Abstract Objective The risk of AS is associated with genomic variants related to antigen presentation and specific cytokine signalling pathways, suggesting the involvement of cellular immunity in disease initiation/progression. The aim of the present study was to explore the repertoire of TCR sequences in healthy donors and AS patients to uncover AS-linked TCR variants. Methods Using quantitative molecular-barcoded 5′-RACE, we performed deep TCR β repertoire profiling of peripheral blood (PB) and SF samples for 25 AS patients and 108 healthy donors. AS-linked TCR variants were identified using a new computational approach that relies on a probabilistic model of the VDJ rearrangement process. Results Using the donor-agnostic probabilistic model, we reveal a TCR β motif characteristic for PB of AS patients, represented by eight highly homologous amino acid sequence variants. Some of these variants were previously reported in SF and PB of patients with ReA and in PB of AS patients. We demonstrate that identified AS-linked clones have a CD8+ phenotype, present at relatively low frequencies in PB, and are significantly enriched in matched SF samples of AS patients. Conclusion Our results suggest the involvement of a particular antigen-specific subset of CD8+ T cells in AS pathogenesis, confirming and expanding earlier findings. The high similarity of the clonotypes with the ones found in ReA implies common mechanisms for the initiation of the diseases. T cell receptor repertoire, ankylosing spondylitis, synovial fluid, spondyloarthropathy Rheumatology key messages TCR β motif characteristic for peripheral blood of AS patients was revealed using the donor-agnostic probabilistic model. The AS-related CD8+ T cell clonotypes were detected and enriched in SF from HLA-B*27+ AS patients. The AS-related clonotypes were not the most expanded in SF and peripheral blood of AS patients. Introduction SpAs are a group of diseases with similar clinical and genetic features, which includes AS, PsA, enteropathic arthritis, undifferentiated SpA and ReA. Strong linkage with certain alleles of genes involved in the antigen-presenting pathway (HLA-B, ERAP1 and ERAP2; reviewed by Brown et al. ) underlies the arthritogenic peptide hypothesis of AS and other HLA-B*27-associated disease initiation . The hypothesis suggests expansion of self-reactive T cell clones triggered by cross-reaction with putative pathogen-derived peptide presented on HLA-B*27. The similarity in clinical manifestations and association of certain MHC alleles with increased disease risk might indicate a similar or identical structure of the putative antigen(s) recognized by the self-reactive T cell clones. This, in turn, implies that the expanded T cell clones, specific to the antigen, might have a similar structure of the antigen-recognizing domain of the TCR. Several studies have revealed the presence of T cell clones with highly homologous TCR identified in peripheral blood and synovial fluid of different patients with ReA [3–5]. However, until the development of high-throughput sequencing, T cell repertoire analysis was restricted to comparisons of tens to hundreds TCRs between several patients. Recently, Faham et al.  have used high-throughput sequencing-based profiling of TCR repertoires from a large collection of peripheral blood (PB) samples from AS patients and healthy donors and identified a CDR3 motif similar to the one previously described for ReA. Here, we report the results of our independent study designed to search for T cell clonotypes linked with AS. We applied deep TCR profiling with unique molecular identifier (UMI)-based quantification and error correction  to peripheral whole blood samples of 25 AS patients and 107 healthy donors and identified on average ∼200 000 TCR β clonotypes per donor. Using a new algorithm based on probabilistic assessment of clonal sharing between donors, we identified the group of CD8+ T clonotypes with highly similar TCR β shared almost exclusively between AS patients. The TCR structure of the clonotypes confirms the findings of Faham et al. . We also demonstrate the presence and enrichment of the clonotypes in SF of inflamed joints of all HLA-B*27+ AS patients studied, suggesting their involvement in pathogenesis. Methods Donors and samples For the present study, PB samples were obtained from 25 AS patients (24 HLA-B*27+ and 1 HLA-B*27−; details of the patient cohorts, including age, sex and treatment, are provided in supplementary Table S1, available at Rheumatology online) and 13 healthy donors (8 HLA-B*27+ and 5 HLA-B*27−). To expand our control cohort, we also used previously published TCR repertoire data for 95 healthy donors obtained using the same TCR profiling technique [8–10]. Three donors of the additional cohort were HLA-B*27−. Altogether in this study, we analysed peripheral whole blood TCR repertoires of 107 healthy donors: 7 HLA-B*27+, 8 HLA-B*27− and 92 donors with unknown HLA-B*27 status. For one HLA-B*27+ healthy donor, only CD8+ TRBV9+ and CD8− TRBV9+ fractions of PB T cells were available. SF samples were collected from five AS patients (four HLA-B*27+ and one HLA-B*27−) during acute synovitis. The HLA-B*27 status for healthy donors and AS patients was determined using an allele-specific qPCR HLA-B*27-typing kit (DNA-Technology, Moscow, Russia). The study was approved by the local ethical committee of Pirogov Russian National Research Medical University, Moscow, Russia and performed in accordance with the Declaration of Helsinki. All donors provided their informed consent for sample collection. T cells and RNA isolation, TCR cDNA library preparation and sequencing PB or SF (5–8 ml) mononuclear cells (PBMCs or SFMCs) were isolated by Ficoll density gradient centrifugation. PBMC/SFMC CD4+ or CD8+ T cells were isolated using Dynabeads CD4/CD8 positive isolation kit (Thermo Fisher Scientific, Waltham, MA, USA). Total RNA from PBMCs and SFMCs was obtained using TRIzol reagent (Thermo Fisher Scientific, Waltham, MA, USA). CD3+ CD8+ TRBV9+ and CD3+ CD8– TRBV9+ cells from HLA-B*27+ healthy donors were FACS sorted from PBMCs using anti-human CD3-APC (Invitrogen, Waltham, MA, USA), CD8-FITC (Miltenyi, Bergisch Gladbach, Germany) and TRBV9-PE (Beckman Coulter, Indianapolis, IN, USA) antibodies. PBMCs were stained according to the manufacturer’s recommendations and sorted directly in 300 μl of RLT (Qiagen, Venlo, The Netherlands). For immunomagnetic isolation of TRBV9+ T cells, anti-human TRBV9-PE antibody and Anti-PE MicroBeads (Myltenyi, Bergisch Gladbach, Germany) were used. Total RNA from sorted cells was isolated using RNAeasy Mini kit (Qiagen, Venlo, The Netherlands) and completely used for TCR library preparation. UMI-tagged 5′-RACE TCR β cDNA libraries were prepared according to the previously described general protocol , with a double-end sample barcoding strategy. Details of the protocol with implementation of double barcoding, including consumables, oligonucleotides, RT and PCR conditions, were described earlier . Libraries were sequenced with Illumina HiSeq 2000/2500 (Illumina, San Diego, CA, USA) in 100 + 100 pair-end mode. Raw sequencing data obtained for the present study are available in the Sequence Read Archive database (SRP111372). Sequencing data analysis and statistical methods Raw sequencing data were pre-processed using MiGEC software . To exclude erroneous clonal sharing that resulted from cross-sample contamination, we used a double-indexing strategy of library preparation with introduction of the first sample-specific barcode directly during cDNA synthesis. All reads with improper pairs of sample barcodes were discarded before repertoire reconstruction. Reads containing UMIs that have been sequenced only once were also discarded, ensuring that all TCR β cDNA molecules that passed to further analysis were sequenced at least twice. This allowed us essentially to reduce the number of erroneous reads, comprising sequencing and PCR errors, before further steps of UMI-based reads error correction that are included in the MiGEC pipeline. TCR β CDR3 sequences were then extracted using MiTCR software with ETE mode for advanced correction of erroneous clonotypes . Further TCR repertoire analysis was performed using the tcR R-package  and R programming language . TCR clonotypes were assigned to the CD4+ or CD8+ T cell subpopulation based on detection of the corresponding TCR β nucleotide sequence in repertoire data obtained for separated CD4+ or CD8+ T cell samples isolated from the same PBMC sample. The ambiguities were solved using the abundance ratio threshold: clonotypes at least 10 times more abundant in one cell subset compared with another were attributed to the first subset; clonotypes with a lower ratio were attributed as undefined. Fisher’s exact test with the Benjamini–Hochberg correction was used to calculate P-values for comparisons of clonotype frequency between cohorts or clonotype abundance. In the analysis of clonal recombination probability, P-values for clonotype sharing were calculated using the density of log10 rearrangements distribution (assuming normal distribution) for clonotypes shared by a certain number of donors (i.e. for clonotypes shared between four donors, five donors, etc.). To obtain a better estimate of distribution parameters, we pooled together clonotypes shared between more than seven individuals (independently of the disease status of the donors who shared the clonotypes), to obtain ⩾30 observations in each bin. The P-value threshold was set up at <0.001, and each clonotype below the threshold was considered as having significantly higher sharing than could be expected based on its recombination probability. Results T cell repertoire profiling strategy Study of T cell clonal sharing between donors could be biased dramatically by cross-contamination between samples and sequencing errors, affecting the detection of clones in samples. Here, to study T cell repertoires of AS patients and healthy donors we used our previously published protocol [7, 9], which relies on dual indexing to eliminate cross-sample contamination and UMI for correction of PCR and sequencing errors. Importantly, tagging of each TCR β mRNA molecule with UMI allowed us to count the number of TCR mRNA molecules passed to the analysis and to estimate the abundance of T cell clonotypes identified [11, 15, 16]. TCR profiling depth also can influence the detection of T cell clones in a sample. For whole blood samples, a comparable TCR repertoire profiling depth was reached for both cohorts: on average we analysed 180 591 and 206 162 unique TCR beta mRNA molecules for the samples from AS patients and HLA-B*27+ healthy donors, respectively. Even, greater averaged depth was achieved for blood samples of the donors in whole control cohort of 107 healthy donors (see next subsection; supplementary Table S2, available at Rheumatology online). Discovery of AS-related T cell clonotypes using assessment of VDJ recombination probability T cell clonotype sharing between samples derived from different donors depends on several sequential events: generation of the particular TCR during VDJ recombination, successful selection of the clone in the thymus and further clonal expansion in the periphery . It can be stated that TCR variants with a relatively high probability of generation will be shared between a higher number of donors compared with those having a lower probability of generation, irrespective of antigen specificity and involvement in the immune response. On the contrary, clonal expansion of T cells specific for a particular antigen increases their chance of getting into a sample and, thus, the probability of their detection in the donor. Hence, the immune response against a particular antigen shared between individuals, especially in a similar MHC context, increases the in-cohort detection frequency (i.e. sharing between donors) of identical or highly homologous TCR variants specific for this antigen [18, 19]. Thus, the expansion can increase sharing of the clonotypes with a relatively low probability of generation. Considering this, we applied a new approach based on probabilistic assessment of the sharing of each TCR β clonotype between donors to find the clonotypes that are shared between a higher number of donors than could be expected in the absence of clonal expansion. As the first step, to provide a rough estimate of clonal sharing, we analysed the presence of 54 000 TCR β clonotypes that were shared between at least three AS patients (i.e. clonotypes, shared between ⩾10% of patients) in PB T cell repertoires of healthy donors (control cohort, n = 107; combined data obtained using same TCR profiling technique from the present study and three others [8–10]). We found only one clonotype, TRBV9–CASSVGLYSTDTQYF–TRBJ2-3, with significantly higher sharing in AS patients compared with healthy donors (supplementary Table S3, available at Rheumatology online). To confirm that unusual sharing of the TRBV9–CASSVGLYSTDTQYF–TRBJ2-3 clonotype between AS patients was attributable to clonal expansion and to increase the analytical power to search for other clonotypes overrepresented between patients, we estimated the probability of generation of CDR3 β amino acid sequences encoded by TRBV9–TRBJ2-3 rearrangements for all clonotypes found in the data. We extended the statistical model of TCR generation originally published by Murugan et al.  by adding the Monte-Carlo approach; the probability of generation of a given TCR amino acid sequence was estimated as the average number of corresponding nucleotide sequences generated in silico. We generated 2 × 109 TRBV9–TRBJ2-3 TCR rearrangements, out of which 1.2 × 107 corresponded to 335 TRBV9–TRBJ2-3 CDR3 amino acid sequences shared between at least any four donors from the cohort: 25 AS patients, 7 HLA-B*27+ and 7 HLA-B*27− healthy donors. As expected, the number of donors positive for each particular clonotype was roughly proportional to the number of in silico rearrangements that resulted in the corresponding CDR3 amino acid sequence, thus demonstrating that clonotypes with a higher probability of generation have higher sharing between donors (Fig. 1). Fig. 1 View largeDownload slide Identification of AS-related T cell clonotypes based on probabilistic model of TCR rearrangement Each dot represents a unique TRBV9–TRBJ2-3 amino acid TCR β sequence shared between the given number of any donors (x-axis) from the cohort of 25 AS patients, 7 HLA-B*27+ and 7 HLA-B*27− healthy donors. Clonotypes found exclusively in AS patients are shown in black. On the y-axis, the log10 of in silico-produced rearrangements corresponding to the CDR3 amino acid sequences is shown. The dashed line represents the threshold of significance for the clonotype sharing deviation above the expected level based on the probability of recombination (P < 0.001). The right panel shows the eight TCR β clonotypes unusually shared between donors. These clonotypes are enriched between AS patients compared with healthy controls. *Previously published ReA and AS clonotypes. **Published with V β segment TRBV13. Fig. 1 View largeDownload slide Identification of AS-related T cell clonotypes based on probabilistic model of TCR rearrangement Each dot represents a unique TRBV9–TRBJ2-3 amino acid TCR β sequence shared between the given number of any donors (x-axis) from the cohort of 25 AS patients, 7 HLA-B*27+ and 7 HLA-B*27− healthy donors. Clonotypes found exclusively in AS patients are shown in black. On the y-axis, the log10 of in silico-produced rearrangements corresponding to the CDR3 amino acid sequences is shown. The dashed line represents the threshold of significance for the clonotype sharing deviation above the expected level based on the probability of recombination (P < 0.001). The right panel shows the eight TCR β clonotypes unusually shared between donors. These clonotypes are enriched between AS patients compared with healthy controls. *Previously published ReA and AS clonotypes. **Published with V β segment TRBV13. Notably, eight TCR β clonotypes had a higher frequency of detection among donors than could be expected by chance based on their generation probability (P < 0.001; dashed line in Fig. 1), suggesting that their expansion was driven by a common antigen. Seven of them had a highly similar sequence and were found exclusively in AS patients. Of these clonotypes, the greatest sharing between patients was observed for the previously detected TRBV9–CASSVGLYSTDTQYF–TRBJ2-3. Consistent with the estimated relatively low probability of generation and thus low probability of being shared between donors, each of the clonotypes was found in PB samples of ≤4 donors in the whole control cohort of 107 healthy individuals (Fig. 2A; supplementary Fig. S1, available at Rheumatology online). Fig. 2 View largeDownload slide Abundance of AS-associated TCR β clonotypes in peripheral blood (A) Proportion of T cells with a particular TCR β (rows) in the peripheral blood (PB) of each donor (columns): AS patients, healthy HLA-B*27+ and HLA-B*27− donors (marked with asterisk), and donors with unknown MHC genotype, (B) compared with the PB proportion of the clonotypes observed after increasing of repertoire analysis depth by TCR profiling of FACS-sorted CD8+ TRBV9+ or CD8− TRBV9+ cells from the PB of healthy HLA-B*27+ donors. Amino acid residues in variable positions of CDR3 are coloured accordig to the physical and chemical properties of the side-chains. Fig. 2 View largeDownload slide Abundance of AS-associated TCR β clonotypes in peripheral blood (A) Proportion of T cells with a particular TCR β (rows) in the peripheral blood (PB) of each donor (columns): AS patients, healthy HLA-B*27+ and HLA-B*27− donors (marked with asterisk), and donors with unknown MHC genotype, (B) compared with the PB proportion of the clonotypes observed after increasing of repertoire analysis depth by TCR profiling of FACS-sorted CD8+ TRBV9+ or CD8− TRBV9+ cells from the PB of healthy HLA-B*27+ donors. Amino acid residues in variable positions of CDR3 are coloured accordig to the physical and chemical properties of the side-chains. PB samples from 15 of 25 AS patients were positive for at least one of the eight clonotypes discovered. Five of the clonotypes were previously published in several studies of SF and blood samples from patients with ReA [3–5]. Clonotypes CASSVGLFSTDTQYF, CASSVGLYSTDTQYF and CASSVGVYSTDTQYF from our study belong to the motif that was recently discovered in blood samples of AS patients . Occurrence of the clonotypes matching the motif was very similar for the two independent studies; the proportion of positive patients was 0.56, compared with 0.57 in our cohort. Thus, in the present study we independently discovered a group of similar TCR β clonotypes identical to or closely resembling the ones published previously for ReA and AS. By sequencing of CD8+ and CD4+ T cell subpopulations isolated from the PB of eight AS patients from our cohort, we attributed the AS-related clonotypes discovered in our study to the subpopulations. Seven of the eight clonotypes were detected exclusively in CD8+ cell samples, and CASSVGGFGDTQYF was found only in one sample of CD4+ cells (Fig. 1). Abundance of the AS-related clonotypes in peripheral blood of patients and healthy donors Involvement of T cell clones in inflammation should lead to clonal expansion and, thus, to increased number of cells with a particular TCR in the donor’s blood. To perform quantitative analysis of TCR β repertoires, we normalized raw sequencing data using UMI. The normalization minimizes the impact of PCR bias on the estimation of clonal abundances . Surprisingly, all eight clonotypes had a relatively low abundance in PB of patients despite their significant sharing in the AS cohort (Fig. 2A). The highest and lowest concentrations in PB observed for the AS-related clonotypes were ∼0.006 and ∼0.00008%, respectively. For comparison, the average proportion of top 1 T cell clonotypes in AS PB samples was 3.86% (interquartile range: 2.7%). Based on previous estimation of the efficiency of the TCR profiling technology used, most of sequenced unique UMI-labelled TCR β mRNA molecules represent distinct T cells . This means that the number of T cells with the most expanded AS-related TCR β was ∼6/100 000 PB T cells of the patient. Low abundance of T cell clones in peripheral blood might account for their absence in some samples, because of sampling or insufficiency of repertoire analysis depth. To increase the depth and test the presence of the AS-related clonotypes in PB of healthy HLA-B*27+ donors we sequenced repertoires of FACS-sorted TRBV9+ T cells from six donors analysed earlier in this study and one additional HLA-B27+ healthy individual. For each donor, we sorted ⩾35 000 TRBV9+ CD8+ T cells and ≥50 000 TRBV9+ CD8− T cells, thus obtaining a 10-fold increase in the number of analysed TRBV9+ cells compared with the whole blood sequencing (where we analysed ∼6700 TRBV9+ cells on average for both the AS cohort and HLA-B27+ healthy donors). Of the eight AS-related clonotypes discovered, we found four in three donors in sum, and only two donors had the clonotypes in the CD8+ T cell subset. The abundance of the clonotypes normalized for whole blood was much lower than in PB samples of AS patients (Fig. 2B). This finding indicates that AS-related clonotypes are also present in the PB of some healthy donors, albeit at extremely low frequencies. We also compared the TCR profiling depth for samples of AS patients where we observed at least one of the eight clonotypes (positive) with the ones where no AS-related clonotypes were detected (negative). The number of sequenced TCR cDNA molecules with the TRBV9–TRBJ2-3 combination was higher for positive samples (Wilcoxon rank sum test P < 0.002), and no difference was observed in the proportion of this combination in the TCR repertoire between positive and negative samples (supplementary Fig. S2, available at Rheumatology online). This result indicates that the proportion of negative samples in the AS cohort could be lowered with the increase in TCR profiling depth. Presence and enrichment of the discovered clonotypes in inflamed joint Targeting of discovered clonotypes to the site of inflammation would provide further support for their involvement in the disease. To test this, we performed TCR sequencing of SF CD8+ and CD4+ T cells obtained from the inflamed knee joint of four AS patients: three HLA-B*27+ and one HLA-B*27−. In each sample of CD8+ SF T cells from HLA-B*27+ AS patients, we found between four and seven of the eight TCR β clonotypes discovered as overrepresented in PB samples of AS cohort. Again, the clonotype TRBV9–CASSVGGFGDTQYF–TRBJ2-3 was detected exclusively in CD4+ SF cell sample and in only one donor. In addition, we analysed the TCR β repertoire of immunomagnetically sorted TRBV9+ T cells from the SF of one additional HLA-B*27+ AS patient (D), and also found four of the AS-related clonotypes. Therefore, we found each of the AS-related clonotypes discovered in the present study in the SF of HLA-B*27+ patients. Furthermore each SF sample from HLA-B*27+ patients (n = 4) turned out to be positive for at least several of the AS-related clonotypes. Notably, we did not observe any of the eight AS-related clonotypes in CD8+ or CD4+ SF T cell samples of the HLA-B*27− AS patient, and only TRBV9–CASSVGVYSTDTQYF–TRBJ2-3 was detected in the PB sample of this patient. Bystander activation of T cells in an inflammatory milieu may lead to the presence in the repertoire of an inflamed site of T cell clones activated without TCR engagement and, thus, not related to antigens represented in the site [21, 22]. Considering this, we analysed clonotype sharing between samples from HLA-B*27+ patients, suggesting that the group of shared clonotypes should be enriched by those recognizing site-specific antigens, whereas the repertoire of bystander T cells should be more diverse between patients. Owing to technical reasons, for patient D the TCR sequencing data were available only for TRBV9+ SF T cells, and this sample was excluded from analysis of clonal sharing and abundance. We identified 10 000–40 000 TCR β amino acid variants for each CD8+ SF T cell sample from HLA-B*27+ patients, but only 195 clonotypes were shared between at least two samples. Most of the shared clonotypes had TRBV9–TRBJ2-3 rearrangement, and six of the eight TRBV9–TRBJ2-3 clonotypes formed a group with a highly similar amino acid CDR3 sequence (Fig. 3A; supplementary Table S4, available at Rheumatology online). This group includes the five AS-related clonotypes discovered in PB samples and one additional variant, TRBV9–CASSAGLYSTDTQYF–TRBJ2-3, which was shared between all three CD8+ SF samples of HLA-B*27+ patients. The clonotype was found in only one unmatched PB sample of AS donors and was not detected in the PB of any of the healthy donors studied. Fig. 3 View largeDownload slide Highly similar CD8+ T cell clonotypes are enriched in SF of AS patients (A) CD8+ SF TCR β clonotypes shared between two (open circles) or all three (black dots) HLA-B*27+ AS patients studied form clusters based on the similarity of the predicted amino acid CDR3 sequence (one mismatch in amino acid CDR3 sequence allowed). The most prominent cluster comprises five clonotypes discovered in the peripheral blood (PB) of AS patients earlier in the present study (underlined) and one additional highly similar CD8+ T cell clonotype shared between all three SF samples. (B) Proportion of AS-related clonotypes from CD8+ T cells in matched SF and PB samples (patients A, B and C). Clonotypes with a significantly higher abundance in SF are marked with an asterisk (P < 0.05). Fig. 3 View largeDownload slide Highly similar CD8+ T cell clonotypes are enriched in SF of AS patients (A) CD8+ SF TCR β clonotypes shared between two (open circles) or all three (black dots) HLA-B*27+ AS patients studied form clusters based on the similarity of the predicted amino acid CDR3 sequence (one mismatch in amino acid CDR3 sequence allowed). The most prominent cluster comprises five clonotypes discovered in the peripheral blood (PB) of AS patients earlier in the present study (underlined) and one additional highly similar CD8+ T cell clonotype shared between all three SF samples. (B) Proportion of AS-related clonotypes from CD8+ T cells in matched SF and PB samples (patients A, B and C). Clonotypes with a significantly higher abundance in SF are marked with an asterisk (P < 0.05). We also compared the difference in abundance of the AS-related clonotypes between the SF and PB samples of same donors. Each of the clonotypes had a significantly higher abundance in SF compared with PB in at least one of the three sample pairs studied (Fig. 3B), suggesting their targeting to the inflamed tissue. Despite the fact that the AS-related clonotypes were enriched in the inflamed joint, they were not the most expanded clonotypes in SF, being ranked below ⩾40 other clonotypes by clonal size in all of the three samples. Discussion By combining high-throughput TCR sequencing and a new computational approach for analysis of the sharing of T cell clonotypes, which relies on estimation of the probability of TCR β VDJ rearrangement, we were able to discover a group of highly similar TCR clonotypes that were significantly overrepresented in AS patients [found in 15 of 25 patients (60%)]. For these clonotypes, the in silico estimated generation probability of TCR β appeared to be relatively low. In good agreement with that result, we found the clonotypes in sum in only 12 of 107 PB samples from the whole cohort of healthy donors (11.2%). Importantly, seven out of the eight clonotypes were not detected in PB of HLA-B27+ healthy donors with comparable analysis depth. Furthermore, these seven clonotypes were detected in SF from an inflamed joint of HLA-B*27+ AS donors (n = 4). Higher abundance of the clonotypes in SF compared with matched PB samples suggests targeting of T cell clones with such TCR β chains into the site of inflammation. The CD8+ status of these clonotypes was confirmed by sequencing of separated T cell subsets. After analysis of the SF samples, we added to the group of AS-related clonotypes one additional clonotype with a highly similar CDR3. The clonotype was detected in the CD8+ subset of all HLA-B*27+ SF samples, but in only one PB sample. Five of the AS-related clonotypes were reported previously in SF and PB of patients with ReA [3–5], and four of them were recently discovered in PB of AS patients . Three of our clonotypes matched the most frequent AS motif from the study by Faham et al. . The proportion of PB samples positive for the clonotypes was almost identical in our AS cohort. Thus, in the present study we demonstrate the independent discovery of T cell clonotypes associated with positive AS status, supporting the recent discovery  in an independent cohort of patients. We also support involvement of the clonotypes in AS pathogenesis by demonstrating their presence and enrichment in an inflamed joint in AS patients. The advanced algorithm, based on estimation of the probability of generation for shared clonotypes, allowed us to find the TCR motif characteristic for AS patients, using a much smaller cohort of donors compared with the study of Faham et al. . This demonstrates the power of the algorithm to search for clonotypes specifically enriched between samples even if an adequate control is not possible. The higher frequency of detection of the AS-related clonotypes in the repertoires of AS patients compared with healthy individuals, in combination with the relatively low probability of generation of such TCR β, reflects expansion of the clonotypes in peripheral blood of AS patients. Higher abundance of the clonotypes in AS patients was confirmed by direct comparison between the PB samples from patients and healthy donors. Several of these clonotypes were detected in the PB of HLA-B*27+ healthy donors only after a substantial increase of the depth of analysis of the repertoire, showing that the difference between healthy and AS individuals resides in expansion of the AS-related clonotypes rather than in the presence of such T cells in the repertoire. The group of AS-related clonotypes was initially discovered independently of their CDR3 amino acid sequence; however, the clonotypes appeared to have similar or identical amino acids in the same positions of CDR3 (Fig. 4A) that can be described by the amino acid motif. The sequences have a greater similarity if we consider the physical and chemical properties of their CDR3 amino acids, e.g. hydrophilicity  and average volume  (Fig. 4B and C). This suggests similar or identical antigen specificity for the clonotypes matching the TCR β motif. This is also supported by simultaneous expansion of several clonotypes with such a motif observed in each positive PB or SF sample. Potential HLA-B*27 restriction of the clonotypes also supports the suggestion. Discovery of the AS-related clonotypes with assumed similar specificity reinforces the arthritogenic peptide hypothesis of AS initiation; however, future studies with the aim of determining the antigen specificity of these clonotypes are necessary to clarify their relationship to the disease. Fig. 4 View largeDownload slide AS-related clonotypes have similar profile for CDR3 amino acid properties (A) Frequency-based logo representation of predicted CDR3 amino acid sequence of the AS-related clonotypes discovered in peripheral blood (PB) and SF samples in the present study (http://weblogo.threeplusone.com). (B) Hopp–Woods hydrophilicity and (C) average volumes of residues were calculated for each amino acid in TCR β CDR3 of the AS-related clonotypes discovered. The clonotypes are divided on two subgroups depending on Y or F at position 8. Fig. 4 View largeDownload slide AS-related clonotypes have similar profile for CDR3 amino acid properties (A) Frequency-based logo representation of predicted CDR3 amino acid sequence of the AS-related clonotypes discovered in peripheral blood (PB) and SF samples in the present study (http://weblogo.threeplusone.com). (B) Hopp–Woods hydrophilicity and (C) average volumes of residues were calculated for each amino acid in TCR β CDR3 of the AS-related clonotypes discovered. The clonotypes are divided on two subgroups depending on Y or F at position 8. The advanced TCR profiling technique with UMI-based normalization allowed us to demonstrate that the discovered AS-related clonotypes had a relatively low abundance in PB and, more importantly, were enriched but not the most abundant in SF during inflammation. Based on this, it can be speculated that these T cell clones are probably not the key players in the late stages of the disease. The high similarity of the CDR3 amino acid sequence between clonotypes detected in the SF of early ReA and AS patients, together with HLA-B*27 restriction and the clinical similarity between AS and ReA , suggests that the mechanism of initiation of AS might be similar to ReA at least in some patients, and these clonotypes could play an active role during the initial stages of AS. Although the AS-related clonotypes were enriched in AS patients, we were unable to detect them in ∼40% of our AS patients. The fraction of donors with no AS-related clonotypes is similar to the one reported by Faham et al. . We suggest that insufficient depth of repertoire sequencing and/or the presence of additional TCR motifs (not linked to TRBV9, or even restricted to the TCR α chain) could be among the reasons explaining our limited recall. Given the present level of knowledge, it is impossible to exclude an alternative explanation for our results, namely the hypothesis that the expansion of the identified AS-related clonotypes is a consequence of inflammatory joint damage during progression of ReA or AS in HLA-B27+ donors. Thus, further studies, designed to test the involvement of the identified TCR β clonotypes in the pathogenesis of AS and other SpAs, promise to shed light on the mechanisms of initiation of SpA. Acknowledgements We thank our donors, physicians and nursing staff of the clinics where samples were collected. Funding: This work was supported by the Russian Ministry of Education and Science [grant ID RFMEFI60716X0158]. Disclosure statement: The authors have declared no conflicts of interest. Supplementary data Supplementary data are available at Rheumatology online. References 1 Brown MA , Kenna T , Wordsworth BP. Genetics of ankylosing spondylitis–insights into pathogenesis . Nat Rev Rheumatol 2016 ; 12 : 81 – 91 . Google Scholar CrossRef Search ADS PubMed 2 Benjamin R , Parham P. Guilt by association: HLA-B27 and ankylosing spondylitis . Immunol Today 1990 ; 11 : 137 – 42 . Google Scholar CrossRef Search ADS PubMed 3 Duchmann R , May E , Ackermann B et al. HLA-B27-restricted cytotoxic T lymphocyte responses to arthritogenic enterobacteria or self-antigens are dominated by closely related TCRBV gene segments. A study in patients with reactive arthritis . Scand J Immunol 1996 ; 43 : 101 – 8 . Google Scholar CrossRef Search ADS PubMed 4 Dulphy N , Peyrat MA , Tieng V et al. Common intra-articular T cell expansions in patients with reactive arthritis: identical β-chain junctional sequences and cytotoxicity toward HLA-B27 . J Immunol 1999 ; 162 : 3830 – 9 . Google Scholar PubMed 5 May E , Dulphy N , Frauendorf E et al. Conserved TCR β chain usage in reactive arthritis; evidence for selection by a putative HLA-B27-associated autoantigen . Tissue Antigens 2002 ; 60 : 299 – 308 . Google Scholar CrossRef Search ADS PubMed 6 Faham M , Carlton V , Moorhead M et al. Discovery of T-cell receptor β motifs specific to HLA-B27-positive ankylosing spondylitis by deep repertoire sequence analysis . Arthritis Rheumatol 2016 ; 11 : 300 – 8 . 7 Mamedov IZ , Britanova OV , Zvyagin IV et al. Preparing unbiased T-cell receptor and antibody cDNA libraries for the deep next generation sequencing profiling . Front Immunol 2013 ; 4 : 456 . Google Scholar CrossRef Search ADS PubMed 8 Britanova OV , Putintseva EV , Shugay M et al. Age-related decrease in TCR repertoire diversity measured with deep and normalized sequence profiling . J Immunol 2014 ; 192 : 2689 – 98 . Google Scholar CrossRef Search ADS PubMed 9 Zvyagin IV , Mamedov IZ , Tatarinova OV et al. Tracking T-cell immune reconstitution after TCRαβ/CD19-depleted hematopoietic cells transplantation in children . Leukemia 2017 ; 31 : 1145 – 53 . Google Scholar CrossRef Search ADS PubMed 10 Pogorelyy MV , Elhanati Y , Marcou Q et al. Persisting fetal clonotypes influence the structure and overlap of adult human T cell receptor repertoires . PLoS Comput Biol 2017 ; 13 : e1005572 . Google Scholar CrossRef Search ADS PubMed 11 Shugay M , Britanova OV , Merzlyak EM et al. Towards error-free profiling of immune repertoires . Nat Methods 2014 ; 11 : 653 – 5 . Google Scholar CrossRef Search ADS PubMed 12 Bolotin DA , Shugay M , Mamedov IZ et al. MiTCR: software for T-cell receptor sequencing data analysis . Nat Methods 2013 ; 10 : 813 – 4 . Google Scholar CrossRef Search ADS PubMed 13 Nazarov VI , Pogorelyy MV , Komech EA et al. tcR: an R package for T cell receptor repertoire advanced data analysis . BMC Bioinformatics 2015 ; 16 : 175 . Google Scholar CrossRef Search ADS PubMed 14 R Core Team . R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. 2014 . 15 Kivioja T , Vähärautio A , Karlsson K et al. Counting absolute numbers of molecules using unique molecular identifiers . Nat Methods 2011 ; 9 : 72 – 4 . Google Scholar CrossRef Search ADS PubMed 16 Egorov ES , Merzlyak EM , Shelenkov AA et al. Quantitative profiling of immune repertoires for minor lymphocyte counts using unique molecular identifiers . J Immunol 2015 ; 194 : 6155 – 63 . Google Scholar CrossRef Search ADS PubMed 17 Elhanati Y , Murugan A , Callan CG , Mora T , Walczak AM. Quantifying selection in immune receptor repertoires . Proc Natl Acad Sci USA 2014 ; 111 : 9875 – 80 . Google Scholar CrossRef Search ADS PubMed 18 Putintseva EV , Britanova OV , Staroverov DB et al. Mother and child T cell receptor repertoires: deep profiling study . Front Immunol 2013 ; 4 : 463 . Google Scholar CrossRef Search ADS PubMed 19 Zvyagin IV , Pogorelyy MV , Ivanova ME et al. Distinctive properties of identical twins’ TCR repertoires revealed by high-throughput sequencing . Proc Natl Acad Sci USA 2014 ; 111 : 5980 – 5 . Google Scholar CrossRef Search ADS PubMed 20 Murugan A , Mora T , Walczak AM , Callan CG. Statistical inference of the generation probability of T-cell receptors from sequence repertoires . Proc Natl Acad Sci USA 2012 ; 109 : 16161 – 6 . Google Scholar CrossRef Search ADS PubMed 21 Doisne J-M , Urrutia A , Lacabaratz-Porret C et al. CD8+ T cells specific for EBV, cytomegalovirus, and influenza virus are activated during primary HIV infection . J Immunol 2004 ; 173 : 2410 – 8 . Google Scholar CrossRef Search ADS PubMed 22 Bou Ghanem EN , D’Orazio SEF. Human CD8+ T cells display a differential ability to undergo cytokine-driven bystander activation . Cell Immunol 2011 ; 272 : 79 – 86 . Google Scholar CrossRef Search ADS PubMed 23 Hopp TP , Woods KR. Prediction of protein antigenic determinants from amino acid sequences . Proc Natl Acad Sci USA 1981 ; 78 : 3824 – 8 . Google Scholar CrossRef Search ADS PubMed 24 Pontius J , Richelle J , Wodak SJ. Deviations from standard atomic volumes as a quality measure for protein crystal structures . J Mol Biol 1996 ; 264 : 121 – 36 . Google Scholar CrossRef Search ADS PubMed 25 Kaarela K , Jäntti JK , Kotaniemi KM. Similarity between chronic reactive arthritis and ankylosing spondylitis.A 32-35-year follow-up study . Clin Exp Rheumatol 2009 ; 27 : 325 – 8 . Google Scholar PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of the British Society for Rheumatology. All rights reserved. For Permissions, please email: firstname.lastname@example.org This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)
Rheumatology – Oxford University Press
Published: Feb 22, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera