Supported by a new emerging team grant on IBD from the Canadian Institutes for Health Research (IIN 84037) and a Grant-in-Aid of research from the Crohn's and Colitis Foundation of Canada. G.H. holds a Michael Smith Foundation for Health Research (MSFHR) postdoctoral fellowship, Y.Y. is supported by a CIHR Transplantation Research Award, MKL holds a Canada Research Chair in Transplantation. C.A.P. holds a Canada Research Chair in Immune Regulation. Core support for flow cytometry sorting provided by Lixin Xu was funded by the Immunity and Infection Research Centre MSFHR Research Unit.
Toll-like receptor 5 (TLR5) is implicated in the innate and adaptive immune responses that are associated with inflammatory bowel disease (IBD). In humans TLR5 is expressed on CD4+ T cells and costimulation with flagellin potentiates effector and regulatory T cell responses. The aim of this study was to determine the role of TLR5 in CD4+ T cell subsets versus other cells in induction of disease in a model of T cell-dependent colitis.
TLR5 expression on CD4+ T cells was assessed by real-time reverse-transcriptase polymerase chain reaction (RT-PCR). Wildtype (WT) or TLR5-deficient (5−/−) CD4+ T conventional cells (Tconv) and T regulatory cells (Treg) were compared for their ability to induce and suppress T cell transfer colitis, respectively. In addition, the role of TLR5 expression in recipient mice was analyzed.
TLR5 is preferentially expressed on mouse Treg compared to Tconv, although expression levels were low. The colitogenic capacity of WT and 5−/− Tconv was found to be similar and Treg from WT or 5−/− donor animals both prevented T cell transfer colitis in TLR-competent hosts. TLR5 deficiency in recipient mice, however, did affect the disease process, as T cell receptor-β (TCRβ) 5−/− recipients had decreased weight loss compared to TCRβ recipient mice when WT Tconv were used.
TLR5 expression on T cells is not required for induction of or protection from T cell-dependent colitis. Expression of TLR5 in non-T cells has a pathogenic role, since TLR5 deficiency in recipient mice protects against weight loss induced by WT T cells. (Inflamm Bowel Dis 2011;)
Ulcerative colitis (UC) and Crohn's disease (CD), the two forms of human Inflammatory bowel disease (IBD), are inflammatory conditions that affect the gastrointestinal tract and have unclear etiology and no definitive cure.1 Different combinations of genetic and environmental factors have been implicated in the development of IBD. One current mainstream hypothesis is that genetic defects or environmental insults alter gut epithelial and/or immune cell function, predisposing individuals to enhanced exposure and responsiveness of the mucosal immune system to commensal microbiota.2 This abnormal responsiveness then drives ongoing inflammatory destruction of intestinal tissues.
Recent work has implicated the bacterial protein flagellin, the major structural component of bacterial flagella, and its receptor Toll-like receptor 5 (TLR5) in regulation of innate and adaptive immune responses that are associated with IBD. About half of CD patients have abnormally elevated levels of anti-flagellin antibodies, which correlate with particular subtypes of severe disease.3, 4 In addition, a TLR5 stop codon polymorphism, which prevents TLR5 signaling, reduces anti-flagellin adaptive immune responses and is protective against CD in certain ethnic groups.5
The role of TLR5 signaling in IBD is also evident from studies in murine IBD models. TLR5-deficient (5−/−) mice show increased sensitivity to dextran sulfate sodium (DSS)-induced colitis, which causes epithelial damage,6 and develop spontaneous colitis in some animal facilities.7 The cell types that express TLR5 and regulate mucosal immunity via a TLR5-dependent mechanism, however, remain unclear. In humans, TLR5 is reportedly expressed on numerous cell types, such as epithelial cells,8, 9 monocytes,10 neutrophils,11 dendritic cells (DCs),10 endothelial cells,12 and T cells.10 In contrast, in mice TLR5 protein and/or mRNA may be restricted to subsets of lamina propria (LP) DCs,13 epithelial cells of the colon,14 osteoclasts,15 alveolar macrophages,16 and T cells.17, 18 Some evidence suggests that TLR5 expression on DCs is important for gut homeostasis since TLR5+CD11c+ and not TLR5−CD11c+ LP DCs induce the differentiation of T helper (Th)1/Th17 CD4+ cells and IgA+ plasma cells.13 A protective effect of TLR5 on epithelial cells may also explain the susceptibility of 5−/− mice to DSS colitis.6 In addition, there is growing evidence that expression of TLRs other than TLR5 on T cells may protect against colitis. For instance, deletion of TLR4 or 9 increases the colitogenic capacity of conventional CD4+ T cells, suggesting that TLR signaling may inhibit pathogenic T cell responses.19, 20 However, a direct role for TLR5 on T cells in murine colitis has never been examined.
We have previously shown that engagement of TLR5 by flagellin on human T regulatory cells (Treg) enhances their suppressive capacity and expression of Foxp3.10 Whether TLR5 is similarly expressed on murine CD4+Foxp3+ Treg and whether TLR5 expression on T cells plays a functional role is unknown. To gain more insight into how cell-type-specific expression of TLR5 affects colitis, we investigated whether a deficiency in TLR5 in pathogenic or regulatory donor CD4+ T cells versus in recipient mice affects colitis in a T cell transfer model. We found that TLR5 expression in T cells is dispensable for T cell transfer colitis and that TLR5 expression in recipient mice induces weight loss independent of the local immune response and known systemic mediators of wasting disease.
MATERIALS AND METHODS
Foxp3-eGFP reporter (generation F11)21 and T cell receptor-β (TCRβ) (generation F12) mice22 were obtained from Jackson Laboratory (Bar Harbor, ME) and TLR5 5−/− mice (generation F8)23 were kindly obtained from A. Aderem (University of Washington). Mice were crossed and bred in-house and maintained under barrier conditions at the animal facility of the Jack Bell Research Centre. The TCRβ 5−/− mice were generated by crossing TCRβ mice with 5−/− mice and intercrossing the heterozygous offspring of this cross; 6–8-week-old female and male mice were used. Cells from male or female Foxp3-eGFP donor mice were transferred into male TCRβ or TCRβ 5−/− mice. For female TCRβ or TCRβ 5−/− mice, only female donor cells were used. The experiments described in this study were approved by the University of British Columbia (UBC) Animal Care and Use Committee.
T cell Transfer
Following red blood cell lysis of spleens, CD4+ T cells were obtained by negative enrichment (Stemcell, Vancouver, BC), stained with a CD4-APC monoclonal antibody (mAb) (BD Biosciences, San Jose, CA), and sorted into T conventional (Tconv) and Treg cells on FACS aria II high speed cell sorter (BD Biosciences) on the basis of eGFP fluorescence. 0.5 × 106 Foxp3−CD4+ Tconv from Foxp3 eGFP or Foxp3 eGFP 5−/− mice were injected intraperitoneally (IP) with or without 0.25 × 106 Foxp3+CD4+ Treg from the same animals.
Mice were monitored for clinical signs of colitis including weight loss, diarrhea, and bleeding. Mice were euthanized by CO2 asphyxiation followed by cervical dislocation when they lost >15% of their initial body weight or if they displayed severe clinical signs. The percentage weight change was calculated relative to the weight at the start of the experiment. Serum was collected via cardiac puncture and stored at −80°C prior to analysis. Mesenteric lymph nodes (mLN), spleen, and intestines were grossly examined, excised, and kept in cold phosphate-buffered saline (PBS) for subsequent procedures (as described below).
Lamina Propria Mononuclear Cell (LPMC) Isolation
For isolation of LPMC, colons were cut longitudinally, washed with PBS followed by PBS/1% bovine serum albumin (BSA), and incubated in a shaking incubator at 37°C in prewarmed RPMI 1640 containing 10% heat inactivated fetal calf serum (FCS) and 5 mM EDTA for 15 minutes (to remove epithelial cells). This last step was performed twice, discarding the supernatant and vigorously vortexing after the incubation. The remaining tissue was incubated in prewarmed RPMI1640/10% FCS and 15 mM HEPES and cut into 1 mm pieces followed by 2 × 30 minutes digestion using the same medium with 0.2 mg/mL type VIII collagenase (Sigma Aldrich, Oakville, ON) in a 37°C shaking incubator. After both incubation steps the digested tissue pieces were vortexed and LP cells were obtained and kept on ice. LP cells were spun down at 2000 rpm, resuspended in Percoll 30 (30% Percoll 100 in PBS/0.1%BSA), layered on a 40-75 Percoll gradient (Amersham Biosciences, Uppsala, Sweden), and centrifuged at 1800 rpm for 20 minutes (without brake and acceleration) (Percoll 100 = 9 vol Percoll + 1 vol PBS 10×). The LPMC were collected from the 40% to 75% interface, washed twice with PBS/0.1%BSA, and spun at 2000 rpm prior to use in subsequent assays.
Antibody Staining and Flow Cytometry
To prepare single-cell suspensions from mLN and spleens, organs were passed through 70-μm nylon mesh. After centrifugation, splenocytes were treated with ammonium chloride solution (Stemcell Technologies, Vancouver, BC, Canada) for 5 minutes and washed with PBS. Surface Ab staining was performed at 4°C in PBS/2% FCS with or without 0.1% sodium azide (FACS buffer) with CD4 Ab. For intracellular cytokine staining, 0.5 × 106 cells were resuspended in culture medium (RPMI1640 containing 10% heat inactivated FCS, 100 U/mL penicillin, 100 μg/mL streptomycin, 2 mM L-glutamine, and 5 μM β-mercaptoethanol) and stimulated with 50 ng/mL phorbol myristate acetate (Calbiochem, Mississauga, ON) and 1 μg/mL ionomycin (Sigma Aldrich) for 4 hours, with the addition of 10 μg/mL brefeldin A (Sigma Aldrich) after 1 hour. Cells were subsequently fixed with 4% formaldehyde in FACS buffer (fixation buffer) for 10 minutes and incubated with fluorescently labeled interferon-γ, tumor necrosis factor-α, and interleukin (IL)-17 Abs (BD Biosciences) in FACS buffer containing 1% saponin (permeabilization buffer). Unbound Abs were washed away using the permeabilization buffer and cells were resuspended in FACS buffer. Ki-67 staining was performed using Fix/Perm solution (eBioscience, San Diego, CA) according to the manufacturer's instructions.
A plasmid expressing His-tagged Fla2 flagellin from Lachnospiraceae bacterium A4 was generously provided by C. Elson (University of Alabama).24 Fla2 was prepared in BL21 (DE3) pLysS and purified by metal affinity chromatography as described.25 Plates were coated with 1 μg/mL Fla2 in PBS overnight at 4°C, then blocked with 1% BSA in PBS/0.05% Tween-20 for 2 hours at 37°C. Diluted mouse sera in 0.1% BSA in PBS/0.05% Tween-20 were added for 2 hours at room temperature (RT), followed by a horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG (Southern Biotech, Birmingham, AL) for 1 hour at RT. TMB substrate was added and the color reaction was read at 650 nm.
The entire mouse intestine was excised and 0.5 mm segments of the small intestine, cecum, proximal, mid, and distal colon were fixed in 10% buffered formalin, embedded in paraffin, and 5-μm sections were stained with hematoxylin and eosin. The sections were scored for histologic injury by a reader blinded to the identity of the mice. The scoring system was a modification of Kullberg et al.26 Each section was given a score from 0–4 for epithelial injury, crypt abscesses, and inflammatory infiltrate into the lamina propria, submucosa, and serosa, to give a maximal score for each section of 20. The scores for the four sections were added to give a total injury score (0–80). The highest total injury score observed in this study was 45.
Real-time Polymerase Chain Reaction (PCR)
Total RNA was extracted using RNeasy plus mini kit (Qiagen, Mississauga, ON), eliminating residual genomic DNA with gDNA eliminator spin columns. Then 100 ng of purified RNA was reverse-transcribed with the qScript cDNA SuperMix system (Quanta Biosciences, Gaithersburg, MD). The resulting cDNA was amplified using Quantitect SYBR Green PCR master mix (Qiagen) on a 7500 Fast Real-Time PCR System (Applied Biosystems, Mississauga, ON) according to the comparative (ddCt) method. The primer pairs used were for Tlr5 (CGCACGGCTTTATCTTCTCC, GGCA AGGTTCAGCATCTTCAA) and 18S rRNA (CAAGACGGACCAGAGCGAAA, GGCGGGTCATGGGAATAAC).
Cytokine Detection by Luminex
GM-CSF, IFN-γ, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-10, IL-12p40/p70, and TNF-α concentrations in mouse serum were determined by Luminex multiplex bead assay according to the manufacturer's instructions (Invitrogen, mouse cytokine magnetic 10-plex).
To determine whether differences between experimental groups were significant, a two-tailed Mann–Whitney U-test was performed using GraphPad Prism 4 for Macintosh (San Diego, CA). P < 0.05 (*) was considered significant. (**P < 0.01, ***P < 0.001).
Expression of TLR5 on Tconv and Treg Cells
It has previously been reported that TLR5 is highly expressed on mouse CD45RBlowCD4+ T cells, which are known to be enriched in Tregs.17, 18 In order to ask whether TLR5 is highly expressed on Tregs, defined as Foxp3+CD4+ T cells, we investigated the relative expression levels of TLR5 by sorting of Foxp3+ (eGFP+) or Foxp3− (eGFP−) CD4+ T cells from Foxp3-eGFP reporter mice to obtain homogeneous populations of cells (>97% purity, Fig. 1A). mRNA isolated from splenocytes and LPMC of the same mice served as a reference and positive control, respectively. We found that TLR5 mRNA is preferentially expressed on Foxp3+CD4+ Treg compared to Foxp3−CD4+ Tconv (Fig. 1B), although the mRNA expression was considerably lower than in LPMC.
Role of TLR5 in CD4+ T Cells in Development and Protection from Colitis
We next tested whether expression of TLR5 on T cells affected the development of disease in a T cell transfer model of colitis. In these experiments, donor T cell populations were obtained from WT or 5−/− (Foxp3-eGFP reporter) mice and transferred in TCRβ−/− recipient mice. TCRβ mice are B cell sufficient, enabling monitoring of anti-flagellin Abs as a disease correlate. Recipient mice were injected IP with WT Tconv, 5−/− Tconv, WT Tconv+WT Treg, or WT Tconv+5−/− Treg, or left untreated (control) and followed for the development of colitis. Mice injected with either WT or 5−/− Tconv developed colitis with similar kinetics of weight loss and similar histology scores, whereas control mice and mice injected with WT Tconv+WT Treg or WT Tconv+5−/− Treg continued to gain weight, without histological signs of colitis. We did observe a slight difference in the histology scores between mice injected with WT or 5−/− Tconv; however, this was not significant (Mann–Whitney U-test, P = 0.13) (Fig. 2A,B).
To see if there were differences in the adaptive immune response, we focused on the B cell and CD4+ T cell compartments. For the former, we analyzed the levels of serum IgG specific for the A4-Fla2 flagellin, a bacterial antigen associated with colitis.27 The A4-Fla2 IgG levels were increased in mice that developed colitis compared to controls, and there was no difference between the recipients of WT or 5−/− Tconv in this respect. Interestingly, mice injected with WT Tconv+WT Treg or WT Tconv+5−/− Treg showed decreased levels of A4-Fla2 IgG compared to mice transferred with WT Tconv alone, but these levels were still higher than in the control animals and did not differ between the recipients of WT Treg or 5−/− Treg (Fig. 2C).
The CD4+ T cell response also did not markedly differ between the recipients of WT Tconv or 5−/− Tconv and WT Tconv+WT Treg or WT Tconv+5−/− Treg. Similar absolute numbers of CD4+ T cells were found in LP and mLN (Fig. 2D) with comparable percentages proliferating, as measured by Ki-67 expression. Although the difference in the percentage of Ki-67+ cells among CD4+ T cells between recipients of WT Tconv and 5−/− Tconv was significant in the LP, it was of small magnitude (Fig. 2E).
In agreement with the similar absolute number of CD4+ T cells between the recipients of WT Tconv or 5−/− Tconv and WT Tconv+WT Treg or WT Tconv+5−/− Treg, the absolute number of proinflammatory cytokine-producing cells between Tconv and Treg treated groups was not different in LP and mLN (Fig. 2F). Upon adoptive transfer of Tconv to immunodeficient mice, part of the Tconv cells will convert to Treg via a process called peripheral conversion.28 The extent of peripheral conversion, however, was not different between these groups, since we retrieved similar absolute numbers of CD4+ eGFP+ T cells in the analyzed organs (Fig. 2G). Overall, these results show that expression of TLR5 on T cells has no effect on the development and protection of colitis in the T cell transfer model.
Role of TLR5 in Recipient Mice in Local Gut Inflammation
Having established that expression of TLR5 on Tconv or Treg is not required for their colitogenic or regulatory function, respectively, we next investigated the role of TLR5 in recipient mice in this T cell-dependent model of colitis. Although it has been reported that 5−/− mice develop spontaneous colitis7 and metabolic syndrome,29 we did not observe differences between unmanipulated TCRβ and TCRβ 5−/− mice in terms of weight gain or colitis score (Fig. 3A,B).
To determine the effect of TLR5 expression in the recipient mice, TCRβ mice and TCRβ 5−/− mice were injected IP with WT Tconv, WT Tconv+WT Treg, or left untreated. TCRβ 5−/− recipients of WT Tconv+WT Treg were completely protected against weight loss and did not show histological signs of colitis (data not shown). More remarkably, the TCRβ 5−/− recipients of WT Tconv were significantly protected from weight loss compared to TCRβ recipients. However, compared to controls the TCRβ 5−/− mice injected with WT Tconv did lose weight and show clinical signs of colitis (Fig. 3A).
The striking difference in weight loss, however, did not correlate with protection from histological signs of colitis or the ensuing B- and T cell responses. No differences were observed between TCRβ and TCRβ 5−/− recipient mice after blinded histologic scoring (Fig. 3B). Even after unblinding, a second review of slides did not uncover any consistent differences in the type or intensity of tissue injury or inflammation. The low level of intestinal inflammation observed in TCRβ and TCRβ 5−/− without T cell transfer is in agreement with an earlier report on the spontaneous development of mild colitis in T cell receptor-deficient mice.22
A4-Fla2 IgG antibodies in the serum were similarly increased in the two groups of mice that received WT Tconv compared to their respective controls (Fig. 3C) and we found corresponding absolute numbers of CD4+ T cells (Fig. 3D), CD4+ eGFP+ T cells (Fig. 3E) and cytokine producing CD4+ T cells (Fig. 3F) in mLN and LP in the transferred mice. The percentage of Ki-67+ cells of CD4+ T cells was also not different between the two groups (Fig. 3G). Thus, despite a lack of protection against local gut inflammation, the TCRβ 5−/− recipient mice were significantly protected against weight loss.
Role of TLR5 in Recipient Mice in Systemic Inflammation
Besides inducing an inflammatory response in the gut, transfer of Tconv in immunodeficient mice can result in systemic effects, such as splenomegaly and elevated levels of inflammatory serum cytokines,30 transfer of Tconv in TCRβ, and TCRβ 5−/− recipients led to increased splenic weight compared to control animals. However, we did not observe differences between the spleen mass of TCRβ and TCRβ 5−/− mice transferred with WT Tconv at the end of the experiment (Fig. 4A).
An increase in systemic proinflammatory cytokines such as TNF-α, IL-6, and IL-1β has been shown to cause a loss of appetite, resulting in wasting disease in various diseases and experimental disease models.31–33 To investigate whether the levels of these or other cytokines differed between TCRβ and TCRβ 5−/− recipient mice, amounts in the sera were measured at the end of the experiment. The concentrations of TNF-α and IL-1β were low and not significantly different between the two recipient groups. We did observe small, but significant, increases in the concentrations of IL-6 and IL-12p40/p70 in the sera of TCRβ 5−/− recipient mice compared to TCRβ recipient mice (Fig. 4B). The IL-6 and IL-12 concentrations in individual mice, however, did not correlate to their percentage weight change (data not shown). These modest differences in serum cytokines are therefore not a likely explanation for the observed weight loss differences. This suggests that mechanisms apart from local and systemic inflammation are responsible for the observed differences in weight loss.
DISCUSSION AND CONCLUSIONS
Compelling evidence from other studies suggests that TLR signaling is important for the maintenance of gut homeostasis during health, but contributes to immune dysregulation during the pathogenesis of IBD.34, 35 To our knowledge our study is the first to look at the effects of TLR5 deficiency in T cells and recipient cells on T cell transfer colitis. We show that TLR5 expression on T cells is dispensable for the induction and prevention of T cell transfer colitis, but that expression in the recipient plays a role in the induction of weight loss, but not colitis.
TLR signaling can influence gut homeostasis in various ways. Initial studies focused on the role of TLRs in macrophages and DC,36 and more recently, on their role in the intestinal epithelium and T cells. TLR signaling in intestinal epithelial cells leads to activation of NF-κB and is important for the maintenance of intestinal barrier integrity, thereby preventing intestinal inflammation.37, 38 The role of TLR expression on T cells in the context of IBD comes from various studies using the T cell transfer model. For example, although TLR2 is expressed on CD4+ T cells,17 its expression on T cells, or in the recipient, is dispensable for T cell-mediated colitis.39 On the other hand, there is a clear role for TLR9 expression on T cells in T cell-mediated intestinal inflammation. Transfer of TLR9-deficient CD4+CD62L+ T cells induced a more severe colitis upon transfer into SCID mice,20 suggesting TLR9 signaling normally dampens the colitogenic potential of T cells. Although to our knowledge the role of TLR9 engagement on Treg has not yet been tested in the T cell transfer model of colitis, TLR9 ligation would be expected to limit Treg suppression.40, 41
TLR5 and other TLRs signal through the adaptor protein MyD88, and the role of this protein in T cells has also been investigated in murine models of IBD. Although two papers18, 42 reported that MyD88 expression in T cells is required for their colitogenic capacity, Asquith et al43 did not find a similar requirement for MyD88, but instead reported the importance of expression of MyD88 by recipient cells for the development of colitis. The absence of an effect of TLR5 expression on T cells, which we observed in our study, is in line with the studies on the role TLR2 and the latter study on the role of MyD88, but contrasts the findings on the effect of TLR9 deficiency, as mentioned above. The reason for this inconsistency remains unknown. The absence of an effect of TLR5 deficiency in recipient mice on intestinal inflammation, while MyD88 deficient mice are protected against colitis, might point to compensatory signaling by other TLRs.
The cause of weight loss (wasting disease) in the T cell transfer model and in other models of colitis is not exactly known. The main hypotheses that have been put forward are that it is a consequence of poor absorption of nutrients by the inflamed gut44 or an effect of the systemic production of inflammatory mediators such as IL-1β, TNF-α, and IL-6 on appetite.31–33 It is, however, apparent from our study and from studies of others45–47 that weight loss does not necessarily correlate with the extent of local gut inflammation and that additional factors besides local gut inflammation play a role. We did not observe a difference in systemic levels of cytokines known to be involved in the regulation of appetite between WT and 5−/− recipient mice. Other systemic mediators might be responsible for the protection from weight loss in 5−/− recipient mice or, alternatively, TLR5 signaling may control body weight via an unknown mechanism. In support of the latter possibility, MyD88 has been implicated in the control of body weight and appetite via a cytokine-independent mechanism via effects on the central nervous system.48 Whether a similar mechanism plays a role downstream of TLR5 requires further investigation.
In conclusion, we found that although Tregs express high levels of TLR5, expression of this protein on Tconv and Tregs is dispensable for the disease pathogenesis in the T cell transfer model of colitis. In contrast, expression of TLR5 in the non-T cell compartment is required for rapid onset wasting disease. These findings point to a role of flagellin-TLR5 interactions in the maintenance of body weight during colitis. Increased knowledge of the role of TLR signaling on the pathways that control body weight might lead to better treatment options in patients with wasting disease.