Altered T cell receptor beta repertoire patterns in pediatric ulcerative colitis

Summary

The antigenic specificity of T cells occurs via generation and rearrangement of different gene segments producing a functional T cell receptor (TCR). High‐throughput sequencing (HTS) allows in‐depth assessment of TCR repertoire patterns. There are limited data concerning whether TCR repertoires are altered in inflammatory bowel disease. We hypothesized that pediatric ulcerative colitis (UC) patients possess unique TCR repertoires, resulting from clonotypical expansions in the gut. Paired blood and rectal samples were collected from nine newly diagnosed treatment‐naive pediatric UC patients and four healthy controls. DNA was isolated to determine the TCR‐β repertoire by HTS. Significant clonal expansion was demonstrated in UC patients, with inverse correlation between clinical disease severity and repertoire diversity in the gut. Using different repertoire variables in rectal biopsies, a clear segregation was observed between patients with severe UC, those with mild–moderate disease and healthy controls. Moreover, the overlap between autologous blood–rectal samples in UC patients was significantly higher compared with overlap among controls. Finally, we identified several clonotypes that were shared in either all or the majority of UC patients in the colon. Clonal expansion of TCR‐β‐expressing T cells among UC patients correlates with disease severity and highlights their involvement in mediating intestinal inflammation.

Introduction

Inflammatory bowel diseases (IBD), comprised of Crohn's disease (CD) and ulcerative colitis (UC), are chronic inflammatory disorders of the gastrointestinal tract. These diseases develop in genetically susceptible hosts as a result of dysregulated immune responses to environmental triggers and microbial alterations 1. Different effector T cells, including T helper type 1 (Th1), Th2 and Th17 cells, which secrete proinflammatory cytokines, play a key role in mediating the inflammatory cascade 2, while regulatory T cells (T_regs) function to induce tolerance and prevent it 2. Alterations in this delicate balance can lead to development of inflammatory conditions accompanied by expansion of specific pathogenic T cell clones.

The antigenic specificity of T cells is determined following generation and rearrangement of different gene segments leading to a functional T cell receptor (TCR). This involves the recombination of variable (Vβ), diversity (Dβ) and joining (Jβ) genes, accompanied by deletions and insertions of random nucleotides in the junctions, generating millions of unique TCR clonotypes 3. Most TCRs are heterodimers composed of α and β chains, expressed on αβT cells, whereas in 1–10% of T cells the γ and δ chains are expressed on γδT cells 4. The primary site of antigen recognition is at the complementarity determining region (CDR)3, which is the area composed of V, D and J gene segments, and therefore the most variable region 5.

Each T cell expresses on its surface a unique TCR that has the potential to bind to a specific antigen, and the collection of all the different receptors is the TCR repertoire. In the past, studies characterizing the TCR repertoire were performed using polymerase chain reaction (PCR)‐based assays, antibodies, radiolabeled blotting assays and fluorescent labeling techniques that provided a superficial view of the repertoire and differential V(D)J gene usage. In contrast, high‐throughput sequencing (HTS) platforms developed in the last decade allow detailed assessment of TCR repertoire patterns. HTS employs massive parallel sequencing to process millions of rearranged TCR products simultaneously and permits an in‐depth analysis of individual TCRs at the nucleotide level.

The intestinal immune system is continuously exposed to countless bacteria, viruses, food particles and other proteins (both self and non‐self), and one of its critical roles is to determine the pathogenic potential and react accordingly. A diverse TCR repertoire is paramount to facilitate both an effector T lymphocytes response, but also tolerance via T_regs. Alterations in TCR diversity were identified in autoimmune conditions in both mice and humans. For example, in specific strains such as non‐obese diabetic mice, which are prone to develop autoimmune features, the diversity of the TCR repertoire of T_regs was found to be restricted compared to the TCR repertoire of the effector T cells 6. Similarly, in humans, a restricted TCR was demonstrated in several autoimmune diseases including myasthenia gravis 7, multiple sclerosis 8, rheumatoid arthritis 9 and psoriasis 10. Moreover, studies utilizing HTS demonstrated altered TCR repertoires in primary immunodeficiencies 11 and malignancies 12. Collectively, these studies suggest that alterations of the repertoire might have an important role in the pathogenesis of immune‐mediated disorders and might serve as a biomarker of disease activity.

There are limited data employing HTS on the TCR diversity and V(D)J gene usage in the intestine, and whether alterations occur in IBD patients. Earlier studies have documented increased clonality of lamina propria‐residing T cells 13 and differential TCR‐β chain usage between IBD subjects and non‐IBD controls 13, 14, 15, 16, 17. In recent years, only few studies using HTS of the TCR repertoire were performed in IBD patients, showing restricted TCR repertoires in CD and UC patients, although these analyses focused on the clones present in lamina propria 18, 19, 20, 21. We hypothesized that pediatric patients with UC possess unique T cell clones, and their expansion may correlate with degree of intestinal inflammation.

Materials and methods

Collection of samples

Patients were enrolled at two sites: the Pediatric Gastroenterology Unit of Edmond and Lily Safra Children's Hospital, Sheba Medical Center and the Juliet Keidan Institute of Pediatric Gastroenterology, Hepatology and Nutrition, Shaare Zedek Medical Center. This study was approved by the local institutional review board committees at Sheba Medical Center and Shaare Zedek Medical Center, and written informed consent was obtained from participating subjects. All experiments were performed in accordance with relevant guidelines and regulations.

Samples were collected during a colonoscopy performed for clinical assessment of bloody diarrhea (for patients with suspected UC) or concern for IBD that was later ruled out (for control subjects). Beside colonic biopsies obtained for routine clinical histopathological assessment, another one to two pinch biopsies were obtained for research purposes from the rectum. In addition, at the time of colonoscopy, a blood sample was collected.

DNA isolation

Biopsies were enzymatically digested with proteinase K prior to DNA isolation. Genomic DNA was extracted from both blood and intestinal samples using a commercially available kit (Wizard kit, Promega, Madison, WI, USA), according to the manufacturer's instructions.

TCR‐β repertoire library generation

For each genomic DNA sample, primers for various V and J gene segments in the TRB loci were used for amplification of the rearranged CDR3β (ImmunoSeq TRB Survey Service, Adaptive Biotechnologies, Seattle, WA, USA). To ensure equal depth of sequencing among the different samples, we used the Survey level (up to 500 000 reads per sample). The resulting libraries were purified, pooled and subjected to HTS using Illumina technology (Illumina Inc., San Diego, CA, USA), according to the manufacturer's protocol (ImmunoSeq, Seattle, WA, USA).

TCR‐β repertoire library analysis

The online tools of ImmunoSeq software were used for determination of productive clonality, sample overlap, pairwise scatters, CDR3β length, V‐, D‐ and J‐ family gene usage, percentage of productiveness and Venn diagrams. Calculated diversity (%) for the normalized measurement of richness of the TCR repertoire (i.e. number of unique rearrangements) was measured by dividing the number of unique rearrangements (i.e. clones) in each sample by the number of total templates obtained by HTS and multiplied by 100 22. Furthermore, additional diversity indices including Shannon's H and Simpson (1‐D) were calculated using the following formulae:

Shannon's H = ‐${\sum }_{i=1}^R{p}_i\text{In}{p}_i$

Simpson's 1‐D = $1-{\sum }_{i=1}^R{p}_i^2$

R = total templates

i = unique rearrangements

p_i = proportion of the total sequences belonging to the “i”^th unique rearrangement

D = dominance, unevenness

Shannon's H measures the overall diversity in a given population, i.e. total sequences in a given sample, by taking into account the number of unique sequences (richness of the repertoire) and how evenly the sequences are distributed. Simpson's 1‐D measures how evenly the unique sequences are distributed in a given sample. Lastly, graphical presentation of the repertoire was presented using hierarchical tree maps using the Treemap software (www.treemap.com).

Statistical analysis

Unless indicated otherwise, values are expressed mean ± standard error of the mean. The unpaired Student's t‐test or Mann–Whitney U‐test were used to test for statistical significance. Significance was determined if the P‐value was < 0·05 (with P‐value summaries as: *P < 0·05, **P < 0·01 and ***P < 0·001). For differential V‐ and J‐ gene usage, Student's t‐test with Bonferoni's adjustment for multiple comparisons was used. Primary component analysis (PCA) was performed and visualized using the R prcomp function and ggbiplot package. Data were normalized by Z‐score.

Results

Characterization of participating subjects

Paired blood and rectal biopsies were obtained from four control subjects and nine treatment‐naive newly diagnosed UC patients (Table 1). Subjects in the control group underwent a colonoscopy for evaluation of suspected IBD due to abdominal pain, diarrhea or iron deficiency anemia. However, endoscopy was macroscopically and histologically normal. Moreover, none of the control subjects were diagnosed with IBD during the following year after the procedure. Patients with UC were referred for colonoscopy for evaluation of bloody diarrhea. Pan‐colitis (involving the entire colon) was demonstrated in seven patients and left‐sided colitis (involving only the rectum, sigmoid and descending colon) in another two subjects. Histological assessment revealed signs of chronic colitis, without evidence of granulomas and was overall consistent with the diagnosis of UC in all of the patients (representative hematoxylin and eosin stains are shown in Supporting information, Fig. S1). None of these patients were treated with immunosuppressive medications prior to the colonoscopy.

Table 1.

Demographic characteristics and general sample overview of participating subjects

						Blood		Rectum
	Subject Number	Age (years)	Gender	PUCAI	Clinical Disease Severity	Total templates	Unique rearrangements	Total templates	Unique rearrangements
Controls	1	17	F			8622	7497	20 489	14 453
	2	16	M			6735	6212	28 274	17 519
	3	13	M			95 257	78 144	37 502	22 147
	4	15	F			16 756	14 355	41 880	25 842
UC patients	1	16	M	25	Mild	7454	6533	58 157	43 724
	2	13	F	35	Mild	22 352	17 610	24 772	14 277
	3	15	F	45	Moderate	140 571	46 495	96 430	60 391
	4	14	M	50	Moderate	19 987	17 913	77 738	47 419
	5	17	F	55	Moderate	16 115	12 681	45 102	22 238
	6	17	M	50	Moderate	145 311	81 883	55 891	25 231
	7	14	M	75	Severe	110 134	84 092	74 498	32 754
	8	16	M	65	Severe	2190	1940	96 813	47 634
	9	12	M	70	Severe	56 652	41 323	64 693	28 783

PUCAI = Pediatric Ulcerative Colitis Activity Index.

The mean age of the control and patient groups was 15·2 ± 0·8 and 14·9 ± 0·6 years, respectively. Two of the patients presented with a Pediatric Ulcerative Colitis Activity Index (PUCAI) score of 10–30, indicating mild disease activity, four with moderate disease (PUCAI 35–60) and three presented with acute severe colitis (ASC), indicated by a PUCAI score of 65 or higher. Histological severity of inflammation among patients was overall similar to the clinical disease activity scores (data not shown).

Skewed TRBV and TRBJ gene utilization in pediatric UC

The data generated by HTS are summarized in Table 1. In the rectal tissue, the TCR‐β repertoire from UC patients had significantly more total templates (66 010 ± 7806 versus 32 036 ± 4780, P < 0·01) and unique rearrangements (representing distinct clones; 35 828 ± 4944 versus 19 990 ± 2511, P < 0·01) compared with healthy controls, suggesting an increased number of T cells. In contrast, in the blood, both total templates and unique rearrangements varied greatly among individuals, without significant differences between the two groups.

Next, we calculated the differential usage of each of the TRBV and TRBJ gene families used in the patients' blood and rectal TCRs and compared them to controls. The usage of the different TRBV and TRBJ genes was non‐stochastic and largely conserved between paired blood and rectal samples, as well as between different individuals (Supporting information, Fig. S2). The most commonly used TRBV gene family in both blood and tissue were TRBV5, TRBV6 and TRBV7, while the top‐ranking TRBJ genes utilized were TRBJ1‐1, TRBJ1‐2, TRBJ2‐1 and TRBJ2‐7. Significant differences in utilization of different TRBV and TRBJ were demonstrated when comparing controls and patients with UC, both in the blood and in the rectum. These results suggest that distinct gene usages may play an important role in the pathogenesis of colitis and set the stage for an in‐depth characterization of the TCR‐β repertoire in UC, using next‐generation sequencing (NGS).

Enhanced clonal expansion in inflamed colonic tissue

In order to gain a broad overview of the full TCR‐β repertoire diversity, the sequencing data of blood and rectal samples were graphically illustrated as tree maps, in which each clone is represented by a differently colored square, and the size correlates with its frequency (Fig. 1 and Supporting information, Fig. S3). These tree map images demonstrate the increased clonal expansion in patients with UC compared with controls, mainly in the rectum but also in the blood in some patients. Tree map images were similar in patients with mild–moderate UC and those presenting with ASC (Supporting information, Fig. S3). However, the cumulative frequency of the top 100 most common clones in ASC patients was significantly higher compared with controls, in both the blood and rectum (Fig. 1). Interestingly, in a single patient (P3) with moderate disease, four different clones constituted nearly 20% of the full TCR‐β repertoire in the blood (Supporting information, Fig. S3), suggesting pronounced specific antigenic activation of T cells.

Figure 1.

Clonal expansion of the T cell receptor (TCR)‐β repertoire in colonic tissue of UC patients. (a) Representative tree map images of blood and rectal TCR repertoire of a control subject and an ulcerative colitis (UC) patient, illustrating increased clonality in inflamed tissue. (b) Cumulative frequencies of the 100 most prevalent clones used by each subject in the blood and rectum.

Association between disease severity and TCR‐β repertoire diversity features

Next, we calculated several diversity measures in blood and intestinal samples. In general, the TCR‐β repertoire in the rectum was more clonal compared with the blood, in both control and patient groups. Patients with ASC showed significantly higher clonality, which reflects the level of clonal expansion in a given repertoire compared with controls (Fig. 2), whereas a trend towards an increase in clonality was seen in rectal samples of all the UC patients compared with controls (9·0 ± 1·2 versus 5·8 ± 1; P = 0·06).

Figure 2.

T cell receptor (TCR)‐β repertoire features correlates with severity of ulcerative colitis (UC). (a) The clonality index is calculated from normalized Shannon's H [clonality index = 1‐ Shannon's H/log₂ (unique rearrangements)] in blood and tissues of control and UC patients. (b) The calculated diversity is the percent of unique rearrangements divided by total templates in the blood and tissues of patients and controls. (c) PCA of rectal samples from healthy subjects and UC patients showing segregation according to degree of clinical severity.

The diversity indices of the Shannon's H and Simpson's 1‐D did not differ between patients and controls, both in blood and in the intestine (Supporting information, Fig. S4). However, the calculated diversity, which is the percentage of unique rearrangements of total templates, showed a significant decrease in rectal TCR repertoires of ASC patients compared with controls (Fig. 2b). Taken together, although both the total templates and unique rearrangements were higher in the patients compared to the controls the calculated diversity was lower due to increased clonality. Nevertheless, diversity was significantly decreased in intestinal samples of both control subjects and UC patients in comparison with autologous blood samples (Fig. 2b).

Given these differences in the overall diversity of the repertoires, we next assessed whether samples can be clustered together based on these features, in correlation with the severity of inflammation. A statistical model of PCA based on productive clonality, top 100 clones, diversity and percentage of T cells showed that blood and tissue samples cluster separately (Supporting information, Fig. S5A). Moreover, a clear segregation between healthy subjects, patients with mild–moderate disease and those with ASC was demonstrated in rectal specimens (Fig. 2), but not in the blood (Supporting information, Fig. S5B). Thus, patients with ASC had distinct features in the tissue that segregated them from patients with milder forms of colitis and healthy controls. Taken together, our data suggest that in patients with ASC the intestinal TCR‐β repertoire is altered, with reduced diversity due to clonal expansions.

UC patients possess shorter CDR3β clones in the rectum

The CDR1β, CDR2β and CDR3β comprise the major domains of the TCR, which fold together to form the antigen‐binding site. Specifically, the CDR3β is encoded by the last few nucleotides of the V segment, the entire D segment, a portion of the J gene segment and the intervening N regions, thus possessing most of the variability of the T cell receptor 23. We compared CDR3β length distribution and mean length, as it can influence folding of the TCR loop and affinity to the antigen‐bound major histocompatibility complex (MHC). The overall distribution of the CDR3β lengths was comparable between blood and rectal TCR repertoires of control subjects and patients. However, the peak of the CDR3β length distribution for the patients' rectal TCR repertoire was a few nucleotides shorter compared with control samples (Fig. 3a). Furthermore, the average CDR3β length in the intestine was significantly shorter than in the blood for both patient and control groups (Fig. 3b). It is noteworthy that there were no differences between mean CDR3β lengths of patients with mild–moderate disease versus those with ASC. Collectively, our findings of shorter CDR3β length in the inflamed rectal TCR repertoire may be a representative feature of UC, similar to findings observed in other autoimmune diseases, such as insulin‐dependent diabetes mellitus 24 and multiple sclerosis 25.

Figure 3.

Shorter rectal CDR3β length in UC patients. (a) Distribution of the number of nucleotides comprising the CDR3β; frequency was calculated as the average of all individuals from each group. Right image zooms in on the most prevalent nucleotide frequencies in each group. (b) Statistical analysis of the average of CDR3β lengths using Student's t‐test with Bonferroni's correction for multiple comparisons. NT = nucleotide.

Greater presence of shared clones between UC patients

Next, we aimed to determine whether specific clones (based on the CDR3 amino acid sequences) are shared between the patients and between the controls, both in the blood and in the gut. Interestingly, we identified five different rectal clonotypes shared between all patients, and six clonotypes were shared between eight of the nine patients (Fig. 4a). These clones were absent or appeared in only some of the controls. Specifically, two clonotypes (CASSLGGNTGELFF and CASSFQETQYF) were present in all the UC patients and in none of the controls. In contrast, among the controls, six different clonotypes were shared between all controls (Fig. 4a). One of these shared clonotypes (CASSGSYNEQFF) was absent in all UC patients, although its frequency in the controls was very low. In the blood, we found limited sharing of specific CDR3 sequences between UC patients and between controls (Fig. 4b).

Figure 4.

Specific clonotypes are shared between ulcerative colitis (UC) patients and controls. Figure depicts sharing of unique clonotypes between patients and between controls in the (a) rectum and (b) blood. Coloring of each box is represented based on the number of templates of each clone identified. We present sequences that were shared between at least eight of the nine UC patients, and clones that were shared between all four healthy controls. *132 templates; #323 templates.

Enhanced blood–rectal clonal sharing among patients with UC

Circulating T cells migrate from the blood to the intestine following expression of different integrins such as α4β7 26. We wanted to determine whether there is evidence of T cell migration in the HTS data by assessing the degree of overlapping blood and rectal clones for each subject. Among patients, this analysis was possible only for seven of nine subjects, as in the remaining two subjects (P1 and P8) substantially fewer transcripts were detected in the blood due to technical issues, leading to almost no overlap with the rectum. Pairwise scatters showed minimal sharing between blood and rectal samples in the control group (Fig. 5a and Supporting information, Fig. S6). However, among UC patients the degree of shared clones was significantly higher than in healthy controls, as was determined using Morisita's Index of overlap (Fig. 5b). Thus, the enhanced clonal sharing among patients suggests increased trafficking of T cells between the blood and the inflamed tissue.

Figure 5.

Increased autologous blood–rectal clonal overlap in ulcerative colitis (UC) patients. (a) Representative blood–rectal pairwise scatters of T cell receptor (TCR) clones followed by (b) Morisita's index of overlap analysis.

Discussion

In the past decade significant progress has been made in understanding the pathogenesis of IBD, and it is clear that T cells play a pivotal role in this process 27. One of the hallmarks of colitis is expansion of T cells in the lamina propria, whether from the influx of lymphocytes from the blood or due to proliferation of resident T cells. Here we provide a comprehensive analysis of the TCR‐β repertoire of paired blood and rectal T cells in pediatric patients with UC and in healthy controls. We show that pediatric UC is characterized by the expansion of TCR‐β clones mainly in the rectum, but also in the blood in some patients, and demonstrate an inverse correlation between clinical disease severity and the repertoire's diversity. Moreover, clonal sharing between patients with UC is enhanced in comparison to sharing between healthy controls, although the overall magnitude of shared clones was low in both groups.

In the early 1990s, several groups demonstrated increased clonality of T cell populations in the gut compared with the blood 28, 29, 30. These were later extended to inflamed intestine of patients with CD and UC with evidence for further perturbation of the relatively restricted intestinal repertoire 15, 16, 17. This restriction is presumably important for tight regulation of the immune response 31. However, it remains unknown whether tissue clonality is driven by adaptive immune responses to particular pathogen(s), self‐antigens or secondary to mucosal injury. The tight regulation of T cell proliferation and migration in steady state conditions can be depicted by the highly comparable measurements of overall repertoire diversity among the healthy controls, which is absent in the patients, and corresponds to the skewed immune response seen in colitis. Moreover, our observations of increased clonality in blood and tissue samples of the patients with UC firmly support the above‐mentioned studies, further extending to suggest an inverse correlation between severity of inflammation and TCR repertoire diversity.

Our analysis indicates a higher degree of clonal sharing between UC patients, both in the blood and in the rectum. Using NGS we were able to compare the frequency of specific intestinal clonotypes and show that a minority of them are shared between patients, but not between controls. Although the frequency of shared clones was overall very low, it is a very rare event in the control groups. It is likely that UC is not a disease driven by single antigens, therefore we do not find a high degree of unique clonal sharing. Furthermore, the overall population of T cells in the affected tissue may be different than in controls. However, these specific clones might have an important role in driving the inflammatory process in the gut. Interestingly, one common clonotype, CASSLGYEQYF, was previously reported as an intestinal gluten‐reactive clonotype in celiac disease 32, and thus might play a general role in mediating intestinal inflammation or might be up‐regulated due to inflammation. Additional studies are required to further characterize these specific clones, determine whether they characterize specific types of disease and whether their frequency changes in response to therapy. Moreover, it will be interesting to compare the TCR repertoire profiles of patients with UC to patients with CD, and specifically those with colonic involvement.

The TCR recombination machinery which determines antigenic specificity is not only embedded in the DNA of the CDR3β, but is also influenced by other biochemical properties such as differential folding of the TCR and its length 33. We found that the average CDR3β length in the blood was 43·7 nucleotides, in accordance with prior reports 23. Furthermore, the average length of the CDR3β was significantly shorter in intestinal tissue than in the blood, and T cells from inflamed mucosa were significantly shorter than those residing in healthy control mucosa. The average CDR3β length is calculated from the total sequences, inclusive of the expanded T cells, which can be influenced by local signals leading to the proliferation of specific T cells with a shorter CDR3β. Although the shorter CDR3β length translates physiologically to a single amino acid, this may affect the folding of the TCR loop, affinity to the MHC 34 or result in conformational changes of the receptor 35. Clonotypes with shorter CDR3β length have been identified in different autoimmune diseases such as insulin‐dependent diabetes mellitus 24 and multiple sclerosis 25, and thus may signify a potentially pathogenic capacity. Of note, most shared rectal clones between UC patients possessed relatively short CDR3β, thus adding to the evidence for their potential inflammatory characteristics.

One limitation of our study is the small number of individuals in the healthy control group. Nevertheless, repertoire patterns were strikingly similar among different subjects in this group. In addition, we analyzed the repertoire patterns of total lymphocytes rather than specific subpopulations such as effector CD4⁺ T cells or cytotoxic CD8⁺ T cells, in order to capture the overall effect on all T cells. It is plausible that less common subpopulations, such as T_regs, display specific TCR repertoire patterns that are not reflected in the overall repertoire, given the low percentage of these cells among total lymphocytes. Finally, we cannot exclude that, in the process of obtaining pinch biopsies from the intestine, blood‐derived T cells were also collected, especially among patients with UC who had overt signs of colitis. Nevertheless, the strength of our study lies in the use of treatment‐naive pediatric patients newly diagnosed with UC, allowing us to comprehensively investigate the TCR‐β repertoire without any effect of therapy. Furthermore, our paired mucosal‐blood samples directly reveal a relationship between these two distinct anatomical sites. It would be interesting to examine the repertoire of autologous mesenteric lymph nodes, which could further reveal the levels of T cell trafficking.

In conclusion, we show that pediatric UC is characterized by clonal expansion of the TCR‐β repertoire, suggesting specialization of unique T cell clones, which probably have a role in mediating tissue damage. In addition, intestinal clonal expansion among UC patients may correlate with severity of inflammation, highlighting a potential pathogenic role of specific clonotypes. Additional studies are required to determine factors that regulate the TCR‐β repertoire and predict disease progression, as well as identify the antigens driving the expansion of T cells.

Author contributions

L. W. designed the study, conducted the analysis and wrote the manuscript; M. Y. N. conducted the analysis; E. R. conducted the analysis; A. L. conducted the analysis; T. B. conducted the analysis; Y. H. contributed to acquisition of samples and analysis of data; A. L. contributed to acquisition of samples; E. S. contributed to acquisition of samples; M. S. conducted the analysis; R. S. contributed to design of the study, critical analysis of the data and manuscript preparation; B. W. contributed to acquisition of samples; Y. N. L. contributed to design of the study, analysis of the data and wrote the manuscript; D. S. S. designed the study, conducted the analysis and wrote the manuscript.

Disclosures

D. S. S. received fees from AbbVie for consultation and presentation and a research grant from Takeda; these were not related to this study. Other authors have no conflicts of interest to disclose.

Supporting information

Fig. S1. Representative Hematoxylin and eosin stains of rectal biopsies. Figure depicts images of a rectal section from a healthy control, a UC patient with mild‐moderate inflammation and a patient with severe inflammation

Fig. S2. Skewed usage of TRBV gene families and TRBJ genes in UC. Heat map analysis representing the frequency of TRBV gene family and TRBJ gene usage for the UC patients and healthy controls, in blood and rectum. Genes that had a frequency of less than 0.1% were excluded from the analysis. Differences in the average of the frequencies between the compartments were summarized as follow: * statistical significance between either rectum or blood of controls vs. UC patients; # statistical significance between blood and tissue of UC patients; ^ statistical significance between blood and tissue of healthy controls.

Fig. S3. TCR repertoire in blood and rectal samples from patients and controls. Tree maps graphically representing the overall TCR repertoire of autologous blood and rectal samples from 4 healthy controls and 9 UC patients. In order to provide a comprehensive view of all samples figure includes also the representative images shown in Fig. 1.

Fig. S4. Comparison of diversity indices between controls and UC patients. (a) Shannon's H and (b) Simpson's 1‐D analyses of blood and rectal samples from controls and UC patients.

Fig. S5. Blood and rectal samples segregate according to origin of the sample. PCA diagrams based on percentage of T cells, productive clonality, top 100 clones, diversity and percent productiveness. (a) Image depicts segregation of blood (red) vs. rectal (blue) samples in the entire cohort (controls + patients). (b) Image showing lack of segregation of blood samples from controls vs. patients.

Fig. S6. Blood‐rectal pair‐wise scatter plots of controls and UC patients. In order to provide a comprehensive view of all samples figure includes also the representative images shown in Fig. 4.