RESULTS
We selected three epitope-specific populations of CD8+T cells from two individuals (CL-MCRL and CL-3089) with acutely
resolving HCV infection for this study. Both individuals were selected
from a cohort of high-risk injecting drug users followed prospectively
for recent HCV infection19,20. CL-MCRL experienced a
brief period of acute HCV infection (Figure 1A) that was bound by
negative HCV RNA PCR tests at 80 days before and 115 days after the
estimated date of infection21. Similarly, we
identified a single time point for individual CL-3089 at 180 days
post-infection (DPI) with a positive HCV RNA PCR test. Two peripheral
HCV-specific CD8+ T cell populations were identified
from CL-MCRL targeting the HLA-B*07:02 GPRLGVRAT (GPR) and
HLA-A*01:01 ATDALMTGF (ATD) epitopes whilst a single population
targeting the HLA-B*07:02 GPRLGVRAT epitope was identified from
CL-3089 (Table 1).
We stained for epitope-specific populations using MHC-I Dextramers and
flow cytometry (Figure 1B) and tracked the stable persistence of the
GPR-specific population from CL-MCRL over at least two years (Figure
1C). We applied TCR reconstruction from single-cell RNA-sequencing using
the VDJPuzzle pipeline22 and found that the
GPR-specific population from CL-MCRL consisted almost completely of a
monoclonal population of CD8+ T cells expressing an
identical paired CDR3αβ clonotype across all four sampled time points
(Figure 1D). More specifically, this monoclonal clonotype (TCRα:
CAVRATGQNFVF, TCRβ: CASSQAPPGQGVDIQYF) accounted for 171/174 of all
reconstructed paired CDR3αβ clonotypes from the GPR-specific population
from CL-MCRL. By comparison, the ATD-specific and GPR-specific
CD8+ T cell populations from individuals CL-MCRL and
CL-3089, respectively, were highly diverse and polyclonal (Supplementary
Figure 1). Notably, we did not observe the monoclonal GPR clonotype from
CL-MCRL in the repertoire derived from CL-3089. Taken together, these
observations point to the expansion of a naturally-occurring monoclonal
population following natural HCV infection.
We next explored whether the recruitment and expansion of a monoclonal
population may have been driven by the affinity characteristics of the
monoclonal TCR for its cognate peptide target. To this end, we employed
the use of reversible MHC Streptamers14 to measure the
dissociation time of the TCR-pMHC interaction and determine its
koff affinity (Figure 1E). Briefly, reversible
Streptamers were formed from conjugating fluorescently labelled pMHC
monomers with a Strep-Tactin backbone via Strep-tag binding sites, and
the addition of biotin dissociates the backbone by competitive binding
and leaves monomeric TCR-pMHC complexes. The dissociation speed of
monomeric TCR-pMHC complexes can subsequently be measured and
corresponds to the koff rate. We applied this assay
directly to cryopreserved peripheral blood mononuclear cells (PBMC) from
CL-MCRL by first co-staining with reversible Streptamers and
non-reversible Dextramers for the GPR-epitope (Figure 1F). Following the
addition of biotin, we observed rapid loss of the backbone signal
(half-life: 26 seconds) and gradual loss of pMHC signal (Figure 1G). We
periodically paused acquisition to preserve sample during the extended
dissociation between the monoclonal TCR population GPR-epitope presented
on pMHC complexes, and after analysis determined the dissociation to
have a half-life of 794 seconds and a koff constant of
1.3 x 10-3 (95% confidence interval: 1.5 x
10-3 to 1.1 x 10-3).
We were unable to directly measure the koff affinities
of the polyclonal populations from CL-MCRL and CL-3089 because of the
breadth of unique TCR arrangements, and thus we applied an in
vitro colony expansion approach to generate pure, monoclonal
populations of epitope-specific CD8+ cells from
single-cell sorted cells. Single clones from the GPR- and ATD-specific
populations from CL-MCRL and CL-3089 respectively, were identified by
staining with non-reversible Dextramers, isolated by
fluorescence-activated cell sorting, and expanded by stimulation with
phytohemagglutinin (PHA) and IL-2, in the presence of gamma irradiated
feeder cells for 4 weeks (Table 1). We also applied this approach to the
monoclonal GPR-specific population from CL-MCRL and successful
expansions typically yielded up to one million CD8+ T
cells. We next repeated the dissociation assay using these colony
expanded cells by first re-identifying epitope-specific cells by
non-reversible multimer staining (Figure 2A and Methods) followed by
measurement of reversible Streptamer dissociation comprising of the
backbone (Figure 2B) and pMHC complex (Figure 2C).
Notably, the dissociation half-life varied in duration between colonies,
epitope specificity, and individual origin. The strongest
koff affinity was recorded from CL-MCRL GPR-specific
colonies (half-lives 1039 to 1321 seconds), followed by CL-3089
GPR-specific colonies (half-lives 247 to 1191 seconds), and the weakest
koff affinities were recorded from CL-MCRL ATD-specific
colonies (half-lives 22 to 196 seconds) (Figure 2D). More specifically,
the koff constants for the CL-MCRL ATD-specific colonies
ranged from 3.54 x 10-3 to 3.14 x
10-2 (Table 2), the koff constants for
the CL-MCRL GPR-specific colonies ranged from 5.25 x
10-4 to 6.67 x 10-4 (Table 3), and
the koff constants for the CL-3089 GPR-specific colonies
ranged from 5.82 x 10-4 to 2.81 x
10‑3 (Table 3). Intriguingly for one pair of colonies
(F4 and F4-2) derived from the polyclonal GPR-specific population, their
dissociation half-lives were within the range of the monoclonal
clonotype (Table 3) which suggested that it may be possible to identify
similarly high affinity TCRs within polyclonal populations.
Our sorting strategy for colony expansion resulted in the random
selection of clonotypes and thus we performed targeted TCR sequencing to
link clonotype identity with measured koff affinities.
Total RNA was extracted from a subset of samples and the CDR3 regions
from TCRα and TCRβ chain transcripts were enriched by nested PCR
followed by Sanger sequencing. All ATD-specific colonies had distinct
clonotypes (Table 2), which was consistent with the polyclonal nature of
the original repertoire (Supplementary Figure 1). Notably, around half
of all clonotypes from the colony expansions had also been observed in
the scRNA-seq dataset (Tables 2 and 3).
Finally, we considered whether, at the population level, transcriptional
phenotypes may be associated with differences in the distribution of
koff affinities. We first performed a differential gene
expression analysis between the GPR- and ATD-specific populations from
the early time points (<120 days post-infection) in CL-MCRL
and identified enrichment for effector-associated signatures
(NKG7 , GZMH , GZMA ) in the former and a
memory-associated signature (IL7R ) in the latter (Figure 3A).
Next, we compared the GPR-specific monoclonal population from CL-MCRL
and the polyclonal one from CL-3089, but identified differences
primarily related to TCR gene usage and mitochondrial gene expression
which may reflect inter-individual differences (Figure 3B). Rather, the
GPR population from CL-3089 resembled an intermediate population between
the ATD- and GPR-specific populations when characterised by expression
for the phenotypic markers NKG7 , GZMH , and IL7R(Figure 3C). We further applied gene set enrichment analysis using
Hallmark pathways from the Molecular Signatures database which confirmed
that the GPR-specific population exhibited an effector-polarised
phenotype with evidence for elevated cell cycling and cellular
metabolism pathways in its enriched genes (Figure 3D and E). Taken
together, our findings suggest an association between affinity and
transcriptional phenotype at the population level whereby higher
affinity responses can also be observed in combination with highly
active effector phenotypes.