We sequenced the CDR3 hypervariable regions alpha and beta chains of HSV-2-reactive CD8 T-cells specific for the most immunodominant known HSV-2 epitope in humans, as defined by both population prevalence and percent tetramer positivity. Among the 68 clones studied, 50 were derived from single cells. This is the first in-depth examination of epitope-specific T-cell CDR3 repertoire to an alphaherpesvirus in it’s natural host. The CD8 response to HSV-2 strongly displays the public TCR phenomenon, as previously observed for CD8 responses to viruses in the betaherpesvirus (CMV) and gammaherpesvirus (EBV) families. Again similar to CMV and EBV,we observed that HSV-2-specific public TCRs are frequently also immunodominant within-subject. The host response to HSV-2 differs from responses to CMV and EBV, in that the peak and memory-phase overall CD8 magnitudes are much lower for HSV-2. In addition, the cell type HSV-2 uses for latent and lytic replication in vivo, neurons and epithelial cells, respectively, are not professional APC. In contrast, monocyte/macrophages and B-cells, cells that are chronically infected by CMV and EBV, respectively, are highly efficient APC. We therefore conclude that the chronicity of herpesvirus infection may be an important driver for selection towards public TCR sequences.
The CD8 response to HSV is functionally important both in humans and in animal models. Infiltration of HSV-2-specific cytotoxic T-cells into the skin is correlated temporally with clearance of infectious virus from recurrent genital HSV lesions (7
). HSV-2-specific CD8 T-cell also persist at the dermal-epidermal junction, in proximity to the sensory nerve endings that are the conduit for virus delivery during recurrences, for several weeks after resolution of recurrent genital herpes in humans (6
). In mice and humans, HSV-specific CD8 T-cells persistently infiltrate sensory ganglia that harbor infected neurons (2
). Ablation of local CD8 cells or interferon-gamma contribute to ganglionic reactivation in mice (32
). In the case of SIV infection, it has been observed that public TCR-using CD8 wells are correlated with protection after vaccination (34
). Vaccines delivering single HSV CD8 epitopes but neither CD4 nor antibody targets can protect mice from lethal HSV infection (35
), and CD8-based HSV vaccine strategies are currently being studied in humans (36
). The possibility that the targeting HSV-2 epitopes that can elicit public/immunodominant TCRs might be more effective than the targeting of epitopes that do not show these phenomena suggest that similar studies of other HSV-2 CD8 epitopes may be of interest.
In MHC H-2b
haplotype mice, the CD8 response largely focuses on a single epitope in HSV glycoprotein B. This response uses two dominant TCRBV genes, but fine clonotypes have not been investigated (37
). While murine neurons are chronically infected after HSV-1 infection, murine HSV-1 differ significantly from human HSV infection in that productive lytic recurrent infection capable of transmitting infection to the periphery is very rare (38
). The human CD8 response during chronic infection may be shaped by chronic antigenic (re)-stimulation in both skin and ganglia to a larger extent than in mice. The subjects studied in this report each had long-established histories of HSV-2 infection. It is now established, based on PCR-based genital shedding studies, that essentially all HSV-2-infected persons periodically shed virus from the periphery (39
). This implies re-exposure of the immune system to viral antigens on a chronic basis.
We observed a strong bias towards usage of specific TCRVA and TCRVB gene segments, the existence of several TCR single-chain and heterodimeric public amino acid sequences, and also TCR3 amino acid motifs in which distinct sequences shared conserved features. These findings are all interrelated. TCRVA19, TCRVA1-1, TCRVA4 together accounted for more than 50% of the observed TCRA chains. Similarly, TCRBV12-4, TCRBV9, and TCRBV5-1 comprised the majority of the TCRB chains. We detected three distinct public TCRA CDR3 sequences and three distinct public TCRB CDR3 sequences amongst a relatively small collection of subjects and T-cell clones.
Debate exists as to the relative contributions of TCRA or TCRB chains to peptide-MHC binding. This appears to vary for different TCR heterodimers and different MHC-peptide complexes, but in general, TCRB is thought to have more interactions with the C terminus of the peptide while TCRA may dominate interactions with the N terminus (40
). There is no crystal structure for HLA B*0702 complexed with peptide/TCR, and important differences in TCR interactions have been noted between different human HLA alleles (41
). Our data support a model in which the specific TCRVA1-1 CAVRDTNTNAGKSTF sequence form functional heterodimers that can bind B7-RPR in conjunction with quite diverse TCRB sequences using TCRBV 5-1, 6-1, 9, and 12-3. In this example, the TCRVA chain may thus be dominant and contribute more to the overall energetics of the interaction.
Occasional T-cell clones expressed two distinct mRNAs with protein-coding sequences without stop or frameshift mutations within the sequenced regions. Amongst the 50 clones directly derived from single cells in PBMC, 3 (6%) had this pattern. In addition, clones 5.L4, 5.H37, and 5.H48 each had the same two TCRA chains. These were derived after in vitro stimulation of PBMC with peptide, and likely represent progeny of a single cell present in PBMC which expressed both mRNAs. Other clones from this bulk expansion only seemed to express one of these two mRNAs. It is not know if expression of the other was lost, or was below the analytical threshold of detection.
We noted that the functional avidities of T-cell clones were generally quite similar, both within-subject and between subjects. The one exception was subject 5, whose T-cell clones had a functional avidity in cytotoxicity assays that was 2 to 3 orders of magnitude higher than for the other subjects. These clones had TCR sequences largely not seen in any other subject. As these clones were derived after in vitro
enrichment of peptide-specific clones, it is not known how representative they are of epitope-specific T-cells from this subject. Possibly, their higher avidity is partially a consequence of selective outgrowth during in vitro
expansion. Tetramer-based flow sorting was used for both direct PBMC- and culture-derived cloning, so this step is unlikely to have selected preferentially for high affinity clones. These exceptional high avidity clonotypes do, however, demonstrate in principle that highly avid HSV-2-specific T-cell clones can occur, and that comparisons of subjects with different phenotypes could disclose differences between groups. It is not known why the functional avidities we observed for IFN-γ release were generally ~100-fold lower than for cytotoxicity assays. In other T-cell systems, higher densities of TCR stimulation have been reported to be required to trigger interferon than cytotoxicity effector functions (42
). It is not known why there was a slight decrease in percent cytotoxicity at high concentrations (). If this was due to slight peptide toxicity for the LCL target cells, then actual cytotoxicity EC50
values might be left-shifted to higher peptide concentrations, more similar to those seen for the IFN-γ assays. We did not, however, observe higher spontaneous 51
Cr release for target cells incubated at 10(-6)
M peptide compared to target cells incubated with lower concentrations of peptide (data not shown). Our IFN-γ release assay also used a lower responder to stimulator ratio (2.4:1) than did the cytotoxicity assay format (20:1), possibly increasing the observed EC50
. For these reasons, direct quantitative comparisons of EC50
values in the two effector assays must be performed with caution.
Our work-flow required clonal outgrowth prior to assay. Some virus-specific T-cells may not make IFN-γ or proliferate well directly ex vivo
). This may correlate with effector vs. central memory status as measured by a variety of markers. Of note, TCR clonotypes within CMV- and EBV-specific CD8 cells with varying effector vs. central memory phenotypes were recently examined directly ex vivo
, and in general, every TCR clonotype was present within each phenotypic subset (16
). B7-RPR-specific CD8 T-cells also display varied memory markers (9
), and it is not yet known if TCR sharing is also true for HSV-2-specific memory CD8 T-cells with different memory phenotypes. In addition, our sample size was relatively limited and we studied a single HSV-2 epitope. T-cells were also made using three work-flows, and for these reasons our conclusions must be regarded as preliminary. Further research will be required to determine if our observations also hold true for population-prevalent responses such as HLA A*0201/HSV-2 UL47 551-559 (45
). Studies of additional epitopes await validation of additional HSV tetramers that are solidly “on-screen” for direct PBMC staining.
In conclusion, the CD8 response to an immunodominant epitope in HSV-2, a chronic human pathogen to which CD8 cells respond at neuronal and epithelial sites of infection, displays within-person dominance and remarkable conservation of dominant TCR clonotypes between individuals. Within a small set of 5 subjects, three separate public TCRA amino acid sequences, three separate TCRB amino acid sequences, and two distinct public TCRA/B combinations were detected. Chronic exposure to an alphaherpesvirus thus appears to be similar to chronic infection by beta- and gamma-herpesviruses and retroviruses, agents with very different cell tropisms and viral loads, and thus appears to be a common pathway during long-term exposure to foreign antigens.