|Home | About | Journals | Submit | Contact Us | Français|
Thymic stromal lymphopoetin (TSLP) influences numerous immune functions, including those in the colonic mucosa. Here we report that TSLP-deficient (Tslp-/-) mice, did not exhibit increased inflammation during dextran sodium sulfate (DSS)-induced colitis, but failed to recover from disease, resulting in death. Increased localized neutrophil elastase (NE) activity during overt inflammation was observed in Tslp-/- mice, and was paralleled by reduced expression of an endogenous inhibitor, secretory leukocyte peptidase inhibitor (SLPI). Pharmacological inhibition of NE, or treatment with rSLPI reduced DSS-induced mortality in Tslp-/- mice. Signaling through TSLPR on non-hematopoietic cells was sufficient for recovery from DSS-induced colitis. Expression of the receptor occurred on intestinal epithelial cells (IEC), with stimulation inducing SLPI expression. Therefore, TSLP is critical in mediating mucosal healing following insult, and functions in a non-redundant capacity that is independent of restraining T helper 1 (Th1) and Th17 cell cytokine production.
Thymic stromal lymphopoetin (TSLP) is critical in the regulation of numerous immunological processes. These include, but are not limited to, conditioning of dendritic cells (DCs) to drive Th2 skewing of T cells (Ito et al., 2005), enhanced class switching in B cells (Xu et al., 2007), and induction of Foxp3+ regulatory T (Treg) cells in humans (Watanabe et al., 2005) but not mice (Sun et al., 2007). TSLP expression is increased in immunopathologies with dysregulated Th2 cell-type cytokine production including atopic dermatitis (Soumelis et al., 2002), and asthma (Ying et al., 2005). While the pathophysiological features of these diseases can be recapitulated in mice over-expressing TSLP (Headley et al., 2009), decreased production of TSLP has been observed in mucosal biopsies from patients with inflammatory bowel disease (IBD) (Rimoldi et al., 2005). Regulation of TSLP is dependent on the transcription factor NF-κB, and is induced by exposure to bacteria (Rimoldi et al., 2005), agonists of toll-like receptor (TLR) or the intracellular sensor NOD, (Lee and Ziegler, 2007; Zaph et al., 2007) and cytokines including TNFα (Lee et al., 2008). Initially suggested as a mechanism to allow the commensal microbiota to induce tolerance, TSLP produced by intestinal epithelial cells is not required for tolerogenic mucosal DC, and Treg cell generation (Iliev et al., 2009).
Although the etiology of IBD has remained elusive, immuno-pathology results from loss of immunosuppressive or enhanced production of pro-inflammatory factors. Self-resolving immuno-pathology that recapitulates the hallmarks of colitis in patients can be induced in mice by addition of dextran sodium sulfate (DSS) to drinking water. Inflammation in this model is due to epithelial cell apoptosis, increasing recruitment and activation of macrophages (Tlaskalova-Hogenova et al., 2005), DCs (Berndt et al., 2007), and neutrophils (Okayasu et al., 1990). While T-cell activation and mucosal homing occur, T-cells are not necessary for disease (Katakura et al., 2005). The resulting inflammation is associated with structural changes in the colon due to increased activity of proteases including neutrophil elastase (NE) and metallomatrix proteases (MMPs) (Castaneda et al., 2005).
Recovery following inflammation is a complex process requiring not only suppression of immune processes, but also balance between proteolytic enzymes and their endogenous inhibitors. Secretory leukocyte peptidase inhibitor (SLPI) is induced by microbial products, functioning as a serine protease inhibitor, anti-microbial peptide, and inhibitor of NF-κB (Taggart et al., 2005). SLPI is increased during resolution from DSS colitis (Mizoguchi et al., 2003), and exerts a non-redundant role following injury, as SLPI deficiency reduces cutaneous wound healing (Ashcroft et al., 2000).
TSLP in the gastrointestinal tract has been described to inhibit Th1 cell type cytokine production following insult. Although in a previous publication increased IFNγ production was observed in mice deficient for the TSLP receptor (Crlf2-/-), potentiating the severity of DSS colitis (Taylor et al., 2009); here we report that deficiency in TSLP ligand did not increase colitis severity, but prevented recovery from disease. Recovery from colitis occurred through signaling transduced by TSLPR on radio-resistant host cells, as mortality was not increased in Crlf2+/+ mice reconstituted with Crlf2-/- bone marrow (BM). Our data demonstrates that TSLP functions as a critical mediator controlling the balance between host defense and wound repair, without constraining production of Th1 cytokines.
The requirement for TSLP in the development and maintenance of B- and T-cell populations was assayed in TSLP wild-type (Tslp+/+), and deficient (Tslp-/-) mice. Consistent with a lack of a requirement for TSLP in murine Treg cell development (Sun et al., 2007), equivalent numbers of CD4+Foxp3+ cells were found in the spleen, and mesenteric lymph nodes (MLNs) (Figure S1A) in Tslp+/+ and Tslp-/- mice. This was mirrored by equivalent colonic Foxp3 expression in both genotypes by qRT-PCR (Figure S1B). TSLP deficiency did not alter the number of DCs (CD11b+ CD11c+, CD11b- CD11c+, or CD11clo PDCA-1+) in the spleen or MLN, or increase activation, as assessed by CD40, CD80, and CD86 (Figure S1C&D). In addition, we noted no spontaneous inflammation in mice aged for 18 months. Thus, the absence of TSLP does not favor the development of spontaneous inflammation or autoimmunity.
In the absence of spontaneous inflammation, we sought to determine if the lack of TSLP would enhance the severity of an induced colonic inflammation. Unlike the increased susceptibility reported in Crlf2-/- mice (Taylor et al., 2009), the severity of DSS colitis was equivalent in Tslp-/- and Tslp+/+ mice. Macroscopic score, colon length, concentration of the acute phase protein serum amyloid A (SAA), and histological damage scores, were not increased in Tslp-/- compared to Tslp+/+ mice at 8 days post-DSS (Figure 1A-E). While Tslp+/+ mice began to recover 9-10 days post-DSS, TSLP deficiency prevented recovery from inflammation (Figure 1F). Progressive weight loss and inability to recover met pre-defined humane endpoints, requiring sacrifice beginning on day 9, with no Tslp-/- mice surviving 14 days post-DSS (Figure 1G). Peripheral blood analysis indicated no anemia, or thrombocytopenia in all genotypes (Table S1). Accelerated disease progression was also excluded, as Tslp-/- mice assessed 4 days post-DSS did not display increased inflammation compared to Tslp+/+ (Figure S2A&B). We further excluded increased susceptibility to DSS, as a lower dose (3% w/v) produced equivalent weight loss, colon shortening, macroscopic and histological damage scores in Tslp-/- and Tslp+/+ mice (Figure S2C&D and data not shown). Therefore, lack of TSLP does not result in enhancement of inflammation, but impairs recovery from disease.
To ascertain if the lack of enhanced colonic inflammation was unique to Tslp-/- mice, we subjected mice lacking TSLP receptor (Crlf2-/-) to DSS-colitis. Onset of disease by weight loss, macroscopic score, colon length, and histological examination, indicated equivalent morbidity in Crlf2-/- compared to WT or Tslp-/- mice (Figure S2E&F and data not shown). However, Crlf2-/- mice experience significantly increased mortality compared to DSS-treated WT mice (Figure S2G). Thus, lack of TSLP or TSLPR reduces the ability to recover from colonic insult without enhanced inflammation.
In order to determine if these data were specific to the DSS model, we utilized a T-cell dependent model of colitis. Disease was induced by adoptive transfer of FACS sorted CD4+CD45RBhi effector T (Teff) cells into Rag1-/- mice, with the effectiveness of Treg cell assessed by co-transfer of CD4+CD45RBlo cells. Transfer of Teff cells from Crlf2+/+ or Crlf2-/- mice resulted in weight loss 5 weeks post-transfer (Figure 2A). Disease onset, macroscopic score, colon length, and histological scores, were similar between recipients of Crlf2+/+ or Crlf2-/- T-cells (Figure 2A-E). Recipients of Teff + Treg cells from Crlf2+/+ or Crlf2-/- mice appeared healthy upon necropsy. Moreover, Treg cells from Crlf2-/- mice remain effective in vivo against WT Teff cells as recipients of Crlf2+/+ Teff + Crlf2-/- Treg cells did not develop disease (Figure 2A-E). These data demonstrate that TSLP signaling in T-cells does not influence the development of T-cell dependent colitis, and that Treg function is not diminished by deficiency in TSLPR.
As TSLP binds a heterodimer of TSLPR and IL-7Rα, and IL-7 can increase colitis severity (Okada et al., 2005), we investigated if lack of TSLP or TSLPR altered IL-7 signaling. Stimulation of isolated splenocytes with rmIL-7 induced equivalent STAT5 phosphorylation in CD4+ and CD8+ T-cells from WT, Tslp-/- and Crlf2-/- mice (Figure 2F). Together, these data demonstrate that absence of TSLP evoked signaling in T-cells does not influence the development of a T-cell dependent colitis, that Treg function is not diminished by deficiency in TSLPR, and that lack of enhanced inflammation is not specific to innate or adaptive immune models of colitis.
Previous work conducted on Crlf2-/- mice implicated increased Th1 type cytokines produced during DSS-induced colitis enhanced the severity of disease. To ascertain if delayed recovery was due to dysregulated Th1-type cytokine production, the immune response to DSS was characterized by several complementary techniques. Cytokines characteristic of Th1, Th2, and Th17 cell immune responses were evaluated in the serum and colons of mice 8 days post-DSS. Mice with overt inflammation had significantly increased serum concentrations (Figure 3A) and colonic mRNA (Figure 3B) for IFNγ, IL-4, IL-6, IL-10, IL-17, TNFα, and KC. No differences were observed between the two genotypes at baseline or during inflammation (Figure 3A&B). CD4+ T-cells from the spleen or MLN also produced equivalent amounts of IFNγ following in vitro re-stimulation irrespective of genotype (data not shown). Therefore the initial severity of intestinal inflammation and cytokine production is not simply enhanced in the absence of TSLP.
We further assessed if lack of TSLP would impact systemic immune functions by analyzing the response following immunization with OVA. Injection of OVA:CFA or OVA:Alum induced potent Th1 and Th2 responses respectively. We detected no difference in OVA specific IgG1 (Figure S3A), or IgG2a (Figure S3B) antibody titers in Tslp-/- mice compared to Tslp+/+ controls. These data clearly indicate that lack of TSLP alone does not increase the production of Th1 type cytokines, basally or following the induction of an immune response.
Given the role of mucosal barrier dysfunction in IBD, it was conceivable that delayed recovery may be due to prolonged exposure to luminal contents, resulting in sustained immune activation. Intestinal barrier function was assessed by translocation of viable bacteria to the MLN and spleen, or by detection of gavaged FITC conjugated dextran probe in the serum. Tslp-/- mice did not have a basal intestinal barrier defect, as naive mice displayed no increased bacterial translocation to the MLN (Figure S4A) or spleen (Figure S4B) compared to Tslp+/+ mice. In contrast, increased bacterial translocation to the MLN and spleen was observed in all mice treated with DSS, irrespective of genotype. Similarly, translocation of FITC-dextran from the lumen to the blood was not enhanced in naive Tslp-/- mice compared to Tslp+/+ controls, whereas DSS significantly enhanced permeability in both genotypes (Figure S4C).
In addition to the IEC monolayer and tight junction proteins between adjacent cells, the mucosal barrier is also comprised of secreted IgA, and mucous produced by goblet cells. There was no reduction in the numbers of IgA+ B-cells in the Peyer's patches (Figure S4D), or in luminal or serum IgA (data not shown) in Tslp-/- mice. No difference in the number of goblet cells was observed in the colon or terminal ileum of naïve Tslp-/- mice (Figure S4E&F). Although DSS treatment reduced the number of colonic goblet cells, there was no difference between genotypes. Deficiency in TSLP does not reduce intestinal barrier function at baseline, or during inflammation.
To determine the impact of TSLP or TSLPR deficiency on the microbiota composition we used qRT-PCR for selected bacterial 16s rDNA. No significant differences in the microbiota were observed in comparing naïve or DSS treated WT, Tslp-/-, or Crfl2-/- mice. Treatment with DSS colitis did however significantly reduce segmented filamentous bacteria (SFB), to below the limit of detection (Figure S4G). Using this method, we detected no Salmonella sp., or Helicobacter sp.; bacteria known to exacerbate or induce murine colitis (data not shown). TSLP ligand or receptor deficiency does not appear to alter the composition of the colonic microbiota in health or disease.
To assess if macrophage or neutrophil recruitment was reduced in Tslp-/- mice, immunohistochemistry (IHC) was conducted. Similar numbers of these cells were observed in the colon following DSS colitis in Tslp-/- and WT mice (data not shown). Neutrophil recruitment after i.p. injection of LPS, or the NOD2 agonist MDP induced equivalent recruitment of neutrophils into the peritoneum of WT and Tslp-/- mice (data not shown). Together these data indicate that the absence of TSLP does not impair host defenses following breach of the epithelial barrier.
In order to determine the defect preventing Tslp-/- mice from recovering following DSS colitis, we performed qRT-PCR on colonic samples for genes critical to resolution of inflammation. Expression of Annexin A1 (ANXA1), PTGES2, iNOS (NOS2), and the negative regulators of NF-κB signaling IRAKM, and TNFAIP3 were increased in all genotypes following DSS administration (Figure 4A). We also detected no difference in TNFAIP8, TFF3, REGIIIβ and REGIIIγ, lipoxygenases (ALOX5, ALOX12, ALOX15), and genes expressed by alternatively activated macrophages; Arg1, Mrc1, Chi3l3, and Fizz1 between genotypes (Figure S5). In contrast, expression of Slpi (encoding secretory leukocyte peptidase inhibitor) mRNA was significantly up-regulated following DSS colitis in Tslp+/+, but not Tslp-/- mice, (Figure 4A) and confirmed by immunoblot on full thickness colon homogenates (Figure 4C).
Several mechanisms have been ascribed to how SLPI aids in resolution from injury, including the inhibition of proteases. Consistent with increased protease activity, colonic sections from DSS-treated Tslp-/- mice showed defined regions below IEC with reduced collagen (blue staining, area indicated by “*”, Figure 4B). Both MMPs and NE can degrade extracellular membrane (ECM) components, although SLPI has been demonstrated to inhibit NE. Using a non-selective MMP assay, increased colonic MMP activity was detected 8 days post-DSS, however no differences between genotypes was observed (Figure 4D). In a similar fashion, use of a NE selective substrate demonstrated significantly increased colonic NE activity in Tslp-/- mice 8 days post-DSS compared to WT (Figure 4E). Increased NE activity was not due to enhanced NE expression (Elane, Figure 4A), neutrophilia, or recruitment following DSS in Tslp-/- mice (data not shown).
NE is known to degrade a number of substrates in vivo. Included in these is progranulin (PGRN), a protein that aids wound healing unless degraded by NE. While increased PGRN protein expression occurred during inflammation in WT mice, decreased protein but not mRNA expression was observed in Tslp-/- mice (Figure 4A&F). In the absence of TSLP, increased NE activity following colonic inflammation causes PGRN degradation, a protein critical to wound repair.
To assess how increased NE activity and lack of SLPI contributed to increased mortality in Tslp-/- mice, a selective NE inhibitor (NEi), SSR69071, was used. Mice were subjected to DSS colitis and orally gavaged with SSR69071, beginning on day 6, prior to the recovery phase. SSR69071 treatment significantly reduced mortality in Tslp-/- mice compared to vehicle control (Figure 5A). Disease was not altered in Tslp+/+ mice, indicating that SSR69071 does not simply reduce the severity of colitis. To ascertain if lack of SLPI alone led to increased NE activity and mortality, Tslp-/- mice were treated with rSLPI or vehicle, at the beginning of recovery. Gavage of rSLPI significantly reduced mortality in Tslp-/- mice following DSS colitis (Figure 5B). These data demonstrate that enhanced NE activity in Tslp-/- mice, due to the absence of SLPI, prevents colonic healing following insult.
Inhibition of NE by SSR69071 and the impact on newly synthesized (non-cross linked) collagen was assessed by Sircol collagen assay. Nascent collagen was significantly increased after DSS administartion of Tslp+/+ but not Tslp-/- mice. NE inhibition had no effect in Tslp+/+, but significantly increased collagen in Tslp-/- mice compared to vehicle control (Figure S6A).
Expression of prominent intestinal collagen progenes assayed for by qRT-PCR indicated increased expression of Col1a1 in Tslp+/+ and Tslp-/- mice 8 days post-DSS, irrespective of NEi treatment (Figure S7). Significantly increased expression of Col4a2 was also observed in DSS-treated Tslp-/- mice receiving vehicle or SSR69071. We also noted reduced Col1a1, and Col4a1 following DSS administration, however this was again independent of genotype, or NEi treatment. These changes therefore were unlikely to account for any difference in mortality or detection of nascent collagen protein.
To address how NEi reduced mortality, we assessed if SSR69071 prevented apoptosis or enhanced IEC proliferation. Apoptosis, as indicated by cleaved caspase 3 (CC3) staining, was not increased in Tslp+/+ or Tslp-/- mice 8 (Figure S6B), or 10 days post-DSS (Figure 5D). Ki67 staining revealed significantly reduced proliferation 8 and 10 days post-DSS in Tslp-/- mice.
While this proliferation defect was not restored by SSR69071 treatment 8 days post-DSS (Figure S6C), IEC proliferation was restored 10 days post-DSS in NEi treated Tslp-/- mice (Figure 5C&E). Inhibition of NE restored nascent collagen 8 and 10 days post-DSS (Figure S6A&E). More importantly, treatment normalized PGRN protein 8 days post-DSS (eg. 2 days of NEi treatment Figure S6D)), and 10 days post-DSS (4 days NEi treatment, Figure 5F). This occurred at a post-transcriptional level as PGRN mRNA expression did not increase with NEi treatment (Figures S6F&S7). Thus, while the NEi was effective within 2 days of treatment, at least 4 days of increased PGRN stability was required before an effect on IEC proliferation could be observed. TSLP-induced SLPI is required to regulate NE activity, and subsequently PGRN availability, leading to healing and ultimately the survival of DSS-treated Tslp-/- mice.
We assessed the sites of TSLP and SLPI production in situ by performing immunofluorescence on colonic tissue. To this end, staining for epithelial cadherin (E-cadherin; green), a protein that localizes to the periphery of IEC, SLPI (red) and TSLP (blue) was performed (Figure 6A). TSLP was predominantly expressed in IEC (E-cadherin+ cells, adjacent to “*”). In contrast, SLPI expression can be found in TSLP+ IEC (adjacent to “†”), and in cells of the lamina propria (arrowheads).
To determine if TSLP-induced signaling in immune cells was required for resolution of colitis, we constructed BM chimeras by reconstituting irradiated WT mice with BM from Crlf2+/+ or Crlf2-/- mice, and subjected them to DSS. The severity of colitis was not enhanced in WT recipients of Crlf2-/- bone marrow (Crlf2-/- → Crlf2+/+) compared to WT (Crlf2+/+→ Crlf2+/+) chimeras, 8 days post-DSS. Weight loss (Figure 6B), colon length (Figure 6C), and histological damage (Figure 6E) were similarin recipients of Crlf2+/+ or Crlf2-/- BM; indicating that the absence of TSLPR on cells of hematopoietic origin does not enhance colonic inflammation. In addition, increased mortality was not observed in recipients of Crlf2-/- BM following DSS colitis compared to Crlf2+/+ BM recipients. Furthermore, no reduction in collagen deposition (Figure 6D), and equivalent SLPI mRNA expression was detected in Crlf2-/-→ Crlf2+/+ and Crlf2+/+→ Crlf2+/+ (Figure 6F) BM chimeras. Together, these data demonstrate that TSLP aids in resolution of colonic inflammation through cells of a non-hematopoietic lineage.
IEC were isolated from the duodenum, jejunum, terminal ileum, and mid-distal colon, and assessed for TSLPR by flow cytometry. Isolated and enriched epithelial cells, from jejunum, terminal ileum, and colon were TSLPR+ (Figure 7A), and confirmed in situ by IHC (Figure 7B). TSLPR was not expressed on fibroblasts (αSMA-, E-cadherin-, vimentin+,) or myofibroblasts (αSMA+, E-cadherin-, vimentin+) isolated from the intestinal stroma (Figure 7C). TSLPR on IEC is functional, as stimulation with rmTSLP induced AKT (S473) and ERK (T208) phosphorylation in Crlf2+/+ but not Crlf2-/- IEC (Figure 7D).
In order to determine the kinetics of signal transduction, we stimulated the mouse IEC line mICcl2, with rmTSLP. Stimulation elicited ERK1/2 and AKT phosphorylation within 10 minutes, and the downstream target, Foxo3a, within 30 minutes (Figure 7E). Phosphorylation of Foxo3a resulted in increased expression of the anti-apoptotic BCL2 within 2h post-stimulation, while MCL-1 was not altered (Figure 7F). This suggests that TSLP can alter apoptosis in IEC by inducing anti-apoptotic genes expression.
This reductionist approach was further utilized to determine if TSLP stimulation of IEC was sufficient to induce SLPI expression. mICcl2 monolayers stimulated with rmTSLP, increased SLPI mRNA was within 1h following stimulation (Figure 7G), and increased protein expression after 2h (Figure 7F). Together these data demonstrate that TSLP can directly elicit signaling in IEC and induce the expression of SLPI.
This study highlights the critical role for TSLP in colonic healing following transient inflammation. Our findings demonstrate that lack of TSLP ligand or receptor does not enhance disease severity, but prevents resolution, resulting in death. Increased mortality was due to unrestrained NE activity increasing turnover of progranulin, a protein that induces IEC proliferation. The underlying mechanism for this dysregulated NE activity was found to be due to an inability to increase SLPI expression following DSS-induced inflammation. TSLP-elicited signaling is not restricted to immune cells, with stimulation through TSLPR on IEC activating intracellular signaling, and SLPI expression.
Immunity was not altered in Tslp-/- mice in the steady state or after the induction of an immune response. No Th1 or Th17 bias was observed in naïve mice, or during DSS-induced colitis systemically (serum or splenocytes), or within the mucosal environment (MLN or colon). Lack of enhanced inflammation was not model specific, as deficiency in TSLP signaling did not increase the severity of a T-cell dependent colitis by enhanced Teff (i.e. increased Th1 cytokine production) or decreased Treg cell function. It is important to note that although TSLP can potentiate Th2 responses (Ito et al., 2005), Tslp-/- or Crlf2-/- mice remained capable of generating Th2 cytokines. In accordance with our data, Th2 responses remained intact following infection of Crlf2-/- mice with an intestinal helminth (Massacand et al., 2009). Moreover, oral vaccination against OVA induces equivalent OVA specific Th2 responses in WT and Crlf2-/- mice (Blazquez et al., 2010). Therefore, TSLP is not required for Th2 responses, and the absence of TSLP does not necessarily result in Th1 biased immune responses.
To determine whether death in Tslp-/- mice was due to overt inflammation, downstream pro-inflammatory signaling cascades were studied. TSLP deficiency did not alter the expression of negative regulators of pro-inflammatory cytokine signaling, eliminating a role for uninhibited NF-κB regulated gene expression (e.g. iNOS, IL-6, TNFα). Although reduced healing of the intestinal mucosa, and an inability to limit colonic inflammation has been observed in Anxa1-/- (Babbin et al., 2008) and Ptges2-/- (Morteau et al., 2000) mice, TSLP deficiency did not impact expression of these genes; indicating that the resolution defect is independent of these factors. As a whole, our data demonstrate that impaired recovery from DSS colitis is not simply due to aberrant cytokine production.
We excluded that impaired host defenses prevented the ability to recover from DSS colitis, and increased mortality in Tslp-/- mice. Naïve and DSS-treated Tslp-/- mice did not have increased bacterial translocation to the MLN or spleen. Increased translocation is a common observation in mice with impaired host defenses due to ablation of genes such as MyD88 (Slack et al., 2009). As the intestinal mucosa is under continuous immune surveillance that results in migration of DC laden with commensal bacteria to the MLN (Fukata et al., 2005; Macpherson and Uhr, 2004), it is improbable that increased bacteria within the colonic mucosa occurred without a paralleled increase in MLN. Altered host defenses can also perturb the composition of the microbiota (Lupp et al., 2007). As an example of this, Nod2-/- mice have reduced antibacterial peptides, increasing specific commensal populations (Petnicki-Ocwieja et al., 2009). We noted no significant effect on the microbiota in Tslp-/- or Crlf2-/- compared to WT mice. In keeping with this, no difference was observed in the REG family of c-type lectins that function as antimicrobial peptides (Pickert et al., 2009). We also noted no deficiency in the recruitment of neutrophils to the colon following DSS colitis or into the peritoneum following challenge with pathogen associated molecular patterns. Critically, mouse norovirus and Helicobacter sp. were not detected in our mice, as both can exacerbate the severity of intestinal inflammation (Cadwell et al., 2010). Thus, it is unlikely that Tslp-/- mice have a defect in host defense, and instead the contribution of microbiota in different animal facilities (Friswell et al., 2010) may play a pivotal role in the differences between previous studies (Taylor et al., 2009).
Repair of the epithelium occurs in at least two phases; the rapid restitution phase, and a slower regeneration period. Whereas restitution involves spreading of epithelial cells to fill gaps in the damaged mucosa, regeneration requires cell proliferation, deposition of proteins including collagen around the wound site (Taupin and Podolsky, 2003), and contraction by myofibroblasts. Deposition of structural proteins is a tightly controlled process; dysregulation results in fibrosis or fistula development due to increased or insufficient ECM protein deposition respectively (Rieder et al., 2007). This process is regulated by the rate of collagen synthesis and the balance between proteolytic enzymes and their endogenous inhibitors. Highlighting that this balance was altered in DSS-treated Tslp-/- mice, reduced collagen was detected by histological staining, and Sircol assay without concomitant decreases in collagen mRNA expression. These data demonstrated that the reduced collagen deposition was due to enhanced, localized enzymatic activity resulting in collagen turnover. In accordance with a previous report (Castaneda et al., 2005), we observed increased MMP activity during overt colonic inflammation; however this occurred irrespective of genotype, indicating that the increased mortality is MMP independent. Degradation of collagen can also occur by NE (Zhu et al., 2001), an enzyme found in abundance in samples from patients with IBD (Langhorst et al., 2008). Colonic tissue assayed at the peak of disease demonstrated significantly enhanced NE activity in DSS-treated Tslp-/- compared to WT mice, without increased NE expression. Increased enzymatic activity without a paralleled increase in NE expression indicated a defect in an endogenous regulator.
A profound reduction in SLPI mRNA and protein were observed in the colon of Tslp-/- mice post-DSS. This inability to increase SLPI expression following DSS colitis and control NE activity decreased mucosal healing, and ultimately increased mortality. Indeed, inhibition of NE with a selective inhibitor, or rSLPI significantly improved survival in Crlf2-/- mice. In support of this hypothesis for regulation of wound healing by SLPI, increased NE activity delayed recovery in Slpi-/- mice following dermal injury (Ashcroft et al., 2000). SLPI influences the balance between host defense and tissue repair by regulating NE-mediated conversion of progranulin to granulin. These proteins exert opposing physiological functions; progranulin induces epithelial cell proliferation, and reduces neutrophil activity, granulin inhibits proliferation, increases IL-8 production (human cell lines), neutrophil recruitment and activation (Kessenbrock et al., 2008; Zhu et al., 2002). To delineate the role of enhanced NE activity in the failure to recover from DSS-colitis in Tslp-/- mice, we assessed the impact of NE inhibition during the recovery phase. IEC proliferation was significantly reduced 8 days post-DSS that was normalized after 4 days of SSR69071 treatment. This inhibition of NE activity restored nascent collagen protein and PGRN in DSS-treated Tslp-/- mice beginning 2 days after treatment. In accordance with previously published in vitro data, where r PGRN increased epithelial cell proliferation after 3 days (Zhu et al., 2002), IEC proliferation occurred after 4 days of treatment following inhibition of NE and stabilization of the PGRN protein. Together our data indicate that TSLP-induced SLPI dictates the balance between host defense and tissue regeneration, and consequently aids mucosal healing following colonic inflammation by modulating NE activity.
Although TSLP has been primarily thought to exert physiological effects through TSLPR on B-cells, T-cells, and DC, the contribution of this receptor on non-hematopoietic cells was evaluated in the resolution of colitis. By using BM chimeras, TSLPR on non-hematopoietic cells was sufficient for SLPI expression and survival. Supporting a role for non-hematopoietic cells, functional TSLPR is expressed on the surface of primary IEC, and a mouse IEC line. Stimulation through this receptor initiates activation of both ERK and AKT signaling, and SLPI expression. Resident colonic fibroblasts or myofibroblasts would seem to be excluded from this population of cells that can directly respond to TSLP, as they lack TSLPR. In assessing TSLP and SLPI production in situ, both proteins were prominently co-expressed in colonic IEC with some SLPI expression localized to cells in the lamina propria. As previously reported, these are likely macrophages (Nakamura et al., 2003) and DC (Samsom et al., 2007), although it would appear that IEC are the predominant source of colonic SLPI. Therefore TSLP can act on IEC in an autocrine manner that is beneficial to intestinal restitution and healing.
Here we propose a model whereby TSLP acts as a key facilitator in wound repair following intestinal injury. As TSLP expression by IEC is enhanced following TLR stimulation, or pro-inflammatory cytokines (Zaph et al., 2007), it is conceivable that following DSS-induced epithelial injury, exposure to the luminal contents increases TSLP expression in adjacent IEC. TSLP produced in this manner can act directly on epithelial or stromal cells to induce SLPI gene expression, creating a local tissue environment that favors healing and resolution of inflammation. This is a novel and non-redundant role of TSLP in the colonic mucosa, and one that does not involve restraining of Th1 or Th17 immune responses.
Tslp-/- mice were generated by a conventional targeting strategy, by inserting an eGFP and a neomycin cassette into exon 2 five codons after the transcription start of the TSLP gene. Insertion of the targeting vector deletes a portion of exon 2, and all of exons 3 and 4. Deletion of TSLP in Tslp-/- mice was confirmed by southern blotting, and PCR with primers spanning the junction between exon 2 and the eGFP-NeoR insert (data not shown). Mice were back-crossed to C57BL/6 for at least 10 generations, with both males and females used. TSLPR deficient (Crlf2-/-) mice were described previously (Al-Shami et al., 2005). Rag1-/- mice (Jackson) were backcrossed to C57BL/6 mice for at least 10 generations.
Blood collected by cardiac puncture was added to an EDTA containing tube, and was analyzed using a HEMAVET instrument (Abaxis, Union City CA).
Colitis was induced by addition of dextran sodium sulphate (DSS; 36,000-50,000 MW, MP Biomedicals, Solon OH) to the drinking water for 5 days, followed by 3 days of normal water. Blood was collected from anaesthetized mice by cardiac puncture.
In subsets of experiments, mice were treated with the NE inhibitor SSR69071 or vehicle (DMSO) by oral gavage twice daily (3mg/kg, Tocris, Ellisville, MS). Mice were treated with rSLPI (R&D systems) by gavage in 0.1M Sodium borate twice daily (Sigma-Aldrich, St. Louis, MO)
Colitis was induced by the transfer of FACS sorted CD4+CD45RBhi effector T-cells (4×105 cells/mouse). Protection from colitis and hence the capacity of immuno-regulatory CD4+CD45RBlo (2×105 cells/mouse) cells was assessed following co-transfer into recipient C57BL/6 Rag1-/- mice.
Colitis was assessed by macroscopic and histological scores published previously (Reardon et al., 2001). All procedures were approved by the University Health Network animal care committee.
Mice were immunized with OVA (Grade V, Sigma-Aldrich) in CFA (Fisher Scientific Ottawa, ON) or Alum AlK(SO4)2 (Sigma-Aldrich) by i.p. injection as described previously (Yip et al., 1999). Serum was obtained 2 weeks later, with OVA specific IgG1, and IgG2a determined by ELISA (Southern Biotechnology, Birmingham AL).
Mice were injected i.p. with 100ng LPS (Sigma Aldrich), or 100μg of MDP (Invivogen, San Diego, CA). The peritoneal cavity was washed with PBS, recovered cells were counted, stained and enumerated by flow cytometry.
Male recipient mice were lethally irradiated (Gammacell 40, MDS Nordion, Ottawa ON) by two exposures of 550 rads prior to receiving 1×106 BM cells from Crlf2-/- mice by i.v. and were maintained on Baytril for 2 weeks.
Tissues were formalin fixed and processed according to standard histological protocols. Staining with PAS reagent or masson's trichrome were conducted according to manufacturers' instructions (Sigma-Aldrich). Sections were stained for Ki67 (Lab Vision, Fremont, CA), EMR1 (Life Span Biologicals, Seattle, WA), NIMP (Santa Cruz Biotechnology, Santa Cruz, CA), Cleaved Caspase3 (Cell Signaling Technology, Danvers, MA) using a standard protocol, for development with DAB substrate (VectorLabs, Burlington, ON)
Formalin fixed paraffin embedded tissues were stained using a previously published protocol (Kaniuk and Brumell). Briefly, rehydrated sections were subjected to antigen retrieval (Citrate buffer: pH6.0, 90°C, 30min). Slides were blocked for 1h and incubated with anti -E-cadherin (ECMbiosciences, Versailles, KY), -SLPI, -TSLP (Life Span Biologicals,) overnight, washed and incubated with appropriate conjugated secondary antibodies Alexafluor-488, -546, -647 antibodies (1h, RT, Invitrogen). Images were acquired with a Fluoroview 1000 confocal microscope (Olympus, Markham ON)
Tissue was homogenized in Trizol (Invitrogen, Burlington ON) using a TissueLyser (Qiagen). RNA was precipitated with LiCl2, and cDNA synthesis was performed (iScript kit; Bio-Rad, Mississauga, ON). Previously published primers (Peterson et al., 2010) or primers selected from the primerbank (Spandidos et al., 2008) (Table S2) were used in qRT-PCR with SyberGreen master mix (ABI). Data was analyzed by the -2ΔΔCT method normalizing to gapdh.
Microbial DNA was extracted using a QIAamp DNA stool kit (Qiagen) according to manufacturer's specifications. DNA eluted from the spin columns was quantified, and qRT-PCR performed using microbial primers to 16s rDNA (Table S2). Data is presented as percentage relative to eubacteria.
Spleens and lymph nodes were dissociated mechanically, and red blood cell lysis was performed on spleens with ACK buffer (Sigma-Aldrich). For splenic DC, spleens were injected with collagenase/DNase (37°C, 30min). Cells were counted, and 1×106 cells were stained with anti-; B220, CD3, CD11b, CD40, CD80, CD86, F4/80, IgA, MHCII (BD Biosciences, Mississauga, ON), CD11c, pDCA-1 (eBioscience, San Diego, CA), TSLPR (R&D systems, Minneapolis, MN).
Intracellular staining for Foxp3 and IFNγ were conducted according to manufacturers' instructions (Biolegend and eBioscience respectively). Phosphoflow was performed using anti-pSTAT5, -pERK1/2 (BD Biosciences), or -pAKT (Cell Signaling Technology). Vimentin (Cell signaling Tech.) and αSmooth Muscle Actin (Milipore) were labeled with Alexafluor -405 and -488 respectively with a Zenon labeling kit (Invitrogen).
Data was acquired on a CantoII (BD Biosciences) and analyzed with FlowJo (Tree star, Ashland OR).
Serum cytokine concentrations were evaluated using a mouse Th1/Th2/Th17 kit (BD Biosciences) according to the manufacturer's instructions. Samples were acquired on a FACsArray and analyzed with CBA software (BD Biosciences).
Mice were starved overnight with free access to water and gavaged the next morning with 4.4KDa FITC-Dextran (Sigma-Aldrich). Blood was collected 4h post-gavage as described above, and read on a Flex Station3 (Molecular Devices, Sunnyvale, CA).
Spleens or MLN were weighed, and homogenized in 500ul of sterile PBS containing 0.1% Triton-X100 (w/v). Samples were serially diluted in PBS, plated on 5% Sheep's blood agar, and grown overnight.
Snap frozen tissue was homogenized by TissueLyser (Qiagen). MMP activity was assayed using a generic MMP assay kit (AnaSpec) according to manufacturer's instructions. Neutrophil elastase activity was assessed in a similar fashion, except that Rhodamine 110, bis-CBZ-L-alanyl-L-alanyl-L-alanyl-L-alanine amide (Invitrogen) was used as the substrate.
Collagen was quantified using the Sircol assay according to manufacturer's instructions. Colonic protein was incubated with Sirius red with gentle agitation (0.5h, RT). Dye bound collagen was centrifuged, washed, and re-solubilized with alkaline buffer. Collagen concentration was determined by measuring the absorbance at 555nm (Flex Station).
The mouse intestinal epithelial cell line mICcl2 (Bens et al., 1996) was kindly provided by Dr. Philpott (University of Toronto). Cells were stimulated with recombinant mouse TSLP (R&D systems).
Intestinal tissues were opened, washed in PBS, and cut into 0.5mm pieces. Tissues were incubated with cell recovery solution (BD Biosciences) (Perreault and Beaulieu, 1998). Epithelial cells were stained immediately for flow cytometry or enriched by negative selection using the BD Imag system (BD Biosciences) to remove CD11b+, CD11c+, CD3+, B220+ cells. For phosphoflow, cells were incubated in optiMEM for 15min and re-stimulated with 50ng/mL of rmTSLP for 30min prior to analysis as described above.
Fibroblasts and myofibroblasts were dissociated by collagenase/dispase as described previously (Evans et al., 1992; Fruchtman et al., 2005). Cells were surface stained for E-cadherin, CD11b, CD11c, F480, CD3, B220, fixed with formaldehyde (4% v/v) and treated with cytoperm (BD Biosciences) for intracellular staining of αSMA and vimentin as described above.
Western blotting was performed according to a standard protocol. Blots were incubated with anti-SLPI (Santa Cruz biotechnology, Santa Cruz, CA), -Foxo3A, -Phospho-Foxo3A (T32), -Phospho-AKT, -Phospho-ERK, -BCL2, (Cell Signaling Technology) and −AKT, -ERK, −Tubulin, -MCL (Santa Cruz Biotechnology, Santa Cruz, CA).
All statistics were calculated in Prism 5.0 (GraphPad, La Jolla, CA), and are two-tailed Student's t-test, or ANOVA with Tukey post-test as indicated in the respective figure legends. All data are presented as mean ± standard deviation.
This work was supported by an operating grant from the Canadian Institutes of Health Research (TWM). CR is a recipient of a Cancer Research Institute/Irvington Institute Fellowship. ML is a recipient of IZKF Erlangen TP A46. Additional support was from AR056113 and AI068731 grants from the NIH (SFZ).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.