Characterizing the T cell Repertoire in the Proximal Airway in Health and Disease

Abstract

Objectives:

Recent translational scientific efforts in subglottic stenosis (SGS) support a disease model where epithelial alterations facilitate microbiome displacement, dysregulated immune activation, and localized fibrosis. Given the observed immune cell infiltrate in SGS, we sought to test the hypothesis that SGS cases possessed a low diversity (highly clonal) adaptive immune response when compared with healthy controls.

Methods:

Single cell RNA sequencing (scRNA-seq) of subglottic mucosal scar in iSGS (n=24), iLTS (n=8) and healthy controls (n=6) was performed. T cell receptor (TCR) sequences were extracted, analyzed, and used to construct repertoire structure, compare diversity, interrogate overlap, and define antigenic targets using the Immunarch bioinformatics pipeline.

Results:

The proximal airway mucosa in health and disease are equally diverse via Hill framework quantitation (iSGS vs iLTS vs Control, p > 0.05). Repertoires do not significantly overlap between individuals (Morisita < 0.02). Among iSGS patients, clonality of the TCR repertoire is driven by CD8+ T cells, and iSGS patients possess numerous TCRs targeting viral and intercellular pathogens. High frequency clonotypes do not map to known targets in public datasets.

Conclusion:

SGS cases do not possess a lower diversity adaptive immune infiltrate when compared with healthy controls. Interestingly, the TCR repertoire in both health and disease contains a restricted number of high frequency clonotypes that do not significantly overlap between individuals. The target of the high frequency clonotypes in health and disease remain unresolved.

LAY SUMMARY

T cell repertoires in the proximal airway in both health and disease are equally diverse and are characterized by high frequency clonotypes that do not overlap between individuals and have unresolved pathogen targets.

INTRODUCTION

Subglottic stenosis (SGS) can occur after iatrogenic injury (post-intubation: iLTS)¹, as a manifestation of collagen vascular disease (Granulomatosis with polyangiitis: GPA)², or without an identifiable associated disease process (termed idiopathic SGS: iSGS)³. Over the last decade, translational science exploring the pathogenesis of SGS has developed a model where epithelial alterations facilitate microbiome displacement^4,5, dysregulated immune activation^6,7, and localized fibrosis^8-10. Given the pronounced immune cell infiltrate¹¹ observed in SGS cases¹², we sought to test the hypothesis that a clonal adaptive immune response against a conserved antigen was associated with tissue remodeling in the subglottic mucosa. Rather than interrogate antigens directly, our unbiased approach applied the principle that an individual’s T cell receptor (TCR) repertoire encodes their antigen exposure history.

T cells are essential effectors of the adaptive immune response, defending against both pathogens and malignant transformation while maintaining tolerance to self¹³. This specification is largely driven by the TCR expressed on the surface of each T cell. The TCR molecule in alpha/beta T cells is composed by one alpha and one beta chain, produced through combinatorial somatic rearrangement of multiple variable (V), diversity (D) (for the β-chain only), joining (J) and constant (C) gene segments¹⁴. The third complementarity determining region (CDR3) is created by the joining of the V-D-J and V-J segments in beta and alpha chain respectively and additional diversity is generated through the removal or insertion of non-germ-line nucleotides at each joining junction. This process generates a TCR repertoire with a theoretical range approaching 2 × 10¹⁹ unique TCRαβ pairs¹⁵. This vast diversity confers an ability to recognize a wide array of foreign and self-peptides presented on major histocompatibility complexes (MHC, in humans also termed HLA)¹⁶. Given the low probability of producing an exact VDJ rearrangement twice in one individual¹⁷, the TCR sequence can be used as a unique identifier of T cell clones. This allows use of an individual’s TCR repertoire as a proxy for their pathogen exposure history and supports profiling to generate meaningful biological interpretation of T cell function in health and disease.

Given prior work suggesting a role for T cell subsets in SGS pathogenesis¹², we sought to expand our understanding of the local adaptive immune response in the subglottic scar of SGS patients. Harnessing new tools in single cell RNA sequencing and expanded patient numbers, we interrogated the structure of the TCR repertoire in iSGS, iLTS, and healthy controls. Analysis of the TCR repertoire allows for comparison of host immune responses within and between disease states. Presence of a pathogen leads to clonal expansion of a T cell with a unique TCR due to stimulation by the cognate antigen (Fig. 1A). This stimulation changes the composition of the overall TCR repertoire, decreasing diversity as it is now dominated by a select few clonotypes (Fig. 1B). Comparing diversity between groups provides insight into the presence or absence of a primary antigen effector (Fig. 1C). Determining which antigens these TCRs target can be accomplished using public repositories of TCR CDR3β amino acid sequences and their known antigen targets (Fig. 1E). Analysis of the repertoire structure and interpatient similarity can support a convergent cause to disease (Fig. 1D, 1F). Our results suggest the proximal airway mucosa in health and disease has abundant infiltrating T cells dominated by a restricted number of high frequency clonotypes that do not significantly overlap between individuals.

Figure 1.

Clonal outgrowth of a T cell with a unique TCR (barcode) due to stimulation with its cognate antigen (A) that generates shifts in the composition of TCR repertoire (B). The change in T cell diversity can be detected using diversity analysis utilizing a wide variety of diversity metrics (C). Repertoire structure is determined based on what percentage of the total T cell population consists of a certain TCR. Repertoire structure conducted alongside diversity analysis provides insights into overall repertoire composition (D). Public databases of know TCR-antigen pairs can be quarried for matches, providing insight into antigen exposure (E). Overall T cell populations can be compared in similar and varied disease states through repertoire overlap analysis (F).

MATERIALS & METHODS

Patients

A total of 24 patients with iSGS, eight patients with iLTS, and seven healthy control patients were included in this study. (Table 1, Table S1). Diagnosis of subglottic stenosis was confirmed using previously described criteria¹⁸. Biopsies of airway scar or healthy subglottic mucosa from iSGS, iLTS, and three controls patients were obtained during routine endoscopic intervention in the operating room. The three control data sets obtained from public data were originally sourced from tracheal tissue of recently deceased organ donors. Consent for biopsy collection and analysis was obtained from each patient in accordance with Institutional IRB guidelines.

Table 1.

iSGS, iLTS, and healthy control samples used in analysis

Disease	Patients (N)	Cells	Unique CDR3β
iSGS	24	18552	13956
iLTS	8	29166	19279
Control	7	8740	5590

TCR Sequencing

TCR sequence data was sourced from in-house sequencing and supplemented by public repositories^{16, 17}. Sequencing techniques included ImmunoSeq (Adaptive biotechnologies Corp.), Illimani Smart-Seq2 assay, and 10XGenomics. A full list of each sample, sample source, data composition, sequence methodology, and data location can be found in supplemental table 1. Brief protocols for each methodology are as follows.

Adaptive ImmunoSeq:

Surgical biopsies of airway tissue was mechanical digested and T cells (CD4+ & CD8+) were sorted using a FACSARIA flow cytometer (BD Biosciences) directly into RLT buffer. Genomic DNA was extracted using a DNeasy Minikit (Qiagen) and high-throughput TCR sequencing was performed using the ImmunoSEQ assay (Adaptive Biotechnologies Corp.)^19-21.

Illumina SmartSeq2:

Surgical samples were prepared and digested into a single cell suspension as previously described¹⁰. 300-500 Cells of interest (e.g. CD8 T cells) are then identified and sorted with multiparameter flow cytometry. Indexed sorting collects the phenotyping data and plate location of each cell. cDNA library creation is carried out as described by Picelli et al. using a template switching oligo (TSO). After amplification, cDNA is fragmented by tagmentation using Nextera XT and individually dual indexed. Pooled libraries are sequenced on a HiSeq3000 (Illumina) in the VANTAGE core. Deconvolved reads in FASTQ format are trimmed and the retained reads are then aligned to the human transcriptome (hg38) using Bowtie2²², and quantified using the RSEM software package²³. Count matrixes from Bowtie2/RSEM are automatically fed into the SCONE software package²⁴ for quality control, evaluation and selection of the most appropriate normalization method. TRAPeS²⁵ is then used in conjunction with TopHat2²⁶ to assemble, align, and validate TCRα & TCRβ sequences present in each cell.

10XGenomics Single Cell Profiling:

We determined the distribution and phenotype of cellular populations present in iSGS airway scar by analyzing a scRNA-seq atlas⁴ containing 9 iSGS patients, 6 iLTS and 6 normal controls. scRNA-seq was performed using the 10x Genomics Chromium single cell immune profiling platform and cDNA libraries in accordance to manufacturers protocols. Suspended cells were loaded onto Chromium Next Gem Single Cell chip for cellular barcode addition, mRNA amplification, and subsequent RNA sequencing. Single cell integration for cellular identification was conducted via Harmony²⁷. Downstream processing was conducted using Seurat^28,29. Unsupervised clustering analysis classified the cell type based on PanglaoDB³⁰.

TCR Repertoire Data Analysis

TCR repertoire scrutinization was performed by integrating TCR sequences from all samples into the Immunarch 0.9.0 immunoinformatic analytic package in R (R Foundation for Statistical Computing, Vienna, Austria). Focus was placed on the TCR CDR3β regions amino acid and nucleotide sequence as it has been previously shown to play the largest part in receptor specificity^31,32. Diversity was calculated using the Hill Foundation number and Chao1 richness index. Repertoire overlap and similarity were calculated using Morisita’s overlap index. Rarefaction interpolation analysis was also performed using Immunarch. TCR antigen specificity was determined using three online databases linking CDR3β amino acid sequences with known antigens ^33-35.

Data Visualization and Statistical Analysis.

Bar and dot plots were visualized in Prism version 9.0 (GraphPad Software Inc., La Jolla, CA). TCR repertoire overlap, rarefaction interpolation, and gene usage heat maps were produced using internal Immunarch visualization based on ggplot2. Statistical significance was set at p value less than 0.05. Differences between x and y groups were determined using the Kruskal-Wallis and Mann-Whitney tests for normal and non-normal distributions, respectively. All statistical analyses were performed with Prism version 9.0.

RESULTS

The T cell repertoire of the proximal airway mucosa in health and disease can be characterized.

Following data curation and quality control filtering, we began characterizing the TCR repertoires in iSGS, iLTS, and healthy controls. Previous studies have shown that in some inflammatory disease states, TCR composition changes and clonality increase in T cells^36,37. To investigate whether this trend applies in iSGS, distribution of T cell clonality was determined. Each patient’s unique TCR clones were ranked based on cell count and their proportion amongst the T cell population was determined (Fig. 2A). This analysis was performed for each patient, and results were averaged for each disease state based on clone ID rank. (Fig. 2B). Dominant, highly expressed clones were seen in all groups. All groups experience a rapid drop-off of cell proportions down to multitudes of singularly expresses TCR sequences. iSGS and iLTS taper off abruptly at approximately clone 3-4, while controls had 10-11 high frequency clones. In order to rigorously quantify the repertoire structures between iSGS, iLTS and controls, diversity analysis was conducted. Hill based diversity estimation was utilized to reduce bias as it places varied emphasis on clonally enriched T cells. Diversity estimations at every Q value showed no significant difference between the three groups (Fig. 2C), indicating equivalently diverse T cell population in both disease (iSGS & iLTS) and healthy controls.

Figure 2.

Repertoire structure characterizes the proportionality of clonotypes in a patient specific TCR repertoire. More highly expanded clones are given lower Clone IDs (A). Repertoire structure characterization conducted for all iSGS, iLTS and healthy control samples. Average percent totals and standard deviations were determined and plotted. Analysis visualized the first 20 clones (majority of samples have fully plateaued, reaching individually expressed T cell clones by 10-15 clones). Each sample contained a highly expressed dominant clonotype (iSGS: 3.32% ± 4.68%, iLTS: 1.91% ± 1.103%, Control: 5.38% ± 7.06%) (B). Rarefaction analysis was first conducted on sample sets to ensure sufficient sample size for analysis. The Hill diversity estimation was calculated as an overall estimate for repertoire diversity. Hill based diversity was chosen as different aspects of the overall repertoire are emphasized at different Q values, providing a less biased diversity estimation. Higher Q values places increased emphasis on more dominant clonotypes and reduced weight on rare species. Four Q values were used. No significant difference was seen between any of the three groups over the range of Q values. (Hill Diversity: iSGS vs iLTS vs Control, Q1: 573.5 vs 748.0 vs 575.6, p = 0.6760, Q2: 372.7 vs 497.8 vs 368.2, p = 0.7105, Q3: 237.5 vs 306.9 vs 252.4, p = 0.8440, Q4: 170.5 vs 225.2 vs 171.1, p = 0.7992) (C).

To further investigate the characteristics of the adaptive immune response in the subglottic scar, specific T cell populations in iSGS were analyzed. Clones of CD4+ and CD8+ T cells were ranked according to proportion of the total population (Fig. 3A). Both groups had several highly expanded clones; however, CD8+ cells had a larger number of highly expanded clones. The CD8+ T cells experience a sharp and consistent decline in clonal expansion around clone 10. The CD4+ T cell population had few clonally expanded cells. Previous work investigating the T cell population of patients with iSGS has shown that the subglottic scar tissue contains increased CD69+ CD103+ CD8+ tissue resident memory cells (T_RM) compared to iLTS patients¹². Given these cells may play a role in the pathogenesis of iSGS, we interrogated the T_RM and tissue non-resident memory cells (T_NonRM). CD8+ T_RM cells contained more highly expressed clones then CD8+ T_nonRM although the diversity of the two populations was not significantly different by Chao1 estimation (Fig. 3B). In all three disease states, single dominant clones were observed in the repertoires. To determine if these dominant clones were retained over time, longitudinal analysis in one iSGS patient was conducted. TCR sequencing of biopsies taken from an iSGS patient 12 months apart showed the persistent of the same dominant T cell clone (Fig. 3C). The lack of experimental replicates, however, limits the ability to draw generalized conclusions on the temporal persistence of dominant T cell clones. No other TCRs matched between populations in the longitudinal analysis.

Figure 3.

Repertoire structure characterization was conducted on the CD4+ and CD8+ T cell population in iSGS patients. Five iSGS patients, patients 11 – 15, were included in this analysis. The CD8+ compartment contains a larger number of high frequency clones (Chao1 iSGS CD8 vs CD4; 3750 vs 270, P<0.001). Clonal expansion drops off after 10 clones in CD8+ T cells and 2 clones in CD4+ T cells Both cell types had a dominant clonotype (iSGS CD4+ T cells: 3.47 ± 5.65%, iSGS CD8+ T cells: 16.19 ± 16.97%) (A). CD8+ T_RM cells contained more highly expressed clones then CD8+ T_nonRM although the diversity of the two populations was not significantly different (Chao1 iSGS CD8_TRM vs CD8_nonTRM; 576 vs 346, P>0.05) (B). Longitudinal analysis in one patient was conducted. Biopsies of scar tissue were sequenced in the same patient 12 months apart. The dominant clone initially accounted for 48% of the population and 28% of the population after 12 months. No other clonotype persisted over time (C).

iSGS patients have T cells with CDR3B sequences that map to known antigenic targets.

We have previously shown that the subglottis is a reservoir for adaptive immunological memory and contains a population of CD8+ T cells with receptors that recognize antigens for viral and intracellular pathogens¹². To provide a more nuanced and complete understanding of these antigen specific T cells, a similar analysis was conducted including our newly acquired iSGS patient data as well as iLTS and healthy control samples. TCR repertoires were interrogated for CDR3β amino acids sequences matching that of public TCR databases. A variety of pathogen-specific TCRs were identified, including cytomegalovirus (CMV), Epstein Barr virus (EBV), influenza (Flu), COVID-19, mycobacterium tuberculosis (M.Tb), hepatitis C virus (HCV), human immunodeficiency virus (HIV), yellow fever virus (YFV), and melanoma (Fig. 4A). ISGS and iLTS groups did not have an increased number of patients with TCRs targeting CMV, EBV, or Flu when compared to the control population (Fig. 4B). Presence of a wide variety of T cells with known antigen targets supports the assertion that the subglottic mucosa harbors viral memory, holding true for both stenotic and healthy patients. With the presence of dominant clones in each patient and the indication of temporal persistence, we sought to investigate potential antigen targets for these clones specifically. However, no patients dominant clonotype mapped to that of a known pathogen specific TCR.

Figure 4.

TCR sequences from all iSGS, iLTS, and Control patients were compared to public databases (VDJdb, McPAS-TCR, and TBAdb) linking TCRs to known cognate antigens. The average number of CDR3β sequences matching a specific disease was determined and plotted. Each point represents a distinct patient. Pathogen hits include: cytomegalovirus (CMV), Epstein Barr virus (EBV), influenza, homo sapiens self-antigens, Covid-19, mycobacterium tuberculosis (M.Tb), hepatitis C virus (HCV), human immunodeficiency virus (HIV), yellow fever virus (YFV), melanoma, and neoantigens (A). Percent of the total population containing at least one TCR clone targeting either CMV, EBV, or Flu was determined. Confidence intervals were determined, and mean values were compared using the chi-squared test. No significant difference between groups was observed in the percent of patients positive for each viral pathogen (B).

iSGS TCR repertoires show limited interpatient similarity.

To understand the relative interpatient homogeneity of the lymphocyte population, TCR repertoire overlap analysis was conducted. The identity of shared TCR CDR3β amino acid sequences between patients was determined (Fig. S1). All groups showed a relatively low number of exact CDR3β replicates between patients, all having less than 10 exact TCR matches between samples. Of the overlapping clonotypes, none were the most highly expanded, dominant clonotype. To account for sample size differences between groups, the Morisita Overlap Index was employed (Fig. 5, Fig. S2). Uniformly low similarity between all groups was observed. iSGS and iLTS patients appear to possess a unique T cell population with limited similarity across patients, having largely private TCR repertoires. Consistent with this finding, there is minimal bias observed in V(D)J gene utilization (Fig. S3).

Figure 5.

Overall similarity of the TCR repertoires in iSGS, iLTS, and control patients was determined and plotted via heat map. Morisita’s overlap index was chosen for repertoire comparison as it accounts for differences in repertoire size between patients, allowing for a more uniform and applicable comparison. Morisita’s overlap index ranges from 0 - 1, no similarity - perfect similarity. Minimal intergroup and intragroup similarity is seen, with overlap never exceeded 0.02. Average Morisita values for each disease was determined and compared by ANOVA test (iSGS vs iLTS: 2.6*10⁻⁴ vs 6.7*10⁻⁴, p = 0.2108, iSGS vs Control: 2.6*10⁻⁴ vs 3.3*10⁻³, p = 0.9806, iLTS vs Control: 6.7*10⁻⁴ vs 3.3*10⁻³, p = 0.958).

DISCUSSION

Subglottic stenosis (SGS) can occur after multiple disparate physiologic insults. Given the pronounced immune cell infiltrate¹¹ observed in SGS cases¹², we sought to test the hypothesis that a clonal adaptive immune response against a conserved antigen was associated with tissue remodeling in the subglottic mucosa. Rather than interrogate antigens directly, our unbiased approach interrogates the (TCR) repertoire in iSGS, iLTS and healthy patients. Interestingly, our results reveal a common structure to the adaptive immune repertoire in both health and disease. In each individual, a small number of unique high frequency clonotypes (of unknown antigen specificity) are accompanied by a diverse group of additional TCRs that appear to target viral and intercellular pathogens.

Linking antigen specificity measurements to T cell clonality is critical for understanding the drivers of the T cell response in subglottic disease. However, several aspects of T cell antigen recognition make mapping TCR sequence to antigen specificity a challenging problem. First, the frequency of rare populations of antigen-specific T cells may be as low as one in a million cells³⁸, which makes detection of these cells difficult. Second, high levels of variability due to the polymorphic nature of the MHC, the polyspecificity of both the TCR and the peptide-MHC complex (pMHC), and the wide variety of potential epitopes encoded by a single antigen substantially add to the complexity of resolving TCR antigen specificities.³⁹ Lastly, the relatively weak affinity and avidity of most TCR–pMHC interactions renders selective isolation of antigen-specific T cell populations nontrivial.^{39, 40}

Prior work from the Rheumatology literature has demonstrated that individual responses to auto-antigens are unique and polyclonal, highlighting the importance of interrogating the TCR repertoire within affected tissue³⁷. While theoretical estimates of TCR clonal diversity may reach 10¹⁹ unique receptors¹⁶, this number is dwarfed by the potential number of antigenic peptides that could be encountered, suggesting cross-reactivity of TCRs as a component of effective immunity. Several studies of TCR cross-reactivity have used combinatorial peptide libraries to estimate how many peptides a TCR can recognize⁴¹ and demonstrated that a single TCR can recognize multiple peptides in the context of a single MHC. These results highlight the potential of TCR cross-reactivity to elicit autoimmunity⁴². They also suggest that additional techniques to map antigen specificity to the identified T cell clones in the proximal airway will be critical to improved understanding of the role of the adaptive immune response in subglottic stenosis.

Although this study enhances our understanding of the immunologic drivers of T-cell activity in iSGS, there are several limitations to address. First, limited data availability hampers our ability to draw further conclusions and ensure elimination of unintended biases. Given the rarity of iSGS, sparse data sets make rare populations of antigen specific T cells difficult to detect. Studies investigating the TCR repertoires of more common disorders such as lung cancer^{43, 44} or viral infections^{45, 46} have access to orders of magnitude more CDR3β sequences. Continued work on expanding patient samples will help to confirm these initial results. The composition of an individual’s TCR repertoire is the result of many forces, including age and HLA genotype. Previous work has shown that the HLA genotype of iSGS patients is similar to matched controls⁴⁷. In addition to the role of HLA in driving overlap in TCR responses, data suggest increased age correlates with more shared low frequency TCR clones⁴⁸. The average age of our iSGS cohort meets immunologic definitions of “increased age” (>40 years) at 54 (+/− 8 years), and yet there was minimal overlap in TCR clones observed. Given the private TCR repertoires observed, neither HLA nor age appears to influence the local TCR repertoire in the proximal airway. This finding may be limited by the small number of patients, or restricted number of sequenced T cells per patient. Furthermore, while iSGS patients have T cells with CDR3β sequences that map to known antigenic targets, the large number of viral epitopes identified may relate to structural biases in the available public datasets. Examining the distributions of antigens, the three public TCR libraries (VDJdb, TRAdb, McPAS-TCR) shows that CMV, EBV, and influenza are the most represented antigens in the curated public data. CMV constitutes 28% of all public library antigenic targets mapped to known CDR3β sequences (Fig. S4).

Additionally, T cell repertoires are frequently characterized by the number and frequency of unique TCR gene sequences they contain using diversity indices (richness, Shannon entropy, and others related to Hill’s framework). While these indices show promise as correlates of immunologic health in infection, aging⁴⁹, and anti-tumor responses⁵⁰, they overlook fundamental features of repertoire function. For example, sequence diversity cannot indicate what antigens a repertoire binds (the indices do not inform us if a TCR contains epitope-binding capacity for many different epitopes or for only a few)⁵¹. The reason for this shortcoming is that sequence diversity measures only the number of different TCRs, but not their basic function: epitope binding. While high frequency clones can be shared between individuals acutely after exposure to a common antigen⁵², the TCRs that emerge in response to a pathogen are not always identical. Rather the repertoires of different individuals can possess non-identical but highly similar sequences, suggesting a response to multiple epitopes of the same antigen⁵³. Interestingly, work in identical twins following vaccination with a live attenuated virus supports this phenomenon. Despite identical genetic backgrounds and antigenic exposures, twins mount non-overlapping but structurally similar responses to common antigens ^53-55. Taken in context, additional work interrogating the local immune response in the proximal airway mucosa will need to integrate new computational tools in network analysis^32,56 and biophysical properties⁵⁷, along with confirmatory in vitro modeling to better define the target of the observed T cell response.

In this study we harness new tools in TCR repertoire analysis to help characterize the immune response in iSGS. The applicability of this study in Otolaryngology is broad. These results offer insight into the presence of a limited number of dominant TCR clones in the proximal airway in states of disease and health, with variability in the identity of these clones across iSGS patients. This study broadens our understanding of the immunologic basis of iSGS as well as the role of the proximal airway in the immune response against respiratory pathogens. Future investigations of the most highly clonal T cell targets may lead to further insights into the etiology of iSGS and alternate mucosal diseases in the head and neck. Most critically, improved understanding of disease pathogenesis may open the door to innovative new treatments.

CONCLUSION

Our results suggest that the TCR repertoire in the subglottic mucosa in both health and disease is diverse yet contains a select number of high-frequency clonotypes whose targets remain unresolved. Additionally, numerous clonotypes are found in distinct patient samples with specificity for viral and intracellular pathogens, suggesting an unappreciated role in the proximal airway for the response to inhaled pathogens.

Supplementary Material

Fig S1

Figure S1. Composition of each of the three public TCR libraries, VDJdb, TBAdb, and McPAS-TCR were determined. Significant bias in overall library composition for CMV, candida albicans, influenza, EBV, and tetanus toxoid are seen.

Fig S2

Figure S2. Total numbers of exact CDR3β amino acid sequence matches between patients in iSGS, iLTS, and control groups. Each number represents the total unique sequence matches. No matches were between dominant clones.

Fig S3

Figure S3. Morisita overlap index between patients in iSGS, iLTS, and control groups.

Fig S4

Figure S4. Somatic V(D)J recombination gene usage was determined for each patient in the three groups and total gene occurrences compared. No significant bias toward a gene or groups of genes was seen.

Tab S1

Supplemental Table 1. Full breakdown of each sample, including information on how the data was obtained, the technique for sequencing, total cell count, total unique CDR3β sequences, and whether certain cell types are differentiated in the population.