Expansion of Unique Hepatitis C Virus–Specific Public CD8⁺ T Cell Clonotypes during Acute Infection and Reinfection

Via their TCRs, an αβ heterodimer, CD8⁺ T cells recognize Ags presented by class I MHC molecules. TCR repertoire diversity is achieved through germline V, D, and J gene segment rearrangements and addition or deletion of nucleotides between segments, resulting in a specific CD8⁺ TCR or clonotype. The TCRAV and TCRBV for the α- and β-chains, respectively, encode CDR1 and CDR2, which interact with the MHC molecule (1). Ag-binding specificity is determined by the most variable region of the TCR, the CDR3 region, formed by the V-J α and V(D)J β junctions. An identical clonotype or CDR3 amino acid sequence detected in several individuals, termed “public,” was associated with better control of viral infections, including CMV and HIV, whereas less common or “private” repertoires were not (2–4).

Hepatitis C virus (HCV) infection has two dichotomous outcomes where ∼25% of acutely infected subjects resolve spontaneously, whereas the rest acquire persistent infection. Expansion of HCV-specific CD8⁺ T cells targeting multiple epitopes (i.e., broad) and producing several cytokines (i.e., polyfunctional) is essential to clear acute primary infection (5–7). Hence HCV represents a unique opportunity to examine the protective capacity of virus-specific CD8⁺ T cell public clonotypes. Cross-sectional studies reported the presence of public TCRs recognizing the hypervariable region 1 (HVR1) ³⁹⁸SLASLFTQGA⁴⁰⁷ epitope of the E2 glycoprotein and the nonstructural 3 (NS3) ¹³⁹⁵HSKKKCDEL¹⁴⁰³ (HSK) or ¹⁴³⁵ATDALMTGY¹⁴⁴³ (ATD) HCV epitopes, irrespective of HCV infectious outcome (8, 9). These studies mostly relied on in vitro expansion of Ag-specific T cells before analysis that may have favored expansion of certain TCR clonotypes. Furthermore, very limited analysis was performed early during acute infection. Whether other immunodominant HCV epitopes favor the expansion of public clonotypes and whether they correlate with spontaneous resolution of acute infection remain unknown.

The HLA-A2–restricted NS3-1073 ¹⁰⁷³CINGVCWTV¹⁰⁸¹ epitope, derived from the highly immunogenic NS3 protein, is one of the immunodominant epitopes during HCV infection (10–12). Spontaneous clearance of acute HCV was associated with an NS3-1073–specific CD8⁺ T cell response (13–15). Long-lived NS3-1073–specific memory CD8⁺ T cells were detected two decades after HCV spontaneous clearance, suggesting it may contribute to long-term protective immunity (10). Indeed, our group has previously demonstrated expansion of memory NS3-1073–specific CD8⁺ T cells in HCV resolvers after virus re-exposure and reinfection (5). In individuals who spontaneously resolved their reinfection, we observed focusing of the TCR repertoire with rapid and selective expansion of memory CD8⁺ T cell clonotypes with the highest functional avidity (16).

CD8⁺ T cells specific to the HCV NS3-1073 epitope are also known to be cross-reactive to several HLA-A2–restricted epitopes. The first cross-reactivity reported was against the NA-231 epitope of influenza virus (17). Other cross-reactivities were later identified against CMV (pp65) and EBV (LMP2) (17–19). In addition, a pre-existing pool of memory HCV NS3-1073–specific CD8⁺ T cells, induced by heterologous infections, was characterized in a cohort of healthy HCV-seronegative subjects (19). These memory HCV NS3-1073–specific CD8⁺ T cells influenced the T cell response to HCV peptide vaccine (19). These studies suggested that this cross-reactivity may contribute to enhanced spontaneous clearance or pathogenesis of acute HCV infection.

In this article, we characterized and compared the HCV NS3-1073–specific TCR repertoire during the very early acute phase of HCV infection and long-term follow-up in spontaneous resolvers (SRs) and during acute infection in subjects who developed chronic infection (CI). Our main objective was to characterize the presence of public clonotypes against this immunodominant epitope and their potential association with spontaneous HCV resolution directly ex vivo. We identified a common set of public clonotypes independent of infection outcome. Although several CDR3 sequences were shared with other non-HCV HLA-A2–restricted epitopes, these potentially cross-reactive clonotypes did not expand during acute HCV infection. The public clonotypes we identified were unique to HCV.

Study subjects were recruited among people who inject drugs enrolled in the Montreal Hepatitis C Cohort (20). Subjects were followed every 3 mo for HCV infection. Acute HCV primary infection was defined as a positive HCV RNA and/or anti-HCV Ab–positive test after a negative test in the past 24 wk. The estimated date of infection (EDI) was calculated as the median time between last negative and first positive RNA or Ab test. Spontaneous resolution was defined as viral clearance (HCV RNA negative) at ≤24 wk (SR, n = 8), and CI was defined as HCV RNA positive at 24 wk (CI, n = 6). For the purpose of this study, two different time points were used: acute (<24 wk or 168 d) and follow-up (>49 wk or 343 d). Two HCV spontaneously resolved subjects who experienced a documented reinfection episode were recruited within the same cohort where reinfection was defined as a positive HCV-RNA test after two negative tests ≥60 d apart. The detailed analysis of their immune response and T cell repertoire analysis were described previously (5, 16). Subjects’ demographics are described in Table I. This study was approved by the Institutional Ethics Committee of the Centre de Recherche du Centre Hospitalier de l’Université de Montréal (CRCHUM; Protocol SL05.014).

The MHC class I monomer, HLA-A2 bound to the HCV NS3 peptide aa 1073–1081 (CINGVCWTV) [A2/NS3-1073], was synthesized by the National Institutes of Health Tetramer Core Facility (Emory University, Atlanta, GA). Tetramers were prepared by adding 10 μl of PE-labeled ExtrAvidin (Sigma, St. Louis, MO) five times, with 10-min incubation at room temperature after each addition. Cryopreserved PBMCs were thawed, and CD8⁺ T cells were negatively selected using the MACS CD8⁺ T cell Isolation Kit (Miltenyi Biotec, Auburn, CA). Tetramer staining and cell surface staining were performed on isolated CD8⁺ T cells as previously described (16). Directly conjugated mAbs against the following cell-surface markers were used: CD3–Pacific Blue (clone UCHT1), CD8–Alexa Fluor 700 (clone RPA-T8), and CD45RO–BB515 (clone UCHL1), all from BD Biosciences (San Diego, CA). The LIVE/DEAD fixable aqua dead cell Stain Kit (Molecular Probes Thermo Fisher Scientific, Burlington, ON, Canada) was used to identify live cells. Cell sorting was performed using a BD Aria II cell sorter equipped with blue (488 nm), red (633 nm), and violet (405 nm) lasers (BD Biosciences), while multiparameter flow cytometry was performed using a BD LSRII instrument with the same setup and an additional yellow-green laser (561 nm) using FACSDiva software version 6.1.3 (BD Biosciences) at the CRCHUM flow cytometry core. Data files were analyzed using FlowJo software version 10.4.11 (BD Biosciences).

FACS-sorted HCV-specific CD8⁺ T cells from each subject were sent to Adaptive Biotechnologies (Seattle, WA) for genomic DNA extraction followed by TCRβ sequencing, as previously described (16). Data analyses were performed using ImmunoSEQ Analyzer software (v3.0) and R v3.5.1. All TCR repertoires were cleaned to remove sequences with a stop codon and those that are out of frame. The numbers of FACS-sorted CD8⁺ T cells, total and unique productive sequences, and clonality for each sample are detailed in Supplemental Table I. Diversity indices (Simpson diversity, Shannon entropy, CDR3 length, and NT additions) were provided by Adaptive Biotechnologies.

HCV RNA was extracted from 500 μl of plasma EDTA samples using QIAamp MinElute Virus spin kit (Qiagen, Germany), following the manufacturer’s protocol. As previously described (21), purified viral RNA was reverse transcribed, PCR amplified, cloned, and sequenced at the McGill University and Genome Quebec Innovation Centre (Montréal, QC).

We extracted HLA-A2–restricted Ag-specific TCR sequences from the VDJ database (https://vdjdb.cdr3.net/). Using R (https://www.R-project.org/) (22), we identified exact CDR3 amino acid sequence matches within this database (23) and our HCV NS3-1073 TCR repertoire dataset. Frequencies of identified CDR3 amino acid sequences in a particular subject’s repertoire were summed up to quantify the percentage of clonotypes identical to other epitope(s)-specific TCR repertoire.

IFN-γ ELISPOT assays were performed using 2 × 10⁵ cryopreserved PBMCs, as previously described (24). The following peptides were used for this assay: HCV (NS3; CINGVCWTV), Flu (NA; CVNGSCFTV), CMV (pp65; NLVPMVATV), EBV (BMLF1; GLCTLVAML), Flu (M1; GILGFVFTL), yellow fever virus (YFV; NS4B; LLWNGPMAV), and Melanoma (MLANA; ELAGIGILTV). The HCV NS3-1073 peptide was synthesized by the Sheldon Biotechnology Centre (McGill University, Montréal, QC, Canada). The Flu NA-231 peptide was synthesized by JPT Peptide Technologies (Berlin, Germany), whereas the other four were synthesized by ProImmune (Sarasota, FL).

HCV-specific CD8⁺ T cells were expanded as previously described (25). In brief, freshly thawed PBMCs were seeded in a 48-well plate at a concentration of 2 × 10⁶ cells/well, in R10 (RPMI 1640 + 10% heat-inactivated FBS [Life Technologies]), supplemented with penicillin + streptomycin (P/S; 1×; Wisent) and 14 mM HEPES (Wisent; R10-P/S-HEPES). Cells were stimulated with 10 μg/ml HCV NS3 peptide (NS3-1073-1081; CINGVCWTV) for 2 wk at 37°C. Half of the medium was replenished every 3 d with R10-P/S-HEPES supplemented with 40 U/ml rIL-2 (National Institutes of Health AIDS Reagents Program) (R10-P/S-Hepes-IL-2, first rIL-2 addition on day 3). Expanded HCV-specific CD8⁺ T cells were then cryopreserved in freeze mix (FBS + 10% DMSO) at a concentration of 10⁷ cells/ml. Thawed cells were restimulated in a 48-well plate at a concentration of 10⁶ cells/well in the presence of 2 × 10⁶ nonautologous irradiated (30 Gy) feeder cells and 0.01 μg/ml anti-CD3 (Beckman Coulter), in R10-P/S-IL-2. Half of the medium was replenished every 3 d with R10-P/S-IL-2. After 2 wk, autologous EBV-transformed B cell lines (BLCLs), suspended in R10 medium, were irradiated at 100 Gy and prepulsed with either no peptide, 15 μg/ml HCV NS3 peptide (NS3-1073-1081; CINGVCWTV), Flu peptide (NA-231-239; CVNGSCFTV), or CMV peptide (pp65-495; NLVPMVATV) for 1 h at 37°C. BLCLs were then washed and incubated with HCV-specific T cells at a ratio of 10:1 (T cells:BLCL) for 6 h in AIM-V medium (Life Technologies) supplemented with 10% human serum (Wisent) and anti-CD107a-BUV395 Ab (clone H4A3; BD Bioscience). After 1 h of stimulation, 10 μg/ml brefeldin A (Sigma) and 6 μg/ml monensin (Sigma) were added to each well. At the end of the stimulation, cells were washed using FACS buffer (PBS, 1% FCS), surface staining was performed using directly conjugated mAbs (CD3–Pacific Blue [clone UCHT1], CD4-BV605 [clone RPA-T4], and CD8-allophycocyanin-H7 [clone SK1]), and live cells were identified as described earlier. Cells were then permeabilized with CytoFix/CytoPerm (BD Biosciences) for 15 min at 4°C in the dark, washed using Perm/Wash buffer (BD Biosciences), and then incubated for 30 min at 4°C in the dark with anti-IFN-γ-PE-Cy7 (clone B27), anti-TNF-α-PerCP-Cy5.5 (clone MAb11), and anti-IL-2-allophycocyanin (clone MQ1-17H12) (all from BD Biosciences). Cells were then washed and fixed with 1% paraformaldehyde (Sigma-Aldrich) in PBS and analyzed by FACS as described earlier. Boolean gating and SPICE software were used for polyfunctionality analysis (26).

Raw data of all TCR sequences are available at https://clients.adaptivebiotech.com/pub/mazouz-2021-ji (DOI: 10.21417/SM2021JI).

We analyzed the TCR repertoire of FACS-sorted MHC class I tetramer-reactive CD8⁺ T cells specific for the immunodominant HLA-A*0201–restricted NS3-1073 epitope in HCV-infected subjects (n = 14). Our cohort included samples collected during acute HCV infection (≤168 d after EDI) from either SRs (SR-Acute; n = 4) or subjects whose infections became chronic (CI-Acute; n = 6), as well as samples at a follow-up time point (≥347 d after EDI) after HCV clearance in the SR group (SR-Follow-up; n = 8) (Table I). Four SR subjects had paired samples at both the acute and follow-up time points. Because the frequencies of NS3-1073–specific CD8⁺ T cells are dramatically reduced with progression toward chronic infection, we could not sort and analyze the repertoire of NS3-1073–specific CD8⁺ T cells in CI subjects at follow-up. The frequencies of HLA-A2/NS3-1073 tetramer⁺CD8⁺ T cells were not significantly different between groups (Fig. 1A, Supplemental Table I). In addition, we sorted and sequenced the TCR repertoire of total naive CD45RO⁻CD8⁺ T cells from two samples: SR.5-Acute and CI.5-Acute as controls for each group. TCR deep sequencing information, including number of sorted cells, productive total/unique sequences, and clonality, is presented in Supplemental Table I.

We compared the diversity of the repertoire in SR-Acute, SR-Follow-up, and CI-Acute samples using the Simpson diversity index, which integrates the number of clonotypes and their abundance (27) and ranges from 0 (high diversity, i.e., large repertoire) to 1 (low diversity, i.e., narrow repertoire) (Fig. 1B). We also examined the Shannon diversity, which integrates both the abundance and evenness (i.e., similarity of frequencies) of all clonotypes (28) and ranges from 0 (low diversity) to 1 (high diversity) (Fig. 1C). Both analyses demonstrated high diversity in all groups but no significant differences between groups (Fig. 1B and (1C). As shown by tree maps, the majority of samples displayed a broad TCR repertoire, with the top 10 sequences representing less than 25% of the total TCR repertoire in most samples (3/4 SR-Acute, 7/8 SR-Follow-up, 4/6 CI-Acute) (Fig. 1D). Only three samples (SR.1-Acute, SR.1-Follow-up, and CI.1-Acute) were more focused, with the top 10 sequences representing 55.8%, 43.5%, and 56.3% of their NS3-1073–specific TCR repertoire, respectively (Fig. 1D). Altogether, the accumulated frequencies of the top 10 TCR clonotypes showed no significant differences between the three groups (Fig. 1E).

Next, we assessed the CDR3 amino acid length distribution. The naive CD45RO⁻CD8⁺ TCR repertoires from SR.5-Acute and CI.5-Acute displayed a normal Gaussian distribution indicated by a coefficient of determination (R²) close to 1 (Supplemental Fig. 1A, right). In contrast, the HCV NS3-1073–specific TCR repertoires were more skewed (R² ranged from 0.6 to 0.96) in all samples, suggesting Ag-specific clonotypic expansion (Supplemental Fig. 1A, left). The CDR3 lengths were skewed, but no particular amino acid length was preferentially overrepresented (Supplemental Fig. 1A, left). There were no significant differences in the CDR3 amino acid length distribution between SR-Acute, SR-Follow-up, and CI-Acute (represented as R²; Supplemental Fig. 1B).

Altogether, these data suggest that SR-Acute, SR-Follow-up, and CI-Acute samples had similar highly diverse HCV NS3-1073–specific TCR repertoire.

To further characterize the HCV NS3-1073–specific TCR repertoire, we analyzed and compared the TCRBV gene usage between all groups. As we previously described (16), we stratified clonotypes at the nucleotide level into four distinct categories according to their abundance within the total repertoire analyzed in each subject: dominant (frequencies ≥ 1%), subdominant (≥0.5%), low-abundance (≥0.1%), and lowest-abundance clonotypes (<0.1%). Hereinafter, all HCV NS3-1073–specific TCR repertoire results presented will focus on dominant, subdominant, and low-abundance clonotypes only. The detailed raw TCR sequencing data for each subject are available online as described in Materials and Methods. Using a threshold frequency of ≥10%, we found that the most frequently used V gene families were V02–V07 and V09, irrespective of HCV infection outcome. (Fig. 2A shows V02–V09 families, while all families are shown in Supplemental Fig. 2A. V04 and V06 were the two V gene families preferentially used in all groups. Also, V04 was significantly more frequently used as compared with V06 (p ≤ 0.05) at the follow-up time point in SR (Fig. 2A). By examining paired naive and acute samples from SR.5 and CI.5, we observed that the preferential usage of V04 and V06 was not associated with their overrepresentation within the naive pool (Supplemental Fig. 2B). V02 was significantly enriched in CI-Acute as compared with SR-Follow-up (p ≤ 0.05), but this was mainly driven by one outlier. Similarly, V09 was significantly enriched in SR-Acute as compared with SR-Follow-up (p ≤ 0.01) and CI-Acute (p ≤ 0.001) samples because of another outlier.

Next, we examined whether preferential usage of these V genes was sustained after resolution of acute HCV infection. We compared the TCR repertoire of paired samples from four SR subjects during acute infection and at a follow-up time point at >49 wk after HCV infection (range, 367–503 d after EDI). The V gene family usage profiles remained comparable at follow-up (Fig. 2B). The V04 and V06 remained within the top four preferentially used families by all subjects (Fig. 2B–D). Altogether, these data demonstrate preferential V04 and V06 usage in all groups during acute infection. This preferential usage was sustained at follow-up in SR subjects.

Next, we performed an in-depth analysis of all CDR3 sequences within the repertoire to identify public TCR clonotypes (CDR3 amino acid sequence) that are present in multiple individuals [reviewed by Li et al. (29)]. First, we sought to identify the number of common identical CDR3 sequences in all potential pairwise combinations among dominant, subdominant, and low-abundance clonotypes (Fig. 3A). As expected, the highest number of shared clonotypes was between samples belonging to the same subject at two different time points (e.g., SR.1-Acute and SR.1-Follow-up). In contrast, CI.4-Acute did not share any clonotypes with SR.5-Acute, SR.5-Follow-up, and SR.8-Follow-up. All other pairwise comparisons revealed the presence of 1–11 identical CDR3 sequences (Fig. 3A). To refine our analyses, we designated a given clonotype as “public” if it was present in at least two subjects at a cumulative frequency of ≥0.5%, irrespective of infection outcome. Using these criteria, we identified a set of 13 public clonotypes (Fig. 3B, Supplemental Table II). These public clonotypes and their cumulative frequencies in each sample are summarized in Supplemental Table II. Six public clonotypes (1, 3, 4, 7, 10, and 13) used the V04 gene family. Two of them (public clonotypes 7 and 10) were highly related because they were both 13 aa long with only 1 aa difference (Supplemental Table II). These public clonotypes did not segregate samples by group, underscoring that their expansion was independent of infection outcome (Fig. 3B). One public TCR clonotype (CASSQEPGAPNTGELFF [V04-02/J02-02]), hereinafter termed public clonotype 1, was detected in most samples (12/14) (Fig. 3B). The cumulative frequencies of this public clonotype were not significantly different between all groups (Fig. 3C). Other public clonotypes were detected in fewer subjects and at lower frequencies. Public clonotypes 2, 7, and 10–12 were primarily detected in resolvers, while public clonotype 13 was detected only in subjects with CI. Public clonotype 1 was long-lived and detectable at the follow-up time point in subjects SR.1, SR.5, SR.7, and SR.8 at a frequency of 0.28–2.65% at 1–2 y postinfection (Fig. 3B, Supplemental Table II). Other public clonotypes that were present during the acute phase at a frequency of >1% were also maintained at a high frequency at the follow-up time point. Altogether, these results suggest that HCV infection can prime the expansion of several public clonotypes. Several of these public clonotypes, including the dominant public clonotype 1, were long-lived and detectable in resolvers after HCV clearance.

The TCR-α-chain is an important determinant of the functionality of a public clonotype. However, it is technically challenging to identify the TCRαβ paired sequences from bulk sequencing data. To gain an insight about the TCRαβ pairing of the identified public clonotypes, we examined publicly available data. Eltahla et al. (30) examined paired TCRαβ sequences via single-cell RNA sequencing of NS3-1073–specific CD8⁺ T cells from one subject who had spontaneously resolved primary HCV infection. Three of the public clonotypes we identified, including the dominant public clonotype 1 and public clonotypes 2 and 5, were also detected by Eltahla et al. (30), together with their corresponding TCRα sequences (Fig. 3D). These data confirm the presence of these public clonotypes in another unrelated study subject and potential paired TCRα sequences. However, additional studies are required to clone and analyze the impact of the different TCR-α-chain sequences on the functionality of these public clonotypes.

To assess whether the preferential expansion of certain public clonotypes was associated with a specific viral variant, we sequenced the region spanning the NS3-1073 epitope in autologous virus isolated from the plasma of subjects who had high viral loads that were high enough to sequence (n = 9; Table I). Compared with the epitope reference sequence (CINGVCWTV), we did not detect any sequence variation in the autologous virus in the CI group. However, we detected variant viral sequences in three of the SR-Acute samples (SR.5, SR.7, and SR.8). All three subjects exhibited an expansion of public clonotype 1. Furthermore, the viral variants were not the same in all subjects, suggesting that a particular HCV NS3-1073 viral sequence is not the main driver of expansion of this public clonotype, and that this public clonotype may have the flexibility to recognize multiple variants of the same epitope.

Next, we sought to examine whether the dominant public clonotypes we detected were enriched because of a pre-existing pool of cross-reactive memory CD8⁺ T cells before HCV infection. Zhang et al. (19) had previously reported the presence of naive and memory HCV NS3-1073–specific CD8⁺ T cells in HCV-seronegative subjects, suggesting that these T cells might, in part, be primed by previous heterologous infections. We searched for exact matches between the data from Zhang et al. (19) and ours. We identified four TCR clonotypes from our dataset that were also present in seronegative subjects. These four clonotypes contained only one public clonotype (public clonotype 4), while the other three clonotypes did not meet our public clonotype criteria (Table II). These results suggest that the public clonotypes detected in our study are unlikely to have originated from a memory pool of cross-reactive CD8⁺ T cells.

Next, we examined whether the public clonotypes that we identified and were long-lived would preferentially expand on HCV re-exposure and reinfection. We had previously published a longitudinal analysis of the HCV NS3-1073–specific TCR repertoires from two HCV resolvers with a documented HCV reinfection episode where one spontaneously cleared (SR/SR-1), while the other became persistently infected (SR/CI-2) (16). This divergent outcome was associated with different levels of expansion of HCV NS3-1073–specific CD8⁺ T cells in each subject (Fig. 4A). In subject SR/SR-1, NS3-1073–specific CD8⁺ T cells expanded from 0.46% to 3.37% during peak reinfection (3 wk after detection of reinfection) where ∼58% of tetramer⁺ cells exhibited a CD127⁻ effector phenotype. The frequency then stabilized at 0.57% at late reinfection (24 wk). In subject SR/CI-2, NS3-1073–specific CD8⁺ T cells were at 0.70% at pre-reinfection, decreased to 0.32% during peak reinfection (week 4), and then stabilized at 0.63% at late reinfection (week 24). Autologous virus sequencing in SR/CI-2 revealed a variant of the NS3-1073 epitope (Table I) that was not recognized by the pre-existing memory CD8⁺ T cells, which may explain the limited expansion of the NS3-1073 population in that subject (5).

Within the NS3-1073–specific FACS-sorted CD8⁺ T cells, public clonotype 1 was detected only in subject SR/CI-2. Although the overall tetramer frequency did not increase, this public clonotype expanded 2.49-fold at peak reinfection and remained within the dominant pool at late reinfection (Fig. 4A). Public clonotypes 2, 3, and 6 were present in both subjects. Public clonotype 2 was detectable at pre-reinfection at a frequency of 1.62% and 1.10% of the repertoire in SR/SR-1 and SR/CI-2, respectively. At peak reinfection, it expanded only in subject SR/SR-1, representing 3.85% of the effector (CD127⁻) and 0.90% of the memory (CD127⁺) CD8⁺ T cells repertoire. Public clonotype 3 slightly expanded (1.2-fold) in subject SR/CI-2. Public clonotype 6 was detectable at pre-reinfection at a higher frequency in SR/SR-1 (2.10%) as compared with SR/CI-2 (0.28%). It expanded to 2.25% of memory CD127⁺ T cells and to 1.53% of effector CD127⁻ T cells in SR/SR-1. It also expanded 1.57-fold in SR/CI-2. Public clonotypes 5 and 12 were detectable only in SR/CI-2 and were at low frequencies (≤0.21%) at either pre-reinfection or late reinfection, respectively, and therefore are unlikely to have played a key role during reinfection (Fig. 4A).

We had previously established T cell clones from SR/SR-1 (clones R1–R5) (16), so we examined them for the presence of the public clonotypes identified in this study. Two clones had public clonotypes: clone R2 carried public clonotype 2, and clone R4 carried a mix of public clonotypes 2 and 6 (Fig. 4B). These T cell clones also displayed high functional avidity (16) and polyfunctionality indices (Fig. 4C). In summary, these data suggest that pre-existing public clonotypes of NS3-1073–specific CD8⁺ T cells can expand on HCV re-exposure and reinfection, are polyfunctional, and may contribute to viral clearance.

Previous studies reported that HCV NS3-1073–specific CD8⁺ T cells can be cross-reactive and can recognize HLA-A2–restricted epitopes from other viruses, including influenza (NA-231), CMV (pp65), and EBV (LMP2) (17–19). However, these studies did not examine the TCR repertoires of cross-reactive T cells, and it is unknown whether the same clonotypes/public TCRs are expanded during these different viral infections. Thus, we sought to identify HCV NS3-1073–specific clonotypes that are common with T cell repertoires specific for other HLA-A2–restricted epitopes and that may be cross-reactive. Using the online VDJ database developed by Shugay et al. (23) and CDR3 sequences from a study of the TCR repertoire after YFV vaccination by Pogorelyy et al. (31), we searched for exact matches between TCR clonotypes identified in our study and known clonotypes specific to various HLA-A2–restricted epitopes. The resulting list of shared CDR3 amino acid sequences, their cumulative frequencies in each subject, and the cognate epitope(s) are presented in (Fig. 5A and Table III. Shared CDR3 amino acid sequences were found in three SR-Acute, six SR-Follow-up, and three CI-Acute samples. These shared clonotypes were specific to different Ags, including pp65 (CMV) (32), BMLF1 (EBV) (33), M1 (influenza) (32), NS4B (YFV) (34), and MLANA (Melanoma) (35) (Fig. 5A, Table III). However, these clonotypes represented only minor fractions of the repertoire of HCV NS3-1073–specific CD8⁺ T cells. The lowest frequency was 0.19% for YFV vaccine TCR in SR.7-Acute, whereas the highest was MLANA (Melanoma) at a frequency of 2.45% in CI.3-Acute (Fig. 5A, Table III).

The majority of the identified clonotypes were private (present in only one subject). Only one clonotype previously reported to be reactive to the M1 epitope (Influenza) (CASSQVQGTYEQYF) (32) was detected in three subjects, but the frequencies were less than 0.5% in two of these subjects, and therefore not meeting our criteria for designation as a public clonotype (Table III). Interestingly, none of the HCV NS3-1073 public clonotypes (Supplemental Table II) were shared with other HLA-A2–restricted Ags, suggesting that these public clonotypes are unique to HCV. Overall, despite the multiple shared clonotypes identified, they represented only a minor fraction of HCV NS3-1073–specific CD8⁺ T cell repertoires (generally <1%), and they were thus unlikely to have contributed to viral clearance.

Next, we assessed the in vitro cross-reactivity of HCV NS3-1073–specific CD8⁺ T cells with other viral Ags. We selected three resolver follow-up samples from subjects SR.1, SR.2, and SR.3 and one chronic acute (CI.3-Acute) sample because they expressed TCR clonotypes shared with other HLA-A2–restricted epitopes (Fig. 5A). SR.4-Follow-up sample was used as a control because it did not contain any shared clonotypes. We used a dose-response IFN-γ ELISPOT assay to compare directly ex vivo the reactivity of PBMC samples against the HCV NS3-1073 epitope and the other potentially cross-reactive epitopes based on the shared TCR repertoire of each subject as depicted in (Fig. 5A and Table III. We also examined the response against the Flu NA-231 epitope known to be cross-reactive with HCV NS3-1073–specific CD8⁺ T cells (17, 19). All samples tested showed a strong response to HCV NS3-1073 and limited responses to the other HLA-A2–restricted epitopes (Fig. 5B).

To test whether any responses were missed because of low frequencies of cross-reactive CD8⁺ T cells, we expanded NS3-1073–specific CD8⁺ T cells in vitro by stimulation with the cognate peptide for 14 d as described in Materials and Methods. We then tested the polyfunctionality of these expanded cells by intracellular cytokine staining after stimulation with NS3-1073, CMV (pp65), or Flu (NA-231) and evaluated the production of IFN-γ, TNF-α, IL-2, and the degranulation marker CD107a (Fig. 5C). We selected two samples, SR.2-Follow-up and SR.4-Follow-up, because they had high frequencies of NS3-1073 tetramer⁺CD8⁺ T cells (Supplemental Table I). They also exhibited different levels of V04 gene usage (21.57% versus 15.9%) and public clonotype 1 (6.86% versus 0.47%). Although SR.4-Acute expressed more IFN-γ at the individual cytokine level, SR.2-Acute displayed a higher frequency of CD8⁺ T cells producing more than one effector function (Fig. 5C). In contrast, very limited responses were detected after stimulation with pp65 (CMV) or NA-231 (Flu) where only low levels of CD107a were detected (Fig. 5C).

In summary, these results demonstrate that HCV NS3-1073–specific CD8⁺ T cells share a very limited number of clonotypes with other HLA-A2–restricted epitopes and consequently exhibit low cross-reactivity, suggesting that they were not primed by other pathogens before HCV infection.

We characterized and tracked directly ex vivo the TCR repertoire of the HCV-1073–specific CD8⁺ T cells during acute HCV in resolvers and subjects progressing to chronic infection, as well as at ∼1 y after HCV resolution in resolvers. TCR repertoires responding to the conserved NS3-1073 epitope were highly diverse with preferential usage of the TRBV04 and V06 families regardless of infection outcome. We identified 13 public clonotypes unique to HCV that were shared across several subjects. Several of these public clonotypes were long-lived in resolvers and re-expanded on reinfection. In addition, we identified a set of TCR clonotypes shared with other HLA-A2–restricted epitopes suggesting potential cross-reactivity. However, these clonotypes were of low frequency and demonstrated limited cross-reactivity in in vitro assays and are thus unlikely to have played a major role in determining the outcome of primary acute HCV infection.

We demonstrated that the HCV NS3-1073–specific TCRβ repertoire was diverse during early acute infection irrespective of infection outcome, and that this diversity persisted at long-term follow-up. Several studies demonstrated that a diverse CD8⁺ TCR repertoire is important in recognizing escape mutations in targeted epitopes that may arise during HIV or HCV infections (36–39). These studies suggested that having a more diverse repertoire early during acute infection allows refined selection of the most effective clonotype against the infecting virus sequence and potential escape mutations. In this study, we identified different viral variants of the HCV NS3-1073 epitope in resolvers. These variants are known to affect either HLA-A2 binding or TCR recognition (40, 41). The fact that subjects SR.5 and SR.8 spontaneously cleared their acute infection suggests that the diverse repertoire and the public TCRs identified in these subjects were flexible and were able to recognize both the reference sequence (used in the tetramers) and the autologous sequences effectively.

We identified a set of 13 NS3-1073–specific public clonotypes with different expansion profiles. This might be explained by the usage of a different α-chain leading to a TCRαβ clonotype of higher or lesser affinity to the cognate epitope. We could not examine the TCRαβ paired sequences in our study but were able to identify three of them from publicly available data (30). This approach remains very limited to one subject and three sequences. Additional studies using single-cell sequencing approaches coupled with cloning of the receptor(s) of interest and functional analysis would provide better insights about the role of the α-chain, the αβ pairing, and a more accurate measure of functional avidity.

NS3-1073–specific public clonotypes were maintained within the memory pool and re-expanded on reinfection. We had previously reported that the NS3-1073–specific TCR repertoire becomes focused on reinfection with preferential expansion of CD8⁺ T cell clonotypes of high functional avidity (16). So, it is also possible that the expanded public TCRs are those with the highest functional avidity. Indeed, data from individual clones generated from subject SR/SR-1 contained two of the public clonotypes identified in this study. These clonotypes were of high functional avidity and polyfunctionality indices. Additional testing using a broader panel of T cell clones will be required to validate that possibility.

The high prevalence of public clonotypes observed during acute HCV was also characterized by preferential usage of the TRBV04 gene family. One single TCR clonotype in particular (TRBV4-02/TRBJ02-02) was identified in 12 of 14 subjects. Other public clonotypes using the V04 gene family, specific to the CMV pp65 (42) and HIV p24 Gag-derived KK10 epitopes (3), were previously reported. The presence of such public clonotypes and their maintenance within the long-term memory pool were found to be shaped by convergent recombination (43). Indeed, our data demonstrate that combinatorial (Supplemental Table II) and/or junctional (data not shown) diversities contribute to generating the same amino acid sequence.

We identified CDR3 amino acid sequences shared with other HLA-A2–restricted epitopes. Wedemeyer et al. (17) were the first to characterize HCV NS3-1073–specific CD8⁺ T cells cross-reactivity to the NA-231 epitope from the influenza virus. Yet, it was later shown that this HCV NS3-1073/NA-231 cross-reactivity is of low affinity (44), and that their structural conformation and their specific TCR repertoires are distinct (45). This is in line with our results where no HCV NS3-1073/FLU NA-231 shared CDR3 amino acid sequences were identified. Zhang et al. (19) had tested the directionality of this cross-reactivity and showed that HCV NS3-1073 peptides can induce CMV (pp65-495), Flu (M1-58), and EBV (LMP2-426) CD8⁺ T cells expansion (i.e., “reverse” cross-reactivity) and not vice versa (19, 44). Altogether, our data and others suggest that cross-reactivity of HCV NS3-1073–specific CD8⁺ T cells with other epitopes is more likely after HCV exposure.

We identified shared CDR3 amino acid sequences with multiple HLA-A2–restricted epitopes. Yet, the majority of shared clonotypes were of low frequency. Although we identified public clonotype 4 that was also present in the seronegative cohort, none of the public clonotypes identified in our study were shared with other HLA-A2–restricted epitopes. Technical confounders, such as the use of different sequencing approaches and in vitro expansion before analysis, may have limited our capacity to identify exactly matched CDR3 sequences. Nevertheless, there was little to no functional cross-reactivity of HCV NS3-1073–specific T cells in response to any of the shared HLA-A2–restricted epitopes. Overall, shared clonotypes represented only a minor fraction of the HCV NS3-1073–specific TCR repertoire. Thus, even though a pool of cross-reactive memory CD8⁺ T cells may exist before HCV infection, they play a limited role during primary HCV infection but may contribute to HCV-related liver disease severity as previously observed (18, 46).

In conclusion, our results demonstrate preferential usage of TRBV04 and V06 gene families, as well as the expansion of public TCR clonotypes unique to HCV, irrespective of infection outcome or autologous virus sequence. Additional studies examining the functional avidity of such public clonotypes and characterization of the TCR repertoire of other immunodominant HCV epitopes are required. Our data contribute to publicly available TCR repertoire databases that can be used to predict specificities of expanded T cells in specific pathological conditions or infections (47) and improve the development of algorithms to identify HCV-specific TCR clonotypes and their relationship to other specificities and to track HCV-specific clonotypes during future clinical trials for HCV vaccines (31, 48, 49).

We thank all study subjects for participating in this study, and we thank Dr. Dominique Gauchat and Philippe St-Onge of the flow cytometry core of the CRCHUM for technical help with cell sorting experiments.

The authors have no financial conflicts of interest.