Cutting Edge: Expression of IRF8 in Gastric Epithelial Cells Confers Protective Innate Immunity against Helicobacter pylori Infection

Interferon regulatory factor (IRF)8 is a transcription factor of the IRF family with a wide range of functions in immune cell development, activation, immunomodulation, and function. Under steady-state conditions, based on an IRF8 reporter, IRF8 is expressed in developing lymphoid and myeloid cells but not in mature neutrophils, T cells, or megakaryocytes (1, 2). A knockout of the mouse gene encoding IRF8 (Irf8^−/− mice) results in a variety of developmental abnormalities of dendritic cells, B cells, Langerhans cells, and monocytes (reviewed in Ref. 3). IRF8 exerts its regulatory functions by binding to specific DNA sequences. Because IRF8 itself has weak DNA binding activity, heterodimerization with a partner molecule is essential to assert its transcriptional regulatory activity (4). Depending on the partner molecule, IRF8 either activates or represses expression of its target genes. Genome-wide analyses of IRF8 target genes in germinal center B cells, myeloid cells, and brain during an inflammatory response against certain pathogens revealed that IRF8 regulates a relatively conserved set of genes involved in Ag presentation, IFN responses, DNA repair, RNA expression, and protein processing (5–7).

Although IRF8 expression and function were once thought to be restricted to the hematopoietic system, more recent evidence indicates a much broader tissue distribution. Studies in the nervous system (8), heart (9), and other muscle cells (10) identified previously unrecognized functions of IRF8. For example, expression of IRF8 in neurons confers protection against ischemic-reperfusion–induced brain damage (8). Expression of IRF8 in cardiomyocytes prevents development of cardiac hypertrophy by inhibiting calcineurin signaling (9). These data underscore the importance of cell context–dependent expression and functions of IRF8. However, it remains unclear how IRF8 is regulated in these nonhematopoietic cells and what gene programs IRF8 controls.

We report that IRF8 is expressed in gastric epithelial cells (GECs), another nonhematopoietic cell type. Expression levels of IRF8 are enhanced following infection with Helicobacter pylori. Mice deficient in IRF8 exhibit higher bacteria loads than wild-type (WT) controls. We also show that IRF8 is induced by IFN-γ and that IRF8, in turn, promotes IFN-γ production, forming a positive regulative loop to amplify inflammation. Furthermore, using chromatin immunoprecipitation (ChIP) sequencing (ChIP-seq) and RNA sequencing (RNA-seq) analyses, we identified IRF8 target genes in GECs, providing a detailed understanding of how IRF8 functions in gastric mucosal innate immunity.

B6 (WT), Irf8^−/−, Irf8^f/f, and IRF8-EGFP reporter mice were described previously (2, 11, 12). Vil-Cre mice (stock #4586) were purchased from The Jackson Laboratory and bred with Irf8^f/f mice to generate Irf8^f/fVil-Cre mice. All mice were maintained in a specific pathogen–free facility at the National Institutes of Health according to guidelines approved by the National Institute of Allergy and Infectious Diseases (ASP LIG-16) and the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK-K052-NIHMD-13) Animal Care and Use Committees. Infection of mice with the H. pylori SS1 strain (kindly provided by Drs. A. Lee and J. O’Rourke, University of New South Wales, Sydney, Australia) was performed as described before (13). Briefly, mice were inoculated with an oral challenge dose of 10⁷ CFU at days 1, 3, and 5. At 4 and 8 wk following infection, stomach tissues were processed for histology and tissue culture for enumeration of H. pylori. For infection of cells in vitro, mouse GECs were prepared by enzyme digestion, as previously described (14). After culture for 5–8 d, subconfluent cells were incubated with H. pylori at a multiplicity of infection of 100 for 2 d.

Paraffin sections of stomach tissues were stained with a polyclonal anti-IRF8 Ab (Santa Cruz Biotechnology) and/or DAPI using standard procedures and imaged using a Nikon ECLIPSE TE2000-U confocal microscope. For intracellular staining, GECs were fixed and permeabilized using a Fix & Perm Kit (Life Technology) and stained with allophycocyanin-conjugated anti-cytokeratin Ab (pan reactive; EXBIO Praha). Cells were analyzed by a FACS LSR II flow cytometer (BD Biosciences) and FlowJo software.

Microarray analysis was done using RNA extracts from stomach tissues of infected and control mice and Affymetrix chips. Data were analyzed by the National Institute of Diabetes and Digestive and Kidney Diseases Bioinformatics facility. Quantitative real-time PCR (qPCR) was performed as described previously (15). The GSM06 GECs (a gift from Dr. Yoshiaki Tabuchi, University of Toyama) were stimulated with 10 ng/ml IFN-γ for different times before analysis. Primers are listed in Supplemental Table I. Western blot and ChIP analyses were performed as previously described (15). ChIP-seq and RNA-seq analyses were performed as described previously (16). The sequences were submitted to the National Center for Biotechnology Information Gene Expression Omnibus under accession number GSE67476 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?token=yrwtacaspzoxlqp&acc=GSE67476).

The unpaired two-tailed Student t test or Mann–Whitney U test was used to compare differences between two groups. The p values <0.05 were considered statistically significant.

Microarray-based gene-expression profiling of gastric tissues from H. pylori–infected and mock-infected mice revealed a 2-fold increase in Irf8 transcripts in infected stomach tissues (Fig. 1A). This was confirmed by qPCR (Fig. 1B). Infection of a mouse GEC line, GSM06, with H. pylori also resulted in a time-dependent increase in Irf8 expression (Fig. 1C). Immunohistochemical analyses revealed that normal stomach tissues expressed low levels of IRF8, as evidenced by barely positive staining of IRF8 in nucleated glandular cells (Fig. 1D). However, infection with H. pylori resulted in a robust increase in IRF8 in GECs (Fig. 1D). By using an IRF8-EGFP reporter mouse in which an IRF8-EGFP fusion protein is expressed under the control of natural regulatory elements of the Irf8 gene (2), we detected a dramatic upregulation of IRF8-EGFP expression in infected gastric glandular cells (Fig. 1E).

In an in vitro assay, cultured primary GECs of IRF8-EGFP and WT mice uniformly expressed cytokeratin, a marker of gastric epithelial identity (Fig. 1F, upper panels). Following coculture with H. pylori, GECs of IRF8-EGFP reporter mice, but not WT mice, upregulated expression of IRF8-EGFP (Fig. 1F, lower panels), consistent with the data from immunohistolochemistry analyses. As a positive control, IFN-γ, a known potent stimulus for IRF8 expression, significantly upregulated expression of IRF8-EGFP. The results demonstrated that IRF8 was readily induced in isolated GECs, indicating that expression induced by exposure to H. pylori shown in the immunohistolochemistry analyses was not dependent on other gastric cell types. From this, we conclude that infection with H. pylori significantly increased expression of IRF8 in GECs.

To determine whether IRF8 contributes to gastric inflammatory responses against infection with H. pylori, we infected Irf8^−/− and WT mice with the H. pylori SS1 strain and assessed changes in pathology and bacterial loads in stomach tissues at later time points. At 4 and 8 wk postinfection, the extent of stomach tissue damage was indistinguishable in Irf8^−/− and WT controls (data not shown), but bacterial loads in Irf8^−/− mice were significantly greater than in controls at both time points (Fig. 2A).

Because Irf8^−/− mice are defective in many aspects of innate and adaptive immunity, including the development and function of dendritic cells, B cells, Th17 cells, and monocytes (reviewed in Refs. 3, 17), the decreased bacterial clearance in Irf8^−/− mice could be due to a collective dysfunction of these immune effector cells. To gain more insights into GEC-intrinsic effects of IRF8 resulting from H. pylori infection, we first generated chimeric mice by reconstituting lethally irradiated Irf8^−/− mice with hematopoietic stem cells (HSCs) of WT mice. Eight weeks after reconstitution, the mice were infected with H. pylori and examined 2 mo later. Irf8^−/− mice reconstituted with WT HSCs still exhibited higher bacterial loads than WT control mice reconstituted with WT HSCs (Fig. 2B). Next, we generated IRF8 conditional-deletion mice using Vil-Cre–mediated deletion of the IRF8 gene in villin-expressing gastric tissues. The bacterial loads in Irf8^f/fVil-Cre mice were significantly higher than in Irf8^f/f control mice at both 4 and 8 wk postinfection (Fig. 2C). These data strongly support a GEC-intrinsic effect of IRF8 in limiting H. pylori colonization. Next, we investigated H. pylori–induced IFN-γ production in GECs in vitro. Previous studies demonstrated that infection with H. pylori induced expression of IFN-γ in gastric tissues (18). Consistent with this result, the expression of IFN-γ transcripts was significantly increased in gastric tissues following H. pylori infection (Fig. 3A). Infection of the GSM06 GEC cell line and WT primary GECs with H. pylori also induced expression of IFN-γ (Fig. 3B). However, infection failed to induce IFN-γ expression in Irf8^−/− GECs (Fig. 3B). Importantly, IFN-γ transcripts were negligible or absent in gastric tissues of Irf8^−/− mice infected with H. pylori for 4 and 8 wk (Fig. 3C). Taken together, these data suggest that expression of IFN-γ in H. pylori–infected GECs is IRF8 dependent.

As noted previously, IFN-γ is a potent inducer of IRF8 expression in immune cells (19). Interestingly, IFN-γ also induced IRF8 expression in GSM06 cells in a time-dependent manner (Fig. 3D). Similarly, stimulation of primary GECs of IRF8-EGFP mice with IFN-γ also increased IRF8-EGFP expression (Fig. 1F, lower panels). It is worth noting that GECs express IFN-γ receptors constitutively, as determined by qPCR analyses of transcript levels for the IFN-γ receptors Ifngr1 and Ifngr2 (data not shown). Taken together, these data suggested that IFN-γ drives IRF8 expression, which, in turn, promotes IFN-γ production. This positive-feedback loop is expected to amplify IFN-γ–mediated immunity in the early phase of infection, which could change the course of later adaptive immunity to H. pylori by priming CD4⁺ T cells for Th1 differentiation (reviewed in Ref. 20).

Previous high-throughput genomic analyses of tissue-specific IRF8 target genes in stomachs were based on analyses of whole tissues, obviating the opportunity to discern the contributions of gastric cells versus infiltrated inflammatory immune cells, which often express high levels of IRF8 (2). To overcome this obstacle, we performed genome-wide analysis of IRF8 target genes using ChIP-seq and RNA-seq and the GEC line (GSM06) stimulated with IFN-γ, a factor that was more efficient than H. pylori in inducing IRF8 protein expression (Fig. 1F), enabling us to generate high-quality ChIP materials for analysis. To demonstrate that this approach is relevant to the biology of a true infection, we validated some of the target genes identified following IFN-γ treatment using cultured mouse primary GECs infected with H. pylori.

We identified 666 IRF8-specific ChIP-seq peaks in DNA from cells stimulated with IFN-γ. The distribution of IRF8 binding sites, with 20% proximal to the transcription start site, 32% intronic, and 35% intergenic (Fig. 4A), was similar to that recently described for IRF8-induced mouse monocyte differentiation (21). De novo motif analysis revealed that IRF8-bound genomic sequences were greatly enriched (71%) for the sequence, GAAANNGAAA, which matches the core consensus IFN-stimulated response element sequence (Fig. 4B). Of the 666 IRF8 ChIP-seq peaks, 314 bound within genes annotated by RefSeq (Fig. 4C).

RNA-seq analyses of gene expression by the IFN-γ–stimulated and control cells identified 1037 IFN-γ–regulated genes, including 605 that were upregulated and 432 that were downregulated (Fig. 4D). The overlap of IRF8-bound genes identified by ChIP-seq with IFN-γ–stimulated genes identified by RNA-seq included 77 genes: 73 were upregulated, and 4 were downregulated (Fig. 4C, Supplemental Table II). Ingenuity functional analysis revealed that 48% of these genes are involved in endocrine system disorders, gastrointestinal diseases, and immunological diseases, whereas 41% of these genes belong to pathways of antimicrobial response, inflammatory response, and endocrine system disorders (Supplemental Table II). Interestingly, the most increased gene (Ifit3), but not the most decreased gene (Ifi202b), in infected gastric tissues (Fig. 1A) was found to be a direct target of IRF8 (Supplemental Table II). Ifit3 was recently identified as a novel antiviral protein (22, 23) and could be an important innate factor against microbial infection.

Previous gene microarray profiling analysis in the lungs of Mycobacterium tuberculosis–infected mice and ChIP-seq analysis in the brain of malaria-infected mice led to the identification of a common core of 53 IRF8-bound genes that were upregulated in both conditions (5). Given that different tissues express significantly different gene profiles and that IRF8 targets can vary with tissue, this core list could be an underestimate. Nevertheless, we compared this “common core” list with our 77-gene list and identified 17 genes that overlapped in all three tissues. These genes ontologically belong to innate immunity (Ifit3, Nlrc5, Oasl2, Trim21), adaptive immunity, Ag processing and presentation (Cd274, H2-T22, Psmb8, Tap2), GTP signaling (Gbp2, Gbp3, Igtp, Irgm1, Irgm2), and ubiquitination (Rnf19b, Usp18). The small number of overlapping genes also indicates that IRF8-controlled gene programs are mostly cell context dependent.

To determine whether IRF8 is required for expression of these genes, we analyzed expression of Usp18, Upp1, Nlrc5, Ifit3, Ifit1, and Ifi35 in primary GECs of WT and Irf8^−/− mice. The cells were treated with IFN-γ or infected with H. pylori for 2 d, and gene-expression levels were analyzed by qPCR. As shown in Fig. 4E, transcript levels of all six genes were significantly higher in primary GECs of WT mice than in IRF8-deficient controls following stimulation with IFN-γ (Fig. 4E, p < 0.001). As expected, deficiency in IRF8 abrogated H. pylori–induced upregulation of these genes (Fig. 4E). In addition, the degree of altered gene expression in IFN-γ–treated GECs was significantly greater than in H. pylori–infected GECs, suggesting that IFN-γ is a prominent stimulator for elicitation of robust GEC gene-expression programs.

Previous studies support a Th1-biased adaptive immune response during H. pylori–induced gastric infection (20). Our study extends this view by providing new evidence that infected GECs exhibit an IRF8–IFN-γ⁺ regulatory circuit that would facilitate Th1 cell differentiation. Because GECs produce IFN-γ (Fig. 3) (14), and IFN-γ expression in GECs is completely dependent on IRF8 (Fig. 3), these data suggest that IRF8 could be the major regulator for IFN-γ production in infected gastric tissues. Importantly, the IRF8–IFN-γ regulatory circuit may amplify local inflammatory responses to promote bacterial clearance. Thus, it is possible that downregulation of IRF8 in GECs may facilitate chronic infection and enhance the likelihood of H. pylori–induced malignant transformation.

We thank Alfonso Macias for maintaining the mouse colony. This article is dedicated to the memory of Dr. William G. Coleman, Jr., who initiated and supervised this collaborative study with H.C.M. and H.W.

The authors have no financial conflicts of interests.