatitis within the ATF3null livers was predominately mononuclear by 3 days postinfection.

Fig. 1.

ATF3null mice show enhanced protection against MCMV infection. (A) ATF3 mRNA levels are increased in MCMV-infected mice relative to uninfected mice, as measured by qRT-PCR. Liver ATF3 expression 3 days postinfection was normalized by EF1α expression in each mouse and then divided by the expression level in uninfected livers. Mean ± SEM for nine infected mice is shown. (B) Mice infected with MCMV have decreased liver viral load after 3 days of infection. The mean viral pfu/g liver for three mice is denoted by a dash (*, P = 0.04). These data are representative of three independent experiments (P < 0.05; n = 10 mice per genotype). (C) ATF3null mice show decreased scores for viral-induced cytomegaly and INIB compared with WT mice. Mean ± SEM is shown; n = 3 mice per genotype. (D) ATF3null mice show reduced viral-induced damage in the liver, with fewer productively infected cytomegalic cells containing INIB (arrow). Representative H&E-stained liver sections are shown. (Scale bars: 10 μm.)

ATF3 Acts Directly in NK Cells to Control MCMV Infection in Vivo.

Studies using immunodeficient mice or cell-depleting antibodies have established that NK cells control liver viral load and hepatitis during the first 3 days of MCMV infection (2, 19–22), a time frame when we observe a significant difference in viral load between WT and ATF3null mice. The ATF3 effect is kinetically compatible with this transcription factor regulating NK cell function. We established that stimulated NK cells express ATF3 mRNA and protein (Fig. 2 and data not shown) and demonstrated that freshly isolated NK cells rapidly induce ATF3 mRNA expression after in vitro activation with IL-12 and anti-CD28 antibody (a surrogate for antigen-presenting cell costimulation that augments IFN-γ production) (23, 24) (Fig. 2). To establish whether ATF3 acts directly in NK cells to mediate the increased resistance of ATF3null mice to MCMV infection, as opposed to indirectly by regulating inflammatory cytokine secretion in macrophages or other cell types, we performed reconstitution experiments in NK-deficient mice. Common γ-chain^−/−/Rag2^−/− mice lack NK cells, and although they also lack T and B lymphocytes, they represent the best available genetic mouse model of NK cell deficiency (25). These NK-deficient mice were injected with equivalent numbers of WT or ATF3null NK cells and infected 24 h later with MCMV. FACS analysis of splenocytes prepared from mice 3 days after NK cell reconstitution showed that the spleens contained equivalent numbers of DX5/CD49b⁺CD3⁻ cells (data not shown). Mice that received ATF3null NK cells had a significant decrease in MCMV pfu at 2 and 3 days postinfection when compared with mice that received their WT counterparts (Fig. 3A and data not shown). After 3 days of MCMV infection, macroscopic white necrotic foci were readily observed on the surface of livers from mice that had received WT NK cells, consistent with hepatitis and characteristic of severe uncontrolled infection. In contrast, these foci were not present on the livers of mice injected with ATF3null NK cells (data not shown). These macroscopic differences correlated with the striking histological differences observed in infected livers between the two groups. Liver sections from mice reconstituted with WT NK cells were characterized by readily observable abundant cytomegalic cells containing well defined eosinophilic INIB (Fig. 3 B and C). These sites of active viral replication were either not surrounded by inflammatory cells (Fig. 3B) or surrounded principally by recruited neutrophils rather than mononuclear cells (Fig. 3C). In contrast, mice reconstituted with ATF3null NK cells had inflammatory foci characterized principally by mononuclear cells with occasional neutrophils and rare cytomegalic cells either lacking INIB or containing ill-defined INIB, suggesting a nonproductive infection (Fig. 3D). Therefore, ATF3 can act intrinsically in NK cells to impact the course of viral infection in vivo.

Fig. 2.

NK cells express ATF3. IL-2-expanded NK cells stimulated with IL-12 and anti-CD28 antibody up-regulate ATF3 mRNA expression. ATF3 expression was determined by qRT-PCR and normalized to EF1α. Relative units are shown, and data are representative of the expression pattern in freshly isolated NK cells for ≥3 independent experiments.

Fig. 3.

NK-deficient mice reconstituted with ATF3null NK cells control MCMV infection better than mice reconstituted with WT NK cells. (A) Mice infected with MCMV have decreased viral load in liver after 2 days of infection. Viral titers were quantified by overlaying fibroblasts with serial dilutions of organ homogenates and enumerating plaques. The mean viral pfu/g liver for three mice is denoted by a dash (*, P = 0.03). (B–D) Representative H&E-stained liver sections are shown. (B and C) Liver sections from mice reconstituted with WT NK cells contained numerous cytomegalic cells containing distinct characteristic eosinophilic INIB, which either lacked inflammatory cells (B) or were surrounded by predominantly polymorphonuclear cells (C). (D) Mice reconstituted with ATF3null NK cells had inflammatory foci characterized by fewer cytomegalic cells, which lack distinct INIB and are surrounded predominantly by mononuclear cells. These data are representative of two independent experiments. (Scale bars: 10 μm.)

ATF3null NK Cells Express More IFN-γ than WT NK Cells.

To understand how ATF3 regulates NK effector function, we compared expression of candidate genes between WT and ATF3null NK cells. It has been shown that NK cells regulate liver viral loads by their production of IFN-γ (6), whereas there is conflicting evidence regarding the role of granzyme B and perforin as additional antiviral effectors in the liver (refs. 6, 26, and 27 and reviewed in ref. 18). We observed that ATF3null NK cells stimulated with IL-12 and anti-CD28 antibody produce significantly more IFN-γ than their WT counterparts (Fig. 4). In contrast, ATF3 expression did not alter perforin or granzyme B expression significantly in stimulated NK cells (Fig. 4). We excluded the possibility that ATF3null NK cells were more responsive to IL-12 by demonstrating that GM-CSF and TNF, two prototypic IL-12-stimulated genes, were equivalently transcribed, expressed, and secreted by IL-12-stimulated WT and ATF3null NK cells (Fig. 4 and data not shown). The selectivity of IFN-γ regulation by ATF3 was replicated in vivo; the livers of MCMV-infected ATF3null mice contained higher levels of IFN-γ mRNA and protein than those from infected WT mice, whereas ATF3 expression did not alter levels of perforin, granzyme B, TNF, and GM-CSF mRNA (data not shown).

Fig. 4.

ATF3null NK cells express more IFN-γ than WT NK cells. IL-2-expanded NK cells were stimulated in vitro with IL-12 and anti-CD28 antibody for 4 h, and qRT-PCR was performed with probes specific for IFN-γ, granzyme B, perforin, GM-CSF, and TNF, and mRNA levels were normalized by using EF1α levels. Values were then divided by the expression level in WT cells, and the displayed units correspond to the relative fold increase in gene expression. The means and SEM of two to six independent experiments are shown (*, P < 0.05).

ATF3 Negatively Regulates IFN-γ Expression in NK Cells.

We performed a series of experiments to determine whether ATF3 functionally regulates IFN-γ expression in NK cells. ATF3 translocates to the nucleus in NK cells from 2 to 6 h after in vitro stimulation, as shown by Western blotting of nuclear extracts and immunofluorescence microscopy studies (data not shown). To determine whether ATF3 can interact directly with a cis-regulatory element of the IFN-γ gene, ChIP experiments were performed with an ATF3-specific antibody (Fig. 5A). A distal element has been shown to regulate early IFN-γ expression by T cells stimulated with IL-12 (28–31). Using the TRANSFAC database we determined that this element contained several sites that could be occupied by ATF3 [namely CREB (−297 to −304) and TAXCREB (−296 to −310 and −199 to −213)]. DNA fragments immunoprecipitated by using an anti-ATF3 antibody were used as templates for quantitative PCR using distal element-specific primers. ChIP experiments revealed inducible ATF3 binding to the distal cis-regulatory element of the IFN-γ gene in IL-12-stimulated NK cells, which was maximal at 4 h and sustained through 12 h poststimulation (Fig. 5A). This binding correlated with repression of IFN-γ expression because ATF3null NK cells transcribed and secreted significantly more IFN-γ after in vitro stimulation than WT NK cells (Fig. 5 B and C). Interestingly, although T cells express ATF3, it does not influence the production or secretion of IFN-γ (Fig. 5D), nor does it bind to the regulatory elements of the IFN-γ gene described above (data not shown).

Fig. 5.

ATF3 negatively regulates IFN-γ expression in NK cells. (A) Chromatin was isolated from IL-2-expanded NK cells at various time points after stimulation with IL-12 and anti-CD28 antibody. ChIP experiments used ATF3-specific antibodies and primers to amplify −375 to −164 of the IFN-γ locus, which contains two predicted ATF3-binding sites. Immunoprecipitated DNA was quantified by qPCR using a Taqman-labeled probe and normalized to the quantity of DNA in 10% of the input used for ChIP for each sample. NRS denotes ChIP performed using normal rabbit serum as a negative control. Data are representative of two to four independent experiments. (B) ATF3null NK cells express more IFN-γ mRNA than WT NK cells. Freshly isolated NK cells were stimulated with IL-12 + anti-CD28 antibody. RNA was isolated and expression of IFN-γ was measured by qRT-PCR and normalized to EF1α expression. Relative units are shown. Data are representative of the expression pattern in IL-2-expanded NK cells; means ± SEM for ≥3 independent experiments are shown. (C) ATF3null NK cells secrete more IFN-γ relative to WT cells. Freshly isolated NK cells were stimulated with IL-12 and anti-CD28 antibody, and secreted IFN-γ was quantified by ELISA. Means ± SEM for ≥3 independent experiments are shown. (D) ATF3 does not regulate IFN-γ secretion by T cells. WT and ATF3null primary T cells were stimulated with IL-12 + anti-CD3 + anti-CD28 antibodies. Supernatants were collected after 24 h, and secreted IFN-γ was quantified by ELISA. Means ± SEM for ≥3 independent experiments are shown.

Discussion

We have shown that ATF3 is a negative regulator of IFN-γ expression in NK cells. This activity correlates with increased resistance of ATF3null mice to MCMV infection, a virus that is uniquely susceptible to control by NK-produced IFN-γ. In this model, hepatic viral load is absolutely controlled by IFN-γ, which is secreted by NK cells responding to IL-12 (2, 32, 33). Previous studies have shown that depletion of either NK cells or IFN-γ exacerbates infection, which leads to an increased liver viral load and hepatitis (2, 3, 19–21, 32). Further evidence that NK cell-derived IFN-γ limits viral replication and liver damage is provided by the observation that adoptively transferred NK cells are protective in WT but not in IFN-γR^−/− recipients (6). The 5- to 15-fold decrease in viral load we observed in ATF3null mice is comparable in magnitude to the 1 log decrease in viral load observed in MCMV-infected mice treated with ectopic IL-12. In those mice, IFN-γ secretion was increased by 2- to 6-fold during the first 3 days of infection. The consequent reduction in viral load and histopathology was completely abrogated by treatment with anti-IFN-γ antibodies, highlighting how modulation of IFN-γ levels can titrate antiviral responses (2).

NK cells are necessary for optimal development of inflammatory foci around infected cells. ATF3null livers showed reduced MCMV-induced hepatic histopathology, suggesting increased containment of MCMV. NK-deficient mice reconstituted with ATF3null NK cells similarly showed fewer INIB and more consistent recruitment of inflammatory cells around infectious foci compared with WT NK reconstituted mice. The abundance of neutrophils in the inflammatory foci in mice receiving WT NK cells suggests an uncontained infection, which is in striking contrast to the predominantly mononuclear infiltrate present in ATF3null NK-reconstituted mice. The NK reconstitution experiments clearly demonstrate that ATF3 acts within the NK compartment to regulate antiviral response; however, these data do not exclude a role for ATF3 within other cell types that could complement the effect observed in NK cells.

We show that ATF3 acts as a negative regulator of IFN-γ expression in NK cells and not in T cells. It has been suggested that chromatin structure may underlie differences in innate and adaptive production of cytokines (11). Epigenetic regulation of IFN-γ gene expression differs between NK and T cells. NK cells are poised to rapidly transcribe IFN-γ as this locus is constitutively and extensively hyperacetylated and hypomethylated, increasing chromatin accessibility for binding of both positive and negative transcriptional regulators (10–12). In contrast, histone acetylation at the IFN-γ locus in T cells depends on activation by Th1-skewing factors (such as IL-12) and proliferation (10, 11, 29). There is precedence for a transcription factor playing distinct roles in IFN-γ expression among cell types. Hlx and NF-κB p50 act as negative regulators of IFN-γ expression in NK cells (14, 15), yet can be positive transcriptional regulators of IFN-γ in T cells (34–36).

It is interesting to speculate as to why the removal of a negative regulator of NK cell function results in improved protection against MCMV infection. Although an increased level of IFN-γ is protective against acute viral infection, excessive production of IFN-γ by NK cells can lead to increased pathology in the Shwartzman reaction to LPS, inflammatory bowel disease, and atherosclerosis (37–42). Negative regulators such as ATF3 serve to modulate inflammation to a level appropriate for clearing a pathogen while preventing pathology caused by unrestrained inflammatory responses.

Materials and Methods

Mice.

C57BL/6 and Rag1^−/− (The Jackson Laboratory), BALB/c (Charles River Laboratories), ATF3^−/− (C57BL/6; T. Hai, Ohio State University, Columbus), and common-γ chain^−/−/Rag2^−/− (C57BL/6J × C57BL/10SgSnAi; Taconic) mice were housed in a specific pathogen-free environment with the approval and supervision of the Institutional Animal Care and Use Committee at the Institute for Systems Biology.

MCMV Infection of Mice.

Mice were infected i.v. with 3 × 10⁴ pfu of MCMV Smith strain (ATCC) isolated from the salivary glands of BALB/c mice. After 3 days, livers were collected and divided for histology, ELISA, quantitative RT-PCR (qRT-PCR), and pfu assay. For pfu assay, tissues were weighed, homogenized in PBS by using a tissue grinder (Omni), and sonicated on ice to release cell-associated virus. Cell debris was removed by centrifugation, and duplicate log dilutions were overlaid on nonconfluent monolayers of BALB/c mouse embryo fibroblasts. Cells were infected by centrifugation (2,000 × g for 30 min) followed by a 30-min incubation at 37°C, media were removed, and cells were overlaid with complete DMEM (Gibco) + 0.5% LMP-agarose (Gibco) and incubated for 3–4 days. Plaques were enumerated after incubation with 0.1% crystal violet in 20% methanol. Calculated pfu were normalized to homogenized tissue weight. Cytokine content of sonicated liver homogenates was determined by ELISA (R&D) and normalized to tissue weight.

Cell Purification and Stimulation.

NK cells were isolated from the spleens and lymph nodes (mesenteric, brachial, inguinal) of C57BL/6 WT and ATF3null mice. Single-cell suspensions were prepared by using collagenase D or Liberase Blendzyme 3 (Roche) and purified by autoMACS by negative selection using CD90 and B220 microbeads followed by positive selection using DX5 microbeads (Miltenyi Biotech). Cell purity was monitored by flow cytometry using NK1.1 (PK136), CD49b (DX5), CD3 (17A2), and B220 (RA3–6B2)-specific antibodies (BD Biosciences), and NK cells were 65–85% NK1.1⁺CD3⁻B220⁻. Cells were cultured in complete RPMI (Gibco) [containing 10% FBS (HyClone) + 2 mM l-glutamine + 100 units penicillin + 100 μg streptomycin + 100 μM β-mercaptoethanol] at 1 × 10⁶ cells per ml and stimulated with 4 ng/ml murine IL-12 (Peprotech) and 2 μg/ml anti-CD28 antibody (clone 37.51; Biolegend). T cells were additionally stimulated with immobilized anti-CD3ε antibodies (5 μg/ml; clone 145–2C11 LE; BD Biosciences). For some experiments, isolated NK cells were expanded by culture with 4,000 units/ml IL-2 (Peprotech) for 7–10 days. NK cells were prepared from Rag1^−/− mice by mechanically dissociating spleens, removing adherent cells, and culturing NK cells with IL-2. Results obtained using IL-2-expanded NK cells were confirmed in freshly isolated NK cells.

qRT-PCR, ELISA, and Western Blotting.

RNA was prepared by using TRIzol (Gibco/BRL), genomic DNA was removed by DNaseI (Ambion), and cDNA synthesis was performed on equivalent quantities of total RNA by using SuperScript II (Gibco/BRL). Gene-specific primers and Taqman probes (ABI) were used for qPCR amplification by using an ABI 7600. Gene expression levels were calculated using cycle numbers in the linear amplification range and normalized to the expression level of elongation factor (EF) 1α. ELISA for IFN-γ and IL-12p40 were performed according to the manufacturer's instructions (R&D). Nuclear extracts were prepared by removing cytosolic proteins by hypotonic lysis in the presence of 0.2% Nonidet P-40 and boiling washed nuclei in SDS/PAGE sample loading buffer. Proteins were resolved by using 4–20% gradient SDS/PAGE gels, transferred to PVDF membrane, blocked by using 5% milk-TBST, and hybridized overnight at 4°C by using ATF3-specific (Santa Cruz) or Lamin B1 (Zymed) antisera. After incubation with secondary antibody, blots were developed by using ECL (Pierce).

ChIP Assay.

NK cells from Rag1^−/− mice were expanded by in vitro culture with 4,000 units/ml IL-2 for 7 days to obtain sufficient cell numbers (5 million to 10 million cells per ChIP). Expanded cells were not secreting IFN-γ immediately before stimulation with 4 ng/ml IL-12 and 2 μg/ml anti-CD28 antibody. Freshly isolated T cells were stimulated with 4 ng/ml IL-12, 2 μg/ml anti-CD28 antibody, and 5 μg/ml immobilized anti-CD3 antibody. After stimulation, NK or T cells were fixed in 1% formaldehyde for 5 min, and cross-linking was stopped with 125 mM glycine, pH 2.5 for 5 min. Cells were washed with ice-cold PBS and pelleted, then lysed for 30 min in RIPA buffer (10 mM Tris, pH 8.0/140 mM NaCl/1% Triton X-100/0.1% SDS/1% SDC) supplemented with proteases inhibitors. Cells were sheared by drawing three times through a 30-g needle and sonicated (7 × 1 min at 30% maximum power), and fragmentation of chromatin was evaluated by agarose gel electrophoresis (majority of fragments were 200–1,000 bp). Lysate was centrifuged to pellet debris, and then brought to 1 ml by using RIPA buffer. Ten percent of the chromatin input was reserved before immunoprecipitation. Extracts were precleared for 15 min at 4°C with 20 μl of a 50% suspension of salmon sperm-saturated protein A. Immunoprecipitations were carried out at 4°C overnight with anti-ATF3 antibody (Santa Cruz) or normal rabbit serum IgG (Santa Cruz). Immune complexes were collected with protein A and washed three times for 5 min each with RIPA buffer, high-salt buffer (10 mM Tris, pH 8.0/0.1% SDS/1% Triton X-100/1 mM EDTA/0.5 mM EGTA/1% SDC/500 mM NaCl), LiCl buffer (0.25 M LiCl/1% Nonidet P-40/1% SDC/1 mM EDTA/10 mM Tris·HCl, pH 8.0), and low-salt buffer [1× Tris/EDTA (TE)]. Immune complexes were extracted in 1× TE containing 1% SDS, and protein–DNA cross-links were reversed by heating at 65°C overnight. After proteinase K digestion (180 μg, 1 h at 45°C), DNA was extracted by phenol-chloroform-isoamyl alcohol followed by chloroform-isoamyl alcohol, then ethanol-precipitated. One thirtieth of the immunoprecipitated DNA was used in each PCR. qPCR was performed (50°C for 2 min, 95°C for 10 min, 40 cycles of 95°C for 15 s, annealing and extension at 60°C for 1 min) to amplify −375 to −164 of the IFN-γ locus: 5′-AACCACAAACAAAGGCTCCCTGTGC-3′ (sense); 5′-AGAGTTTCCTTTCGACTCCTTGGG-3′ (antisense); 5′-FAM-AGAATCCCACAAGAATGGCACAGGTG-TAMRA-3′ (probe).

NK Cell Reconstitution.

NK cells were isolated from spleens and lymph nodes by enzymatic digestion using Liberase Blendzyme 3 (Roche) and purified by autoMACS as described, and cell purity was monitored by flow cytometry using NK1.1, CD49b, CD3, and B220-specific antibodies. WT and ATF3null NK cells were of similar purity, and each contained 10% contaminating CD3⁺ T cells. A total of 1 × 10⁶ NK cells were injected i.v. 20 h before i.p. infection with 1 × 10⁵ pfu MCMV. Samples were collected from mice 2 or 3 days postinfection. The efficiency of reconstitution was assessed by preparing single-cell suspensions from spleens of reconstituted mice 2–3 days postinfection and staining with CD49b and NK1.1-specific antibodies.

Histology.

Five-micrometer sections from neutral buffered formalin-fixed livers were stained with hematoxylin and eosin and examined by a board-certified veterinary pathologist who was blinded to genotype. Sections were scored on a four-point scale for distribution of hepatitis, individual cell necrosis, coagulative necrosis, apoptosis, cytomegaly, intranuclear inclusions, lymphohistiocytic inflammation, and neutrophilic inflammation. Cytomegaly and intranuclear inclusions were scored individually and then combined (0 = none, 1 = few foci contain enlarged cells/inclusions, 2 = average 1–2 cells per foci, 3 = average 3–5 cells per foci, 4 = average 6+ cells per foci). For NK reconstitution experiments, sections were examined by two veterinary pathologists who were blind to genotype.

Statistics.

The unpaired two-tailed Student's t test was used for statistical analysis. Results were considered statistically significant at P < 0.05.