Short Report

Longitudinal high-throughput TCR repertoire profiling reveals the dynamics of T-cell memory formation after mild COVID-19 infection

Shemyakin-Ovchinnikov Institute of Bioorganic Chemistry, Russian Federation
Pirogov Russian National Research Medical University, Russian Federation
National Research Center for Hematology, Russian Federation
Laboratoire de physique de l'École Normale Supérieure, ENS, PSL, Sorbonne Universite, Universite de Paris, and CNRS, France
Institute of Clinical Molecular Biology, Kiel University, Germany
Masaryk University, Central European Institute of Technology, Czech Republic
V.I. Kulakov National Medical Research Center for Obstetrics, Gynecology and Perinatology, Russian Federation
Moscow State University, Russian Federation

Jan 5, 2021

https://doi.org/10.7554/eLife.63502

Version of Record

Accepted for publication after peer review and revision.

Download
Cite
Share
CommentOpen annotations (there are currently 0 annotations on this page).

Version of Record published: January 13, 2021 (This version)
Accepted: January 5, 2021
Accepted Manuscript published: January 5, 2021 (Go to version)
Received: September 27, 2020

1. Part of Collection
COVID-19: A Collection of Articles

Edited by Diane M Harper et al.
Further reading

Article
Figures and data
Abstract
Introduction
Results
Discussion
Materials and methods
Data availability
References
Decision letter
Author response
Article and author information
Metrics

Abstract

COVID-19 is a global pandemic caused by the SARS-CoV-2 coronavirus. T cells play a key role in the adaptive antiviral immune response by killing infected cells and facilitating the selection of virus-specific antibodies. However, neither the dynamics and cross-reactivity of the SARS-CoV-2-specific T-cell response nor the diversity of resulting immune memory is well understood. In this study, we use longitudinal high-throughput T-cell receptor (TCR) sequencing to track changes in the T-cell repertoire following two mild cases of COVID-19. In both donors, we identified CD4⁺ and CD8⁺ T-cell clones with transient clonal expansion after infection. We describe characteristic motifs in TCR sequences of COVID-19-reactive clones and show preferential occurrence of these motifs in publicly available large dataset of repertoires from COVID-19 patients. We show that in both donors, the majority of infection-reactive clonotypes acquire memory phenotypes. Certain T-cell clones were detected in the memory fraction at the pre-infection time point, suggesting participation of pre-existing cross-reactive memory T cells in the immune response to SARS-CoV-2.

Introduction

COVID-19 is a global pandemic caused by the novel SARS-CoV-2 betacoronavirus (Vabret et al., 2020). T cells are crucial for clearing respiratory viral infections and providing long-term immune memory (Schmidt and Varga, 2018; Swain et al., 2012). Two major subsets of T cells participate in the immune response to viral infection in different ways: activated CD8⁺ T cells directly kill infected cells, while subpopulations of CD4⁺ T cells produce signalling molecules that regulate myeloid cell behaviour, drive, and support CD8 response and the formation of long-term CD8 memory, and participate in the selection and affinity maturation of antigen-specific B cells, ultimately leading to the generation of neutralizing antibodies. In SARS-1 survivors, antigen-specific memory T cells were detected up to 11 years after the initial infection, when viral-specific antibodies were undetectable (Ng et al., 2016; Oh et al., 2011). The T-cell response was shown to be critical for protection in SARS-1-infected mice (Zhao et al., 2010). Patients with X-linked agammaglobulinemia, a genetic disorder associated with lack of B cells, have been reported to recover from symptomatic COVID-19 (Quinti et al., 2020; Soresina et al., 2020), suggesting that in some cases, T cells are sufficient for viral clearance. Thevarajan et al. showed that activated CD8⁺HLA^–DR⁺CD38⁺ T cells in a mild case of COVID-19 significantly expand following symptom onset, reaching their peak frequency of 12% of CD8⁺ T cells on day 9 after symptom onset, and contract thereafter (Thevarajan et al., 2020). Given the average time of 5 days from infection to the onset of symptoms (Bi et al., 2020), the dynamics and magnitude of T-cell response to SARS-CoV-2 is similar to that observed after immunization with live vaccines (Miller et al., 2008). SARS-CoV-2-specific T cells were detected in COVID-19 survivors by activation following stimulation with SARS-CoV-2 proteins (Ni et al., 2020) or by viral protein-derived peptide pools (Weiskopf et al., 2020; Braun et al., 2020; Snyder et al., 2020; Le Bert et al., 2020; Meckiff et al., 2020; Bacher et al., 2020; Peng et al., 2020). Some of the T cells activated by peptide stimulation were shown to have a memory phenotype (Weiskopf et al., 2020; Le Bert et al., 2020; Mateus et al., 2020), and some potentially cross-reactive CD4⁺ T cells were found in healthy donors (Braun et al., 2020; Grifoni et al., 2020; Sekine et al., 2020; Bacher et al., 2020).

T cells recognise short pathogen-derived peptides presented on the cell surface of the major histocompatibility complex (MHC) using hypervariable T-cell receptors (TCR). TCR repertoire sequencing allows for the quantitative tracking of T-cell clones in time, as they go through the expansion and contraction phases of the response. It was previously shown that quantitative longitudinal TCR sequencing is able to identify antigen-specific expanding and contracting T cells in response to yellow fever vaccination with high sensitivity and specificity (Minervina et al., 2020; Pogorelyy et al., 2018; DeWitt et al., 2015). Not only clonal expansion but also significant contraction from the peak of the response are distinctive traits of T-cell clones specific to the virus (Pogorelyy et al., 2018).

In this study, we use longitudinal TCRalpha and TCRbeta repertoire sequencing to quantitatively track T-cell clones that significantly expand and contract after recovery from a mild COVID-19 infection, and determine their phenotype. We reveal the dynamics and the phenotype of the memory cells formed after infection, identify pre-existing T-cell memory clones participating in the response, and describe public TCR sequence motifs of SARS-CoV-2-reactive clones, suggesting a response to immunodominant epitopes.

Results

Longitudinal tracking of TCR repertoires of COVID-19 patients

In the middle of March (day 0), donor W female and donor M male (both healthy young adults) returned to their home country from the one of the centres of the COVID-19 outbreak in Europe at the time. Upon arrival, according to local regulations, they were put into strict self-quarantine for 14 days. On day 3 of self-isolation, both developed low-grade fever, fatigue, and myalgia, which lasted 4 days and was followed by a temporary loss of smell for donor M. On days 15, 30, 37, 45, and 85, we collected peripheral blood samples from both donors (Figure 1a). The presence of IgG and IgM SARS-CoV-2-specific antibodies in the plasma was measured at all time points using SARS-CoV-2 S-RBD domain-specific ELISA (Figure 1—figure supplement 1). From each blood sample, we isolated peripheral blood mononuclear cells (PBMCs, in two biological replicates), CD4⁺, and CD8⁺ T cells. Additionally, on days 30, 45, and 85, we isolated four T-cell memory subpopulations (Figure 1—figure supplement 2a): effector memory (EM: CCR7⁻CD45RA⁻), effector memory with CD45RA re-expression (EMRA: CCR7⁻CD45RA⁺), central memory (CM: CCR7⁺CD45RA⁻), and stem cell-like memory (SCM: CCR7⁺CD45RA⁺CD95⁺). From all samples, we isolated RNA and performed TCRalpha and TCRbeta repertoire sequencing as previously described (Pogorelyy et al., 2017). For both donors, TCRalpha and TCRbeta repertoires were obtained for other projects 1 and 2 years prior to infection. Additionally, TCR repertoires of multiple samples for donor M – including sorted memory subpopulations – are available from a published longitudinal TCR sequencing study after yellow fever vaccination (donor M1 samples in Minervina et al., 2020).

Figure 1 with 5 supplements see all

Download asset Open asset

Longitudinal tracking of T-cell clones after mild COVID-19.

(a) Study design. Peripheral blood of two donors was sampled longitudinally on days 15, 30, 37, 45, and 85 after arrival in Russia. At each time point, we evaluated SARS-CoV-2-specific antibodies using ELISA (Figure 1—figure supplement 1) and isolated PBMCs in two biological replicates. Additionally, CD4⁺ and CD8⁺ cells were isolated from a separate portion of blood, and EM, CM, EMRA, and SCM memory subpopulations were FACS sorted on days 30, 45, and 85. For each sample, we sequenced TCRalpha and TCRbeta repertoires. For both donors, pre-infection PBMC repertoires were sampled in 2018 and 2019 for other projects. (**b, c**) PCA of clonal temporal trajectories identifies three groups of clones with distinctive behaviours. Left: First two principal components of the 1000 most abundant TCRbeta clonotype frequencies normalized by maximum value for each clonotype in PBMC at post-infection time points. Colour indicates hierarchical clustering results of principal components; symbol indicates if clonotype was called as significantly contracted from day 15 to day 85 (triangles) or expanded from day 15 to day 37 (circles) by both edgeR and NoisET (Figure 1—figure supplement 5 shows overlap between clonal trajectory clusters and edgeR/NoisET hits). Right: Each curve shows the average ±2.96 SE of normalized clonal frequencies from each cluster. Contracting (d) and expanding (e) clones include both CD4⁺ and CD8⁺ T cells and are less abundant in pre-infection repertoires. T-cell clones significantly contracted from day 15 to day 85 (d) and significantly expanded from day 15 to day 37 (e) were identified in both donors. The fraction of contracting (d) and expanding (e) TCRbeta clonotypes in the total repertoire (calculated as the sum of frequencies of these clonotypes in the second PBMC replicate at a given time point and corresponding to the fraction of responding cells of all T cells) is plotted in log-scale for all reactive clones (left) and reactive clones with the CD4 (middle) and CD8 (right) phenotypes. Similar dynamics were observed in TCRalpha repertoires (Figure 1—figure supplement 3) and for significantly expanded/contracted clones identified with the NoisET Bayesian differential expansion statistical model alone (Figure 1—figure supplement 4).

Two waves of T-cell clone response

From previously described activated T-cell dynamics for SARS-CoV-2 (Thevarajan et al., 2020), and immunization with live vaccines (Miller et al., 2008), the peak of the T-cell expansion is expected around day 15 post-infection, and responding T cells significantly contract afterwards. However, Weiskopf et al., 2020 reports an increase of SARS-CoV-2-reactive T cells at later time points, peaking in some donors after 30 days following symptom onset. To identify groups of T-cell clones with similar dynamics in an unbiased way, we used principal component analysis (PCA) in the space of T-cell clonal trajectories (Figure 1b,c). This exploratory data analysis method allows us to visualize major trends in the dynamics of abundant TCR clonotypes (occurring within top 1000 on any post-infection time points) between multiple time points.

In both donors, and in both TCRalpha and TCRbeta repertoires, we identified three clusters of clones with distinct dynamics. The first cluster (Figure 1b,c, purple) corresponded to abundant TCR clonotypes that had constant concentrations across time points, the second cluster (Figure 1b,c, green) showed contraction dynamics from day 15 to day 85, and the third cluster (Figure 1b,c, yellow) showed an unexpected clonal expansion from day 15 with a peak on day 37 followed by contraction. The clustering and dynamics are similar in both donors and are reproduced in TCRbeta (Figure 1b,c) and TCRalpha (Figure 1—figure supplement 3a,b) repertoires. We next used edgeR, a software for differential gene expression analysis (Robinson et al., 2010) and NoisET, a Bayesian differential expansion model (Puelma Touzel et al., 2020), to specifically detect changes in clonotype concentration between pairs of time points in a statistically reliable way and without limiting the analysis to the most abundant clonotypes. Both NoisET and edgeR use biological replicate samples collected at each time point to train a noise model for sequence counts. Results for the two models were similar (Figure 1—figure supplement 4), and we conservatively defined as expanded or contracted the clonotypes that were called by both models simultaneously. We identified 291 TCRalpha and 295 TCRbeta clonotypes in donor W and 607 TCRalpha and 616 TCRbeta in donor M significantly contracted from day 15 to day 85 (largely overlapping with cluster 2 of clonal trajectories, Figure 1—figure supplement 5). One hundred and seventy six TCRalpha and 278 TCRbeta for donor W and 293 TCRalpha and 427 TCRbeta clonotypes for donor M were significantly expanded from days 15 to 37 (corresponding to cluster 3 of clonal trajectories).

Note that, to identify putatively SARS-CoV-2-reactive clones, we only used post-infection time points, so that our analysis can be reproduced in other patients and studies where pre-infection time points are unavailable. However, tracking the identified responding clones back to pre-infection time points reveals strong clonal expansions from pre- to post-infection (Figure 1d,e, Figure 1—figure supplement 3c,d). For brevity, we further refer to clonotypes significantly contracted from day 15 to 85 as contracting clones and clonotypes significantly expanding from day 15 to 37 as expanding clones. Contracting clones corresponded to 2.5% and 0.9% of T cells on day 15 post-infection, expanding clones reached 1.1% and 0.8% on day 37 for donors M and W, respectively (Figure 1d,e, left). This magnitude of the T-cell response is of the same order of magnitude as previously observed after live yellow fever vaccine immunization of donor M (6.7% T cells on day 15 post-vaccination). For each contracting and expanding clone, we determined their CD4/CD8 phenotype using separately sequenced repertoires of CD4⁺ and CD8⁺ subpopulations (see Materials and methods). Both CD4⁺ and CD8⁺ subsets participated actively in the response (Figure 1d,e). Interestingly, clonotypes expanding after day 15 were significantly biased towards the CD4⁺ phenotype, while contracting clones had balanced CD4/CD8 phenotype fractions in both donors (Fisher’s exact test, $p < 0.01$ for both donors). A recent study by Parrot et al., 2020 reported activation of mucosal-associated invariant T (MAIT) cells in COVID-19 patients. MAIT cells are innate-like T cells recognising bacterial metabolites with TCRs featuring an invariant TCRalpha chain. We used a set of 121 unique MAIT TCRalpha amino acid sequences from Loh et al., 2020 to check whether MAIT cells contributed significantly to contracting or expanding clonotype groups identified in our study. We found that a small number of low-frequency contracting (one clonotype for each donor) and expanding (two clonotypes for each donor) TCRalpha clonotypes have MAIT invariant alpha chains, and the contribution of these clonotypes to the response was relatively small (<1% of responding T cells are MAIT).

Memory formation and pre-existing memory

On days 30, 45, and 85, we identified both contracting (Figure 2a–c) and expanding (Figure 2—figure supplement 1a–c) T-cell clones in the memory subpopulations of peripheral blood. Both CD4⁺ and CD8⁺ responding clones were found in the CM and EM subsets; however, CD4⁺ were more biased towards CM (with exception of donor W day 30 time point, where a considerable fraction of CD8⁺ clones were found in CM), and CD8⁺ clones more represented in the EMRA subset. A small number of both CD4⁺ and CD8⁺ responding clonotypes were also identified in the SCM subpopulation, which was previously shown to be a long-lived T-cell memory subset (Fuertes Marraco et al., 2015). Note that we sequenced more cells from PBMC than from the memory subpopulations (Supplementary file 1), so that some low-abundant responding T-cell clones are not sampled in the memory subpopulations. Intriguingly, a number of responding CD4⁺ clones, and fewer CD8⁺ clones, were also represented in the repertoires of both donors 1 and 2 years before the infection. The majority of the pre-existing clones had low concentrations at pre-infection and day 85 post-infection time points, and high concentrations on day 15, suggesting that they expanded after the infection and later contracted for both donors (Figure 2—figure supplement 2). For donor M, for whom we had previously sequenced memory subpopulations before the infection (Minervina et al., 2020), we were able to identify pre-existing SARS-CoV-2-reactive CD4⁺ clones in the CM subpopulation 1 year before the infection and a group of CD8⁺ clones in the pre-infection EM subpopulation. Interestingly, on day 30 after infection, the majority of pre-infection CM clones were detected in the EM subpopulation, suggesting recent T-cell activation and a switch of the phenotype from memory to effector. These clones might represent memory T cells cross-reactive for other infections, for example other human coronaviruses.

Figure 2 with 2 supplements see all

Download asset Open asset

Memory phenotypes of responding clonotypes contracting after day 15.

(a) A large fraction of contracting clonotypes is identified in T-cell memory subsets after infection. Bars show the fraction of contracting CD4⁺ and CD8⁺ TCRbeta clonotypes present in 2 year; 1 year pre-infection PBMC; in at least one of memory subpopulation sampled on day 30, day 37, and day 85 post-infection. (**b, c**) Responding clones are found in different memory subsets. Fraction of CD4⁺ (left panels) and CD8⁺ (right panels) contracting clones of donor W (b) and M (c) that were identified in each memory subpopulation repertoire at each time point. For both donors, CD4⁺ clonotypes were found predominantly in central memory (CM) and effector memory (EM), while CD8⁺ T cells were enriched in EMRA compartment. (c) For donor M, CD4⁺ contracting clonotypes are also identified in memory subsets 1 year before the infection, with a bias towards the CM subpopulation and a group of CD8⁺ clones is found in the pre-infection EM subpopulation.

A search for TCRbeta amino acid sequences of responding clones in VDJdb (Bagaev et al., 2020) — a database of TCRs with known specificities — resulted in essentially no overlap with TCRs not specific for SARS-CoV-2 epitopes: only two clonotypes matched. One match corresponded to the cytomegalovirus epitope presented by the HLA-A*03 MHC allele, which is absent in both donors (Supplementary file 2), and a second match was for Influenza A virus epitope presented by HLA-A*02 allele. The absence of matches suggests that contracting and expanding clones are unlikely to be specific for immunodominant epitopes of common pathogens covered in VDJdb. We next asked if we could map specificites of our responding clones to SARS-CoV-2 epitopes.

Validation by MHC tetramer-staining assay

On day 25, post-infection donor M participated in a study by Shomuradova et al., 2020 (as donor p1434), where his CD8⁺ T cells were stained with a HLA-A*02:01-YLQPRTFLL MHC-I tetramer (Figure 1—figure supplement 2b). TCRalpha and TCRbeta of FACS-sorted tetramer-positive cells were sequenced and deposited to VDJdb (see Shomuradova et al., 2020 for the experimental details). We matched these tetramer-specific TCR sequences to our longitudinal dataset (Figure 3a for TCRbeta and Figure 3—figure supplement 1 for TCRalpha). We found that their frequencies were very low on pre-infection time points and monotonically decreased from their peak on day 15 ( $7.1 \times 10^{- 4}$ fraction of bulk TCRbeta repertoire) to day 85 ( $1.3 \times 10^{- 5}$ fraction), in close analogy to our contracting clone set. Among the tetramer-positive clones that were abundant on day 15 (with bulk frequency $> 10^{- 5}$ ), 17 of 18 TCRbetas and 12 of 15 TCRalphas were independently identified as contracting by our method (as expected for TCRs specifically binding MHC-I multimer, all matching clonotypes in our datasets had a CD8⁺ phenotype). No tetramer-positive clonotypes matched with expanding clonotypes peaking on day 37.

Figure 3 with 2 supplements see all

Download asset Open asset

Distinctive TCR sequence motifs and epitope specificity of contracting CD8⁺ T cell clones.

(a) SARS-CoV-2-specific TCRs are independently identified by clonal contraction. On day 25, donor M participated in a study by Shomuradova et al., 2020, where TCR sequences responding to the HLA-A*02:01-YLQPRTFLL-tetramer T cells were determined. Here we matched the resulting tetramer-specific TCRbeta clonotypes to the longitudinal repertoire dataset obtained in the current study. Each dot corresponds to the frequency of HLA-A*02:01-YLQPRTFLL-tetramer-specific TCRbeta clonotype in bulk repertoire of donor M (donor p1434 from Shomuradova et al., 2020) at each time point. Green dots correspond to clonotypes independently identified as contracting in our longitudinal dataset. Blue line shows the cumulative frequency of tetramer-specific TCRbeta clonotypes. (**b, c**) Analysis of TCR amino acid sequences of contracting CD8⁺ clones reveals distinctive motifs. For each set of CD8alpha and CD8beta contracted clonotypes, we constructed a separate similarity network. Each vertex in the similarity network corresponds to a contracting clonotype. An edge indicates two or less amino acid mismatches in the CDR3 region and identical V and J segments. Networks are plotted separately for CD8alpha (b) and CD8beta (c) contracting clonotypes. Clonotypes without neighbours are not shown. Sequence logos corresponding to the largest clusters are shown under the corresponding network plots. ‘T’ on vertices indicate TCRbeta clonotypes confirmed by HLA-A*02:01-YLQPRTFLL tetramer staining. Shaded coloured circles (c) indicate clonotypes with large number of matches to CD8⁺ TCRs recognising SARS-CoV-2 peptides pools from Snyder et al., 2020 (MIRA peptide dataset). Lists of peptides in YLQ and ORF7b peptide pools are shown on the right. (d) Sequence convergence among contracting clonotypes. The number of edges in each group is shown by pink dots and is compared to the distribution of that number in 1000 random samples of the same size from the relevant repertoires at day 15 (blue boxplots). (e) Fraction of TCRbeta clonotypes with matches in the MIRA dataset (coloured rectangles) out of all responding CD8⁺ TCRbeta clonotypes in donor M on day 15.

TCR sequence motifs of responding clones

It was previously shown that TCRs recognising the same antigens frequently have highly similar TCR sequences (Dash et al., 2017; Glanville et al., 2017). To identify motifs in TCR amino acid sequences, we plotted similarity networks for significantly contracted (Figures 3b,c and 4a,b) and expanded (Figure 4—figure supplement 1b–e) clonotypes. The number of edges in all similarity networks except CD8⁺ expanding clones was significantly larger than would expected by randomly sampling the same number of clonotypes from the corresponding repertoire (Figure 3d and Figure 4—figure supplement 1a). In both donors, we found clusters of highly similar clones in both CD4⁺ and CD8⁺ subsets for expanding and contracting clonotypes. Clusters were largely donor specific, as expected, since our donors have dissimilar HLA alleles (Supplementary file 2) and thus each is likely to present a non-overlapping set of T-cell antigens. The largest cluster, described by the motif TRAV35-CAGXNYGGSQGNLIF-TRAJ42, was identified in donor M’s CD4⁺-contracting alpha chains. Clones from this cluster constituted 16.3% of all of donor M’s CD4⁺-responding cells on day 15, suggesting a response to an immunodominant CD4⁺ epitope in the SARS-CoV-2 proteome. The high similarity of the TCR sequences of responding clones in this cluster allowed us to independently identify motifs from donor M’s CD4 alpha contracting clones using the ALICE algorithm (Pogorelyy et al., 2019; Figure 3—figure supplement 2). While the time-dependent methods (Figure 1) identify abundant clones, the ALICE approach is complementary to both edgeR and NoisET as it identifies clusters of T cells with similar sequences independent of their individual abundances.

Figure 4 with 2 supplements see all

Download asset Open asset

Analysis of TCR amino acid sequences of CD4⁺contracting clones reveals distinctive motifs.

(a) Each vertex in the similarity network corresponds to a contracting clonotype. An edge indicates two or less amino acid mismatches in the CDR3 region (and identical V and J segments). Networks are plotted separately for CD4alpha (a) and CD4beta (b) contracting clonotypes. Clonotypes without neighbours are not shown. Sequence logos corresponding to the largest clusters are shown under the corresponding network plots. (**c, d**) Clonotypes forming the two largest motifs are significantly more clonally expanded (p<0.001, one-sided t-test) in a cohort of COVID-19 patients (Snyder et al., 2020) than in a cohort of control donors (Emerson et al., 2017). Each dot corresponds to the total frequency of clonotypes from motifs shaded on (b) in the TCRbeta repertoire of a given donor. Coloured dots show donors predicted to share HLA-DRB1*07:01 allele with donor W (c) or HLA-DRB1*03:01-DQB1*02:01 haplotype with donor M (d).

Mapping TCR motifs to SARS-CoV-2 epitopes

In CD8⁺ T cells, three clusters of highly similar TCRbeta clonotypes in donor M and one cluster of TCRalpha clonotypes correspond to YLQPRTFLL-tetramer-specific TCR sequences described above. To map additional specificities for CD8⁺ TCRbetas, we used a large set of SARS-CoV-2-peptide-specific TCRbeta sequences from Snyder et al., 2020 obtained using Multiplex Identification of Antigen-specific T-cell Receptors Assay (MIRA) with combinatorial peptide pools (Klinger et al., 2015). For each responding CD8⁺ TCRbeta, we searched for the identical or highly similar (same VJ combination, up to one mismatch in CDR3aa) TCRbeta sequences specific for given SARS-CoV-2 peptides. A TCRbeta sequence from our set was considered mapped to a given peptide if it had at least two highly similar TCRbeta sequences specific for this peptide in the MIRA experiment. This procedure yielded unambiguous matches for 32 CD8⁺ TCRbetas – just one clonotype was paired to two peptide pools (Supplementary file 3). The vast majority of matches to MIRA corresponded to groups of contracting clones. As expected, we found that all clusters corresponding to HLA-A*02:01-YLQPRTFLL MHC-I tetramer-specific TCRs were matched to the peptide pool YLQPRTFL,YLQPRTFLL,YYVGYLQPRTF in the MIRA dataset. Another large group of matches corresponded to the HLA-B*15:01-restricted (Stamatakis et al., 2020) NQKLIANQF epitope. Interestingly, clonotypes corresponging to this cluster together made up 21% of the CD8⁺ immune response on day 15, suggesting immunodominance of this epitope. Two TCRbeta clonotypes mapped to this epitope were identified in Effector Memory subset 1 year before the infection, suggesting potential cross-reactive response. We speculate that this response might be initially triggered by NQKLIANAF, a homologous HLA-B*15:01 epitope from HKU1 or OC43, common human betacoronaviruses. Importantly, no matches for the NQKLIANQF epitope were observed among contracting or expanding clonotypes of HLA-B*15-negative donor W (Supplementary file 2). To predict potential pairings between TCRalpha and TCRbeta motifs, we used a method of alpha/beta clonal trajectory matching described in Minervina et al., 2020 (see Materials and methods for details). We found consistent pairing between one of the motifs in TCRalpha to the largest motif in TCRbeta T cells, which is associated with HLA-B*15:01-NQKLIANQF.

Validation of CD4⁺ COVID-19 HLA-restricted specificity by cohort association analysis

At the time of writing, no data on TCR sequences specific to MHC-II class epitopes exist to map specificities of CD4⁺ T cells in a similar way as we did with MIRA-specific TCRs. However, a recently published database of 1414 bulk TCRbeta repertoires from COVID-19 patients allowed us to confirm the SARS-CoV-2 specificity of contracting clones indirectly. Public TCRbeta sequences that can recognise SARS-CoV-2 epitopes are expected to be clonally expanded and thus sampled more frequently in the repertoires of COVID-19 patients than in control donors. In Figure 4c,d, we show that the total frequency of TCRbeta sequences forming the largest cluster in donor M (Figure 4c) and donor W (Figure 4d) is significantly larger in the COVID-19 cohort than in the healthy donor cohort from Emerson et al., 2017, suggesting antigen-dependent clonal expansion. We hypothesized that the difference between control and COVID-19 donors in motif abundance should be even larger if we restrict the analysis to donors sharing the HLA allele that presents the epitope. Unfortunately, HLA-typing information is not yet available for the COVID-19 cohort. However, using sets of HLA-associated TCRbeta sequences from DeWitt et al., 2018, we could build a simple classifier to predict the HLA alleles of donors from both the control and COVID-19 cohorts exploiting the presence of TCRbeta sequences associated with certain HLA alleles (see Materials and methods for details). We found that the CD4⁺ TCRbeta motif from donor W occurs preferentially in donors predicted to have DRB1*07:01 allele, while the motif from donor M appears to be associated with HLA-DRB1*03:01-DQB1*02:01 haplotype. The frequency of sequences corresponding to these motifs can then be used to identify SARS-CoV-2 infected donors with matching HLA alleles (Figure 4—figure supplement 2).

Discussion

Using longitudinal repertoire sequencing, we identified a group of CD4⁺ and CD8⁺ T cell clones that contract after recovery from a SARS-CoV-2 infection. Our response timelines agree with T-cell dynamics reported by Thevarajan et al., 2020 for mild COVID-19, as well as with dynamics of T-cell response to live vaccines (Miller et al., 2008). We further mapped the specificities of contracting CD8⁺ T cells using sequences of SARS-CoV-2-specific T cells identified with tetramer staining in the same donor, as well as the large set of SARS-CoV-2 peptide stimulated TCRbeta sequences from Snyder et al., 2020. For large CD4⁺ TCRbeta motifs, we show strong association with COVID-19 by analysing the occurrence patterns and frequencies of these sequences in a large cohort of COVID-19 patients. Surprisingly, in both donors, we also identified a group of predominantly CD4⁺ clonotypes that expanded from day 15 to day 37 after the infection. One possible explanation for this second wave of expansion is the priming of CD4⁺ T cells by antigen-specific B cells, but there might be other mechanisms such as the migration of SARS-CoV-2-specific T cells from lymphoid organs or bystander activation of non-SARS-CoV-2 specific T cells. It is also possible that later expanding T cells are triggered by another infection, simultaneously and asymptomatically occurring in both donors around day 30. In contrast to the first wave of response identified by contracting clones, for now we do not have confirmation that this second wave of expansion corresponds to SARS-CoV-2-specific T cells. The accumulation of TCR sequences for CD4⁺ SARS-CoV-2 epitope-specific T cells may further address this question. We showed that a large fraction of putatively SARS-CoV-2-reactive T cell clones are later found in memory subpopulations and remain there at least 3 months after infection. Importantly, some of responding clones are found in long-lived stem cell-like memory (SCM) subset, as also reported for SARS-CoV-2 convalescent patients in Sekine et al., 2020. A subset of CD4⁺ clones were identified in pre-infection central memory subsets, and a subset of CD8⁺ T cells were found in effector memory. Among these are CD8⁺ clones recognising NQKLIANQF, an immunodominant HLA-B*15:01 restricted SARS-CoV-2 epitope, for which homologous epitope differing by 1 aa mismatch exists in common human betacoronaviruses. The presence of SARS-CoV-2 cross-reactive CD4⁺ T cells in healthy individuals was recently demonstrated (Braun et al., 2020; Grifoni et al., 2020; Le Bert et al., 2020; Meckiff et al., 2020; Bacher et al., 2020; Peng et al., 2020). Our data further suggests that cross-reactive CD4⁺ and CD8⁺ T cells can participate in the response in vivo. It is interesting to ask whether the presence of cross-reactive T cells before infection is linked to the mildness of the disease (with predicted HLA-B*15:01 cross-reactive epitope described above as a good starting point). Larger studies with cohorts of severe and mild cases with pre-infection time points are needed to address this question.

Materials and methods

Key resources table

Reagent type (species) or resource	Designation	Source or reference	Identifiers	Additional information
Antibody	Anti-CD3-FITC (mouse monoclonal)	eBioscience	CAT# 11-0038-42	FACS (5 µl per test)
Antibody	Anti-CD45RA-eFluor450 (mouse monoclonal)	eBioscience	CAT# 48-0458-42	FACS (5 µl per test)
Antibody	Anti-CCR7-APC (rat monoclonal)	eBioscience	CAT# 17-1979-42	FACS (5 µl per test)
Antibody	Anti-CD95-PE (mouse monoclonal)	eBioscience	CAT# 12-0959-42	FACS (5 µl per test)
Commercial assay or kit	Dynabeads CD4 Positive Isolation Kit	Invitrogen	CAT# 11331D
Commercial assay or kit	Dynabeads CD8 Positive Isolation Kit	Invitrogen	CAT# 11333D

Donors and blood samples

Request a detailed protocol

Peripheral blood samples from two young healthy adult volunteers, donor W (female) and donor M (male), were collected with written informed consent in a certified diagnostics laboratory. Both donors gave written informed consent to participate in the study under the declaration of Helsinki. HLA alleles of both donors (Supplementary file 2) were determined by an in-house cDNA high-throughput sequencing method.

SARS-CoV-2 S-RBD domain-specific ELISA

Request a detailed protocol

An ELISA assay kit developed by the National Research Centre for Hematology was used for detection of anti-S-RBD IgG according to the manufacturer’s protocol. The relative IgG level (OD/CO) was calculated by dividing the OD (optical density) values by the mean OD value of the cut-off positive control serum supplied with the Kit (CO). OD values of d37, d45, and d85 samples for donor M exceeded the limit of linearity for the Kit. In order to properly compare the relative IgG levels between d30, d37, d45, and d85, these samples were diluted 1:400 instead of 1:100; the ratios d37:d30 and d45:d30 and d85:d30 were calculated and used to calculate the relative IgG level of d37, d45, and d85 by multiplying d30 OD/CO value by the corresponding ratio. Relative anti-S-RBD IgM level was calculated using the same protocol with anti-human IgM-HRP-conjugated secondary antibody. Since the control cut-off serum for IgM was not available from the Kit, in Figure 1—figure supplement 1b, we show OD values for nine biobanked pre-pandemic serum samples from healthy donors.

Isolation of PBMCs and T-cell subpopulations

Request a detailed protocol

PBMCs were isolated with the Ficoll-Paque density gradient centrifugation protocol. CD4⁺ and CD8⁺ T cells were isolated from PBMCs with Dynabeads CD4⁺- and CD8⁺-positive selection kits (Invitrogen), respectively. For isolation of EM, EMRA, CM, and SCM memory subpopulations, we stained PBMCs with the following antibody mix: anti-CD3-FITC (UCHT1, eBioscience), anti-CD45RA-eFluor450 (HI100, eBioscience), anti-CCR7-APC (3D12, eBioscience), and anti-CD95-PE (DX2, eBioscience). Cell sorting was performed on FACS Aria III, and all four isolated subpopulations were lysed with Trizol reagent immediately after sorting.

TCR library preparation and sequencing

Request a detailed protocol

TCRalpha and TCRbeta cDNA libraries preparation was performed as previously described in Pogorelyy et al., 2017. RNA was isolated from each sample using Trizol reagent according to the manufacturer’s instructions. A universal primer binding site, sample barcode, and unique molecular identifier (UMI) sequences were introduced using the 5′RACE technology with TCRalpha and TCRbeta constant segment-specific primers for cDNA synthesis. cDNA libraries were amplified in two PCR steps, with the introduction of the second sample barcode and Illumina TruSeq adapter sequences at the second PCR step. Libraries were sequenced using the Illumina NovaSeq platform (2 × 150 bp read length).

TCR repertoire data analysis

Raw data preprocessing

Request a detailed protocol

Raw sequencing data was demultiplexed, and UMI-guided consensuses were built using migec v.1.2.7 (Shugay et al., 2014). Resulting UMI consensuses were aligned to V and J genomic templates of the TRA and TRB locus and assembled into clonotypes with mixcr v.2.1.11 (Bolotin et al., 2015). See Supplementary file 1 for the number of cells, UMIs, and unique clonotypes for each sample.

Identification of clonotypes with active dynamics

Request a detailed protocol

Principal component analysis (PCA) of clonal trajectories was performed as described before Minervina et al., 2020. First, we selected clones that were present among the top 1000 abundant in any of post-infection PBMC repertoires, including biological replicates, i.e. considered clone abundant if it was found within top 1000 most abundant clonotypes in at least one of the replicate samples at one time point. Next, for each such abundant clone, we calculated its frequency at each post-infection time point and divided this frequency by the maximum frequency of this clone for normalization. Then we performed PCA on the resulting normalized clonal trajectory matrix and identified three clusters of trajectories with hierarchical clustering with average linkage, using Euclidean distances between trajectories. We identify statistically significant contractions and expansions with edgeR as previously described (Pogorelyy et al., 2018), using false discovery rate-adjusted $p < 0.01$ and log₂ fold-change threshold of 1. NoisET implements the Bayesian detection method described in Puelma Touzel et al., 2020. Briefly, a two-step noise model accounting for cell sampling and expression noise is inferred from replicates, and a second model of expansion is learned from the two time points to be compared. The procedure outputs the posterior probability of expansion or contraction and the median estimated log₂ fold change, whose thresholds are set to 0.05 and 1, respectively.

Mapping of COVID-19-associated TCRs to the MIRA database

Request a detailed protocol

TCRbeta sequences from T cells specific for SARS-CoV-2 peptide pools MIRA (ImmuneCODE release 2) were downloaded from https://clients.adaptivebiotech.com/pub/covid-2020. V and J genomic templates were aligned to TCR nucleotide sequences from the MIRA database using mixcr 2.1.11. We consider a TCRbeta from MIRA matched to a TCRbeta from our data, if it had the same V and J and at most one mismatch in CDR3 amino acid sequence. We consider a TCRbeta sequence mapped to an epitope if it has at least two identical or highly similar (same V, J and up to one mismatch in CDR3 amino acid sequence) TCRbeta clonotypes reactive for this epitope in the MIRA database.

Computational alpha/beta pairing by clonal trajectories

Request a detailed protocol

Computational alpha/beta pairing was performed as described in Minervina et al., 2020. For each TCRbeta, we determine the TCRalpha with the closest clonal trajectory (Supplementary files 3 and 5). We observe no stringent pairings between TCRbeta and TCRbeta motifs with exception of two contracting CD8 TCRbeta clusters: TRBV7-2/TRBJ1-2 NQKLIANQF-associated clones from donor M paired to TRAV21/TRAJ40 alphas from the same cluster (CASSLEDTNYGYTF-CAVHSSGTYKYIF and CASSLEDTIYGYTF-CAALTSGTYKYIF), and TRBV7-9/TRBJ2-3 beta cluster paired to largest alpha cluster (CASSPTGRGRTDTQYF-CAYRSGGSEKLVF and CASSPTGRGGTDTQYF-CAYRRPGGEKLTF).

Computational prediction of HLA types

Request a detailed protocol

To predict HLA types from TCR repertoires of COVID-19 cohort, we used sets of HLA-associated TCR sequences from DeWitt et al., 2018. We use TCRbeta repertoires of 666 donors from cohort from Emerson et al., 2017, for which HLA-typing information is available in DeWitt et al., 2018 as a training set to fit logistic regression model, where the presence or absense of given HLA allele is an outcome, and the number of allele-associated sequences in repertoire, as well as the total number of unique sequences in the repertoire, are the predictors. A separate logistic regression model was fitted for each set of HLA-associated sequences from DeWitt et al., 2018 and then used to predict the probability $p$ that a donor from the COVID-19 cohort has this allele. Donors with $p < 0.2$ were considered negative for a given allele.

Data availability

Request a detailed protocol

Raw sequencing data are deposited to the Short Read Archive (SRA) accession: PRJNA633317. Processed TCRalpha and TCRbeta repertoire datasets, resulting repertoires of SARS-CoV-2-reactive clones, and raw data preprocessing instructions can be accessed from: https://github.com/pogorely/Minervina_COVID; copy archived at https://archive.softwareheritage.org/swh:1:rev:d5b3953d3739f21a1ccc23114bfffa002e11951e/.

Data availability

Raw sequencing data are deposited to the Short Read Archive (SRA) accession: PRJNA633317. Resulting repertoires of SARS-CoV-2-reactive clones can be found in SI Tables 3-6 and also accessed from: https://github.com/pogorely/Minervina_COVID (Copy archived at https://archive.softwareheritage.org/swh:1:rev:d5b3953d3739f21a1ccc23114bfffa002e11951e/). Processed TCRalpha and TCRbeta repertoire datasets are available at : https://zenodo.org/record/3835955.

The following data sets were generated

1. Minervina AA
2. Pogorelyy MV
3. Komech EA
4. Karnaukhov VK
5. Bacher P
6. Rosati E
7. Franke A
8. Chudakov DM
9. Mamedov IZ
10. Lebedev YB
11. Mora T
12. Walczak AM
(2020) NCBI BioProject
ID PRJNA633317. Longitudinal high-throughput TCR repertoire profiling reveals the dynamics of T cell memory formation after mild COVID-19 infection.

https://www.ncbi.nlm.nih.gov/bioproject/PRJNA633317/
1. Minervina AA
2. Pogorelyy MV
3. Komech EA
4. Karnaukhov VK
5. Bacher P
6. Rosati E
7. Franke A
8. Chudakov DM
9. Mamedov IZ
10. Lebedev YB
11. Mora T
12. Walczak AM
(2020) Zenodo
Longitudinal high-throughput TCR repertoire profiling reveals the dynamics of T cell memory formation after mild COVID-19 infection.
https://doi.org/10.5281/zenodo.3835955

The following previously published data sets were used

1. Minervina AA
2. Pogorelyy MV
3. Komech EA
4. Karnaukhov VK
5. Bacher P
6. Rosati E
7. Franke A
8. Chudakov DM
9. Mamedov IZ
10. Lebedev YB
11. Mora T
12. Walczak AM
(2019) NCBI BioProject
ID PRJNA577794. Comprehensive analysis of antiviral adaptive immunity formation and reactivation down to single cell level.

https://www.ncbi.nlm.nih.gov/bioproject/PRJNA577794
1. Nolan S
2. Vignali M
3. Klinger M
4. Dines J
5. Kaplan I
6. Svejnoha E
7. Craft T
8. Boland K
9. Pesesky M
10. Gittelman RM
11. Snyder TM
12. Gooley CJ
13. Semprini S
14. Cerchione C
15. Mazza M
16. Delmonte OM
17. Dobbs K
18. Carreño-Tarragona G
19. Barrio S
20. Sambri V
21. Martinelli G
22. Goldman J
23. Heath JR
24. Notarangelo LD
25. Carlson JM
26. Martinez-Lopez J
27. Robins H
(2020) ImmuneAccess
A large-scale database of T-cell receptor beta (TCRb) sequences and binding associations from natural and synthetic exposure to SARS-CoV-2.
https://doi.org/10.21417/ADPT2020COVID
1. Emerson R
2. DeWitt W
3. Vignali M
4. Gravley J
5. Hu J
6. Osborne E
7. Desmarais C
8. Klinger M
9. Carlson C
10. Hansen J
11. Rieder M
12. Robins H
(2017) ImmuneAccess
Immunosequencing identifies signatures of cytomegalovirus exposure history and HLA-mediated effects on the T-cell repertoire.
https://doi.org/10.21417/B7001Z

References

1. Bacher P
2. Rosati E
3. Esser D
4. Martini GR
5. Saggau C
6. Schiminsky E
7. Dargvainiene J
8. Schröder I
9. Wieters I
10. Khodamoradi Y
11. Eberhardt F
12. Vehreschild M
13. Neb H
14. Sonntagbauer M
15. Conrad C
16. Tran F
17. Rosenstiel P
18. Markewitz R
19. Wandinger KP
20. Augustin M
21. Rybniker J
22. Kochanek M
23. Leypoldt F
24. Cornely OA
25. Koehler P
26. Franke A
27. Scheffold A
(2020) Low-Avidity CD4⁺ T cell responses to SARS-CoV-2 in unexposed individuals and humans with severe COVID-19
Immunity 53:1258–1271.
https://doi.org/10.1016/j.immuni.2020.11.016
- PubMed
- Google Scholar
1. Bagaev DV
2. Vroomans RMA
3. Samir J
4. Stervbo U
5. Rius C
6. Dolton G
7. Greenshields-Watson A
8. Attaf M
9. Egorov ES
10. Zvyagin IV
11. Babel N
12. Cole DK
13. Godkin AJ
14. Sewell AK
15. Kesmir C
16. Chudakov DM
17. Luciani F
18. Shugay M
(2020) VDJdb in 2019: database extension, new analysis infrastructure and a T-cell receptor motif compendium
Nucleic Acids Research 48:D1057–D1062.
https://doi.org/10.1093/nar/gkz874
- PubMed
- Google Scholar
1. Bi Q
2. Wu Y
3. Mei S
4. Ye C
5. Zou X
6. Zhang Z
7. Liu X
8. Wei L
9. Truelove SA
10. Zhang T
11. Gao W
12. Cheng C
13. Tang X
14. Wu X
15. Wu Y
16. Sun B
17. Huang S
18. Sun Y
19. Zhang J
20. Ma T
21. Lessler J
22. Feng T
(2020) Epidemiology and transmission of COVID-19 in 391 cases and 1286 of their close contacts in Shenzhen, China: a retrospective cohort study
The Lancet Infectious Diseases 20:911–919.
https://doi.org/10.1016/S1473-3099(20)30287-5
- PubMed
- Google Scholar
(2015) MiXCR: software for comprehensive adaptive immunity profiling
Nature Methods 12:380–381.
https://doi.org/10.1038/nmeth.3364
- PubMed
- Google Scholar
Preprint
1. Braun J
2. Loyal L
3. Frentsch M
4. Wendisch D
5. Georg P
6. Kurth F
7. Hippenstiel S
8. Dingeldey M
9. Kruse B
10. Fauchere F
11. Baysal E
12. Mangold M
13. Henze L
14. Lauster R
15. Mall MA
16. Beyer K
17. Röhmel J
18. Voigt S
19. Schmitz J
20. Miltenyi S
21. Demuth I
22. Müller MA
23. Hocke A
24. Witzenrath M
25. Suttorp N
26. Kern F
27. Reimer U
28. Wenschuh H
29. Drosten C
30. Corman VM
31. Giesecke-Thiel C
32. Sander LE
33. Thiel A
(2020) Presence of SARS-CoV-2-reactive T cells in COVID-19 patients and healthy donors
medRxiv.
https://doi.org/10.1101/2020.04.17.20061440
- Google Scholar
1. Dash P
2. Fiore-Gartland AJ
3. Hertz T
4. Wang GC
5. Sharma S
6. Souquette A
7. Crawford JC
8. Clemens EB
9. Nguyen THO
10. Kedzierska K
11. La Gruta NL
12. Bradley P
13. Thomas PG
(2017) Quantifiable predictive features define epitope-specific T cell receptor repertoires
Nature 547:89–93.
https://doi.org/10.1038/nature22383
- PubMed
- Google Scholar
1. DeWitt WS
2. Emerson RO
3. Lindau P
4. Vignali M
5. Snyder TM
6. Desmarais C
7. Sanders C
8. Utsugi H
9. Warren EH
10. McElrath J
11. Makar KW
12. Wald A
13. Robins HS
(2015) Dynamics of the cytotoxic T cell response to a model of acute viral infection
Journal of Virology 89:4517–4526.
https://doi.org/10.1128/JVI.03474-14
- PubMed
- Google Scholar
1. DeWitt WS
2. Smith A
3. Schoch G
4. Hansen JA
5. Matsen FA
6. Bradley P
(2018) Human T cell receptor occurrence patterns encode immune history, genetic background, and receptor specificity
eLife 7:e38358.
https://doi.org/10.7554/eLife.38358
- PubMed
- Google Scholar
1. Emerson RO
2. DeWitt WS
3. Vignali M
4. Gravley J
5. Hu JK
6. Osborne EJ
7. Desmarais C
8. Klinger M
9. Carlson CS
10. Hansen JA
11. Rieder M
12. Robins HS
(2017) Immunosequencing identifies signatures of Cytomegalovirus exposure history and HLA-mediated effects on the T cell repertoire
Nature Genetics 49:659–665.
https://doi.org/10.1038/ng.3822
- PubMed
- Google Scholar
(2015) Long-lasting stem cell-like memory CD8⁺ T cells with a naïve-like profile upon yellow fever vaccination
Science Translational Medicine 7:282ra48.
https://doi.org/10.1126/scitranslmed.aaa3700
- PubMed
- Google Scholar
1. Glanville J
2. Huang H
3. Nau A
4. Hatton O
5. Wagar LE
6. Rubelt F
7. Ji X
8. Han A
9. Krams SM
10. Pettus C
11. Haas N
12. Arlehamn CSL
13. Sette A
14. Boyd SD
15. Scriba TJ
16. Martinez OM
17. Davis MM
(2017) Identifying specificity groups in the T cell receptor repertoire
Nature 547:94–98.
https://doi.org/10.1038/nature22976
- PubMed
- Google Scholar
1. Grifoni A
2. Weiskopf D
3. Ramirez SI
4. Mateus J
5. Dan JM
6. Moderbacher CR
7. Rawlings SA
8. Sutherland A
9. Premkumar L
10. Jadi RS
11. Marrama D
12. de Silva AM
13. Frazier A
14. Carlin AF
15. Greenbaum JA
16. Peters B
17. Krammer F
18. Smith DM
19. Crotty S
20. Sette A
(2020) Targets of T cell responses to SARS-CoV-2 coronavirus in humans with COVID-19 disease and unexposed individuals
Cell 181:1489–1501.
https://doi.org/10.1016/j.cell.2020.05.015
- PubMed
- Google Scholar
1. Klinger M
2. Pepin F
3. Wilkins J
4. Asbury T
5. Wittkop T
6. Zheng J
7. Moorhead M
8. Faham M
(2015) Multiplex identification of Antigen-Specific T cell receptors using a combination of immune assays and immune receptor sequencing
PLOS ONE 10:e0141561.
https://doi.org/10.1371/journal.pone.0141561
- PubMed
- Google Scholar
1. Le Bert N
2. Tan AT
3. Kunasegaran K
4. Tham CYL
5. Hafezi M
6. Chia A
7. Chng MHY
8. Lin M
9. Tan N
10. Linster M
11. Chia WN
12. Chen MI
13. Wang LF
14. Ooi EE
15. Kalimuddin S
16. Tambyah PA
17. Low JG
18. Tan YJ
19. Bertoletti A
(2020) SARS-CoV-2-specific T cell immunity in cases of COVID-19 and SARS, and uninfected controls
Nature 584:457–462.
https://doi.org/10.1038/s41586-020-2550-z
- PubMed
- Google Scholar
1. Loh L
2. Gherardin NA
3. Sant S
4. Grzelak L
5. Crawford JC
6. Bird NL
7. Koay HF
8. van de Sandt CE
9. Moreira ML
10. Lappas M
11. Allen EK
12. Crowe J
13. Loudovaris T
14. Flanagan KL
15. Quinn KM
16. Rossjohn J
17. Thomas PG
18. Eckle SBG
19. McCluskey J
20. Godfrey DI
21. Kedzierska K
(2020) Human Mucosal-Associated invariant T cells in older individuals display expanded tcrαβ clonotypes with potent antimicrobial responses
The Journal of Immunology 204:1119–1133.
https://doi.org/10.4049/jimmunol.1900774
- PubMed
- Google Scholar
1. Mateus J
2. Grifoni A
3. Tarke A
4. Sidney J
5. Ramirez SI
6. Dan JM
7. Burger ZC
8. Rawlings SA
9. Smith DM
10. Phillips E
11. Mallal S
12. Lammers M
13. Rubiro P
14. Quiambao L
15. Sutherland A
16. Yu ED
17. da Silva Antunes R
18. Greenbaum J
19. Frazier A
20. Markmann AJ
21. Premkumar L
22. de Silva A
23. Peters B
24. Crotty S
25. Sette A
26. Weiskopf D
(2020) Selective and cross-reactive SARS-CoV-2 T cell epitopes in unexposed humans
Science 370:89–94.
https://doi.org/10.1126/science.abd3871
- PubMed
- Google Scholar
1. Meckiff BJ
2. Ramírez-Suástegui C
3. Fajardo V
4. Chee SJ
5. Kusnadi A
6. Simon H
7. Eschweiler S
8. Grifoni A
9. Pelosi E
10. Weiskopf D
11. Sette A
12. Ay F
13. Seumois G
14. Ottensmeier CH
15. Vijayanand P
(2020) Imbalance of regulatory and cytotoxic SARS-CoV-2-Reactive CD4⁺ T Cells in COVID-19
Cell 183:1340–1353.
https://doi.org/10.1016/j.cell.2020.10.001
- PubMed
- Google Scholar
1. Miller JD
2. van der Most RG
3. Akondy RS
4. Glidewell JT
5. Albott S
6. Masopust D
7. Murali-Krishna K
8. Mahar PL
9. Edupuganti S
10. Lalor S
11. Germon S
12. Del Rio C
13. Mulligan MJ
14. Staprans SI
15. Altman JD
16. Feinberg MB
17. Ahmed R
(2008) Human effector and memory CD8+ T cell responses to Smallpox and Yellow Fever Vaccines
Immunity 28:710–722.
https://doi.org/10.1016/j.immuni.2008.02.020
- PubMed
- Google Scholar
1. Minervina AA
2. Pogorelyy MV
3. Komech EA
4. Karnaukhov VK
5. Bacher P
6. Rosati E
7. Franke A
8. Chudakov DM
9. Mamedov IZ
10. Lebedev YB
11. Mora T
12. Walczak AM
(2020) Primary and secondary anti-viral response captured by the dynamics and phenotype of individual T cell clones
eLife 9:e53704.
https://doi.org/10.7554/eLife.53704
- PubMed
- Google Scholar
1. Ng OW
2. Chia A
3. Tan AT
4. Jadi RS
5. Leong HN
6. Bertoletti A
7. Tan YJ
(2016) Memory T cell responses targeting the SARS coronavirus persist up to 11 years post-infection
Vaccine 34:2008–2014.
https://doi.org/10.1016/j.vaccine.2016.02.063
- PubMed
- Google Scholar
1. Ni L
2. Ye F
3. Cheng ML
4. Feng Y
5. Deng YQ
6. Zhao H
7. Wei P
8. Ge J
9. Gou M
10. Li X
11. Sun L
12. Cao T
13. Wang P
14. Zhou C
15. Zhang R
16. Liang P
17. Guo H
18. Wang X
19. Qin CF
20. Chen F
21. Dong C
(2020) Detection of SARS-CoV-2-Specific humoral and cellular immunity in COVID-19 convalescent individuals
Immunity 52:971–977.
https://doi.org/10.1016/j.immuni.2020.04.023
- PubMed
- Google Scholar
1. Oh HL
2. Chia A
3. Chang CX
4. Leong HN
5. Ling KL
6. Grotenbreg GM
7. Gehring AJ
8. Tan YJ
9. Bertoletti A
(2011) Engineering T cells specific for a dominant severe acute respiratory syndrome coronavirus CD8 T cell epitope
Journal of Virology 85:10464–10471.
https://doi.org/10.1128/JVI.05039-11
- PubMed
- Google Scholar
(2020) MAIT cell activation and dynamics associated with COVID-19 disease severity
Science Immunology 5:eabe1670.
https://doi.org/10.1126/sciimmunol.abe1670
- PubMed
- Google Scholar
Preprint
1. Peng Y
2. Mentzer AJ
3. Liu G
4. Yao X
5. Yin Z
6. Dong D
7. Dejnirattisai W
8. Rostron T
9. Supasa P
10. Liu C
11. López-Camacho C
12. Slon-Campos J
13. Zhao Y
14. Stuart DI
15. Paesen GC
16. Grimes JM
17. Antson AA
18. Bayfield OW
19. Hawkins D
20. Ker DS
21. Wang B
22. Turtle L
23. Subramaniam K
24. Thomson P
25. Zhang P
26. Dold C
27. Ratcliff J
28. Simmonds P
29. de Silva T
30. Sopp P
31. Wellington D
32. Rajapaksa U
33. Chen YL
34. Salio M
35. Napolitani G
36. Paes W
37. Borrow P
38. Kessler BM
39. Fry JW
40. Schwabe NF
41. Semple MG
42. Baillie JK
43. Moore SC
44. Openshaw PJM
45. Ansari MA
46. Dunachie S
47. Barnes E
48. Frater J
49. Kerr G
50. Goulder P
51. Lockett T
52. Levin R
53. Zhang Y
54. Jing R
55. Ho LP
56. Cornall RJ
57. Conlon CP
58. Klenerman P
59. Screaton GR
60. Mongkolsapaya J
61. McMichael A
62. Knight JC
63. Ogg G
64. Dong T
65. Oxford Immunology Network Covid-19 Response T cell Consortium
66. ISARIC4C Investigators
(2020) Broad and strong memory CD4+ and CD8+ T cells induced by SARS-CoV-2 in UK convalescent COVID-19 patients
bioRxiv.
https://doi.org/10.1101/2020.06.05.134551
- Google Scholar
1. Pogorelyy MV
2. Elhanati Y
3. Marcou Q
4. Sycheva AL
5. Komech EA
6. Nazarov VI
7. Britanova OV
8. Chudakov DM
9. Mamedov IZ
10. Lebedev YB
11. Mora T
12. Walczak AM
(2017) Persisting fetal clonotypes influence the structure and overlap of adult human T cell receptor repertoires
PLOS Computational Biology 13:e1005572.
https://doi.org/10.1371/journal.pcbi.1005572
- PubMed
- Google Scholar
1. Pogorelyy MV
2. Minervina AA
3. Touzel MP
4. Sycheva AL
5. Komech EA
6. Kovalenko EI
7. Karganova GG
8. Egorov ES
9. Komkov AY
10. Chudakov DM
11. Mamedov IZ
12. Mora T
13. Walczak AM
14. Lebedev YB
(2018) Precise tracking of vaccine-responding T cell clones reveals convergent and personalized response in identical twins
PNAS 115:12704–12709.
https://doi.org/10.1073/pnas.1809642115
- PubMed
- Google Scholar
(2019) Detecting T cell receptors involved in immune responses from single repertoire snapshots
PLOS Biology 17:e3000314.
https://doi.org/10.1371/journal.pbio.3000314
- PubMed
- Google Scholar
(2020) Inferring the immune response from repertoire sequencing
PLOS Computational Biology 16:e1007873.
https://doi.org/10.1371/journal.pcbi.1007873
- PubMed
- Google Scholar
1. Quinti I
2. Lougaris V
3. Milito C
4. Cinetto F
5. Pecoraro A
6. Mezzaroma I
7. Mastroianni CM
8. Turriziani O
9. Bondioni MP
10. Filippini M
11. Soresina A
12. Spadaro G
13. Agostini C
14. Carsetti R
15. Plebani A
(2020) A possible role for B cells in COVID-19? lesson from patients with agammaglobulinemia
Journal of Allergy and Clinical Immunology 146:211–213.
https://doi.org/10.1016/j.jaci.2020.04.013
- PubMed
- Google Scholar
(2010) edgeR: a bioconductor package for differential expression analysis of digital gene expression data
Bioinformatics 26:139–140.
https://doi.org/10.1093/bioinformatics/btp616
- PubMed
- Google Scholar
1. Schmidt ME
2. Varga SM
(2018) The CD8 T cell response to respiratory virus infections
Frontiers in Immunology 9:678.
https://doi.org/10.3389/fimmu.2018.00678
- PubMed
- Google Scholar
(2020) Robust T cell immunity in convalescent individuals with asymptomatic or mild COVID-19
Cell 183:158–168.
https://doi.org/10.1016/j.cell.2020.08.017
- PubMed
- Google Scholar
1. Shomuradova AS
2. Vagida MS
3. Sheetikov SA
4. Zornikova KV
5. Kiryukhin D
6. Titov A
7. Peshkova IO
8. Khmelevskaya A
9. Dianov DV
10. Malasheva M
11. Shmelev A
12. Serdyuk Y
13. Bagaev DV
14. Pivnyuk A
15. Shcherbinin DS
16. Maleeva AV
17. Shakirova NT
18. Pilunov A
19. Malko DB
20. Khamaganova EG
21. Biderman B
22. Ivanov A
23. Shugay M
24. Efimov GA
(2020) SARS-CoV-2 epitopes are recognized by a public and diverse repertoire of human T cell receptors
Immunity 53:1245–1257.
https://doi.org/10.1016/j.immuni.2020.11.004
- PubMed
- Google Scholar
(2014) Towards error-free profiling of immune repertoires
Nature Methods 11:653–655.
https://doi.org/10.1038/nmeth.2960
- PubMed
- Google Scholar
Preprint
1. Snyder TM
2. Gittelman RM
3. Klinger M
4. May DH
5. Osborne EJ
6. Taniguchi R
7. Zahid HJ
8. Kaplan IM
9. Dines JN
10. Noakes MN
11. Pandya R
12. Chen X
13. Elasady S
14. Svejnoha E
15. Ebert P
16. Pesesky MW
17. De Almeida P
18. O'Donnell H
19. DeGottardi Q
20. Keitany G
21. Lu J
22. Vong A
23. Elyanow R
24. Fields P
25. Greissl J
26. Baldo L
27. Semprini S
28. Cerchione C
29. Mazza M
30. Delmonte OM
31. Dobbs K
32. Carreño-Tarragona G
33. Barrio S
34. Imberti L
35. Sottini A
36. Quiros-Roldan E
37. Rossi C
38. Biondi A
39. Bettini LR
40. D'Angio M
41. Bonfanti P
42. Tompkins MF
43. Alba C
44. Dalgard C
45. Sambri V
46. Martinelli G
47. Goldman JD
48. Heath JR
49. Su HC
50. Notarangelo LD
51. Martinez-Lopez J
52. Carlson JM
53. Robins HS
(2020) Magnitude and dynamics of the T-Cell response to SARS-CoV-2 infection at both individual and population levels
medRxiv.
https://doi.org/10.1101/2020.07.31.20165647
- Google Scholar
1. Soresina A
2. Moratto D
3. Chiarini M
4. Paolillo C
5. Baresi G
6. Focà E
7. Bezzi M
8. Baronio B
9. Giacomelli M
10. Badolato R
(2020) Two X-linked agammaglobulinemia patients develop pneumonia as COVID-19 manifestation but recover
Pediatric Allergy and Immunology 31:565–569.
https://doi.org/10.1111/pai.13263
- PubMed
- Google Scholar
Preprint
(2020) Generation of SARS-CoV-2 S1 spike glycoprotein putative antigenic epitopes in vitro by intracellular aminopeptidases
bioRxiv.
https://doi.org/10.1101/2020.06.22.164681
- Google Scholar
(2012) Expanding roles for CD4⁺ T cells in immunity to viruses
Nature Reviews Immunology 12:136–148.
https://doi.org/10.1038/nri3152
- PubMed
- Google Scholar
1. Thevarajan I
2. Nguyen THO
3. Koutsakos M
4. Druce J
5. Caly L
6. van de Sandt CE
7. Jia X
8. Nicholson S
9. Catton M
10. Cowie B
11. Tong SYC
12. Lewin SR
13. Kedzierska K
(2020) Breadth of concomitant immune responses prior to patient recovery: a case report of non-severe COVID-19
Nature Medicine 26:453–455.
https://doi.org/10.1038/s41591-020-0819-2
- PubMed
- Google Scholar
1. Vabret N
2. Britton GJ
3. Gruber C
4. Hegde S
5. Kim J
6. Kuksin M
7. Levantovsky R
8. Malle L
9. Moreira A
10. Park MD
11. Pia L
12. Risson E
13. Saffern M
14. Salomé B
15. Esai Selvan M
16. Spindler MP
17. Tan J
18. van der Heide V
19. Gregory JK
20. Alexandropoulos K
21. Bhardwaj N
22. Brown BD
23. Greenbaum B
24. Gümüş ZH
25. Homann D
26. Horowitz A
27. Kamphorst AO
28. Curotto de Lafaille MA
29. Mehandru S
30. Merad M
31. Samstein RM
32. Sinai Immunology Review Project
(2020) Immunology of COVID-19: current state of the science
Immunity 52:910–941.
https://doi.org/10.1016/j.immuni.2020.05.002
- PubMed
- Google Scholar
Preprint
(2020) Phenotype and kinetics of SARS-CoV-2-specific T cells in COVID-19 patients with acute respiratory distress syndrome
medRxiv.
https://doi.org/10.1101/2020.04.11.20062349
- Google Scholar
1. Zhao J
2. Zhao J
3. Perlman S
(2010) T cell responses are required for protection from clinical disease and for virus clearance in severe acute respiratory syndrome coronavirus-infected mice
Journal of Virology 84:9318–9325.
https://doi.org/10.1128/JVI.01049-10
- PubMed
- Google Scholar

Decision letter

Sandeep Krishna

Reviewing Editor; National Centre for Biological Sciences‐Tata Institute of Fundamental Research, India
Miles P Davenport

Senior Editor; University of New South Wales, Australia
Lindsay Cowell

Reviewer

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This paper uses high throughput longitudinal TCR sequencing to understand the TCR dynamics in two persons infected with SARS-CoV-2. In particular, they find two peaks of T cell clonal expansion at 15 and 37 days post infection. The authors also identify TCR sequence motifs that are likely to be specific to SARS-CoV-2 and thereby show that some T cell clones are present pre-infection suggesting the existence of cross-reactive memory T cells prior to the infection.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting your work entitled "Longitudinal TCR repertoire profiling reveals the dynamics of T cell memory formation after mild COVID-19" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Phil Bradley (Reviewer #2).

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

While we found the questions raised in this study interesting, we concluded that the manuscript would not be suitable for eLife without substantially more evidence that the clonotypes have TCR specificity for SARS-CoV-2. The individual reviews are appended below.

Reviewer #1:

This work investigates the T cell receptor (TCR) repertoires of 2 individuals diagnosed with mild COVID-19 infection. The authors use high-throughput sequencing of 2 biological replicate samples obtained at each of multiple pre-infection and post-infection timepoints to identify TCRalpha and TCRbeta clonotypes that contract or expand post-infection and to investigate potential reactivation of pre-existing memory cells. This is a potentially interesting work that may provide novel insights into T cell responses to SARS-CoV-2. However, some of the specific details of the various analyses reported are not clear and I have several major concerns about the reported work.

1) The primary concern is the TCR specificity of the clonotypes that were determined to be contracting or expanding post-SARS-CoV-2-infection and therefore identified as responding to or reactive to SARS-CoV-2. There is no verification that these expanding or contracting clonotypes have TCR specificity for SARS-CoV-2. One alternative possibility is that some, maybe even many, of these expanding or contracting clonotypes are bystander-activated T cells with TCRs that are not specific for SARS-CoV-2. Similarly, the clonotypes that were identified as contracting or expanding post-SARS-CoV-2 infection and also detected in the memory pool prior to SARS-CoV-2 infection may not be cross-reactive (i.e. specificity for another infection + SARS-CoV-2), as suggested by the authors, but rather non-SARS-CoV-2-specific bystander-activated memory T cells.

While the dynamics of the T cell populations following SARS-CoV-2 infection may be informative regardless of the mode of activation of the T cells (i.e. TCR-mediated vs. bystander activated), the reported TCR clonotype motifs are only informative if these TCRs have SARS-CoV-2 specificity.

2) Another concern is the substantial variation between the various approaches used to identify the contracting and expanding clonotypes post-infection that are associated with COVID-19 infection. The manuscript text states that the EdgeR and NoiseET approaches for identifying expanding and contracting clonotypes yielded similar results. Figure 1—figure supplement 4A, D suggest that the two approaches yield similar trajectories for the identified expanding and contracting clonotype subsets (i.e. fraction of reactive clonotypes). However, the Venn diagrams in Figure 1—figure supplement 4B, C, E, F show that the two approaches are, in some cases, identifying substantially different subsets of expanding or contracting clonotypes. For example, for Donor M in Figure 1—figure supplement 4F, of the 1044 expanded clonotypes identified by NoiseET, only 478 were also identified by EdgeR.

The text also states that the contracting and expanding clonotypes identified using EdgeR largely overlap/correspond to the clusters 2 and 3 of clonal trajectories yielded using PCA (Figure 1B-E) but no quantitative evidence is provided to support this. Venn diagrams, similar to those in Figure 1—figure supplement 4, could be provided that compare the expanding and contracting clonotypes identified using the three different approaches (i.e. EdgeR, NoiseET, and PCA) as applied to TCRα as well as TCRβ clonotypes.

While these differences between methods may not have significant consequences for some of the reported results (eg. temporal clonal trajectories), these differences raise concerns about the results that depend on specific clonotype sequences (eg. Figure 2D-G, Figure 3—figure supplement 2 and Figure S5D-G that report amino acid motifs for contracting and expanding clonotypes).

Reviewer #2:

This manuscript describes a longitudinal study of TCR repertoires in two individuals with mild COVID-19. TCRα and β repertoires at 4 time points post-infection are used to identify T cell clonotypes likely responding to COVID-19. These responding clones fall into two groups, a set of monotonically contracting clones and a set of clones whose frequencies peak (at day ~37) and then contract. Sequencing of memory populations at two time points and availability of TCR repertoire data from both individuals prior to infection allow the authors to map clonotypes to memory phenotypes and to identify a handful of responding clones that existed in the memory compartment prior to infection. Clusters of sequence-similar clonotypes are identified that suggest focused responses to immunodominant epitopes. This is a succinct and timely study and I have no major concerns, just a few questions/suggestions/typos detailed below.

How unexpected is the TCR clustering evident in Figure 2D-G? For example if the same number of equally high Pgen sequences were selected at random? I wonder whether the authors could run ALICE on just the responding clones (not the full dataset) to assess which neighborhoods are very unlikely to occur by chance.

Could the "computational chain pairing" method of Minervina et al. be applied to this data? If only to try to connect some of the sequence motifs between the α and β chains?

Reviewer #3:

This is a case report analysing TCR repertoire on two individuals with suspected COVID-19 infection. The report shows that a set of TCR sequences expands between days 15 and day 30/37 and another set contract. The amount of expansion/contraction is not clearly shown. Most of these sequences are found in the memory phenotype. A few (especially CD4) are found before immunisation. As the authors point out, the evidence that the TCRs recognise COVID-19 is purely circumstantial. Even if they do, I do not see that this study contributes significantly to understanding either the protective or the pathological immune response to COVID-19.

1) The Abstract is full of unsubstantiated claims. For example "T cell response is a critical part of both individual and herd immunity to SARS-CoV-2 and the eﬃcacy of developed vaccines. " Or "In both donors we identiﬁed SARS-CoV-2-responding CD4⁺ and CD8⁺ T cell clones. We describe characteristic motifs in TCR sequences of COVID- 19-reactive clones, suggesting the existence of immunodominant epitopes." The authors do not identify COVID-19 responding clones; nor do they show any evidence that there are immunodominant epitopes.

2) Figure 1 What does "normalized trajectory of TCR clones in each cluster" mean ? It would be interesting to see the magnitude of the responses. Similarly, I don't really understand the y axis in panels d and e.

3) Figure 3. I don't understand panels a and b. Is this the proportion of contracting TCR sequences which are memory phenotype? If so, what are the rest ? Or are they simply not captured. The figure legend is obscure.

https://doi.org/10.7554/eLife.63502.sa1

Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

While we found the questions raised in this study interesting, we concluded that the manuscript would not be suitable for eLife without substantially more evidence that the clonotypes have TCR specificity for SARS-CoV-2. The individual reviews are appended below.

In this resubmission, we now provide such validation using three different approaches:

1) For the HLA-A02 positive donor M, we have acquired TCRbeta and TCRalpha sequences of TCRs binding a HLA-A02 tetramer loaded with a known SARS-CoV-2 peptide. We show that TCRs identified in this assay are independently found in our longitudinal analysis (see Figure 3A in the manuscript).

2) Additionally to our data, a large dataset of SARS-CoV-2 specific TCRs from SARSCoV-2 peptide stimulation experiments of CD8⁺ T cells became available just recently (Snyder et al., 2020, medarxiv). We used this large set of TCR-peptide pairs to further map the specificities of COVID-19-associated TCRs identified by our longitudinal approach. Importantly, the results from this computational mapping agree well with the results we have from analysing HLA-A02 tetramer-specific clones (see Figure 3C in the manuscript). This new analysis also allows us to estimate the fraction of expanding CD8⁺ T cells corresponding to TCRs specific for certain SARS-CoV-2 peptides in vivo.

3) Finally, to validate CD4⁺ clonotypes (points 1 and 2 above only concerned CD8⁺ cells), we used a large dataset of bulk TCRbeta repertoires from 1414 COVID-19 patients reported in the same study (Snyder et al., 2020). We show that the CD4⁺ motifs we report are significantly expanded in COVID-19 patients in comparison to healthy controls (see Figure 4C,D in the manuscript). Moreover, we show that these CD4⁺ motifs are in fact present in the majority of COVID-19 patients with matching HLA-context.

Another point raised by the reviewers was a lack of diagnostic potential of our method. We now show that the motifs identified from the longitudinal analysis are able to identify COVID positive patients.

In this revision we have also added a new timepoint at 3 months for each donor (both bulk and memory repertoires). The analysis of this timepoint provides additional information about the decay of COVID-19 associated T cell clones and their transition to long-lived memory subpopulations on a larger time scale.

Reviewer #1:

This work investigates the T cell receptor (TCR) repertoires of 2 individuals diagnosed with mild COVID-19 infection. The authors use high-throughput sequencing of 2 biological replicate samples obtained at each of multiple pre-infection and post-infection timepoints to identify TCRalpha and TCRbeta clonotypes that contract or expand post-infection and to investigate potential reactivation of pre-existing memory cells. This is a potentially interesting work that may provide novel insights into T cell responses to SARS-CoV-2. However, some of the specific details of the various analyses reported are not clear and I have several major concerns about the reported work.

1) The primary concern is the TCR specificity of the clonotypes that were determined to be contracting or expanding post-SARS-CoV-2-infection and therefore identified as responding to or reactive to SARS-CoV-2. There is no verification that these expanding or contracting clonotypes have TCR specificity for SARS-CoV-2. One alternative possibility is that some, maybe even many, of these expanding or contracting clonotypes are bystander-activated T cells with TCRs that are not specific for SARS-CoV-2. Similarly, the clonotypes that were identified as contracting or expanding post-SARS-CoV-2 infection and also detected in the memory pool prior to SARS-CoV-2 infection may not be cross-reactive (i.e. specificity for another infection + SARS-CoV-2), as suggested by the authors, but rather non-SARS-CoV-2-specific bystander-activated memory T cells.

While the dynamics of the T cell populations following SARS-CoV-2 infection may be informative regardless of the mode of activation of the T cells (i.e. TCR-mediated vs. bystander activated), the reported TCR clonotype motifs are only informative if these TCRs have SARS-CoV-2 specificity.

We thank the reviewer for the valuable comments. We agree that the lack of evidence of SARS-CoV-2 specificity of identified responding clonotypes was the main limitation of the previous version of the manuscript. In the revised version we provide both experimental and computational proof of the SARS-CoV-2 specificity of reported clonotypes. Donor M from the current study also participated in the study by Shomuradova et al. 2020 (as donor p1434), where T cells were sorted using a HLA-A02 tetramer loaded with a SARS-CoV-2 epitope from the spike protein. Sequences of TCRs specific for this SARS-CoV-2 epitope (YLQPRTFLL) were deposited to VDJdb after we shared our original preprint. We now use these sequences to prove that our approach is capable of identifying clonotypes specific to a certain SARS-CoV-2 epitope by clonal dynamics only, see Figure 3A. We show that a) tetramer-specific TCRs show temporal dynamics corresponding to the contracting clones’ group (monotonically decreasing in frequency after day 15 post-infection). b) Our approach identifies almost all tetramerspecific TCRs which are abundant on day 15. c) some groups of highly similar TCR sequences of contracting clones in TCRbeta and TCRalpha repertoire correspond to tetramer-specific clones.

We next compared our CD8⁺ TCRbeta sequences with a recently published large set of TCRbetas from T cells stimulated with SARS-CoV-2 peptide pools (MIRA assay from Snyder et al., 2020, medarxiv). This allowed us to map specificities of our TCRbeta sequences to certain peptides/or pools of overlapping peptides from the MIRA assay. Importantly, this procedure correctly mapped the specificity of TCRbetas from tetramerspecific clones described above. TCR specific for one of MIRA epitopes (NQKLIANQF) were found in the pre-existing memory of donor M. Interestingly this epitope is just one mismatch away from the homologous peptide (NQKLIANAF) of common cold betacoronaviruses, suggesting potential cross-reactivity between these two epitopes.

Finally, we prove SARS-CoV-2 association of discovered CD4⁺ TCRbeta contracting motifs in both donors using a large cohort of COVID-19 patients from ref. Snyder et al., 2020. We show that occurrence of TCRbetas from the largest sequence similarity groups of both donors is predictive of COVID-19 positivity in donors with matching HLA-context.

1) New Figure 3A shows longitudinal tracking of A*02:01-YLQPRTFLL tetramer clones (Figure 3—figure supplement 1 for TCRalpha).

2) Occurence of tetramer clones in clusters of similar contracted TCRbetas and TCRalpha is shown on Figure 3B and c respectively.

3) Mapping of clusters of the similar CD8 TCRbetas to SARS-CoV-2 specific peptides from MIRA experiment is shown on Figure 3C as shaded circles.

4) Occurrence of main CD4⁺ motifs from contracting clones of both donors in large cohorts of COVID-19 patients and pre-pandemic controls is shown on Figure 4C,D.

5) Separate columns in Supplementary file 3 shows if clones were detected in the tetramer enrichment experiment from Shomuradova et al., as well as specificity for peptide/peptide pool obtained by comparison with MIRA database.

6) Both Results and Materials and methods sections are updated to describe new results.

7) We note that none of these methods proved SARS-CoV-2 specificity of clones with a peak on day 37 (either because they are not specific for SARS-CoV-2, or because our validation methods are more suitable for CD8 clones). We discuss this issue in the Discussion.

2) Another concern is the substantial variation between the various approaches used to identify the contracting and expanding clonotypes post-infection that are associated with COVID-19 infection. The manuscript text states that the EdgeR and NoiseET approaches for identifying expanding and contracting clonotypes yielded similar results. Figure 1—figure supplement 4A, D suggest that the two approaches yield similar trajectories for the identified expanding and contracting clonotype subsets (i.e. fraction of reactive clonotypes). However, the Venn diagrams in Figure 1—figure supplement 4B, C, E, F show that the two approaches are, in some cases, identifying substantially different subsets of expanding or contracting clonotypes. For example, for Donor M in Figure 1—figure supplement 4F, of the 1044 expanded clonotypes identified by NoiseET, only 478 were also identified by EdgeR.

The text also states that the contracting and expanding clonotypes identified using EdgeR largely overlap/correspond to the clusters 2 and 3 of clonal trajectories yielded using PCA (Figure 1B-E) but no quantitative evidence is provided to support this. Venn diagrams, similar to those in Figure 1—figure supplement 4, could be provided that compare the expanding and contracting clonotypes identified using the three different approaches (i.e. EdgeR, NoiseET, and PCA) as applied to TCRα as well as TCRβ clonotypes.

While these differences between methods may not have significant consequences for some of the reported results (eg. temporal clonal trajectories), these differences raise concerns about the results that depend on specific clonotype sequences (eg. Figure 2D-G, Figure 3—figure supplement 2 and Figure S5 D-G that report amino acid motifs for contracting and expanding clonotypes).

To ensure that all of our results are reproducible with both the NOISET and edgeR approaches, we redid all the analysis and now call a clonotype “contracting” or “expanded” if it was called significant by both approaches simultaneously. Hence we show that active clonal dynamics, participation of the pre-existing memory in the response and SARSCoV-2 TCR sequence motifs we report in our initial preprint are successfully captured by both edgeR and NoisET noise models.

All figures and values in the manuscript are updated with significantly contracting and expanding clonotype groups identified with both models.

We used PCA as an exploratory data analysis technique to identify general patterns in clonotype dynamics and inform the selection of timepoints for more precise analysis with edgeR/NoisET. We do not use PCA to identify significantly contracted/expanded clones, as PCA does not take noise into account.

The overlap between clusters identified by clustering of the principal components and significant clonotypes called by both edgeR and NoisET simultaneously is now shown on Figure 1—figure supplement 5.

We also added different shapes in Figure 1BC and Figure 1—figure supplement 3A-B to indicate clonotypes called as expanding/contracting by both edgeR and NoisET.

Reviewer #2:

This manuscript describes a longitudinal study of TCR repertoires in two individuals with mild COVID-19. TCRα and β repertoires at 4 time points post-infection are used to identify T cell clonotypes likely responding to COVID-19. These responding clones fall into two groups, a set of monotonically contracting clones and a set of clones whose frequencies peak (at day ~37) and then contract. Sequencing of memory populations at two time points and availability of TCR repertoire data from both individuals prior to infection allow the authors to map clonotypes to memory phenotypes and to identify a handful of responding clones that existed in the memory compartment prior to infection. Clusters of sequence-similar clonotypes are identified that suggest focused responses to immunodominant epitopes. This is a succinct and timely study and I have no major concerns, just a few questions/suggestions/typos detailed below.

How unexpected is the TCR clustering evident in Figure 2D-G? For example if the same number of equally high Pgen sequences were selected at random? I wonder whether the authors could run ALICE on just the responding clones (not the full dataset) to assess which neighborhoods are very unlikely to occur by chance.

We show that clustering events are indeed unlikely to occur by chance. This can be assessed by creating a random sample of sequences from the repertoire of the same size as the responding subset, and by comparing the average number of edges in their similarity networks. That number is significantly lower in the random subset than in the responding subset (as expected, with exception of expanding CD8⁺ clones group, there is no strong motif there).

We added new Figure 3D and Figure 4—figure supplement 1A showing the distribution of the number of edges expected under a random sampling assumption.

Could the "computational chain pairing" method of Minervina et al. be applied to this data? If only to try to connect some of the sequence motifs between the α and β chains?

We thank the reviewer for this suggestion. We applied a pairing method from Minervina et al., and found that most clusters do not show consistent pairing with each other. However, we identified two cases in CD8⁺ contracting clones with repeated cluster-to cluster pairings.

We have added this information in the Materials and methods. The sequence of the predicted α chain for each TCRbeta clone is given in Supplementary file 3 and Supplementary file 5.

Reviewer #3:

This is a case report analysing TCR repertoire on two individuals with suspected COVID-19 infection. The report shows that a set of TCR sequences expands between days 15 and day 30/37 and another set contract. The amount of expansion/contraction is not clearly shown. Most of these sequences are found in the memory phenotype. A few (especially CD4) are found before immunisation. As the authors point out, the evidence that the TCRs recognise COVID-19 is purely circumstantial. Even if they do, I do not see that this study contributes significantly to understanding either the protective or the pathological immune response to COVID-19.

1) The Abstract is full of unsubstantiated claims. For example "T cell response is a critical part of both individual and herd immunity to SARS-CoV-2 and the eﬃcacy of developed vaccines. " Or "In both donors we identiﬁed SARS-CoV-2-responding CD4⁺ and CD8⁺ T cell clones. We describe characteristic motifs in TCR sequences of COVID- 19-reactive clones, suggesting the existence of immunodominant epitopes." The authors do not identify COVID-19 responding clones; nor do they show any evidence that there are immunodominant epitopes.

We agree that evidence for these claims was suggestive in the initial preprint. However we believe we now provide strong evidence for these points in the revised version:

1) We provide evidence that clones identified by our longitudinal analysis are indeed COVID-19 specific with 3 different methods: by the sequencing of TCRbeta and TCRalpha of T cell clones stained with HLA*A02-tetramers loaded with YLQPRTFLL (peptide from SARS-CoV-2 spike, see Figure 3A); by matching of TCRbeta sequences to a large recently published dataset of SARS-CoV-2 reactive CD8⁺ T cells (MIRA peptide assay, Snyder et al., 2020, see Figure 3C); and by showing preferential occurrence of both CD8 and CD4 T cell motifs in a large TCRbeta dataset of 1414 COVID-19 patients (Snyder et al., 2020), Figures 4CD.

2) In the original preprint we showed that T cell clones with active dynamics after COVID-19 frequently feature highly similar TCRbeta/TCRalpha chains. It was previously shown by multiple groups that similar TCR sequences frequently recognise the same epitope. We hypothesized that clusters of highly similar TCRs, corresponding to the large fraction of expanding cells, recognise certain immunodominant epitopes.

By sequencing TCRs from tetramer-binding sorted cells, and by searching for TCRbeta clones in databases of epitope-specific TCRs, we were able to confirm this suggestion. The new Figure 3E shows the fraction of the CD8⁺ response mapped to certain peptides, showing that the CD8⁺ T cells responding to the spike-derived HLA-B*15:01 restricted NQKLIANQF epitope corresponds to 21% of CD8⁺ response at its peak. Another prominent epitope is HLA-A*02:01-YLQPRTFLL, which corresponded to 4% of CD8⁺ response at the peak. The occurrence of the described CD4 motifs in a substantial part of HLA-matched COVID-19 patients from Snyder et al., 2020 also points towards the existence of immunodominant CD4 epitopes in the SARS-CoV-2 genome, which elicit a response in the majority of COVID-19 patients with matching HLA-context.

1) We modified the Abstract and reformulated the claims pointed out by the reviewer to be more specific.

2) We added additional analyses proving the COVID-19 specificity of the TCRs we identified and mapping them to certain epitopes:

a) New Figure 3A shows longitudinal tracking of A*02:01-YLQPRTFLL tetramer specific clones (Figure 3—figure supplement 1 for TCRalpha).

b) The occurrence of tetramer-positive clones in clusters of similar contracted TCRalpha and TCRbeta clonotypes is shown in Figure 3B and C.

c) The mapping of TCRbeta clusters to SARS-CoV-2 specific peptides from the MIRA experiment is shown in Figure 3C (shaded circles).

d) The occurrence of the main CD4⁺ motifs of contracting clones from both donors in large cohorts of COVID-19 patients and the pre-pandemic control cohort is shown in Figure 4C, D.

e) Separate columns in Supplementary file 3 report if clones were detected in the tetramer enrichment experiment from Shomuradova et al., as well as the specificity for peptides from the MIRA database.

f) Both the Results and Materials and methods sections are updated to describe the new analyses.

2) Figure 1 What does "normalized trajectory of TCR clones in each cluster" mean ? It would be interesting to see the magnitude of the responses. Similarly, I don't really understand the y axis in panels d and e.

1) We reformulated and expanded Figure 1’s caption to make it clear that clonal frequencies are normalized by the maximum value for each clone in Figure 1B-C, and that Figure 1B-C’s left panels show the average of such normalized frequencies within each cluster (and thus show the dynamical pattern of an average clone in that cluster).

2) We now discuss the magnitude of the responses in the text.

3) We reformulated the Figure 1D-E caption to explain the y-axis more clearly: it is the sum of the frequencies of all clonotypes detected as “contracting” (D) and “expanding” (E).

3) Figure 3. I don't understand panels a and b. Is this the proportion of contracting TCR sequences which are memory phenotype? If so, what are the rest ? Or are they simply not captured. The figure legend is obscure.

Yes, this is correct. These data are now shown in Figure 2, and we have expanded its caption to explain it more clearly. We profile PBMC repertoires more deeply than repertoires of FACS sorted memory cells, and thus indeed low-abundant responding clones are captured in PBMC, but not in memory subpopulations. We have added a discussion of this sampling issue in the text.

https://doi.org/10.7554/eLife.63502.sa2

Article and author information

Author details

Anastasia A Minervina

Shemyakin-Ovchinnikov Institute of Bioorganic Chemistry, Moscow, Russian Federation

Contribution
Conceptualization, Data curation, Software, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing

Competing interests
No competing interests declared

"This ORCID iD identifies the author of this article:" 0000-0001-9884-6351
Ekaterina A Komech
1. Shemyakin-Ovchinnikov Institute of Bioorganic Chemistry, Moscow, Russian Federation
2. Pirogov Russian National Research Medical University, Moscow, Russian Federation
Contribution
Validation, Investigation, Methodology, Writing - review and editing

Competing interests
No competing interests declared
Aleksei Titov

National Research Center for Hematology, Moscow, Russian Federation

Contribution
Validation, Investigation, Visualization, Methodology, Writing - review and editing

Competing interests
No competing interests declared
Meriem Bensouda Koraichi

Laboratoire de physique de l'École Normale Supérieure, ENS, PSL, Sorbonne Universite, Universite de Paris, and CNRS, Paris, France

Contribution
Software, Formal analysis, Investigation, Methodology

Competing interests
No competing interests declared
Elisa Rosati

Institute of Clinical Molecular Biology, Kiel University, Kiel, Germany

Contribution
Resources, Investigation, Methodology, Project administration, Writing - review and editing

Competing interests
No competing interests declared

"This ORCID iD identifies the author of this article:" 0000-0002-2635-6422
Ilgar Z Mamedov
1. Shemyakin-Ovchinnikov Institute of Bioorganic Chemistry, Moscow, Russian Federation
2. Pirogov Russian National Research Medical University, Moscow, Russian Federation
3. Masaryk University, Central European Institute of Technology, Brno, Czech Republic
4. V.I. Kulakov National Medical Research Center for Obstetrics, Gynecology and Perinatology, Moscow, Russian Federation
Contribution
Resources, Supervision, Funding acquisition, Methodology, Project administration, Writing - review and editing

Competing interests
No competing interests declared
Andre Franke

Institute of Clinical Molecular Biology, Kiel University, Kiel, Germany

Contribution
Resources, Supervision, Funding acquisition, Project administration

Competing interests
No competing interests declared
Grigory A Efimov

National Research Center for Hematology, Moscow, Russian Federation

Contribution
Resources, Supervision, Investigation, Methodology, Writing - review and editing

Competing interests
No competing interests declared
Dmitriy M Chudakov
1. Shemyakin-Ovchinnikov Institute of Bioorganic Chemistry, Moscow, Russian Federation
2. Pirogov Russian National Research Medical University, Moscow, Russian Federation
3. Masaryk University, Central European Institute of Technology, Brno, Czech Republic
Contribution
Conceptualization, Resources, Supervision, Funding acquisition, Methodology, Project administration, Writing - review and editing

Competing interests
No competing interests declared

"This ORCID iD identifies the author of this article:" 0000-0003-0430-790X
Thierry Mora

Laboratoire de physique de l'École Normale Supérieure, ENS, PSL, Sorbonne Universite, Universite de Paris, and CNRS, Paris, France

Contribution
Conceptualization, Formal analysis, Supervision, Investigation, Methodology, Writing - original draft, Project administration, Writing - review and editing

Competing interests
No competing interests declared

"This ORCID iD identifies the author of this article:" 0000-0002-5456-9361
Aleksandra M Walczak

Laboratoire de physique de l'École Normale Supérieure, ENS, PSL, Sorbonne Universite, Universite de Paris, and CNRS, Paris, France

Contribution
Conceptualization, Formal analysis, Supervision, Funding acquisition, Investigation, Methodology, Writing - original draft, Project administration, Writing - review and editing

Competing interests
Senior editor, eLife

"This ORCID iD identifies the author of this article:" 0000-0002-2686-5702
Yuri B Lebedev
1. Shemyakin-Ovchinnikov Institute of Bioorganic Chemistry, Moscow, Russian Federation
2. Moscow State University, Moscow, Russian Federation
Contribution
Conceptualization, Resources, Supervision, Funding acquisition, Methodology, Project administration, Writing - review and editing

Competing interests
No competing interests declared

"This ORCID iD identifies the author of this article:" 0000-0003-4554-4733
Mikhail V Pogorelyy
1. Shemyakin-Ovchinnikov Institute of Bioorganic Chemistry, Moscow, Russian Federation
2. Pirogov Russian National Research Medical University, Moscow, Russian Federation
Contribution
Conceptualization, Data curation, Software, Formal analysis, Supervision, Investigation, Methodology, Writing - original draft, Project administration, Writing - review and editing

For correspondence
m.pogorely@gmail.com

Competing interests
No competing interests declared

"This ORCID iD identifies the author of this article:" 0000-0003-0773-1204

Funding

Russian Science Foundation (RSF 20-15-00351)

Yuri B Lebedev

Deutsche Forschungsgemeinschaft (Exc2167)

Andre Franke

Deutsche Forschungsgemeinschaft (4096610003)

Andre Franke

H2020 European Research Council (COG 724208)

Aleksandra M Walczak

Russian Foundation for Basic Research (19-54-12-011)

Ilgar Z Mamedov

Russian Foundation for Basic Research (18-19-09132)

Ilgar Z Mamedov

Ministry of Science and Higher Education (075-15-2019-1789)

Dmitriy M Chudakov

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

The study was supported by RSF 20-15-00351. ER and AF are supported by the DFG Excellence Cluster Precision Medicine in Chronic Inflammation (Exc2167) and the DFG grant n. 4096610003, MBK, TM, and AMW are supported by European Research Council COG 724208. IZM is supported by RFBR 19-54-12011 and 18-29-09132. DMC is supported by the grant from Ministry of Science and Higher Education of the Russian Federation 075-15-2019-1789.

Ethics

Human subjects: All subjects gave written informed consent in accordance with the Declaration of Helsinki. The study protocol was approved by the Pirogov Russian National Research Medical University local ethics committee (#194 granted on March 16, 2020).

Senior Editor

Miles P Davenport, University of New South Wales, Australia

Reviewing Editor

Sandeep Krishna, National Centre for Biological Sciences‐Tata Institute of Fundamental Research, India

Reviewer

Lindsay Cowell

Publication history

Received: September 27, 2020
Accepted: January 5, 2021
Accepted Manuscript published: January 5, 2021 (version 1)
Version of Record published: January 13, 2021 (version 2)

Copyright

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.