Interleukin-7 Links T Lymphocyte and Intestinal Epithelial Cell Homeostasis

Interleukin-7 (IL-7) is a major survival factor for mature T cells. Therefore, the degree of IL-7 availability determines the size of the peripheral T cell pool and regulates T cell homeostasis. Here we provide evidence that IL-7 also regulates the homeostasis of intestinal epithelial cells (IEC), colon function and the composition of the commensal microflora. In the colon of T cell-deficient, lymphopenic mice, IL-7-producing IEC accumulate. IEC hyperplasia can be blocked by IL-7-consuming T cells or the inactivation of the IL-7/IL-7R signaling pathway. However, the blockade of the IL-7/IL-7R signaling pathway renders T cell-deficient mice more sensitive to chemically-induced IEC damage and subsequent colitis. In summary, our data demonstrate that IL-7 promotes IEC hyperplasia under lymphopenic conditions. Under non-lymphopenic conditions, however, T cells consume IL-7 thereby limiting IEC expansion and survival. Hence, the degree of IL-7 availability regulates both, T cell and IEC homeostasis.


Introduction
Interleukin-7 (IL-7) is a crucial survival factor for T cells and the competition for IL-7 is the major regulatory principle that stabilizes peripheral T cell homeostasis [1,2]. T cells express the IL-7 receptor (IL-7R) and remove IL-7 from the system continuously [3]. As soon as IL-7 production and consumption reach the equilibrium, the size of the peripheral T cell pool becomes self-limiting [1,2]. Consequently, the lack of IL-7consuming T cells is associated with increased levels of serum IL-7 in lymphopenic humans and mice [4,5].
Host survival depends on the tight regulation of IL-7 availability. For example, mice lacking IL-7 suffer from severe immunodeficiency [6]. In contrast, elevated levels of IL-7 promote spontaneous T cell activation [7] and T cell-mediated inflammation in the intestine and other organs [8][9][10]. Similarly, the overabundance of IL-7 under lymphopenic conditions contributes to the activation of adoptively transferred, naïve T lymphocytes, which undergo lymphopenia-induced proliferation (LIP), convert into effector/memory T cells and cause inflammation in the large intestine [11,12]. Based on the aforementioned observations, the blockade of IL-7R signaling in pathogenic T cells is considered as a therapeutic option for the treatment of T cell-mediated autoimmunity [13,14].
However, recent evidence suggests that the maintenance of immunological self-tolerance in the intestine is not only controlled by cytokine receptor signaling in immune cells. For example, cell autonomous cytokine receptor signals regulate intestinal epithelial cell (IEC) homeostasis and protect mice from immune-mediated colitis [15][16][17][18][19]. We have shown recently that IEC are the major source of IL-7 in the murine intestine [20]. However, it remained open whether and how IL-7 affects IEC homeostasis and intestinal physiology. Here we show that murine IEC express functional IL-7R and expand in response to IL-7 in vivo. Furthermore, we demonstrate that IEC accumulate in the colon of lymphopenic mice in an IL-7/IL-7R-dependent fashion correlating with decreased colitis induction. Importantly, IEC hyperplasia and protection from colitis are blocked, if naive CD8 + T cells consume IL-7 or IL-7R signaling is inactive in IEC. In summary, our data demonstrate that IL-7 regulates IEC homeostasis and intestinal integrity. Under lymphopenic conditions, elevated levels of IL-7 promote IEC hyperplasia and protect mice from colitis. In contrast, IL-7 consumption by T cells prevents IEC hyperplasia and facilitates colitis induction. Hence, an as yet unknown IL-7-dependent regulatory feedback loop links T cell and IEC homeostasis and controls intestinal integrity.

Lymphopenia-associated IEC hyperplasia leads to the accumulation of IL-7 + cells
We have shown recently that the commensal microflora and Interferon-ß (IFN-ß) promote IL-7 production in the intestine [20]. However, it remained unclear whether T cells contribute to the regulation of intestinal IL-7 production. In order to visualize T cell-dependent il-7 gene regulation in vivo, we made use of bacterial artificial chromosome (BAC)-transgenic IL-7-reporter mice. This mouse encodes enhanced green fluorescent protein (G), recombinase Cre (C), the human diphteria toxin receptor (D) and click beetle green luciferase 99 (L) under control of the il-7 promoter and is termed IL-7GCDL hereafter. As we have shown previously, luciferase activity in IL-7GCDL mice correlates closely with il-7 gene activity. IL-7GCDL mice were crossed to the Rag-deficient (Rag 2 ) genetic background. As shown in Figure 1A, bioluminescence (BL) signals were most pronounced in the thorax and the abdomen of Rag + and Rag 2 IL-7GCDL mice. However, BL was strongly increased in the abdomen, but not the thorax, of Rag 2 IL-7GCDL mice ( Figure 1A). This resulted mainly from elevated levels of transgene expression in the intestine with particularly high levels in the colon ( Figure 1B). In contrast, BL levels in thymus, skin and lung were similar in Rag + and Rag 2 IL-7GCDL mice ( Figure 1C). Histological analysis revealed that the colon of Rag 2 IL-7GCDL mice was hyperplastic ( Figure 1D) and contained increased numbers of IL-7 + Epcam + IEC ( Figure 1E). Importantly, IEC hyperplasia was independent of the GCDL transgene and equally pronounced in non-transgenic Rag 2 mice ( Figure S1). Thus, lymphopenia-associated IEC hyperplasia is associated with the accumulation of IL-7 + IEC in the colon.
Lymphopenia-associated IEC hyperplasia is IL-7/IL-7Rdependent IL-7 is a potent cell cycle-promoting and anti-apoptotic cytokine [21]. We therefore asked next, whether lymphopenia-associated IEC hyperplasia is IL-7-dependent. To test this, colon sections from IL-7GCDL non-transgenic WT, Rag 2 and IL-7 receptor a chain (IL-7Ra)-deficient Rag 2 (Rag 2 IL-7R 2 ) mice were analyzed. As compared to WT mice, Rag 2 crypts contained more Ki67 + proliferating cells (Figure 2A and C), were elongated ( Figure 2A and B) and contained fewer cleaved caspase 3 + apoptotic cells (Figure 2A). Additionally, the number of IEC expressing Gob5 (mCLCA3) was strongly reduced in Rag 2 mice (Figure 2A), indicating altered IEC differentiation. In contrast, colons of Rag 2 IL-7R 2 did not show any signs of IEC hyperplasia. The frequency of Ki67 + cells (Figure 2A and C) and colon wall thickness (Figure 2A and B) were increased while numbers of cleaved caspase 3 + apoptotic and Gob5 + cells ( Figure 2A) were reduced. Thus, IL-7R signaling promotes hyperplasia of IEC in the colon of Rag 2 mice. A similar tendency was observed in the small intestine ( Figure S2). However, these effects were far less pronounced than in the colon, probably due to lower levels of il-7 expression ( Figure 1B).
Next we asked whether IL-7R-dependent changes in IEC homeostasis were associated with alterations in gut physiology. For this purpose, transepithelial resistance (TER) and apparent permeabilities for Na + and Cl 2 were measured in the colon. As compared to T cell-competent WT mice, TER was elevated in Rag 2 samples ( Figure 2D), correlating well with the simultaneous reduction of Na + and Cl 2 permeability ( Figure 2E, F). Compared to Rag 2 samples, TER and permeabilities were restored again in Rag 2 IL-7R 2 mice and nearly reached WT levels ( Figure 2D-F). In effect, IL-7R-dependent IEC hyperplasia in Rag 2 mice is associated with decreased intestinal permeability.

IL-7 overabundance promotes IEC hyperplasia
Freshly isolated colonic IEC from Rag 2 mice expressed the IL-7R ( Figure 3A) and phosphorylated signal transducer and activator of transcription (Stat) 5 after IL-7 treatment ( Figure 3B). This suggests a direct action of IL-7 on IEC homeostasis. To test whether the restoration of IL-7 signaling is sufficient to induce IEC hyperplasia, IL-7-deficient Rag 2 (Rag 2 IL-7 2 ) mice were treated with recombinant mouse IL-7. To exclude IL-7R-independent side effects, Rag 2 IL-7R 2 mice were treated in parallel. PBStreated mice served as negative controls. As shown in Figure 4A and B, colon wall thickness and the numbers of Ki67 + cells/crypt were similar in the colons of PBS-treated Rag 2 IL-7 2 and Rag 2 IL-7R 2 mice. Hence, untreated Rag 2 IL-7 2 mice, similar to Rag 2 IL-7R 2 mice (Figure 2), did not show signs of IEC hyperplasia. In contrast, colons of IL-7-treated Rag 2 IL-7 2 mice contained elevated numbers of IEC ( Figure 4A and E) and showed a strong increase in IEC proliferation ( Figure 4B and E). Additionally, the frequency of apoptotic cleaved caspase 3 + IEC was reduced ( Figure 4E), indicating that IL-7 promotes both, IEC proliferation and survival. This resulted, at least partially, from a direct effect of IL-7 on IEC, which showed increased levels of nuclear Stat5 ( Figure S3). These effects were IL-7R-dependent, since IEC homeostasis remained unaltered in IL-7-treated Rag 2 IL-7R 2 mice ( Figure 4A and B). These results indicate, that the reduction of IEC numbers in Rag 2 IL-7 2 and Rag 2 IL-7R 2 mice does not result from unknown developmental defects but from the lack of IL-7 signaling.
Due to the lack of IL-7-consuming T cells, IL-7 availability is increased in Rag 2 mice [5]. This suggested to us that the intestinal epithelium responds to elevated IL-7 levels. To test this hypothesis, colon sections from IL-7 transgenic (tg) mice [22] were analyzed. As shown in Figure 4C, their colonic epithelium was hyperplastic and contained far more Ki67 + cells than that of non-transgenic WT mice ( Figure 4D). Thus, IL-7 overabundance is sufficient to cause IEC hyperplasia. IL-7 can induce nuclear translocation of b-catenin [23], a central regulator of IEC homeostasis [24]. In the nucleus, bcatenin binds to transcription factors of the T cell factor/ lymphocyte enhancer factor (TCF/LEF) family to activate genes promoting proliferation, survival, differentiation and positioning of IEC [24]. In the WT colon, nuclear b-catenin is mainly restricted to the crypt base [24] (data not shown). Accordingly, luminal IEC of PBS-treated Rag 2 IL-7 2 mice were nearly devoid of nuclear bcatenin ( Figure 4E). In contrast, IL-7-treatment caused the accumulation of b-catenin in the nucleus of luminal IEC ( Figure 4E). Similar results were obtained with IL-7tg mice ( Figure 4F). Hence, IL-7 overabundance is associated with the accumulation of nuclear b-catenin in luminal IEC and IEC hyperplasia.
T lymphocytes prevent IEC hyperplasia and promote colitis in an antigen-independent, IL-7R-dependent fashion Having shown that lymphopenia-associated IL-7 overabundance promotes IEC hyperplasia, we asked next whether IL-7 consumption by T cells is sufficient to normalize IEC homeostasis in Rag 2 mice. For this purpose, Rag 2 mice were reconstituted with polyclonal CD4 + and CD8 + T lymphocytes and colon sections were analyzed 85 days later. As compared to untreated Rag 2 controls, colon wall thickness ( Figure 5A) and the number of Ki67 + cells ( Figure 5B and D) were reduced in T cell-reconstituted Rag 2 mice. Simultaneously, the number of apoptotic cleaved caspase 3 + and Gob5 + cells ( Figure 5D) was elevated after T cell reconstitution. Additionally, nuclear b-catenin was hardly detect-able in luminal IEC of T cell-reconstituted Rag 2 mice ( Figure 5D). In agreement with reduced IEC numbers and a normalization of IEC homeostasis, BL was strongly reduced in the intestine of T cell-reconstituted Rag 2 IL-7GCDL mice ( Figure S4). In contrast, T cell reconstitution did not lead to any overt changes in the colonic epithelium of Rag 2 IL-7R 2 control mice ( Figure 5A and B). Similarly, IEC homeostasis remained unaltered in Rag 2 mice reconstituted with IL-7R 2 T cells ( Figure S5). However, at this stage we could not exclude that antigen-recognition and activation of the transferred T cells, in conjunction with IL-7R signaling, caused the subsequent regulation of IEC homeostasis. To test this possibility, the colon of Rag 2 OT-I + TCR-transgenic (Rag 2 OT-I + ) mice was analyzed. These mice are devoid of T and B lymphocytes except for a monoclonal population of CD8 + T cells, which are specific for chicken ovalbumin and therefore inert to foreign antigens. Importantly, CD8 + T cells from untreated Rag 2 OT-I + mice did not show any signs of activation and had a naïve phenotype ( Figure S7). As shown in Figure 5A, colon wall thickness in Rag 2 OT-I + mice was significantly lower than in Rag 2 mice. Furthermore, fewer proliferating Ki67 + ( Figure 5B and D) and more apoptotic cleaved caspase 3 + IEC were found in Rag 2 OT-I + mice ( Figure 5D). At the same time, high numbers of Gob5 + cells were found in Rag 2 OT-I + colons and luminal IEC contained hardly any nuclear b-catenin ( Figure 5D). Thus, IEC homeostasis in Rag 2 OT-I + mice was very similar to WT (Figure 2A-C) and T cell-reconstituted Rag 2 mice ( Figure 5A, B and D). We therefore conclude that naïve CD8 + T cells can regulate IEC homeostasis in an antigen-independent fashion. However, T cell-mediated IEC regulation requires IL-7R expression on both cells types, T cells ( Figure S5) and IEC ( Figure 5A-C). It is important to stress that CD8 + OT-I T cells expressed much higher levels of the IL-7Ra chain ( Figure 5E) than IEC ( Figure 3A). This suggests that naïve CD8 + OT-I T cells consume more IL-7 than IEC. If so, the presence or absence of T cells should affect IL-7R-dependent gene expression in IEC. In this context it was shown recently, that IL-7R signaling counterregulates il-7 gene activity in a negative feedback loop [5]. Consequently, IL-7 overabundance in Rag 2 mice is associated with decreased rates of il-7 transcription in lymph node stroma cells [5]. As shown in Figure 5C, IL-7 mRNA levels in the colon of untreated Rag 2 mice were significantly lower than in untreated Rag 2 IL-7R 2 mice. This indicates that IL-7 overabundance in Rag 2 mice reduces il-7 gene expression in IEC similar to lymph node stroma cells [5]. After T cell reconstitution, however, IL-7 mRNA levels increased in Rag 2 mice and reached levels similar to those found in untreated Rag 2 IL-7R 2 mice. Importantly, T cell reconstitution did not alter IL-7 mRNA levels in the colon of Rag 2 IL-7R 2 mice.
In Rag 2 OT-I + mice, IL-7 mRNA levels in the colon were significantly higher than in untreated Rag 2 mice but comparable to T cell-reconstituted Rag 2 mice. Hence, the presence of naïve CD8 + T cells is sufficient to increase il7 expression in IL-7Rcompetent IEC. This suggests that T cells withdraw IL-7 from IEC thereby regulating IL-7R-dependent gene expression and subsequent IEC homeostasis.
To test for the physiological relevance of our findings, we made use of a well-defined disease model. Mice were treated for 5 days with dextran sulfate sodium (DSS) via the drinking water to induce IEC damage and subsequent colitis. As shown in Figure 6A, WT mice lost weight rapidly without any apparent signs of recovery after DSS withdrawal at day 5. In contrast, body weight loss was only modest and transient in Rag 2 mice ( Figure 6A). However, DSS-resistance of Rag 2 mice was IL-7R-dependent as shown by the fact that Rag 2 IL-7R 2 mice lost around 20-25% of their body weight and hardly recovered ( Figure 6A).
DSS-induced colitis is more severe in WT than in Rag 2 mice, because IEC damage facilitates priming of commensal-specific, pro-inflammatory T cells [25], which aggravate disease. However, based on the aforementioned results we hypothesized that T cells may promote DSS-induced colitis in an antigen-independent fashion, namely via the blockade of IL-7R signaling in IEC. To test this, Rag 2 OT-I + mice were treated with DSS. The kinetics of body weight loss was nearly identical to Rag 2 IL-7R 2 mice ( Figure 6A). Importantly, CD8 + T cells from DSS-treated Rag 2 OT-I + mice did not show any signs of activation and had a naïve phenotype ( Figure S7). Hence, the presence of naive, IL-7consuming CD8 + T cells is sufficient to block the protective effect of IL-7 in DSS-treated Rag 2 mice.
As expected, tissue damage and leukocyte infiltration were most pronounced in WT mice ( Figure 6B and C). In contrast, Rag 2 colon samples showed only mild alterations as compared to untreated controls. However, tissue damage and leukocyte infiltration were clearly more pronounced in Rag 2 IL-7R 2 and Rag 2 OT-I + mice as compared to DSS-treated Rag 2 mice ( Figure 6B and C). In summary, our results demonstrate that IL-7 protects Rag 2 mice from DSS-induced tissue damage and subsequent colitis.

Discussion
CD4 + T cells recognizing antigens derived from the commensal microflora cause colitis under lymphopenic conditions [26]. Since IL-7 is a potent activation, growth and survival factor for T cells, it promotes intestinal damage via its action on T cells [9,11,12,27].
Here we provide evidence that IL-7 regulates IEC homeostasis in the colon and protects lymphopenic mice from DSS-induced colitis.
We show that IL-7-producing IEC accumulate in the intestine of Rag 2 mice ( Figure 1). This requires IL-7R signaling, correlates with increased rates of IEC proliferation and survival and results in IEC hyperplasia (Figure 2A-C). IEC are the major source of IL-7 in the colon [20] and express IL-7R, which are functional in vitro ( Figure 3) and in vivo ( Figure S3). This suggests that IL-7R signaling in IEC contributes to the regulation of intestinal homeostasis. Nevertheless, we cannot formally exclude indirect effects of IL-7, e.g. on innate lymphoid cells (ILC). IL-22-producing ILC are important regulators of intestinal homeostasis [28], they are generated in an IL-7/IL-7R-dependent fashion [29,30] and protect Rag 2 mice from DSS-induced colitis [31,32]. Therefore, the lack of ILC may promote alterations in IEC homeostasis and aggravate DSS-induced colitis in Rag 2 IL-7R 2 mice (Figure 2 and  6).
Nevertheless, the lack of ILC does not explain the results we have obtained with IL-7R-competent Rag 2 mice. For example, only Rag 2 mice reconstituted with IL-7R-competent ( Figure 5) but not -deficient T cells normalized IEC homeostasis ( Figure 5 and S5A-C) although ILC were similarly abundant in both experimental groups ( Figure S5D). Furthermore, IL-7R-competent Rag 2 OT-I + mice contained ILC ( Figure S5E) but nevertheless showed reduced rates of IEC proliferation/survival ( Figure 5A-D) and more severe symptoms after DSS treatment ( Figure 6). Finally, IEC numbers and proliferation were indistinguishable between lethally irradiated Rag 2 mice reconstituted with Rag 2 or Rag 2 IL-7R 2 bone marrow (BM) (Information S1 and Figure S6). This shows that lymphopenia-associated IEC hyperplasia can occur independently of BM-derived IL-7R + ILC and might result form IL-7R signaling in IEC.
Hence, our results indicate that IL-7R expression by naïve T cells is sufficient to limit IEC proliferation/survival and aggravate DSS colitis even in the presence of ILC. Together with the fact that the T cell-mediated normalization of IEC homeostasis required IL-7R expression by both, T and host cells ( Figure  S5A-C and Figure 5A and B), our data suggest that T cells are sufficient to regulate intestinal homeostasis via the withdrawal of IL-7 from the intestinal epithelium. Nevertheless, we do not want to exclude that T cells may produce an as yet unknown, IL-7induced factor, that also contributes to the regulation of IEC homeostasis.
T cell reconstitution was associated with two seemingly contradictory observations in the intestine of Rag 2 mice. On the one hand it increased il-7 gene activity in IEC (Fig. 5C), on the other, reporter gene activity was decreased in the intestine of Rag 2  IL-7GCDL mice ( Figure S4D). It is important to emphasize that the bioluminescent signal of an organ is determined by both, the transcriptional activity of the il-7 gene and the total number of IL-7-producing cells per organ. Since IEC numbers were strongly reduced in the intestine of T cell-reconstituted Rag 2 mice ( Figure 5A, B, D and Figure S4A-C) the comparably weak upregulation of il-7 gene activity ( Figure 5D) could not prevent the overall reduction of intestinal IL-7 production ( Figure S4D). Hence, T cells appear to regulate steady state IL-7 levels in the intestine of Rag 2 mice mainly via the modulation of IEC numbers.
It is important to note that IEC homeostasis in Rag 2 IL-7R 2 mice was similar to WT mice (Figure 2A-C). This shows that IL-7R signaling is dispensable for IEC proliferation and survival in the steady state. Nevertheless, IL-7 overabundance promotes IL-7R signaling in a considerable proportion of IEC ( Figure S3) correlating with elevated levels of IEC proliferation and survival ( Figure 4). We therefore conclude that lymphopenia-related IL-7 overabundance [5] is the major reason for IEC hyperplasia in Rag 2 mice (Figures 2 and 4) and protection from DSS-induced colitis ( Figure 6). Whether the protective effect of IL-7 is due to increased rates of (i) steady state IEC proliferation/survival or (ii) more efficient tissue regeneration in Rag 2 mice is an important question for the future.
In previous experiments we have analyzed the colon of IL-7GCDL mice after crossing them to reporter mice expressing eGFP only after the Cre-mediated deletion of a DNA stop cassette [20]. Since the IL-7GCDL mouse expresses Cre (but not eGFP) under control of the il-7 promoter, this approach enabled us to visualize IL-7 producing IEC and their progeny based on eGFP expression. In the colon of such F1 animals, entire crypts were eGFP positive [20] suggesting that the il-7 promoter is active in epithelial stem cells. This interpretation is supported by our current findings shown in Figure 1E and S1. Here, IL-7 expression is most prominent at the crypt base where stem cells are located [24]. Hence, colonic epithelial stem cells are a putative source of IL-7. However, it remains to be shown which other IEC subtypes express IL-7 and its receptor to elucidate whether IL-7 regulates IEC homeostasis in lymphopenic mice in an autocrine and/or paracrine fashion.
We have shown previously that the commensal microflora promotes il-7 gene expression in the intestine [20]. Consequently, intestinal BL and corresponding IL-7 levels in the colon (data not shown) were strongly reduced in antibiotic-treated Rag 2 IL-7GCDL mice ( Figure S8A). This was associated with reduced colon wall thickness ( Figure S8B) and lower levels of IEC proliferation ( Figure S8C). Similar results were obtained with germ-free Rag 2 mice ( Figure S8B and C) showing that the commensal microflora promotes IEC hyperplasia in Rag 2 mice, probably via the induction of il-7 gene expression [20].
It is known that the bacterial content of the large intestine is higher than in the small intestine. This may explain why il-7 gene activity is comparably low in the latter ( Figure 1B) and why IEC homeostasis in this part of the gut is less severely affected by IL-7 ( Figure S2). However, IL-7-dependent hyperplasia of the colonic epithelium correlates with changes in colon function ( Figure 2D-F) and alterations in the commensal microflora (Information S1 and Figure S8). IL-7 levels are elevated in HIV patients [4], which frequently suffer from diarrhea [33]. Similarly, severe combined immunodeficiency (SCID) patients often develop intestinal complications [34]. Due the broad impact of IL-7 on intestinal physiology shown here, it is tempting to speculate that the lymphopenia-associated overabundance of IL-7 promotes intestinal alterations frequently observed in lymphopenic patients.
The nuclear translocation of b-catenin is crucial for the maintenance of IEC homeostasis [24]. In a healthy colon, nuclear b-catenin is mainly restricted to IEC located at the crypt base [24]. However, in the colon of Rag 2 mice, nuclear b-catenin was also found in luminal IEC ( Figure 5D). This was not the case for Rag 2 IL-7 2 ( Figure 4E) and Rag 2 IL-7R 2 mice (data not shown) suggesting that IL-7R signaling promotes nuclear translocation of b-catenin in Rag 2 IEC, similar to what was described for myeloma cells [23]. In line with this, IL-7 injection and its transgenic overexpression were both sufficient to cause the accumulation of nuclear b-catenin in IEC ( Figure 4E, F). In the nucleus, b-catenin initiates a transcriptional program that promotes cell cycle progression [35,36] and survival [37]. This may facilitate wound healing after DSS-induced IEC damage and may explain why Rag 2 mice were protected from colitis in an IL-7R-dependent fashion ( Figure 6).
IEC hyperplasia is an early step in colon carcinogenesis [38] and facilitates malignant transformation in the mouse intestine [18,39]. Hence, IL-7/IL-7R-dependent IEC hyperplasia in Rag 2 mice may be a double-edged sword. While it improves resistance to chemically-induced IEC damage, it may increase the risk for colon cancer development at the same time.
Due to their high levels of IL-7R expression ( Figure 5E), T lymphocytes are major IL-7 consumers in the body. It is well accepted that T lymphocytes compete for IL-7 and that the degree of IL-7 availability determines the size of the peripheral T cell pool [1,2]. Here we provide evidence that the degree of IL-7 availability also determines the size of the IEC pool and that competition for IL-7 is not restricted to T cells. As we have shown here, IL-7R + T cells alter b-catenin expression, IEC homeostasis and il-7 gene expression in Rag 2 but not Rag 2 IL-7R 2 mice ( Figure 5). This process seems to be independent of TCR specificity, since naïve ovalbumin-specific CD8 + OT-I T cells normalized the indicated parameters similarly efficient like polyclonal T lymphocytes ( Figure 5). Based on these results, we propose that T lymphocytes and IEC homeostasis are linked via the competition for IL-7. In the presence of an intact T cell pool, IL-7 levels are too low to affect IEC homeostasis. Under lymphopenic conditions, however, IL-7 overabundance promotes IEC hyperplasia and protects the epithelium from tissue damage ( Figure 6). Hence, the immune status determines whether and how IL-7 affects intestinal homeostasis.
In summary, our results provide new insights into the regulation of intestinal homeostasis and have important implications for the design of clinical protocols targeting the IL-7/IL-7R signaling pathway to treat T cell-mediated autoimmunity and cancer [13,14,40].

Ethics Statement
This study was performed in strict accordance with the recommendations for the Care and Use of Laboratory Animals at the Charité-Universitä tsmedizin Berlin. The protocol was approved by the Landesamt für Gesundheit und Soziales-Berlin (Permit Number: G0170/08). Every effort was made to minimize suffering.

Adoptive T cell transfer
CD4 + and CD8 + T cells were isolated from single cell suspensions prepared from spleens and lymph nodes of Thy1.1congenic WT mice using CD4-and CD8a-specific microbeads and autoMACS (both Miltenyi Biotec GmbH). 5610 6 cells were injected i.v.

Bioluminescence (BL) detection
D-luciferin (Synchem OHG) and the IVIS Imaging Systems (either Series 100 or 200) (Xenogen) were used. For BL detection in live animals, 3 mg D-luciferin were injected i.p. into shaved mice. For BL detection in isolated organs, mice were injected i.v. with 3 mg D-luciferin 2 minutes prior to organ removal. Organs were collected in phosphate-buffered saline (PBS) containing 1.5 mg/ml D-luciferin and analyzed subsequently.

Immunostaining
Immunostainings were done as described previously [20]. To determine colon wall thickness, the distance between submucosa and lumen was measured. Six to eight areas were analyzed per mouse and the mean colon wall thickness was calculated. The percentage of Ki67 + cells/crypt was determined for 5 to 6 crypts per mouse and mean values were calculated.

Transepithelial resistance and permeability measurements
Colon samples were rinsed with Ringer's solution, glued on plastic rings (serosal area of the tissue without correction: 0.049 cm 2 ) with histoacryl tissue glue and mounted in Ussing chambers for measuring transepithelial resistance (TER, V ?cm 2 ) and dilution potentials. Resistance of the bathing solution was measured prior to the experiment and subtracted. Ussing hemichambers and water-jacketed gas lifts were filled with 10 ml standard Ringer's solution (in mM: Na + 140; Cl 2 149.8; K + 5.4; Ca 2+ 1.2; Mg 2+ 1; HEPES 10; D(+)-glucose 10. pH was adjusted to 7.4 with NaOH. The solution was equilibrated with 100% O 2 at 37uC. NaCl dilution potentials were measured by switching one hemichamber to a solution containing a reduced concentration of NaCl and all other components identical to standard Ringer's. Osmolality was balanced by mannitol. Apparent ion permeabilities for sodium and chloride were determined from dilution potentials using the Goldman-Hodgkin-Katz equation. For relating TER and permeabilities to the effective epithelial area, the area of crypt epithelium was determined. For this, colon samples were glued to plastic rings as used in the Ussing chamber. Crypt number per mm 2 (n crypt /A serosa ) was determined by counting crypts of Clarke's reagent fixed tissue under a light microscope. To evaluate the crypt length (L crypt ) and inner and outer crypt diameter (ID crypt ; OD crypt ), cross sections of tissues were fixed with 10% formalin, embedded in paraffin, HE-stained, and analyzed. From the evaluated parameters area correction factors were calculated by equation 1: Quantification of IL-7 mRNA IL-7 mRNA was quantified as described previously [20].

Isolation of colonic IEC and flow cytometric analysis of IL-7R expression
Freshly isolated colons were incised longitudinally, washed in PBS/EDTA (2 mM), and subsequently incubated for 30 min at 37uC in RPMI1640 supplemented with 10% FCS, Penicillin/ Streptomycin, 0.2 mg/ml Collagenase D, 0.2 mg/ml Dispase II and 10 mg/ml Dnase I (Roche Diagnostics). After incubation, epithelial cells were detached with the help of a syringe plunger and free crypts were incubated in PBS/EDTA (5 mM) for 10 min at 4uC. Dissociated cells were passed through a 40 mm cell strainer (BD Biosciences) and washed in PBS/EDTA (5 mM). Cells were stained with directly labeled monoclonal antibodies against CD45 (30F-11, BD Biosciences), CD326 (G8.8, Biolegend) and CD127 (A7R34, eBioscience) for 45 min at 4uC. Two rounds of signal amplification were performed with the anti-APC FASER kit (Miltenyi Biotec GmbH) to increase the fluorescence intensity for CD127. Data were acquired on a Canto II flow cytometer (BD Biosciences) and analyzed with FlowJo software (Tree Star).

Colitis induction and histopathological analysis
Acute colitis was induced by feeding mice with 4% dextran sulfate sodium (DSS, MW: 36000-50000, MP Biomedicals) dissolved in drinking water and administered ad libitum for 5 days followed by 3 days of regular drinking water, resulting in a 8day experimental period. For histological analysis, colon samples were fixed with 4% paraformaldehyde and embedded in paraffin. 2 mm sections were cut, deparaffinized, stained with hematoxylin and eosin (H&E), and scored in a blinded manner. The DSSinduced histological colitis score is the sum of individual scores for inflammatory cell infiltration and tissue damage. Individual scores for infiltration were given as follows: 0 points, no inflammatory cell infiltration; 1 point, increased number of inflammatory cells in lamina propria; 2 points, inflammatory cell infiltration extends into the submucosa; 3 points, transmural inflammatory infiltrates; and for tissue damage: 0 points, no mucosal damage; 1 points, discrete epithelial lesions; 2 points, erosions or focal ulcerations; 3 points, severe mucosal damage with extended ulcerations extending into bowel wall.

Statistics
Data are given as means+SEM and tested for significance by Student's t-test or the Mann-Whitney U-test, as appropriate. In cases of multiple testing, the Bonferroni-Holm correction was applied. P,0.05 was considered significant.  Figure S3 IL-7R signaling induces nuclear accumulation of Stat5 in colonic IEC. Rag 2 IL-7 2 (n = 3) and Rag 2 IL-7R 2 (n = 3) mice were treated with PBS or IL-7/anti-IL-7 (IL-7) twice a week for 2 weeks as described in Figure 4. Colon sections were stained with DAPI and antibodies for Stat5 (green) and EpCam (red). The white arrow indicates a nucleus containing Stat5. Data represent are representative for 2 independent staining reactions per mouse. (TIF) Figure S4 IEC homeostasis and IL-7 reporter activity are normalized in T cell-reconstituted Rag 2 IL-7GCDL mice. (A-C) Rag 2 IL-7GCDL mice (Rag 2 ) were reconstituted with 5610 6 MACS-sorted CD4 + and CD8 + T lymphocytes isolated from spleens and lymph nodes of Rag + IL-7GCDL mice (Rag + ). Colon sections were analyzed 56 days later. Colon sections from untreated Rag + IL-7GCDL mice (Rag + ), untreated Rag 2 IL-7GCDL mice (Rag 2 ) and T cell-reconstituted Rag 2 IL-7GCDL mice (Rag 2 +T cells) were stained with DAPI and antibodies for  Figure S8 The commensal microflora promotes intestinal IL-7 production, which induces IEC hyperplasia and alterations in the commensal microflora. (A) Rag 2 IL-7GCDL mice (n = 6) were treated with antibiotics for 49 days as described previously [20]. The same mouse prior to (untr) and after antibiotic treatment (anti) is shown. BL is shown in photons per s per cm 2 per steradian. Similar results were obtained after antibiotic treatment for 21 days in 3 additional experiments (n = 21). (B) Colon wall thickness (mm) and (C) the percentage of Ki67 + cells in crypts were determined in colon sections from untreated (untr; n = 3) and antibiotic-treated Rag 2 IL-7GCDL mice (anti; n = 6) as well as specific-pathogen-free (spf; n = 3) and germ-free Rag 2 mice (gf; n = 10). 5-7 individual crypts per mouse were analyzed. Shown are mean values+SEM. Statistically significant values are indicated: * p,0.05 and *** p,0.001 (Student's t test). (B, C) For antibiotic-treated mice, crypt length and the percentage of Ki67 + cells in crypts were analyzed for the representative experiment shown in (A). For germ-free mice, pooled data from two independent experiments are shown. Data are representative for 2-3 independent staining reactions per mouse. (D) Feces samples from WT (n = 18), Rag 2 (n = 18) and Rag 2 IL-7R 2 (n = 12) mice were analyzed by qRT-PCR to determine the copy number of eubacterial 16 S rDNA (EUBV3) and 16 S rRNA from Lactobacillus-group (LACTO), Mouse Intestinal Bacteroides (MIB), Bacteroides/Prevotella-group (BACT), gamma Proteobacteria/Enterobacteriaceae (ENTERO), Clostridium leptum subgroup (CLEP), Clostridium coccoides subgroup and Enterococcus (ECCOC). Shown are 16 S copy numbers per ng DNA (log10) for individual mice and the median for each experimental group. Statistically significant values are indicated: * p,0.05, ** p,0.01, *** p,0.001 (Mann-Whitney-U-Test). (TIF) Information S1 Experimental procedures for the generation of bone marrow chimeras ( Figure S6) and the analysis of the intestinal microflora ( Figure S8) are described. Specific references are included. (DOCX)