Homeostatic interferon-lambda response to bacterial microbiota stimulates preemptive antiviral defense within discrete pockets of intestinal epithelium

  1. Jacob A Van Winkle
  2. Stefan T Peterson
  3. Elizabeth A Kennedy
  4. Michael J Wheadon
  5. Harshad Ingle
  6. Chandni Desai
  7. Rachel Rodgers
  8. David A Constant
  9. Austin P Wright
  10. Lena Li
  11. Maxim N Artyomov
  12. Sanghyun Lee
  13. Megan T Baldridge  Is a corresponding author
  14. Timothy J Nice  Is a corresponding author
  1. Department of Molecular Microbiology and Immunology, Oregon Health & Science University, United States
  2. Department of Medicine, Washington University School of Medicine, United States
  3. Department of Pathology and Immunology, Washington University School of Medicine, United States

Abstract

Interferon-lambda (IFN-λ) protects intestinal epithelial cells (IECs) from enteric viruses by inducing expression of antiviral IFN-stimulated genes (ISGs). Here, we find that bacterial microbiota stimulate a homeostatic ISG signature in the intestine of specific pathogen-free mice. This homeostatic ISG expression is restricted to IECs, depends on IEC-intrinsic expression of IFN-λ receptor (Ifnlr1), and is associated with IFN-λ production by leukocytes. Strikingly, imaging of these homeostatic ISGs reveals localization to pockets of the epithelium and concentration in mature IECs. Correspondingly, a minority of mature IECs express these ISGs in public single-cell RNA sequencing datasets from mice and humans. Furthermore, we assessed the ability of orally administered bacterial components to restore localized ISGs in mice lacking bacterial microbiota. Lastly, we find that IECs lacking Ifnlr1 are hyper-susceptible to initiation of murine rotavirus infection. These observations indicate that bacterial microbiota stimulate ISGs in localized regions of the intestinal epithelium at homeostasis, thereby preemptively activating antiviral defenses in vulnerable IECs to improve host defense against enteric viruses.

Editor's evaluation

The paper shows that homeostatic interferon-stimulated gene expression in the mouse intestinal epithelium (which is not uniform but concentrated in mature epithelial pockets) depends on the presence of bacterial microbiota and intestinal epithelial cell-intrinsic expression of the IFN-λ receptor associated with leucocyte IFN-λ production. Mouse rotavirus infection (an intestinal epithelial pathogen) is more effectively initiated in the absence of this homeostatic IFN-λ response although it remains unclear how the localized pockets of interferon-stimulated gene expression impart protection.

https://doi.org/10.7554/eLife.74072.sa0

Introduction

Interferons (IFNs) are a family of cytokines produced in response to infection that signal IFN receptor-bearing cells to induce transcription of hundreds of IFN-stimulated genes (ISGs). These ISGs perform diverse functions, but many cooperate to induce an antiviral state (Sadler and Williams, 2008; Schoggins and Rice, 2011). There are three types of IFNs: type I IFNs (IFN-αs, IFN-β, others), type II IFN (IFN-γ), and type III IFNs (IFN-λs). These three types are differentiated by receptor usage (type I IFN receptor: Ifnar1/Ifnar2; type II IFN receptor: Ifngr1/Ifngr2; type III IFN receptor: Ifnlr1/Il10rb), but all three receptor complexes signal through Janus-kinase (JAK) and signal transducer and activator of transcription (STAT) factors to stimulate ISG transcription (Ingle et al., 2018; Schneider et al., 2014). Type I and III IFNs are directly stimulated by host detection of microbe-associated molecular patterns (MAMPs) such as viral nucleic acids, and the prominent contribution of these IFN types to antiviral defense is reflected by the breadth of evasion strategies used by diverse viral families to prevent their production or action (Levy and García-Sastre, 2001; Rojas et al., 2021; Taylor and Mossman, 2013). The type I IFN receptor is expressed broadly across most cell types, whereas the type III IFN receptor, Ifnlr1, is primarily restricted to epithelial cells (Kotenko and Durbin, 2017; Sommereyns et al., 2008). Accordingly, IFN-λ is of particular importance in effective antiviral defense of barrier tissues.

Interestingly, previous studies in mice have noted that intestinal epithelial cells (IECs) are hyporesponsive to type I IFN (Lin et al., 2016; Hernández et al., 2015; Pott et al., 2011; Van Winkle et al., 2020). The responsiveness of IECs to type I IFN appears to be developmentally regulated because the IECs of adult mice exhibit weaker type I responses than IECs of neonatal mice (Lin et al., 2016). Additionally, infection of mice deficient in IFN receptors (Ifnar1, Ifngr1, Ifnlr1; single or double knockouts) with enteric viruses indicates that IFN-λ is the predominant antiviral IFN type that controls viral replication in the gastrointestinal epithelium (Hernández et al., 2015; Nice et al., 2015; Pott et al., 2011). IECs can robustly respond to IFN-λ with upregulation of canonical antiviral ISGs and increased resistance to infection by enteric viruses, such as rotaviruses and noroviruses (Baldridge et al., 2017; Nice et al., 2015; Pott et al., 2011). Mouse rotavirus is a natural pathogen of mice that infects enterocytes located at the tips of villi in the small intestine (Burns et al., 1995). Rotaviruses have developed mechanisms to antagonize the induction of IFN by infected cells, suggesting that evasion of the IFN response is necessary for efficient epithelial infection (Arnold et al., 2013). Indeed, prophylactic administration of IFN-λ significantly reduces the burden of mRV infection, demonstrating its potential for mediating epithelial antiviral immunity to this pathogen (Mahlakõiv et al., 2015; Lin et al., 2016; Pott et al., 2011; Van Winkle et al., 2020). However, a protective role of the endogenous IFN-λ response to mRV infection is less clear, perhaps reflecting the success of mRV evasion mechanisms.

Epithelial immunity in the gut must be appropriately balanced to protect the intestinal epithelium while preventing loss of barrier integrity and intrusion by microbes that are abundant within the gastrointestinal tract. The bacterial microbiota in the intestine perform critical functions by aiding in host metabolism (Krajmalnik-Brown et al., 2012), providing a competitive environment to defend against pathogens (Kim et al., 2017), and initiating and maintaining host immune function during homeostasis (Honda and Littman, 2016; Rooks and Garrett, 2016). In this complex environment host epithelial and immune cells detect bacteria and viruses using a suite of pattern recognition receptors (PRRs) that sense the presence of MAMPs. Stimulation of PRRs, such as the toll-like receptor (TLR) family, activates antimicrobial and antiviral defenses providing local protection in many tissues (Chu and Mazmanian, 2013; Iwasaki and Medzhitov, 2015; Thompson et al., 2011). TLR-dependent pathways induce production of IFNs, primarily by signaling through TIR-domain-containing adapter-inducing interferon-β (TRIF) and myeloid differentiation primary response 88 (MYD88) adapter proteins (Monroe et al., 2010; Odendall et al., 2017), providing a mechanism by which bacterial MAMPs can initiate IFN responses.

Signals from the bacterial microbiota have been shown to elicit a steady-state type I IFN response in systemic tissues and cell types that can prime antiviral immunity by several independent studies (Abt et al., 2012; Bradley et al., 2019; Ganal et al., 2012; Steed et al., 2017; Stefan et al., 2020; Winkler et al., 2020). Additionally, a steady-state ISG signal has been observed in the intestine of uninfected mice (Baldridge et al., 2015; Lin et al., 2016; Stockinger et al., 2014), but this intestinal response remained poorly characterized. Together with the observed hypo-responsiveness of IECs to type I IFN, these findings suggested that bacterial microbiota may stimulate an IFN-λ response in the gut. To explore this interaction, we undertook the present study to assess the role of bacterial microbiota in induction of enteric ISGs at homeostasis using a combination of broad-spectrum antibiotics (ABX) and genetically modified mice.

In this study, we uncovered an ISG signature in IECs that was dependent on the presence of bacteria and IFN-λ signaling (hereafter referred to as ‘homeostatic ISGs’). This panel of genes was present in wild-type (WT) mice with conventional microbiota and was reduced in WT mice treated with ABX and in mice lacking Ifnlr1. We revealed that homeostatic ISG expression is (i) restricted to the intestinal epithelium across both the ileum and colon, (ii) independent of type I IFN signaling, and (iii) associated with IFN-λ transcript expression by epithelium-associated CD45+ leukocytes. Surprisingly, we found that homeostatic ISGs are not expressed uniformly by all IECs; rather, expression is concentrated in localized pockets of IECs and in differentiated IECs relative to crypt-resident progenitors. These patterns of localized ISG expression are corroborated by independently generated single-cell RNA sequencing (scRNA-seq) data from mouse and human IECs. We also found that ISG expression can be increased in ABX-treated mice by reconstitution of bacterial microbiota or administration of bacterial lipopolysaccharide (LPS). Finally, we found that this microbiota-stimulated ISG signature provides protection from initiation of mRV infection. Cumulatively, this study found that bacteria initiate preemptive IFN-λ signaling in localized areas to protect IECs from enteric viruses.

Results

Bacterial microbiota stimulate IFN-λ response genes in the ileum at homeostasis

To determine the effect of bacterial microbiota on homeostatic IFN-λ responses, we compared gene expression in whole ileum tissue for the following experimental groups: (i) wild-type (WT) C57BL/6J mice intraperitoneally injected with IFN-λ as compared to unstimulated WT mice, (ii) WT mice with conventional microbiota as compared to WT mice treated with an antibiotic cocktail (ABX) to deplete bacteria, (iii) WT mice with conventional microbiota as compared to Ifnlr1-/- mice with conventional microbiota, and (iv) Ifnlr1-/- mice with conventional microbiota compared to Ifnlr1-/- mice treated with ABX (Figure 1A). To rule out contributions of Ifnlr1 toward an altered intestinal bacterial microbiota, we performed 16S rRNA sequencing on stool from Ifnlr1+/+, Ifnlr1+/-, and Ifnlr1-/- mice and did not find statistically significant differences in alpha-diversity and beta-diversity measurements with the statistical power available from 12 mice per group (Figure 1—figure supplement 1). For each comparison of ileum gene expression shown in Figure 1A, we performed gene set enrichment analysis (GSEA) to determine enrichment or depletion of hallmark gene sets (Liberzon et al., 2015). The hallmark gene set that was most enriched in IFN-λ-treated WT mice compared to unstimulated WT mice was INTERFERON_ALPHA_RESPONSE (Figure 1B, Figure 1—figure supplement 2). Therefore, this gene set reflects differences in IFN-λ responses between experimental conditions.

Figure 1 with 3 supplements see all
Bacterial microbiota stimulate interferon-lambda (IFN-λ) response genes in the ileum at homeostasis.

(A) Depiction of experimental treatments and comparison groups. Following the indicated treatments, a segment of whole ileum tissue was harvested and analyzed by RNA sequencing for differentially expressed genes between paired conditions. (B–E) Gene set enrichment analysis of INTERFERON_ALPHA_RESPONSE hallmark genes (interferon-stimulated genes [ISGs]) was performed with the following comparisons: wild-type (WT) mice treated with 25 µg of IFN-λ relative to WT mice treated with PBS (B), WT mice treated with antibiotics (ABX) relative to untreated WT mice (C), Ifnlr1-/- mice relative to WT mice (D), and Ifnlr1-/- mice treated with ABX relative to untreated Ifnlr1-/- mice (E). Normalized enrichment score (NES) and false discovery rate (FDR) are overlaid for each comparison with significant FDR’s highlighted (red). (F) A Venn diagram depicting the total number of differentially expressed genes that are (i) increased with IFN-λ stimulation (orange), (ii) decreased with ABX treatment (green), or (iii) decreased in Ifnlr1-/- mice relative to WT (blue). An overlapping subset of 21 genes was shared among all three comparisons (red box). (G) A graph and heatmap of the relative expression of the 21 genes that overlap in all experimental groups in (F) (‘homeostatic ISGs’). Statistical significance in (G) was determined by one-way ANOVA with Tukey’s multiple comparisons where ns = p > 0.05 and **** = p < 0.0001.

Figure 1—source data 1

Gene set enrichment analysis, differential gene expression analysis, and counts of homeostatic interferon-stimulated genes (ISGs) from RNA sequencing analysis.

https://cdn.elifesciences.org/articles/74072/elife-74072-fig1-data1-v2.xlsx

To deplete bacterial microbiota, we administered a cocktail of broad-spectrum ABX, and demonstrated that this treatment reduced bacterial 16S gene copies in stool to below the limit of detection (Figure 1—figure supplement 3). GSEA of WT mice treated with ABX showed a significant depletion (negative enrichment score) of INTERFERON_ALPHA_RESPONSE hallmark genes in the ileum relative to WT mice with conventional microbiota (Figure 1C). These data indicate that ISGs are present at steady state in the ileum of specific-pathogen-free mice with conventional microbiota and suggest that microbiota stimulate expression of these genes in the ileum at homeostasis.

Type I, II, and III IFN responses have substantial gene expression overlap; therefore, to prove that the IFN-λ receptor was necessary for expression of ISGs at homeostasis, we analyzed gene expression in the ileum of Ifnlr1-/- mice. Indeed, GSEA showed significant reduction of INTERFERON_ALPHA_RESPONSE hallmark genes in the ileum of Ifnlr1-/- mice compared to WT mice (Figure 1D). Furthermore, expression of INTERFERON_ALPHA_RESPONSE hallmark genes was not significantly decreased in Ifnlr1-/- mice treated with ABX compared to Ifnlr1-/- mice with conventional microbiota (Figure 1E). In contrast to INTERFERON_ALPHA_RESPONSE, other hallmark gene sets such as INFLAMMATORY_RESPONSE and IL6_JAK_STAT3_SIGNALING were significantly decreased by ABX treatment in both WT mice and Ifnlr1-/- mice (Figure 1—figure supplement 2), which indicates a selective requirement for Ifnlr1 to elicit ISGs and a minimal effect of Ifnlr1 deficiency on other microbiota-dependent genes. Together, these findings suggest that Ifnlr1 is necessary for the bacterial microbiota-dependent expression of ISGs in the ileum at homeostasis. Since ISGs are enriched in mice upon IFN-λ stimulation and are decreased in Ifnlr1-/- and ABX-treated mice, we conclude that the bacterial microbiota stimulates IFN-λ responses in the ileum at homeostasis.

To define a core set of bacterial microbiota-dependent, Ifnlr1-dependent ‘homeostatic ISGs’, we determined the overlap of differentially expressed genes (DEGs) by defining genes with: (i) increased expression upon IFN-λ stimulation in WT mice, (ii) decreased expression upon ABX treatment in WT mice, and (iii) decreased expression in Ifnlr1-/- mice relative to WT mice. The DEGs shared by each of these comparisons comprised a set of 21 genes that are decreased upon treatment with ABX and loss of Ifnlr1, and are induced in response to IFN-λ (Figure 1F). This set of homeostatic ISGs includes antiviral genes that are dependent on bacterial microbiota and the IFN-λ pathway. Comparison of homeostatic ISG transcript counts between experimental treatments revealed similar insights as prior GSEA. WT mice treated with IFN-λ had higher expression of all homeostatic ISGs than untreated mice, whereas WT mice with a conventional microbiota had higher expression of homeostatic ISGs than Ifnlr1-/- mice and ABX-treated mice of both genotypes (Figure 1G). We did not detect additional decreases in these homeostatic ISGs in ABX-treated Ifnlr1-/- mice relative to conventional Ifnlr1-/- mice, suggesting that Ifnlr1 is necessary for expression of homeostatic ISGs (Figure 1G). These results indicate that there is modest but significant expression of ISGs at homeostasis that is lost with Ifnlr1 deficiency or ABX treatment. Together, these analyses revealed a homeostatic signature of ISGs in the ileum that depends upon the presence of bacterial microbiota and on intact IFN-λ signaling.

Homeostatic, microbiota- and Ifnlr1-dependent ISGs are primarily expressed in intestinal tissues

To complement the results of the RNA-seq and extend this analysis to other tissue sites, we quantified tissue-level expression of a panel of three ISGs by quantitative PCR (qPCR): Ifit1, Oas1a, and Stat1. Ifit1 and Stat1 were present among the 21 homeostatic ISGs in the preceding analysis and Oas1a was included as a representative canonical ISG that we hypothesized would be present in the homeostatic signature, but did not meet the statistical criteria used to define the core set of 21 homeostatic ISGs (Figure 1F–G). We assessed absolute abundance of these ISG transcripts in the ileum, colon, mesenteric lymph nodes (MLN), and spleen tissue of WT and Ifnlr1-/- mice with or without ABX treatment (Figure 2A–C). Consistent with our RNA-seq data, these ISGs were reduced in the ilea of Ifnlr1-/- mice and ABX-treated WT mice compared to WT mice with conventional microbiota (Figure 2A). Second, homeostatic ISGs were expressed in WT colonic tissue and were significantly decreased in colonic tissue of Ifnlr1-/- mice and ABX-treated WT mice (Figure 2B). These data indicate that homeostatic ISGs in both the ileum and colon were dependent on Ifnlr1 and the bacterial microbiota; however, these homeostatic ISGs were more abundantly expressed in ileal tissue than colonic tissue (Figure 2—figure supplement 1). To confirm that treatment with ABX does not ablate the ability of the intestine to respond to IFN-λ, we stimulated ABX-treated mice with intraperitoneal IFN-λ and harvested ileum tissue. Stimulation with small amounts of IFN-λ rescued ISG expression in whole tissue (Figure 2—figure supplement 2), indicating that that reduction of homeostatic ISG expression upon treatment with ABX is not due to an inability of the intestine to respond to IFN-λ.

Figure 2 with 2 supplements see all
Homeostatic, microbiota- and Ifnlr1-dependent interferon-stimulated genes (ISGs) are primarily expressed in intestinal tissues.

A segment of ileum or colon tissue from wild-type (WT) or Ifnlr1-/- mice was harvested following H2O or antibiotics (ABX) treatment and the ISGs Ifit1, Stat1, and Oas1a were analyzed by quantitative PCR (qPCR). Transcripts were quantified in ileum (A), colon (B), or mesenteric lymph node (MLN) and spleen (C) with normalization to untreated WT mice. Data points represent individual mice and data are pooled from 10 to 20 independent experiments in (A–B) and 15 independent experiments in (C). Statistical significance was determined by one-way ANOVA with Tukey’s multiple comparisons in (A–B) and Dunnett’s multiple comparisons in (C). * = p < 0.05, ** = p < 0.01, *** = p < 0.001, **** = p < 0.0001.

Figure 2—source data 1

Values and statistical tests of interferon-stimulated genes (ISG) expression in ileum, colon, mesenteric lymph node (MLN), and spleen.

https://cdn.elifesciences.org/articles/74072/elife-74072-fig2-data1-v2.xlsx

Enteric colonization by bacteria was shown to stimulate systemic type I IFN responses (Abt et al., 2012; Bradley et al., 2019; Ganal et al., 2012; Steed et al., 2017; Stefan et al., 2020; Winkler et al., 2020), so we assessed whether the decreases in ISGs upon ABX treatment or loss of Ifnlr1 in the ileum were recapitulated in systemic immune tissues. We quantified Ifit1, Stat1, and Oas1a expression in the MLN and the spleen and found that ABX treatment and Ifnlr1 deletion did not reduce ISGs in these tissues (Figure 2C). Although we observe increases in Ifit1 expression in MLN upon treatment with ABX, these results are not recapitulated by Stat1 and Oas1a expression, and no significant changes in ISG expression were detected in the spleen. Cumulatively, these data indicate that homeostatic ISGs include genes beyond the core signature identified in Figure 1 (e.g. Oas1a), that homeostatic ISGs are present in colonic tissue, and that homeostatic IFN-λ-stimulated genes are most prominent in enteric tissues.

Homeostatic ISG expression in the intestine is independent of type I IFN

To determine whether detection of enteric bacteria by TLRs stimulates homeostatic ISGs, we measured tissue ISG expression in mice that were deficient in TRIF or MYD88. Signaling through TRIF results in activation of interferon regulatory factors (IRFs), such as IRF3 and IRF7, that commonly contribute to IFN induction (Honda et al., 2005a; Osterlund et al., 2007; Schmid et al., 2010). Additionally, signaling through MYD88 can aid initiation of IFN expression, through nuclear factor kappa-light-chain-enhancer of activated B cells transcription factor family members (Osterlund et al., 2007) and IRF7 (Honda et al., 2005b; Tomasello et al., 2018). Ifit1 expression in the ileum and colon of Trif-/- mice was not significantly different than in WT mice, and Ifit1 expression was reduced with ABX in Trif-/- mice (Figure 3A). However, we found that mice lacking Myd88 exhibited tissue-specific decreases in Ifit1 expression, with significant decreases in the ileum, but not in the colon, relative to WT mice (Figure 3A). These data are consistent with a previous report (Stockinger et al., 2014) and suggest that signaling through MYD88, but not TRIF, is necessary for homeostatic ISG expression in the ileum, whereas other factors may dominate in the colon.

Homeostatic interferon-stimulated gene (ISG) expression is independent of type I interferon (IFN).

Ifit1 expression levels were assessed by quantitative PCR (qPCR) from the ileum or colon of (A) Trif-/- and Myd88-/-, (B) Irf3-/- and Irf7-/-, or (C) Ifnar1-/- and Stat1-/- mice treated with or without antibiotics (ABX) normalized to untreated, wild-type (WT) mice. Some WT controls are shared across experiments in (A–C). (D) Ifit1 expression was measured by qPCR in ileum tissue of indicated genotypes. Data points represent individual mice and are pooled from at least six independent experiments in (A–C) and two independent experiments in (D). Statistical significance was determined by one-way ANOVA with Dunnett’s multiple comparisons in (A–C) and one-way ANOVA with Tukey’s multiple comparisons in D, where * = p < 0.05, ** = p < 0.01, *** = p < 0.001, **** = p < 0.0001.

Figure 3—source data 1

Values and statistical tests of interferon-stimulated gene (ISG) expression in the ileum and colon of genetic knockout mice.

https://cdn.elifesciences.org/articles/74072/elife-74072-fig3-data1-v2.xlsx

To expand these findings, we assessed the role of IRF3 and IRF7 transcription factors that are commonly activated downstream of MYD88 and TRIF. We did not observe significant decreases in Ifit1 expression in the ileum or the colon of Irf3-/- mice as compared to WT (Figure 3B), indicating that IRF3 is not required for homeostatic IFN-λ induction. However, we observed a modest (twofold) decrease in Ifit1 expression in both the ileum and colon of Irf7-/- mice when compared to WT mice (Figure 3B). Although IRF7 is implicated by these data, it does not appear to be strictly required for homeostatic expression of Ifit1 because expression is further reduced by ABX treatment (Figure 3B). These data suggest that IRF7 is not necessary for the homeostatic ISG response to the bacterial microbiota.

Type I and III IFNs stimulate overlapping ISG responses that are both dependent on STAT1. Therefore, to further investigate the contribution of type I IFN signaling to homeostatic ISG expression, we quantified Ifit1 expression in ileum and colon tissue from Ifnar1-/- or Stat1-/- mice. Ifit1 expression was not significantly different in either ileum or colon tissue of Ifnar1-/- mice compared to WT (Figure 3C). However, Ifit1 expression was significantly lower in ileum and colon tissue of Stat1-/- mice compared to Ifnar1-/- and WT mice. Importantly, treatment of Stat1-/- mice with ABX did not further reduce Ifit1 expression, emphasizing the necessity of STAT1 for this homeostatic ISG response.

Lastly, to determine whether type I IFN signaling plays a compensatory role in homeostatic ISG expression in the absence of Ifnlr1, we bred mice with heterozygous expression of both Ifnlr1 and Ifnar1 (Ifnlr1-/+/Ifnar1-/+) with mice that lack Ifnlr1 and Ifnar1 (Ifnlr1-/-/Ifnar1-/-). This breeding scheme produced littermate-matched mice that were Ifnlr1-/+/Ifnar1-/+, Ifnlr1-/-/Ifnar1-/-, and mice singly deficient in Ifnlr1 (Ifnlr1-/-/Ifnar1-/+) or Ifnar1 (Ifnlr1-/+/Ifnar1-/-). We found that Ifit1 expression was not significantly different in Ifnlr1-/+/Ifnar1-/- relative to Ifnlr1-/+/Ifnar1-/+ controls in ileal tissue (Figure 3D). Ifit1 expression in Ifnlr1-/-/Ifnar1-/+ was significantly lower compared to Ifnlr1-/+/Ifnar1-/+ controls, but was not significantly different from Ifnlr1-/-/Ifnar1-/- mice (Figure 3D). Cumulatively, these data indicate that homeostatic ISG expression in the intestine is partly dependent on MYD88, and is independent of type I IFN signaling.

Homeostatic ISG expression in the intestine is restricted to epithelial cells

Given the primarily epithelial expression of Ifnlr1, we assessed which compartment of the ileum expresses homeostatic ISGs by isolating a stripped intestinal epithelial fraction and digesting the underlying lamina propria. We assessed ISG expression in these two fractions from WT and Ifnlr1-/- mice treated with or without ABX. Treatment with ABX or loss of Ifnlr1 reduced homeostatic ISGs in the IEC fraction, but the lamina propria had low expression of these ISGs relative to the epithelium of untreated WT mice, regardless of treatment or genotype (Figure 4A).

Homeostatic interferon-stimulated gene (ISG) expression in the intestine is restricted to epithelial cells.

(A) Ifit1, Stat1, and Oas1a expression was quantified in stripped epithelial cells or in the lamina propria cells of the ileum. Comparisons were performed between wild-type (WT) and Ifnlr1-/- mice with or without antibiotics (ABX) treatment, and ISG expression was normalized to WT values. (B–C) Ifit1 and Oas1a expression from the (B) ileum or (C) colon of mice with conditional presence (flox) or absence (cKO) of Ifnlr1 in intestinal epithelial cells. Data points represent individual mice and are pooled from three independent experiments in (A) and two independent experiments in (B–C). Statistical significance was determined by one-way ANOVA with Dunnett’s multiple comparisons. * = p < 0.05, ** = p < 0.01, *** = p < 0.001.

Figure 4—source data 1

Values and statistical tests of interferon-stimulated gene (ISG) expression in the ileum and colon of genetic knockout mice.

https://cdn.elifesciences.org/articles/74072/elife-74072-fig4-data1-v2.xlsx

IECs express abundant Ifnlr1, but other intraepithelial cell types do not (Hernández et al., 2015; Sommereyns et al., 2008). To determine whether Ifnlr1 expression by IECs was required for homeostatic ISG expression, we used mice with IECs that are conditionally deficient in Ifnlr1 (Ifnlr1 IEC-cKO) and littermates that retain normal Ifnlr1 expression (Ifnlr1flox/flox) (Baldridge et al., 2017). Ifit1 and Oas1a expression in the ileum (Figure 4B) and colon (Figure 4C) of Ifnlr1flox/flox mice was decreased upon treatment with ABX, consistent with the phenotype observed in WT mice. Conditional deletion of Ifnlr1 in IECs reduced Ifit1 and Oas1a to a similar extent as the reduction observed in Ifnlr1-/- animals, above (Figure 2A–B). Together, these findings indicate that homeostatic ISGs are dependent on Ifnlr1 expression by IECs.

Bacterial microbiota stimulate expression of Ifnl2/3 by CD45+ cells

A previous study (Hernández et al., 2015) noted the presence of IFN-λ transcripts (Ifnl2/3) at homeostasis in CD45+ cells within the stripped intestinal epithelium, but not in the lamina propria. To extend these findings and to determine whether CD45+ cells in the intestinal epithelium produce IFN-λ in response to bacterial microbiota at homeostasis, we enriched cell subsets from epithelial or lamina propria fractions of the ileum for quantification of Ifnl2/3 by qPCR. We treated WT mice with control, ABX, or stimulation with a synthetic dsRNA analogue (poly I:C). We then magnet-enriched EpCAM+ or CD45+ cells from stripped intestinal epithelium or digested lamina propria ileal tissue (Figure 5—figure supplement 1). We found that CD45+ cells from the intestinal epithelial and lamina propria fraction expressed detectable Ifnl2/3 at homeostasis, but EpCAM+ cells did not, which is consistent with Mahlakõiv et al. Furthermore, we found that expression of Ifnl2/3 in CD45+ cells from the epithelial fraction was significantly reduced with ABX treatment, and CD45+ cells from lamina propria had a non-statistically significant (p = 0.0523) reduction in Ifnl2/3 expression with ABX treatment (Figure 5A). From these data, we conclude that epithelium-associated CD45+ leukocytes are the likely source of homeostatic IFN-λ in response to bacterial microbiota, but we do not rule out additional involvement of CD45+ cells in the lamina propria.

Figure 5 with 1 supplement see all
Bacterial microbiota stimulate expression of Ifnl2/3 by CD45+ cells.

(A–B) Cellular suspensions from the ileal epithelium and lamina propria were harvested from wild-type (WT) mice treated with H2O, ABX, or stimulated with poly I:C. Resulting cells were enriched for EpCAM-positive and CD45+ cells by magnetic separation. Ifnl2/3 expression (A) and Ifnb1 expression (B) were quantified from each enriched cellular fraction by quantitative PCR (qPCR). Data points represent individual mice and are pooled from two independent experiments. Statistical significance was determined by two-way ANOVA with Dunnett’s multiple comparisons, where * = p < 0.05 and **** = p < 0.0001.

Figure 5—source data 1

Values and statistical tests of interferon expression by CD45+ cells in the ileum.

https://cdn.elifesciences.org/articles/74072/elife-74072-fig5-data1-v2.xlsx

Similar to Ifnl2/3, we found that CD45+ cells of the epithelial fraction modestly expressed IFN-β transcript (Ifnb1) at homeostasis. Additionally, CD45+ cells of the lamina propria robustly expressed Ifnb1 at homeostasis. However, unlike Ifnl2/3, Ifnb1 was not decreased in mice treated with ABX (Figure 5B). These data indicate that Ifnb1 expression by CD45+ cells in the intestine is less dependent on stimulation by bacterial microbiota relative to Ifnl2/3, consistent with the dominant role of IFN-λ responses in driving homeostatic ISG expression in the epithelium.

Homeostatic ISG expression in the small intestine is highly localized

Homeostatic ISG expression in the ileum was of relatively low magnitude when compared to IFN-λ treatment (Figures 12). Therefore, we hypothesized that a low abundance of homeostatic ISG expression would be uniformly distributed between IECs of the intestinal epithelium, and we sought to assess the distribution of the homeostatic ISG, Ifit1, using in situ hybridization (RNAscope). Contrary to our hypothesis, RNAscope staining of the ileum from untreated WT mice revealed localized pockets of robust Ifit1 expression in individual villi rather than ubiquitously low expression throughout the intestinal epithelium (Figure 6A). Additionally, Ifit1 localization was skewed away from the crypt and toward the tips of individual villi within the ileum, and this localization was not specific to Ifit1 because the distinct ISG Usp18 co-localized with Ifit1 (Figure 6A). These data indicate that homeostatic ISGs are sporadically expressed in individual villi and are primarily localized to mature enterocytes that are most distally located in villi. We determined that localized ISG expression within individual villi was not due to a localized ability to respond to IFN-λ because stimulation with exogenous IFN-λ resulted in Ifit1 expression within all intestinal villi, but not intestinal crypts (Figure 6—figure supplement 1). The minimal expression of Ifit1 within intestinal crypts following exogenous IFN-λ treatment suggests that homeostatic ISGs are localized to mature enterocytes because intestinal crypts do not exhibit robust responses to IFN-λ. Additionally, the non-uniform distribution of homeostatic Ifit1 expression was ablated in the ileum of mice treated with ABX (Figure 6B–D), consistent with a dependency on bacterial microbiota. To determine whether localized Ifit1 expression was dependent on IEC expression of Ifnlr1, we assessed the distribution of Ifit1 expression in the ilea of littermate Ifnlr1flox/flox and Ifnlr1 IEC-cKO mice. We found that the localized Ifit1 expression observed in untreated WT mice was recapitulated in Ifnlr1flox/flox mice (Figure 6E), but these areas of ISG expression were ablated in Ifnlr1 IEC-cKO mice (Figure 6F–G). Lastly, we found that this discrete localization of the homeostatic ISG response is not limited to the ileum, as localized Ifit1 staining is also observed in the colonic epithelium (Figure 6—figure supplement 2). Together, our analyses indicated that homeostatic ISGs are expressed in a highly localized manner within the intestinal epithelium.

Figure 6 with 2 supplements see all
Homeostatic interferon-stimulated gene (ISG) expression in the small intestine is highly localized.

The ilea of wild-type (WT), Ifnlr1flox/flox, or Ifnlr1 intestinal epithelial cell (IEC)-cKO mice were harvested, processed into a Swiss rolls, and stained by in situ hybridization for the ISGs, Ifit1 (red), and Usp18 (green), with a DAPI (blue) counterstain.(A) Representative high-magnification images of co-localized Ifit1 (red) and Usp18 (green) expression in the ileum of WT mice, see arrows. (B–D) WT mice treated with H2O control or antibiotics (ABX) with quantification of Ifit1 area relative to the total area of the section across replicate mice. (E–G) Ifnlr1flox/flox or Ifnlr1 IEC-cKO mice at homeostasis with quantification of Ifit1 area relative to the total area of the section across replicate mice. Scale bar = 100 µm in (A) and 500 µm in (B–G). Statistical significance was calculated by unpaired t-test where ** = p < 0.01. Each data point in (D) and (G) represents an individual mouse from two independent experiments.

Figure 6—source data 1

Values and statistical tests of Ifit1 area in the ileum of treated and genetic knockout mice.

https://cdn.elifesciences.org/articles/74072/elife-74072-fig6-data1-v2.xlsx

Mature enterocytes express homeostatic ISGs in public single-cell datasets from mouse and human

To determine the extent of conservation of homeostatic ISG expression by IECs, we performed orthogonal analyses of publicly available scRNA-seq datasets from mouse (Haber et al., 2017) and human (Elmentaite et al., 2020) IECs. Recently, Haber et al. published a large scRNA-seq dataset that profiled sorted IECs from the small intestine of specific pathogen-free mice and defined 15 distinct IEC subtypes (Haber et al., 2017). We analyzed IECs from this dataset, and found that of the 21 homeostatic ISGs identified in Figure 1, 19 were present in the Haber et al. single-cell dataset. We determined the percentage of each epithelial cell subtype that expresses each individual homeostatic ISG, and generated a heatmap with hierarchical clustering to group IEC subtypes that have similar ISG expression patterns (Figure 7A). Homeostatic ISGs were predominantly expressed in mature enterocyte subtypes, which clustered separately from crypt-resident progenitor IECs such as transit amplifying (TA) cells and stem cells (Figure 7A). To compare homeostatic ISG expression between polar extremes of the crypt-villus axis, we grouped IEC subtypes that represented enterocytes (mature enterocyte cells) and crypt-associated cells (TA cells and stem cells) to compare the overall proportions of these cells with homeostatic ISG expression. We found that a significantly higher percentage of enterocytes express homeostatic ISGs than crypt-associated cells, but that these homeostatic ISGs were expressed in a relatively small proportion of enterocytes (<20%) (Figure 7B). Notably, Ifit1 (highlighted in red) was present in ~5% of enterocytes by scRNA-seq, which is consistent with our observation of 1–4% Ifit1-positive area by imaging the mouse small intestine (Figure 6B–G). Furthermore, the relative absence of ISG-positive crypt-associated cells in this scRNA-seq data is consistent with our observation that intestinal crypts lacked Ifit1 and Usp18 expression by imaging.

Mature enterocytes express homeostatic interferon-stimulated genes (ISGs) in public single-cell datasets from mouse and human.

(A–B) A mouse intestinal epithelial cell (IEC) single-cell transcriptional dataset (Haber et al., 2017) (GSE92332) was analyzed to determine the percentage of each epithelial cell subtypes that express homeostatic ISGs. (A) Heatmap depicting the proportion of each epithelial cell type expressing 19 of the 21 homeostatic ISGs identified in Figure 1. (B) Enterocyte subtypes (blue text) and crypt-resident progenitor subtypes (green text) cells were grouped and the percentage of cells that express each homeostatic ISG was compared. (C–D) A human IEC single-cell transcriptional dataset (Elmentaite et al., 2020) (E-MTAB-8901) was analyzed for the percentage of epithelial cell subtypes that express homeostatic ISGs. (C) A heatmap depicting the percentage of IEC subtypes that express human orthologs of murine homeostatic ISGs identified in Figure 1. (D) The mature enterocyte subtype (blue text) and crypt-resident progenitor subtypes (green text) were grouped and the percentage of cells that express each homeostatic ISG was compared. Lines in (B) and (D) link paired ISGs in each IEC subset. Statistical significance in (B) and (D) was calculated by Wilcoxon test where * = p < 0.05, and *** = p < 0.001.

Figure 7—source data 1

Values and statistical tests of the percentage of cells expressing individual interferon-stimulated genes (ISGs) in single-cell RNA sequencing datasets.

https://cdn.elifesciences.org/articles/74072/elife-74072-fig7-data1-v2.xlsx

We expanded our investigation to an scRNA-seq dataset from the ileum of healthy, human, pediatric patients that was previously described (Elmentaite et al., 2020). IECs from this dataset were previously clustered by Elmentaite et al. and annotated as: enterocytes, early enterocytes, tuft cells, enteroendrocrine cells, BEST4 enterocytes, goblet cells, TA cells, and crypt. For our analysis of homeostatic ISG expression, we excluded IEC subtypes with fewer than 20 constituent cells, which retained five annotated groups: enterocytes, early enterocytes, goblet cells, TA cells, and crypt. Of the 21 homeostatic ISGs identified in Figure 1, 14 orthologous human genes were present in these data. Similar to analysis in Figure 7A, we determined the percentage of each group that expressed each individual homeostatic ISG and generated a heatmap of these data with hierarchical clustering to group IEC subtypes that have similar ISG expression patterns (Figure 7C). Enterocyte, early enterocyte, and goblet subtypes clustered separately from TA cells and crypt, with the highest proportion of homeostatic ISGs being present in the enterocyte subtype (Figure 7C). As with mouse IEC data, above, we grouped annotated cells that localize in the crypt (TA cells and crypt) and compared overall proportions of homeostatic ISG expression with the mature enterocyte subtype (Figure 7D). Similar to mice, homeostatic ISGs were present in significantly more enterocytes than crypt-associated cells, and most homeostatic ISGs in this human dataset were present in a relatively small proportion of cells (<20%). These human data suggest that ISGs may be present in a small proportion of IECs from healthy, human ileal tissue at homeostasis and may share the localization observed in our murine analyses. Our analysis of a murine scRNA-seq dataset support our previous observations that homeostatic ISGs are not ubiquitously expressed throughout the intestinal epithelium; rather, they are expressed in a minority of IECs and skewed toward mature enterocytes along the crypt-villus axis. Our analysis of a human scRNA-seq dataset are consistent with observations in the murine model, though future studies will be required to definitively address the existence of homeostatic ISGs in human tissue.

Assessing the effect of peroral bacterial products on homeostatic ISGs in ABX-treated mice

To further define the relationship between bacteria and homeostatic ISGs, we assessed whether oral administration of fecal contents or purified LPS (a bacterial MAMP and TLR4 agonist) could restore localized ISG expression in mice with depleted bacterial microbiota. Control groups of Ifnlr1flox/flox mice with conventional microbiota (Figure 8A) retained a localized Ifit1 expression pattern, whereas ABX-treated Ifnlr1flox/flox mice (Figure 8B) and Ifnlr1 IEC-cKO mice (Figure 8C) lacked Ifit1 expression. Oral LPS administered to conventional Ifnlr1flox/flox mice did not significantly alter the distribution or frequency of localized Ifit1 expression (Figure 8D). However, localized Ifit1 expression was visible in 4/12 ABX-treated Ifnlr1flox/flox mice administered LPS (Figure 8E and H–I). In ABX-treated Ifnlr1flox/flox mice treated with fecal transplant of conventional microbes, localized expression of Ifit1 was visible in 4/8 mice (Figure 8F and H–I). Importantly, we did not observe localized Ifit1 expression in Ifnlr1 IEC-cKO mice following LPS administration (Figure 8G), indicating that LPS-stimulated Ifit1 depends on IEC expression of Ifnlr1.

Figure 8 with 1 supplement see all
Assessing the effect of peroral bacterial products on localized homeostatic interferon-stimulated genes (ISGs) in antibiotics (ABX)-treated mice.

The ilea of treated wild-type (WT), Ifnlr1flox/flox, and Ifnlr1 intestinal epithelial cell (IEC)-cKO mice were harvested, processed into Swiss rolls, and stained by in situ hybridization for the ISG, Ifit1 (red), with a DAPI (blue) counterstain. (A–C) Representative images from Ifnlr1flox/flox mice treated with H2O control followed by PBS stimulation (A), ABX followed by PBS stimulation (B), or from Ifnlr1 IEC-cKO mice (B). (D) Representative images of Ifnlr1flox/flox mice treated with H2O control followed by lipopolysaccharide (LPS) stimulation. (E–F) Two representative images of Ifnlr1flox/flox mice treated with ABX followed by LPS stimulation (E) or ABX followed by fecal transplantation (F). (G) A representative image of Ifnlr1 IEC-cKO mice treated with ABX followed by LPS stimulation. (H) Quantification of Ifit1 area relative to the total area of each tissue section with a dashed line at the highest Ifnlr1 IEC-cKO value. The proportion of Ifit1-positive mice (above dashed line) and Ifit1-negative mice (below dashed line) are tabulated (I) and graphed (J) for Ifnlr1flox/flox mice of each condition. (K) rDNA was isolated from the luminal contents of mice at endpoint harvest. 16S gene copies were assessed by quantitative PCR (qPCR) and normalized to input. Limit of detection: dashed line. Where depicted, scale bar = 500 µm. Data points represent individual mice and are pooled from four independent experiments in (A–J) and from two independent experiments in (K). Statistical significance was determined by Kruskal-Wallis with Dunn’s multiple comparisons in (H), by one-way ANOVA with Dunnett’s multiple comparisons in (K), and by Fisher’s exact tests in (J) where * = p < 0.05, ** = p < 0.01, and **** = p < 0.0001.

Figure 8—source data 1

Values and statistical tests of Ifit1 area, the proportion of Ifit1-positive cells, and 16S gene copies.

https://cdn.elifesciences.org/articles/74072/elife-74072-fig8-data1-v2.xlsx

We noted that the visibility of localized Ifit1 signal following LPS administration or fecal transplant appeared largely binary (i.e. present or absent) in our imaging data (Figure 8E and F, representative Ifit1-positive and Ifit1-negative images). Using quantification of Ifit1 area (Figure 8H), we stratified mice into Ifit1-postive and Ifit1-negative groups (Figure 8I–J) based on a cutoff set at the maximal Ifit1 area value of Ifnlr1 IEC-cKO mice (dashed line in Figure 8H). Results of this unbiased stratification were consistent with visible Ifit1 staining and indicated that 8/8 conventional Ifnlr1flox/flox mice, 0/8 ABX-treated control mice, 4/12 LPS-treated mice, and 4/8 fecal transplant mice were Ifit1-positive (Figure 8I–J). Statistical analysis by Fisher’s exact test indicated that LPS administration non-significantly increased (p = 0.1022) the proportion of mice that were Ifit1-positive, whereas fecal reconstitution of ABX-treated mice significantly increased the likelihood of these mice being Ifit1-positive (Figure 8J). Importantly, mice that received fecal transplant had full restoration of 16S gene copies (Figure 8K) despite only 4/8 having homeostatic Ifit1 expression. These findings suggest that reconstitution of the homeostatic ISG signal by fecal transfer has incomplete penetrance at this timepoint, underscoring the incomplete presence of localized Ifit1 expression (4/12) in ABX-treated mice administered peroral LPS.

To corroborate and extend these findings, we performed orthogonal analyses of Ifit1, Stat1, and Oas1a expression in ileum tissue of WT mice treated with ABX followed by fecal transplant, LPS administration, administration of the TLR5 agonist: flagellin, or administration of TLR9 agonist: CpG DNA (Figure 8—figure supplement 1). Similar to imaging data, these qPCR data exhibited high variance. However, stratification of tissues into positive and negative for each ISG indicated that peroral administration of LPS to ABX-treated mice increased the proportion of ISG-positive tissues by 21–36% (Ifit1: p = 0.065; Stat1, p < 0.05; Oas1a, p < 0.05). Additionally, peroral administration of flagellin significantly increased the proportion of ISG-positive mice by 25–44% (Ifit1: p = 0.052; Stat1, p < 0.01; Oas1a, p < 0.05), whereas CpG DNA did not significantly increase the proportion of ISG-positive mice (Figure 8—figure supplement 1). Together, these data suggest that LPS and flagellin are sufficient to stimulate homeostatic ISG expression in a significant proportion of ABX-treated mice. The ability of multiple PRR ligands to stimulate homeostatic ISGs suggests that exposure to a variety of bacterial MAMPs is the basis for localized, homeostatic ISG expression.

The homeostatic IFN-λ response preemptively protects IECs from murine rotavirus infection

To assess the capacity of homeostatic ISGs to protect IECs from viral infection, we utilized infection with murine rotavirus (mRV), an IEC-tropic pathogen. Prior studies of rotaviruses have identified viral immune evasion genes that block IFN induction through multiple mechanisms (Arnold et al., 2013). However, we reasoned that preexisting ISG expression stimulated by the bacterial microbiome at homeostasis may preemptively protect IECs from the infection before viral gene expression is initiated. To determine the role of an epithelial IFN-λ response over the course of mRV infection in adult mice, we monitored daily shedding of viral genomes in the stool of Ifnlr1flox/flox mice and Ifnlr1 IEC-cKO littermates. We first detected mRV shedding in the stool on day 2 after inoculation and, at this early timepoint, Ifnlr1 IEC-cKO mice shed 20-fold more mRV genomes into their stool than Ifnlr1flox/flox mice (Figure 9A). However, at the peak of viral shedding between days 3 and 5, there were no significant differences between Ifnlr1flox/flox and Ifnlr1 IEC-cKO littermates (Figure 9A). This similarity at peak of viral shedding was consistent with an ability of mRV to evade the host IFN response once infection is established. Together, these findings suggest that Ifnlr1 IEC-cKO mice have defects in protection against initiation of mRV infection and that the protective capacity of endogenous IFN-λ signaling against mRV is primarily prophylactic in nature.

Figure 9 with 3 supplements see all
The homeostatic interferon-lambda (IFN-λ) response preemptively protects intestinal epithelial cells (IECs) from murine rotavirus (mRV) infection.

Ifnlr1flox/flox and Ifnlr1 IEC-cKO mice were infected with 100 SD50 (A) or 5000 SD50 (B–H) of mRV and stool (A) or stripped ileal IECs (B–G) were assessed for mRV infection. (A) Timecourse of mRV genome copies detected in the stool of Ifnlr1flox/flox and Ifnlr1 IEC-cKO mice by quantitative PCR (qPCR). (B–G) Mechanically stripped IEC fractions were analyzed by qPCR for mRV genome copies (B) or flow cytometry for mRV antigen-positive IECs (C–G). Representative flow cytometry plots of naïve (C), Ifnlr1flox/flox (D), and Ifnlr1 IEC-cKO (E) mice infected with mRV with quantification in (F). (G) The median fluorescence intensity (MFI) of mRV antigen in infected cells relative to uninfected cells. (H) A linear correlation plot of mRV+ cells and mRV MFI with 95% confidence intervals (dashed lines). (I–J) Wild-type (WT) mice were infected with 5000 SD50 of mRV and the ilea were processed into Swiss rolls and stained by in situ hybridization for the interferon-stimulated gene (ISG), Ifit1, or mRV. The percentage of mRV-infected cells that were Ifit1-positive and Ifit1-negative were determined. Additional representative images for (I–J) are depicted in Figure 9—figure supplement 2. Where designated in (A–G), dashed lines = limit of detection (LOD) as set by naïve mice. Data points represent individual mice and are pooled from two to three independent experiments. Statistical significance was determined by two-way ANOVA with Sidak’s multiple comparisons (A), by Mann Whitney (B, F), unpaired t-test (G), or Kruskal-Wallis with Dunn’s multiple comparisons (J), where * = p < 0.05 and *** = p < 0.001.

Figure 9—source data 1

Values and statistical tests of murine rotavirus (mRV) genomes, percentages of cells infected, median fluorescence intensity (MFI), and % of cells mRV+ cells co-staining with Ifit1 expression.

https://cdn.elifesciences.org/articles/74072/elife-74072-fig9-data1-v2.xlsx

To more stringently assess the capacity of Ifnlr1 to protect IECs against the earliest stages of mRV infection, we inoculated mice with 5000 SD50 (50% shedding dose) of mRV to maximize the likelihood of uniform viral exposure throughout the intestine. At 24 hr post-inoculation, we quantified mRV genomes in the epithelial fraction and the proportion of infected IECs (live, EpCAM-positive, CD45-negative, and mRV-positive) by flow cytometry. At 24 hr post-inoculation, we found that Ifnlr1 IEC-cKO mice had 20-fold more mRV genomes than Ifnlr1flox/flox mice (Figure 9B). In addition, we found that a threefold greater proportion of IECs were infected with mRV in Ifnlr1 IEC-cKO mice than Ifnlr1flox/flox mice at 24 hr post-infection (Figure 9C–F). However, the median fluorescence intensity (MFI) of mRV antigen was equivalent in mRV-infected IECs from Ifnlr1flox/flox and Ifnlr1 IEC-cKO mice (Figure 9G) and the MFI of mRV did not correlate (r2 = 0.0003) with the percentage of infected IECs (Figure 9H). This equivalent mRV antigen burden in infected cells from Ifnlr1flox/flox and Ifnlr1 IEC-cKO mice in combination with the lack of correlation between antigen burden and percentage of infected cells suggests that the protective role of Ifnlr1 is to prevent infection of IECs rather than to limit replication within IECs after they are infected. We assessed an earlier timepoint at 12 hr post-inoculation and observed similar trends toward an increased proportion of IEC infection in Ifnlr1 IEC-cKO relative to Ifnlr1flox/flox littermates, but the extent of infection was 10- to 100-fold lower and near the limit of detection (Figure 9—figure supplement 1). Thus, the 12 and 24 hr timepoints capture the earliest detectable infection of IECs by mRV, and this early infection is significantly reduced by IFN-λ signaling.

To further contextualize the localization of ISGs and mRV-infected cells over time, we performed in situ hybridization for mRV genomes and Ifit1 in the ileum of WT mice at 12, 24, and 96 hr post-inoculation. Quantification of co-staining for Ifit1 in mRV-infected cells indicated that a minority (~30%) were Ifit1+ at 12 or 24 hr post-inoculation, whereas a majority (~63%) of mRV-infected cells were Ifit1+ at 96 hr post-inoculation (Figure 9I–J, Figure 9—figure supplement 2). These co-staining data are consistent with early mRV evasion of IFN responses within infected IECs at the same timepoints that we observed increased infection of Ifnlr1 IEC-cKO mice (Figure 9B–F, Figure 9—figure supplement 1). Furthermore, we found mRV inoculation did not increase the area of Ifit1 expression in the intestine at 12–24 hr post-inoculation. However, at 96 hr post-inoculation, we found a significant increase in epithelial Ifit1 expression in the ileum that coincided with increased viral genomes and antigen (Figure 9—figure supplement 3), suggesting that Ifit1 expression at 12 and 24 hr post-inoculation is primarily due to the preexisting homeostatic response rather than in response to mRV infection. Therefore, we propose that homeostatic ISGs stimulated by Ifnlr1 expression on IECs play an early protective role against viral infection that preempts viral IFN evasion mechanisms.

Discussion

Here, we report that bacterial microbiota induce an enteric IFN-λ response in IECs at homeostasis (Figures 14). Although there are previous reports of basal, non-receptor-dependent ISG expression in immortalized and primary cell lines (Hakim et al., 2018), the homeostatic response that we report here is dependent on the IFN-λ receptor. Furthermore, we find this response is independent of type I IFN signaling (Figure 3), indicating that IFN-λ signaling plays a dominant and active role in the gastrointestinal epithelium. We also find minimal changes in ISG expression within the spleen and MLN after ABX treatment, suggesting that homeostatic ISGs are predominantly expressed in the intestine. These findings differ slightly from prior descriptions of systemic type I IFN responses that are dependent on bacterial microbiota (Abt et al., 2012; Bradley et al., 2019; Ganal et al., 2012; Steed et al., 2017; Stefan et al., 2020; Winkler et al., 2020). However, differences in the specific tissues and cell types analyzed make it difficult to draw direct comparisons across these studies. Therefore, we conclude that homeostatic ISGs are substantially present in IECs, but do not dispute the prior findings that relatively low homeostatic ISG expression induced by type I IFN plays an important role in extra-intestinal tissues and non-epithelial cell types.

A prior study by Mahlakõiv et al. noted the presence of Ifnl2/3 transcripts in CD45+ cells within the stripped intestinal epithelium, but not the lamina propria, at homeostasis. We have confirmed these findings and have extended them to show that this homeostatic expression of Ifnl2/3, but not Ifnb1, is dependent on bacterial microbiota (Figure 5). However, we have been unable to detect IFN-λ transcripts by RNA scope in situ hybridization in mice at homeostasis, consistent with recently published data (Ingle et al., 2021). This suggests that the production of IFN-λ is below the limit of detection by imaging or highly transient in nature. It is also unclear which CD45+ cell type produces homeostatic IFN-λ. Swamy et al. showed that T cell receptor stimulation led intraepithelial lymphocytes to produce IFN-λ (Swamy et al., 2015) and Mahlakõiv et al. suggested that the primary producers of IFN-λ at steady state are intraepithelial lymphocytes due to the abundance of intraepithelial lymphocytes in the epithelial fraction. However, myeloid cells also reside near the intestinal epithelium and can sample luminal contents by various mechanisms (Chieppa et al., 2006; McDole et al., 2012; Niess et al., 2005). Although the cell type that produces homeostatic IFN-λ is unknown, enrichment for Ifnl2/3 in CD45+ cells within the intestinal epithelial fraction suggests that proximity to bacterial stimuli may be a primary determinant in this response. We suggest that these cells may be actively surveying the intestinal epithelium to detect bacterial MAMPs. The specific CD45+ cell types responsible for producing homeostatic IFN-λ will be a topic of interest for future studies.

We initially anticipated that the distribution of homeostatic ISGs would be uniform among IECs. Instead, we found that this IFN-λ response is highly localized within enteric tissues. Homeostatic ISGs are observed in a minority of small intestinal villi and are primarily present in mature epithelium toward the villus tips (Figure 6). Likewise, homeostatic ISGs are present within patches of the mature epithelium in the colon (Figure 6—figure supplement 2). The surprising finding of localized ISGs in the ileum are supported by analysis of independently generated scRNA-seq datasets from mouse and human small intestinal IECs that depict expression of homeostatic ISGs in a minority of cells with predominant expression in mature enterocytes (Figure 7).

The basis for localized ISG expression is unknown; however, it may reflect the distribution of cells capable of sensing bacterial microbiota, distinct microenvironments within the gastrointestinal tract, or qualitative differences in bacterial colonization. Given the results of our data in Figure 8 and Figure 8—figure supplement 1, we suggest that LPS administration, flagellin administration, or fecal transplant can partially restore the expression of homeostatic ISGs in the small intestine. This interpretation supports the concept that localized ISG-positive regions may be uniquely exposed or responsive to a variety of bacterial MAMPs. Indeed, we find that MYD88 is required for WT levels of homeostatic ISG expression in the small intestine (Figure 3). However, MYD88 is dispensable for homeostatic ISG expression in the colon, and TRIF is not required in either small intestine or colon (Figure 3). These data suggest that the bacterial microbiota broadly stimulates homeostatic IFN-λ through multiple, redundant PRRs. Furthermore, the presence of localized Ifit1 expression in ABX-treated mice following LPS administration (Figure 8) suggests that localization is an intrinsic property of homeostatic ISG stimulation and is not likely to be due to qualitative differences in bacterial colonization.

ISG localization may be indicative of regional differences in access of luminal bacterial MAMPs to IFN-λ-producing cells. One host mechanism to limit bacterial interactions with the intestinal epithelium is the presence of mucus layers (Atuma et al., 2001; Johansson et al., 2011). Intriguingly, the single mucus layer in the small intestine is much less adherent than the mucus layers present in the colon (Johansson et al., 2011), which might allow occasional direct bacterial interactions with IECs or other cells near the intestinal epithelium. However, soluble components from enteric bacteria may also readily diffuse through mucus layers. In this case, there may be sporadic defects in tight junction proteins that are required to maintain the intestinal epithelium. Tight junction remodeling is essential to maintain intestinal integrity during apoptosis and extrusion of IECs that are regularly shed from the intestinal epithelium (Williams et al., 2015). Future studies will be necessary to determine whether defects in epithelial barrier integrity during extrusion events are linked to the local ISG responses that we observe and further delineation of the factors that render specific regions ‘responsive’ will be of great interest for follow-up studies.

The preceding findings (Figure 4) suggested that the homeostatic ISG response in IECs would provide protection against IEC-tropic viruses, such as mRV. Although bacterial-associated, IFN-independent mechanisms of mRV clearance have been reported (Shi et al., 2019; Zhang et al., 2014), we found that the signature of homeostatic ISGs in ileum tissue included well-characterized antiviral ISGs (Figures 12). To investigate whether these homeostatic ISGs protect against MRV, we used Ifnlr1 IEC-cKO mice that lack homeostatic ISGs (Figures 4 and 6) rather than using ABX treatment, which introduces pleiotropic effects on rotavirus infection (Uchiyama et al., 2014) and dramatically increases transit time through the intestine (Baldridge et al., 2015). Using Ifnlr1 IEC-cKO mice, we found increase in IEC infection by mRV at early stages of infection compared to Ifnlr1flox/flox littermates (Figure 9 and Figure 9—figure supplement 1). However, the protection offered by IEC expression of Ifnlr1 was lost by the middle and late stages of infection, consistent with the ability of mRV to antagonize induction of IFN responses once infection is established (Arnold et al., 2013). Our observations during initiation of infection may provide important context to observations in other studies that report differing capacity for infection-induced IFN-λ to protect against mRV infection (Lin et al., 2016; Pott et al., 2011). Ultimately, it is clear that prophylactic administration of exogenous IFN-λ protects against mRV infection (Lin et al., 2016; Pott et al., 2011; Van Winkle et al., 2020), providing precedent that homeostatic IFN-λ would also be protective when induced by bacterial microbiota prior to infection.

Although we found that homeostatic ISGs provide protection during initiation of mRV infection, it remains unclear how these localized ISG pockets impart this protection. Given our findings in Figure 8 and Figure 8—figure supplement 1, we suggest that these localized ISGs may be indicative of locations that are particularly vulnerable to viral infection if ISGs were not present at the time of viral exposure. However, alternative explanations for the protective effects that we observe are also plausible, such as: (i) an inability to detect the full magnitude of ISG expression by imaging, or (ii) an unknown temporal component to homeostatic ISG signaling that may be coincident with durable ISG protein expression. Given the magnitude of signal amplification in RNAscope in situ hybridization and the similarity of Ifit1 expression in our imaging data to Ifit1 expression in public single-cell sequencing datasets, we find it unlikely that there is more widespread homeostatic ISG expression below the limit of our detection by imaging. However, we do think an uncharacterized temporal component of homeostatic ISG signaling is plausible, wherein individual villi may be rapidly and transiently expressing homeostatic ISGs in response to sensing of bacterial microbiota. This temporal model may also include a more durable ISG protein response that is not fully concordant with expression of ISG transcripts. In sum, these findings indicate that preexisting, homeostatic ISGs present in Ifnlr1-sufficient mice are protective during initiation of mRV infection, but that endogenous IFN-λ does not reduce mRV burden in infected cells. These data highlight the possibility that detection of bacterial microbiota in particularly exposed areas may preemptively activate homeostatic ISGs as a form of anticipatory immunity to protect the intestinal epithelium from enteric viruses.

Materials and methods

Mice

All mice were bred using the C57BL/6 background and used within the age of 8–12 weeks; C57BL/6J mice (stock #000664) were purchased from Jackson Laboratories (Bar Harbor, ME) and used as WT. Genetically modified mice included Ifnlr1-/- and Ifnlr1flox/flox (generated from Ifnlr1tm1a(EUCOMM)Wtsi as published; Baldridge et al., 2017), Ifnar1-/- (B6.129.Ifnar1tm1), Trif-/- (JAX C57BL/6J-Ticam1Lps2/J, stock #005037), Myd88-/- (JAX B6.129P2(SJL)-Myd88tm1.1Defr/J, stock #009088), Irf3-/- (B6.129S/SvEv-Bcl2l12/Irf3tm1Ttg), Irf7-/ (B6.129P2-Irf7tm1Ttg/TtgRbrc), Stat1-/- (B6.129.-Stat1tm1Dlv), and Villin-cre (B6.Cg-Tg(Vil1-cre)997Gum/J) mice. Ifnlr1-/- and Ifnar1-/- mice were bred to Ifnlr1-/+/Ifnar1-/+ mice to generate littermate Ifnlr1-/+/Ifnar1-/+, Ifnlr1-/+/Ifnar1-/-, Ifnlr1-/-/Ifnar1-/+, Ifnlr1-/-/Ifnar1-/- offspring.

All mice were maintained in specific pathogen-free facilities at Oregon Health & Science University (OHSU) and Washington University in St Louis (WUSTL). Animal protocols were approved by the Institutional Animal Care and Use Committee at OHSU (protocol #IP00000228) and WUSTL (protocol #20190162) in accordance with standards provided in the Animal Welfare Act.

For all experiments, mice were allocated into experimental groups based on genotype with equal representation of individual litters and equal sex ratios.

Mouse treatments

Request a detailed protocol

Mice were administered an ad libitum antibiotic cocktail consisting of: 1 g/L ampicillin, 1 g/L metronidazole, 1 g/L neomycin, and 0.5 g/L vancomycin (Sigma, St Louis, MO) in autoclaved H2O (OHSU) or in 20 mg/mL grape Kool-Aid (Kraft Foods, Northfield, IL) (WUSTL). Sterile H2O (OHSU) or Kool-Aid (WUSTL) alone was used as a control. Mice were maintained on ABX or control for 2 weeks prior to harvest.

Recombinant IFN-λ was provided by Bristol-Myers Squibb (New York City, NY) as a monomeric conjugate comprised of 20 kDa linear PEG attached to the amino-terminus of murine IFN-λ. Mice were injected intraperitoneally with IFN-λ or an equal volume of PBS vehicle as indicated in figure legends at the indicated time prior to analysis.

Mice were stimulated with of 100 µg of the synthetic dsRNA analogue, poly I:C (R&D, #4287) or PBS by intraperitoneal injection in a 200 µL volume, 2 hr prior to harvest.

Twenty-five µg of the bacterial product LPS (Sigma #L4391), flagellin (Invivogen #tlrl-bsfla; from Bacillus subtilis) or CpG (Invivogen #tlrl-1585; Class A CpG oligonucleotide), were perorally administered to mice in 25 µL of sterile PBS. Mice were treated on days 15 and 16 of antibiotic treatment or H2O control prior to harvest on day 17.

For transplantation of fecal material, antibiotic treatment was stopped and mice were fed 25 µL of fecal mixture by pipet for 2 consecutive days. Fecal mixture was prepared by collecting fecal samples from control mice; a single stool pellet was resuspended in 200 µL of sterile PBS, stool was broken apart by pipetting, and large particulate was allowed to settle for several minutes prior to administration.

Cell isolation

Request a detailed protocol

Epithelial fractions were prepared by non-enzymatic dissociation as previously described (Nice et al., 2016). Briefly, mouse ileum was opened longitudinally and agitated by shaking in stripping buffer (10% bovine calf serum, 15 mM HEPES, 5 mM EDTA, and 5 mM dithiothreitol in PBS) for 20 min at 37°C. Lamina propria fractions were prepared by enzymatic digestion and dissociation with the Lamina Propria Dissociation Kit and GentleMacs Dissociator (Miltenyi Biotec, Bergisch Gladbach, Germany). Dissociated cells were collected for use in qPCR analysis, flow cytometry, and magnet enrichment.

Rotavirus infection of mice

Request a detailed protocol

Mouse rotavirus (EC strain) was graciously provided by Andrew Gewirtz (Georgia State University). Viral stocks were generated by inoculating 4- to 6-day-old neonatal BALB/c mice and harvesting the entire gastrointestinal tract upon presentation of diarrheal symptoms 4–7 days later. Intestines were freeze-and-thawed, suspended in PBS, and homogenized in a bead beater using 1.0 mm zirconia-silica beads (BioSpec Products). These homogenates were clarified of debris, aliquoted, and stored at –70°C. The 50% shedding dose (SD50) was determined by inoculation of 10-fold serial dilutions in adult C57BL/6J mice. For stool timecourse studies, mice were inoculated by peroral route with 100 SD50 and a single stool pellet was collected daily for viral quantitation by qPCR. For protection studies, mice were inoculated by intragastric gavage with 5000 SD50, and ileum was isolated and mechanically stripped 12- and 24 hr later for quantitation of viral burden by qPCR and flow cytometry.

RNA isolation, rDNA isolation, qPCR, and analysis

Request a detailed protocol

RNA from tissue and stripped IECs was isolated with TRIzol (Life Technologies, Carlsbad, CA) according to the manufacturer’s protocol. RNA from magnet-enriched cells was purified by Zymo Quick-RNA Viral Kit (Zymo Research, Irvine, CA). The larger of either 1 μg of RNA or 5 μL of RNA were used as a template for cDNA synthesis by the ImProm-II reverse transcriptase system (Promega, Madison, WI) after DNA contamination was removed with the DNAfree kit (Life Technologies). 16S bacterial rDNA was isolated from stool and intestinal contents with a ZymoBIOMICS DNA kit (Zymo Research, Irvine, CA) kit. Quantitative PCR was performed using PerfeCTa qPCR FastMix II (QuantaBio, Beverly, MA) and the absolute quantities of transcript were determined using standard curves composed of gBlocks (IDT) containing target sequences. Absolute copy numbers from tissue samples were normalized to the housekeeping gene, ribosomal protein S29 (Rps29). Taqman assays for selected genes were ordered from IDT (Coralville, IA): Rps29 (Mm.PT.58.21577577), Ifit1 (Mm.PT.58.32674307), Oas1a (Mm.PT.58.30459792), Mx2 (Mm.PT.58.11386814), Stat1 (Mm.PT.58.23792152), Isg15 (Mm.PT.58.41476392.g). Taqman assays for Ifnl2/3 and Ifnb1 were designed previously (Van Winkle et al., 2020) and consisted of the following primer-probe sequences: Ifnl2/3 (Primer 1 – GTTCTCCCAGACCTTCAGG, Primer 2 – CCTGGGACCTGAAGCAG, Probe – CCTTGCAGGCTGAGGTGGC); Ifnb1 (Primer 1 – CTCCAGCTCCAAGAAAGGAC, Primer 2 – GCCCTGTAGGTGAGGTTGAT, Probe – CAGGAGCTCCTGGAGCAGCTGA). Murine rotavirus was detected using Taqman primer-probe sets specific for 422–521 of GeneBank sequence DQ391187 as previously described (Fenaux et al., 2006) with the following sequences: Primer 1 – GTTCGTTGTGCCTCATTCG, Primer 2 – TCGGAACGTACTTCTGGAC, Probe – AGGAATGCTTCAGCGCTG; and universal bacterial 16S rDNA was detected using Taqman primer-probe sets with previously designed sequences (Nadkarni et al., 2002): Primer 1 – GGACTACCAGGGTATCTAATCCTGTT, Primer 2 – TCCTACGGGAGGCAGCAGT, Probe – CGTATTACCGCGGCTGCTGGCAC.

RNAscope

Request a detailed protocol

Swiss rolls of intestinal tissue were fixed in 10% neutral-buffered formalin for 18–24 hr and paraffin-embedded. Tissue sections (5 μm) were cut and maintained at room temperature with desiccant until staining. RNA in situ hybridization was performed using the RNAscope Multiplex Fluorescent v2 kit (Advanced Cell Diagnostics, Newark, CA) per protocol guidelines. Staining with anti-sense probes for detection of Ifit1 (ACD, #500071-C2), Usp18 (ACD, #524651-C1), and MRV (ACD, #1030611-C1) was performed using ACDBio protocols and reagents. MRV probes were designed to target 2–1683 of DQ391187 against NSP3, VP7, and NP4. Slides were stained with DAPI and mounted with ProLong Gold antifade reagent (ThermoFisher), and imaged using a Zeiss ApoTome2 on an Axio Imager, with a Zeiss AxioCam 506 (Zeiss).

Collected images were batch processed in Zeiss Zen 3.1 using unstained control slides to set background values and quantified using ImageJ. Area of Ifit1 was determined by positive Ifit1 fluorescent area relative to the total fluorescent area of the tissue section. Ifit1+ mRV-infected cells were defined as mRV+ particles with greater than 5 µm area with a maximum Ifit1 intensity greater than 10% above background. These Ifit1+ cells were then divided by the total number mRV+ cells to determine the percentage of Ifit1+ mRV-infected cells.

Flow cytometry and magnet enrichment

Request a detailed protocol

Dissociated cells were collected and stained for flow cytometry. Cells were stained with Zombie Aqua viability dye (BioLegend), Fc receptor-blocking antibody (CD16/CD32; BioLegend), anti-EpCAM (clone G8.8; BioLegend), and anti-CD45 (clone 30-F11; BioLegend). For analysis of murine rotavirus infection, cells were stained with anti-rotavirus (polyclonal; ThermoFisher, #PA1-7241) followed by goat anti-rabbit secondary (ThermoFisher). All data were analyzed using FlowJo software (BD Biosciences). Gates were set based on unstained and single-fluorophore stains. IECs were selected by gating on live, EpCAM-positive, CD45-negative cells. Gates for murine rotavirus infection were set based on naïve samples.

Where indicated, dissociated cells were enriched using MojoSort Mouse anti-APC Nanobeads (BioLegend, #480072) after flow cytometry staining for anti-EpCAM and anti-CD45 with APC fluorophores by following manufacturer protocols.

RNA sequencing and expression analysis

Request a detailed protocol

WT C57BL/6J or Ifnlr1-/- mice were administered ad libitum Kool-aid or ABX for 2 weeks, or WT mice were administered 25 µg recombinant IFN-λ for 1 day, then ileal segments lacking Peyer’s patches were harvested and RNA-seq was performed as prior (Park et al., 2016). mRNA from ilea was purified with oligo-dT beads (Invitrogen, Carlsbad, CA) and cDNA was synthesized using a custom oligo-dT primer containing a barcode and adaptor-linker sequence, degradation of RNA-DNA hybrid following single-strand synthesis, and ligation of a second sequencing linker with T4 ligase (New England Biolabs, Ipswich, MA). These reactions were cleaned up by solid phase reversible immobilization (SPRI), followed by enrichment by PCR and further SPRI to yield strand-specific RNA-seq libraries. Libraries were sequenced with an Illumina HiSeq 2500 with three to four mice were included in each group. Samples were demultiplexed with second mate, reads were aligned with STAR aligner and then counted with HT-Seq. DEGs were identified using DESeq2 (Love et al., 2014) based on cutoffs of twofold change, and an inclusive p-value < 0.5. Standard GSEA was performed to identify enrichments in IFN-λ response genes. RNA-seq data were uploaded to the European Nucleotide Archive (accession #PRJEB43446).

16S rRNA gene illumina sequencing and analysis

Request a detailed protocol

For sequencing of the 16S rRNA gene, primer selection and PCRs were performed as described previously (Caporaso et al., 2011). Briefly, each sample was amplified in triplicate with Golay-barcoded primers specific for the V4 region (F515/R806), combined, and confirmed by gel electrophoresis. PCRs contained 18.8 μL RNase/DNase-free water, 2.5 μL of 10× High Fidelity PCR Buffer (Invitrogen, 11304–102), 0.5 μL of 10 mM dNTPs, 1 μL 50 of mM MgSO4, 0.5 μL each of forward and reverse primers (10 μM final concentration), 0.1 μL of Platinum High Fidelity Taq (Invitrogen, 11304–102), and 1.0 μL genomic DNA. Reactions were held at 94°C for 2 min to denature the DNA, with amplification proceeding for 26 cycles at 94°C for 15 s, 50°C for 30 s, and 68°C for 30 s; a final extension of 2 min at 68°C was added to ensure complete amplification. Amplicons were pooled and purified with 0.6× Agencourt AMPure XP beads (Beckman-Coulter, A63882) according to the manufacturer’s instructions. The final pooled samples, along with aliquots of the three sequencing primers, were sent to the DNA Sequencing Innovation Lab (Washington University School of Medicine) for sequencing by the 2 × 250 bp protocol with the Illumina MiSeq platform.

Read quality control and the resolution of amplicon sequence variants were performed in R with DADA2 (Callahan et al., 2016). Non-bacterial amplicon sequence variants were filtered out. The remaining reads were assigned taxonomy using the Ribosomal Database Project (RDP trainset 16/release 11.5) 16S rRNA gene sequence database (Cole et al., 2014). Ecological analyses, such as alpha-diversity (richness, Shannon diversity) and beta-diversity analyses (UniFrac distances), were performed using PhyloSeq and additional R packages (McMurdie and Holmes, 2013). 16S sequencing data have been uploaded to the European Nucleotide Archive (accession #PRJEB43446).

scRNA-seq analyses

Request a detailed protocol

A mouse scRNA-seq dataset generated by Haber et al. was accessed from NCBI’s Gene Expression Omnibus (GEO), accession #GSE92332. A human pediatric scRNA-seq dataset generated by Elmentaite et al. was accessed in processed form from The Gut Cell Atlas, with raw data also available on EMBL-EBI Array Express, accession #E-MTAB-8901. Files were analyzed in R using Seurat (v. 3.2.2) (Stuart et al., 2019). For the Haber et al. dataset, UMI counts were normalized to counts per million, log2 transformed, and homeostatic ISGs were selected for analysis. Data was collated by previously annotated cell type and proportion of cells expressing each individual ISG was calculated. For Elmentaite et al., data was restricted to healthy controls and then collated by previously annotated cell type to determine the proportion of cells expressing individual ISGs.

Statistical analyses

Request a detailed protocol

Sample size estimation was performed based on historical data. Data were analyzed with Prism software (GraphPad Prism software), with specified tests as noted in the figure legends.

Data availability

RNA-seq data were uploaded to the European Nucleotide Archive under accession #PRJEB43446. Source data files are provided for each figure and contain the numerical data used to generate the figures.

The following data sets were generated
    1. Baldridge et al
    (2021) EBI
    ID PRJEB43446. A homeostatic interferon lambda response to the bacterial microbiome stimulates preemptive antiviral defense within discrete pockets of intestinal epithelium.
The following previously published data sets were used
    1. Haber et al
    (2017) NCBI Gene Expression Omnibus
    ID GSE92332. A single-cell survey of the small intestinal epithelium.
    1. Elmentaite et al
    (2020) EBI
    ID PRJEB37689. Single-cell sequencing of developing human gut reveals transcriptional links to childhood Crohn's disease.

References

Decision letter

  1. Andrew J MacPherson
    Reviewing Editor; University of Bern, Switzerland
  2. Carla V Rothlin
    Senior Editor; Yale School of Medicine, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting the paper "A homeostatic interferon-λ response to bacterial microbiota stimulates preemptive antiviral defense within discrete pockets of intestinal epithelium" for consideration by eLife. Your article has been reviewed by 4 peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by a Senior Editor. The reviewers have opted to remain anonymous.

We are sorry to say that, after consultation with the reviewers, we have decided that this work cannot be considered in its current form for publication by eLife. The reviewers engaged in a careful consultation round before this decision was reached. Although there was substantial interest in the findings, there were concerns over the level of novelty, the underlying mechanisms of the localized interferon response and how these may be influenced by microbiota variation. The eLife policy is not to invite a revision if significant further experimental work is considered necessary. Although the reviewers concluded that the work necessary was outside the boundaries of a revision request, we would be interested to receive a further submission as a new paper if you are able to address the concerns.

Reviewer #1 :

In this paper the authors provide new information about the relative importance of the type I and type III interferon-driven gene expression and anti-viral responses, particularly focused on the role of the Intestinal microbiota to maintain background levels of type III (interferon λ) signaling. They show with interferon λ administration as a positive control, and antibiotic-mediated microbiota biomass depletion that low background levels of type III interferon-driven gene expression are mediated by the microbiota. Heterozygous mouse strain combinations for epithelial specific either type I or type III interferon receptor deficiency shows that the effect is type III mediated. In-situ hypbridisation shows that type III-driven gene expression is highly discontinuous in the epithelial layer and mainly at the villous tips. Rotavirus infection shows slightly accelerated kinetics in the absence of the type III receptor-signaling. Since intestinal type I and type III interferon responses are well described to occur, this provides a distinction between the two signaling pathways the consequent antiviral responses and the role of the microbiota in maintaining a basal level of type III signaling.

This is a highly mature paper with a consistent set of experiments supporting the role of type III interferon signaling as a consequence of microbiota colonization in the intestine. The strengths of the paper derives from experimental contrasting use of epithelial-specific IFNLR deficiency and IFNAR deficiency, control treatments of interferon λ in verifying gene sets, localization of the homeostatic and IFNL-stimulated response by in situ hybridisation and the use of the functional rotavirus response in the strain combination context.

I consider this is almost publication ready.

Questions and suggestions that I have regarding the data and interpretation are as follows.

1. Do the authors consider that the relative importance of type I and type III signaling depends on the age of the mice.

2. What happens to the localization of the type III signal as control stimulation or viral infection proceeds? Presumably the time trajectory of in situs in these different contexts has not been done.

3. Do microbiota composition, bacteriophage lytic phases, or diet influence the homeostatic response.

4. I think that the authors should consider citing the basal signaling in Caco2 cells and organoids (Hakim Sci. Rep. 2018 8:8341) and the interferon-independent (Shi 2019, Cell 179, 644-658) papers in the discussion.

Reviewer #2:

In this work, Van Winkle et al. examine contributions of interferons and microbiota to innate responses in the intestine, and which cell types are involved. Previous work by this team and others demonstrated that microbiota influence interferon responses in the intestine, which can affect infection with several enteric viruses, either increasing infection or decreasing infection depending on the viral system. Here, the team profiles expression of ISGs, examines which cells produce IFNs and express ISGs, whether these responses are microbiota dependent, and examines the effect on infection with murine rotavirus.

Strengths:

The most interesting aspect of this study is the observation that ISG expression in the intestine is extremely patchy and limited to a few mature enterocytes. Figure 6 is stunning, and these data are supported by beautiful controls (robust/broad ISG expression everywhere in the intestine of mice treated IP with IFNlambda in Figure S4, but loss of expression in mice treated with ABX, and confirmation in public single cell RNA-seq data sets). This is the most unique and significant contribution of the study.

Other aspects of the study are also well done with appropriate controls (multiple panels with ISG levels throughout, etc.). The team uses a variety of mouse strains, treatments, etc. to support their claims.

Weaknesses:

The primary weakness lies in significance, partially based on past work. The observation about patchy ISG expression in mature enterocytes is very cool, but it remains unknown why/how this happens and whether there are any functional consequences. This reviewer understands that this is a very tough problem and may take time to figure out. Additionally, much work has already been done with microbiota and IFNlambda effects on enteric virus infection, making many of the findings here overlapping or redundant with prior work: IFNlambda effects on IECs and viral infection (Sommereyns Plos Path 2008, Baldridge JVI 2017), leukocytes as the source of IFNlambda (Mahlakoiv Plos Path 2015), and microbiota-mediated innate immune modulation and effects of IFNlambda (e.g., Baldridge Science 2015, Nice Science 2015).

For Figure 5 and S3 data, the authors examine gene expression in intestinal cells collected from mice and then enriched for epithelial vs. lamina propria cell populations. They see significantly reduced expression of IFN upon ABX treatment in just one case-harvested epithelial cells that have been enriched for CD45+ cells (leukocytes) have reduced IFNlambda expression. Lamina propria CD45+ cells do not show significantly reduced IFNlambda. This is surprising. While there are low levels of CD45+ cells in epithelial cell preps (as they show in Figure S3), there are higher levels of CD45+ cells in the LP preps. It seems odd that the only significant difference lies with these immune cells in the epithelial cell prep, and this effect is not observed in LP preps, which should contain many more CD45+ cells. Given the authors' conclusion that leukocytes are the source of IFNlambda, this deserves more follow-up or at least some discussion. Is there something special about CD45+ cells from epithelial preps? Is it possible these are contaminating cells? Or is it simply an issue with spread in the data?

In Figures 8 and S6, the authors examine Ifit1 expression in WT or mutant mice with different treatments (ABX, LPS, fecal transfer), to determine if LPS or fecal transfer can restore gene expression. Due to low expression levels and a high degree of spread in the data, they resort to binning the primary gene expression data (8H and S6A; ABX vs. ABX/LPS not significantly different) into + vs. – categories (Figure 8J, S6E; ABX vs. ABX/LPS now are significantly different) to allow different statistical analysis. From this, they conclude that ISG expression is partially restored by LPS treatment in ABX-treated mice. Since the data are just around the detection limit and it's a high bar to restore gene expression, it's perhaps understandable that the data are binary, but this isn't the most convincing data set.

Reviewer #3:

The authors convincingly show by RNAseq, that microbiota ablation by antibiotics treatment deprives WT mice of a tonic and interferon λ induced interferon stimulated gene (ISG) response. Alongside they test interferon λ receptor ko mice, which do not show a tonic induction of ISGs or a response to antibiotics dependent removal of microbiota. Induction of ISGs in intestinal tissue depends on TLR signaling, which would support a microbiota dependent induction. Intriguingly this ISG response is highly localised in very focussed areas of the intestine. Fecal transfer or LPS treatment restore the antibiotics induced phenotype of basal ISG expression loss and partially restore antiviral protection in intestinal epithelial cells.

These data complement recent publications [PMID: 32380006][PMID: 31269444] of tonic type I interferon signaling induced by microbiota could explain a series of publications showing the importance of microbiota for antiviral defence in the gut.

The authors build a line of arguments based on the correlation of data from IFNLR -/- mice and ABX treated mice. They omit however an important alternative explanation, which might be qualitative differences in the microbiota composition between IFNLR -/- and WT mice.

An explanation for the focussed ISG response in the intestine and how this explains the reduced resistance to enteric viruses is not provided.

An alternative explanation for the observed ISG expression phenotype could be a difference in baseline microbiota composition between WT and IFNLR -/- mice. This should be checked by 16S rRNA gene sequencing of small and large intestinal samples and could be complemented by a fecal transfer from IFNLR-/- mice into ABX treated WT mice followed by measurement of baseline ISGs in the intestine.

Qualitative differences in colonisation could also explain the binary response of WT mice to fecal transfer.

The focussed ISG response shown in Figure 8 is intriguing but difficult to reconcile with the 20x reduced viral titres in WT mice. Does the virus only replicate in these sites? And are commensal microbiota specifically dense in these areas. A multicolour FISH approach could answer both of these points.

Reviewer #4:

In this paper from Van Winkle et al., the authors determined that murine intestinal epithelial cells produce several homeostatic ISGs. These homeostatic ISGs are located in the mature enterocytes and are stimulated by the presence of bacterial microbiota. This is an exciting concept as it becomes clearer that all cells in a tissue are not the same. The authors nicely demonstrate that these homeostatic ISGs are only present when the commensal microbiota is also present and that they depend on IFNl signaling.

A main weakness of the paper is that the authors use IFNl treatment to stimulate ISGs and then use this as a basis to determine homeostatic gene levels. However, it has been shown that stem cells have basal ISGs that are IFN independent (Wu…Rice, Cell 2018), which means that the only cells that could respond to IFN would be the mature cells in the villi. Additionally, as the bacteria would normally only be in contact with the enterocytes and not the crypts then one would not expect to find bacterial induced homeostatic ISGs lower in the crypt-villi axis. Given these, the authors findings while interesting are expected due to the experimental set-up. While several experiments contain complete controls to make all interpretations, several experiments lack controls and many conclusions suffer from over statements, lack of convincing imaging and correlations between infections and ISGs.

– In Figure 1 the authors claim that there is no significant difference between ifnlr-/- +/- ABX. However, this is hard to say from the presented data. There are no clear differences in the Enrichment plots D-F and the values given in B seem to be wrong as it seems very unlikely that both the FDR and the Nominal p-value are the exact same number. Without the correct numbers it is hard to judge the conclusion of this figure.

– On Figure 1, the authors claim that WT mice separate from ifnlr-/- mice, however in Figure 1H, one of the WT mice and ifnlr-/- mice cluster together. The authors should add more mice to show this more clearly or tone down this statement.

– On Figure S2, the IFNL without ABX control is missing to be able to interpret the data set.

– In Figure 3, why are the het samples shown as stripped epithelium and not full tissue as the others in the experiment. It would be good to have all tissue samples or all stripped epithelium to be able to make a full comparison between the data sets.

– Additionally, in Figure 3 the ifr7-/- effect on ISGs levels is very modest and is further decreased with ABX (which is normally considered a criteria for it to not be key for this response) the authors should provide more evidence that irf7 is involved or tone down their conclusion in this regard.

– The impact of ABX and ifnlr-/- seems to be less apparent in the colon than the ileum (e.g. Figure 2A vs B). This is curious since the colon will have the higher bacterial load. Can the authors comment on this?

– Figure 5B, it is curious that there is no induction of ifnb1 in the epithelial fraction upon poly IC treatment. While several studies have shown that the cells are not responsive to type I IFNs, there is no data showing that they cannot produce type I IFNs upon MAMP stimulation. Can the authors explain this?

– Figure S3, why are the values of CD45+ cells not closer to 100% post-enrichment? If there are still 50% of non-CD45+ cells that are being analyzed, what are they and can they impact the results shown in Figure 5?

– Figure S4, high magnification images are needed to see if the crypts are able to respond to IFNs in a murine model to clearly demonstrate whether it would be possible for them to have IFN dependent homeostatic ISGs or whether the authors are already selecting for a phenotype only present in mature enterocytes.

– Figure S5, the colon data is not very convincing as the ifnlrKO shows some background staining which does not look much different from the other images. Additionally, it is not clear why only floxed and dIEC mice are used in this figure and there are no WT mice used as in Figure 6. It would be good to have the matching animals to better compare the data. It would also be important to include high mag images to clearly see if the ISG distribution is the same (as it does not seem to be the same tip distribution that is seen in the ileum). This figure should also include a quantification as provided in other figures.

– Figure 7, crypt cells are not a cell type. There are stem cells, TA and other progenitors present that make up the crypt. This should be renamed to correctly indicate the type of cells in this population.

– Additionally, the scRNA-Seq used for human analysis was a bit strange. This paper used mainly embryos and children to derive the cell types. The fact that only a few cell types were highly present seems to indicate that maybe this was not the best paper. It would be good to compare to stronger papers such as Wang et al. JEM 2020, Fuji et al. Cell Stem Cell 2018. There is also a paper using human ileum showing that stem cells and enterocytes each have their own basal ISGs which are stem cell specific or enterocytes specific (Triana, 2020). The statements about the human data should be toned down or a more robust scRNA data set should be used.

– Figure S6D, why are there different numbers of mice used for each ISG? If the RNA is harvested, then the same RNA can be used to assess all three ISGs. It seems strange that of 30 mice tested for ifit1, only 17 were tested for stat1. These missing numbers seems strange and should be included.

– The homeostatic ISGs are only found in the tip of the villi, and rotavirus infection normally occurs at the tips of the villi, the authors claim that these homeostatic ISGs are more protective against infection. To make this claim the authors need to perform a co-staining of rotavirus infected tissue to show that the villi that are infected are not the villi which express the homeostatic ISGs. In general, rotavirus staining is needed as the FACS data suggests that a very low numbers of cells are infected which makes it hard to justify the authors claims.

– The authors claim that IFN is being produced from CD45+ epithelium derived lymphocytes. To support and strengthen this claim, it would be important to perform RNA scope with markers for these cells to determine if the sparse ISG induction seen in the tissue is correlates with the localization of these cells.

– Since the authors claim is that the homeostatic ISGs come from sensing of bacterial components, and that LPS can rescue this phenotype, then the authors should determine if tlr4-/- mice also lack the presence of homeostatic ISGs.

– Furthermore, the authors claim in the discussion that microbiota components are responsible for the homeostatic levels and that they signal through several MAMPs. However, this is not shown in the paper. In the manuscript the authors only tested LPS. They also used CpG which will signal through MyD88 but does not induce this response. To make the claim that multiple MAMPs can be used, then several other MAMP agonists should be used to show that this is not only a LPS phenotype (e.g. flagellin).

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for sending your article entitled "A homeostatic interferon-λ response to bacterial microbiota stimulates preemptive antiviral defense within discrete pockets of intestinal epithelium" for peer review at eLife. Your article is being evaluated by 3 peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation is being overseen by Carla Rothlin as the Senior Editor.

Reviewer #1:

The authors have completed a careful revision of the manuscript in accordance with the original critiques.

Apart from the duplication of the Figure 3 as Figure 2 which is presumably a submission error, I still have some concerns.

The antibiotic treatments in combination with the strain combinations leave overlapping ranges of interferon-stimulated genes (ISGs) that gives some uncertainty of interpretation (for example in the Ifnlr1-deficient for Stat1 transcripts which is non-significant with respect to wild-type for the left panel of 4A). Presumably the noise in the system also makes the Irf7-deficient result challenging to interpret in the left panel of 3B.

The patchy expression of the ISGs in the histological panels is clear, but I am worried about the security of the interpretations of LPS or flagellin recovery of the effects through stratification of those animals where the effect is or is not seen when there may be as few as 4 signals in the positive control (Figure 8A).

Finally, the rotavirus 'preemptive' protection is at best partial and likely context (for example microbiota composition) dependent, so it would be a surprise to me if all investigators trying to repeat this obtained exactly the same result. I think that the authors would have to be more cautious in their conclusions and in their abstract to make the paper publishable.

Reviewer #2:

The authors convincingly excluded that qualitative differences in the microbiota between WT and IFNLR ko mice are responsible for the observed differences in baseline ISG expression.

The data in Figure 9I are however not convincingly showing that RV preferentially replicates in IFIT1 negative cells. The statement "we generally found that mRV RNA was not present in villi that have IFIT1 expression" in line 603 is not supported by the provided images. If there is a biological effect, the authors would have to substantiate it by providing a quantification of relative number of RV positive cells in presence or absence of IFIT1 co-staining.

This is problematic since the authors central claim is the increased defence against viral pathogens as a direct consequence of microbiota induced IFNL The current dat a would suggest a rather indirect mechanism.

Reviewer #3:

The revisions to the manuscript have improved the clarity and strengthened several concerns. I believe the manuscript is close to be ready for publication however there are still a few questions that could use a bit more quantifications to improve their strength.

Figure 8 and S9. While I appreciate the difficulty in these experiments by adding back the fecal transplant or adding the TLR agonists. I am still concerned about the strength of the data and claiming that they can partially restore the phenotype.

Figure 9- Thanks for including the co-stainings, this helps to see how virus infection is relative to the IFIT1 stainings, however these images are lacking quantification. Additionally, it would be important to plot the % of co-localization of IFIT and rotavirus as it seems that in several examples there is more overlap than suggested in the text.

Figure 10 – The 96h time was performed only 1 time. It would be important to have this repeated more than one time and to have the colocalization of IFIT and rotavirus for this time point as well. As Figure 9 indicates that there is a lot of virus shedding still taking place at this time, it would be important to clarify how this correlates with the ISG expression.

https://doi.org/10.7554/eLife.74072.sa1

Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

Reviewer #1:

[…]

Questions and suggestions that I have regarding the data and interpretation are as follows.

1. Do the authors consider that the relative importance of type I and type III signaling depends on the age of the mice.

This is an important point, and we are aware of studies that indicate age of mice can influence IEC responsiveness to Type I IFN. To emphasize these findings, we have updated our introduction to highlight studies that explored this topic on (Lines 54-56): “The responsiveness of IECs to type I IFN appears to be developmentally regulated because the IECs of adult mice exhibit weaker type I responses than IECs of neonatal mice (Lin et al., 2016).”

2. What happens to the localization of the type III signal as control stimulation or viral infection proceeds? Presumably the time trajectory of in situs in these different contexts has not been done.

We find the temporal aspect of this localized homeostatic type III signaling to be incredibly interesting. We are unable to perform time trajectories of individual mice by in situ hybridization due to the terminal nature of this approach. However, we have now included ileal Ifit1 in situ hybridization imaging of representative mice at 12hr, 24hr, and 96hr postinoculation with murine rotavirus. These data are presented in Figure S10, are described in Lines 603-605: “Additionally, we found mRV infection did not dramatically increase area of Ifit1 expression in the intestine at 24 hours post-inoculation. However, at 96 hours post-inoculation, we found a significant increase in epithelial Ifit1 expression in the ileum relative to early times post-inoculation.”

3. Do microbiota composition, bacteriophage lytic phases, or diet influence the homeostatic response.

To address this important point, we now provide evidence that microbiota composition does not influence the homeostatic ISG response detailed in this manuscript. We find no difference in the number of observed bacterial species, Shannon Diversity Index, or UniFrac distance when we performed 16S rRNA sequencing on stool from Ifnlr1 +/+, Ifnlr1 +/, and Ifnlr1 -/- mice. We have included these data in Figure S1 and described them in text (Lines 117-120): “To rule out contributions of Ifnlr1 toward an altered intestinal bacterial microbiota, we performed 16S rRNA sequencing on stool from Ifnlr1+/+, Ifnlr1+/-, and Ifnlr1-/- mice and did not find differences in α-diversity and β-diversity measurements.”

Secondly, we find the possibility of bacteriophage mechanisms to be intriguing, but have collected no data on this topic. We briefly collaborated with Dr. Brooke Napier’s group at Portland State University to assess whether ketogenic diet affects homeostatic ISG expression by altering LPS tolerance, but did not observe any interactions in our limited studies. For the reviewer’s benefit, in Author response image 1 we have data from stripped intestinal epithelium of mice fed standard chow and a keto diet.

Author response image 1

4. I think that the authors should consider citing the basal signaling in Caco2 cells and organoids (Hakim Sci. Rep. 2018 8:8341) and the interferon-independent (Shi 2019, Cell 179, 644-658) papers in the discussion.

We agree that basal interferon responses and interferon-independent bacterial clearance of murine rotavirus is very interesting. We have expanded the discussion to include citations to both references.

We mention Hakim et al. in lines 659-661: “ Although there are previous reports of basal, non-receptor-dependent ISG expression in immortalized and primary cell lines (Hakim et al., 2018), the homeostatic response that we report here is dependent on the IFN-λ receptor.” and Shi et al. in lines 726-729: “Although bacterial-associated, IFN-independent mechanisms of mRV clearance have been reported (Shi et al., 2019; Zhang et al., 2014), we found that the signature of homeostatic ISGs in ileum tissue included well-characterized antiviral ISGs (Figures 1-2).”

Reviewer #2:

[…]

For Figure 5 and S3 data, the authors examine gene expression in intestinal cells collected from mice and then enriched for epithelial vs. lamina propria cell populations. They see significantly reduced expression of IFN upon ABX treatment in just one case-harvested epithelial cells that have been enriched for CD45+ cells (leukocytes) have reduced IFNlambda expression. Lamina propria CD45+ cells do not show significantly reduced IFNlambda. This is surprising. While there are low levels of CD45+ cells in epithelial cell preps (as they show in Figure S3), there are higher levels of CD45+ cells in the LP preps. It seems odd that the only significant difference lies with these immune cells in the epithelial cell prep, and this effect is not observed in LP preps, which should contain many more CD45+ cells. Given the authors' conclusion that leukocytes are the source of IFNlambda, this deserves more follow-up or at least some discussion. Is there something special about CD45+ cells from epithelial preps? Is it possible these are contaminating cells? Or is it simply an issue with spread in the data?

Previous work by Mahlakoiv et al. (Figure S3 of Mahlakoiv 2015) showed that CD45+ cells in intestinal epithelial cell preps express IFN-λ transcripts at homeostasis and therefore, we do not find it surprising to detect IFN-λ transcripts in this cellular fraction. The data depicted in Figure 5 and Figure S6 (previously S3) build upon the findings of Mahlakoiv et al. by confirming that these homeostatic IFN-λ transcripts are dependent upon the presence of bacterial microbiota. Although CD45+ cells are more abundant in LP preps, our and Mahlakoiv et al’s findings suggest that the proximity of CD45+ cells to bacterial stimuli in the intestinal epithelial layer is more indicative of IFN-λ expression than the relative abundance of CD45+ cells. To clarify these points, we have updated the Results section to elaborate on the findings of Mahlakoiv et al. (lines 336-338): “A previous study (Mahlakõiv et al., 2015) noted the presence of IFN-λ transcripts (Ifnl2/3) at homeostasis in CD45+ cells within the stripped intestinal epithelium, but not in the lamina propria.”

We additionally emphasize the potential role of CD45+ cells from the lamina propria in our conclusion of Figure 5 (lines 348-350): “From these data, we conclude that epithelium-associated CD45+ leukocytes are the likely source of homeostatic IFN-λ in response to bacterial microbiota, but we do not rule out additional involvement of CD45+ cells in the lamina propria.”

We have also expanded our discussion to include further rationale for CD45+ cell surveillance of the intestinal epithelium for stimuli from the bacterial microbiota (lines 685-690): “Although the cell type that produces homeostatic IFN-λ is unknown, enrichment for Ifnl2/3 in CD45+ cells within the intestinal epithelial fraction suggests that proximity to bacterial stimuli may be a primary determinant in this response. We suggest that these cells may be actively surveying the intestinal epithelium to detect bacterial MAMPs. The specific CD45+ cell types responsible for producing homeostatic IFN-λ will be a topic of interest for future studies.”

Our future work beyond the scope of this study aims to definitively characterize the IFN-λ source cell type and its requirement for homeostatic ISG expression.

In Figures 8 and S6, the authors examine Ifit1 expression in WT or mutant mice with different treatments (ABX, LPS, fecal transfer), to determine if LPS or fecal transfer can restore gene expression. Due to low expression levels and a high degree of spread in the data, they resort to binning the primary gene expression data (8H and S6A; ABX vs. ABX/LPS not significantly different) into + vs. – categories (Figure 8J, S6E; ABX vs. ABX/LPS now are significantly different) to allow different statistical analysis. From this, they conclude that ISG expression is partially restored by LPS treatment in ABX-treated mice. Since the data are just around the detection limit and it's a high bar to restore gene expression, it's perhaps understandable that the data are binary, but this isn't the most convincing data set.

We appreciate that Reviewer #2 recognizes the difficulties associated with analyzing the datasets present in Figure 8 and Figure S9 (formerly S6). Because control reconstitution with fecal transfer results in incomplete restoration of homeostatic ISGs, we find it compelling that restoration of Ifit1 signal by LPS in ABX-treated mice is also localized. Although these results are variable, we think that the use of three candidate ISGs and two experimental approaches increases the weight of their evidence. Additionally, we now include experiments testing a role for a distinct MAMP (flagellin) in restoration of homeostatic ISGs in Figure S9. Flagellin results in partial restitution, similar to LPS, further increasing the collective strength of these data.

Reviewer #3:

[…]

An alternative explanation for the observed ISG expression phenotype could be a difference in baseline microbiota composition between WT and IFNLR -/- mice. This should be checked by 16S rRNA gene sequencing of small and large intestinal samples and could be complemented by a fecal transfer from IFNLR-/- mice into ABX treated WT mice followed by measurement of baseline ISGs in the intestine.

We now provide evidence that microbiota composition does not influence the homeostatic ISG response detailed in this manuscript. We have found no difference in the number of observed bacterial species, Shannon Diversity Index, or UniFrac distance when we performed 16S rRNA sequencing on stool from Ifnlr1 +/+, Ifnlr1 +/-, and Ifnlr1 -/- mice. We have included these data in Figure S1 and described them in text (Lines 117-120): “To rule out contributions of Ifnlr1 toward an altered intestinal bacterial microbiota, we performed 16S rRNA sequencing on stool from Ifnlr1+/+, Ifnlr1+/-, and Ifnlr1-/- mice and did not find differences in α-diversity and β-diversity measurements.”

Furthermore, although early figures used separately bred WT and Ifnlr1-/- mouse lines, we went on to validate findings using littermate Ifnlr1flox/flox and Ifnlr1ΔIEC mice in subsequent figures (Figures 4 and 6). These littermate comparisons in combination with 16S rRNA sequencing suggest that unique microbiota composition is not the primary determinant of the homeostatic IFN-λ response.

Qualitative differences in colonisation could also explain the binary response of WT mice to fecal transfer.

It is plausible that localized differences in colonization by specific members of the bacterial microbiota may exist and may explain the binary response of WT mice to fecal transfer. However, because administration of free LPS (and also flagellin in revised manuscript) to ABX-treated mice results in binary restoration of Ifit1 signal, our model is that bacterial colonization is not necessary for localization of ISG signal. However, we have not fully ruled out a contributory role for qualitative differences in bacterial colonization, so we now include discussion of this possibility into the discussion in Lines 698-701: “The basis for this pattern is unknown; however, it may reflect the distribution of cells capable of sensing bacterial microbiota, distinct microenvironments within the gastrointestinal tract, or qualitative differences in bacterial colonization.”

The focussed ISG response shown in Figure 8 is intriguing but difficult to reconcile with the 20x reduced viral titres in WT mice. Does the virus only replicate in these sites? And are commensal microbiota specifically dense in these areas. A multicolour FISH approach could answer both of these points.

We agree that experiments to visualize sites of virus replication are important context. Therefore, we have now included multicolor FISH depicting the distribution of Ifit1 and murine rotavirus transcript at 24 hours post-infection (Figure 9I) and have included description of these results in-text (Lines 598-602): “To further contextualize the role of homeostatic ISGs in protection against mRV, we performed in situ hybridization for both mRV and Ifit1 in the ileum of WT mice at 24 hours post-inoculation (Figure 9I). We observed variability in the magnitude of mRV infection between mice at this timepoint; however, we generally found that mRV RNA was not present in villi that have Ifit1 expression.” To summarize, we observe that mRV-infected IECs are primarily Ifit1 negative, and Ifit1-positive villi are rarely infected by mRV. These observations are consistent with mRV evasion of ISG production by infected cells, and imply (but do not prove) a protective role for preexisting ISG expression. We propose that regions with homeostatic ISG expression may have increased chance of initial encounter with mRV and are resistant to infection, but that these regions are not exclusive sites of mRV encounter. In addition to imaging, this model is partly informed by our observations that luminal MAMP administration results in localized ISG response and suggests differential access of luminal contents (Figure 8, Figure S9).

Local variation in bacterial microbiota concentration is an attractive hypothesis for localized ISG expression, but we have been unable to collect sufficient evidence to support it. Our attempts to retain mucus and bacteria within our tissue sections were not consistently successful across the large tissue sections needed to properly survey for localized ISG expression. Additionally, as mentioned above, administration of luminal MAMP to ABX-treated mice partially restores Ifit1 signal, so we infer that qualitative differences in bacterial colonization are insufficient to explain the localization of ISG signal.

Reviewer #4:

[…]

– In Figure 1 the authors claim that there is no significant difference between ifnlr-/- +/- ABX. However, this is hard to say from the presented data. There are no clear differences in the Enrichment plots D-F and the values given in B seem to be wrong as it seems very unlikely that both the FDR and the Nominal p-value are the exact same number. Without the correct numbers it is hard to judge the conclusion of this figure.

In response to these and other reviewer comments, we have made multiple changes listed below to more clearly present our GSEA analysis. Together, our conclusions are unchanged from our prior submission, but the presentation is improved in clarity and includes additional information about gene set changes in our treatment groups.

1) We have removed genes with 0 counts in all samples because they do not provide information for enrichment comparisons and decreased visual clarity of the prior enrichment plots.

2) To improve upon our prior “circular” approach of using IFN-λ treatment response (Lee et al.) to test for ISG enrichment, our revised manuscript analyzes our data by GSEA for enrichment of the independently-curated HALLMARK gene sets, which include ISG sets and other pathways relevant to intestinal homeostasis (Figure S2) (Liberzon et al., 2016).

3) We now report only FDR, which is adjusted for multiple comparison testing such as the hallmark gene sets. The values in the previous Figure 1B (both FDR and p-value) were equal because of the usage of a single gene set, where FDR takes into account multiple hypotheses testing with multiple gene sets.

4) In our new analysis using HALLMARK gene sets (Figure S2), we find that INTERFERON_Α_RESPONSE genes are most upregulated in the ileum with IFN-λ stimulation, consistent with the known overlap between type I and III IFN signaling pathways. Therefore, we use this independently-curated set of ISGs for enrichment plots in Figure 1B-E.

– On Figure 1, the authors claim that WT mice separate from ifnlr-/- mice, however in Figure 1H, one of the WT mice and ifnlr-/- mice cluster together. The authors should add more mice to show this more clearly or tone down this statement.

We agree that our prior use of hierarchical clustering to support the claim that there is no significant difference between Ifnlr1-/- mice regardless of ABX-treatment was not convincing. Therefore, we have instead included statistical analysis of count values of individual homeostatic ISGs in Figure 1G. We find there is no statistically significant difference between the ISGs of ABX-treated Ifnlr1-/- mice and conventional Ifnlr1-/- mice. These data provide additional quantification that augments our initial gene set enrichment analysis. In light of our new analysis, we have amended our commentary on the results of the heatmap in Figure 1G (formerly Figure 1H) on lines 154-162: “Comparison of homeostatic ISG transcript counts between experimental treatments revealed similar insights as prior GSEA analysis. WT mice treated with IFN-λ had higher expression of all homeostatic ISGs than untreated mice, whereas WT mice with a conventional microbiota had higher expression of homeostatic ISGs than Ifnlr1-/- mice and ABX-treated mice of both genotypes (Figure 1G). We did not detect additional decreases in these homeostatic ISGs in ABX-treated Ifnlr1-/- mice relative to conventional Ifnlr1-/- mice, suggesting that Ifnlr1 is necessary for expression of homeostatic ISGs (Figure 1G). These results indicate that there is modest but significant expression of ISGs at homeostasis that is lost with Ifnlr1 deficiency or ABX treatment.”

– On Figure S2, the IFNL without ABX control is missing to be able to interpret the data set.

We do not think that the omission of IFN-λ-treated control mice precludes interpretation of this dataset. We have reported these supplementary data in lines 217-220 where we write, “Stimulation with small amounts of IFN-λ rescued ISG expression in whole tissue (Figure S5, formerly Figure S2), indicating that reduction of homeostatic ISG expression upon treatment with ABX is not due to an inability of the intestine to respond to IFN-λ”.

Our data are sufficient to show that IFN-λ injection of ABX-treated mice can significantly increase three candidate ISGs in the ileum, which suggests that the effect of ABX is not on the responsiveness of the intestine to IFN-λ, but rather, the production of endogenous IFN-λ.

– In Figure 3, why are the het samples shown as stripped epithelium and not full tissue as the others in the experiment. It would be good to have all tissue samples or all stripped epithelium to be able to make a full comparison between the data sets.

We have updated Figure 3D to display analysis of full tissue rather than stripped epithelium in the interest of consistency with data in Figure 3A-C.

– Additionally, in Figure 3 the ifr7-/- effect on ISGs levels is very modest and is further decreased with ABX (which is normally considered a criteria for it to not be key for this response) the authors should provide more evidence that irf7 is involved or tone down their conclusion in this regard.

We have toned down our conclusion when describing Figure 3B (Lines 273-279): “However, we observed a modest (twofold) decrease in Ifit1 expression in both the ileum and colon of Irf7-/- mice when compared to WT mice (Figure 3B). Although IRF7 is implicated by these data, it does not appear to be strictly required for homeostatic expression of Ifit1 because expression is further reduced by ABX treatment (Figure 3B). These data suggest that IRF7 is not necessary for the homeostatic ISG response to the bacterial microbiota.”

– The impact of ABX and ifnlr-/- seems to be less apparent in the colon than the ileum (e.g. Figure 2A vs B). This is curious since the colon will have the higher bacterial load. Can the authors comment on this?

We observe that homeostatic ISGs are expressed at lower levels in colonic tissue as compared to ileal tissue; therefore, the decreases of these ISGs with ABX and in Ifnlr1-/- mice are less robust than their analogous condition in ileal tissue. We have now included paired comparisons of Ifit1 expression from the ileum and colon of WT mice in Figure S4, and thank

the reviewer for drawing our attention to clarify this point

– Figure 5B, it is curious that there is no induction of ifnb1 in the epithelial fraction upon poly IC treatment. While several studies have shown that the cells are not responsive to type I IFNs, there is no data showing that they cannot produce type I IFNs upon MAMP stimulation. Can the authors explain this?

The lack of Ifnb1 induction by IECs in the epithelial fraction upon poly I:C treatment is consistent with a prior publication by Mahlakoiv et al. (Mahlakoiv 2015, Figure 2B), which reported that the bulk epithelial fraction lacks Ifnb1 induction upon poly I:C injection. Epithelial cells make up the majority of the epithelial cell fraction; therefore, we expected enriched EpCAM+ cells to lack Ifnb1 induction, consistent with Mahlakoiv et al.

– Figure S3, why are the values of CD45+ cells not closer to 100% post-enrichment? If there are still 50% of non-CD45+ cells that are being analyzed, what are they and can they impact the results shown in Figure 5?

The initial abundance of CD45+ cells in the intestinal epithelial strip is relatively low (~5%), so our enrichment method does not yield near 100% puruty. Upon purity analysis by flow cytometry, we have noted an abundance of dead EpCAM+ cells. In the interest of transparency regarding the cellular material used for qPCR, we have included all live and dead flow events for our purity analysis (Figure S6), and this point is now further clarified in the revised figure legend. We would suggest that the inclusion of CD45-negative events in the CD45-enriched population does not lead to erroneous conclusions: we find the purity of EpCAM+ cells is quite high (> 90%) post-enrichment, but we cannot detect IFN-λ transcripts in these cells; therefore, the contamination of EpCAM+ events in the CD45-enriched populations does not confound our conclusions from Figure 5.

– Figure S4, high magnification images are needed to see if the crypts are able to respond to IFNs in a murine model to clearly demonstrate whether it would be possible for them to have IFN dependent homeostatic ISGs or whether the authors are already selecting for a phenotype only present in mature enterocytes.

We have updated Figure S7 (formerly figure S4) and now show increased magnification fields as insets. We have also clarified the language in the Results section of Figure 6 to reflect that this stimulation depicts minimal Ifit1 expression in intestinal crypts, consistent with the lack of homeostatic ISG expression by crypt-associated cells in Figure 6A and, subsequently, Figure 7 (Lines 390-395): “ We determined that localized ISG expression within individual villi was not due to a localized ability to respond to IFN-λ because stimulation with exogenous IFN-λ resulted in Ifit1 expression within all intestinal villi, but not intestinal crypts (Figure S7). The minimal expression of Ifit1 within intestinal crypts following exogenous IFN-λ treatment suggests that homeostatic ISGs are localized to mature enterocytes because intestinal crypts do not exhibit robust responses to IFN-λ.”

– Figure S5, the colon data is not very convincing as the ifnlrKO shows some background staining which does not look much different from the other images. Additionally, it is not clear why only floxed and dIEC mice are used in this figure and there are no WT mice used as in Figure 6. It would be good to have the matching animals to better compare the data. It would also be important to include high mag images to clearly see if the ISG distribution is the same (as it does not seem to be the same tip distribution that is seen in the ileum). This figure should also include a quantification as provided in other figures.

In Figure S8 (formerly figure S5), we provided images of Ifit1 expression in the colon of Ifnlr1Het/Ifnar1Het and littermate mice singly deficient in Ifnlr1 (Ifnlr1KO/Ifnar1Het) or Ifnar1 (Ifnlr1Het/Ifnar1KO). Our manuscript thus includes images from both full and conditional KO mice as well as relevant controls. We are confused regarding the reviewer’s comments about Ifnlr1flox/flox and Ifnlr1ΔIEC mice as these mice do not appear in our former Figure S5.

To increase ease of viewing and ability to distinguish signal from background, we have provided higher magnification insets of areas of interest for these images, consequently these insets also increase appreciation of the lack of Ifit1 staining in mice lacking Ifnlr1 (Ifnlr1KO/Ifnar1Het). We have also added double-knockout (Ifnlr1KO/Ifnar1KO) images to emphasize our conclusions. Finally, we have added quantification of the area of Ifit1 expression in tissue sections of Ifnlr1Het/Ifnar1Het, Ifnlr1KO/Ifnar1Het, Ifnlr1Het/Ifnar1KO , and Ifnlr1KO/Ifnar1KO mice.

– Figure 7, crypt cells are not a cell type. There are stem cells, TA and other progenitors present that make up the crypt. This should be renamed to correctly indicate the type of cells in this population.

We have updated our language in the Results section of Figure 7 to more accurately convey the cell types present in the “crypt-associated” grouping that we defined. In certain cases, we refer to a group of cells from the single-cell sequencing data produced by Elmentaite et al. that were annotated as “crypt”. In the interest of accuracy and readability, where applicable in the results of Figure 7, we have clarified the decision to retain the cellular annotations bestowed by the original publishing group and removed any reference to this crypt group as a type of cell.

– Additionally, the scRNA-Seq used for human analysis was a bit strange. This paper used mainly embryos and children to derive the cell types. The fact that only a few cell types were highly present seems to indicate that maybe this was not the best paper. It would be good to compare to stronger papers such as Wang et al. JEM 2020, Fuji et al. Cell Stem Cell 2018. There is also a paper using human ileum showing that stem cells and enterocytes each have their own basal ISGs which are stem cell specific or enterocytes specific (Triana, 2020). The statements about the human data should be toned down or a more robust scRNA data set should be used.

We have not used the data gathered from embryos and have only analyzed the data from healthy children. Rather than add more datasets to the manuscript, we have chosen to tone down our conclusions regarding our analysis of the human scRNA-sequencing data by Elmentaite et al. (Lines 480-482): “Our analysis of a human scRNA-seq dataset is consistent with observations in the murine model, though future studies will be required to definitively address the existence of homeostatic ISGs in human tissue.”

– Figure S6D, why are there different numbers of mice used for each ISG? If the RNA is harvested, then the same RNA can be used to assess all three ISGs. It seems strange that of 30 mice tested for ifit1, only 17 were tested for stat1. These missing numbers seems strange and should be included.

We thank the reviewer for noting the values in Figure S9D (formerly S6D). Upon bringing it to our attention, we realized that the table in question had transposed columns. We have updated the table to display the correct values – for FT rescue, we have tested 30 mice for Ifit1, and have tested slightly fewer mice (27 and 24) for Stat1 and Oas1a, respectively.

– The homeostatic ISGs are only found in the tip of the villi, and rotavirus infection normally occurs at the tips of the villi, the authors claim that these homeostatic ISGs are more protective against infection. To make this claim the authors need to perform a co-staining of rotavirus infected tissue to show that the villi that are infected are not the villi which express the homeostatic ISGs. In general, rotavirus staining is needed as the FACS data suggests that a very low numbers of cells are infected which makes it hard to justify the authors claims.

We have now included multicolor in situ hybridization depicting the distribution of Ifit1 and murine rotavirus at 24hr postinfection (Figure 9I) and have included description of these results in-text (Lines 598-602): “To further contextualize the role of homeostatic ISGs in protection against mRV, we performed in situ hybridization for both mRV and Ifit1 in the ileum of WT mice at 24 hours post-inoculation (Figure 9I). We observed variability in the magnitude of mRV infection between mice at this timepoint; however, we generally found that mRV RNA was not present in villi that have Ifit1 expression.”

– The authors claim that IFN is being produced from CD45+ epithelium derived lymphocytes. To support and strengthen this claim, it would be important to perform RNA scope with markers for these cells to determine if the sparse ISG induction seen in the tissue is correlates with the localization of these cells.

Our data showing that IFN-λ is being produced by CD45+ cells associated with the intestinal epithelium is consistent with previous observations by Mahlakoiv et al. (Mahlakoiv 2015, Figure S3). We agree that further defining the cellular source of IFN-λ in response to microbial stimuli is a priority. However, we believe that this question demands robust experimental verification beyond that of correlative imaging and aim to pursue these studies in the future.

– Since the authors claim is that the homeostatic ISGs come from sensing of bacterial components, and that LPS can rescue this phenotype, then the authors should determine if tlr4-/- mice also lack the presence of homeostatic ISGs.

As suggested by the reviewer below, we now include additional data with flagellin stimulation, which is capable of partially restoring homeostatic ISGs. Therefore, we would not expect deficiency of a single TLR to prevent expression of all homeostatic ISGs. In fact, MyD88 deficiency does not uniformly prevent expression of homeostatic ISGs across tissue sites (ileum and colon). Separate studies beyond the scope of the current manuscript will be needed to further fully define the cell types and signaling pathways responsible for homeostatic IFN-λ production.

– Furthermore, the authors claim in the discussion that microbiota components are responsible for the homeostatic levels and that they signal through several MAMPs. However, this is not shown in the paper. In the manuscript the authors only tested LPS. They also used CpG which will signal through MyD88 but does not induce this response. To make the claim that multiple MAMPs can be used, then several other MAMP agonists should be used to show that this is not only a LPS phenotype (e.g. flagellin).

We now include evidence that peroral flagellin is able to stimulate the homeostatic IFN-λ response to a similar extent as LPS in antibiotic-treated mice (Figure S9, formerly Figure S6). We have now included additional commentary regarding this finding in our results (Lines 526-535): “To corroborate these findings, we performed orthogonal analyses of Ifit1, Stat1, and Oas1a expression in ileum tissue of WT mice treated with ABX followed by fecal transplant, LPS administration, administration of the TLR5 agonist: flagellin, or administration of TLR9 agonist: CpG DNA (Figure S9). These data exhibited high variance, but were consistent with imaging data in Figure 8, indicating a partial restoration of homeostatic ISGs in 2050% of ABX-treated mice by LPS, flagellin, or fecal transplant, but not CpG DNA (Figure S9). Together, these data indicate that both LPS and flagellin are sufficient to restore homeostatic ISG expression to a similar extent as total microbiota. The ability of multiple TLR stimuli to partially restore homeostatic ISGs suggests that exposure to a variety of bacterial MAMPs is the basis for localized, homeostatic ISG expression.”

[Editors’ note: what follows is the authors’ response to the second round of review.]

Essential revisions:

Functional protection against rotavirus

The reviewers noted the need for quantification of the colocalization of murine rotavirus and Ifit1 expression in our co-staining images. Reviewer 3 also requested additional experimental replicate of Ifit1 imaging at the 96hr post-inoculation timepoint in Figure 9 —figure supplement 2 (previously Figure 9 —figure supplement 1), and MRV colocalization imaging of infection at 96hr post-inoculation.

To address these concerns, we have taken the following actions:

1) We performed an additional experimental repeat of 96hr post-inoculation data in WT mice and included these data in Figure 9 and Figure 9 —figure supplement 2.

2) We now provide quantification of murine rotavirus and Ifit1 from images (12, 24, and 96hr p.i.) in Figure 9 and Figure 9 —figure supplement 2.

3) Though not specifically requested by reviewers, we also added genome qPCR and quantification of %mRVpositive cells by flow cytometry for the 96hr post-inoculation dataset to increase uniformity in the data presentation across timepoints (Figure 9 —figure supplement 3).

The reviewers noted concerns regarding our interpretation that pre-existing microbiota-dependent Ifnlr1dependent signaling is responsible for protection against murine rotavirus infection.

Our understanding from reading the specific comments of reviewers is that quantitation of co-localization in our images, above, may address their concerns. However, we have also revised the Results section to include quantification of Ifit1 and mRV co-incidence on lines 665-668; “Quantification of co-staining for Ifit1 in mRVinfected cells indicated that a minority (~30%) were Ifit1+ at 12hr or 24hr post-inoculation, whereas a majority (~63%) of mRV-infected cells were Ifit1+ at 96hr post-inoculation (Figure 9I-J and Figure 9 —figure supplement 2).”

Ultimately, we are sympathetic to the reviewers’ puzzlement about how localized, ISG-expressing IECs could provide meaningful protection against viral infection and understand that the reviewers are unconvinced by our discussion of our model in the paper.

We have now addressed alternative models that explain our data in the Discussion section on lines 830844.

“Although we found that homeostatic ISGs provide protection during initiation of mRV infection, it remains unclear how these localized ISG pockets impart this protection. Given our findings in Figure 8 and Figure 8 —figure supplement 1, we suggest that these localized ISGs may be indicative of locations that are particularly vulnerable to viral infection if ISGs were not present at the time of viral exposure. However, alternative explanations for the protective effects that we observe are also plausible, such as: (i) an inability to detect the full magnitude of ISG expression by imaging, or (ii) an unknown temporal component to homeostatic ISG signaling that may be coincident with durable ISG protein expression. Given the magnitude of signal amplification in RNAscope in situ hybridization and the similarity of Ifit1 expression in our imaging data to Ifit1 expression in public singlecell sequencing datasets, we find it unlikely that there is more widespread homeostatic ISG expression below the limit of our detection by imaging. However, we do think an uncharacterized temporal component of homeostatic ISG signaling is plausible, wherein individual villi may be rapidly and transiently expressing homeostatic ISGs in response to sensing of bacterial microbiota. This temporal model may also include a more durable ISG protein response that is not fully concordant with expression of ISG transcripts.“

Partial effect of LPS and Flagellin

Reviewers voiced concerns regarding the interpretation of LPS, flagellin, and fecal-transfer experiments to assess ISG restoration in ABX-treated mice. The primary concern with these data was the validity of interpretation and no experiments were suggested by reviewers.

We have substantially updated our interpretation of data from Figure 8 and Figure 8 —figure supplement 1 in the abstract, results, and discussion by softening conclusions, replacing interpretation with statements of fact in the results, and clarifying our interpretation in the discussion.

Examples include:

i) In the abstract:

Updated Text Lines 28-30: “Furthermore, we assessed the ability of orally-administered bacterial components to restore localized ISGs in mice lacking bacterial microbiota.”

ii) In the figure and section headings of Figure 8 and Figure 8 —figure supplement 1:

Updated Text: Assessing the effect of peroral bacterial products on homeostatic ISGs in ABX-treated mice”.

iii) In the results on Lines 552-554:

Updated Text: “However, localized Ifit1 expression was visible in 4/12 ABX-treated Ifnlr1flox/flox mice administered LPS (Figure 8E, 8H-I).” (iv) In the discussion for example on lines 786-789:

Updated Text: “Given the results of our data in Figure 8 and Figure 8 —figure supplement 1, we suggest that LPS administration, flagellin administration, or fecal transplant can partially restore the expression of homeostatic ISGs in the small intestine. This interpretation supports the concept…”.

https://doi.org/10.7554/eLife.74072.sa2

Article and author information

Author details

  1. Jacob A Van Winkle

    Department of Molecular Microbiology and Immunology, Oregon Health & Science University, Portland, United States
    Contribution
    Conceptualization, Investigation, Methodology, Visualization, Writing - original draft, Writing - review and editing
    Competing interests
    No competing interests declared
  2. Stefan T Peterson

    Department of Medicine, Washington University School of Medicine, St. Louis, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-2984-6400
  3. Elizabeth A Kennedy

    Department of Medicine, Washington University School of Medicine, St. Louis, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  4. Michael J Wheadon

    Department of Medicine, Washington University School of Medicine, St. Louis, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-2579-3786
  5. Harshad Ingle

    Department of Medicine, Washington University School of Medicine, St. Louis, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  6. Chandni Desai

    Department of Pathology and Immunology, Washington University School of Medicine, St. Louis, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  7. Rachel Rodgers

    Department of Medicine, Washington University School of Medicine, St. Louis, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  8. David A Constant

    Department of Molecular Microbiology and Immunology, Oregon Health & Science University, Portland, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  9. Austin P Wright

    Department of Molecular Microbiology and Immunology, Oregon Health & Science University, Portland, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1447-5205
  10. Lena Li

    Department of Molecular Microbiology and Immunology, Oregon Health & Science University, Portland, United States
    Contribution
    Project administration
    Competing interests
    No competing interests declared
  11. Maxim N Artyomov

    Department of Pathology and Immunology, Washington University School of Medicine, St Louis, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  12. Sanghyun Lee

    Department of Medicine, Washington University School of Medicine, St. Louis, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  13. Megan T Baldridge

    Department of Medicine, Washington University School of Medicine, St Louis, United States
    Contribution
    Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Supervision, Writing - review and editing
    For correspondence
    mbaldridge@wustl.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-7030-6131
  14. Timothy J Nice

    Department of Molecular Microbiology and Immunology, Oregon Health & Science University, Portland, United States
    Contribution
    Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Supervision, Writing - review and editing
    For correspondence
    nice@ohsu.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4471-7666

Funding

National Institutes of Health (R01-AI130055)

  • Timothy J Nice

National Institutes of Health (T32-GM071338)

  • Jacob A Van Winkle

National Institutes of Health (T32-AI007472)

  • Jacob A Van Winkle

National Institutes of Health (R01-AI139314)

  • Megan T Baldridge

National Institutes of Health (R01-AI141716)

  • Megan T Baldridge

National Institutes of Health (R01-AI141478)

  • Megan T Baldridge

Pew Charitable Trusts

  • Megan T Baldridge

Washington University School of Medicine in St. Louis (MI-II-2019-790)

  • Megan T Baldridge

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

The authors would like to thank the following OHSU core facilities: Integrated Genomics Laboratory, Advanced Light Microscopy Core, and Histopathology Core; and the following WUSTL core facilities: Genome Technology Access Center. TJN was supported by NIH grant R01-AI130055 and by a faculty development award from the Sunlin & Priscilla Chou Foundation. JAV was supported by NIH grants T32-GM071338 and T32-AI007472. MTB was supported by NIH grants R01-AI139314, R01-AI141716, and R01-AI141478, the Pew Biomedical Scholars Program of the Pew Charitable Trusts, and a Children’s Discovery Institute of Washington University and St Louis Children’s Hospital Interdisciplinary Research Initiative grant (MI-II-2019–790). The funders had no role in study design, data collection, and interpretation, or the decision to submit the work for publication.

Ethics

All mice were maintained in specific-pathogen-free facilities at Oregon Health & Science University (OHSU) and Washington University in St. Louis (WUSTL). Animal protocols were approved by the Institutional Animal Care and Use Committee at OHSU (protocol #IP00000228) and WUSTL (protocol #20190162) in accordance with standards provided in the Animal Welfare Act.

Senior Editor

  1. Carla V Rothlin, Yale School of Medicine, United States

Reviewing Editor

  1. Andrew J MacPherson, University of Bern, Switzerland

Publication history

  1. Preprint posted: June 2, 2021 (view preprint)
  2. Received: September 21, 2021
  3. Accepted: February 4, 2022
  4. Accepted Manuscript published: February 9, 2022 (version 1)
  5. Version of Record published: February 17, 2022 (version 2)

Copyright

© 2022, Van Winkle et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 912
    Page views
  • 122
    Downloads
  • 1
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Jacob A Van Winkle
  2. Stefan T Peterson
  3. Elizabeth A Kennedy
  4. Michael J Wheadon
  5. Harshad Ingle
  6. Chandni Desai
  7. Rachel Rodgers
  8. David A Constant
  9. Austin P Wright
  10. Lena Li
  11. Maxim N Artyomov
  12. Sanghyun Lee
  13. Megan T Baldridge
  14. Timothy J Nice
(2022)
Homeostatic interferon-lambda response to bacterial microbiota stimulates preemptive antiviral defense within discrete pockets of intestinal epithelium
eLife 11:e74072.
https://doi.org/10.7554/eLife.74072

Further reading

    1. Immunology and Inflammation
    Jaime James et al.
    Research Article

    Chronic autoimmune diseases are associated with mutations in PTPN22, a modifier of T cell receptor (TCR) signaling. As with all protein tyrosine phosphatases, the activity of PTPN22 is redox regulated, but if or how such regulation can modulate inflammatory pathways in vivo is not known. To determine this, we created a mouse with a cysteine-to-serine mutation at position 129 in PTPN22 (C129S), a residue proposed to alter the redox regulatory properties of PTPN22 by forming a disulfide with the catalytic C227 residue. The C129S mutant mouse showed a stronger T-cell-dependent inflammatory response and development of T-cell-dependent autoimmune arthritis due to enhanced TCR signaling and activation of T cells, an effect neutralized by a mutation in Ncf1, a component of the NOX2 complex. Activity assays with purified proteins suggest that the functional results can be explained by an increased sensitivity to oxidation of the C129S mutated PTPN22 protein. We also observed that the disulfide of native PTPN22 can be directly reduced by the thioredoxin system, while the C129S mutant lacking this disulfide was less amenable to reductive reactivation. In conclusion, we show that PTPN22 functionally interacts with Ncf1 and is regulated by oxidation via the noncatalytic C129 residue and oxidation-prone PTPN22 leads to increased severity in the development of T-cell-dependent autoimmunity.

    1. Immunology and Inflammation
    Magdalena Shumanska, Ivan Bogeski
    Insight

    The oxidative state of a critical cysteine residue determines the enzymatic activity of a phosphatase involved in T-cell immune responses.