Macrophage Coordination of the Interferon Lambda Immune Response

Lambda interferons (IFN-λs) are a major component of the innate immune defense to viruses, bacteria, and fungi. In human liver, IFN-λ not only drives antiviral responses, but also promotes inflammation and fibrosis in viral and non-viral diseases. Here we demonstrate that macrophages are primary responders to IFN-λ, uniquely positioned to bridge the gap between IFN-λ producing cells and lymphocyte populations that are not intrinsically responsive to IFN-λ. While CD14+ monocytes do not express the IFN-λ receptor, IFNLR1, sensitivity is quickly gained upon differentiation to macrophages in vitro. IFN-λ stimulates macrophage cytotoxicity and phagocytosis as well as the secretion of pro-inflammatory cytokines and interferon stimulated genes that mediate immune cell chemotaxis and effector functions. In particular, IFN-λ induced CCR5 and CXCR3 chemokines, stimulating T and NK cell migration, as well as subsequent NK cell cytotoxicity. Using immunofluorescence and cell sorting techniques, we confirmed that human liver macrophages expressing CD14 and CD68 are highly responsive to IFN-λ ex vivo. Together, these data highlight a novel role for macrophages in shaping IFN-λ dependent immune responses both directly through pro-inflammatory activity and indirectly by recruiting and activating IFN-λ unresponsive lymphocytes.


INTRODUCTION
Lambda interferons (IFNL and IFN-λ), also known as type III IFNs, are a family of cytokines comprising four members: IFN-λ1 (IL29), IFN-λ2 (IL28A), IFN-λ3 (IL28B), and IFN-λ4. While all IFN-λs signal through a unique IFNLR1:IL10Rβ receptor complex, they activate a gene signature similar to type I IFNs, IFN-α, and IFN-β (1). Both type I and III IFNs activate the transcription of hundreds of interferon stimulated genes (ISGs) (1) that exhibit numerous autocrine and paracrine antiviral roles. Although IFNs are required to clear most viral infections, prolonged expression due to environmental or genetic factors can stimulate sustained immune activation, driving tissue damage, and fibrosis (2,3).
Elevated IFN-λ3 production has demonstrated a strong association with IFNL genotype and hepatic inflammation, increasing the risk of both viral (HBV and HCV) and nonviral (non-alcoholic steatohepatitis, NASH) related progressive liver inflammation and fibrosis (4). Furthermore, these effects appear to be independent of IFN-λ4 activity, suggesting that IFN-λ3 may be a primary mediator of inflammation (5). While the precise mechanisms remain uncertain, peripheral and hepatic immune cell populations vary according to the IFNL polymorphism in patients with chronic HCV infection, suggesting that IFN-λs can prompt immune cell migration to tissues (5,6).
Here, we demonstrate that human macrophages, not monocytes, are a dominant, physiologically relevant IFNλ responsive population capable of orchestrating tissue inflammation. This is achieved through a direct immunostimulatory response to IFN-λ and subsequent NK and T cell chemotaxis and activation. In vivo, macrophages are responsive to IFN-λ3 and accumulate in inflamed human liver. These data suggest a novel role of macrophages as key players in modulating the IFN-λ response in acute infection, as well as chronic disease.

Macrophages Not Monocytes Are Responsive to IFN-λ
To address the uncertainty surrounding monocyte and macrophage (Mφ) IFN-λ-responsiveness, we measured mRNA expression of the IFN-λ receptor, IFNLR1, in blood leukocytes by digital droplet PCR (ddPCR). DdPCR enables the precise quantification of RNA transcripts by performing the PCR reaction within >10,000 oil droplets, and calculating transcript copies using Poisson's law of small numbers (34). IFNLR1 expression in freshly isolated monocytes and in Mφs cultured for 7 days with GM-CSF was compared to IFN-λ responsive cells (pDCs) and "unresponsive" cells (NK and T cells). Similar to NK and T cells, monocytes expressed minimal IFNLR1 transcript. Mφ and pDC IFNLR1 expression was significantly increased compared to other populations, suggesting IFN-λ responsiveness ( Figure 1A). Increased abundance of IFNLR1 was confirmed following monocyte to macrophage differentiation using seven datasets from the NCBI Gene Expression Omnibus (35) (Figure S1). To assess IFNLR1 expression during differentiation, IFNLR1 expression was quantified over 24 h (qPCR, no differentiation stimulus) and 7 days (flow cytometry, GM-CSF differentiation) following monocyte plating. Expression of the IFNLR1 transcript was quickly increased as early as 16 h post-plating, reaching a 30-fold increase at 24 h ( Figure 1B). IFNLR1 surface expression was significantly increased at day 3 (monocyte IFNLR1 MFI 927 vs. day 3 Mφ 3199) and further increased at day 7 (day 7 Mφ IFNLR1 MFI 10,412) ( Figure 1C).
To test monocyte and Mφ responsiveness to IFN-λ, cells isolated and cultured as in Figure 1A were treated with 100 ng/ml IFN-λ3 for 8 h. This concentration is not a saturation dose, but is high enough to evoke a strong interferon response in Mφs ( Figure S2). Consistent with increased IFNLR1 expression, Mφ and pDC mRNA expression of viperin and ISG15 were markedly increased (Figure 1D), whereas monocytes and NK cells demonstrated negligible responses.

Differentiation Method Regulates IFN-λ Responsiveness
Mφ differentiation medium containing IFN-γ and LPS or interleukin 4 (IL-4) and IL-13 are often used to generate pro-inflammatory (M1) or anti-inflammatory (M2) Mφs, respectively, but do not reflect the spectrum of macrophage activation in vivo (36). To avoid generating Mφs whose IFN-λ sensitivity is influenced by phenotype skewing, monocytes were differentiated for 7 days with GM-CSF or M-CSF alone, as previously described (37,38). The resulting Mφ populations are differentially responsive to inflammatory stimuli, and are thus M1-or M2-shifted while maintaining some baseline characteristics of polarized Mφs ( Figure S3).When compared to monocyte derived DCs (MDDCs) generated using IL-4 and GM-CSF, the resulting Mφs express elevated surface expression of CD14 and CD16, reduced CD1C, and unlike MDDCs, adhere strongly to culture dishes ( Figure S4). M1-and M2-shifted Mφs will be termed GM-Mφs and M-Mφs for the remainder of the manuscript.
IFN receptor expression and response to type I and III IFNs was examined in monocytes and Mφs. Mφ differentiation increases the abundance of the type I IFN receptor, IFNAR1 transcript (Figure 2A), and protein ( Figure 2B) ∼2-fold as compared to monocytes irrespective of stimulus. IFNLR1 transcript abundance was increased in M-Mφs and GM-Mφs over 30-and 60-fold, respectively. The IFN-λ co-receptor IL10RB was also measured, and was not significantly modulated following macrophage differentiation ( Figure S5). Consistent FIGURE 1 | Macrophages but not monocytes are highly IFN-λ3 responsive. To investigate IFN-λ responsiveness, immune cell subsets were magnetically isolated and IFNLR1 expression was quantified by ddPCR (A). Mφ and pDC IFNLR1 expression was significantly higher than monocyte, NK and T cell populations (p < 0.05, n = 8). Time course analysis demonstrated that IFNLR1 expression quickly rises following monocyte plating, reaching a 30× increase at 24 h even in the absence of GM-CSF addition (p < 0.001, n ≥ 5) (B). Similarly, IFNLR1 surface expression during macrophage differentiation (MFI) increased at days 3 and 7 (p < 0.001, n ≥ 5) (C). Isolated immune cell subsets were treated with 100 ng/ml IFN-λ3 for 8 h and the expression of ISGs viperin and ISG15 were examined (n ≥ 7) (D). Consistent with IFNLR1 expression, Mφs and pDCs were highly responsive to IFN-λ3, whereas monocytes and NK cells were not (n ≥ 5). Data are representative of two (B,C) and three independent experiments (A,D). One-way ANOVA (A), Mann-Whitney test (B,D), paired t-test (C), *p < 0.05, **p < 0.01, ***p < 0.001 (mean ± SE).
with gene expression, IFNLR1 protein was absent in monocytes, and increased in GM-Mφs compared to M-Mφs. The IFNLR1 bands at 70 and 45 kDa represent the full length and soluble isoforms of IFNLR1, respectively (24).
To confirm that elevated IFNLR1 expression confers response to IFN-λs, monocytes and Mφs from three healthy subjects were treated with IFN-λ3 for 15 min and STAT1 phosphorylation (Y701) was examined by Western blot. Monocytes did not phosphorylate STAT1 in response to IFN-λ3, whereas both M-Mφs and GM-Mφs were highly sensitive ( Figure 2C). All cells demonstrated no STAT1 phosphorylation pre-treatment ( Figure S6).
Mφs were subsequently treated with either interferon-alpha (IFN-α) or IFN-λ3 to determine whether cognate receptor expression defines sensitivity. Following 8 h of IFN-α or IFN-λ3, all measured ISGs were significantly increased by both IFNs (Figure 2D). M-Mφs were more sensitive to IFN-α, demonstrating stronger induction of all ISGs, particularly CD80 and TRAIL. In contrast, GM-Mφs were more sensitive to IFN-λ3, increasing the expression of both ISG15 and viperin compared to M-Mφs. To confirm that macrophage differentiation and not treatment with M-or GM-CSF specifically induce IFNLR1 expression, monocytes were also differentiated using 10% autologous human serum from healthy individuals. Compared monocytes, human serum differentiated macrophages (HS-Mφs) possess increased IFNLR1 transcript abundance and possessed similar IFN-λ3 responsiveness as MDDCs and GM-Mφs, both of which express high levels of IFNLR1 ( Figure S7).

IFN-λ3 Drives a Pro-inflammatory Macrophage Phenotype
The robust induction of IFNLR1 following monocyte plating suggests that monocytes quickly become IFN-λ responsive upon differentiation and transmigration into tissues. Consequently, in the context of chronic antigen exposure, IFN-λ expression at sites of inflammation will likely influence monocyte differentiation and subsequent Mφ phenotype due to their prolonged exposure throughout the differentiation process. To test this hypothesis, we differentiated monocytes for 7 days with either M-CSF or GM-CSF alone (differentiation stimulus), or in combination with IFN-λ3 (activation stimulus), followed by transcriptional and functional assessment of Mφ phenotype ( Figure 3A). FIGURE 2 | M-CSF and GM-CSF differentiated macrophages respond differently to IFN-α and IFN-λ3. Following M-and GM-CSF stimulated differentiation, IFN-λ responsiveness of Mφ populations was examined, and compared to IFN-α. M-CSF and GM-CSF Mφ subsets both increased the expression of IFNAR1 ∼2-fold following differentiation from monocytes, and IFNLR1 transcripts ∼30-and 60-fold, respectively (A) (n ≥ 9). Western blot of IFNAR1 and IFNLR1 from four healthy subjects confirmed these findings (B) (n = 7, total). Supporting these findings, phosphorylation of STAT1 was detected by Western blot in both monocytes and macrophages following 15 min of IFN-α treatment and macrophages only following IFN-λ3 (C). ISG transcripts for ISG15, viperin, CD80, and TRAIL were examined to measure Mφ sensitivity to IFN-α (50 U/ml) and IFN-λ3 (100 ng/ml) (D) (n = 8). M-CSF differentiated Mφs were more responsive to IFN-α, whereas GM-CSF differentiated Mφs, IFN-λ3. Data are representative of two independent experiments. Paired t-test * /# p < 0.05, ** /## p < 0.01, ***p < 0.001 (mean ± SE). *Mock vs. IFN treatments, # M-Mφ vs. GM-Mφ.
RNA sequencing of M-Mφs and GM-Mφs from three donors was undertaken followed by paired analysis of transformed gene counts (Log 2) between untreated and IFN-λ3 treated Mφs (Tables S1, S2). The resulting smear plot demonstrates significantly up and down-regulated genes following differentiation of M-Mφs (blue) and GM-Mφs (red) with IFN-λ3 ( Figure 3B). GM-Mφs were significantly more responsive to IFN-λ3, up-regulating 463 genes ≥2-fold compared to 184 genes in M-Mφs. Similarly, GM-Mφs downregulated 467 genes ≥2-fold compared to 252 genes in M-Mφs. IFN-λ driven ISG induction was also collectively higher in GM-Mφs as demonstrated by the heat map of gene expression LogFC ( Figure 3C).
Functional analysis of data from Mφs differentiated with IFN-λ3 revealed numerous well defined ISGs (e.g., IFI27, MX1, and TLR3) and transcription factors (STAT and IRF gene families) responsible for ISG gene transcription ( Figure 3D). In addition, a Th1 chemokine profile (CCL3, 4, and 5 and CXCL9, 10, and 11) responsible for CCR5 and CXCR3 mediated immune cell chemotaxis (39) was found in response to maturation with IFN-λ3, with stronger induction in GM-Mφs. Up-regulation of immune cell interaction and activation (CD80, CD86, and IL15) as well as antigen presentation [major histocompatibility complex (MHC) class I HLA genes] was also observed. Using transcriptomic markers of M1 and M2 Mφ differentiation, IFN-λ3 was found to induce the expression of the majority of M1, but not M2 markers, in both M and GM-differentiated Mφs, supporting the movement toward an M1 phenotype (GM-CSF p < 0.001, M-CSF p < 0.05, Sign test null hypothesis of 0.5) ( Figure S8).
Gene induction was confirmed by qPCR from a larger group of donors including individuals used for RNA sequencing data, and compared to the effects of IFN-α differentiation. M-Mφs were considerably more sensitive to IFN-α, whereas Frontiers in Immunology | www.frontiersin.org FIGURE 3 | demonstrated that GM-CSF Mφs were significantly more responsive to IFN-λ3, as demonstrated by smear plot and Venn diagram of genes regulated above 2-fold (B). GM-CSF Mφ ISG induction was significantly stronger, as demonstrated by heat map of gene log 2-fold change (logFC) (C). IFN-λ3 increased transcript abundance of numerous ISGs and transcription factors (TFs) in both Mφ sets, as well as numerous chemokines, cytokines, and genes responsible for antigen (Ag) presentation and co-stimulation, particularly in GM-CSF Mφs (D). IFN-λ3 stimulated genes were confirmed by qPCR, using IFN-α differentiated Mφs as a comparison (E) (n = 8). Quantitative PCR data are representative of three independent experiments. Paired t-test * /# p < 0.05, ** /## p < 0.01, ***p < 0.001 (mean ± SE). *Mock vs. IFN treatments, # M-Mφ vs. GM-Mφ. GM-Mφs responded strongly to IFN-λ ( Figure 3E). In addition to chemokines identified by RNA sequencing, inflammatory mediators including CCL2, IL1B, and TNF transcripts were increased by IFN-λ in both Mφ subsets. To assess the role of differentiation (M-vs. GM-CSF), interferon treatment (IFN-α and -λ3), and their subsequent interaction, a 2-way ANOVA was additionally performed. As expected, all ISGs measured were significantly affected by IFN treatment (p < 0.01), however only CD80 expression was influenced by differentiation (p < 0.01).
In agreement with RNA-Seq analysis, a significant interaction between IFN treatment and differentiation was observed for all measured genes (CXCL10, CCL8, IL15; p < 0.001, CD80; p < 0.01, TRAIL, TNF; p < 0.05) apart from IL1B and CCL2. Analysis using Metacore functional annotation software demonstrated that similar networks were activated by IFN-λ3 in both M-Mφs and GM-Mφs ( Table 1). Immune activation was considerably stronger in GM-Mφs, with highly significant pvalues in networks such as antigen presentation and lymphocyte proliferation. Down-regulated gene networks were primarily associated with the cell cycle and protein translation (Table S3).
This analysis is consistent with Mφ BrdU assays, which demonstrated a reduction in BrdU incorporation following IFNα (p < 0.05) and IFN-λ3 (NS) treatment ( Figure S9).
Co-culture experiments were next performed to assess the capacity of IFN-λ matured Mφs to stimulate NK cells in vitro. NK cells were incubated with Mφs overnight, removed, and cocultured with K562 cells. K562 cells lack MHC class I expression, making them targets for NK cell killing. A significant increase in NK cell degranulation (CD107a), particularly within the CD56 dim population, was observed following co-culture with IFN-λ3 treated GM-Mφs ( Figure 4D). NK cell IFN-γ production was also increased following co-culture with IFN-λ3 treated GM-Mφs, but significance was lost within subgroup analysis. Minimal effect on NK cell function was observed following M-Mφ co-culture.

IFN-λ Stimulates Macrophage Phagocytosis and Cytotoxicity
To examine the effect of IFN-λ3 on Mφ effector function that is not associated with an inflammatory phenotype, phagocytosis was examined by flow cytometry. Tissue resident Mφs that demonstrate an M2 phenotype are highly phagocytic and efficient at presenting antigen (40), a phenotype that can be replicated in vitro (41,42). UV induced apoptotic K562 cells stained with Zombie Yellow viability stain or DAPI labeled E. coli were incubated with Mφs for 1 h at a ratio of 2:1 and 4:1, respectively. The ratio of double-fluorescent (phagocytic, CD14+) cells to mono-fluorescent (non-phagocytic, CD14+) cells, as measured by flow cytometry (Figure 5A) was calculated to determine the phagocytic Mφ percentage (Figure 5B). Confocal microscopy was additionally used to confirm cell engulfment. IFN-λ3 increased phagocytosis of K562 cells (30% increase) and E. coli bacteria (10% increase) in M-Mφ alone. Mφ MFI, indicating of the number of engulfed target cells, was consistent among populations when K562 cells were used as targets, likely reflecting Frontiers in Immunology | www.frontiersin.org  (Figure S10A). IFN-λ3 had minimal effect on the expression of most phagocytic receptors, but significantly increased key members of the complement cascade (C1S and C1R) ( Figure S10B). These data suggest that activation of the complement system by IFN-λ3 may stimulate M-Mφs phagocytosis of both bacterial and apoptotic cells (43,44), however further functional analysis is required.
To quantify the ability of Mφs to kill virus infected cells (cytotoxicity), Mφs were co-cultured with HCV infected (JFH1 strain) Huh-7 cells for 24 h. Following incubation, Huh-7 cells were labeled with Epcam, Annexin V, and propidium iodide (PI) to quantify cells undergoing apoptosis ( Figure 5C). Additionally, Huh-7 and JFH1 infected Huh-7 cultures were used as controls to confirm HCV mediated Huh-7 cell apoptosis, as previously described (45). M-and GM-Mφs differentiated with IFN-λ3 stimulated an increase of early apoptotic (Annexin V+, PI − ) cells, by ∼20 and 90%, respectively, compared to mock treated controls ( Figure 5D). GM-Mφs alone increased Annexin V+, PI+ cells, representing late apoptosis by ∼20% following IFN-λ3 treatment. The cytotoxic mechanism by which Mφs killed infected cells has not been determined, but is likely mediated by soluble factors such as TRAIL that is highly inducible following IFN-λ3 treatment in GM-Mφs in particular ( Figure 5E). Low expression of TRAIL in untreated Mφs may explain the apparent lack of apoptosis following co-culture. In addition, no nitric oxide production by M-or GM-Mφs was found in response to IFN-λ3, bacterial or infected cell stimulus. To validate Huh-7 cell apoptosis results, qPCR for apoptosis markers Caspase 3 (Casp3), Caspase 7 (Casp7), and Bax were performed ( Figure S11). Coculture with IFN-λ3 differentiated GM-Mφs increased Casp3 and 7 expression by ∼6-fold in addition to increasing the antiviral response of Huh-7 cells as demonstrated by strong induction of ISGs viperin and ISG15.

Liver Macrophages Are IFN-λ3 Responsive in vivo
To assess IFN-λ production in vivo, we measured the expression of IFNL genes in liver biopsies of chronic HBV, HCV, and NAFLD/NASH patients and compared them to normal liver tissue from benign liver resections. IFNLs were increased in both viral (>10-fold IFNL1, IFNL2/3 HCV vs. healthy) and nonviral (e.g., ∼2-fold IFNL1, IFNL2/3 NAFLD/NASH vs. healthy) liver disease (Figure 6A), indicating that chronic inflammatory conditions can increase hepatic IFN-λ expression to facilitate the generation of inflammatory macrophages in vivo.
To demonstrate the presence of IFN-λ responsive Mφs in vivo, we performed immunofluorescent labeling of liver tissue from a patient with autoimmune hepatitis (AIH), chronic HCV infection, and normal liver obtained from a cancer resection. Biopsies were labeled with IFNLR1 and CD68 or CD11b antibodies to identify IFN-λ responsive liver Mφs (Kupffer cells) or myeloid populations (monocytes/macrophages/neutrophils), respectively. As demonstrated in Figure 6B, all CD68+ and a fraction of CD11b+ cells were labeled with IFNLR1.
Immuno-labeling was also performed using CD68 or CD11b in combination with CD3 to demonstrate immune cell proximity in inflamed tissue ( Figure S12). CD3+ T cells localized in proximity to CD68 Mφs, supporting a role for Mφ derived chemokines as mediators of immune cell trafficking.
To compare IFN-λ3 sensitivity, T cells and Mφs from each individual were cultured for 10 h in media alone or with IFN-λ3, followed by quantification of ISG expression. T cells were unresponsive to IFN-λ3 as demonstrated by a lack of ISG15 and viperin induction ( Figure 6E). Conversely, IFNLR1 expressing Mφs were highly responsive to IFN-λ3, increasing the abundance of both transcripts ∼6-fold.

DISCUSSION
The cellular and molecular mechanisms by which IFNλ modulates host responses to viral infections and tissue inflammation remains unclear. Here we undertook comprehensive functional characterization to demonstrate both in vitro and ex vivo, that macrophages are likely immune cell drivers of IFN-λ mediated hepatic antiviral and inflammatory activities. Unlike monocytes, macrophages are highly sensitive to IFN-λ through the induction of IFNLR1 Frontiers in Immunology | www.frontiersin.org FIGURE 6 | infected liver tissue were labeled with monocyte/Mφ markers CD11b and CD68 as well as IFNLR1 antibodies, and examined by confocal microscopy (B).
To assess the responsiveness of liver Mφs to IFN-λ3, immune cells were isolated from fresh liver tissue, sorted by FACS and cultured in the presence of IFN-λ3. Live CD45+ immune cells were sorted based on the expression of CD3 and CD56 into NK and T cell subsets, and CD14 and CD68 into Mφ subsets (C). IFNLR1 expression, as determined based on IFNLR1 MFI was compared among liver immune cell subsets, and was significantly higher in CD14+, CD68+ liver Mφs, as compared to NK (CD56+), NKT (CD3+, CD56+), and T cells (CD3+, CD56−) (D) (n = 6). Sorted T cells and Mφs were cultured with 100 ng/ml IFN-λ3 for 10 h and ISG mRNA expression was compared to mock treated cells (E) (n = 8). Scale bars represent 100 µm. Data are representative of one cell sorting experiment. Wilcoxon matched pairs signed rank test, *p < 0.05, **p < 0.01, and ***p < 0.001 (mean ± SE). expression. As such, monocytes likely become IFN-λ responsive upon movement into tissue and subsequent differentiation. Upon IFN-λ stimulation, macrophages develop a robust immune-stimulatory gene signature, expressing hundreds of ISGs, cytokines, chemokines, and co-stimulatory molecules to stimulate both autocrine and paracrine immune cell activation (Figure 7). IFN-λs are inducible cytokines that drastically increase in abundance upon viral infections, but can also effectively protect against bacterial and fungal insults (46,47). Activation of TLRs 3,4,5,7,9 (48,49) can drive IFN-λ expression, which is dependent factors including cell type and cellular environment. IFN-λs are necessary for epithelial barrier protection in the lungs, liver and gastrointestinal tract, however their dysregulation has been associated with a number of diseases that lack an obvious association with microbial infection. These include chronic inflammatory diseases such as psoriasis (50), systemic lupus erythematosus (51), and asthma (52). Consequently, it is important to understand the direct and indirect molecular mechanisms by which IFN-λs are induced, as well as the responding cellular identities.
The effectiveness of direct acting antiviral therapy for chronic HCV infection has ultimately overshadowed the antiviral role of IFN-λs in the liver, however there remains much to be understood regarding the immuno-stimulatory and potentially destructive roles of these unique cytokines. Recent evidence suggests that IFNL genotype influences IFN-λ expression in the liver to facilitate immune cell migration and subsequent inflammation (4,5). Unlike IFN-λ4 which is weakly secreted by hepatocytes (53,54), IFN-λs 1-3 are highly expressed, and can exert paracrine effects on surrounding immune cells. This is consistent with reports showing increased Mφ activation in patients possessing the favorable IFNL genotype (55). The importance of the Mφ response is additionally underscored by the fact that Mφ but not hepatocyte ISG expression is positively associated with both the favorable IFNL genotype that produces increased IFN-λ3 and antiviral response (56,57).
Both PCR and RNA-Seq analysis support the IFN-λ sensitive nature of GM-Mφs and highlight the stimulatory role of IFN-λ3. Increased expression of pattern recognition receptors [TLR3, IFIH1 [MDA5], DDX58 [RIG-I]] in response to IFN-λ can increase antigen recognition, as we have shown in Figure 5. Numerous pro-inflammatory transcription factors including STATs 1-3, IRFs 1, 7, and 9, AP-1, and NFκB components were activated in response to IFN-λ, as demonstrated by an enrichment of their respective target genes (Table S4). Inflammatory cytokines TNF and IL1B that are known mediators of hepatocyte apoptosis and liver injury were moderately, albeit significantly induced by IFN-λ treatment alone (Figure 3E), though our data supports additional inflammatory effects of IFN-λ. By strengthening Mφ recognition and response to pathogen associated molecular patterns, IFN-λs likely sensitize Mφs to inflammatory stimuli, thus amplifying the strength and/or duration of the inflammatory cascade. This has been demonstrated by Liu et al. who showed that IFN-λ1 can promote IL-12 production in TLR7 stimulated Mφs (28).
In response to IFN-λ, GM-Mφs potently express Th1 chemokines including CXCL 9, 10, and 11 as well as IL-15 and IL-27, notable drivers of T and NK cell activation and proliferation. In agreement with RNA-Seq gene expression data, we demonstrated that IFN-λ3 treated GM-Mφs stimulate T and NK cell chemotaxis and subsequent NK cell cytotoxicity. These data suggest that IFN-λs are strong mediators of the Th1 response and provides a rationale for works by Morrow et al. who showed that IFN-λ3 can increase IFN-γ secretion and degranulation despite T cell insensitivity to IFN-λ (58). A similar phenotype has been observed in tumor model NK cells, where IFN-λ signaling drives NK cell cytotoxicity, suppression of tumor growth and metastases (33,59). This Th1 skewing effect has been further validated using murine models of Th2 diseases where IFN-λ alleviated symptoms of airway disease (60), intestinal inflammation (61), and conjunctivitis (62).
Interestingly, GM-Mφs were significantly more responsive to IFN-λ, whereas M-Mφs where more responsive to IFN-α. These data suggest that while type I and type III IFNs induce a similar gene signature, their respective response is dependent not only on the cell type, but also, the inflammatory phenotype of the responsive cell. This data is consistent with work by Fleetwood et al. that demonstrate a strong dependence on type I IFN signaling in M-CSF over GM-CSF cultured Mφs (37). Consistent with an M2 phenotype (63), M-Mφs were not particularly efficient at driving immune cell chemotaxis and activation upon IFN-λ3 stimulation, but were significantly more phagocytic than GM-Mφs both at baseline and in response to IFN-λ3. These data suggest that IFN-λs are perhaps not inherently inflammatory, but instead promote macrophage effector functions based on location or developmental phenotype.
Our data fills a current gap in knowledge concerning the cellular identities and mechanisms that regulate local IFNλ mediated inflammation. Because IFN-λ signaling is longer lasting and unlike IFN-α does not become refractory following chronic exposure (64), continuous IFN-λ expression from chronic infections can drive prolonged immune activation. While DCs have a strong IFN-λ response, they are a small population in the liver and migrate toward proximal lymph nodes in response to infection (65). Liver Mφs (Kupffer cells) on the other hand consist of the ∼3/4 of hepatic immune cells, and remain locally to become crucial drivers of localized tissue inflammation (65). Neutrophils are the only other immune cell subset with a defined IFN-λ response, and respond with reduced migration and suppression of inflammation (22,23,66).
In summary, we have demonstrated a novel concept whereby Mφs gain IFN-λ sensitivity quickly following differentiation from monocytes. These data support a pro-inflammatory role for IFN-λs, particularly via recruitment of NK and T cells, chief promoters of inflammation in chronic liver disease. Mφs bridge the gap between IFN-λ responsive and non-responsive effector cells, and are likely implicated in the elimination of acute infection and the promotion of tissue damage in chronic disease.

Patient Cohort
Blood samples were obtained from healthy volunteers at the Westmead Institute of Medical Research. Data points represent individual donors from a cohort of ∼20 healthy individuals. Different cohorts of donors were used for individual experiments based on availability. Liver tissue was obtained at Westmead Hospital, Sydney, at the time of needle biopsy [chronic HBV/HCV infection, non-alcoholic fatty liver disease (NAFLD)/NASH, autoimmune hepatitis] or from patients undergoing liver resections (normal tissue). Ethics approval was obtained from the Sydney West Area Health Service and University of Sydney. Informed consent was obtained for all subjects [HREC2002/12/4.9(1564)].

RNA Sequencing and Bioinformatics
RNA was extracted using the Favorgen Tissue Total RNA Kit and the sequencing library was prepared using the TruSeq Stranded mRNA Library Prep Kit (Illumina). Single ended RNA sequencing (RNA-Seq) was performed at the Australian Genome Research Facility using the Illumina HiSeq 2500 platform (50 bp read length; minimum of 10 7 reads per sample). Sequence alignments and gene counts were performed using STAR RNA-Seq aligner version 2.5.1b (67) and paired comparisons were performed using EdgeR version 3.16.2 (68). Heat map visualization of RNA-Seq data was performed using GITools (69). Functional analysis of IFN-λ3 mediated gene expression was conducted using Metacore version 6.29 (Thomson Reuters).

Digital Droplet PCR (ddPCR)
Immune cell RNA was quantified using the Qubit fluorometer and RNA BR assay kit (Thermo Fisher), and cDNA was synthesized from ≥10 ng of RNA per sample using qScript cDNA supermix (Quantabio). cDNA was combined with ddPCR supermix and droplet generation oil for probes (Bio-Rad), and droplets were generated using the Bio-Rad QX200 Droplet Generator. PCR was performed using IFNLR1 and GAPDH probes according to the manufacturer's instructions, and droplet fluorescence was analyzed using the Bio-Rad QX200 Droplet Reader. Absolute quantification of transcript number was determined using QuantaSoft Analysis Pro software.

Phagocytosis Assays
To stimulate apoptosis, K562 cells were exposed to UV light for 15 min. Apoptotic K562 cells were stained with ZombieYellow viability stain (Biolegend), after which apoptosis was confirmed with >90% of cells staining positive. Culture media was removed from Mφ cultures and target cells in RPMI + 10% FCS were added at a ratio of 2:1 (K562). Culture plates were centrifuged at 450 g for 2 min to synchronize phagocytosis. Following 1 h of incubation at 37 • C, cells were washed and labeled with BV711 mouse anti-human CD14 antibody (BD Biosciences), and analyzed using the BD Biosciences LSR Fortessa cell analyzer.

Statistics
Statistics were performed using GraphPad Prism and were chosen based on the normality of the data, with p < 0.05 deemed significant. Student t-tests or Mann-Whitney tests were performed on unpaired samples based on data normality. Paired t-tests and Wilcoxon matched pairs signed rank test were performed on paired samples based on Gaussian distribution.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Sydney West Area Health Service. The patients/participants provided their written informed consent to participate in this study.