TH17, TH22 and TReg Cells Are Enriched in the Healthy Human Cecum

There is increasing evidence that dysregulation of CD4+ T cell populations leads to intestinal inflammation, but the regional distribution of these populations throughout the intestinal tract in healthy individuals remains unclear. Here, we show that TH17, TH22 and TReg cells are enriched in the healthy human cecum compared to the terminal ileum and sigmoid colon, whereas TH1 and TH2 cells do not significantly vary by location. Transcriptional profiling analysis of paired pinch biopsies from different regions of the intestine identified significant differences in the metabolic state of the terminal ileum, cecum, and sigmoid colon. An increased proportion of TH17 cells was positively associated with expression of resistin (RETN) and negatively associated with expression of trefoil factor 1 (TFF1). These results suggest that CD4+ T helper cells that are important in maintaining mucosal barrier function may be enriched in the cecum as a result of metabolic differences of the surrounding microenvironment.


Introduction
The human intestinal epithelium represents a critical interface between our entire internal milieu and the outside world [1]. Appropriate mucosal homeostasis depends on interaction between the commensal microbiota, the intestinal epithelium, and the mucosal immune system [2]. Of mucosal immune cells, T H 17 cells are particularly important in regulating intestinal immunity and have a complex role in human disease [1,2]. For example, T H 17 cells have been shown to be increased in active Crohn's disease [3] and are reduced in HIV infection [4]. Induced FoxP3 + regulatory T cells (T Regs ) are developmentally linked to T H 17 cells, with both requiring TGFb for differentiation [5]. Recent evidence suggests that these two cell populations may arise from the same naïve precursor cells and exhibit plasticity [6]. For example, FoxP3 + IL-17 + CD4 + cells have now been observed under several different inflammatory conditions [7,8]. One of the cytokines produced by T H 17 cells that is important in regulating mucosal barrier function is IL-22, which has been shown to promote epithelial proliferation and increase mucus production [9]. Not all T H 17 cells produce IL-22, however, and a subset of CD4 + helper T cells that produce IL-22 but not IL-17 have been identified in humans and are termed T H 22 cells [10]. While these different CD4 + T cell populations have been suggested to play important roles in regulating intestinal immunity, their exact functions in inflammation during human disease remain unclear, perhaps because an appropriate balance of these cells is required for healthy homeostasis.
In order to better understand how dysregulation among populations of CD4 + T cells in the intestinal lamina propria may be important during inflammatory conditions of the gastrointestinal tract, it is necessary to fully characterize these populations in healthy individuals. Specifically, the regional variations of these lymphocyte populations within the small and large intestine may provide important clues to their function in vivo. In order to address this question, we analyzed mucosal pinch biopsies collected from the terminal ileum (TI), ileocecal valve (ICV), appendiceal orifice (AO), and sigmoid colon (SC) from a healthy population presenting for screening colonoscopy. We were particularly interested in evaluating the cecal appendix given its implication in a number of inflammatory and infectious disease states, including ulcerative colitis and Clostridium difficile infection [11,12]. Here we report that the proximal colon (cecum) is enriched in T H 17, T H 22, and T Reg (but not T H 1 or T H 2) cell populations compared to the terminal ileum and distal large intestine in healthy individuals. We hypothesize that these differences are related to variations in metabolism and immune activation among different regions of the colon and small intestine.

T H 17 and T Reg cells are enriched in the healthy human cecum
Different regions of the intestinal tract perform diverse dietary functions and are colonized with distinct concentrations of commensal bacteria [13]. We therefore hypothesized that there are important differences in the distribution of lymphocyte populations in the healthy human gut. Based on mounting evidence of the immunologic function of the human cecum [12,14], our objective was to characterize the distribution of CD4 + T helper cell populations in the cecum relative to the terminal ileum and sigmoid colon in healthy individuals. We recruited 26 unique patients at average-risk for colon cancer who were undergoing screening colonoscopy and consented to participate in this study ( Table 1). The majority of the subjects were black men with an average age of 58.9+6.2 (Table 1) and therefore may not be representative for other gender, age, and ethnic groups. Hematologic and renal function parameters were within normal range for the majority of subjects. As shown in Table 2, the prevalence of adenomatous polyp tissue obtained at colonoscopy in this cohort (30.7%) is typical of the average prevalence in the age-adjusted general population [15]. To assess regional differences in CD4 + T helper cell populations, lamina propria mononuclear cells (LPMCs) were isolated from pinch biopsies obtained at the terminal ileum (TI), ileocecal valve (ICV), appendiceal orifice (AO), and the sigmoid colon (SC). The ICV and AO are anatomical landmarks of the most proximal portion of the large intestine (i.e. the cecum). We were particularly interested in comparing immune parameters of the cecum with the TI because it is a critical site of luminal sampling and is rich in mucosal associated lymphoid tissue [16]. Biopsies were additionally taken from the SC (defined arbitrarily as 20cm from the anal verge in all subjects), because it represents a hindgut-derived distal site in the human colon that is easily accessible in all subjects. Biopsy-derived LPMCs were stimulated with phorbol 12-myristate 13-acetate (PMA) and ionomycin and then assessed for a number of immune markers by polychromatic flow cytometry. The flow cytometry panels used for this analysis are shown in Tables 3 and 4. By gating on singlet, live, CD3 + CD4 + T cells ( Figure 1A), we compared the frequency of IL-17 producing CD4 + cells among different regions of the small and large intestine ( Figure 1B). There was a significantly elevated frequency ( Figure 1C) of IL-17 producing CD4 + cells in both parts of the cecum including the ICV (Median = 12.2%) and AO (10.4%), compared to the TI (8.2%) and SC (6.7%). This was somewhat unexpected since T H 17 cells have previously been reported to be enriched in the terminal ileum of mice [16,17].
To determine if there were significant non-CD4 + sources of IL-17 such as innate lymphocytes [18], we modified our gating strategy to first select for all of the IL-17 + cells (Figure 2A) in the live cell gate. Overall, there was an almost equivalent number of total IL-17 + cells among the different regions examined, with a slight decrease in the number of IL-17 + cells in the SC compared to the TI, ICV and AO ( Figure 2B). While a large majority of all the IL-17 + cells were CD3 + (Figure 2A), there were also some CD3 -CD56cells that were IL-17 + and these were significantly greater in numbers in the TI and SC ( Figure 2C). This indicates a possible inverse relationship in the cecum between IL-17 production by CD3 2 CD56 2 cells and IL-17 production by CD3 + CD56 2 cells, which were predominantly CD4 + (Figure 2A & 2B). Whereas there were more T H 17 cells in the cecum, there may be more innate lymphocytes producing IL-17 in the TI and the SC.
Previous studies of mucosal biopsies from HIV infected individuals have indicated that increased frequency of T H 17 cells may be accompanied by reduced frequencies of T Regs because of the developmental link between these two populations [4]. However, when we assessed the frequency of CD4 + CD25 + FoxP3 + cells, we found that T Regs were also significantly enriched in the cecum ( Figure 1D & E). The proportion of CD25 + FoxP3 + cells in the CD4 + compartment was significantly higher in the ICV (8.8%) and AO (8.5%) compared to the TI (4.8%) and SC (6.3%) ( Figure 1D & E). These results indicate that T H 17/T Reg balance is maintained across different locations of the healthy human intestine. Though T H 17 and T Regs are enriched in the proximal colon, there were no significant differences in the regional variation of the ratio of T H 17/T Regs in healthy individuals ( Figure 1F).   Because it was not possible to standardize the quantity of tissue that was processed for isolating LPMCs, the results here are presented as proportional data to total CD4 + cells. Absolute cell counts could vary based on variation in the size of processed pinch biopsies. CD4 + cell viability was relatively uniform across all locations, with the exception being a slight increase in CD4 + viability from the appendiceal orifice compared to the sigmoid colon (Table 5).

T H 22 and IL-17 producing T Reg cells are also enriched in the healthy human cecum
Recently, a population of CD4 + helper T cells that produces IL-22 but not IL-17 has been identified [10] and termed T H 22 cells. We previously found that IL-22 producing CD4 + cells were associated with helminth infection and disease remission in a patient with ulcerative colitis [19]. However, the normal variation and distribution of T H 22 cells across the intestine under baseline conditions is not known. In this study, we found that CD4 + IL-22 + IL17 2 T cells were also significantly elevated in the ICV (5.7%) and AO (5.9%), compared to the TI (3.1%) and the SC (3.6%), mirroring the results from T H 17 and T Reg cells ( Figure 3A & 3B). Since IL-22 is important in controlling bacterial infections in the gut, such as Citrobacter rodentium [9], the enrichment of T H 22 as well as T H 17 cells in the cecum may indicate the presence of potentially pathogenic bacteria in this region that are actively restrained by the mucosal immune response. Additionally, IL-22 producing T H 17 cells were also enriched in the cecum ( Figure 3C).
Finally, we examined the distribution of IL-17 producing T Regs , which have been reported to be elevated in the inflamed mucosa of Crohn's disease patients [7] as well as ulcerative colitis patients [8].     We found that CD4 + IL-17 + FoxP3 + cells showed a similar pattern to T H 17, T H 22 and T Regs , being significantly elevated in the ICV (1.4%) and the AO (1.5%) compared to the TI (0.5%) and the SC (0.4%) ( Figure 3D & 3E). Together, these results from lamina propria CD4 + T cells suggest that there may be more immune activation in the cecum compared to both the small intestine (TI) and the distal colon (SC). However, only those CD4 + cell populations that have been reported to be important for mucosal homeostasis were enriched in the cecum. When we examined T H 1 and T H 2 populations that produce IFNc and IL-4, there were no significant differences among the different sites that were biopsied in this study ( Figure 4).

Polycytokine production in lamina propria CD4 + T cells within the intestine
We next analyzed the different combinations of cytokines produced by CD4 + T cell populations among different regions of the gut. We utilized Boolean gate analysis ( Figure 5A) to expand our five-cytokine measurements (IL-17, IL-22, IL-4, TNFa and IFNc) into 32 unique functional combinations of cytokineproducing CD4 + cells. The pie charts shown in Figure 5B illustrate the averaged overall pattern of cytokine production among lamina propria CD4 + cells in the different regions of the gut. Each slice within the pie chart represents a specific combination of cytokine production (denoted in Figure 5C). Not surprisingly, the most common CD4 + cells are the ones not producing any of the five measured cytokines, followed by cells producing a combination of IFNc and TNFa and then IFNc and TNFa individually ( Figure 5B, C). No significant differences in the composition of cytokine-producing CD4 + cells among the various biopsy locations were found using permutation testing in SPICE. However, we noted that populations of CD4 + cells producing IFNc, TNFa, and IL-22 in combination, as well as TNFa and IL-22 together, appear to be more prevalent in the cecum compared to the terminal ileum and sigmoid colon.

Transcriptional profiling analysis of biopsies from different regions of the intestine
To determine if the microenvironment from which we isolated lamina propria CD4 + cells may provide an explanation for the regional variation in the T H 17, T H 22 and T Reg populations, we performed gene expression profiling experiments of paired pinch biopsies collected from the same regions that were analyzed by flow cytometry. RNA from a subset of matched samples (total N = 21, TI = 6, ICV = 5, AO = 5, SC = 5) from five subjects was isolated for DNA microarray analysis on the Agilent platform, in order to identify molecular signatures that may help to explain compositional differences in the CD4 + compartment. Unsupervised hierarchical clustering analysis was performed on 2,351 gene probes that were filtered to have a standard deviation of at least 1.0 between all 21 samples ( Figure 6A). This analysis showed that the gene expression profiles of biopsies from the terminal ileum clustered very distinctly from biopsies of the cecum (including AO and ICV) and the sigmoid colon (SC). The appendiceal orifice and ileocecal valve were less distinct from each other and clustered together separately from the sigmoid colon and the terminal ileum. Principal component analysis of this dataset ( Figure 6B) provided a similar picture, segregating the terminal ileum from the large intestine based on PC1, with the proximal colon (cecum, including AO and ICV) segregating from the sigmoid colon on PC2 ( Figure 6B).
We then conducted a supervised analysis using multiclass statistical analysis of microarrays (SAM) to identify genes that were differentially expressed among regions of the intestine. 2,079 unique gene probes were identified at a False Discovery Rate (FDR) of 0% that varied significantly based on biopsy location ( Figure 6C). Gene ontology (GO) analysis of this list of genes indicated that differences in lipid, protein, and carbohydrate metabolism varied most significantly among the different regions of the intestine that were biopsied ( Figure 6D). Of note, 304 genes were grouped into the immunologic processes category (Table 6).
We were particularly interested in the differential expression of mucin genes among different regions of the gastrointestinal tract because of their important role in mucosal barrier function [20]. Furthermore, based on microarray analysis of an individual who treated his ulcerative colitis with Trichuris trichiura, we have hypothesized that increased mucus production could mitigate inflammatory pathology and bacterial attachment in the colon [19,20]. We confirmed by real-time PCR on an expanded subset of samples (N = 40, TI = 11, ICV = 9, AO = 10, SC = 10) that MUC1 was most highly expressed in the sigmoid colon and least expressed in the terminal ileum, whereas MUC3 was most highly expressed in the terminal ileum and least expressed in the distal colon ( Figure 7). MUC1 and MUC3 are membrane bound mucins. MUC1 is a tumor-associated molecule frequently overexpressed in carcinomas [21] and MUC3 is the major component of the glycocalyx of the small intestine [22]. Variants of MUC3 have previously been associated with inflammatory bowel diseases [23]. MUC5B had a similar expression profile to MUC1, being most elevated in the sigmoid colon (Figure 7), but is a secreted polymeric mucin that can be dysregulated in the lung of subjects with idiopathic pulmonary fibrosis [24].
We also explored the differential expression of the peroxisome proliferator-activated receptor (PPAR) members of the nuclear hormone receptor superfamily because they are important regulators of inflammation [25]. We confirmed that PPARa was more highly expressed in the terminal ileum (Figure 7), whereas PPARc was more highly expressed in the cecum (AO and ICV). This was surprising because PPARc has been reported to inhibit T H 17 differentiation [26] whereas here we find that the cecum has relatively more T H 17 cells as well as higher expression of PPARc. Perhaps the anti-inflammatory effects of PPARc also require enrichment in the cecum in order to balance higher levels of immune activation. We performed additional two-class analysis by comparing the proximal colon to the distal colon and excluding the terminal ileum, with similar results (Figure 8).
Consistent with the hypothesis that there is more bacterial mediated activation in the cecum, we confirmed elevated expression of the anti-microbial peptide DEFB1 (Figure 7). DEFB1 is the only defensin to be expressed in the uninflamed colon and certainly plays an important role in mucosal defense against microbes [27]. Although we speculated that the increased expression of DEFB1 and the enrichment of CD4 + T cells important in regulating mucosal barrier function may be driven by increased bacterial attachment to the intestinal wall in the cecum, we did not find significant differences in bacterial load present in these biopsies by qPCR analysis of bacterial specific ribosomal 16S DNA (Figure 9). Perhaps a more detailed deep sequencing analysis may identify specific differences in bacterial species that may be enriched on the intestinal wall of the cecum compared to other regions of the gastrointestinal tract.

Correlating transcriptional profiling data with frequency of T H 17 and T Reg cells
Having generated a combination of flow cytometry and transcriptional profiling datasets from paired biopsy samples, we sought to identify genes that have expression levels that are directly correlated with the frequency of lamina propria T H 17 cells and T Regs in these healthy individuals ( Figure 10). We defined the frequency of T H 17 cells and T Regs obtained by flow cytometry as quantitative variable inputs to the statistical analysis of microarrays (SAM) algorithm in order to determine the strength of their relationship with gene expression values. Because of the relatively small number of samples, we used a conservative FDR of 0% as the cutoff for significance ( Figure 10A-B and Table 7). For T H 17 cells, we identified four gene probes that were positively correlated with T H 17 cell frequency (including resistin (RETN)) and 12 gene probes that were negatively correlated with T H 17 cell frequency ( Figure 10A). Interestingly, many of the probes negatively associated with T H 17 cell frequency encoded long non-coding RNA (lincRNA), for which the significance is unclear (Table 7). We did note with interest that trefoil factor 1 (TFF1) was among the most negatively associated genes with T H 17 cell frequency. The positive association of RETN (r 2 = 0.2275, P = ,0.01) and negative association of TFF1 (r 2 = 0.3308, P = ,0.0001) with T H 17 cell frequency were confirmed by RT-PCR analysis of additional biopsy samples (N = 40) from different regions of the intestine ( Figure 10C). For T Regs (CD4 + CD25 + FoxP3 + cells), we identified 22 gene probes that were positively correlated, but none that were negatively associated with T Reg frequency ( Figure 10B). Of these genes, Fc receptor-like protein 5 (FCRL5) was of particular interest because there was a strong association to T Reg frequency and a closely related FCRL family member was reported to be associated with T Reg function [28]. We confirmed by RT-PCR analysis that FCRL5 (r 2 = 0.3671, P = ,0.001) was indeed positively associated with T Reg frequency ( Figure 10C).

Discussion
In this study, we have systematically characterized regional differences of intestinal lamina propria CD4 + T cell populations by flow cytometry in a healthy population and identified gene expression patterns that correlate with their frequencies. We found the cecum to have significantly elevated frequencies of T H 17, T H 22, and T Reg lymphocyte populations (but not T H 1 or T H 2 cells). Interestingly, these are the CD4 + subsets that are most important in regulating intestinal homeostasis. While T H 17 cells are enriched in the terminal ileum in mice, we find here that the human cecum has a higher frequency of T H 17 cells than does the ileum, reflecting important differences between mice and humans. Since intestinal mucosal CD4 + T cells are being increasingly studied under disease conditions (e.g. inflammatory bowel disease and HIV), it is important to rigorously investigate the baseline mucosal immune environment throughout the gut in healthy individuals. When mucosal biopsies are collected from different parts of the human intestine, there must be an awareness that inherent regional differences among CD4 + T cell populations are present in order to interpret study findings.
The results of our study are consistent with a previous histopathology study showing that the cecum is a site of inherent inflammation even in a normal healthy population without Table 6. Differential expression of gene probes involved in immunologic processes.
Recently, innate lymphoid cells (ILCs) have emerged as an important non-CD4 + source of effector cytokines that may be particularly important at mucosal barrier surfaces [18]. IL-17 and IL-22 producing human ILCs are responsive to IL-23 signaling and could be important mediators of inflammatory bowel diseases [34]. Of additional interest was our observation that the increase in T H 17 cells in the cecum (compared to the terminal ileum and sigmoid colon) was offset by having relatively fewer numbers of IL-17 producing CD3 2 CD56 2 cells in this region. This inverse relationship raises the possibility of regulatory feedback between innate and adaptive sources of IL-17 in the healthy human intestinal tract.
We attempted to determine if the biopsies collected from the cecum had a greater bacterial load attached to the intestinal wall compared with those obtained from the terminal ileum and the sigmoid colon, but no significant differences were observed based on qPCR analysis of several bacterial ribosomal 16S DNA sequences (relative to host DNA). It is certainly conceivable that a deep sequencing analysis of bacterial ribosomal 16S units may lead to the identification of novel bacterial species or operational taxonomic units (OTUs) that are more abundant on the cecal wall compared to other parts of the intestine. Micro-heterogeneity between sample sites has been reported based on the very limited number of individuals who have been evaluated by deep sequencing studies of the colonic mucosa [35][36][37], but large variations in bacterial populations among intestinal regions has not been found. However, the cecum was not analyzed in detail in these studies. It would be extremely interesting to conduct deep sequencing analyses to determine the nature of the cecal microbiota attached to the mucosa using paired biopsy samples for which the gene expression profiles, as well as flow cytometry profiles of CD4 + T cell populations, is well characterized.
In a previous human study, gene expression profiling analysis was performed along the entire proximal-distal axis of the colon, but the terminal ileum was not included [38]. This study revealed two distinct patterns of transcript abundance along the proximaldistal axis of the large intestine. The colon was described as having a (i) dichotomous expression pattern consistent with its embryologic origins from both the midgut and the hindgut and hence segregating genes that are proximal or distal to a point that is two thirds the length of the transverse colon measured from the hepatic flexure and (ii) a gradual change in transcript levels from the cecum to the rectum, likely reflecting luminal flow and alterations in commensal microbiota. Our transcriptional profiling analysis of mucosal pinch biopsies confirms these results, with similar segregation of gene transcripts from both the proximal and distal colon and from the colon and terminal ileum. Additionally, we related transcriptional differences with the frequencies of CD4 + subsets along the intestine.
By correlating our flow cytometry analysis with gene expression profiling data from paired biopsy samples we discovered a positive relationship between the expression of resistin (RETN) and the Horizontal bars indicate median expression levels. Statistical significance between groups was determined by two-tailed Mann Whitney test. *P,0.05; **P,0.01; ***P,0.001. doi:10.1371/journal.pone.0041373.g007 frequency of T H 17 cells. Resistin is a cytokine produced by adipose tissue and was originally discovered as a potential mediator of metabolic functions involved in insulin resistance [39]. More recently, it has been characterized as an inflammatory cytokine that is implicated in several pathologic intestinal conditions (e.g. colon cancer and inflammatory bowel disease) likely through induction of IL-1b, IL-6, IL-8, IL-12, TNFa, and other pro-inflammatory cytokines [40]. As a pro-inflammatory cytokine the positive association between the frequency of T H 17 cells in the intestinal mucosa and expression levels of RETN is not surprising. PPARc has been shown to regulate RETN activity in several studies [41,42]. In this study we found that PPARc expression and T H 17 cell frequencies are both elevated in the human cecum, hence a relationship with RETN function is plausible. In addition, bacterial products like lipopolysaccharide (LPS) have been found to induce RETN expression in human and murine macrophages [43]. Future studies to establish a more mechanistic relationship between RETN and T H 17 cells will need to be conducted before any conclusions can be made about the biological significance of this relationship. We also found a negative relationship between the frequency of T H 17 cells and the expression levels of TFF1. The trefoil factors are mucin-associated peptides associated with a number of human tissues, with TFF1 and TFF2 primarily located in the gastric mucosa and TFF3 predominantly present in the mucus cells of the small and large intestine [44]. Up-regulation of all three TFF members occurs at sites of mucosal injury and has been observed in patients with peptic ulcer disease and inflammatory bowel disease [44,45]. TFF1 plays a critical role in mucosal protection and repair, with loss of TFF1 being associated with NF-kBmediated inflammation and gastric carcinogenesis [46]. Based on these observations, it was surprising to observe a negative relationship between the frequencies of T H 17 cells and TFF1, since both factors could be important for maintaining intestinal homeostasis. One possibility could be that the anti-microbial effects of TFF1 are acting through an independent pathway from T H 17 cells. Hence, when bacterial loads are controlled through a TFF1 mediated pathway, fewer T H 17 cells are required. This hypothesis could be tested further through the careful localization of TFF1 expression and presence of T H 17 cells by immunohistochemistry of the intestinal tissues.
Finally, we found that T Reg frequencies were positively associated with expression of FCRL5 (CD307). FCRL3, a closely related FCRL family member, is highly expressed on dysfunctional T Regs that express high levels of PD-1 and have a memory phenotype [28]. While FCRL5 has mostly been characterized in B cells and B cell lymphomas [47,48], polymorphisms have also been reported to be associated with autoimmune diseases such as ankylosing spondylitis [49], suggesting an immunoregulatory role [47]. The relationship between FCRL5 and T Reg cells in regulating intestinal homeostasis warrants further investigation. This study has important limitations. The majority of the enrolled subjects were older black men and so their mucosal environment may not be representative of younger patients in a disease state. Furthermore, complete colonoscopies requiring a preparative purge were performed to obtain biopsies from the cecum, a feature that may impact the analysis of the intestinal microenvironment. While the pinch biopsy samples obtained were approximately equal in size from all locations and from all subjects, we did not directly measure the absolute numbers of cells per gram of pinch biopsy tissue from each location. Importantly, CD4 + viability was relatively similar independent of biopsy location. Lastly, we recognize that differences in gene expression profiles of the pinch biopsies may be caused by heterogeneity of cellular populations and cellular density between samples. Nonetheless, this study is a step towards understanding locationspecific variations in lamina propria CD4 + T cell populations in healthy individuals. This information may provide a basis for comparison when characterizing disease conditions that have characteristic distribution patterns along the intestinal tract, such as Crohn's disease and ulcerative colitis.

Ethics statement
New York Harbor Veterans Affairs Hospital Institutional Review Board approval was obtained before enrolling subjects in the study. Written informed consent was received from all participants.

Research subjects
Consecutive subjects over the age of 50 who were at an average risk for colon cancer presented for screening colonoscopy and were enrolled in the study. Subjects were excluded who required endoscopy for an indication other than screening (e.g. family history of colon cancer, iron deficiency anemia, or hematochezia), were positive for human immunodeficiency virus (HIV), or who had absolute or relative contraindications to acquisition of biopsies for research purposes at the time of colonoscopy (e.g. systemic anticoagulation at the time of the procedure).

Biopsy acquisition protocol
Six to eight pinch biopsies were obtained at colonoscopy from the terminal ileum (TI), the ileocecal valve (ICV), the appendiceal orifice (AO), and the sigmoid colon defined as 20 centimeters from the anal verge (SC) using sterile cold biopsy forceps. Not every location was biopsied in all subjects due to technical factors (e.g. a tortuous colon did not permit terminal ileum intubation). The mucosa was directly observed and irrigated after biopsy acquisition to ensure resolution of bleeding. One biopsy from each location was snap-frozen for RNA extraction in Trizol reagent (Invitrogen). The remaining biopsies were placed in complete RPMI containing 10% fetal calf serum, 0.05 mM 2-Mercaptoethanol, and penicillin/streptomycin/glutamine (Invitrogen).

Isolation of lamina propria mononuclear cells
Biopsies were digested at 37uC for 1 hour in 100units/mL of collagenase Type VIII (Sigma) and 150 mg/mL DNase (Sigma) in complete RPMI. Cells were filtered through a 50 micron filter, washed with 5 mL 1X PBS, and pelleted. Cells were then resuspended in 40% Percoll (GE Healthcare), under-layered with

Boolean gate analysis of flow cytometric data
Five cytokine-producing CD4 + cell populations were analyzed in Flowjo using Boolean gates in order to calculate 32 possible cytokine populations. Data was exported to Simplified Presentation of Incredibly Complex Evaluations (SPICE) and filtered at a 2% cutoff to simplify the representation of the data (http://exon. niaid.nih.gov/spice/).

Determination of bacterial attachment
DNA was isolated from intestinal pinch biopsies stored in Trizol using the DNeasy Blood and Tissue Kit (Qiagen) according to the manufacturer's instructions. After a chloroform purification and ethanol extraction step, DNA was probed with Taqman probes designed for bacterial genes of interest (Applied Biosystems). Bacterial probes were selected for universal bacteria 16S, Clostridiales, Bacteroides, Prevotella, and Fusobacterium as previously described [50][51][52][53]. The RNase P copy number assay was utilized as a basis to compare bacterial DNA content to host (human) DNA content (Applied Biosystems) by performing a logarithmic ratio of the cycle thresholds obtained for bacterial and host DNA.

Transcriptional profiling and real-time PCR analysis of intestinal biopsies
Complementary DNA (cDNA) was reverse transcribed from RNA isolated from intestinal pinch biopsies according to manufacturer's instructions (Agilent Technologies). Cyanine 3labeled, linearly amplified complementary RNA (cRNA) was then synthesized, purified, hybridized to microarray slides, and scanned using the High Resolution C Scanner according to the manufacturer's instructions (Agilent Technologies). The raw data was then uploaded to Microsoft Excel, where it was log-transformed quantile-normalized and median centered. Expression profiling data were filtered at a standard deviation of 1.0 before undergoing unsupervised hierarchical clustering analysis in Cluster (Eisenlab). Results were then visualized with Java Treeview. Principal component analysis (PCA) was performed on the gene expression parameters using Cluster to generate eigenvalues, and the resulting datasets were further compiled and plotted on Excel. Filtered data sets were also analyzed for statistically significant genes (either through unpaired two-way comparison or multiclass analysis) with the Statistical Analysis of Microarrays (SAM) software version 2.23 A (http://www-stat.stanford.edu/tibs/SAM). Correlations between gene expression data and flow cytometric parameters were performed using quantitative analysis in SAM. Gene ontology (GO) and pathway analyses were performed with Protein Analysis Through Evolutionary Relationships (PANTHER) software (http://www.pantherdb.org).

Real-time quantitative PCR of intestinal biopsies
1mg of RNA from each sample was reverse-transcribed using SuperScript III Reverse Transcriptase (Invitrogen) and the resulting cDNA was used for quantitative real-time PCR with Taqman probes ordered directly from the manufacturer for the following genes of interest: DEFB1 (Hs00608345_m1), FCRL5 (Hs01070204_m1), MUC1 (Hs00159357_m1), MUC3 (Hs03649367_mH), MUC5 (Hs00861588_m1), PPARA (Hs00947539_m1), PPARG (Hs01115513_m1), TFF1 (Hs00907239_m1), and RETN (Hs00220767_m1) (Applied Biosystems). All cycle threshold values were normalized to GAPDH values. No significant differences in GAPDH expression were observed among the biopsy sites (data not shown). Data was analyzed with the t test using Prism 5.0 (GraphPad Software).

Statistical analysis
Statistical analysis was performed using Prism 5.0 (GraphPad Software). The two-tailed Mann Whitney non-parametric t test was used to assess statistical significance for all samples.