The Citrobacter rodentium model mimics the pathogenesis of infectious colitis and requires sequential contributions from different immune cell populations, including innate lymphoid cells (ILCs) and CD4+ lymphocytes. In this study, we addressed the role of STAT3 activation in CD4+ cells during host defense in mice against C. rodentium. In mice with defective STAT3 in CD4+ cells (Stat3ΔCD4), the course of infection was unchanged during the innate lymphoid cell–dependent early phase, but significantly altered during the lymphocyte-dependent later phase. Stat3ΔCD4 mice exhibited intestinal epithelial barrier defects, including downregulation of antimicrobial peptides, increased systemic distribution of bacteria, and prolonged reduction in the overall burden of C. rodentium infection. Immunomonitoring of lamina propria cells revealed loss of virtually all IL-22–producing CD4+ lymphocytes, suggesting that STAT3 activation was required for IL-22 production not only in Th17 cells, but also in Th22 cells. Notably, the defective host defense against C. rodentium in Stat3∆CD4 mice could be fully restored by specific overexpression of IL-22 through a minicircle vector–based technology. Moreover, expression of a constitutive active STAT3 in CD4+ cells shaped strong intestinal epithelial barrier function in vitro and in vivo through IL-22, and it promoted protection from enteropathogenic bacteria. Thus, our work indicates a critical role of STAT3 activation in Th17 and Th22 cells for control of the IL-22–mediated host defense, and strategies expanding STAT3-activated CD4+ lymphocytes may be considered as future therapeutic options for improving intestinal barrier function in infectious colitis.
Intestinal inflammation caused by pathogenic bacteria is a frequent and potentially life-threatening disease. Besides the fitness and diversity in the host’s immune system, there are several influences that can modulate the colonization of pathogens and the course of disease, including bacterial virulence factors and competition with the resident gut flora (1). Moreover, compromised intestinal barrier function can render individuals susceptible to intestinal inflammation and infection (2, 3). Antimicrobial peptides are typically produced by crypt-residing Paneth cells in the small intestine (3, 4). Upon challenge, however, intestinal epithelial cells (IECs) of the large intestine can also produce high amounts of antimicrobial peptides (4, 5).
Genome-wide association studies have connected the transcription factor STAT3, which is widely expressed by different tissues and cell types, to intestinal pathology (6). In response to extracellular stimuli, such as IL-22, IL-6, or IL-23, STAT3 molecules are phosphorylated, translocate to the nucleus, and can influence the transcription of target genes related to cellular key processes, including proliferation, cell survival, activation, and differentiation (7). STAT3 has cell type–specific functions that are regulated in a context-dependent manner. On one hand, conditional deletion of STAT3 activity in macrophages and neutrophils results in spontaneous enterocolitis, and defective STAT3 activation in IECs impairs barrier function and mucosal healing (8, 9). On the other hand, T cell–specific inactivation of STAT3 protects from intestinal inflammation in a transfer model of colitis (10). Additionally, activation of STAT3 is critically linked to the development of Th17 cells, which can promote intestinal pathology (11, 12). Conversely, dominant negative mutations of STAT3 are causative for the hyper-IgE syndrome, which is characterized by Th17 deficiency and impaired defense against extracellular bacteria (13, 14).
Infection of mice with Citrobacter rodentium is a well-established disease model mimicking infectious colitis by enterohemorrhagic or enteropathogenic Escherichia coli in humans. All of these pathogens use attaching and effacing lesions to colonize the gastrointestinal tract of the host and share the type III secretion system (15).
In wild-type mice, infection with C. rodentium is mostly self-limiting and the bacterial levels in the colon peak around days 8–11 with variations depending on the mouse strain and the resident flora (15). The host defense against C. rodentium comprises sequential contributions from different immune cell populations, including a variety of type 3 ILC (ILC3) subsets, dendritic cells, and adaptive immune cells, including T and B lymphocytes (15–20). Several studies have suggested an important role for Th17 cells during the course of C. rodentium infection (21–23).
Although CD4 expression is frequently found in lymphocytes, other immune cell subsets such as dendritic cells or ILC3s can also express CD4. Indeed, CD4+ ILC3s are crucial mediators in host defense upon infection with enteropathogenic bacteria (24). Strikingly, several features critically important during the host defense against C. rodentium such as expression of IL-17, IL-22, IL-23R, AHR, and CD4 are shared by both Th17 cells and ILC3 subsets (5, 16, 20, 21, 24–26). Thus, studies with mice deficient in such molecules are restricted in the analysis of differential contributions by ILC3s or Th17 cells. However, ILC3s act rather early upon C. rodentium infection (16, 24, 25), whereas major contributions of T cells occur with delayed kinetics (15, 27). For the analysis of differential contributions of Th17 cells, it would be interesting to study gene-modified mice with intact ILC3 responses, but specific lack of Th17 cells.
In this study, we have analyzed the role of STAT3 in CD4+ cells during the course of C. rodentium infection, revealing a critical contribution of STAT3 activation in Th17 and Th22 cells for control of the IL-22–mediated host defense against enteropathogenic bacteria in vivo.
Materials and Methods
Stat3fl/fl and Stat3∆CD4 mice were previously described (28). R26Stat3Cstopfl/fl mice were already published (29). In brief, they were generated by transfecting Bruce4 embryonic stem cells with a vector targeting Stat3-C cDNA and a 5′ loxP-flanked NeoR stop cassette into the Rosa26 gene. The modified Rosa26 vector contained FLAG-tagged Stat3-C cDNA together with a 3′ FRT-flanked IRES-EGFP cloned downstream of the floxed NeoR stop cassette and was fully sequenced and linearized before being electroporated into the C57BL/6-derived embryonic stem cells. For CD4-specific constitutive activation of STAT3, R26Stat3Cstopfl/fl mice were crossed with CD4Cre mice. The analysis included animals with one or two alleles expressing transgenic Stat3-C, as specified in the figure legends.
All mice were bred and maintained in individually ventilated cages. Experiments were performed with cohoused littermates. Animal care was within Institutional Care Committee guidelines, and all animal experiments were performed in accordance with protocols approved by the governments of Rhineland-Palatinate and Middle Franconia, Germany.
C. rodentium infection and bioluminescence imaging
For our study, the bioluminescent strain ICC 169 of C. rodentium was used (30). Inoculation was performed via oral gavage of 8 × 109 CFU in a total volume of 200 μl PBS. Mice were starved for 8 h prior to infection. Body weights were measured daily and the bacterial burden per mouse and time point were quantified by bioluminescence imaging with an IVIS Lumina II system and Living Image Software v3.0 (Xenogen). Per measurement, bioluminescent counts were recorded for 3 min from the abdomen as the region of interest.
Isolation of lamina propria cells
Colons were washed with ice-cold PBS, cut longitudinally, and separated into ∼1-cm-long pieces. The epithelium was removed by incubating the tissue for 20 min at 37°C and gentle shaking twice in Ca2+- and Mg2+-free HBSS containing 5 mM EDTA (Carl Roth), 5% FBS (PAA Laboratories), and 1 mM DTT (Carl Roth) and once in Ca2+- and Mg2+-free HBSS containing 10 mM HEPES (PAA Laboratories). The remaining tissue was further digested with a lamina propria dissociation kit (Miltenyi Biotec) according to the manufacturer’s instructions.
In vitro stimulation of CD4+ splenocytes
CD4+ T cells were freshly isolated from spleens with MACS isolation kits (Miltenyi Biotec). For polyclonal in vitro stimulation, hamster anti-mouse CD3 Ab (clone 145-2C11)–coated wells (10 μg/ml in PBS for 3 h at 37°C) were used together with soluble hamster anti-mouse CD28 (2 μg/ml, clone 37.51). For T cell polarizations, the following supplements were used: rat anti-mouse IFN-γ (4 μg/ml, clone XMG1.2), rat anti-mouse IL-4 (4 μg/ml, clone 11B11), TGF-β (5 ng/ml, R&D Systems), IL-6 (50 ng/ml, ImmunoTools), and IL-23 (30 ng/ml, R&D Systems). TGF-β inhibitor SB431542 (10 nmol/ml) was purchased from Sigma-Aldrich; all Abs were from Bio X Cell.
Organoids were cultured as described previously (31). In short, crypts were isolated from the small intestine with ice-cold PBS containing 2 mM EDTA. The crypts were suspended in Matrigel (BD Biosciences), seeded onto a culture plate, and grown in advanced DMEM/F12 medium (Invitrogen) containing 2 mM GlutaMAX (Invitrogen), N2 supplement (Invitrogen), B27 supplement (Invitrogen), 1 mM N-acetylcysteine (Sigma-Aldrich), 10 mM HEPES, 100 U/ml penicillin/streptomycin (Life Technologies), 50 ng/ml murine epidermal growth factor (ImmunoTools), 1 μg/ml recombinant human R-spondin (R&D Systems), and 100 ng/ml recombinant murine noggin (PeproTech). Organoids were cultured for 7–10 d before stimulation. Medium and Matrigel were changed every third day. For stimulation, medium was replaced for 12 h with supernatant from polarized CD4+ T cells.
Construction of in vivo overexpression vectors
Expression constructs for sustained in vivo overexpression of IL-17A, IL-22, IFN-γ, and IL-4 were generated by cloning cDNA fragments encoding for murine full-length cDNAs as described (32). DNA was isolated with Qiagen plasmid maxi kits including endotoxin removal. To further ensure efficient endotoxin removal, DNA was treated with a MiraClean endotoxin removal kit (Mirus Bio, Madison, WI). Five to 10 μg constructs was administered in Krebs–Ringer solution to mice via hydrodynamic tail vein injection as described previously (33).
Histopathological analyses and immunohistochemical stainings
Histopathological analyses were performed on colon tissue that was either instantly frozen in liquid nitrogen and embedded in OCT compound (Sakura Finetek) or fixed in 4.5% formaldehyde (Carl Roth) and embedded in paraffin. Cross-sections were stained with H&E. Crypt lengths were measured with ImageJ software.
Immunostainings of cryosections were performed using the TSA Cy3 system (PerkinElmer). Primary Abs were hamster anti-mouse CD11c (clone HL3, BD Biosciences), rat anti-mouse CD4 (clone RM4-5, BD Biosciences), rabbit anti-mouse MPO (polyclonal, Abcam), rat anti-mouse F4/80 (clone BM8, eBioscience), rat anti-mouse B220 (clone RA3-6B2, BioLegend), and rabbit anti-mouse p-STAT3 (clone D3A7, Cell Signaling Technology). Nuclei were counterstained with DAPI (Life Technologies), and fluorescence analysis was performed with DMI 4000B (Leica) or TCS SP5 (Leica). For quantitative analyses, four to five representative sections (10×) were evaluated in a blinded fashion.
Flow cytometry analyses
For FACS analyses, the following directly conjugated Abs were used: rat anti-mouse CD4 (clone GK1.5), rat anti-mouse CD8 (clone 53-6.7), rat anti-mouse CD3e (clone 17A2), rat anti-mouse CD62L (clone MEL14-H2.100), rat anti-mouse CD44 (clone IM7), rat anti-mouse IL-17A (clone TC11-18H10), goat anti-mouse IL-22 (Poly5146), rat anti-mouse IFN-γ (clone XMG1.2), rat anti-mouse Ly6G (clone 1A8), rat anti-mouse Ly6C (clone HK1.4), rat anti-mouse CD11b (clone M1/70), rat anti-mouse F4/80 (clone BM8), rat anti-mouse B220 (clone RA3-6B2), hamster anti-mouse CD11c (clone HL3), rat anti-retinoic acid–related orphan receptor (ROR)γt (clone B2D), and mouse anti T-bet (eBIO4B10). All FACS Abs were purchased from BioLegend except for anti-CD8a and CD62L, which were from Miltenyi Biotec, and anti-CD11c, which was from BD Biosciences. Cells were stimulated with 50 ng/ml PMA (Merck), 500 ng/ml ionomycin (Sigma-Aldrich), and 1 μg/ml brefeldin A (Sigma-Aldrich) at 37°C for 4 h prior to intracellular cytokine staining with a Fix/Perm cell permeabilization kit (An der Grub) according to the manufacturer’s instructions. For FACS studies that included RORγt or T-bet, a staining set for transcriptions factors from eBioscience was used. Cells were analyzed on a LSR Fortessa (BD Biosciences) and data analysis was performed with FlowJo v7.6.5 (Tree Star).
RNA isolation and quantitative real-time PCR
Total RNA was isolated from the distal colon with the NucleoSpin RNA kit (Macherey Nagel) or from organoids with the RNeasy micro kit (Qiagen) and quantified and quality checked with a Nanodrop ND-1000 (Thermo Scientific). cDNA was subsequently synthesized with the iScript cDNA synthesis kit (Bio-Rad). Real-time quantitative PCR (qPCR) was performed with SsoFast EvaGreen supermix (Bio-Rad) according to the manufacturer’s instructions on a CFX96 thermal cycler (Bio-Rad). Each sample was run in duplicates and GAPDH was used as reference gene. Data were analyzed with the CFX Manager v2.1 (Bio-Rad) using the ΔΔCT method.
Analyses of organ lysates
Organs were aseptically removed at the peak of the infection and homogenized in 200 μl sterile PBS. Serial dilutions were plated on Luria–Bertani agar supplemented with 500 μg/ml erythromycin, and bioluminescent colonies were counted after 1 d in culture at 37°C.
Quantitative data are shown as mean values, and error bars represent SDs. Unless otherwise specified, significance analyses were performed using a Student t test, and p values <0.05 or <0.01 were considered significant or highly significant, respectively.
Stat3ΔCD4 mice are compromised in limiting infection with C. rodentium
The expression of IL-17 in Th17 cells requires STAT3 (12), whereas IL-17 production from ILC3s was reported to occur partly independent of STAT3 (34), which prompted us to study STAT3 activation in the lamina propria upon infection with enteropathogenic bacteria that are known to be controlled by both ILC3s and Th17 cells.
Therefore, we analyzed colonic cross-sections from the C. rodentium model by immunofluorescence for p-STAT3. In uninfected animals, few cells stained positive for active STAT3 in the colon of wild-type mice (Fig. 1A, 1B). Around days 6–8 after bacterial challenge with C. rodentium, we observed a strong infiltration with lymphocytes, and the number of cells expressing p-STAT3 highly increased in the colonic lamina propria of wild-type mice (Fig. 1A, 1B). Importantly, a large number of CD4+ cells stained also positive for p-STAT3 (Fig. 1C, upper panel). Of note, we observed only a minor increase in p-STAT3+ mononuclear cells during the early effector phase (days 1–5) upon infection with C. rodentium (data not shown). To test the functional relevance of STAT3 activation in CD4+ cells in this model, we used conditional knockout mice defective for STAT3 activation in CD4+ cells (Stat3∆CD4). In contrast to CD4+ cells from Stat3fl/fl controls, there were no CD4+ cells in Stat3∆CD4 mice that stained positive for p-STAT3 (Fig. 1C, lower panel). Upon infection with bioluminescent C. rodentium, both Stat3∆CD4 and Stat3fl/fl littermates showed bioluminescent signals increasing with similar kinetics and peaking at days 8–11 (Fig. 1D, 1E). Of note, the bioluminescence top levels were similar in both groups. Importantly, however, whereas Stat3fl/fl mice displayed a rapid decrease in bioluminescent bacterial counts after the peak of infection (Fig. 1D, 1E), Stat3∆CD4 mice did not manage to fight the enteropathogenic bacteria efficiently. In detail, high bioluminescent signals could still be detected in Stat3∆CD4 animals at day 16, that is, at time points when the bacterial counts in controls had already decreased to background level (Fig. 1D, 1E). Interestingly, the great majority of Stat3∆CD4 animals survived and reached a bioluminescent intensity at the threshold level around day 30 (data not shown).
Of note, when Stat3∆CD4 mice on a Rag1-deficient background (Stat3∆CD4;Rag1−/−) and their littermate controls (Stat3fl/fl;Rag1−/−) were compared with each other, we observed similar bioluminescent signals throughout the course of infection, providing additional evidence that STAT3 activation is critically important in adaptive immune cells (data not shown).
Thus, defective STAT3 activation in CD4+ cells does not significantly alter the early phase of infection with C. rodentium, but is required for efficient limiting of bacteria during the lymphocyte-dependent phase.
Loss of STAT3 activation in CD4+ cells does not affect leukocyte infiltration to the colon
Histopathological analysis at the top of infection (days 8–10) demonstrated crypt hyperplasia in both Stat3∆CD4 and Stat3fl/fl control mice. Although we did not detect a significant change, the hyperplasia appeared to be slightly less prominent in Stat3∆CD4mice as compared with Stat3fl/fl littermates (Fig. 2A, 2B). Additionally, comparison of colon cross-sections did not reveal major changes regarding the recruitment and infiltration of immune cells as assessed by markers for neutrophils, macrophages, dendritic cells, and B cells by immunofluorescence and flow cytometry (Fig. 2C–F). Moreover, the presence of CD4+ cells was not significantly changed in Stat3∆CD4 animals, as demonstrated by immunofluorescence and FACS, suggesting altered effector functions rather than differences in cell recruitment, survival, or proliferation in vivo (Fig. 2C–F). Crypt hyperplasia and immune cell infiltration was also similar between Stat3∆CD4and control mice at days 15–16 upon infection with C. rodentium (data not shown).
Hence, the loss of STAT3 activation in CD4+ cells did not significantly affect the presence of leukocytes in the colon upon infection with C. rodentium.
Defective STAT3 activation in CD4+ cells results in nearly complete loss of IL-22 production by CD3+CD4+ lamina propria cells
Taking into account previous work showing high upregulation of IL-22 at the peak of infection in wild-type mice (27), we studied the cytokine production by lamina propria cells from colons at that time. Interestingly, FACS analysis of those lamina propria cells revealed that in Stat3fl/fl control mice, the large majority of IL-22–producing cells were CD4+, suggesting that CD4+ cells are the major source of IL-22 at the peak of C. rodentium infection (Fig. 3A). In contrast, IL-22 was strongly reduced in CD4+ cells of Stat3∆CD4 mice (Fig. 3A). Similar data were obtained for the production of IL-17a, whereas IFN-γ production was increased in CD4+ lamina propria cells of Stat3∆CD4 mice (Fig. 3A). To exclude a contribution of CD8+ T cells, which also lack STAT3 activation in Stat3∆CD4 mice, we tested the IL-22 production by CD8+ T cells, which was very low in wild-type mice at that time (data not shown).
To further differentiate the relative contribution of CD3+CD4+ T cells versus CD3−CD4+ cells, we studied the coexpression of CD3 with IL-17a or IL-22 in CD4+ gated lamina propria cells. In this study, we could show that >90% of IL-17–producing CD4+ cells and >80% of IL-22–producing CD4+ cells also expressed CD3, demonstrating T lymphocytes as the dominant source of IL-17 and IL-22 in wild-type mice at the peak of infection (Fig. 3B). In accordance with STAT3 activation mandatorily needed for Th17 development, IL-17 production was strongly decreased in CD3+CD4+ cells of Stat3∆CD4 mice (Fig. 3B). Importantly, IL-22 production was also fully abrogated in CD3+CD4+ lymphocytes of Stat3∆CD4 mice, suggesting that IL-22 production was highly dependent on STAT3 activation not only in Th17 cells, but also in Th22 cells (Fig. 3B). In contrast to IL-22 production from CD3+CD4+ lymphocytes, including both Th22 and Th17 cells, IL-22 production from CD3−CD4+ cells was diminished in a small proportion of cells only (Fig. 3B).
Notably, we observed that more than 75% of the CD3+CD4+IL-22+ cells stained positive for both IL-22 and IL-17, suggesting that Th17 cells (IL-22+IL-17+), but not Th22 (IL-22+IL-17−) cells, were the dominant source of IL-22 production at the peak of infection with C. rodentium in our setup (Fig. 3C). Similar results were obtained for the population of intraepithelial lymphocytes, in which CD3+CD4+IL-17+IL-22+ cells were also the dominant subset expressing IL-22 (data not shown).
Next, we addressed the expression of the transcription factors RORγt and T-bet in our setup by FACS. In this study we could show that the vast majority (>90%) of IL-22–producing CD4+CD3+ lamina propria cells also expressed RORγt in Stat3fl/fl control mice (Fig. 3D, upper panel). In contrast, CD4+CD3+ RORγt+IL-22+ cells were not detectable and CD4+CD3+RORγt+IL-22− cells were strongly diminished in Stat3∆CD4 mice (Fig. 3D, upper panel).
Interestingly, most of the IL-22–producing CD4+CD3+ cells (>75%) from control mice also expressed the transcription factor T-bet (Fig. 3D, lower panel). In Stat3∆CD4 mice, the number of T-bet+CD4+CD3+ lamina propria cells was increased (i.e., 82–90.3%), which is in accordance with results demonstrating elevated IFN-γ production in Stat3∆CD4 mice (Fig. 3B–D). Further FACS analysis revealed that most CD4+CD3+RORγt+IL-22+ cells produced IL-17 (Th17 cells), but we also detected a substantial population of CD4+CD3+RORγt+IL-22+ cells (>25%) that did not show IL-17 production (Th22 cells, Fig. 3E, upper panel). By studying the expression patterns of CD4+CD3+T-bet+ lamina propria cells, we observed some overlap of IL-22 production with IFN-γ expression, but most of the cells stained positive for either IL-22 or IFN-γ (Fig. 3E, lower panel).
Of note, we did not detect any major changes in Citrobacter-specific or total Ig production between Stat3∆CD4 mice and Stat3fl/fl controls, as evaluated in the serum and feces (data not shown).
In summary, Stat3∆CD4 mice show a nearly complete loss of IL-22 production by CD3+CD4+ lamina propria cells during infectious colitis.
STAT3 activation in CD4+ splenocytes controls IL-22 production by Th17 and Th22 cells
Stat3∆CD4 mice showed compromised reduction in the overall burden of C. rodentium and loss of IL-22 production from CD4+ lamina propria lymphocytes. Similarly, IL-22 production was also strongly diminished upon in vitro polarization of CD4+ splenocytes from unchallenged mice, providing additional evidence that STAT3 activation is required for IL-22 production in CD4+ T cells (Fig. 4A).
To further evaluate the role of STAT3 for IL-22 expression in T cells, we intended to analyze how elevated STAT3 activation in CD4+ T cells may affect cytokine production. Therefore, conditional transgenic mice were used expressing constitutively active STAT3 protein (STAT3C) in CD4+ cells (heterozygous R26Stat3Cstopwt/fl CD4Cre and homozygous R26Stat3Cstopfl/fl CD4Cre) (29). STAT3C protein is a STAT3 homodimer formed by disulfide bonds between two cysteine substitutions, leading to activation-independent tyrosine phosphorylation. To test the influence of STAT3C expression on cytokine production in Th17 cells, we performed Th17 polarization with freshly purified CD4+ splenocytes. Strikingly, the number of IL-22–expressing cells was moderately increased in cultures from heterozygous R26Stat3Cstopwt/fl CD4Cre mice and strongly increased in cultures from homozygous R26Stat3Cstopfl/fl CD4Cre mice as compared with R26Stat3Cstopfl/fl littermate controls, suggesting a direct link between STAT3 activation and IL-22 expression (Fig. 4B). Analog expression differences were observed for IL-17 (Fig. 4B). In contrast, the production of IFN-γ was similar in all groups, indicating a selective regulation of IL-22 and IL-17 by STAT3C (Fig. 4B). Upon Th17 polarization, the upregulation of IL-22 production by STAT3C-bearing cells was more prominent for IL-22+IL-17+ cells (i.e., Th17 cells) than for IL-22+IL-17− cells (i.e., Th22 cells) (Fig. 3C). However, the selective inhibition of TGF-β signaling resulted in a strong decrease of Th17 cells, whereas the number of Th22 cells was not compromised but even increased, indicating that STAT3 activation can promote IL-22 production in CD4+ T cells independent of TGF-β signaling (Fig. 4C).
Thus, our data suggest that STAT3 activation controls IL-22 production in both Th17 and Th22 cells.
Defective STAT3 activation in CD4+ cells causes intestinal epithelial barrier defects and facilitates systemic distribution of bacteria
Based on our observations hitherto and previous reports showing that both IL-22 and IL-17 can influence the expression of antimicrobial peptides by IECs (5, 9, 22), we hypothesized that defective STAT3 activation in CD4+ cells may affect intestinal barrier function in vivo. Strikingly, Stat3∆CD4 animals showed signs of elevated bacterial invasion and systemic distribution as demonstrated by bacterial cultures of lysates from the liver and the spleen in infectious colitis. Colonies of C. rodentium could rarely be detected in such lysates from Stat3fl/fl control mice at days 8–10 (Fig. 5A, 5B), that is, at a time when the colonization of C. rodentium was similar in Stat3∆CD4 mice and Stat3fl/fl controls. However, Stat3∆CD4 mice were susceptible to systemic bacterial distribution and showed significantly higher numbers of C. rodentium colonies, indicating that Stat3∆CD4 mice were more prone to systemic dissemination despite equivalent C. rodentium burden in the colon (Fig. 5A, 5B). Similar results with increased dissemination to the liver and spleen in Stat3∆CD4 mice were obtained at days 15–16.
Moreover, the production of some IEC-derived antimicrobial peptides, including defensin β1, regenerating islet-derived 3 β and γ (RegIIIβ, RegIIIγ), and calgranulin A (S100A8), was reduced in the colon of Stat3∆CD4 mice at day 8 upon infection with C. rodentium (Fig. 5C). In contrast, other molecules related to IEC-mediated barrier function such as mucin 2 and 13 (Muc2, Muc13), serum amyloid 1, or trefoil factor 3 did not show major changes (Fig. 5C). At days 15–16, when the overall burden of C. rodentium was strongly decreased in control mice already, antimicrobial peptides were low in control mice as well (data not shown).
Thus, our data suggest that defective STAT3 activation in CD4+ lymphocytes causes intestinal epithelial barrier defects and renders individuals susceptible to systemic bacterial distribution during infectious colitis.
Systemic overexpression of IL-22 restores the host defense against C. rodentium in Stat3∆CD4 mice
To further address the differential role of Th cell–derived cytokines on the intestinal epithelium, we used a recently established minicircle vector-based method in combination with hydrodynamic tail vein injection for the systemic overexpression of cytokines (32). Vectors were constructed for IL-17A, IL-17F, IFN-γ, and IL-4, and one injection of 10 μg vector DNA resulted in high expression (>5 ng/ml) of the respective cytokine as measured in the blood (data not shown). Next, we tested the expression of IEC-derived peptides in the distal colon tissue upon systemic expression of those cytokines in uninfected mice. In this study, we could directly demonstrate that delivery of IL-22, but not IL-17A, IL-17F, IFN-γ, or IL-4, caused high expression of antimicrobial peptides, including RegIIIβ and RegIIIγ at day 3 upon vector injection, as analyzed by qPCR (Fig. 6A). On the contrary, other genes related to epithelial defense mechanisms in IECs, including S100A8, Muc2, defensin β1, or serum amyloid 1, remained without major changes (Fig. 6A and data not shown).
To investigate whether IL-22 delivery might influence the phenotype of Stat3∆CD4 mice, IL-22 vector was applied to Stat3∆CD4 mice at day 10 upon C. rodentium infection. Strikingly, Stat3∆CD4 mice overexpressing IL-22 could clear the bacteria with similar kinetics as Stat3fl/fl control mice, indicating that reintroduction of IL-22 can rescue the compromised host defense of Stat3∆CD4 mice (Fig. 6B, 6C). In contrast, systemic overexpression of IL-17A and IL-17F failed to rescue the phenotype of Stat3∆CD4 mice regarding C. rodentium infection (data not shown). In line with our observations in uninfected mice at day 3, we detected increased expression of RegIIIβ and RegIIIγ in mice infected with C. rodentium and overexpressing IL-22 (Fig. 6D).
Thus, our data provide evidence that STAT3 activation in CD4+ lymphocyte subsets can shape strong epithelial defense mechanisms through IL-22 production and induction of antimicrobial peptides in IECs.
Constitutive activation of STAT3 in CD4+ cells causes high induction of antimicrobial peptides in IECs in vitro and in vivo and may offer a therapeutic approach for infectious colitis
To further evaluate the role of STAT3 in T cells for the cross-talk to IECs, intestinal epithelial organoids were grown from wild-type mice as previously described (31) and subsequently cultured in the presence of conditioned medium derived from supernatants of CD4+ splenocytes from R26Stat3Cstopfl/fl CD4Cre or control mice polarized under Th17 or Th22 conditions (Fig. 7A). Strikingly, conditioned medium from R26Stat3Cstopfl/fl CD4Cre mice, which contained elevated levels of IL-22, resulted in high expression of RegIIIβ and RegIIIγ, but not of Muc2 or S100A8, by intestinal organoids, providing direct evidence that activation of STAT3 in CD4+ T cells controls the production of some antimicrobial peptides through cross-talk to IECs (Fig. 7B, 7C, and data not shown). In line with our previous results (Figs. 5, 6), the induction of RegIIIβ and RegIIIγ by conditioned medium was almost completely abrogated by neutralizing anti–IL-22 Ab, indicating that cross-talk from STAT3C-expressing CD4+ T cells to IECs in vitro was mainly mediated through IL-22 production (Fig. 7D).
Next, we hypothesized that an increase of active STAT3 in CD4+ cells may be useful as a therapeutic approach in infections with enteropathogenic bacteria. Consistent with our hypothesis, we observed that primary IECs purified from the colons of 6-wk-old R26Stat3Cstopwt/fl CD4Cre mice and R26Stat3Cstopfl/fl CD4Cre mice showed very high production of antimicrobial peptides even under unchallenged conditions (Fig. 7E). To further study the in vivo relevance and therapeutic potential for infections with enteropathogenic bacteria, we tested the course of C. rodentium infection in mice expressing STAT3C. Remarkably, R26Stat3Cstopwt/fl CD4Cre mice were protected and cleared the infection faster than did wild-type controls (Fig. 7F, 7G), suggesting that the specific elevation of active STAT3 in CD4+ cells may serve as a therapeutic strategy in infections with enteropathogenic bacteria.
Interestingly, some differences in the expression of antimicrobial peptides by IECs were still detectable at day 8 upon infection with C. rodentium (Fig. 7H). Additionally, the analysis of colon tissue lysates from mice infected with C. rodentium confirmed higher protein levels for IL-22 and antimicrobial peptides in R26Stat3Cstopwt/fl CD4Cre mice compared with controls at day 8, providing further support for the therapeutic approach (Fig. 7I).
Of note, R26Stat3Cstopfl/fl CD4Cre mice could not be analyzed in the C. rodentium model, as they suffer from chronic airway inflammation and have a shortened lifespan with a mean survival of <40 d (29). Therefore, although STAT3 activation in CD4+ cells may act as a promising therapeutic strategy, the consequences of its modulation need to be carefully controlled.
In summary, our data suggest that the expression of STAT3C in CD4+ cells increases the production of antimicrobial peptides in IECs through IL-22 and accelerates the reduction in the overall burden of enteropathogenic bacteria. Thus, STAT3 activation in CD4+ lymphocytes may offer a future therapeutic option for promoting cross-talk to IECs and improving intestinal barrier function in infectious colitis.
Finally, we propose a model for the role of STAT3 activation in CD4+ cells during infection with enteropathogenic bacteria such as C. rodentium (Fig. 8). In individuals with intact STAT3 phosphorylation, such infections can be successfully counterbalanced through efficient cross-talk to IECs, which includes high production of IL-22 by CD4+ lamina propria lymphocytes shaping strong epithelial defense mechanisms in the colon. In individuals that cannot activate STAT3 in CD4+ cells, the early phase of infection is not altered because subsets of ILC3 and possibly other CD4− cells could provide substantial amounts of IL-22. Additionally, IL-22 production by CD3−CD4+ cells might partly occur via STAT3-independent mechanisms. At the peak of infection, however, individuals that cannot activate STAT3 in CD4+ cells fully lack their major source of IL-22, that is, CD4+ lymphocytes (Th17 and Th22 cells), resulting in increased bacterial invasion and delayed reduction of bacterial burden.
Hence, individuals with defective STAT3 in CD4+ cells display impaired intestinal epithelial barrier function, and protection from enteropathogenic bacteria is severely compromised during the T cell–dependent phase of infection.
Our work analyzed the specific contribution of STAT3 activation in CD4+ cells to host defense upon infection with enteropathogenic bacteria. We found that, compared with littermate controls, Stat3∆CD4 mice show a similar course of infection during the early phase, which is known to require critical contributions by ILC3s, including CD4+ lymphoid tissue inducer cells and NK22 cells (16, 17, 20, 24). From the peak of infection on, however, Stat3∆CD4 mice are heavily compromised in reducing the burden of C. rodentium. Moreover, Stat3∆CD4 animals were more susceptible to bacterial distribution of C. rodentium to distant organs such as liver and spleen. Our study was performed with cohoused mice, indicating that the differences were stably related to genotypes and not driven by a dominant transmissible flora.
In accordance with a mandatory role of STAT3 for Th17 cell development (12), Stat3∆CD4 mice are also lacking nearly all Th17 cells in the C. rodentium model. Hence, our observations are consistent with previous studies suggesting Th17 cells to be important in the C. rodentium model (21, 23). However, previous studies were often restricted in addressing the specific role of Th17 cells in vivo, as they included mice deficient in genes that are already critically important for the development or effector functions of ILC3 subsets such as IL-22, IL-23R, or AHR (5, 16, 21, 25). In contrast, our investigations using conditional knockout mice allowed for more specific investigations into the role of Th17 and Th22 cells at the peak of infection.
It was demonstrated that STAT3 activation correlates with IL-22 transcript expression in CD4+ T cells (12, 35). Strikingly, our data strongly suggest that STAT3 activation is mandatorily needed for IL-22 production from CD3+CD4+ lamina propria cells in the C. rodentium model. In line with previous reports showing that STAT3 is required for Th17 cell development and early Th17 lineage differentiation (12, 36), our findings provide evidence that the lack of active STAT3 not only interferes with the expression of IL-22, but also with the population of CD4+CD3+RORγt+IL-17+IL-22+ cells (Th17 cells). Additionally, our data might indicate that STAT3 also interferes with the population of CD4+CD3+IL-17−IL-22+ cells, which are usually referred to as Th22 cells.
It remains unresolved whether Th22 cells represent a truly unique lineage equivalent to other types of Th cells or rather a different activation state. Our analyses of primary lymphocytes from conditional STAT3-deficient mice and novel mice expressing constitutively active STAT3 under control of the CD4 promoter are consistent with the concept that Th22 cells, in contrast to Th17 cells, develop independently of TGF-β signaling (27, 37). Additionally, a reciprocal need for RORγt and T-bet was recently suggested for Th17 cells and Th22 cells, respectively (27). However, whereas CD3+CD4+ lamina propria cells from Tbx-21−/− (encoding for T-bet) mice showed a diminished, but still clearly detectable, population of IL-22–producing CD4+ T cells (27), our work could demonstrate that nearly all Th22 cells were absent in Stat3∆CD4 mice, suggesting an even more important role for STAT3 than for T-bet. On the basis of these findings and recent work analyzing regulatory networks for Th17 specification (36), it may be interesting to learn how other transcription factors that also guide early Th17 lineage differentiation such as BATF and IFN regulatory factor (IRF) 4 influence the development of Th22 cells. Both BATF and IRF4 were reported being similarly expressed in Th17 cells and Th22 cells (27). Binding of BATF to the IL-22 promoter was shown, suggesting modulation of IL-22 expression independent of promoting Th17 lineage specification (38). Interestingly, IRF4 deficiency was recently correlated with high IL-22 expression (39), suggesting divergent roles for IRF4 and STAT3.
In our experimental setup, IL-17+IL-22+CD3+CD4+ lymphocytes represented the most frequent IL-22–producing population at the peak of infection in the lamina propria, which differs from a recent study arguing for Th22 cells as the predominant source of IL-22 (27). In contrast to Basu et al. (27), our ex vivo FACS analyses of lamina propria cells were performed without preceding IL-23 restimulation. Additionally, differences may be caused by the resident gut microbiota, which can influence the course of C. rodentium infection and the level of Th17 cells (23, 40).
Although this study did not directly address the upstream signaling responsible for STAT3 activation in Th17 and Th22 cells in vivo, IL-6 seems to be a likely candidate molecule. It has been previously linked to the production of IL-22 (41), development of Th22 cells (42), and host defense upon C. rodentium infection (21, 43). Other cytokines that are also highly expressed at the top of infection, including IL-23, may additionally contribute to STAT3 activation.
Our data demonstrate that Stat3∆CD4 mice compared with cohoused littermate controls show a similar course of infection during the early phase upon inoculation with C. rodentium. Whereas our work demonstrated a critical role for STAT3 in CD4 T cells for efficient reduction in the overall burden of C. rodentium from the peak of infection on, a very recent study by Guo et al. (44) reported that STAT3 activation in ILC3s is very important for protection during the early phase of infectious colitis. Notably, ILC3s comprise various subsets, and it appears possible that subpopulations of ILC3 might not require STAT3 activation for IL-22 production. In line with that hypothesis, our molecular analyses could suggest that CD3−CD4+ cells might partly produce IL-22 independent of STAT3. However, the efficiency of the CD4-driven Cre recombinase in ILCs was challenged (45). Nevertheless, differential requirements for the transcription factor IRF4 were recently reported between Th17 cells and ILC3s for the production of both IL-22 and IL-17 (46). Moreover, previous work demonstrated that IL-17 production in ILC3s can occur in both a STAT3-dependent and STAT3-independent manner (34).
Collectively, our observations are in line with sequential waves of IL-22 expression produced by different immune cell populations upon infection with enteropathogenic bacteria (27). Our work is consistent with studies demonstrating that IL-22–producing ILC3s are critical during the early phase of immune defense in the C. rodentium model (17, 24, 25).
Stat3∆CD4 mice show decreased levels of IEC-derived antimicrobial peptides in the colon at the peak of C. rodentium infection, which is in accordance with IL-22, IL-17A, and IL-17F as potent inducers of antimicrobial peptides (5, 9, 22). Importantly, systemic overexpression of IL-22, but not IL-17A and IL-17F, could restore efficient reduction in the overall burden of C. rodentium in Stat3∆CD4 mice. Hence, our data argue for an important role of STAT3-dependent IL-22 release from CD4+ lymphocytes, which is consistent with a more severe phenotype in IL-22−/− mice than in IL-17A−/− and IL-17F−/− mice (5, 22).
We and others have previously shown that IL-22 signaling induces STAT3 activation in IECs upon mucosal damage, thereby promoting restoration of intestinal homeostasis (9, 47, 48). Strong STAT3 activation in IECs is also found upon infection with C. rodentium (49). Thus, our data support a model in which STAT3 activation in CD4+ lymphocytes links STAT3 activation in IECs via IL-22, thereby shaping the host defense in the colon during infection with enteropathogenic bacteria.
As this work did not involve human studies, potential conclusions regarding implications for human diseases remain speculative. In contrast to mice, no CD4+ ILC3s have been identified in humans so far (20). Our work revealed a key role for STAT3 activation in CD4+ lymphocytes for efficient reduction in the overall burden of enteropathogenic bacteria in a well-established disease model mimicking infections by enterohemorrhagic and enteropathogenic E. coli in humans. As in mice, STAT3 activation is essential for Th17 cell biology in humans (13). Consistently, genetic defects in STAT3 have been connected to human diseases rendering individuals susceptible to infections (14). Of note, single nucleotide polymorphisms of STAT3 and numerous other genes related to STAT3 signaling in CD4+ T cells have been associated with chronic intestinal pathology in inflammatory bowel disease (IBD) (6), and a pathogenic role of bacteria has been proposed for the onset and perpetuation of IBD (50, 51). Additionally, the contribution of a defective intestinal barrier function to the multifactorial pathogenesis of IBD has become increasingly acknowledged in the past few years. Thus, STAT3 activation in human CD4+ cells may guide host defense mechanisms by IECs in a similar way as in mice.
In conclusion, our data suggest that approaches specifically elevating STAT3 activation in CD4+ lymphocytes can shape strong epithelial defense mechanisms against infections with enteropathogenic bacteria. It is intriguing to speculate that the promotion of STAT3 phosphorylation in mucosal CD4+ T cells or the supply of colon-tropic STAT3 activated CD4+ cells might serve as a future therapeutic option for individuals suffering from infectious colitis or intestinal diseases associated with compromised epithelial defense mechanisms.
Bioluminescent C. rodentium were provided by C.U. Riedel (University of Ulm, Ulm, Germany). We thank K. Enderle, I. Zoeller-Utz, C. Lindner, and A. Taut for excellent technical assistance and M. McLaughlin for critically reading the manuscript.
This work was supported by the Interdisciplinary Center for Clinical Research Erlangen (to C.N.), by Deutsche Forschungsgemeinschaft Grants NE1927 (to C.N.), CEDER KFO257 (to M.F.N. and C.B.), and SPP1656 (to S.W.), and by the Friedrich-Alexander-Universität Emerging Fields Initiative (to C.N.) and the Erlanger Leistungsbezogene Anschubfinanzierung und Nachwuchsförderung-Fonds Erlangen (to C.N.).
Abbreviations used in this article:
inflammatory bowel disease
intestinal epithelial cell
innate lymphoid cell
IFN regulatory factor
regenerating islet-derived 3
retinoic acid–related orphan receptor
constitutively active STAT3.
The authors have no financial conflicts of interest.