Altered IFN-γ–Mediated Immunity and Transcriptional Expression Patterns in N-Ethyl-N-Nitrosourea–Induced STAT4 Mutants Confer Susceptibility to Acute Typhoid-like Disease

Acute foodborne bacterial infections remain a major public health problem associated with high morbidity and mortality worldwide. The intracellular bacterium Salmonella enterica continues to pose a global challenge to human health (1). In humans, Salmonella infections cause a range of foodborne and waterborne illnesses, from a self-limiting, localized gastroenteritis to the more severe, potentially fatal systemic disease of typhoid fever (2). Certain infected individuals (1–4%) may further develop recurrent infections or become asymptomatic chronic carriers acting as a reservoir for pathogen persistence and dissemination in the population (3, 4). Typhoid fever is strictly caused by an infection with the human-restricted S. enterica serovars Typhi and Paratyphi. It is primarily endemic in developing areas of the world where poor sanitation and lack of access to clean drinking water are common (5). In contrast, nontyphoidal Salmonella (NTS) serovars, such as Salmonella Typhimurium and Salmonella Enteritidis, are capable of infecting a broad spectrum of hosts, resulting in intestinal and diarrheal disease (salmonellosis). NTS serovars are the second leading cause of foodborne illnesses in North America. Approximately 5% of patients with salmonellosis are at increased risk of further developing invasive transient bacteremia and sepsis (6). Indeed, disease manifestation of Salmonella-related infections in humans is dependent on the complex interaction between environmental factors, bacterial serotype, and host genetic factors.

The outcome of S. enterica infection relies on the activation of both early innate functions and adaptive humoral and cell-mediated immune responses of the host (7, 8). During systemic Salmonella infection, rapid neutrophilic infiltration and phagocytosis by tissue macrophages are crucial in limiting hepatosplenic infection. Activated macrophages limit Salmonella replication and dissemination to new foci by phagolysosomal bacterial killing and secretion of inflammatory chemokines and cytokines, including TNF-α, IL-12, IL-18, and IFN-γ (7). The inflammatory environment results in the rapid recruitment of leukocytes that increase the ability for intracellular pathogen killing and formation of granulomas. Importantly, IL-12 induces IFN-γ secretion from NK cells, which is critical in activating macrophages in the early response against systemic Salmonella infections (9, 10). Protective immunity is further mediated by IFN-γ production from activated CD4⁺ Th1 cells as well as the cytotoxic response of CD8⁺ T lymphocytes (11). Indeed, immunodeficient mice lacking IFN-γ or IFN-γR fail to resolve primary infection with an attenuated Salmonella strain (12). Moreover, neutralization of IL-12 in a mouse model of typhoid fever abrogates host immunity to primary Salmonella infection (13). In addition, clinical reports have shown that patients immunocompromised as a consequence of HIV infection, chronic granulomatous disease, or functional genetic defects in the IL-12/IL-23 (IL-12β, IL-12Rβ1) and IFN-γ (IFN-γR1, IFN-γR2, STAT1) pathways are predisposed to Mendelian susceptibility to mycobacterial disease and/or disseminated Salmonella infection (14–19). The following highlights the importance of systemic immunity to control invasive S. enterica infections.

In mice, S. Typhimurium infection is a recognized experimental model for studying acute systemic disease resembling clinical features of typhoid fever in humans (20, 21). Alternatively, the use of humanized mice (Rag2^−/−γ_c^−/−; [NOD]-scid IL2rγ^null) and, more recently, Tlr11^−/− mice, which have been shown to be efficiently colonized by S. Typhi, are providing novel tools to study typhoid fever (22–25). Following oral infection of mice, Salmonella invade the microfold cells in the intestinal epithelium and are taken up by dendritic cells (DCs) and macrophages in the underlying Peyer’s patches before infecting the mesenteric lymph nodes and eventually disseminating via the circulation to replicate in resident phagocytes of the spleen and liver (26, 27). Previously, the genetic and molecular basis of several mutations, including Nramp1/Slc11a1, Tlr4, and Pklr, has been shown to be important in resistance to Salmonella infection in mice (28–33). However, as the mouse genome has a low frequency of naturally occurring spontaneous mutations, new approaches to identify host susceptibility or resistance genes in the mouse are essential to study the host response to infectious diseases. As such, we have used an unbiased forward genomics approach by N-ethyl-N-nitrosourea (ENU) germline mutagenesis to screen and identify novel mutations critical in antibacterial immunity, as well as to investigate gene function in vivo in the context of Salmonella infection.

In this article, we report the identification of the Ity14 (Immunity to Typhimurium locus 14) pedigree carrying an ENU-induced mutation in Stat4 (Signal Transducer and Activator of Transcription Factor 4) conferring increased susceptibility to sublethal primary Salmonella infection. The transcription factor STAT4 is specifically expressed in myeloid and lymphoid cells and is a critical mediator of IL-12 signaling in development of inflammatory immune responses (34, 35). IL-12 production from activated macrophages and DCs activates TYK2 and JAK2 receptor–associated kinases, resulting in dimerization, phosphorylation, and activation of STAT4. STAT4 has been shown to be involved in both innate and adaptive immunity by regulating the transcription of target genes, including IFN-γ; NK cell cytotoxicity; Th1 cell differentiation from naive CD4⁺ T cells; and Ig isotype switching to IgG1. Consistent with these findings, STAT4 is a central determinant in host resistance to various bacterial, viral, and protozoan infections while playing a critical role in regulating inflammatory immune-mediated diseases (36).

The physiological function of STAT4 in vivo in response to S. Typhimurium infection remains poorly characterized. In this paper, we elucidate the impact of Stat4^Ity14/Ity14 on innate immunity during Salmonella infection. We show that Ity14 mice with a hypomorphic mutation in Stat4 have an increased innate susceptibility following sublethal invasive S. Typhimurium challenge. Overall, Stat4^Ity14/Ity14 mice have increased mortality following infection, with a concomitant progressive increase in splenic and hepatic bacterial load. Using genome-wide expression microarrays in spleen tissue from wild-type and Ity14 mutants, we studied the impact of Stat4 on the inflammatory immune response to Salmonella infection. We validated the importance of IFN-γ–mediated immunity during systemic Salmonella infection. We further show that NK and NKT cells play an important early role in controlling Salmonella in Stat4^Ity14/Ity14 mice. During oral infection using a model of intestinal typhlitis, mice lacking STAT4 also develop early dissemination of S. Typhimurium and systemic disease. In addition, increased expression of genes involved in cell–cell interactions and communication, as well as increased CD11b expression on a subset of splenic myeloid DCs, shows compromised recruitment of inflammatory cells to the spleen during Salmonella infection in Stat4 mutants.

All animal experiments were conducted following specific conditions outlined by the Canadian Council on Animal Care and protocols approved by the McGill University Animal Care Committee. The 129S1/SvImJ, DBA/2J, C57BL/6J, and C.129S2-Stat4^tm1Gru/J mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Experiments were performed using mice between 7 and 15 wk of age of both sexes.

Generation 0 (G0) males on a 129S1 background between 8 and 12 wk of age were i.p. injected with a single ENU dose of 150 mg per kg body weight. Efficiently mutagenized G0 males temporarily experienced a period of infertility and only regained fertility after 11 wk. ENU-induced mutations were brought to homozygosity using a three-generation breeding scheme. N3 progeny on a resistant 129S1-129X1 background were initially screened for susceptibility to S. Typhimurium infection.

DNA was extracted from a tail piece of mice by proteinase K digestion and phenol-chloroform extraction. A genome scan was performed on DNA samples from 21 mice (9 susceptible, 12 resistant) and a panel of 708 single nucleotide polymorphisms (SNPs) for 129 and DBA/2 strains of mice (Medium Density SNP Panel; Illumina GoldenGate, The Centre for Applied Genomics, University of Toronto, Toronto, ON, Canada). Mice were further genotyped by SNP sequencing (McGill University and Genome Quebec). The mutation in Stat4 in Ity14 mice (c.1335+5 G > A) was genotyped using a Custom TaqMan SNP Genotyping Assay (Applied Biosystems, Streetville, ON, Canada).

We followed standard manufacturer protocols to perform target capture with the SureSelect Mouse All Exon Kit (Agilent Technologies, Santa Clara, CA) and sequencing of 100-bp paired end reads on Illumina HiSEquation 2000. This process generated > 8 Gb of sequence for each of the two susceptible Ity14 samples. Reads were aligned to mm9 with BWA (37) and coverage was assessed with BEDTools, showing an average of 58.9 reads covering each base of the consensus coding sequence genes for the mouse genome (38). Single nucleotide variants and short insertions and deletions were called using samtools pileup and varFilter (39) with the base alignment quality adjustment disabled, and were then quality filtered to require ≥ 20% of reads supporting the variant call. Variants were annotated using both Annovar (40) and custom scripts to identify whether they affected the protein coding sequence, and whether they had previously been seen in mouse dbSNP128 or in any of seven mouse exomes sequenced in parallel. To detect splice site mutations, the threshold of detection was increased to 6 bp instead of the standard 2-bp flanking exons.

Mice were challenged i.v. in the caudal vein with an infectious dose of 5 × 10³ CFUs S. Typhimurium strain Keller. Dose preparation and method of infections have been previously described (41, 42). For the complementation assay, mice were infected with a lower dose of 2.5 × 10³ CFUs, given that the C.129S2-Stat4^tm1Gru/J mice are on a BALB/c background (33). Infected mice were monitored over the course of 14 d, and susceptible mutants defined by a body score index < 2 were euthanized by CO₂ asphyxiation. To determine bacterial load in organs at specific time points postinfection, aseptically harvested organs were weighed and homogenized in 0.9% saline using a Polytron (Kinematica, Bohemia, NY). Serial dilutions of organ homogenates were then plated on trypticase soy agar to determine bacterial counts.

At 1 d prior to infection, mice were fasted for 6 h. At 4 h into the fast, mice were gavaged with 20 mg streptomycin sulfate diluted in 100 μL sterile water. An overnight culture of S. Typhimurium SL1344 was prepared in 5 ml TSB supplemented with 50 μg/ml streptomycin sulfate and incubated at 37°C. The day of the infection, 2 ml of the overnight culture was inoculated in 50 ml TSB containing streptomycin, and grown to an OD of 0.9 at 600 nm. Streptomycin-pretreated mice were each gavaged with 5 × 10⁷ CFU bacteria resuspended in 100 μl sterile saline. For CFU enumeration, see the detailed method above.

H&E-stained cecum and colon slides were analyzed in duplicate and blinded for genotype. A histopathological scoring system was adapted from a previously described model (43). Briefly, the score evaluated the submucosal edema (0 = 0, 1 ≤ 10%, 2 = 10–40%, 3 = >40%), the polymorphonuclear cell (PMN) infiltration of the lamina propria (0 = ≤ 5 PMN, 1 = 5–20, 2 = 21–60, 3 = 61–100, 4 = > 100), the goblet cell count (0 = >10, 1 = 6–10, 2 = 1–5, 3 = 0 [per high-power field]), and the epithelial integrity (0 = no change, 1 = desquamation, 2 = erosion of the epithelial surface, 3 = ulceration) of cecum cross-sections (43).

Mice were injected i.v. with 6 × 10⁴ CFUs S. Typhimurium strain XEN26. Whole-body imaging of mice was performed daily, beginning on day 3 to day 8 postinfection on the xenogen IVIS Spectrum System (Caliper, Alameda, CA). Mice were anesthetized with isoflurane and anteriorly shaved prior to imaging. Bioluminescent images were acquired with an open emission filter, binning factor of 16, and exposure time ranging from 2 to 5 min. The regions of interest were selected, normalized across time points, and quantified using Living Image software v.4.3.1 (Caliper).

Protein extracts from spleen were prepared by homogenizing tissue using a Polytron (Kinematica) in lysis buffer (1 M Tris-HCl, pH = 8; 0.5 M KCl; 0.1 M MgCl₂; 0.1% Triton X-100; 10% glycerol) with protease inhibitor (P2714; Sigma-Aldrich) and further sonicated for 15 s on ice at 60% amplitude. Protein extracts were centrifuged at 13,000 rpm at 4°C for 15 min. Supernatants were collected and protein concentration quantified by Bradford assay (500-0006; Bio-Rad). Whole-cell lysates were extracted in Laemmli buffer and probed by immunoblot using Abs for STAT4 and GAPDH (Cell Signaling). The STAT4 Ab recognizes an epitope around lysine 151 upstream of the Ity14 mutation.

Spleens were harvested and collected from uninfected and day 4 S. Typhimurium–infected mice in 4 ml PBS under sterile conditions. Spleens were macerated with 70-μm cell strainers into single-cell suspensions, treated with ACK lysis buffer, and washed in cold PBS. Cell counts were determined using a Coulter Z2 particle counter (Beckman Coulter). For splenocyte stimulation, 5 × 10⁵ cells were plated in 96-well round-bottom plates and stimulated with recombinant mouse IL-12 (20 ng/ml; R&D Systems) or LPS (1 μg/ml; Sigma-Aldrich) for 2, 4, 8, and 24 h at 37°C, 5% CO₂. For surface staining, 5 × 10⁶ splenocytes were plated in 96-well round-bottom plates and stained for various cell surface markers, using fluorochrome-labeled mAbs (eBioscience). The following anti-mouse Abs were purchased from eBioscience: CD49b (clone DX5), CD4 (GK1.5), CD8α (53-6.7), CD45R (RA3-6B2), Ly-6G (RB6-8C5), F4/80 (BM8), CD11c (HL3), and CD11b (M1/70). CD3e (500A2) was purchased from BD. Fixable Viability Dye (eBioscience) to irreversibly stain dead cells was used in accordance with the manufacturer's protocol. Cells were acquired on a FACSCanto II (BD) flow cytometer, and results were analyzed using FlowJo (v9.4.10) software.

For intracellular cytokine staining, 10 × 10⁶ splenocytes were plated in complete RPMI 1640 media in six-well tissue culture–treated plates. Cells were stimulated and activated ex vivo with 50 μg/ml PMA and 1 μg/ml ionomycin (Sigma-Aldrich) for 4 h in the presence of GolgiStop (BD). IFN-γ (XMG1.2) and IL-4 (eBioscience) were stained according to the manufacturer’s protocol (Cytofix/Cytoperm Plus Fixation/Permeabilization Kit with GolgiStop; BD). Cells were acquired on a FACSCanto II, and results were analyzed as described above. IL-6, TNF-α, and IL-10 cytokines in the supernatant of splenocytes, as well as IFN-γ and IL-12p70 in serum, were measured by sandwich ELISA (eBioscience).

Spleens of Ity14 mice were collected for microarrays and frozen at −80°C. Total RNA was extracted from ∼ 50 mg splenic tissue, using TRIzol reagent (Invitrogen Canada, Burlington, ON, Canada). RNA yield was determined using a NanoDrop spectrophotometer (ThermoFisher Scientific, Waltham, MA) and the overall quality assessed by denaturing agarose gel electrophoresis. Four uninfected and four day 4 S. Typhimurium–infected age-matched mice (two females, two males) per genotype were used for whole-genome expression profiling on Illumina Mouse-Ref-8 v2.0 BeadChip (Illumina, San Diego, CA). The quality control, hybridization, and array analysis were performed at the McGill University and Genome Québec Innovation Centre (Montreal, QC, Canada).

Raw data are available through the ArrayExpress database (www.ebi.ac.uk/arrayexpress) under accession number E-MTAB-1931. Preliminary expression data analysis was done using FlexArray 1.4.1 software, as previously described (44, 45) (Genome Québec, Montreal, QC Canada) (http://genomequebec.mcgill.ca/FlexArray). Principal component analysis was generated in FlexArray. Briefly, gene lists for each sample group were generated by selecting for genes with a false discovery rate p < 0.05 by the unpaired Student t test and > or < 2-fold change. The canonical pathway analyses were generated through the use of Ingenuity pathway analysis (IPA; Ingenuity Systems, www.ingenuity.com). Heatmaps were generated using R package (46).

RNA quantification by the SYBR Green–based detection kit was performed using the Applied Biosystem StepOnePlus (Applied Biosystems, Carlsbad, CA). Complementary DNAs were amplified by PCR using the following primer pairs: Usp: 18 5′-AGTGCTGTCTAGAGACCTCTGC-3′ and 5′-GGAGTTAAGGAACACGTCTG-3′; Il4: 5′-TGAGAGAGATCATCGGCATTT-3′ and 5′-GTGAGGACGTTTGGCACATC-3′; Il10: 5′-AGTGGAGCAGGTGAAGAGTGA-3′ and 5′-ATGCAGTTGATGAAGATGTCAAA-3′; Ifng: 5′-ACTGGCAAAAGGATGGTGAC-3′ and 5′-ATCCTTTTTCGCCTTGCTGT-3′; Lcn2: 5′-CAGAAGGCAGCTTTACGATGT-3′ and 5′-TGTTCTGATCCAGTAGCGACA-3′; Mpo: 5′-ATGCTTCAGACCTCCAATGGT-3′ and 5′-CTCTGTCCACTAGCTGCTTGG-3′; Gbp5: 5′-CAGGCAAATCCTACCTGATGA-3′ and 5′-ACCAAAGTGTGGTCTGGCTTT-3′; Stat4: 5′-GCGTCCATTGACAAGAATGTT-3′ and 5′-CCTTGGGTTGCAAATGTCTAA; Stat1: 5′-ACAACATGCTGGTGACAGAGCC-3′ and 5′-TGAAAACTGCCAACTCAACACCTC-3′; and normalized to the reference Hprt gene: Hprt 5′-GTTGGATACAGGCCAGACTTTGTTG-3′ and 5′-GATTCAACTTGCGCTCATCTTAGGC-3′. The data obtained were expressed as 2^−ΔΔCt.

Statistical analyses were done using GraphPad Prism v5.0 (GraphPad Software, La Jolla).

The Ity14 pedigree was identified in an in vivo recessive screen of ENU-mutagenized mice for innate susceptibility to Salmonella infection, as measured by survival analysis (Fig. 1). Mutagenized 129S1 G0 males were crossed to wild-type 129X1 females to generate G1 mice. G1 males were further outcrossed to 129X1 females to produce G2 females. For each G1 pedigree, two G2 daughters were backcrossed to generate N3 offspring (Fig. 1A). We initially screened N3 mice on a resistant 129S1/129X1 mixed background with a sublethal dose of S. Typhimurium and monitored mice for 14 d for development of clinical disease. We observed 25% mortality in Ity14 N3 progeny by day 6 postinfection (data not shown). To map and identify the mutation responsible for the Salmonella-susceptibility phenotype, we outcrossed both the G1 male and the G2 female to DBA/2J mice in an F2 cross. We observed 25% mortality by day 5 postinfection (Fig. 1B), consistent with the survival phenotype observed in N3 animals. Initial genotyping was performed using a total of 21 F2 mice (9 susceptible, 12 resistant) and 708 SNPs (polymorphic between 129 and DBA/2J). We detected linkage to a 25.6-Mb region on chromosome 1 with a maximum LOD score of 6.23 at RS1347566. Fine mapping refined the Ity14 locus to a 2.7-Mb interval (Fig. 1C). At the peak marker, F2 mice homozygous for the 129S1 allele were highly susceptible to S. Typhimurium infection by day 6 postinfection, whereas heterozygous and mice homozygous for the DBA/2J allele were significantly more resistant (Fig. 1D). Overall, we identified the Ity14 pedigree with increased susceptibility to Salmonella infection contributed by a 2.7-Mb interval on proximal chromosome 1.

Whole-exome sequencing of coding exons and flanking splice junctions in two susceptible Ity14 mice identified a novel ENU-induced mutation in Stat4. The average coverage across the exome was 83-fold. The Stat4 mutation was the only one both validated and segregated with the survival phenotype from a list of predicted missense and splicing variants identified by exome sequencing (Supplemental Table II). The mutation consisted of a guanosine to adenosine substitution within the splice donor site of exon 15, at position +5 of intron 15 (c.1335+5G > A) and was confirmed by Sanger sequencing (Fig. 2A). This guanosine is a highly conserved nucleotide involved in the splice donor sequence recognition by the splicesome U1 subunit, and the maximum entropy score predicts significantly weakened splice site recognition (from 7.07 to 1.48) (http://genes.mit.edu/burgelab/maxent/Xmaxentscan_scoreseq.html). Therefore, it was predicted that the mutation would interfere with normal splicing function. PCR amplification of cDNA isolated from wild-type, heterozygous, and mutant spleens using primers located in flanking exons 14 and 16 resulted in a smaller PCR product in mutant mice (Fig. 2B) as a result of an 84-bp deletion corresponding to exon 15. The mutation resulted in a 28-aa deletion in the DNA binding domain of STAT4 (p.G418_E445) (Fig. 2C). We observed decreased expression of a smaller STAT4 protein product at days 0 and 4 postinfection in Stat4^Ity14/Ity14 splenic tissue, compared with littermate controls (Fig. 2D). In contrast, at the transcript level, Stat4 gene expression was downregulated to a lesser extent in Stat4^Ity14/Ity14 mice compared with wild-type at day 4 postinfection (Fig. 2E). We further validated STAT4 as the candidate gene responsible for the susceptibility phenotype in the Ity14 pedigree by allelic complementation assays. Stat4^Ity14/+ mice were crossed to Stat4^−/− mice, and susceptibility to infection was assessed by survival analysis in F2 progeny. We observed a lack of complementation in Stat4^Ity14/− mice with a mean survival time equivalent to that in Stat4^Ity14/Ity14 and Stat4^−/− animals, thereby confirming that the mutation within Stat4 was responsible for the Ity14 phenotype (Fig. 2F ).

Increased mortality following S. Typhimurium infection in Stat4^Ity14/Ity14 mutants paralleled the significantly higher bacterial load in target organs, specifically in the spleen and liver at day 4 postinfection. At day 4 following infection, we observed a 17.2-fold change in Salmonella replication in the spleen and a 87.3-fold change in the liver in Stat4^Ity14/Ity14, compared with wild-type littermate controls (Fig. 3A). In vivo imaging using luminescent S. Typhimurium strain XEN26 validated progressive increased bacterial load in Ity14 mutant mice over the course of infection (Supplemental Fig. 1A, 1B). The in vivo systemic bacterial replication observed in Stat4^Ity14/Ity14 at day 4 after Salmonella infection parallels the low levels of circulating IFN-γ measured in the serum of these mice (Fig. 3B). Moreover, we observed increased IL-12p70 levels in the serum of Stat4^Ity14/Ity14 mice at day 4 postinfection, suggesting a lack of negative feedback (Fig. 3C). Consistent with the essential function of STAT4 in IFN-γ induction downstream of IL-12 signaling, we observed significantly lower levels of IFN-γ production in the supernatant of Stat4^Ity14/Ity14 splenocytes following 24-h stimulation with recombinant mouse IL-12 (Fig. 3D,). IFN-γ is an important mediator of granuloma formation, critical to control and prevent intracellular bacterial dissemination. Consequently, reduced levels of IFN-γ are consistent with the increased bacterial load observed in Stat4^Ity14/Ity14 mice. Although we did not observe any significant differences in proinflammatory cytokine production, IL-6 and TNF-α, or anti-inflammatory cytokine IL-10 following LPS stimulation in explanted splenocytes (Fig. 3E–G), histopathological H&E staining of infected spleen and liver sections further demonstrates that Ity14 mutants have impaired immune cell–mediated recruitment to the site of infection. During Salmonella infection, Stat4^Ity14/Ity14 have both fewer and smaller multifocal microabscesses of inflammatory infiltrates composed of neutrophilic granulocytes and macrophages in liver, as well as in both white and red pulps of spleen, compared with wild-type controls (Supplemental Fig. 1C).

To further demonstrate the importance of STAT4 in controlling systemic S. Typhimurium infection, streptomycin-pretreated wild-type and Ity14 mutant mice were infected per os to study intestinal pathological changes in the cecum, as well as bacterial dissemination to the spleen and liver at days 1 and 4 postinfection. No histopathological differences in the cecum were observed at day 1 postinfection; however, at day 4 postinfection we observed less submucosal edema, less PMN cell infiltration of the lamina propria, increased goblet cell count, and decreased epithelial integrity in Ity14 mutants, compared with wild-type (Fig. 4E). In contrast, Stat4 knockout (KO) mice had increased pathological features at day 1 postinfection, with no differences at day 4 postinfection (Supplemental Fig. 1D). We can clearly detect an effect of the background on the severity of the pathological lesions: control mice on a BALB/cJ background were more affected than control mice on a mixed 129S1:DBA/2J background at day 1 postinfection, whereas the lesions were more prominent in Ity14 mutants compared with Stat4 KO at day 4. These data are consistent with our previous observation that inbred strains present different clinicopathological features of Salmonella-induced typhlitis (45). Despite these differences in kinetics and amplitude of the histopathological scores between Ity14 mutants and Stat4 KO, we observed dissemination of Salmonella to systemic sites in both mutants. Indeed, Ity14 mutant mice had a significantly higher bacterial load in the spleen at day 4 postinfection (Fig. 4A), as well as higher bacterial burden in the liver compared with wild-type littermates (Fig.4B). As expected, Ity14 mutants had significantly decreased serum IFN-γ levels and increased IL-12p70 levels in serum at day 4 postinfection compared with wild-type littermates, as observed during systemic infection (Fig. 4C, 4D). The above data are consistent with what we observed in Stat4 KO mice (Supplemental Fig. 1E–G). In the context of the oral Salmonella infection model, this set of experiments demonstrates that STAT4 is important in controlling bacterial dissemination to the spleen and liver in the presence of intestinal disease.

To assess the impact of the ENU-induced Stat4 mutation in Ity14 mice on innate immunity to Salmonella infection, we performed expression microarrays. Genome-wide expression analysis was done on Stat4^+/+ and Stat4^Ity14/Ity14 mRNA isolated from the spleen of uninfected mice versus day 4 Salmonella-infected mice. Venn diagram results showed that 111 genes were transcriptionally modulated at least 2-fold in a similar way in both wild-type and Ity14 mutants (76 genes upregulated and 35 genes downregulated) at day 4 postinfection. In addition, subsets of genes were differentially regulated uniquely in wild-type (46 genes upregulated and 40 genes downregulated) and Ity14 mutants (50 genes upregulated and 29 genes downregulated) (Fig. 5A, Supplemental Table IA–C). By looking at overall expression patterns of genes significantly up- or downregulated in wild-type or Ity14 mutants after Salmonella infection, we identified two clusters of genes that either are unique to wild-type or Ity14 mutants, which we further focused on (Fig. 5B).

The transcriptional signature common to both wild-type and Stat4^Ity14/Ity14 consisted of genes critical in inflammation. Clusters of genes identified were involved in cell recruitment and cytokine signaling (Ccl3, Ccl4, Cxcl1, Cxcl9, Cxcl10, Ccl21a, Ccl21c, Socs3), type 1 IFN pathway (Usp18, Upp1, Stat1, Irf1, Ifi27, Ifi47, Irgm1, Oasl2), acute phase response (Saa3), proinflammatory signaling (Il1b, Il1rn, Casp1, Casp4), cell growth (Slfn1), and innate immunity to bacterial infections (Lcn2, Cebpb, Cd14, Usp18, Mpo).

Consistent with the role of STAT4 in IFN-γ–mediated immunity, the global gene expression pattern observed solely in wild-type mice over the course of Salmonella infection was predominantly IFN-γ driven. In addition to IFN-γ, a significant number of genes that make up the IFN-γ resistome were differentially regulated following infection, including the guanylate-binding proteins (Gbp1, Gbp2, Gbp5, Gbp6, Aif1, Fcgr3, Igfbp4, Ligp2, Mpo, Mt1, Psmb10, Scd1, Sepp1, Timp2, Trafd1, Ubd, Wars) (Fig. 5C). Furthermore, we observed upregulation of genes critical in costimulatory signal transduction and lymphocyte activation (Slamf8, Nkg7), neutrophil recruitment and cytotoxicity (S100a8, Plac8), Ag presentation (Psmb10), cell adhesion (Emilin2, Lgals9), apoptosis (Batf2, Mfge8), and serine protease activity (Prtn3, Ctsc, Ctsg, Serpinaf). The shift toward development of a Th1 immune response triggered by Salmonella infection in wild-type mice coincided with downregulation of certain genes implicated in Th2 immunity (Sepp1, Ltbp4, Fcna). Although significantly upregulated in both Stat4^+/+ and Stat4^Ity14/Ity14, a number of genes were upregulated at day 4 postinfection but to a lesser extent in Stat4^Ity14/Ity14 mice compared with Stat4^+/+ littermates. The following included T lymphocyte chemoattractants (Cxcl9 and Cxcl10), genes involved in IFN signaling (Irf1, Irgm1, Ifi47, Stat1) and lymphocyte activation and maturation (Ly6c, Ly6a). Overall, the canonical pathways enriched [−log(p value)] in wild-type mice compared with Ity14 mutants consisted of communication between innate and adaptive immunity, cell-mediated apoptosis, costimulation and activation of T cell response, as well as cytokine/chemokine signaling (IPA) (Fig. 5E).

Of interest, various pathogen recognition receptors critical in innate immune signaling and activation of host defense mechanisms were significantly upregulated in Stat4^Ity14/ty14 mice, but not in Stat4^+/+ mice, upon Salmonella infection. The following included members of the C-type lectin family (Clec4a3, Clec4n), scavenger receptor family (Marco), SIRP family (Sirpb1), TLR family (Tlr2), and paired Ig-like receptors (Pira3, Pira4, Pira11) (Fig. 5D). There was significant enrichment [−log(p value)] of innate immune recognition canonical pathways in Ity14 mutants versus wild-type controls (IPA) (Fig. 5E). The NF-κB pathway has previously been reported to be induced during Salmonella infection. We observed significant upregulation of transcripts clustering in the NF-κB pathway, including Irak3, Irf7, Myd88, and Nfkbia, as well as downstream antiapoptotic transcripts (Ier3, Bcl2a1c) unique to Stat4^Ity14/Ity14 mice. In addition, Ity14 mutants upregulated expression of proinflammatory mediators (Il1a, Iftm2, Iftm6, Mmp3, Mmp9, Mmp14), cell growth factors (Anxa2, Anxa3), and inhibitors of cathepsins (Cstb, Stfa1). Although significantly upregulated in both Stat4^+/+ and Stat4^Ity14/Ity14, a set of transcripts were upregulated to a greater extent in Stat4^Ity14/Ity14 mice compared with wild-type mice. The following included genes involved in cytokine and chemokine signaling (Ccl3, Ccl4, Cxcl1, Socs3), innate immunity (Cd14, Cebpb, Chi3l1, Fcgr4, Timp1), type 1 IFN signaling (Oas2), acute phase response (Saa3), and IL-1 signaling (Il1b, Il1rn). A subset of transcripts were further selected, including previously known inflammatory mediators with IFN-γ–independent regulation following Salmonella infection (Lcn2, Mpo, Usp18), anti-inflammatory cytokine Il10, and IFN-γ–regulated genes (Gbp5, Stat1), to validate the expression data using quantitative RT-PCR (Supplemental Fig. 2). Taken together with the enhanced susceptibility to Salmonella infection in Ity14 mutants, upregulation of various innate immune recognition receptors suggests a possible coping strategy for impaired immunity in these mice.

Given the heterogeneity of IFN-γ–producing lymphocytes that contribute to the control of S. Typhimurium infection, we examined cell-specific IFN-γ contribution by multistain flow cytometry (47–49). Upon Salmonella infection, total spleen cell numbers increase similarly both in wild-type [226.1 ± 71.4 (×10⁶)] and in mutant [195.3 ± 71.6 (×10⁶)] mice (Fig. 6A). We did not observe differences in the percentages of CD3^-DX5⁺ NK cells, CD3⁺DX5⁺ NKT cells, CD4⁺ or CD8⁺ T lymphocytes, and CD45R⁺ B lymphocytes between wild-type and mutants prior to and at day 4 after Salmonella infection (data not shown). We further confirmed decreased IFN-γ and increased IL-4 expression at the transcript level in the spleen at day 4 following Salmonella infection in Stat4^Ity14/Ity14 mice (Fig. 6B, 6C). To study intracellular cell-specific cytokine secretion, splenocytes from uninfected and day 4 Salmonella-infected mice were stimulated ex vivo with PMA and ionomycin. Salmonella-infected NK and NKT cells isolated from Stat4^Ity14/Ity14 mice produced significantly less IFN-γ [15.8% ± 9.4 (NK) and 12.6+1.9% (NKT)], compared with littermate controls [45.9%±10.6 (NK) and 41%±8.2 (NKT)] (Fig. 6D, 6E). Of interest, we did not observe any difference in the percentage of IFN-γ secreted from CD4⁺ or CD8⁺ T cells in Salmonella-infected splenocytes. However, Ity14 mutants had significantly lower IFN-γ mean fluorescence intensity (MFI) in CD4⁺ T cells (Fig. 6F). Furthermore, a trend toward lower IFN-γ MFI from CD8⁺ T cells from Ity14 mutants was observed in Salmonella-infected splenocytes (data not shown). In addition, there was increased IL-4 secretion and IL-4 expression by MFI in CD4⁺ T cells from Stat4^Ity14/Ity14 mice after Salmonella infection (Fig. 6G). Previously, IFN-γ⁺CD4⁺ and IFN-γ⁺CD8⁺ T lymphocytes were reported to acquire an activated phenotype (CD44^hi, CD62L^low), and reach maximal IFN-γ production, only about 2–3 wk following virulent S. Typhimurium (SL1344) infection in mice (50). Therefore, early susceptibility in Stat4^Ity14/Ity14 mice, prior to day 5 postinfection, likely explains the observed minimal T lymphocyte IFN-γ response. The following suggests that the impairment of IFN-γ secretion in Stat4^Ity14/Ity14 mice upon Salmonella infection was primarily contributed by NK and NKT cell deficiency.

Several of the genes identified in the microarray analysis that are upregulated in Stat4^Ity14/Ity14 mice during infection are reported in BioGPS to be expressed in myeloid CD8α⁻ DCs and/or granulocytes, as well as macrophages stimulated with LPS (Clec4a3, Clec4n, Il1b, Oas2, Fcgr4, Sirpb1a) (51). Another subset of the above genes is mainly expressed only in macrophages stimulated with LPS (Tlr2, Pira3, Pira4, Pira11, Myd88, Mmp14, Anxa2, Anxa3, Cstb, Ccl3, Ccl4, Cxcl1). We further characterized different myeloid cell populations in the spleen of infected wild-type and Stat4^Ity14/Ity14 mutants by surface staining markers and flow cytometry. We did not observe any differences in the percentage of splenic macrophages, granulocytes, or DC populations during Salmonella infection (data not shown, Fig. 7A). Of note, we observed significantly greater expression of CD11b (by MFI) in a CD3⁻CD8α⁻CD45R⁻Gr1⁻CD11c^lowCD11b⁺ splenic myeloid DC subset in Salmonella-infected Stat4^Ity14/Ity14 versus Stat4^+/+ mice (Fig. 7B). Expression of CD11b is involved in leukocyte adhesion and migration, as well as complement-mediated opsonization of bacteria. These results provide insight into innate immune regulation in Stat4^Ity14/Ity14 mice and suggest that altered myeloid cell receptor expression mediates pathogen recognition and limits Salmonella replication.

The use of ENU chemical mutagenesis in the mouse has proved an effective means to identify and study mutations in genes that affect the host immunological response to microbial challenge (52–54). We used an ENU mutagenesis approach to screen for recessive germline mutations conferring increased mortality following S. Typhimurium infection, a globally relevant human pathogen (55). In this article, we report the identification of the Ity14 deviant pedigree. A combination of mapping and exome sequencing was used to identify a splice site point mutation in STAT4 (p.G418_E445). The STAT4 mutation was the only homozygous mutation that segregated in all susceptible mice and within the 2.7-Mb mapped interval. This mutation likely interferes with the normal DNA-binding function of STAT4 and in regulating transcription of target genes. Stat4 is expressed at the transcript level in Ity14 mutants, but is a hypomorph with significantly decreased STAT4 expression at the protein level, as shown by Western blot in uninfected and infected splenic tissue, suggesting that the protein is either rapidly degraded or its translation is inhibited. The Ity14 allele is a recessive loss of function of Stat4 that was confirmed by complementation assays wherein Stat4^Ity14/Ity14, Stat4 KO, and compound heterozygous mice present similar survival curves following Salmonella infection.

Intact IFN signaling is essential in the host immune response to viral, bacterial, and fungal disease. The importance of IFN signaling has been highlighted by patients carrying mutations in genes in the IFN-α/β and IFN-γ pathways, which present with increased susceptibility to infection. Individuals with a defect in production or response to IFN-γ have been shown to have increased susceptibility to Mendelian susceptibility to mycobacterial disease, viral disease, or Salmonella infection. These genetic deficiencies have been identified in the following genes: IFNGR1, IFNGR2, STAT1, IL12B, IL12RB1, NEMO, and TYK2 (15, 56, 57). Deficient IL-12 signaling results in impaired Th1 differentiation and IFN-γ production. In fact, multiple episodes of salmonellosis are reportedly frequent in IL-12Rβ1–deficient patients, illustrating the importance of the IL-12/IFN-γ axis in mounting an efficient immune response to primary and secondary infections (17). However, no mutations in STAT4 have been identified in humans to date.

We have previously reported another ENU-induced mouse mutant, Usp18^Ity9, a negative regulator of type 1 IFN signaling, which emphasizes the importance of IFN signaling in susceptibility to Salmonella infection (42, 45). We have shown in this model that transient suppression of STAT4-induced IFN-γ production contributes in part to the pathogenesis of the disease. In USP18 mutants, disease susceptibility was mainly driven by hyperactivation of the type 1 IFN pathway, resulting in increased levels of IL-6 and IFN-β in circulation and development of septic shock. In the current article, we specifically demonstrate the impact of Stat4 in Salmonella disease susceptibility.

Over the past decade, studies in Stat4 KO mice have illustrated the importance of functional STAT4 in both infectious and noninfectious diseases (36, 58, 59). Stat4 KO mice have increased susceptibility to infections primarily driven by a Th1 immune response, including those caused by Mycobacterium tuberculosis, Leishmania major, Trypanosoma cruzi, Toxoplasma gondii, Babesia, and Listeria monocytogenes (36). In contrast, STAT4 deficiency has generally been shown to protect from T cell–mediated autoimmune diseases, including experimental allergic encephalomyelitis, a model for multiple sclerosis and collagen-induced arthritis, and a model for rheumatoid arthritis (60, 61). Furthermore, Stat4 KO mice injected with LPS were shown to be somewhat protected from endotoxemia (62).

STAT4 is a transcription factor and has been shown to bind to the promoter and/or drive expression of several target genes (63–65). Recent genomic approaches by chromatin immunoprecipitation–on–chip and chromatin immunoprecipitation–sequencing have elucidated STAT4 binding sites, target genes, and STAT4-dependent epigenetic modifications on a genome-wide level (65–67). These studies have been done in activated CD4⁺ T cells cultured under Th1 cell conditions. As such, comparisons between our microarray data done in uninfected and day 4 postinfected spleen, and previous studies, must be interpreted with caution. The above potentially explains why we did not detect other known STAT4 targets, including Furin, Il18r1, and Il12rb2, and signature Th1 cell genes, including Tbx21. These discrepancies highlight the difference between in vitro and in vivo studies, and the importance of studying the impact of STAT4 at the whole-organism level.

The transcriptional signature was different in Stat4^Ity14/Ity14 mice compared with Stat4^+/+ following Salmonella infection, which correlates with the histopathological results. As expected, IFN-γ and IFN-γ–regulated gene expressions were downregulated in Ity14 mutants, including a subset of guanylate-binding proteins (Gbp1,2,5,6) previously reported to protect against bacterial infections (68, 69). We also observed impaired upregulation of Stat1, consistent with STAT1 as a target of STAT4. IFN-γ has been reported to have anti-inflammatory properties in suppressing IL-1α,β production by myeloid cells during Mycobacterium tuberculosis infection (70). Consistent with these findings, we detected increased expression of genes in the IL-1 signaling pathway (Il1a, Il1b, Ilrn) in Stat4^Ity14/Ity14 mice that have low levels of circulating IFN-γ.

Stat4^Ity14/Ity14 mice had significantly lower levels of IFN-γ in circulation, which contributed to impaired microabscess formation and increased systemic bacterial burden in the spleen and liver. This finding is consistent with previous data showing increased bacterial burden in STAT4 KO mice compared with littermate controls in spleen and liver following S. Typhimurium infection (42). The following further extends to a model of intestinal disease, wherein the absence of STAT4 leads to bacterial dissemination to the spleen and liver with intestinal disease. It has previously been shown that positive feedback regulation exists between IL-12 and IFN-γ (71). However, excessive amplification of the IL-12/IFN-γ axis can lead to immunopathological changes, and therefore several mechanisms are involved in regulating the IL-12 pathway (72). In fact, increased IFN-γ correlated with lower IL-12p70 serum levels in wild-type mice. In contrast, Ity14 mutants with impaired IFN-γ expression had significantly increased IL-12p70 levels in serum at day 4 following Salmonella infection, consistent with the lack of negative feedback regulation. Previous studies have suggested differing roles of neutrophils, macrophages, T lymphocytes, and NK and NKT cells in contributing to IFN-γ production during primary Salmonella infection (47–49, 73, 74). Recently, Kupz et al. (75) showed that Thy1-expressing precursor and immature NK cells are critical in antibacterial immunity via IFN-γ–dependent control of S. Typhimurium. In addition, NK and NKT cells are reported to be early producers of IFN-γ from healthy human adult blood lymphocytes stimulated ex vivo with NTS contributing to protection from bacteremia (76). In the current study, we demonstrate that a deficiency in NK and NKT cell IFN-γ secretion correlates with early susceptibility to S. Typhimurium infection in Stat4^Ity14/Ity14 mice.

In Stat4^Ity14/Ity14 mice, we observed upregulation of various genes following Salmonella infection involved in pathogen recognition. For example, we observed an increase in expression of Pira3, Pira4, and Pira11 genes. The murine paired Ig-like receptors PIR-A and PIR-B, activating and inhibitory receptors, respectively, are maintained in balance to regulate the host inflammatory response. PirB^−/− mice have been reported to have increased susceptibility to Salmonella infection (77). These mice also have impaired DC maturation, increased Th2 immunity, and upregulation of PIR-A (78). In Stat4^Ity14/Ity14 mutants, upregulation of the PirA group of genes suggests a potential shift in inflammatory response. Indeed, lymphocytes from Stat4 KO mice cultured under Th1 conditions produce higher levels of Th2 cytokines (59). Consistent with these data, Stat4^Ity14/Ity14 mice had increased IL-4 mRNA expression following Salmonella challenge.

Furthermore, we also observed upregulation of innate immune cell receptors like C-type lectins, macrophage scavenger receptors (Marco), TLR signaling (Tlr2), proinflammatory mediators (Il1a, Mmp3, Mmp9, Mmp14), and NF-κB signaling. A proportion of these upregulated genes in Stat4^Ity14/Ity14 mutants clustered together, showing high expression levels in myeloid CD8α⁻ DCs and/or granulocytes, and in LPS-stimulated macrophages. In addition, CD11b is a β2 integrin involved in regulating leukocyte adhesion and migration during inflammation. CD11b is also a receptor of C3bi that coats bacteria to drive complement-mediated opsonization. Increased expression of CD11b on DCs in Stat4^Ity14/Ity14 mice suggests a possible overwhelming and compensatory mechanism in the mutants to facilitate complement-mediated Salmonella opsonization and killing. Overall, the upregulation of expression of immune signaling genes suggests that Stat4^Ity14/Ity14 mutants have adapted alternative mechanisms to counteract the lack of STAT4-induced IFN-γ–mediated immunity. Ultimately, the lack of IFN-γ is detrimental to the host, as Stat4^Ity14/Ity14 mutants develop fatal bacteremia following Salmonella infection.

We thank Geneviève Perreault, Nadia Prud'homme, Vanessa Guay, Marie Chevenon, Lei Zhu, and Line Larivière for technical assistance, and Sean Wiltshire and Sean Beatty for help with R code.

The authors have no financial conflicts of interest.