Type I Interferon Induced Epigenetic Regulation of Macrophages Suppresses Innate and Adaptive Immunity in Acute Respiratory Viral Infection

Influenza A virus (IAV) is an airborne pathogen that causes significant morbidity and mortality each year. Macrophages (Mϕ) are the first immune population to encounter IAV virions in the lungs and are required to control infection. In the present study, we explored the mechanism by which cytokine signaling regulates the phenotype and function of Mϕ via epigenetic modification of chromatin. We have found that type I interferon (IFN-I) potently upregulates the lysine methyltransferase Setdb2 in murine and human Mϕ, and in turn Setdb2 regulates Mϕ-mediated immunity in response to IAV. The induction of Setdb2 by IFN-I was significantly impaired upon inhibition of the JAK-STAT signaling cascade, and chromatin immunoprecipitation revealed that both STAT1 and interferon regulatory factor 7 bind upstream of the transcription start site to induce expression. The generation of Setdb2 LacZ reporter mice revealed that IAV infection results in systemic upregulation of Setdb2 in myeloid cells. In the lungs, alveolar Mϕ expressed the highest level of Setdb2, with greater than 70% lacZ positive on day 4 post-infection. Silencing Setdb2 activity in Mϕ in vivo enhanced survival in lethal IAV infection. Enhanced host protection correlated with an amplified antiviral response and less obstruction to the airways. By tri-methylating H3K9, Setdb2 silenced the transcription of Mx1 and Isg15, antiviral effectors that inhibit IAV replication. Accordingly, a reduced viral load in knockout mice on day 8 post-infection was linked to elevated Isg15 and Mx1 transcript in the lungs. In addition, Setdb2 suppressed the expression of a large number of other genes with proinflammatory or immunomodulatory function. This included Ccl2, a chemokine that signals through CCR2 to regulate monocyte recruitment to infectious sites. Consistently, knockout mice produced more CCL2 upon IAV infection and this correlated with a 2-fold increase in the number of inflammatory monocytes and alveolar Mϕ in the lungs. Finally, Setdb2 expression by Mϕ suppressed IL-2, IL-10, and IFN-γ production by CD4+ T cells in vitro, as well as proliferation in IAV-infected lungs. Collectively, these findings identify Setdb2 as a novel regulator of the immune system in acute respiratory viral infection.


Introduction
IAV is an airborne pathogen that is responsible for significant mortality in humans [1]. Infection with seasonal strains of IAV is typically limited to the upper respiratory tract and causes mild to moderately severe respiratory disease. In contrast, highly pathogenic strains of IAV can spread to distal airways and alveolar spaces causing pneumonia that can be lethal. Alveolar macrophages (Mϕ) are the first immune population to encounter IAV virions in the lungs and are required for host protection [2][3][4][5]. Following activation, alveolar Mϕ become highly phagocytic and are a major source of proinflammatory cytokines, including type I interferon (IFN-I) [6,7]. Viral detection by pattern recognition receptors (PRRs) initiates a signaling cascade that activates interferon regulatory factor (IRF) 3 and IRF7, transcription factors involved in the initiation and amplification of the IFN-I response [8,9]. IFN-I binds to the IFN-α receptor (IFNAR) to induce the transcription of more than 300 IFN-stimulated genes (ISGs) with antiviral and immunomodulatory functions [10]. However, the production of IFN-I and other proinflammatory cytokines must be tightly regulated to avoid respiratory failure. Cytokineinduced lung injury, rather than uncontrolled viral replication, is the most common cause of severe morbidity and mortality in individuals exposed to highly pathogenic strains of IAV [11][12][13].
Several other functions of resident and recruited Mϕ in infection caused by respiratory pathogens have been described. Early production of chemokines by alveolar Mϕ promotes the infiltration of inflammatory cells to the site of infection [14]. Additionally, Mϕ directly initiate adaptive immune responses during infection. It has been shown that alveolar Mϕ rapidly transport antigen to draining lymph nodes in Streptococcus pneumoniae infection [15]. Within the lungs, Mϕ present antigen and activate virus-specific T cells [16]. Expression of the Notch ligand Delta-like 1 by Mϕ regulates the production of the antiviral cytokine IFN-γ by CD4 + and CD8 + T cells in IAV infection [5]. Mϕ further enhance T cell-mediated immunity by undergoing apoptosis, resulting in cross-presentation of antigen to cytotoxic CD8 + T cells by DCs [17,18]. Finally, Mϕ play a pivotal role in the resolution of infection and restoration of an anti-inflammatory environment in the lungs. Internalization of residual infected-apoptotic cells and cellular debris by Mϕ inhibits viral dissemination and tissue damage by dampening inflammation and maintaining lung function [19][20][21].
Epigenetic modifications control gene transcription by altering residues in histone tails of chromatin. It has been shown that specific chromatin-modifying enzymes influence the phenotype and function of Mϕ [22][23][24][25]. The SET (Su(var)3-9, Enhancer-of-zeste, Trithorax)-domain superfamily consists of histone-modifying enzymes that transfer a methyl group from S-adenosyl-L-methionine to specific lysine residues in histone tails to either activate or block transcription [26]. Setdb2 (SET-domain bifurcated 2) tri-methylates lysine 9 of histone H3 (H3K9me3) to silence gene expression and was first implicated in the induction of B cell chronic lymphocytic leukemia [27]. Consistent with a recent publication by Schliehe et al., we demonstrate that cytokine-dependent signal transduction following IFN-I treatment upregulates Setdb2 in myeloid cells in a STAT1-and IRF7-dependent manner. However, despite this overlapping observation, we uncovered a role for Setdb2 in the regulation of the innate and adaptive immune system in primary IAV infection. Generation of mice lacking Setdb2 specifically in Mϕ revealed that Setdb2 controlled the recruitment of inflammatory monocytes to infected lungs and suppressed the expression of a large number of antiviral genes. Setdb2 expression by Mϕ also influenced cytokine production by CD4 + T cells, as well as proliferation of both CD4 + and CD8 + T cells in infected lungs. These results highlight the impact of histone modification in dictating the severity of infection and may represent a potential therapeutic target for controlling pulmonary infection and other diseases associated with IFN-I activity.

IFN-I upregulates the lysine methyltransferase Setdb2 in murine and human Mϕ
Specific cytokines can induce the expression of histone-modifying enzymes, which, in turn, regulate the transcription of target genes in a variety of immune responses [23,24]. Since IFN-I is rapidly produced following infection with a number of viral pathogens, we asked if IFN-I induced the expression of histone-modifying enzymes in bone marrow-derived Mϕ (BM-Mϕ). Notably, the lysine methyltransferase Setdb2 was upregulated by more than 700-fold relative to unstimulated BM-Mϕ (1.0 ± 0.17 vs. 781.0 ± 108.2; p<0.001) at 24 hours post-stimulation ( Fig  1A). IFN-I-dependent induction was specific to Setdb2, as related histone methyltransferases containing a SET-domain were unaltered following cytokine stimulation ( Fig 1B). To determine if the induction of Setdb2 occurred in a dose-dependent manner, BM-Mϕ were stimulated with increasing doses of cytokine. When normalized to unstimulated BM-Mϕ, a direct correlation between the concentration of IFN-I and Setdb2 transcript was observed ( Fig 1C). To characterize the kinetics of Setdb2 expression, BM-Mϕ were treated with IFN-I over a time course. Setdb2 transcription peaked at 5 hours post-stimulation, began declining by 24 hours, and returned to baseline levels by 48 hours (Fig 1D). We next examined whether cytokines related to IFN-I could upregulate Setdb2. BM-Mϕ were treated with IFN-γ (type II IFN) or IFN-λ (type III IFN) in parallel with IFN-I. While IFN-γ upregulated Setdb2 relative to unstimulated cells (16.8 ± 1.48 vs. 1.0 ± 0.33; p<0.001), it was significantly less potent than IFN-I ( Fig 1E).
To further characterize Setdb2 expression in vitro and in vivo, we generated Setdb2 LacZ reporter mice. To measure the degree of transcription from the Setdb2 promoter, a gene trap vector was used to incorporate the E. coli gene lacZ, which encodes β-galactosidase, into recombinant DNA to generate lacZ fusion transcripts as previously described [28]. BM-Mϕ from reporter mice were treated with a vehicle control or IFN-I and β-galactosidase activity was measured by flow cytometry. At 24 hours post-stimulation, 5% of control cells expressed lacZ. IFN-I treatment increased Setdb2 expression, with more than 20% of BM-Mϕ lacZ positive (5.17 ± 0.48% vs. 22.3 ± 1.32%; p<0.001) (Fig 1F). Enhanced lacZ expression following IFN-I treatment correlated with an increase in mean fluorescent intensity (MFI) (4921 ± 35.2 vs. 6464 ± 52.3; p<0.001) (Fig 1G).
These results prompted us to characterize SETDB2 expression in human cells. Peripheral blood mononuclear cells (PBMCs) were isolated from healthy donors and stimulated with IFN-I. When normalized to unstimulated PBMCs, IFN-I treatment resulted in a 3-fold increase in SETDB2 at 5 and 48 hours post-stimulation ( Fig 1J). Since we only observed a slight increase in SETDB2 in PBMCs, CD14 + monocytes were skewed toward a Mϕ phenotype and stimulated with IFN-I. Initially, IFN-I did not induce SETDB2 expression in human Mϕ. However, a slight increase was observed by 24 hours and SETDB2 expression continued to increase through 72 hours post-stimulation ( Fig 1K). In contrast to murine BM-Mϕ, IFN-I upregulated SETD2, SETD5, and SUV420 in human Mϕ at 72 hours post-stimulation ( Fig 1L). Despite being the primary target for IAV, Setdb2 was not induced in murine airway epithelial cells (AECs) or normal human bronchial epithelial cells (NHBEs) treated with IFN-I (Fig 1M and 1N).

Setdb2 expression by Mϕ is dependent on JAK-STAT pathway and IRF7
IFN-I signals through the JAK-STAT signaling pathway to promote the transcription of ISGs involved in antiviral immunity. This prompted us to explore the signaling pathway regulating Setdb2 transcription. We initially tested if IFN-I-dependent induction of Setdb2 was dependent on the JAK-STAT pathway using the JAK inhibitor tofacitinib [29]. In comparison to control BM-Mϕ, Setdb2 transcript was undetected in both unstimulated and IFN-I-stimulated BM-Mϕ treated with tofacitinib (Fig 2A). Next, we examined Setdb2 expression in Stat1 -/mice since IFN-I signals predominantly through a STAT1-STAT2 heterodimer. Stimulation of wild-type BM-Mϕ with IFN-I resulted in nearly a 1500-fold increase Setdb2 transcription (1463 ± 250; p<0.001). A deficiency in Stat1 significantly dampened the induction of Setdb2, with less than   (Fig 2B). In addition, Setdb2 expression was impaired in the lungs of Stat1 -/mice infected with IAV. On day 4 post-infection, a 90% reduction in Setdb2 transcript was observed in CD11b + cells from Stat1 -/lungs relative to control lungs ( Fig 2C). IRF3 and IRF7 are critical transcription factors involved in IFN-I production and the induction of ISGs [30,31]. Whereas IRF3 is constitutively expressed at low levels and initiates IFN-I production after viral detection, IRF7 is an ISG that is expressed at high levels in infection and is the master regulator of the IFN-I response. Consistent with published data, stimulation of BM-Mϕ with IFN-I resulted in a dramatic upregulation of Irf7, but not Irf3 (Fig 2D and 2E). To determine if Setdb2 expression was dependent on either transcription factor, BM-Mϕ were transfected with control, IRF3, or IRF7 siRNA and stimulated with IFN-I. Silencing of IRF7, but not IRF3, resulted in a significant reduction in Setdb2 in comparison to control cells (0.44 ± 0.08 vs. 1.0 ± 0.09; p<0.01). Transcription of Irf3 and Irf7 was diminished by at least 60% and 80%, respectively, when treated with respective siRNA indicating the knockdown was successful (Fig 2F and 2G).
Since Setdb2 expression was diminished in BM-Mϕ deficient in Stat1 or treated with IRF7 siRNA, we performed ChIP to determine if these transcription factors regulated expression by binding to the Setbd2 promoter. Using published binding site sequences, we identified STAT1 and IRF7 binding sites upstream of the Setdb2 transcription start site. Prior to cytokine stimulation, STAT1 and IRF7 were absent in the Setdb2 promoter (Fig 2H and 2I). However, a dramatic increase in STAT1 and IRF7 binding was observed in BM-Mϕ stimulated with IFN-I ( Fig 2H and 2I).

Setdb2 is systemically upregulated in myeloid cells in IAV infection
Since IAV infection results in the influx of inflammatory cells to the lungs, we further characterized Setdb2 expression using Setdb2 LacZ reporter mice. In naïve lungs, alveolar Mϕ expressed the highest level of Setdb2, with 30% of the population lacZ positive. Less than 10% of other myeloid and lymphoid cellular populations expressed lacZ in steady-state conditions ( Fig 3A). IAV infection resulted in upregulation of Setdb2 in multiple cellular populations. Similar to uninfected lungs, alveolar Mϕ were the predominant population expressing Setdb2 in IAV infection. On day 4 post-infection, the proportion of alveolar Mϕ expressing Setdb2 nearly doubled, with greater than 70% expressing lacZ (31.3 ± 1.74% vs. 74.6 ± 2.69%; p<0.001) ( Fig  3A). In addition, infection enhanced lacZ expression in other myeloid populations, including inflammatory monocytes, tissue Mϕ, neutrophils, and DCs. In contrast, CD4 + T cells were the only lymphoid population to express more lacZ following infection. Comparable to naïve lungs, less than 10% of NK cells, CD8 + T cells, and B cells were lacZ positive in infected lungs ( Fig 3A). IAV-dependent induction of Setdb2 transcription was not limited to the lungs, as lacZ expression was significantly elevated in inflammatory monocytes in the blood (6.51 ± 1.16% vs. 46.3 ± 5.01%; p<0.01) and spleen (18.1 ± 2.48% vs. 33.9 ± 4.19%; p<0.01) (Fig 3B and 3C).

Setdb2 expression by specific myeloid populations controls the severity of IAV infection
To determine the role of Setdb2 in immunity during respiratory viral infection, we generated mice deficient for Setdb2 in myeloid cells with lysosomes (monocytes, Mϕ, and granulocytes) using the Cre-lox system. For validation, Setdb2 transcription was measured in BM-Mϕ from Setdb2 ff Lyz2 cre-(control) and Setdb2 ff Lyz2 cre+ (knockout) mice. Relative to control BM-Mϕ, a significant reduction in transcript was observed in Setdb2 -/-BM-Mϕ treated with a vehicle control or IFN-I ( Fig 4A). To examine if Setdb2 influenced the outcome of infection, mice were infected with a lethal dose of IAV and survival was monitored for two weeks. While both groups of mice began to succumb to infection on day 7, survival was enhanced in Setdb2 ff Lyz2 cre+ mice. Whereas only 20% of control mice were alive by day 9, greater than 65% of knockout mice survived ( Fig 4B). To determine viral load, we quantified the number of copies of IAV proteins in the lungs. While we observed a comparable fold-increase of non-structural protein 1 (NS1) and matrix protein 1 (M1) at day 4 post-infection, Setdb2 ff Lyz2 cre+ lungs had a reduction in both viral proteins by day 8 post-infection ( Fig 4C).
Since tissue damage is often the cause of morbidity and mortality in IAV infection, we next examined lung histology in naïve and infected mice. Prior to infection, control and Setdb2 ff Lyz2 cre+ lung histology was comparable ( Fig 4D). Indicative of respiratory viral infection, the airways of control lungs were filled, likely with dead epithelial and inflammatory cells, cellular debris, and virus on day 4 post-infection. In contrast, Setdb2 ff Lyz2 cre+ lungs had less obstruction to the airways. In addition, we observed dense clusters of lymphoid cells near blood vessels in knockout mice that were absent in wild-type lungs ( Fig 4D).

Setdb2 suppresses the expression of proinflammatory and antiviral genes
IAV sensing by the innate immune system results in robust production of IFN-I and other proinflammatory cytokines and chemokines. Since knockout mice controlled infection better than their wild-type counterparts, we postulated that Setbd2 dictates the severity of infection by regulating the transcription of antiviral genes. To identify potential target genes, we screened antiviral genes in control and Setdb2 -/-BM-Mϕ stimulated with IFN-I using a PCR array. In the absence of Setdb2, the overall gene profile in BM-Mϕ was altered at 24 hours post-stimulation (Fig 5A; p<0.001). A 5.8-and 6.8-fold increase in Ifna2 and Ifnb1 transcript, respectively, was observed in Setdb2 -/-BM-Mϕ. Notably, the CCR2 and CCR5 ligands Ccl2 and Ccl5, respectively, were upregulated 23-and 29-fold in Setdb2 -/-BM-Mϕ (Figs 5A and 6D). Additional chemokines and cytokines, including Cxcl1, Cxcl2, Cxcl9, Cxcl10, Cxcl11, Il1b, Il6, Il10, Il12a, Il12b, and Il15 were upregulated by 2-fold or more in the absence of Setdb2. To confirm the cytokine transcription data, we measured the concentration of select cytokines in the supernatants of BM-Mϕ stimulated with IFN-I. Consistently, more IL-1β, IL-6, IL-10, IL-12p40, and G-CSF were detected in the absence of Setdb2. TNF-α transcript and protein was the only cytokine examined that was reduced in Setdb2 -/-BM-Mϕ (Fig 5B).
IFN-I induces the expression of ISGs that inhibit viral replication and spreading [32,33]. Elevated Ifna2 and Ifnb1 in Setdb2 -/-BM-Mϕ correlated with a 22.7-and 8.5-fold increase in Isg15 and Mx1 transcript, respectively (Fig 5C). Since Setdb2 silences gene expression, we performed ChIP to determine the presence of Setdb2 and H3K9me3 in the promoter region of both genes. In control BM-Mϕ, IFN-I stimulation resulted in a 40-and 300-fold increase in Setdb2 bound to the Isg15 and Mx1 promoters, respectively (Fig 5D and 5E). This correlated with high levels of H3K9me3 in the promoter region of both genes (Fig 5D and 5E). Relative to control BM-Mϕ a significant reduction in Setdb2 and H3K9me3 was observed in Setdb2 -/-BM-Mϕ (Fig 5D and 5E). In addition to less Setdb2 and H3K9me3, enhanced Mx1 expression in Setdb2 -/-BM-Mϕ was associated with STAT1 and IRF7 bound to the promoter (Fig 5F and  5G). This was specific to IFN-I, as stimulation with IFN-II did not increase STAT1 or IRF7 binding in the Mx1 promoter (Fig 5F and 5G). To determine if enhanced survival and a reduced viral load was associated with an enhanced antiviral response, RNA was isolated from the lungs on day 4 post-infection. Consistent with BM-Mϕ data, Ifna2 (1.36 ± 0.27 vs. 33.1 ± 20.8; p<0.05) and Ifnb1 (1.47 ± 0.36 vs. 488.9 ± 241.4; p<0.05) transcript was elevated in the lungs from Setdb2 ff Lyz2 cre+ mice. This correlated with a 5-and 200-fold increase in Mx1 and Isg15, respectively. Heightened expression of antiviral effectors was specific to Mx1 and Isg15, as Rnasel (Ribonuclease L) and Pkr (Protein kinase R) expression was comparable in control and knockout animals infected with IAV ( Fig 5H).

Setdb2 expression does not influence transcription factor activation in BM-Mϕ
An augmented antiviral response in Setdb2 -/-BM-Mϕ was associated with higher expression of a variety of upstream genes linked to the JAK-STAT, TBK1-IRF7, and Iκκ-NF-κB signaling pathways ( Fig 5A). This correlated with upregulation of PRRs, as well as downstream signaling molecules that drive the induction of IFN-I and other proinflammatory cytokines and chemokines. This included upregulation of Ddx58, Ifih1, Aim2, Nlpr3, Tlr3, Tlr8, Tlr9, Mavs, Irf7, Myd88, Stat1 and multiple other genes. Moreover, elevated Mx1 mRNA in knockout BM-Mϕ was associated with enhanced recruitment of STAT1 and IRF7 to the gene promoter. Thus, we postulated that Setdb2 indirectly suppresses proinflammatory gene expression by inhibiting the activation and nuclear translocation of transcription factors. Despite a 3.4-fold increase in Stat1 mRNA (Fig 5A), no difference in total STAT1 protein based on absorbance at 450-nm was observed in knockout BM-Mϕ stimulated treated a media alone (2.31 ± 0.12 vs. 2.39 ± 0.09) or IFN-I (2.32 ± 0.07 vs. 2.30 ± 0.11). At 30 minutes post-cytokine treatment, we unable to detect phosphorylated STAT1 in whole-cell extracts in either group. In contrast to STAT1, a higher concentration of NF-κB p65 protein was detected in knockout BM-Mϕ (211.4 ± 17.64 ng/mL vs. 290 ± 16.6 ng/mL; p<0.01). However, no correlation between Setdb2 expression and NF-κB p65 activation was observed, as an equal concentration of phosphorylated NF-κB p65 was detected in whole-cell lysates from control and knockout BM-Mϕ treated with a vehicle control (16.92 ± 0.89 ng/mL vs. 18.84 ± 3.21 ng/mL) or stimulated with IFN-I (21.29 ± 1.27 ng/mL vs. 24.65 ± 1.66 ng/mL) for a half hour.

Setdb2 expression by myeloid cells controls inflammatory cell recruitment in IAV infection
Since several chemokines were upregulated in Setdb2 -/-BM-Mϕ, we asked if Setdb2 controls the influx of immune cells to the lungs. Setdb2 ff Lyz2 cre+ mice had a 2-fold increase in the number of inflammatory monocytes and alveolar Mϕ on day 4 post-infection (Fig 6A). More CCR2 + monocytes in knockout lungs correlated with an increase in Ccl2 transcript and protein, Setdb2 Regulates Macrophages in Viral Infection a potent chemotactic mediator of inflammatory monocytes (Fig 6B and 6C). Similar to the lungs, Setdb2 -/-BM-Mϕ stimulated with IFN-I transcribed more Ccl2 than control cells ( Fig  6D) and ChIP analysis revealed Ccl2 is a Setdb2 target gene (Fig 6E and 6F). In respect to control cells, a 10-and 30-fold reduction in Setdb2 and H3K9me3, respectively, was observed in the Ccl2 promoter in Setdb2 -/-BM-Mϕ (Fig 6E and 6F).

Setdb2 expression by Mϕ influences T cell proliferation in IAV-infected lungs
Transcriptional analysis of IFN-I-stimulated BM-Mϕ identified Setdb2 as a negative regulator of antiviral immunity. This included the suppression of a number of cytokines and chemokines implicated in the migration and subsequent activation of T cells in the lungs following viral infection (Fig 5A and 5B). Although the number of CD4 + and CD8 + T cells in IAV-infected lungs was comparable on day 4 post-infection (Fig 6A), it is possible that this time point occurred prior to the initiation of the adaptive immune system. This is supported by the observation that reduced disease severity in Setdb2 ff Lyz2 Cre+ mice was not observed until day 8 post-infection. To address this concern, control and Setdb2 ff Lyz2 Cre+ mice infected with a sublethal dose of IAV were euthanized on days 6, 8, and 10 post-infection to determine the proportion and number of T cells in the lungs. In both groups, the proportion of CD4 + T cells represented 10 to 15% of cells in the lungs at all days examined. A very modest, yet significant increase in CD4 + T cells was observed in knockout lungs in comparison to controls (9.73 ± 0.70% vs. 12.90 ± 1.02%; p<0.05). In contrast, the proportion of CD8 + T cells increased over time in IAV-infected lungs. On day 6 post-infection, the proportion of CD8 + T cells were comparable in control and mutant lungs, with CD8 + T cells composing approximately 6% of total cells. Relative to wild-type mice, knockout mice had an increase proportion of CD8 + T cells in the lungs on day 8 (13.25 ± 1.00% vs. 16.87 ± 0.79%; p<0.01) and day 10 (15.92 ± 1.44% vs. 21.32 ± 1.37%; p<0.05). Despite only modest changes in the percentage of T cells, the absolute number of T cells was significantly altered in the absence of myeloid Setdb2. Notably, Setdb2 ff Lyz2 Cre+ mice had 3 times the number of both CD4 + T cells (5.81 ± 0.50 x 10 5 vs. 15.43 ± 0.84 x 10 5 ; p<0.001) and CD8 + T cells (3.82 ± 0.37 x 10 5 vs. 9.69 ± 0.59 x 10 5 ; p<0.001) in the lungs on day 6 post-infection. While the number of both T cell populations were similar in control and knockout lungs on day 8 post-infection, a 2-fold increase in CD8 + T cells (6.54 ± 1.05 x 10 5 vs. 11.80 ± 1.47 x 10 5 ; p<0.01) was observed on day 10 in mice lacking Setdb2 (Fig 7A).
Following homing to the lungs, IAV triggers the activation and subsequent expansion of T cells at the site of infection. Since knockout mice had a greater number of both CD4 + and CD8 + T cells, we asked whether changes in proliferative capability contributed to the altered T cell prolife in IAV-infected lungs. To determine the extent of expansion, IAV-infected mice were injected with EdU (5-ethynyl-2'-deoxyuridine) and euthanized on the indicated days. While proliferation of both CD4 + and CD8 + T cells peaked on day 8 post-infection, the extent of EdU incorporation was significantly higher in knockout lungs. In wild-type lungs, 39.55 ± 2.06% of CD4 + T cells and 34.72 ± 2.20% of CD8 + T cells were EdU positive. In the absence of myeloid Setdb2, EdU incorporation was enhanced by greater than 10% in both CD4 + T cells (39.55 ± 2.06% vs. 51.00 ± 2.62%; p<0.01) and CD8 + T cells (34.72 ± 2.20% vs. 48.20 ± 1.83%; p<0.01). In contrast, gating on either CD4 + or CD8 + T cells on days 6 and 10 post-infection revealed no changes in the extent of EdU incorporation between the two groups of animals. However, despite this observation, the absolute number of proliferating T cells was enhanced on two or all three time points depending on the T cell subset. In respect to controls, knockout lungs had approximately 3 x 10 5 and 1 x 10 5 more CD4 + EdU + T cells in the lungs on days 6 and 8 post-infection, respectively. Similarly, Setdb2 ff Lyz2 Cre+ mice had a greater number of CD8 + EdU + T cells in the lungs at each time point examined. The greatest difference in cell number was observed on day 8 post-infection (2.60 ± 0.34 x 10 5 vs. 4.44 ± 0.47 x 10 5 ; p<0.01), with nearly 2 x 10 5 more proliferating CD8 + T cells in IAV-infected lungs. A significant increase in the number of CD8 + EdU + T cells was also observed in knockout lungs on day 6 (0.99 ± 0.11 x 10 5 vs. 2.32 ± 0.66 x 10 5 ; p<0.05) and day 10 (2.22 ± 0.44 x 10 5 vs. 3.95 ± 0.52 x 10 5 ; p<0.05) post-infection (Fig 7B).

Setdb2 expression by Mϕ influences antigen-specific cytokine production by CD4 + T cells
Mϕ, along with DCs and B cells, process and subsequently present viral antigen to CD4 + T cells to stimulate the production of cytokines with antiviral and/or immunomodulatory activity. This prompted us to ask whether Setdb2 expression by Mϕ influences antigen presentation and downstream T cell responses. Since this process is dependent on the expression of MHC class II, as well as several co-stimulatory molecules, we initially examined cell surface markers on IFN-I-stimulated BM-Mϕ by flow cytometry. The level of MHC class II, CD40, and CD86 expression on the cell surface was comparable in both groups. In contrast, changes in IFNAR and CD80 expression were observed in the absence of Setdb2. Based on MFI, IFNAR was reduced (959.0 ± 7.0 vs. 679.5 ± 10.5; p<0.001) and CD80 was elevated (1868.0 ± 8.0 vs. 2464.8 ± 58.4; p<0.001) in Setdb2 -/-BM-Mϕ stimulated with IFN-I. Despite similar levels on the cell surface, greater than a 2-fold increase in Cd40 and Cd86 transcript was detected in Setdb2 -/-BM-Mϕ stimulated with IFN-I (Fig 5A).
To examine antigen-specific T cell responses, CD4 + T cells from naive OT-II mice were cultured alone or at a 5:1 ratio with either wild-type or Setdb2 -/-BM-Mϕ stimulated with IFN-I in the presence of ovalbumin peptide. When co-cultured with either control or knockout BM-Mϕ, CD4 + T cells produced more IL-2, IL-5, IL-10, and IL-17 (Fig 8A). While the concentration of IL-5 and IL-17 was comparable between groups, the absence of Setdb2 in BM-Mϕ was associated with enhanced IL-2 and IL-10 production (Fig 8A). In contrast, IFN-γ Setdb2 Regulates Macrophages in Viral Infection production was only enhanced in co-cultures containing Setdb2 -/-BM-Mϕ (Fig 8A). This correlated with a 4-fold increase in IFN-γ relative to CD4 + T cells cultured alone or with control BM-Mϕ. IL-4 was the only cytokine examined that was unaltered in all experimental conditions ( Fig 8A).

Discussion
This study uncovered a novel role for a histone-modifying enzyme in regulating immunity in acute respiratory viral infection. We have demonstrated IFN-I-dependent induction of Setdb2 in myeloid cell, most notably alveolar Mϕ and inflammatory monocytes, controls the severity of IAV infection by suppressing innate and adaptive immune responses.
Robust production of IFN-I is required to control infection with several viral pathogens. IFN-I signaling results in the transcription of ISGs involved in antiviral immunity. We identified IFN-I as a potent inducer of Setdb2 expression in Mϕ. The induction of Setdb2 by IFN-I was dependent on the JAK-STAT pathway, as blocking JAK1 or STAT1 activity diminished expression. IFN-γ and IFN-λ also signal through STAT1; however, these cytokines had little to no effect on Setdb2 expression, suggesting a downstream molecule in the signaling cascade drives IFN-I-dependent expression. In addition to STAT1, the transcription factor IRF7 regulated Setdb2 expression in in BM-Mϕ. While both STAT1 and IRF7 bound to the Setdb2 promoter, it is unclear if the transcription factors independently regulate transcription or work in concert for optimal expression. It is somewhat unexpected that IFN-λ did not upregulate

Setdb2 Regulates Macrophages in Viral Infection
Setdb2 since IFN-I and IFN-III have overlapping roles in viral infection and despite signaling through different receptors, initiate a similar signaling cascade resulting in ISGF3 activation [34]. The degree of Setdb2 transcription correlated with the concentration of IFN-I, suggesting the extent of expression may be a useful biomarker for determining the severity of infection and autoimmune diseases that result in copious IFN-I production.
Alveolar Mϕ are the first line of defense against inhaled pathogens and are required to control IAV infection [2][3][4][5]. We found that the majority of alveolar Mϕ in IAV-infected lungs expressed Setdb2 and mice deficient for Setdb2 in Mϕ displayed prolonged survival when challenged with a lethal dose of IAV. One mechanism by which alveolar Mϕ control infection is through the production of IFN-I, which, in turn, induces the expression of antiviral effectors that inhibit viral replication and dissemination. It is well documented that IFN-I-driven induction of Mx1, a dynamin-like GTPase that blocks viral transcription and replication, is critical for controlling IAV infection [35]. In addition to Mx1, the ubiquitin-like protein ISG15 has been shown to have an important role in protecting cells from viral pathogens. In IAV infection, ISG15 limits viral replication by binding to the N-terminal RNA-binding domain of NS1, blocking nuclear import [36]. Setdb2 selectively repressed the expression of Mx1 and Isg15 in BM-Mϕ stimulated with IFN-I and accelerated viral clearance in Setdb2 ff Lyz2 cre+ lungs correlated with enhanced transcription of both genes. Thus, by regulating the expression of specific ISGs, Setdb2 dictated the severity of infection. Since the selectivity of the identified Setdb2-regulated antiviral genes is not limited to IAV, it is plausible that Setdb2 controls the resolution of infection caused by other viral pathogens.
IFN-I upregulated Setdb2 in dose-dependent manner and, in turn, Setdb2 regulated the amplitude of the IFN-I response. We propose three mechanisms by which Setdb2 may repress the expression of IFN-I and downstream ISGs. First, Setdb2 regulated the expression of key transcription factors in the IFN-I signaling cascade, including Stat1 and Irf7. Second, Setdb2 may control the antiviral response by regulating the expression of PRRs since viral sensing by the innate immune system results in cytokine and chemokine production. Enhanced expression of the RNA helicases Ddx58 and Ifih1, as well as downstream genes including Tbk1, Mavs, Myd88, Irf7, among others was observed in Setdb2 -/-BM-Mϕ This finding suggests Setdb2 suppresses these genes to weaken the IFN-I response. Third, in addition to targeting viral proteins, ISG15 can bind to IFN-associated transcription factors and antiviral effectors to enhance innate immunity [37][38][39], suggesting Setdb2-dependent regulation of ISG15 may reduce the stability of components involved in IFN-I production and downstream responses. Although IFN-I is essential for controlling IAV infection, its production must be tightly regulated to prevent tissue damage. Multiple genes involved in the suppression of IFN-I were upregulated in parallel with Ifna2 and Ifnb1 in Setdb2 -/-BM-Mϕ. One example is Pin1 (Peptidyl-prolyl cistrans isomerase NIMA-interacting 1), which negatively regulates the antiviral response by promoting the degradation of IRF3 [40]. The transcription factor Foxo3 (Forkhead box O3) negatively regulates the expression of antiviral genes by forming a regulatory circuit involving IRF7 and IFN-I [41]. The expression profile of antiviral genes in Setdb2 -/-BM-Mϕ closely mimicked Foxo3 -/-BM-Mϕ. This includes enhanced expression of Ccl5, Irf7, Ifnb1, Ddx58, Stat1, as well as additional ISGs suggesting Setdb2 may have a similar role in balancing the beneficial and detrimental consequences of IFN-I.
In addition to IFN-I, Setdb2 influenced the expression of proinflammatory cytokines and the immunoregulatory cytokine IL-10. Of the cytokines examined, TNF-α was the only one repressed in Setdb2 -/-BM-Mϕ. This finding is of interest based on reports demonstrating cross-regulation of IFN-I and TNF-α. It is unclear whether Setdb2 promotes TNF-α expression directly or indirectly through other proteins that regulate transcription and/or translation. Since tri-methylation of H3K9 imprints a repressive mark in chromatin, it is unlikely that Setdb2 directly induces Tnf transcription. Indirect regulation of TNF-α may be cytokinedependent, as reduced TNF-α was associated with overexpression of IFN-I and IL-10. While it is unclear if IFN-I itself can directly suppress TNF-α expression, IL-10 dampens expression by inhibiting the activation of NF-κB [42]. Another possibility is that Setdb2 influences TNF-α expression by regulating proteins that repress transcription. For example, Twist proteins bind to the Tnf promoter to inhibit transcription and IFN-I suppresses expression by activating the receptor tyrosine kinase Axl upstream of Twist1/2 [43]. Finally, IFNs can suppress TNF-α expression in Mϕ by promoting tristetraprolin-mediated mRNA decay in a STAT1-and p38-dependent manner, suggesting Setdb2 may negatively regulate the expression of tristetraprolin or similar proteins [44]. Although IAV infection results in robust TNF-α production, it is unnecessary for viral clearance. Rather, TNF-α, along with Nos2, is the major culprit of immunopathology [45,46]. Furthermore, imbalanced production of either IFN-I or TNF-α is linked to autoimmunity. Rheumatoid arthritis patients, as well as children with chronic arthritis being treated with TNF-α antagonists can develop lupus-like symptoms due to overexpression of IFN-α and ISGs [47,48]. Therefore, targeting Setdb2 in parallel with anti-TNF-α therapy may be beneficial for repressing exaggerated IFN-I activity in autoimmunity.
Mortality in individuals exposed to highly pathogenic strains of IAV is often due to lung injury, rather than uncontrolled viral replication. Infection caused by respiratory pathogens that target epithelial cells results in significant cell death and as a consequence, the airways become clogged with dead cells, cellular debris, surfactant material, and virus. Respiratory failure characterized by defective gas exchange and fatal hypoxia is observed in mice lacking alveolar Mϕ [21], highlighting the importance of sustained Mϕ viability in infection. Consistent with reports demonstrating that the level of morbidity in IAV infection is dependent on the number of resident Mϕ [49,50], a greater number of alveolar Mϕ in Setdb2 ff Lyz2 cre+ mice correlated with less damage to the airways and prolonged survival. There are several potential mechanisms by which Setdb2 may dictate the number of alveolar Mϕ. First, fewer alveolar Mϕ in control mice may indicate Setdb2 regulates genes that promote cell death. In IAV infection, the majority of resident Mϕ are depleted by one week due to necrosis [51], indicating Setdb2 expression by alveolar Mϕ may facilitate the necrotic process. Elevated Ccl5 expression may promote Mϕ survival since the CCL5-CCR5 nexus inhibits apoptosis. Accordingly, mortality is increased in Ccl5 -/or Ccr5 -/in IAV infection [19]. Second, Setdb2 may inhibit Mϕ proliferation by suppressing local GM-CSF production [52]. Third, since alveolar Mϕ can arise from blood monocytes [53], enhanced infiltration in the absence of Setdb2 may re-populate the number of alveolar Mϕ in IAV infection. Furthermore, since the removal of apoptotic cells, cellular debris, and surfactant material limits tissue damage, Setdb2 may control the extent of injury by diminishing the phagocytic capacity of Mϕ. Collectively, Setdb2 expression by Mϕ may control IAV-induced lethality by regulating airway integrity and minimizing tissue damage.
Alveolar Mϕ are a major source of chemokines following infection and as a result, promote the infiltration of inflammatory cells to the lungs. The CCR2-CCL2 axis regulates the emigration of monocytes from the bone marrow and subsequent recruitment to infectious sites [54]. An enhanced number of CCR2 + monocytes correlated with more CCL2 in Setdb2 ff Lyz2 Cre+ lungs and Setdb2 directly regulated Ccl2 transcription in BM-Mϕ However, despite this clear correlation, it does not exclude the possibility that Setdb2 -/cells are highly potent in chemotaxis. In tissue, monocytes differentiate into DCs and exudate Mϕ and these cells are linked to severe disease due to overwhelming proinflammatory cytokine production [45,46,55,56]. In contrast, we found that enhanced monocyte recruitment was associated with prolonged survival. How knockout mice avoided immunopathology needs to be further explored. Although the aforementioned studies implicate monocyte-derived cells as the culprit of IAV-induced tissue damage and death, others have shown that tissue Mϕ can restore lung homeostasis and limit injury by developing an immunoregulatory phenotype that resembles that of alveolar Mϕ [57]. In addition to monocytes, neutrophils cause significant tissue damage in IAV infection. It has been proposed that IFN-I-dependent generation of monocytes attenuates neutrophil infiltration and as a consequence, reduces tissue damage [58]. Consistent with a recent study, the absence of Setdb2 in BM-Mϕ resulted in increased expression of neutrophil chemoattractants [59]. However, neutrophil recruitment was comparable in control and Setdb2 ff Lyz2 Cre+ lungs. This inconsistency may be due to the model system, as neutrophils are more critical in bacterial infection. Moreover, the mouse strain used may account for the discrepancy. Whereas they used mice deficient for Setdb2 in all cells, our mice specifically lacked Setdb2 in monocytes, Mϕ, and granulocytes.
In addition to their central role in innate immune responses, Mϕ are involved in the initiation and maintenance of the adaptive immunity. Resident and recruited Mϕ populations can present antigen to CD4 + T cells and we showed that Setdb2 expression by Mϕ suppresses T cell proliferation and cytokine production. In IAV infection, Th1 cells are characterized by co-production of IFN-γ and IL-10 [60,61] and CD4 + T cells cultured with Setdb2 -/-BM-Mϕ secreted more of both cytokines. In addition to IFN-γ and IL-10, CD4 + T cells produced more IL-2, which likely contributes to enhanced T cell proliferation in infected lungs. Whether Setdb2 expression regulates CD4 + T cell responses through antigen presentation is unclear. Several lines of evidence suggest that Setdb2 may regulate the magnitude of T cell responses indirectly through altered cytokine and chemokine production. The polarization of naïve CD4 + T cells to Th1 cells is dependent on IL-12 and Setdb2 -/-BM-Mϕ secreted more IL-12p40 than control cells. Moreover, IFN-I and CCL5 can upregulate the expression of co-stimulatory molecules on antigen presenting cells, as well as regulate the phenotype of CD4 + T cells by promoting cytokine production and proliferation [62][63][64][65][66]. Since transcription of both genes was enhanced in the absence of Setdb2, heightened IFN-I and CCL5 production may further amplify CD4 + T cell responses. Other lymphoid populations, including CD8 + T cells and B cells, are important for the eradication of virus. Respiratory pathogens can trigger the formation of densely packed clusters of lymphocytes known as inducible bronchus associated lymphoid tissue (BALT). In IAV infection, BALT primes virusspecific T and B cells in the lungs and as a result, accelerates viral clearance [67]. Lung histology revealed the presence of potential lymphoid structures in infected Setdb2 ff Lyz2 cre+ mice, but not control animals. Future studies are required to determine if the accumulation of lymphoid cells in knockout lungs is in fact BALT and if it facilitates prolonged survival.
Understanding why a host would upregulate a histone-modifying enzyme that suppresses antiviral immunity is a challenging question. One possible reason is IAV hijacks host machinery to 'turn on' Setdb2, thereby diminishing IFN-I expression and downstream responses to evade the immune system. By mimicking a sequence in the tail of histone H3, the viral protein NS1 allows H3N2 to utilize transcriptional regulators to suppress the antiviral response [68]. Additionally, NS1 allows IAV to evade the immune system by preventing apoptosis through PI3K (phosphatidylinositide 3-kinase) activation and limiting IFN-β production by inhibiting IRF3 activity [69,70]. Recently, it was demonstrated that IAV escapes the IFN-I response by triggering the production of prostaglandin E 2 by Mϕ, which suppresses both innate and adaptive immunity allowing the virus to replicate more efficiently [71]. It is also plausible that even if Setdb2 suppresses antiviral immunity, it is not dramatic enough to cause significant damage to the host. IAV-associated death is rarely caused by primary infection alone. Rather, secondary bacterial pneumonia is the leading cause of mortality caused by an infectious agent [72]. While the mechanism responsible for enhanced susceptibility to bacterial superinfection is not fully elucidated, IFN-I sensitizes the host to bacterial pneumonia [73][74][75]. Since Setdb2 regulates the IFN-I response in viral infection, it is possible the host has evolved to diminish the antiviral response enough to control viral infection, yet prevent secondary complications. While preparing this manuscript, it was shown that Setdb2 regulates the crosstalk between IFN-I and the NF-κB pathway to control the neutrophil response in bacterial superinfection [59]. Together with our findings, these results implicate Setdb2 as a promising therapeutic target in respiratory viral infection and potentially, in secondary complications and autoimmune diseases linked to IFN-I activity.

Ethics statement
All animal procedures were approved by the University Committee on the Use and Care of Animals at the University of Michigan (PRO00004191) and done in accordance with the Animal Welfare Act guidelines of the National Institutes of Health. Experiments using human samples were approved by the Institutional Review Board of the University of Michigan (HUM00075841) and conducted in accordance with the principles expressed in the Declaration of Helsinki. Written informed consent was obtained from all adult subjects.

Mice and IAV infection
C57BL/6, 129S5, and Stat1 -/mice were purchased from Taconic (Germantown, NY). B6.129P2-Lyz2 tm1(cre)Ifo /J (Lyz2 cre mice), and B6.Cg-Tg(TcraTcrb)425Cbn/J (OT-II) transgenic mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Setdb2 gene targeted embryonic stem (ES) cell clones EPD0164_4-B10, -E09, and -E12 were obtained from the trans-NIH Knockout Mouse Project (KOMP Repository). The JM8.N4 C57BL/6N ES cell clones [76] carried the knockout first tm1a(KOMP)Wtsi Setdb2 allele [77]. The ES cells were expanded in cell culture and chromosome counts were performed. Correct targeting of the Setdb2 gene was confirmed by genetic analysis of DNA from the ES cell clones. Germline transmission of the Setdb2 tm1a allele was obtained by breeding ES cell-mouse chimeras produced by the microinjection of C57BL/6/BrdCrHsd-Tyr c albino C57BL/6 blastocysts with the ES cell clones. Chimeras were mated with FLPo recombinase mice to remove the drug selection cassette and produce mice carrying the conditional floxed Setdb2 tm1c allele. C57BL/6-Tg(CAG-Flpo) 1Afst /Mmucd FLPo recombinase mice [78] were obtained from the Mutant Mouse Resource and Research Centers (Stock Number: 032247-UCD). FLPo mice were backcrossed onto albino C57BL/6 mice so that coat color selection could be used to chimeras to identify germline transmission and maintain an inbred C57BL/6 genetic background. The resulting progeny with a floxed Setdb2 allele were bred with Lyz2 cre mice to generate control and mice deficient for Setdb2 in monocytes, Mϕ, and granulocytes [79]. Setdb2 ff Lyz2 cre and Setdb2 LacZ reporter mice were bred in-house and genotyped with custom primers (S1 Table). For infection, the IAV strain A/PR8/34; H1N1 isotype was used (ATCC). Age-matched female mice were inoculated intranasally with 1 x 10 4 PFU for sublethal infection and 1 x 10 5 PFU for lethal infection.

Cell isolation
Murine CD11b microbeads, human CD14 microbeads, and the murine CD4 + T cell Isolation Kit were purchased from Miltenyi Biotec. Magnetic separation yielded 95% purity of each population. PBMCs were isolated from the blood using Ficoll (GE Healthcare).

BM-Mϕ and human Mϕ differentiation
For the generation of BM-Mϕ, bone marrow was differentiated in L929 cell-conditioned media as previously described [80]. For the differentiation of human of Mϕ, CD14 + monocytes were cultured in complete medium supplemented with 50 ng/mL of M-CSF (R&D systems) for one week. Adherent cells were washed and harvested with Trypsin/EDTA (Lonza).

Epithelial cell culture
Murine AECs were isolated from naïve mice as previously described [81]. Briefly, Dispasedigested lungs were depleted of CD16/CD32 + and CD45 + cells using biotinylated antibodies (BD Biosciences) and anti-biotin microbeads (Miltenyi Biotec). Non-adherent cells were cultured in fibronectin-coated wells and AECs were harvested after 4 days. NHBEs and the BEGM BulletKit were purchased from Lonza. NHBEs were cultured in 25 cm 2 flasks following the manufacturers recommendations. NHBEs were subcultured using Trypsin/EDTA when 80% confluent.

Cytokine and tofacitinib treatment
Murine IFN-α, IFN-β, IFN-γ, IFN-λ2, and IFN-λ3, as well as human IFN-α and IFN-β were purchased from R&D Systems. For IFN-I treatment, cells were given 10 units/mL of IFN-α and IFN-β unless otherwise noted. Cells were treated with 10 ng/mL of IFN-II (IFN-γ) or IFN-III (IFN-λ2 and IFN-λ3). For JAK inhibition, cells were treated with 50 nM tofacitinib (Cayman Chemical) at the time of stimulation. siRNA IRF3 and IRF7 siRNA were purchased from Santa Cruz Biotechnology. Non-targeting siRNA was purchased from GE Dharmacon. BM-Mϕ were transfected with siRNA using the Amaxa Mouse Mϕ Nucleofactor Kit (Lonza). Transfected cells were cultured for 18 hours before stimulation.

Mϕ-T cell co-culture
CD4 + T cells isolated from the spleens of naïve OT-II mice were cultured alone or with BM-Mϕ at a 5:1 ratio. For cytokine analysis, co-cultures were incubated for 48 hours in the presence of 10 ng/mL of ovalbumin peptide (Peptides International).

PCR arrays
The murine/human epigenetic chromatin-modifying enzyme and murine antiviral PCR arrays were purchased from SABiosciences. RNA was DNase-digested using the RNeasy Mini Kit and reverse transcribed with the RT 2 First Strand Kit (Qiagen). RT-PCR was performed according to the manufacturer's instructions and gene expression was normalized to multiple housekeeping genes.

Cytokine and chemokine production
The concentration of CCL2 was measured using the mouse CCL2 ELISA Ready-SET-Go! (eBioscience). All other cytokines were quantified using a Bio-Plex 200 (Bio-Rad Laboratories).

ChIP
A total of 1 x 10 7 BM-Mϕ were treated with a vehicle control or cytokine for 24 hours and ChIP was performed as previously described [24]. DNA was fragmented by sonication using a Branson Sonifier 450 (Branson Ultrasonics). For immunoprecipitation, the following antibodies were used: IRF7 (Santa Cruz Biotechnology), STAT1 (Abcam), H3K9me3 (Abcam), Setdb2 (Dr. Yali Dou, University of Michigan), and rabbit polyclonal IgG (Millipore). DNA was assessed by RT-PCR using custom primers (S2 Table).

STAT1 and NF-κB expression
For transcription factor analysis, whole-cell lysates were collected from a total of 1 x 10 7 BM-Mϕ treated with a vehicle control or IFN-I for 30 minutes. The concentration of protein in each sample was adjusted to a concentration of 100 μg/mL in assay buffer. Total and phosphorylated protein was measured using the following kits from abcam: STAT1 (pY701) + total STAT1 ELISA Kit, STAT1 (pS727) + total STAT1 ELISA Kit, and NF-κB p65 (pS536) + total NF-κB p65 SimpleStep ELISA Kit.

Histology
Lungs were inflated with 10% formalin and processed using routine histological techniques. Tissue sections were stained with H&E and visualized by light microscopy.

Statistical analysis
For the survival study, the p value was determined by a log-rank survival test. Analysis of the antiviral gene profile in BM-Mϕ was characterized by one-way ANOVA. Differences for remaining experiments were analyzed by Student's t-test or two-way ANOVA. A p value of 0.05 was considered significant.
Supporting Information S1