Background
Inflammatory bowel disease (IBD) is characterized by chronic and relapsing inflammation of the gastrointestinal tract. In humans, Crohn’s disease and ulcerative colitis are driven by a complex interplay between genetic and environmental factors, which contribute to chronic inflammation. Mouse models of colitis have been useful in determining the contributions of specific adaptive and innate immune mechanisms involved in the pathogenesis of IBD [
1]. Experimental studies in mice have revealed that a combination of genetic factors, diet, and immune responses to microbial organisms can contribute to IBD susceptibility.
In both humans and mouse models, several lines of evidence suggest a significant role for IL-7 in inflammatory diseases, including IBD. Firstly, under normal conditions, local IL-7 production acts on resident T cells through IL-7 receptor (IL-7R; a heterodimer composed of IL-7Rα and IL-2Rγ chains) to promote their differentiation, survival and homeostasis [
2]. IL-7 is produced by stromal cells and intestinal epithelial cells, suggesting a role for IL-7 in modulating immune responses in the intestinal microenvironment. Secondly, elevated serum concentrations of IL-7 have been detected in patients with ulcerative colitis [
3] and Crohn’s disease [
4], and gastric tissue biopsies from patients infected with
H.
pylori have increased IL-7 message [
5]. Polymorphisms in
IL 7RA are also associated with several inflammatory diseases including ulcerative colitis [
6]. Thirdly, the importance of IL-7 as a mediator of intestinal inflammation has been demonstrated in IL-7 transgenic mice which develop colitis resembling ulcerative colitis [
7], and in IL-7 deficient mice, which are resistant to the development of non-T-, non-B-cell-mediated colitis [
8]. Finally, the importance of IL-7-responsive-T cells in colitis has been demonstrated in TCRα
−/− mice, which develop spontaneous colitis driven by IL-7R
high memory T cells [
9,
10]. Adoptive transfer studies in these mice suggest that the colitogenic T cells are primarily memory CD4
+ cells which express high IL-7Rα and are maintained and expanded with IL-7 [
9,
10], indicating that IL-7 signaling is important in IBD pathogenesis. In an adoptive T-cell-transfer model of colitis, Yamazaki and colleagues successfully treated ongoing colitis using a saporin-conjugated anti-IL-7Rα antibody, selectively eliminating lamina propria lymphocytes (LPL) with high expression of IL-7R [
10]. These and other studies suggest that therapies interfering with IL-7R signaling could abrogate intestinal inflammation in IBD.
The aim of this study was to determine whether inhibition of IL-7Rα signaling would ameliorate colon inflammation induced by a bacterial trigger rather than with adoptive transfer of T cells. We used Helicobacter bilis (Hb) infection to induce colitis in T-cell-sufficient (multiple drug resistance 1a null; Mdr1a−/−) mice, and T- and B-cell-deficient (recombination activating gene 2; Rag2−/−) mice. This method has the advantage of not requiring adoptive transfer of T cells into immunodeficient hosts, which requires IL-7 for T-cell expansion and survival. In addition, our colitis models provided the opportunity to assess the cell types and functions affected by blocking IL-7Rα in mice where intestinal immunopathology is associated with T cells (in Mdr1a−/− mice), and innate immune cells (in Rag2−/− mice). Our results indicated that pharmacological inhibition of IL-7Rα reduced inflammation and subsequent disease development in Hb-infected mice by controlling the expansion of multiple leukocyte populations, as well as the activity of these immune cells. Our findings demonstrate an important function of IL-7R-driven immunity in experimental colitis involving both adaptive and innate immunity.
Methods
Mice, induction of colitis, and antibody treatments
Specific-pathogen free, 4–11 week old
Mdr1a−/− mice (FVB.129P2-
Abcb1a
tm1Bor
, Taconic Farms, Albany, NY) [
11,
12]; bred in house), and
Rag2−/− mice (129S6/SvEvTac-Rag2) were verified to be free of
Helicobacter species by fecal PCR [
13]. Mice were treated by intraperitoneal (IP) injection with anti-IL-7Rα M595 (rat IgG2b; t
1/2 = 3 days; Amgen Inc.) or isotype control M495 (rat IgG2b; Amgen Inc.) antibodies twice weekly, one week prior to oral gavage with
Hb and continuing for the study duration (total of 10 weeks for
Mdr1a−/− mice and 12 weeks for
Rag2−/− mice).
Hb-infection was confirmed by fecal PCR [
13]. Mice were weighed weekly and euthanized when they developed 20% body-weight loss or signs of severe IBD, and tissue samples were taken. All mice were used according to procedures approved by Amgen’s and UW’s Animal Care and Use Committees.
Pathology and immunohistochemistry
Necropsy, tissue sampling, processing and histologic examination was performed as previously described [
11]. Cecum and colon IBD scores were based on severity of mucosal epithelial changes, degree of inflammation and extent of lesions [
11]. 4 μm sections of formalin fixed, paraffin embedded cecum and colon were stained with biotinylated F4/80 antibody (clone CI:a3-1;AbD Serotec, Raleigh, NC) or isotype control, followed by development with horseradish peroxidase-labeled streptavidin and diaminobenzidine substrate solution. F4/80
+ cells were scored based on numbers of cells with membrane staining as follows: 0 = no staining, 1 = minimal to few faintly positive cells, 2 = scattered single positive cells, 3 = clusters of two or more positive cells, 4 = larger clusters of positive cells, multifocal to coalescing. The mucosa, lamina propria, submucosa and tunica muscularis/serosa of both cecum and colon were scored and summed (maximum score = 32).
Flow cytometric analysis
Single cell suspensions were made from spleen and MLN harvested at necropsy [
14]. Cellularity was determined using a hemocytometer or with an Advia 120 Hematology Analyzer (Siemens, Deerfield, IL). Cells were blocked with anti-CD16/CD32 (BD Biosciences, San Jose, CA) then stained with fluorescein isothiocyanate-, phycoerythrin-, peridinin-chlorophyll-protein complex-Cy5.5-, or allophycocyanin-labeled antibodies (BD) to the following surface markers: anti-CD44, CD4, CD8a, CD62L, CD127 (clone B12-1 binds a different epitope than 5 anti-IL-7Rα M59), CD49D (clone DX5), CD11b, and CD11c. Data were collected on BD’s FACSCalibur or LSRII and analyzed using FlowJo (Tree Star, Inc, Ashland, OR).
Serum cytokines and anti-Hb antibodies
Serum obtained at necropsy was analyzed using the Rodent MAP version 2.0 antigen panel (RBM, Austin, TX). Serum
Hb-specific IgG2a was determined by ELISA as previously described [
11] with the exception of using HRP-conjugated anti-mouse IgG2a (Southern Biotech, Birmingham, AL).
Colonic-explant culture cytokines
5 mm of proximal colon was flushed with PBS containing penicillin/streptomycin, blotted on sterile gauze and then cultured in RPMI supplemented with 10% FCS, 1X non-essential amino acids (Irvine Scientific, Santa Ana, CA), 20 mM Hepes, 100 U/ml Penicillin, 100 ug/ml Streptomycin, 50 μM 2-mercaptoethanol, 2 mM L-glutamine, 1 mM sodium pyruvate at 37°C and 5% CO2 for 24 hours. Supernatants were harvested, clarified by centrifugation, and analytes were detected using the LINCOplex Mouse Cytokine/Chemokine Panel-32 plex (Millipore, Billerica, MA) and a Luminex 100 IS (Luminex, Austin, TX). Data were analyzed using the Data Interpretation Application (Luminex). Values were corrected for the weight of the tissue and are presented as concentration of protein/mg tissue in the cultures.
Quantitative RT-PCR analyses
RNA was extracted from 5 mm of proximal colon (flushed with PBS) using the RNeasy kit (Qiagen, Valencia, CA). RNA was converted to cDNA using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA). Gene expression (relative to HPRT) was determined using TaqMan® Mouse Immune Array (Applied Biosystems). Duplicate 150 μg cDNA samples were loaded per reservoir. Data were collected on an ABI 7900 LDA and analyzed using SDS v2.2.2, Spotfire DecisionSite v19.3.1006 (TIBCO Software Inc., Somerville, MA), Microsoft Excel 2003, and GraphPad Prism 5. IL-7Rα expression (relative to Gapdh) was determined on an Mx3005P (Stratagene, La Jolla, CA) using Power SYBR Green PCR Master Mix (Applied Biosystems). Samples were run in duplicate. Primer sequences were as follows: IL-7Rα-forward: 5′-ACAAGAACAACAATCCCACAGAG, IL-7Rα-reverse: 5′-TCGCTCCAGAAGCCTTTGAAG Gapdh forward: 5′-TTCCGTGTTCCTACCCCCAATGTG, and Gapdh reverse: 5′-TAGCCCAAGATGCCCTTCAGTG.
Statistical methods
The Analysis of Variance (ANOVA) method was used to determine differences between treatment groups in colon mRNA expression, colon and serum protein panels, and FACS data. Comparisons between individual treatment groups were then performed with multiple comparison adjustment using either Tukey, Dunnett, or multivariate t methods. Homogeneous variance and normality assumptions were tested using Brown-Forsythe’s test and Shapiro-Wilk test, respectively. If either test showed a significant result (p < 0.01), a logarithm (logarithm base 2 for mRNA data) or square root transformation was considered. The transformation that satisfied both assumption tests was applied to the data. If both transformations failed, nonparametric ANOVA was used. The analyses were performed using SAS© 9.2 (SAS Institute Inc., Cary, NC). Heat maps were generated using Spotfire DecisionSite software. For IBD scores, a similar analysis approach was applied using Graphpad Prism 5 software. An ANOVA analysis was performed. A square root transformation was applied to correct heterogeneous variance among groups. Tukey's multiple comparison test was applied for group comparisons.
Discussion
We used two mouse models of bacterial-induced colitis to study the effects of blocking IL-7Rα on IBD: a T- and B-cell-sufficient strain (Mdr1a−/−) and a T- and B-cell-deficient strain (Rag2−/−). These models were used to circumvent the confounding variable of IL-7-dependent expansion of T cells as occurs in colitis models that rely on adoptive transfer of T cells into a lymphopenic environment. Our results show that IL-7Rα blockade can ameliorate bacterial-induced colitis, and that this effect involves not only T cells but also innate immune cells such as macrophages, DC, and NK cells. Successful treatment of colitis with an anti-IL-7Rα antibody was associated with decreases in T-cell and non-T-cell populations, as well as a reduction of inflammatory cytokines and chemokines. Our results confirm that IL-7Rα+ T-cell-mediated activities are likely key players in IBD. However, our results differ from some reports in that we demonstrate that IL-7R-mediated activity on non-T and non-B cells also contributes to experimental colitis.
In TCRα
−/− mice and T-cell-transfer-induced colitis, high expression of IL-7Rα on T cells is associated with colitis [
9,
10,
15], and IL-7-producing bone marrow cells may harbor colitogenic memory CD4
+ T cells [
16]. The use of a toxin-conjugated anti-IL-7Rα antibody (A7R34) by Yamazaki and colleagues suggests that selective elimination of IL-7Rα
high CD4
+ LPL is required to treat colitis [
10]. In our studies with
Mdr1a−/− mice, we noted that all T-cell subsets in draining MLN were decreased with anti-IL-7Rα M595 treatment, indicating that decreasing total T cells (which likely included CD4
+IL-7Rα
high cells) was in large part responsible for the lack of colitis seen in anti-IL-7Rα M595 antibody-treated mice. We did not detect increased colonic expression of IL-7Rα mRNA in
Hb-infected
Mdr1a−/− mice. The apparent discrepancy between our results of reducing total T cells vs. that of other’s work specifically eliminating IL-7Rα
high CD4
+ T cells may be due to the different model systems. In our models,
Hb is driving chronic inflammation by both adaptive and innate cells, whereas the adoptive transfer models are skewed towards promoting IL-7Rα
high T-cell expression and expansion. For example, in the 500 μg M595-treated
Hb-infected
Mdr1a−/− mice, we saw a greater decrease in total CD4
+ vs. CD8
+ T cells, as well as in naïve vs. activated/memory CD4
+ and CD8
+ T cells. Naïve T cells express high levels of IL-7Rα, whereas activated T cells down regulate IL-7Rα [
17]. Our FACS analysis did not distinguish between activated effector cells and memory cells, but this total population was less affected by IL-7Rα blockade than naïve T cells, perhaps due to lower expression of IL-7Rα. Reducing the pool of naïve T-cells may help control colitis by removing cells that would become activated or develop into memory cells following exposure to
Hb. Regardless of whether anti-IL-7Rα efficacy is due to specifically decreasing IL-7Rα
high or IL-7Rα
+ T cells in general, the combined effect of reducing both naïve and activated/memory T cells was associated with reduced colitis in
Mdr1a−/− mice.
In
Hb-infected
Mdr1a−/− mice, the lower dose of anti-IL-7Rα M595 (50 μg) was as effective in preventing colitis as the higher dose (500 μg), despite the lower dose having less of an effect on reducing T-cell numbers (in MLN). In fact, the lower dose preserved ratios of CD4
+ to CD8
+ and naïve to memory CD4
+ cells in MLN similar to control levels. In the colon however, the reduction in CD4 mRNA expression was the same with both doses of M595. The lower dose of M595 was also effective in reducing inflammatory mediators as measured in colon and serum, indicating that a relatively low dose of M595 can also inhibit the function of T cells (and likely non-T cells). The mechanism by which M595 antibody decreases cellularity is likely due more to population decay by inhibiting survival signals [
18,
19] rather than direct depletion of target cells. However, antibody-dependent cell-mediated cytotoxicity (ADCC) measurements of M595 showed it to be weakly lytic (Amgen Inc.; data not shown). It would be interesting to perform these colitis studies with a re-engineered anti-IL-7Rα antibody that does not have the confounding variable of ADCC. Further dose–response studies would be needed, but these results suggest there may be a therapeutic window for anti-IL-7Rα in inflammatory disease settings whereby T-cell function would be controlled without significant lymphopenia.
Given the significant alterations of chemokines and cytokines associated with innate immune cells in the serum and colon in the
Mdr1a−/− model, we explored anti-IL-7Rα treatment in a
Rag2−/− model of colitis. Amelioration of disease with anti-IL-7Rα treatment in the
Rag2−/− colitis model suggested that IL-7Rα expressing non-T cells, including macrophages, DCs, and NK cells are important in bacterial-induced disease. In fact, there were moderate (but significant) increases in surface expression of IL-7Rα on splenic macrophage and DC populations in
Hb-infected-
Rag2−/− mice. Multiple myeloid cell populations, along with inflammatory cytokines and chemokines associated with their function, were decreased with anti-IL-7Rα antibody treatment in
Hb-infected
Rag2−/− mice. Our findings differ from the conclusions reported by Shinohara and colleagues in which they used an adoptive transfer model of colitis to show that expression of IL-7Rα on CD4
+ T cells, but not on other cells (NK cells, granulocytes, macrophages, and DC), was essential for development of colitis [
20]. However, the adoptive transfer model of colitis is dependent on IL-7R-mediated expansion of T cells for induction of colitis, in contrast to our models wherein bacterial antigens drive inflammation and colitis involving resident host T cells (when present) and other innate immune cells. It is possible that IL-7R expression by cells other than CD4
+ T cells has a modest effect in their model as they noted a trend of increased colitis scores in
IL 7R−/− x
Rag2−/− recipients of wildtype naïve T cells compared to
Rag2−/− -only recipients, although these differences were not significant [
20].
Our studies using
Mdr1a−/− and
Rag2−/− mice strongly suggest that cells other than CD4
+ T cells are important in bacterial-induced colitis. DCs along with other APCs, are likely involved in the development of colitis [
21]. In
Hb-infected
Rag2−/− mice, we found that both pDCs and mDCs were significantly increased with colitis and dramatically decreased with anti-IL-7Rα treatment, especially IL-7Rα
+ pDCs. In
Mdr1a−/− mice, DC-derived cytokines and DC-activation markers CD80 and CD86 were increased with disease and significantly decreased with anti-IL-7Rα antibody treatment. These results are highly consistent with other studies in T-cell sufficient mice, where IL-7 production from DCs and/or IL-7R signaling in DCs may regulate the proliferation and activation status of CD4
+ T cells [
22] involved in colitis.
Our results showed that NK cells may be associated with
Hb-induced colitis. We found that lymphotactin, produced by and chemotactic for NK cells, was increased with disease and decreased following anti-IL-7Rα M595 treatment in both
Rag2−/− and
Mdr1a−/− mice infected with
Hb. A regulatory role for NK cells was demonstrated in one adoptive transfer model of IBD [
23], whereas another study showed NK cells neither suppressed nor exacerbated adoptive transfer-induced IBD [
24]. Our findings using
Rag2−/− mice infected with
Hb indicated that total NK and IL-7Rα
+ NK cells were associated with promoting IBD rather than suppression. Further studies using various IBD models are needed to clarify the function of NK cells in experimental colitis.
IL-7 promotes inflammation in part via activation of monocytes and macrophages [
25] and induces proinflammatory cytokines and chemokines. In our two models, blockade of IL-7Rα decreased colonic F4/80
+ cells and CD68 mRNA in
Hb-infected
Mdr1a−/− mice, decreased subpopulations of macrophages in
Hb-infected
Rag2−/− mice, and reduced monocyte-derived chemokines in both
Hb-infected
Mdr1a−/− and
Rag2−/− mice. Shinohara and colleagues report that only IL-7R
+CD4
+ T cells (and not NK cells, granulocytes, DCs, or macrophages) contribute to colitis in their adoptive transfer model [
20], whereas von Freeden-Jeffry and colleagues reported that in the absence of T and B cells, IL-7 dramatically increases F4/80
+ cell infiltration into the intestinal mucosa upon
H.
hepaticus infection resulting in chronic intestinal inflammation [
8]. The main difference between these studies is the use of adoptive transfer of T cells into a lymphopenic environment vs. use of a bacterium to trigger IBD. Whereas T cells are important for IBD pathogenesis, our conclusions that non-T cells are also important in intestinal inflammation is aligned with von Freeden-Jeffry’s results. Our results using pharmacological blockade of IL-7Rα in T-cell sufficient and T- and B-cell-deficient mice clearly establish the contributions of both T cells and non-T cells such as macrophages, DCs and NK cells in IBD pathogenesis.
Competing interests
CRW, LN, CLS, JM, and GC were employed by Amgen, Inc. at the time the studies were completed. This work was sponsored by Amgen, Inc. LMP received funding from Amgen, Inc. for portions of the studies conducted at the University of Washington. AS, TB, PT, SP, and BMI declare that they have no competing interests.
Authors’ contributions
CRW and AS designed, conducted, analyzed and interpreted data, prepared figures, drafted and revised the manuscript. JM, TB, LMP designed, analyzed and interpreted data, and edited the manuscript. PT performed histopathology analysis and interpretation, and edited the manuscript. LN designed, performed, and analyzed flow cytometry data. GC performed statistical analysis and interpretation and designed the heat maps. SP performed the in vivo portions of the studies and tissue processing. CLS analyzed serum antibody concentrations. BMI contributed to data interpretation and editing of the manuscript. All authors have read and approved this manuscript.