Resistance to lethal ectromelia virus infection requires Type I interferon receptor in natural killer cells and monocytes but not in adaptive immune or parenchymal cells

Type I interferons (IFN-I) are antiviral cytokines that signal through the ubiquitous IFN-I receptor (IFNAR). Following footpad infection with ectromelia virus (ECTV), a mouse-specific pathogen, C57BL/6 (B6) mice survive without disease, while B6 mice broadly deficient in IFNAR succumb rapidly. We now show that for survival to ECTV, only hematopoietic cells require IFNAR expression. Survival to ECTV specifically requires IFNAR in both natural killer (NK) cells and monocytes. However, intrinsic IFNAR signaling is not essential for adaptive immune cell responses or to directly protect non-hematopoietic cells such as hepatocytes, which are principal ECTV targets. Mechanistically, IFNAR-deficient NK cells have reduced cytolytic function, while lack of IFNAR in monocytes dampens IFN-I production and hastens virus dissemination. Thus, during a pathogenic viral infection, IFN-I coordinates innate immunity by stimulating monocytes in a positive feedback loop and by inducing NK cell cytolytic function.


Introduction
Type I interferons (IFN-I) are a family of highly conserved antiviral cytokines. IFN-I bind to the IFN-I receptor IFNAR, a heterodimer composed of IFNAR1 and IFNAR2. Binding of IFN-I to IFNAR induces a signaling cascade that culminates in the upregulation of interferonstimulated genes (ISG) [1]. ISGs induce an antiviral state in cells that curtail viral replication [2]. IFN-I is also thought to activate innate and adaptive immune responses through direct and indirect IFNAR signaling. Consequently, resistance to many viral pathogens requires IFNAR as determined by infection of mice deficient in IFNAR (Ifnar1 -/-) [3][4][5] and increased adverse reactions to attenuated viral vaccines observed in humans with inherited IFNAR deficiency [6].
The discovery of many mechanisms for IFN-I induction was made in fibroblasts [7]. Also, it is well established that virtually any cell type can produce IFN-I in vitro. However, it is now apparent that in vivo, the cells that mostly produce IFN-I are myeloid lineage, such as conventional dendritic cells (cDC), plasmacytoid dendritic cells (pDC), or Ly6C high inflammatory monocytes (iMOs) [8]. Similarly, most cell types express IFNAR, and when exposed to IFN-I in vitro, upregulate ISGs and become more resistant to viral infection [2]. Therefore, the general thought is that the stimulation of IFNAR by IFN-I is protective as it can induce ISGs and the antiviral state in virtually any cell, and because it activates multiple innate and adaptive immune cell types [9]. However, whether this is true in vivo has not been thoroughly studied.
Ectromelia virus (ECTV) is the etiologic agent of mousepox, which causes fulminant hepatitis in mice and resembles human smallpox. C57BL/6 (B6) mice are naturally resistant to lethal mousepox, and this resistance requires IFN-I, which in the draining lymph node (dLN) is mostly produced by infected iMOs recruited by dendritic cells (DCs) [10,11]. Defects in the IFN-I response in vivo such as the absence of IFNAR [12,13], the DNA sensor Cyclic GMP-AMP synthase (cGAS) [14,15], the signaling adapter Stimulator of Interferon Genes (STING) or the transcription factor Interferon regulatory factor 7 (IRF7) [11], results in unchecked ECTV replication, impairment of the adaptive immune response, and death.
Given that IFN-I signaling is crucial for mousepox resistance, it is an excellent model to uncover which cells need to receive IFN-I signaling through IFNAR to resist a lethal viral infection, and to determine whether hematopoietic and non-hematopoietic cells require intrinsic IFNAR. This question is especially intriguing in the context of hepatocytes, which are non-hematopoietic and a key target of ECTV pathogenesis. Furthermore, considering that resistance to mousepox needs a robust cytotoxic response by both innate Natural Killer (NK) and adaptive T cells, it is important to investigate whether resistance to viral disease requires intrinsic IFNAR in these cell types. Also, given the critical role of iMOs in IFN-I production, it is of interest to define the role of IFNAR in these cells.

IFNAR in hematopoietic cells is necessary and sufficient for resistance to lethal mousepox
To test whether hematopoietic or non-hematopoietic cells require IFNAR signaling for resistance to mousepox, we lethally irradiated B6 and Ifnar1 -/mice and reconstituted them with either B6 or Ifnar1 -/bone marrow cells to generate B6! B6, Ifnar1 -/-! B6, B6! Ifnar1 -/and Ifnar1 -/-! Ifnar1 -/bone marrow chimeras (BMC). Following ECTV infection in the footpad, all Ifnar1 -/-! B6 and Ifnar1 -/-! Ifnar1 -/mice succumbed rapidly to the infection, whereas most B6! Ifnar1 -/and B6! B6 mice survived (Fig 1A). IFNAR deficiency restricted to radioresistant cells resulted in only mild susceptibility to mousepox, given that~70% of B6! Ifnar1 -/mice survived. The finding that most B6! Ifnar1 -/mice were resistant to mousepox was surprising because the radioresistant cell compartment includes hepatocytes, which are an important target of ECTV [16], and epidermal Langerhans cells (LCs), which are required for optimal innate immune responses to ECTV in the draining lymph node (dLN) and resistance to mousepox [17]. Thus, we also tested whether the surviving B6! Ifnar1 -/mice had a viral Positive control: liver sample harvested from infected Ifnar1 -/mice at 5 dpi. (C-D) ECTV titers in the livers and spleens of each BMC group were quantified by plaque assay at 5 (C) and 7 dpi (D). The dashed line indicates the detection limit. (E) Number of live splenocytes in BMCs naïve or at 5 and 7 dpi with ECTV. Data are represented as mean ± SEM from three pooled independent experiments (N = 11 per group, multiple comparison ANOVA with Tukey correction). (F) Isg15 transcripts in the spleen at 5 dpi quantified by qPCR, normalized to Gapdh, and adjusted to fold-change to naïve. Data are represented as mean ± SEM from two independent experiments (N = 6 per group, multiple comparison ANOVA with Tukey correction). (G-H) Cre-driven excision of Ifnar1 gene exon10 in the livers (G) and the spleens (H) of the indicated naïve mice measured by qPCR, normalized by Gapdh transcription, and adjusted to fold-change to B6. (I) Survival of the indicated mice infected as in (A). Data pooled from two independent experiments (Log-rank Mantel-Cox compared to Ifnar1 fl/fl mice). (J-K) ECTV titers in the livers and spleens of indicated mice at 5 (J) and 7 dpi (K) quantified as in (C-D). Data are represented as mean ± SEM from two or three pooled independent experiments (N = 8-9 for each Cre + mice and N = 12-18 for Ifnar1 fl/fl mice, ANOVA with Tukey correction compared to Ifnar1 fl/fl mice).
We also found that compared to B6! B6 and B6! Ifnar1 -/mice, the viral loads in Ifnar1 -/-! B6 and Ifnar1 -/-! Ifnar1 -/mice were increased 10-fold at 5 days post-infection (dpi) in the spleen (Fig 1C) and 100-fold in the liver and the spleen at 7 dpi (Fig 1D). The uncontrolled viral replication observed in Ifnar1 -/-! B6 and Ifnar1 -/-! Ifnar1 -/mice correlated with high pathogenicity in the spleen with a 50 to 100-fold reduction in the number of splenocytes at 7 dpi when compared to B6! B6 and B6! Ifnar1 -/mice (Fig 1E). The presence of IFNAR in the hematopoietic compartment was critical for the induction of ISGs because at 5 dpi Isg15 mRNA, as a readout for IFN-I activity, increased to similar levels in the spleens of B6! B6 and B6! Ifnar1 -/mice, whereas there was marginal or no Isg15 upregulation in the spleens of Ifnar1 -/-! B6 and Ifnar1 -/-! Ifnar1 -/mice (Fig 1F) at a timepoint when numbers of splenocytes were similar among all analyzed groups (Fig 1E). The low upregulation of Isg15 in Ifnar1 -/-! B6 mice also indicates that irradiation successfully eliminated most IFN-responding cells in the spleens of Ifnar1 -/-! B6 mice.
To confirm our findings in another system, we crossed Ifnar1 fl/fl mice [18] with mice expressing Cre recombinase from the Vav1 promoter [19] to delete IFNAR in all hematopoietic cells (Ifnar1 ΔVav1 mice) or from the Alb promoter [20] to delete IFNAR exclusively in hepatocytes (Ifnar1 ΔAlb1 mice). The specificity of the deletion was confirmed by quantitative PCR (qPCR) for Ifnar1 transcripts in the liver (Fig 1G) and the spleen (Fig 1H). In agreement with the experiments with BMC, all infected Ifnar1 ΔVav1 mice succumbed rapidly to infection while most Ifnar1 ΔAlb1 and Ifnar1 fl/fl littermates survived (Fig 1I). Moreover, when compared to Ifnar1 fl/fl littermates, Ifnar1 ΔVav1 but not Ifnar1 ΔAlb1 mice had a significant increase in viral loads in the livers and spleens at 5 (Fig 1J) and 7 dpi (Fig 1K). Based on these results, we conclude that IFNAR in hematopoietic cells is essential and generally sufficient for ECTV control and resistance to lethal mousepox, but is mostly dispensable in hepatocytes, which are critical targets of ECTV.

IFNAR in hematopoietic cells is necessary for efficient T cell responses to ECTV
Resistance to mousepox requires functional T and B cells [21][22][23][24][25]. Thus, we analyzed the T cell responses in B6/Ifnar1 -/-BMC. Notably, Ifnar1 -/-! B6 and Ifnar1 -/-! Ifnar1 -/had a significant reduction in the frequency (>10 fold) and absolute numbers (100-to 1000-fold) of CD8 and CD4 T cells expressing granzyme B (GzmB), a marker of activation and cytotoxic potential (Fig 2A and 2B), and CD44, a marker of antigen-experienced T cells (Fig 2A and 2C) when compared to B6! B6 mice. In contrast, in B6! Ifnar1 -/mice, the CD8 T cell responses were not significantly affected, and those of CD4 T cells were significantly but only slightly reduced. This indicates that IFNAR in hematopoietic cells is necessary for efficient T cell responses to ECTV.

Intrinsic IFNAR on adaptive lymphocytes is dispensable for efficient T cell responses and resistance to mousepox
The data above indicated that IFNAR is required in hematopoietic cells for T cell responses to ECTV but did not indicate whether this requirement is intrinsic or extrinsic for T cells. To test for these possibilities, we adoptively transferred CFSE-labeled WT F1 (B6.CD45.1 x B6. CD45.2) and Ifnar1 -/-(CD45.2) cells into IFNAR-sufficient mice (B6.CD45.1). Subsequently, we infected them with ECTV in the footpad and harvested their spleens at 8 dpi (Fig 3A). In this context, where the host efficiently controls virus replication and spread, Ifnar1 -/and WT CD8 and CD4 T cells responded similarly to the infection as determined by CFSE dilution (Fig 3B and 3C). The frequencies of cells expressing CD44 or GzmB (Fig 3D and 3E), and the expansion of CD8 T cells specific for the poxvirus immunodominant K b -restricted epitope TSYKFESV [26] were also similar (Fig 3D and 3F).

IFNAR on NK cells is essential for resistance to mousepox
Resistance to mousepox requires NK cell cytolytic activity [28][29][30][31]. To test whether NK cell antiviral function requires intrinsic IFNAR, we crossed mice expressing Cre as a knock-in in the Ncr1 locus (NKp46 Cre ) with Ifnar1 fl/fl mice to generate Ifnar1 ΔNKp46 mice, which directs Cre to NK cells and other group 1 innate lymphoid cells (ILCs). By flow cytometry, anti-IFNAR1 monoclonal antibody (mAb) stained T cells and NK cells in Ifnar1 fl/fl mice, T cells but not NK cells in Ifnar1 ΔNKp46 mice, and neither T cells nor NK cells in Ifnar1 ΔVav1 mice (Fig  4A), demonstrating the NK cell-specific and widespread deficiency of IFNAR1 in Ifnar1 ΔNKp46 and Ifnar1 ΔVav1 mice, respectively. After ECTV infection, most Ifnar1 ΔNKp46 mice succumbed to mousepox (Fig 4B), with a median survival time of 9.5 days, while most Ifnar1 fl/fl littermates

PLOS PATHOGENS
Type I interferon receptor in resistance to lethal ectromelia virus infection survived, indicating that intrinsic IFNAR signaling in NK cells is vital for optimal resistance to mousepox.
To understand how the lack of intrinsic IFNAR signaling in NK cells predisposes mice to ECTV lethality, Ifnar1 ΔNKp46 , Ifnar1 ΔVav1 , and Ifnar1 fl/fl mice were infected with ECTV and analyzed at 6 and 8 dpi. Viral loads in the liver and spleen of Ifnar1 ΔNKp46 and Ifnar1 fl/fl mice were similar and significantly lower than in Ifnar1 ΔVav1 mice, suggesting that during the first 6 dpi, intrinsic IFNAR signaling in NK cells may not be essential to control ECTV but could be necessary in other hematopoietic cells (Fig 4C). The NK cells' intrinsic IFNAR defect resulted in increased viral loads in the spleen and the liver at 8 dpi (Fig 4D). Compared to infected Ifnar1 fl/fl mice, the numbers of splenocytes were significantly reduced in ECTV-infected Ifnar1 ΔVav1 but not in Ifnar1 ΔNKp46 at 6 dpi, but at day 8 dpi, Ifnar1 ΔNKp46 had reduced numbers of live splenocytes compared to Ifnar1 fl/fl mice (Fig 4E). Notably, at 6 dpi, the frequency and absolute numbers of NK cells were significantly lower in the spleens of Ifnar1 ΔVav1 compared to Ifnar1 ΔNKp46 and Ifnar1 fl/fl mice, while at 8 but not 6 dpi, there were significant decreases in the frequency and numbers of NK cells in Ifnar1 ΔNKp46 mice compared to Ifnar1 fl/fl mice (Fig 4F and 4G).
Considering that the mean time to death in Ifnar1 ΔNKp46 mice infected with ECTV was 9.5 days, we analyzed the T cell response at 8 dpi in the spleens of these mice as a sign of pathogenicity. We found that the frequencies of antigen-experienced (CD44 + ), ECTV-specific CD8 (CD44 + K b -TSYKFESV) and activated cytotoxic CD8 and CD4 (CD44 + GzmB + ) T cells were similar for Ifnar1 ΔNKp46 and Ifnar1 fl/fl mice (Fig 4H), suggesting that, as expected, IFNAR deficiency in NK cells does not directly compromise T cell activation in vivo. However, given the reduced numbers of live splenocytes in ECTV-infected Ifnar1 ΔNKp46 mice at 8 dpi, the total number of antigen-experienced CD8 T cells (Fig 4I), of TSYKFESV-specific CD8 (Fig 4J), and GzmB + CD8 T cells (Fig 4K) was reduced in Ifnar1 ΔNKp46 mice compared to Ifnar1 fl/fl mice.

NK cells require IFNAR for optimal maturation, activation, and cytolytic killing
ECTV-infected Ifnar1 fl/fl and Ifnar1 ΔNKp46 mice had no differences in frequency of NK cells that expressed the activation marker CD69, the receptors NKG2D and CD94, which are required to resist to mousepox [28,29], or the maturation marker KLRG1 [32] There were also no differences in the mean fluorescence intensity (MFI) of NK cells that expressed the non-classical Major Histocompatibility Class I (MHC-I) molecule Qa-1 b , which has been suggested to increase NK cell survival [33] (S1A Fig). However, we observed higher frequencies of immature CD27 + CD11b -NK cells [34] in Ifnar1 ΔNKp46 compared to Ifnar1 fl/fl mice (S1B Fig), suggesting a defect in the ability of NK cells to mature in response to ECTV infection in the absence of intrinsic IFNAR signaling.
We have previously shown that NK cells upregulate GzmB after infection with ECTV [28]. We now find that different to T cells, the upregulation of GzmB in NK cells in response to ECTV is partly dependent on intrinsic IFNAR because at 6 dpi a smaller fraction of NK cells upregulated GzmB in Ifnar1 ΔNKp46 compared to Ifnar1 fl/fl infected mice (Fig 5A, left). Also, using a mAb to perforin (Prf), we found that most NK cells produce intermediate levels of Prf at steady state. At 6 dpi with ECTV, Prf expression was upregulated in NK cells, many of them to high levels. Notably, most NK cells also upregulated Prf in Ifnar1 ΔNKp46 mice, but several remained at an intermediate stage, similar to naïve Ifnar1 fl/fl (Fig 5A, right). Consequently, at 6 dpi, a significantly lower fraction of NK cells was GzmB high Prf high in Ifnar1 ΔNKp46 compared to Ifnar1 fl/fl mice (Fig 5B and 5C).
The previous data suggested a possible defect in NK cell cytolytic function in the absence of intrinsic IFNAR signaling. TAP1-deficient splenocytes (Tap1 -/-) adoptively transferred into naïve mice are killed by NK cells due to their low expression of MHC-I at the cell surface [35,36]. We have previously shown that the preferential killing of Tap1 -/over WT splenocytes is significantly higher in mice infected with ECTV for 6 days as compared to naïve mice [37]. Notably, we found that NK cells were significantly less effective at killing Tap1 -/cells in vivo in Ifnar1 ΔNKp46 mice than in Ifnar1 fl/fl mice following ECTV infection (Fig 5E and 5F). These results indicate that NK cell intrinsic IFNAR signaling is necessary for optimal NK cell killing during an acute viral infection and required for host protection and survival.

Intrinsic IFNAR in Lysozyme-expressing myeloid cells is necessary for resistance to mousepox
The data above demonstrates that intrinsic IFNAR is required in NK cells but not in T cells for efficient ECTV control and survival to mousepox. Nevertheless, it did not escape our attention that Ifnar1 ΔVav1 succumbed faster to ECTV ( ���� P<0.0001) and suffered much higher virus loads than Ifnar1 ΔNKp46 mice. This observation suggested that in addition to NK cells, other innate immune cells require intrinsic IFNAR signaling to resist to mousepox. We have previously shown that iMOs [11] and LCs, but not conventional dendritic cells 1 (cDC1) [17], play vital roles in the development of the innate immune response in the dLN, which is required to restrain early virus spread and to resist lethal mousepox. However, it was unlikely that intrinsic IFNAR signaling was required in LCs since they are radioresistant and B6! Ifnar1 -/mice were resistant to mousepox (Fig 1A). MO, macrophages (Mf), and neutrophils express Lysozyme C-2, encoded by the Lyz2 gene (ImmGen ULI RNASeq and ImmGen MNP OpenSource). Thus, we crossed Ifnar1 fl/fl mice with mice expressing Cre from the Lyz2 promoter [38] to generate Ifnar1 ΔLyz2 mice. Consistent with previous reports [39,40], flow cytometric analysis of splenocytes from naïve Ifnar1 ΔLyz2 mice showed that IFNAR was partially deleted in cells of the MO/Mf lineage and also in neutrophils but unaffected in T cells, B cells, NK cells, cDCs, pDCs and eosinophils (Fig 6A). When we subdivided MOs according to Ly6C expression, we found that most Ly6C hi but not Ly6C int or Ly6C low MO lost expression of IFNAR in Ifnar1 ΔLyz2 mice (Fig 6B). We also attempted to determine IFNAR in subpopulations of cells in the dLN of infected mice, but, unfortunately, anti-IFNAR1 mAb failed to stain any cells in WT or Ifnar1 ΔLyz2 mice, probably due to the high concentration of type I IFNs in the dLN [11] which compete with anti-IFNAR1 (MAR1-5A3) for binding to the extracellular domain of IFNAR [41].
Strikingly, most Ifnar1 ΔLyz2 mice succumbed to ECTV infection with a mean time of death of 8.5 dpi (Fig 6C), indicating that IFNAR expression in Lyz2 + cells is necessary for optimal resistance to lethal mousepox. When we compared virus replication between Ifnar1 ΔLyz2 and control Ifnar1 fl/fl mice, we found they did not differ in viral loads in the dLN at 1-2 dpi (Fig  6D). At 3 dpi, the virus loads in Ifnar1 ΔLyz2 were slightly higher in the dLN (Fig 6D), but 100-fold higher in the spleen (Fig 6E), indicating faster virus spread. At 5 dpi, Ifnar1 ΔLyz2 had increased virus loads in the liver (Fig 6F) and, at 7 dpi, in both liver and spleen (Fig 6G). Consistent with the pathological changes usually observed in mice susceptible to mousepox, Ifnar1 ΔLyz2 mice had significant reductions in the number of splenocytes, and numbers of CD44 + GzmB + and CD44 + K b -TSYKFESV + CD8 T cells in their spleens (Fig 6H-6J). Therefore, absence of IFNAR in Lyz2 + cells leads to early viral spread from the dLN to target organs, exacerbates viral replication at later stages of infection, and indirectly affects the strength of the T cell response. Of note, the phenotype in Ifnar1 ΔLyz2 was less intense than in Ifnar1 ΔVav1 mice, suggesting that the roles of IFNAR in Lyz2 + , NK cells, and possible other innate immune cells in resistance to mousepox are complementary and additive.

iMOs require intrinsic IFNAR for optimal transcription of IFN-I genes and for efficient Ly6C and MHC-II expression but not to migrate to dLNs, produce CXCL9, or resist infection
The data above shows that in Ifnar1 ΔLyz2 mice, neutrophils and Ly6C + MO partially lose IFNAR expression. We and others have previously shown that depletion of phagocytes, including iMOs, render mice susceptible to mousepox [11,21] and that Ly6C + iMOs play a critical role in the early innate immune response to ECTV in the dLN, which is essential to control early virus spread and resistance to mousepox. On the other hand, neutrophil depletion does not increase susceptibility to ECTV [42]. Furthermore, while Lyz2-cre is known to delete floxed genes in LCs [43] and we have shown that LCs are also critical for the innate immune response to ECTV in the dLN [17], LCs are radioresistant and of host origin in B6Ifnar1 -/mice, which were resistant to mousepox (Fig 1A). Thus, it was unlikely that the reason for the We have previously shown that iMOs are recruited to the dLN at 2-4 dpi and have two critical roles for the control of early systemic virus spread: 1) uninfected iMOs secrete CXCL9, which promotes NK cell recruitment to the dLN [10] and 2) infected iMOs produce type I IFNs [11,17]. Thus, we compared the iMOs response in dLN of Ifnar1 ΔLyz2 and Ifnar1 fl/fl mice at 2 dpi with ECTV-expressing green fluorescent protein (ECTV-GFP) [28] to distinguish uninfected (GFP -) and infected (GFP + ) cells by flow cytometry. We found that Ifnar1 ΔLyz2 and Ifnar1 fl/fl mice did not differ either in the number of iMOs recruited to the dLN (S2A and S2B  Fig), the frequency of infected iMOs (S2A and S2C Fig), the number of uninfected iMOs that produced CXCL9 (S2A and S2D Fig) or the frequency and number of NK cells recruited to the dLN by the CXCL9 produced by infected iMOs (S2A and S2E Fig). These findings indicate that iMOs do not require intrinsic IFNAR to be recruited to the dLN, resist infection, or recruit NK cells.
Because Ly6C + MOs did not fully delete IFNAR in Ifnar1 ΔLyz2 mice, we analyzed the accumulation of iMOs at 3 dpi in the dLN of B6.  (S2I Fig, top). WT but not Ifnar1 -/infected iMOs also trended to upregulate MHC-I after infection, but the differences were not significant (S2I Fig,  bottom).
Our previous work has shown that ECTV infection induces high levels of IFN-I in the dLN of Ifnar1 -/mice. However, Ifnar1 -/mice also endure high virus loads, which may affect the IFN-I response [14]. To test whether intrinsic IFNAR deficiency affects the ability of iMOs to express IFN-I and replicate virus without possible effects of total virus loads in the dLN, we infected B6.CD45.1 + Ifnar1 -/-! F1 [B6.CD45.1 x B6.CD45.2] mixed BMC with ECTV-GFP and at 3 dpi, we sorted infected (GFP + ) and uninfected (GFP -) WT CD45.1 and Ifnar1 -/-CD45.2 iMOs from dLNs of infected mice (Fig 7A). In this context, WT and Ifnar1 -/-iMOs were exposed to the same viral loads within the dLN, and IFN-I production and viral replication was only dependent on IFNAR expression. As baseline controls, we used MOs from the spleens of naïve animals. Using qPCR, we found that compared to naïve MOs, both WT and Ifnar1 -/infected iMOs upregulated IFN-I genes determined as fold-change over naïve using primers specific for Ifnb1, Ifna4, and Ifna5 or for conserved sequences among all Ifna genes except Ifna4 (Ifna non-a4) [7]. However, Ifnar1 -/infected iMOs upregulated the transcription of IFN-I genes to much lower levels than infected WT iMOs (Fig 7B). This major reduction in IFN-I gene expression by infected Ifnar1 -/-iMOs suggests the existence of a positive feedback compared to Ifnar1 fl/fl mice). (D-E) ECTV titers in the dLN (D) and the spleen (E) were quantified by plaque assay at the indicated dpi. The dashed line indicates the detection limit. Data are represented as mean ± SEM from two or three pooled independent experiments (N = 5 to 7 for each Cre + mice and N = 11 to 24 for Ifnar1 fl/fl mice; ANOVA with Tukey correction compared to Ifnar1 fl/fl mice). (F-G) ECTV titers in the liver and spleen were quantified by plaque assay at 5 (F) and 7 dpi (G). The dashed line indicates the detection limit. Data are represented as mean ± SEM from two pooled independent experiments (N = 8 to 9 for each Cre + mice and N = 18 for Ifnar1 fl/fl mice, ANOVA with Tukey correction compared to Ifnar1 fl/fl mice). (H) Representative flow cytometry plots showing levels of CD44 and GzmB expression and levels of K b -TSYKFESV-specific CD8 T cells in the spleens of infected mice. (I) Total cell numbers, and percentages and total numbers of CD44 + GzmB + CD8 T cells (NK1.1 -TCRβ + CD8 + ) in spleens of the indicated mice at 7 dpi. (J) Numbers of CD44 + K b -TSYKFESV-specific CD8 T cells in the spleens of the indicated mice at 7 dpi. Data are represented as mean ± SEM from two pooled independent experiments (N = 8 to 9 for each Cre + mice and N = 18 for Ifnar1 fl/fl mice, ANOVA with Tukey correction compared to Ifnar1 fl/fl mice).
https://doi.org/10.1371/journal.ppat.1009593.g006 loop for in vivo expression of IFN-I which, when impaired, culminates in uncontrolled viral spread from the dLN to target organs and ultimately in death. Notably, sorted infected WT and Ifnar1 -/-iMOs had similar levels of infection when measured as viral transcripts by qPCR (Fig 7C) or as infectious particles/cell by plaque assay (Fig 7D). This finding indicates that intrinsic IFNAR signaling does not inhibit viral replication within infected iMOs. Thus, iMOs require IFNAR to for optimal IFN-I gene transcription but not to be directly protected from viral infection.

Discussion
Here we used a lethal acute mouse model of viral infection to show that IFNAR mediates protection mostly by signaling in hematopoietic cells, as indicated by results obtained with Ifnar1 -/-BMC (Fig 1A) and Cre-LoxP conditional deletions (Fig 1I). The observation that IFNAR expression in hepatocytes does not play a significant role in resistance was unexpected, considering that ECTV infects hepatocytes and causes severe hepatitis. In agreement with our observations, others have shown that compared to WT mice, Ifnar1 ΔAlb1 mice have similar levels of inflammation and virus loads, whereas Ifnar1 ΔLyz2 mice have increased viral loads in the liver after intravenous challenge with the closely related poxvirus vaccinia virus [40], suggesting a conserved mechanism for IFNAR protection against poxviruses. We have previously shown that cGAS, which is required for optimal IFN-I production after ECTV infection, is not necessary in parenchymal cells (including hepatocytes) for resistance to mousepox. We now show that resistance to mousepox is also independent of IFNAR in hepatocytes. Whether these characteristics of hepatocytes is the reason for ECTV tropism to the liver is unknown and would be of interest to explore. Studies using other viral models have shown that IFNAR expression in both hematopoietic and non-hematopoietic cells may contribute to virus control [44][45][46], suggesting that viral tropism and pathogenicity might determine which cells need IFNAR expression for protection. Notably, it was also previously shown that IFNAR expression in hematopoietic cells is detrimental for host recovery from infection with a mouse-adapted strain of SARS-CoV [47]. However, in this same study, it was also reported that early treatment of susceptible mice with IFNβ was protective and prevented mice from developing lethal pneumonia, suggesting that the timing of IFN-I response might be critical and very early for airborne diseases.
Optimal activation of adaptive lymphocytes requires more than one signal. Signal 1 is provided by the activation of the T or B cell receptor whereas Signal 2 is provided by the activation of co-stimulatory molecules. It has been demonstrated that intrinsic IFNAR expression is essential for optimal T cell function in some contexts [18] suggesting that IFN-I provides a necessary Signal 3 [48]. Accordingly, IFN-I stimulation on T and B cells results in the upregulation of activation markers and cytokine secretion [49,50]. However, we observed that, in vivo, Ifnar1 -/-T and B cells were still capable of protecting the host from a primary ECTV infection and of promoting viral clearance when innate immune cells were IFNAR sufficient (Fig 3G and 3H). These findings indicate that T and B cells can exert their full antiviral function independently of IFNAR stimulation and that during a highly inflammatory viral infection such as ECTV, Signal 3 may not be required or may be provided by alternative or redundant mechanisms, such as IL-2, IL-6, IL-12, IL-18 or IFN-γ. In agreement with our observations, it has been shown that conditional deletion of IFNAR in T cells does not affect CD8 T cell responses to murine gammaherpesvirus infection [51]. Moreover, using a mixed BMC approach similar to ours (Fig 3G), it has been shown that the adaptive compartment does not require IFNAR signaling for tumor rejection, whereas the innate immune compartment does [27]. Here we did not explore whether intrinsic IFNAR on adaptive lymphocytes is required for the generation of memory CD8 T cells or for protection against secondary infection. However, others have previously shown that Ifnar1 is not required for resistance to secondary ECTV infection [12].
NK cell function is essential for mousepox resistance, and we have shown here that optimal protection by NK cells requires intrinsic IFNAR (Fig 4B). The impaired upregulation of GzmB that we observed in Ifnar1 -/-NK cells (Fig 5A) also occurs during non-lethal infection with mouse cytomegalovirus [33]. It has been shown that intrinsic IFN-I signaling is necessary for optimal NK cell responses to influenza virus and vaccinia virus [52,53], but how this affects virus control was not assessed. Here we observed not only a reduction in GzmB expression but also a decrease in the NK cell pool that expresses both Prf and GzmB (Fig 5B). The finding that the preferential killing of Tap1 -/cells in ECTV-infected Ifnar1 ΔNKp46 mice was significantly reduced (Fig 5E) highlights the importance of the crosstalk among innate immune cells in the dLN where both iMOs and NK cells migrate to upon infection, and where iMOs express IFN-I. However, IFNAR in iMOs is not required for iMOs or NK cell accumulation in the dLN (S2A-S2E Fig).
There were no consistent changes in the expression of various NK cells receptors (S1A Fig), but there was an increase in the frequency of immature NK cells in the spleen of infected Ifnar1 ΔNKp46 mice (S1B Fig). Whether this arrest in development contributes to the reduction in NK cell cytotoxicity and decreased resistance to mousepox in Ifnar1 ΔNKp46 mice is unclear. It is interesting to note that different from NK cells, the upregulation of GzmB in CD8 T cells does not require intrinsic IFNAR (Fig 3E). A possible explanation for this difference is that adaptive lymphocytes may rely more on antigen-specific signals for GzmB upregulation, while NK cells, which lack antigen receptors, could be more dependent on direct IFN-I signals.
The finding that partial IFNAR deletion in the monocytic myeloid lineage and neutrophils (Fig 6A and 6B) results in a significant increase in susceptibility to lethal mousepox (Fig 6C) is intriguing given the ubiquitous expression of IFNAR and the relatively low frequency of myeloid cells in tissues in general. Of note, neutrophil depletion does not increase susceptibility to mousepox [42], cDCs and pDCs express normal levels of IFNAR in Ifnar1 ΔLyz2 mice (Fig  6A), and LCs and Kupffer cells are of host origin in mousepox resistant B6!Ifnar1 -/mice (Fig 1A) indicating that none of these cell types are the culprits for the increased susceptibility of Ifnar1 ΔLyz2 mice to mousepox.
We have previously shown that infected DCs and iMOs are the primary producers of IFN-I during ECTV infection [11,17,54]. Here, we have used mixed BMCs where WT and Ifnar1 -/-iMOs were exposed to the same environment during infection to show that IFNAR signaling in infected iMOs is essential to provide a positive feedback loop for IFN-I expression in vivo (Fig 7B). A positive feedback loop in IFN-I expression was described before in vitro using fibroblasts infected with Newcastle disease virus [7]. In that context, virus infection readily induces the transcription of the Ifnb1 and Ifna4 genes. The resulting IFN-β and IFN-α4 signal through IFNAR to induce IRF7 expression, which leads to the transcription of non-Ifna4 genes. Moreover, Ifnb1 -/mouse fibroblasts were shown to produce less Ifna genes compared to WT fibroblasts after Sendai virus infection [55]. It is unclear whether a similar mechanism also regulates IFN-I production in iMOs in vivo or whether different cell types have different regulatory mechanisms. Considering that Irf7 and Cgas are both ISG and necessary for resistance to ECTV [11,14], it is possible that the IFNAR-dependent positive feedback loop of IFN-I might be mediated by the increased expression of these genes. Indeed, mRNA-Seq data of infected and uninfected WT and Ifnar1 -/-iMOs support this hypothesis and also confirm the IFNAR-dependent positive feedback of IFN-I (manuscript in preparation).
We showed that intrinsic IFNAR signaling does not curb virus transcription (Fig 7C) or production of infectious virus (Fig 7D) in infected iMOs in vivo as a possible mechanism of IFN-I protection from mousepox. Based on this finding, we speculate that the major role of IFN-I signaling in resistance to mousepox is regulating the immune response rather than directly blocking virus replication inside infected cells.
Our data strongly suggest that NK cells and iMOs exert IFNAR-dependent roles that are indispensable for resistance to ECTV infection. We propose that iMOs must sense paracrine or autocrine IFN-I through IFNAR to upregulate IFN-I production, which stimulates NK cells for optimal killing of infected cells and, possibly, other critical protective functions. Our findings contribute to understanding how innate immunity is orchestrated to control viruses and is also of interest in guiding the development of IFN-I-based therapeutic interventions for viral pathogens.

Ethics statement
All the procedures involving mice were carried out in strict accordance with the recommendations in the Eighth Edition of the Guide for the Care and Use of Laboratory Animals of the National Research Council of the National Academies. All experiments were approved by Thomas Jefferson University's Institutional Animal Care and Use Committee under protocol number 01727 "Innate Control of Viral Infections."

Reagents
All reagents used are listed in Box 1. /J were purchased from the Jackson Laboratory. Colonies were bred at Thomas Jefferson University under specific pathogen-free conditions.

Viruses and infection
ECTV Moscow strain (WT) (ATCC VR-1374) and ECTV-GFP [28] were propagated in tissue culture, as previously described [13]. Briefly, BS-C-1 cells grown in DMEM media supplemented with 10% fetal bovine serum (FBS), 100 IU penicillin, 100 μg/mL streptomycin, 1x GlutaMAX, and 10mM HEPES to confluency were infected at MOI 0.01, and viruses were harvested 4 to 5 days later. For that, cells were rinsed with phosphate-buffered solution (PBS), scraped and concentrated by centrifugation. ECTV was released by multiple freezing and thawing cycles. Viruses stocks were sonicated and titrated by plaque assay in BS-C-1 cells.
Mice were infected subcutaneously in the rear footpad with 3000 plaque-forming units (PFU) of ECTV WT or ECTV-GFP as indicated. For sorting monocytes from dLN, mice were infected in both rear footpads with 3000 PFU of ECTV-GFP in each footpad. In survival curves, mice were monitored daily and sacrificed whenever illness resulted in a lack of activity and unresponsiveness to touch. Euthanasia was performed according to the 2013 edition of the AVMA Guideline for the Euthanasia of Animals. For virus titer quantification, entire or portions of spleens and portions of livers were homogenized in supplemented DMEM media using a TissueLyzer II (Qiagen), and organ titers were determined by plaque assay on BS-C-1 cells as described before [13]. Shortly, BS-C-1 cells at confluency were infected with 10-fold dilutions of each virus sample for 2 hours at 37˚C. Infected cells were grown in 1% carboxymethyl cellulose overlay containing DMEM media supplemented with 2.5% FBS, 100 IU/mL penicillin, 100 μg/mL streptomycin, 1x GlutaMAX and 10mM HEPES. Virus plaques were quantified at 5 dpi after staining with 0.2% crystal violet dissolved in 20% methanol.

Bone marrow chimeras
Were prepared as previously described [56]. Briefly, 4-to 6-week-old mice previously treated for three days with acidified water (pH 2.5) were irradiated with 900 Gy using a GammaCell 40 apparatus (Nordion Inc.). Bone marrow cells were isolated in RPMI media supplemented with 10% FBS, 100 IU/mL penicillin, and 100 μg/mL streptomycin. Red blood cells were lysed with 0.84% NH 4 Cl and cells were filtered and counted. Irradiated mice were reconstituted intravenously with 5-10 million bone marrow cells from donors. Chimeras were given acidified water for four weeks and rested for eight weeks after reconstitution.

Flow cytometry and sorting
Flow cytometry was performed as previously described [13]. Briefly, organs were processed into single-cell suspensions. Spleens were smashed with frosted slides, and cells were washed with RPMI media supplemented with 10% FBS, 100 IU/mL penicillin, and 100 μg/mL streptomycin. Livers were cut into smaller pieces and smashed using syringe plungers and metal strainers. White and red blood cells were percoll-enriched, and hepatocytes were discarded. Red blood cells were lysed with 0.84% NH 4

CFSE labeling and adoptive transfers
Splenocytes from indicated donor mice were isolated in RPMI media supplemented with 10% FBS, 100 IU/mL penicillin, and 100 μg/mL streptomycin. Red blood cells were lysed with 0.84% NH 4 Cl. Splenocytes were washed with a pre-warmed phosphate-buffered solution (PBS) containing 0.1% bovine serum albumin (BSA) and filtered in 70μm sterile nylon mesh. Cells were adjusted to 50 million cells/ml concentration, and a 1:1 ratio of donor mix was labeled with 4μM CellTrace™ CFSE Cell Proliferation Kit for 10 min at 37˚C. Recipient mice were injected intravenously with 0.4 to 2 x 10 7 of CFSE-labeled cells and infected at indicated times with ECTV.

RNA preparation and Reverse Transcription-qPCR
Total RNA from spleens and livers was purified with the RNeasy Mini Kit (Qiagen) and treated with DNase I (Qiagen). RNA quantified using NanoDrop, and 300 ng were used as a template for cDNA synthesis with High Capacity cDNA Reverse Transcription Kit in a 20μL reaction. Quantitative PCR (qPCR) was performed as previously described [11,54]. Shortly, 1μL of cDNA was used as a template for amplification using SYBR Green PCR Master Mix through 92˚C for 3 min, 25 cycles of 92˚C for 30 sec, 56˚C for 45 sec, 65˚C for 50 sec, plate read in a Thermocycler CFX96 Real-Time System (BioRad). For purifying RNA from sorted populations, cells were sorted into Trizol, and the RNA was purified with the RNA Clean and Concentrator (Zymo Research). RNA was eluted in 10μL, and all RNA was used as a template for cDNA synthesis. Gene expression was normalized by Gapdh levels.