{"paper_id":"ab1a87a2-9d4f-4165-a5f5-53ba591bb2dd","body_text":"T cell receptor sharing in hypersensitivity \npneumonitis \nWezi Sendama1,2, Wendy Funston2, Richard CH Davidson1,2, Anthony J Rostron1,3* , A \nJohn Simpson1,2* \n1Translational and Clinical Research Institute, Newcastle University, Newcastle upon \nTyne, United Kingdom \n2The Newcastle upon Tyne Hospitals NHS Foundation Trust, Newcastle upon Tyne, \nUnited Kingdom \n3South Tyneside and Sunderland NHS Foundation Trust, Sunderland, United Kingdom \n*These authors contributed equally. \nORCID IDs: Wezi Sendama 0000-0002-4686-3368; Wendy Funston 0000-0002-6521-\n1103; Richard CH Davidson 0000-0001-7118-4941; Anthony J Rostron 0000-0002-9336-\n1723; A John Simpson 0000-0003-4731-7294. \nCorresponding author: Wezi Sendama, 2nd floor William Leech Building, Medical \nSchool, Framlington Place, Newcastle upon Tyne NE2 4HH. Email: \nwezi.sendama@newcastle.ac.uk \nAuthor contributions: WS devised the study and performed analysis; WS, WF , RCHD, \nAJR and AJS interpreted results and drafted the manuscript. \nFunding support: WS is a National Institute for Health and Care Research (NIHR) \nAcademic Clinical Lecturer. AJS is an NIHR Senior Investigator. The views expressed are \nthose of the authors and not necessarily those of the NIHR or the Department of Health \nand Social Care.\n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.11.13.623392doi: bioRxiv preprint \n\n1 \n \nAbstract \nHypersensitivity pneumonitis (HP) is characterised by an excessive pulmonary T cell \nresponse in susceptible individuals after exposure to inhaled antigens. The most \neffective treatment for the condition is antigen avoidance, but in most cases an antigen \ncannot be identified. Profiling antigen-specific T cell responses may form the basis of a \nstrategy to identify causative antigens of the disease. We used public RNA sequencing \ndata and reconstructed T cell receptor (TCR) repertoires from blood and \nbronchoalveolar lavage samples from patients with HP , idiopathic pulmonary fibrosis \nand healthy controls. After excluding TCR sequences likely to be related to common \nmicrobial exposures in patients with HP , we identified TCRs shared between patients \nwith shared human leukocyte antigen alleles, indicating patients likely to have a \ncommon causative antigen. We also identified clusters of TCRs containing identical and \nsimilar TCR clones in individual patients that are plausibly related to the causative \nantigens in those patients. These results establish the feasibility of profiling TCR \nrepertoires to identify antigens in HP.  \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.11.13.623392doi: bioRxiv preprint \n\n2 \n \nIntroduction \nHypersensitivity pneumonitis (HP) results from a disordered pulmonary immune \nresponse to antigens inhaled by a susceptible individual. The antigens that can be \nassociated with the development of HP are diverse.1 A key pillar of disease treatment is \nantigen avoidance, but it is also the case that in up to 60% of cases a causative antigen \ncannot be identified despite rigorous efforts.2 Failure to identify a causative antigen is \nassociated with poorer disease outcomes.3 \nThe pulmonary immune response in HP is characterised by a lymphocyte-rich alveolitis, \nand the cellular portion of bronchoalveolar lavage (BAL) fluid from patients with HP can \ncomprise up to 80% lymphocytes (compared to 15% or lower in control subjects).1 The T \nlymphocytes that accumulate in the lungs of patients with HP do so in an oligoclonal \nmanner, suggesting that immune recognition of the causative antigen is responsible for \nthe lymphocytic alveolitis.4 \nT cell-mediated immune recognition depends upon the presence of a T cell expressing a \nT cell receptor (TCR) with the appropriate structure to bind the complex of an antigenic \npeptide and the major histocompatibility complex (MHC; also known as human \nleukocyte antigen or HLA in humans) molecule upon which it is presented to the T cell. \nThe TCRs expressed by an individual’s naïve T cells are greatly diverse to allow for \nrecognition of the diverse antigens that an individual may encounter, with the diversity \nresulting from the process of V(D)J recombination.5 V(D)J recombination is the process \nby which the genes encoding TCRs are assembled in each naïve T cell from the \nstochastic selection of variable (V), diversity (D) and joining (J) TCR gene segments in \nthe germline.6 The diversity afforded by the process means that each individual’s T cell \nrepertoire can contain in the region of 20 million unique TCR beta chain (TCRβ) amino \nacid sequences.7 \nDespite the richness of TCR repertoires within individuals, it is sometimes the case that \nthe T cells bearing identical TCR amino acid sequences arise independently in multiple \nindividuals. When this TCR sequence sharing is noted in the context of an immune \nresponse it is known as a public T cell response. Public T cell responses occur more \nfrequently than would be explained by chance in part because V(D)J recombination is \nstochastic but not completely random.8,9 Antigen-specific public T cell responses have \nbeen observed in multiple individuals in infectious diseases, in malignancies and in \nautoimmune diseases.9 \nIt is not known whether individuals with HP with the same causative antigen exhibit \nantigen-related public T cell responses. There is circumstantial evidence that this may \nbe the case, including the observation of a shared bias in TCR gene segment expression \nin multiple patients with HP10, and the observation of beryllium-specific public T cell \nresponses in chronic beryllium disease (an exposure-related interstitial lung disease \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.11.13.623392doi: bioRxiv preprint \n\n3 \n \nclinically similar to HP).11 If this phenomenon were replicated in HP , patients with an \nunknown causative antigen could have the antigen identified if they shared clonally \nexpanded TCR sequences with a patient whose antigen was known, providing that the \nexpanded TCRs were not identifiably associated with exposures to antigens unrelated to \nthe disease. \nBy reanalysing RNA sequencing data from online repositories, we found evidence of \nlung T cell receptor sequences shared between patients with HP. After eliminating \nsequences previously documented to be associated with public T cell responses to \ncommon pathogens, we identified TCR sequences plausibly specific to HP antigens, as \nwell as candidates for the HLA alleles necessary for immune recognition of the cognate \nantigens. These findings suggest that an approach of surveying T cell repertoires to \nidentify causative antigens in HP is feasible. \nMethods \nRNA sequencing data \nRNA sequencing data were downloaded from the NCBI Gene Expression Omnibus \n(GEO; https://www.ncbi.nlm.nih.gov/geo/). The bulk of the data analysed were obtained \nfrom GEO accession GSE271789. To minimise the influence of technical factors on the \ncomposition of the reconstructed TCR repertoires, samples that were pooled prior to \nsequencing were excluded. Peripheral blood mononuclear cell (PBMC) samples from 12 \npatients with HP (all with fibrotic HP), 15 patients with idiopathic pulmonary fibrosis \n(IPF) and 15 healthy controls were analysed. BAL samples from 10 patients with HP (6 \nwith fibrotic HP and 4 with non-fibrotic HP) and 10 patients with IPF were also analysed \nfrom this dataset. PBMC and BAL samples were not paired samples from the same \nindividuals in this dataset. \nBecause GSE271789 did not contain BAL samples from healthy controls, data from BAL \nsamples from 6 healthy controls were downloaded from GSE136587. \nT cell receptor repertoire reconstruction and analysis \nImmune receptor repertoires were reconstructed using TRUST4 version 1.1.2, with \ndownloaded fastq files as input.12 TCRβ sequences were used for analysis, with the \namino acid sequences of the third complementarity determining region (CDR3) taken as \nthe sequences in question. Incompletely sequenced CDR3β polypeptide chains were \nexcluded, as were sequences that did not begin with an N-terminal cysteine (C) residue \nor end with a C-terminal phenylalanine (F) residue. \nRepertoire metrics (number of clones, number of unique clonotypes) were calculated \nusing the immunarch version 0.9.1 package in R (version 4.2.2; R Foundation for \nStatistical Computing). Groups were compared using Wilcoxon rank sum tests, with \nBonferroni-Holm adjusted p < 0.05 set as the threshold for statistical significance. \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.11.13.623392doi: bioRxiv preprint \n\n4 \n \nPublic and semi-public TCRβ sequences were identified using tcrdist3 version 0.2.2, \nwith concatenated TCRβ repertoires as input.13,14 Public sequences were defined as \nidentical TCRβ sequences occurring in more than one individual. A sequence was \ndeemed semi-public if a sufficiently similar sequence occurred in at least one other \nindividual. The threshold for similarity was 18 TCRdist units. TCRdist units are a \nweighted measure of distance between TCR sequences, with the greatest weight placed \non differences in CDR3 regions and lesser weight on differences between CDR1, CDR2 \nand CDR2.5 sequences.14 Penalty scores of between 0 and 4 units for each amino acid \nsubstitution are applied (according to a BLOSUM62 substitution matrix).15 Where \nsequences are different lengths consecutive gaps are inserted into the shorter \nsequence at positions that minimise substitution penalties, but each gap carries the \nmaximum penalty of four units. As an example, with CDR3 penalties carrying threefold \nweighting compared to other CDRs (by our criteria to determine semi-publicity), a single \namino acid substitution in the CDR3 would result in two otherwise identical TCR \nsequences being up to 12 TCRdist units apart. \nIdentification of clones under antigenic selection \nTCRs were clustered into similarity neighbourhood groups, with the rationale that \nepitope-specific T cell responses involve expansion of both identical and highly similar \nT cell clones.13 In individual patient repertoires TCRs were placed in the same cluster if \nthey were within 48 TCRdist units of one another, with CDR3 dissimilarities attracting a \nsixfold weighting in TCRdist score compared to other CDRs. This weighting allowed for \nneighbour dissimilarities of up to two CDR3 amino acid substitutions or gaps with \nidentical TCR V-gene segments, or identical CDR3 sequences with differing TCR V-gene \nsegments.16 \nThe number of within-repertoire neighbours was compared to the expected number of \nneighbours based on a reference repertoire. 9.6 × 105 TCR sequences were sampled \nuniformly from 8 human umbilical cord blood samples to provide an ostensibly antigen-\nunstimulated control reference.16 The cord blood samples are derived from experiments \nperformed by Britanova and colleagues17, and are available through the NCBI Sequence \nRead Archive under accession PRJNA316572. The model comparing numbers of within-\nrepertoire neighbours to numbers of neighbours in cord blood reference repertoires was \nbased on the expectation of the number of neighbours following a Poisson \ndistribution.16,18 Numbers of within-repertoire neighbours were considered different to \nexpected numbers of neighbours where the p-values computed by the model (adjusted \nfor multiple comparisons)19 were less than 0.001. Probabilities of CDR3 generation were \nestimated using the OLGA algorithm.20 Sequences occurring in expanded clones \nexclusive to HP patients with greater than expected numbers of within-repertoire \nneighbours were considered to be under antigenic selection. \nExclusion of previously annotated TCRβ sequences \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.11.13.623392doi: bioRxiv preprint \n\n5 \n \nTCRβ sequences within 48 TCRdist units of entries in VDJdb21 (sixfold CDR3 weighting) \nwere deemed to be related to the epitopes in the VDJdb annotations (mostly common \nviral and bacterial pathogens) and thus unrelated to HP. A snapshot of the VDJdb \ndatabase was downloaded in May 2024 and used for the analyses. \nPrediction of HLA genotypes \nHLA genotypes were predicted with T1K version 1.0.6, with fastq files as input.22 \nPredictions were made for HLA-A, HLA-B, HLA-C, HLA-DRB1, HLA-DPB1 and HLA-DQB1 \ngenes. \nCode availability \nThe code to produce the analyses is available at \nhttps://github.com/wezisendama/HP_TCRsharing. \nResults \nBAL samples in HP contain greater numbers of T cell clones and unique clonotypes than \ncontrol or IPF samples \nAs suggested by the lymphocytic alveolitis seen in HP , the TCR repertoires \nreconstructed from BAL samples from patients with HP contained greater numbers of \nclones compared to BAL samples from patients with IPF or healthy controls (Figure 1). \nRepertoires from patients with HP also contained greater numbers of unique \nclonotypes, suggesting concurrent expansions of multiple T cell clones in HP. There \nwere no differences in numbers of individual clones or unique clonotypes in the \nrepertoires reconstructed from PBMC samples, suggesting at least a partially \ncompartmentalised pulmonary T cell immune response in HP . We therefore analysed \nBAL samples further. \nSelected public TCRβ sequences shared between patients with HP suggest shared \ncausative antigens \nAnalysis with tcrdist3 identified 15 identical CDR3β sequences shared between at least \ntwo patients with HP in BAL samples (Table 1). All patients with shared sequences also \nshared at least one HLA allele to second field resolution (Table 2). \nAfter excluding public sequences that were similar to VDJdb entries, sequences that \noccurred in BAL or PBMC samples from study participants who did not have HP , and \nsequences that had a number of within-repertoire neighbours not significantly different \nto the expected number of neighbours, two sequences shared between two pairs of \npatients with HP remained (Table 1, shaded rows). Although only single clones of these \nsequences were detected in the samples, the likelihood of identical sequences with low \nprobabilities of generation (as determined by the OLGA estimates of generation \nprobability as well as the absence of neighbours in cord blood samples) being detected \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.11.13.623392doi: bioRxiv preprint \n\n6 \n \nexperimentally in two individuals without clonal proliferation in the sampled repertoires \nis low.23 These sequences are therefore plausibly related to HP antigens, with the \nimplication being that the pairs of patients share causative antigens. \n13 semi-public sequences (non-identical, but within 18 TCRdist units, as described in \nthe methods section) were identified. None of these groups of sequences met the \ncriteria above to be considered candidate HP-related sequences. \nExpanded T cell clones with greater than expected within-group neighbours represent \nclones under antigenic selection \nDash and colleagues observed that TCR sub-repertoires involved in an antigen-specific \nimmune response are composed of clusters of highly similar receptors alongside \noutlying receptors that are more distinct.13 We therefore considered whether we could \nuse this principle to identify which clones within a patient’s repertoire could be related \nto the causative HP antigen even without evidence of sequence sharing with another \npatient with HP . Bearing in mind that sequences with a higher probability of being \ngenerated by V(D)J recombination are more likely to cluster even without the influence \nof antigen recognition24, we also compared each sequence’s number of within-\nrepertoire neighbours to its expected number of neighbours (using cord blood samples \nas the antigen-naïve reference). \nAfter excluding sequences within 48 TCRdist units of VDJdb entries and sequences that \nalso occurred in samples from participants who did not have HP , several examples were \nidentified of clones in the BAL repertoires of individual patients that may be related to \ncausative antigens in HP (Table 3). \nDiscussion \nWe present evidence of TCR sharing in HP that is likely to be related to the causative HP \nantigens in the patients sharing the receptor sequences. To our knowledge, TCR sharing \nin HP has not been reported previously. Antigen identification in HP remains a \nchallenge, but we are hopeful that the principle established here might allow \nidentification of a patient’s causative antigen where HP-related TCRs are shared with a \npatient whose antigen is known. \nWe also identify clusters of TCR sequences in individual patients that are likely to be \nrelated to disease in HP . While we cannot present evidence of these sequences being \nshared with other patients, we suggest that there is value in establishing a database \nsimilar to VDJdb that annotates such sequences with the patients’ causative antigens \nand HLA genotype where known. Where the sequences are subsequently identified in \nother patients with HP sharing the same antigen (and at least one HLA allele) an \nannotation denoting the level of confidence in the association can be updated. \nConversely, where the sequence is confirmed experimentally to be associated with a \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.11.13.623392doi: bioRxiv preprint \n\n7 \n \ndifferent pathogen the database entry may be removed. Antigen identification could \ntherefore be aided by cross-references against the database. Open access of the \ndatabase could allow contributions from clinicians and researchers globally, \nmaximising its utility. \nThere are limitations in our study. The absence of BAL samples from control participants \nin the GSE271789 dataset means that data from separate experiments had to be \nconsidered. It appeared that the T cell repertoires reconstructed from the healthy \ncontrol BAL samples from GSE136587 were of similar size to the repertoires from \npatients with IPF from GSE271789. This is expected and gives some confidence in the \nvalidity of the control data given that there are similar numbers of lymphocytes per unit \nof BAL volume in patients with IPF and healthy controls25, but we cannot exclude the \nimpact of technical differences between the experiments that yielded the repository \ndata. \nAnother limitation was the absence of paired BAL and PBMC samples from participants \nin the GSE271789 dataset. This could have increased the confidence in our finding of a \ncompartmentalised immune response in HP by allowing clonotypes to be tracked \nbetween tissue compartments. There is prior evidence that supports a proportionally \ngreater T cell response in the lung compared to peripheral blood in HP including the \nobservation of a lung-specific lymphocyte expansion that abates when the causative \nantigen is removed4, and our own evidence of lung-resident T cells providing an \nappreciable portion of the HP immune response.26 These study results support our \nfindings from the comparisons of BAL and PBMC repertoire sizes despite us not having \naccess to paired samples. \nWhile VDJdb represents a vast database of experimentally confirmed TCR specificities, \nby definition it cannot be complete. We aimed to account for this by excluding TCR \nsequences from consideration as HP-related if they were within a liberally defined \ndistance of VDJdb entries, and then by considering shared TCRs that have a low \nprobability of generation and did not occur in non-HP samples. While we therefore \ncannot completely exclude the possibility that the pairs of individuals share the \nunexpected TCRs because of responses to an HP-unrelated epitope, we believe that a \nlung compartment TCR without neighbours in an antigen-naïve reference sample \nshared by two individuals with the same lung disease is unlikely to be explained by \nchance. \nOur work here establishes the feasibility of using T cell repertoires to identify antigens in \nHP , and our use of publicly available RNA sequencing data is a strength as it implies that \nit can be done at relatively low-cost using samples collected for other indications. A \nlow-cost method to identify antigens in HP has the potential to improve outcomes in the \ndisease, and future studies should aim to establish the clinical utility of such an \napproach in the management of the disease. \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.11.13.623392doi: bioRxiv preprint \n\n8 \n \nReferences \n1. Greenberger, P . A. Hypersensitivity pneumonitis: A fibrosing alveolitis produced by \ninhalation of diverse antigens. Journal of Allergy and Clinical Immunology 143, \n1295–1301 (2019). \n2. Nogueira, R., Melo, N., Novais e Bastos, H., Martins, N., Delgado, L., Morais, A. & C. \nMota, P . Hypersensitivity pneumonitis: Antigen diversity and disease implications. \nPulmonology 25, 97–108 (2019). \n3. Petnak, T., Thongprayoon, C., Baqir, M., Ryu, J. H. & Moua, T. Antigen identification \nand avoidance on outcomes in fibrotic hypersensitivity pneumonitis. European \nRespiratory Journal 60, 2101336 (2022). \n4. Facco, M., Trentin, L., Nicolardi, L., Miorin, M., Scquizzato, E., Carollo, D., Baesso, I., \nBortoli, M., Zambello, R., Marcer, G., Agostini, C. & Semenzato, G. T cells in the lung \nof patients with hypersensitivity pneumonitis accumulate in a clonal manner. \nJournal of Leukocyte Biology 75, 798–804 (2004). \n5. Livák, F . In vitro and in vivo studies on the generation of the primary T-cell receptor \nrepertoire. Immunological Reviews 200, 23–35 (2004). \n6. Schatz, D. G. & Ji, Y . Recombination centres and the orchestration of V(D)J \nrecombination. Nat Rev Immunol 11, 251–263 (2011). \n7. Qi, Q., Liu, Y ., Cheng, Y ., Glanville, J., Zhang, D., Lee, J.-Y ., Olshen, R. A., Weyand, C. \nM., Boyd, S. D. & Goronzy, J. J. Diversity and clonal selection in the human T-cell \nrepertoire. Proceedings of the National Academy of Sciences 111, 13139–13144 \n(2014). \n8. Venturi, V ., Quigley, M. F ., Greenaway, H. Y ., Ng, P . C., Ende, Z. S., McIntosh, T., \nAsher, T. E., Almeida, J. R., Levy, S., Price, D. A., Davenport, M. P . & Douek, D. C. A \nMechanism for TCR Sharing between T Cell Subsets and Individuals Revealed by \nPyrosequencing. The Journal of Immunology 186, 4285–4294 (2011). \n9. Li, H., Ye, C., Ji, G. & Han, J. Determinants of public T cell responses. Cell Res 22, \n33–42 (2012). \n10. Trentin, L., Zambello, R., Facco, M., Tassinari, C., Sancetta, R., Siviero, M., Cerutti, \nA., Cipriani, A., Marcer, G., Majori, M., Pesci, A., Agostini, C. & Semenzato, G. \nSelection of T lymphocytes bearing limited TCR-Vbeta regions in the lung of \nhypersensitivity pneumonitis and sarcoidosis. Am J Respir Crit Care Med 155, 587–\n596 (1997). \n11. Bowerman, N. A., Falta, M. T., Mack, D. G., Wehrmann, F ., Crawford, F ., Mroz, M. M., \nMaier, L. A., Kappler, J. W. & Fontenot, A. P . Identification of Multiple Public TCR \nRepertoires in Chronic Beryllium Disease. The Journal of Immunology 192, 4571–\n4580 (2014). \n12. Song, L., Cohen, D., Ouyang, Z., Cao, Y ., Hu, X. & Liu, X. S. TRUST4: immune \nrepertoire reconstruction from bulk and single-cell RNA-seq data. Nat Methods 18, \n627–630 (2021). \n13. Dash, P ., Fiore-Gartland, A. J., Hertz, T., Wang, G. C., Sharma, S., Souquette, A., \nCrawford, J. C., Clemens, E. B., Nguyen, T. H. O., Kedzierska, K., La Gruta, N. L., \nBradley, P . & Thomas, P . G. Quantifiable predictive features define epitope-specific T \ncell receptor repertoires. Nature 547, 89–93 (2017). \n14. Mayer-Blackwell, K., Schattgen, S., Cohen-Lavi, L., Crawford, J. C., Souquette, A., \nGaevert, J. A., Hertz, T., Thomas, P . G., Bradley, P . & Fiore-Gartland, A. TCR meta-\n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.11.13.623392doi: bioRxiv preprint \n\n9 \n \nclonotypes for biomarker discovery with tcrdist3 enabled identification of public, \nHLA-restricted clusters of SARS-CoV-2 TCRs. eLife 10, e68605 (2021). \n15. Henikoff, S. & Henikoff, J. G. Amino acid substitution matrices from protein blocks. \nProc. Natl. Acad. Sci. U.S.A. 89, 10915–10919 (1992). \n16. Mayer-Blackwell, K., Fiore-Gartland, A. & Thomas, P . G. Flexible Distance-Based \nTCR Analysis in Python with tcrdist3. in T-Cell Repertoire Characterization (eds. \nHuang, H. & Davis, M. M.) 309–366 (Springer US, New York, NY , 2022). \ndoi:10.1007/978-1-0716-2712-9_16. \n17. Britanova, O. V ., Shugay, M., Merzlyak, E. M., Staroverov, D. B., Putintseva, E. V ., \nTurchaninova, M. A., Mamedov, I. Z., Pogorelyy, M. V ., Bolotin, D. A., Izraelson, M., \nDavydov, A. N., Egorov, E. S., Kasatskaya, S. A., Rebrikov, D. V ., Lukyanov, S. & \nChudakov, D. M. Dynamics of Individual T Cell Repertoires: From Cord Blood to \nCentenarians. The Journal of Immunology 196, 5005–5013 (2016). \n18. Pogorelyy, M. V ., Minervina, A. A., Shugay, M., Chudakov, D. M., Lebedev, Y . B., Mora, \nT. & Walczak, A. M. Detecting T cell receptors involved in immune responses from \nsingle repertoire snapshots. PLOS Biology 17, e3000314 (2019). \n19. Benjamini, Y . & Hochberg, Y . Controlling the False Discovery Rate: A Practical and \nPowerful Approach to Multiple Testing. Journal of the Royal Statistical Society: \nSeries B (Methodological) 57, 289–300 (1995). \n20. Sethna, Z., Elhanati, Y ., Callan, C. G., Jr, Walczak, A. M. & Mora, T. OLGA: fast \ncomputation of generation probabilities of B- and T-cell receptor amino acid \nsequences and motifs. Bioinformatics 35, 2974–2981 (2019). \n21. Goncharov, M., Bagaev, D., Shcherbinin, D., Zvyagin, I., Bolotin, D., Thomas, P . G., \nMinervina, A. A., Pogorelyy, M. V ., Ladell, K., McLaren, J. E., Price, D. A., Nguyen, T. H. \nO., Rowntree, L. C., Clemens, E. B., Kedzierska, K., Dolton, G., Rius, C. R., Sewell, \nA., Samir, J., Luciani, F ., Zornikova, K. V ., Khmelevskaya, A. A., Sheetikov, S. A., \nEfimov, G. A., Chudakov, D. & Shugay, M. VDJdb in the pandemic era: a compendium \nof T cell receptors specific for SARS-CoV-2. Nat Methods 19, 1017–1019 (2022). \n22. Song, L., Bai, G., Liu, X. S., Li, B. & Li, H. Efficient and accurate KIR and HLA \ngenotyping with massively parallel sequencing data. Genome Res. 33, 923–931 \n(2023). \n23. Venturi, V ., Kedzierska, K., Price, D. A., Doherty, P . C., Douek, D. C., Turner, S. J. & \nDavenport, M. P . Sharing of T cell receptors in antigen-specific responses is driven \nby convergent recombination. Proceedings of the National Academy of Sciences \n103, 18691–18696 (2006). \n24. Elhanati, Y ., Sethna, Z., Callan Jr, C. G., Mora, T. & Walczak, A. M. Predicting the \nspectrum of TCR repertoire sharing with a data-driven model of recombination. \nImmunological Reviews 284, 167–179 (2018). \n25. Boomars, K. A., Wagenaar, S. S., Mulder, P . G., Velzen-Blad, H. van & Bosch, J. M. van \nden. Relationship between cells obtained by bronchoalveolar lavage and survival in \nidiopathic pulmonary fibrosis. Thorax 50, 1087–1092 (1995). \n26. Sendama, W., Funston, W., Rostron, A. J. & Simpson, A. J. Tissue-Resident Memory T \nCells Are Implicated in the Pathogenesis of Hypersensitivity Pneumonitis. Am J \nRespir Crit Care Med 207, 1246–1249 (2023). \n \n  \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.11.13.623392doi: bioRxiv preprint \n\n10 \n \nFigure 1. Mean number of clones and unique clonotypes in reconstructed TCRβ repertoires from \nBAL samples (A and B) and PBMC samples (C and D). Error bars indicate 2.5% and 97.5% \nquantiles. Bonferroni-Holm adjusted p values shown. \nA           B \n \n \n \n \nC             D \n  \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.11.13.623392doi: bioRxiv preprint \n\n11 \n \n \n  \nClone \nID \nPatient ID V gene CDR3β sequence J gene Clone \ncount \nExclusion \ncriterion? \n1619 \n2172 \nBALHN1 \nBALHF3 \nTRBV20-1*02 \nTRBV20-1*01 \nCSAPPRGRGAPVGQETQYF \nCSAPPRGRGAPVGQETQYF \nTRBJ2-5*01 \nTRBJ2-5*01 \n1 \n1 \nNo \n2108 \n2495 \nBALHN2 \nBALHF6 \nTRBV3-1*01 \nTRBV3-1*01 \nCASSQAPSGRIHEQYF \nCASSQAPSGRIHEQYF \nTRBJ2-7*01 \nTRBJ2-7*01 \n1 \n1 \nNo \n64 \n2089 \nBALHN4 \nBALHN2 \nTRBV20-1*05 \nTRBV20-1*04 \nCSASRSPSGNTIYF \nCSASRSPSGNTIYF \nTRBJ1-3*01 \nTRBJ1-3*01 \n1 \n1 \nCord blood \nneighbours \n268 \n2112 \nBALHN4 \nBALHN2 \nTRBV11-1*01 \nTRBV11-1*01 \nCASSPRRDTEAFF \nCASSPRRDTEAFF \nTRBJ1-1*01 \nTRBJ1-1*01 \n1 \n1 \nCord blood \nneighbours \n348 \n2126 \n \nBALHN4 \nBALHN2 \nTRBV7-6*01 \nTRBV7-6*01 \nCASSLAPLETQYF \nCASSLAPLETQYF \nTRBJ2-5*01 \nTRBJ2-5*01 \n1 \n1 \nCord blood \nneighbours \n396 \n2389 \nBALHN4 \nBALHF3 \nTRBV28*01 \nTRBV28*01 \nCASRGGGNTGELFF \nCASRGGGNTGELFF \nTRBJ2-2*01 \nTRBJ2-2*01 \n1 \n2 \nCord blood \nneighbours \n461 \n2142 \nBALHN4 \nBALHN2 \nTRBV6-6*01 \nTRBV6-6*02 \nCASSYSFWGENYGYTF \nCASSYSFWGENYGYTF \nTRBJ1-2*01 \nTRBJ1-2*01 \n1 \n1 \nCord blood \nneighbours \n502 \n2658 \nBALHN4 \nBALHN2 \nTRBV28*01 \nTRBV28*01 \nCASSLTGTGGRETQYF \nCASSLTGTGGRETQYF \nTRBJ2-5*01 \nTRBJ2-5*01 \n1 \n68 \nIdentified in IPF \nsample \n509 \n2582 \nBALHN4 \nBALHF4 \nTRBV28*01 \nTRBV28*01 \nCASSLGPHYEQYF \nCASSLGPHYEQYF \nTRBJ2-7*01 \nTRBJ2-7*01 \n1 \n1 \nVDJdb entry \n588 \n2615 \nBALHN4 \nBALHN2 \nTRBV12-5*01 \nTRBV12-5*01 \nCASGLSETQYF \nCASGLSETQYF \nTRBJ2-5*01 \nTRBJ2-5*01 \n1 \n5 \nCord blood \nneighbours \n972 \n2595 \nBALHN3 \nBALHN2 \n \nTRBV6-5*01 \nTRBV6-5*01 \nCASSYGGVGANVLTF \nCASSYGGVGANVLTF \nTRBJ2-6*01 \nTRBJ2-6*01 \n1 \n4 \nCord blood \nneighbours \n1066 \n2532 \nBALHN3 \nBALHN2 \nTRBV19*01 \nTRBV19*01 \nCASSTAGGVSTEAFF \nCASSTAGGVSTEAFF \nTRBJ1-1*01 \nTRBJ1-1*01 \n1 \n3 \nCord blood \nneighbours \n1598 \n2384 \nBALHN1 \nBALHF3 \nTRBV20-1*01 \nTRBV20-1*01 \nCSEEAGGEQYF \nCSEEAGGEQYF \nTRBJ2-7*01 \nTRBJ2-7*01 \n1 \n2 \nVDJdb entry \n1734 \n2577 \nBALHN1 \nBALHF4 \nTRBV6-5*01 \nTRBV6-5*01 \nCASSSRGDGYTF \nCASSSRGDGYTF \nTRBJ1-2*01 \nTRBJ1-2*01 \n1 \n1 \nCord blood \nneighbours \n2155 \n2522 \n \nBALHN2 \nBALHF6 \nTRBV19*01 \nTRBV19*01 \nCASSSPGGGLGNTEAFF \nCASSSPGGGLGNTEAFF \nTRBJ1-1*01 \nTRBJ1-1*01 \n1 \n1 \nIdentified in IPF \nsample \nTable 1. Public CDR3β sequences shared between pairs of patients with HP . Bold underlined text denotes \ndifferences in V gene segment selection despite identical CDR3β sequences. Entries in final column indicate \nreasons (if any) sequences could not be considered HP-related. \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.11.13.623392doi: bioRxiv preprint \n\n12 \n \n \n  \nPatient \nID \nHLA-A HLA-B HLA-C HLA-DRB1 HLA-DQB1 HLA-DPB1 \nBALHF3 *26:207 \n*23:01 \n*07:458 \n*07:02 \n*15:02 \n*15:06 \n*08:77 \n*08:01 \n*03:01 \n*03:72 \n*1317:01Q \nBALHF4 *23:01 \n*31:153 \n*39:150Q \n*14:01:10 \n*07:809 \n*07:906 \n*04:01 \n*12:111 \n*03:01 \n*03:02 \n*1317:01Q \nBALHF6 *11:01 \n*03:312 \n*38:115 Could not be \ninferred \n*03:01 \n*04:01 \n*03:01 \n*03:02 \n*02:01 \nBALHN1 *24:02 \n*24:95 \n*38:115 \n*35:471 \n*04:01 \n*07:628 \n*03:01 \n*08:77 \n*03:01 *02:01 \nBALHN2 *02:01 \n*03:350 \n*13:179 \n*44:02 \n*01:174 \n*02:190 \n*08:01 \n*15:198 \n*03:01 \n*03:72 \n*02:01 \nBALHN3 *24:02 *38:115 *04:01 \n*03:02 \n*04:01 \n*04:03 \n*03:02 *02:01 \nBALHN4 *68:01 \n*03:01 \n*14:02 \n*38:115 \n*05:01 \n*07:01 \n*04:01 \n*03:01 \n*03:01 \n*03:02 \n*02:01 \nTable 2. HLA alleles of patients with HP with TCRβ repertoire overlap. HLA genotype was inferred to \ntwo-field resolution using T1K. \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.11.13.623392doi: bioRxiv preprint \n\n13 \n \n \nPatient ID V gene CDR3β sequence J gene Clone \ncount \nVDJdb \nsimilarity? \nBALHF1 TRBV6-4*01 \nTRBV6-4*01 \nTRBV6-4*01 \nTRBV6-4*01 \nTRBV6-6*05 \nCASSLGGEEDTQYF* \nCASSLGGEEDTQYF* \nCASSLGGEPDTQYF \nCASSLGGEADTQYF \nCASSLGGEEDTQYF* \nTRBJ2-3*01 \nTRBJ2-3*01 \nTRBJ2-3*01 \nTRBJ2-3*01 \nTRBJ2-3*01 \n1 \n17 \n1 \n1 \n1 \nN \nBALHN1 TRBV11-1*01 \nTRBV7-2*01 \nTRBV7-2*01 \nTRBV7-2*01 \nCASSWAERKTQYF* \nCASSWAGRKTQYF \nCAGSWAERKTQYF* \nCAGSWAERKTQYF* \nTRBJ2-5*01 \nTRBJ2-5*01 \nTRBJ2-5*01 \nTRBJ2-5*01 \n1 \n2 \n2 \n44 \nN \nBALHN3 TRBV2*01 \nTRBV2*01 \nTRBV2*01 \nTRBV2*01 \nTRBV2*01 \nTRBV2*01 \nTRBV2*01 \nCASSEEAVKETKYF* \nCASSEEAVKETQYF* \nCASSEEAVKETQYF* \nCASSEEAVKEPQYF \nCASSEGAVKETQYF \nCASSEAAVKETQDF \nCASSEEAVKETQYF* \nTRBJ2-5*01 \nTRBJ2-1*01 \nTRBJ2-7*01 \nTRBJ2-5*01 \nTRBJ2-5*01 \nTRBJ2-5*01 \nTRBJ2-5*01 \n1 \n1 \n1 \n1 \n1 \n1 \n63 \nN \nBALHN4 TRBV20-1*01 \nTRBV20-1*01 \nTRBV20-1*01 \nTRBV20-1*01 \nTRBV20-1*01 \nTRBV20-1*05 \nCSARDEEEPKRTQYF \nCSARDEEEPGRTQYF \nCSARDEEDPERTQYF \nCSAGDEEEPERTQYF \nCSARDEEEPERTQYF* \nCSARDEEEPERTQYF* \nTRBJ2-3*01 \nTRBJ2-3*01 \nTRBJ2-3*01 \nTRBJ2-3*01 \nTRBJ2-3*01 \nTRBJ2-3*01 \n1 \n1 \n1 \n1 \n14 \n14 \nN \nTable 3. Clusters of highly similar T cell clones presumed to be under antigenic selection. Bold \nunderlined text indicates differing V gene selection and CDR3β amino acid substitution. Asterisks \ndenote identical CDR3β amino acid sequence with differing CDR3β nucleotide sequence or V \ngene. \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted November 15, 2024. ; https://doi.org/10.1101/2024.11.13.623392doi: bioRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}