Comprehensive Analysis of CD8⁺-T-Cell Responses against Hepatitis C Virus Reveals Multiple Unpredicted Specificities

Abstract

The hepatitis C virus (HCV)-specific CD8⁺-T-cell response is thought to play a critical role in HCV infection. Studies of these responses have largely relied on the analysis of a small number of previously described or predicted HCV epitopes, mostly restricted by HLA A2. In order to determine the actual breadth and magnitude of CD8⁺-T-cell responses in the context of diverse HLA class I alleles, we performed a comprehensive analysis of responses to all expressed HCV proteins. By using a panel of 301 overlapping peptides, we analyzed peripheral blood mononuclear cells (PBMC) from a cohort of 14 anti-HCV-positive, HLA A2-positive individuals in an enzyme-linked immunospot assay. Only four subjects had detectable HLA A2-restricted responses in PBMC, and only 3 of 19 predicted A2 epitopes were targeted, all of which were confirmed by tetramer analysis. In contrast, 9 of 14 persons showed responses with more comprehensive analyses, with many responses directed against previously unreported epitopes. These results indicate that circulating HCV-specific CD8⁺-T-cell responses can be detected in PBMC in the majority of infected persons and that these responses are heterogeneous with no immunodominant epitopes consistently recognized. Since responses to epitopes restricted by single HLA alleles such as HLA A2 do not predict the overall response in an individual, more comprehensive approaches, as shown here, should facilitate definition of the role of the CD8⁺-T-cell response in HCV infection. Moreover, the low level or absence of responses to many predicted epitopes provides a rationale for immunotherapeutic interventions to broaden cytotoxic-T-lymphocyte recognition.

The CD8⁺ T-cell response is thought to play a crucial role in the course of hepatitis C virus (HCV) infection. In humans and in chimpanzees, a strong and broadly directed HCV-specific cytoxoxic-T-lymphocyte (CTL) response has been associated with viral clearance during acute HCV infection (7, 29). In contrast, individuals with chronic infection are often found to have a relatively weak and narrowly directed CD8⁺-T-cell response against HCV (reviewed in reference 27). Whether these responses in chronic disease are still beneficial in containing viral replication or whether they are mediators of hepatic injury and disease is unclear. In several studies, the magnitude of the HCV-specific CD8⁺-T-cell response has been correlated to HCV viral load and to liver histology, but the results of these studies have been mixed (17, 31, 33, 34, 41). Overall, the levels of responses detected by most investigators, using a variety of methods, have been fairly low compared to those found in many other viral infections.

Thus far, most studies of the HCV-specific CD8⁺-T-cell response have been undertaken in HLA A2-positive individuals, and the analysis usually has been restricted to a set of previously described HLA A2-restricted HCV epitopes or to HCV peptides shown to bind strongly to HLA A2 and therefore predicted to elicit responses (2, 4, 5, 14, 15, 28, 30, 33, 34, 37-39). However, HLA A2 represents just one of up to six different class I alleles expressed in a given individual. Studies in human immunodeficiency virus type 1 (HIV-1) infection have shown that, although HLA A2-restricted CTLs are present in >70% of persons with chronic HIV infection (8, 12), HLA A2 is rarely the dominant allele mediating recognition of viral epitopes by CTLs (10). A contribution of non-A2 alleles in mediating CD8 T-cell responses in HCV infection has been shown in some studies (21, 22, 24, 41, 42), but the relative contribution of HLA A2 and other alleles and the overall strength of such responses have not been systematically studied. Also, the relative immunogenicity of HCV proteins has not been addressed in a comprehensive fashion.

To test our hypothesis that HCV-specific CTL responses against selected HLA A2-restricted epitopes are not indicative of the total response, we examined the breadth and specificity of CD8⁺-T-cell responses by using a comprehensive analysis of all viral proteins. Using an enzyme-linked immunospot (Elispot) assay to determine the recognition of a panel of peptides spanning all expressed HCV proteins, we addressed the relative contribution of HLA and non-HLA A2 alleles in presenting viral proteins for recognition by CD8⁺ T cells and addressed the overall magnitude and breadth of responses in the peripheral blood in HCV-infected individuals.

MATERIALS AND METHODS

Study subjects.

A total of 14 anti-HCV-positive individuals were selected for study based on expression of the HLA A2 allele (Table 1). One additional subject (P0) had been previously studied and was shown to have CTL responses by limiting dilution cloning (29). This person had an acute HCV infection in 1998, followed by spontaneous clearance of the virus, and contemporary samples were used for comparative assessment of immune responses with overlapping peptides. The study was approved by the Massachusetts General Hospital Institutional Review Board. All subjects gave written informed consent.

TABLE 1.

Subject characteristics

Subject	Disease	Viral load or quantity of HCV	HCV GTd	HLAa
P0	Resolved HCV	Negb	NDc	A2,25; B37,44; Cw5,6
P1	Acute HCV	>1,000,000	1b	A1,2; B8,44; Cw5,7
P2	Acute HCV	599,000	1b	A1,2; B40,57; Cw7,15
P3	Resolved HCV	Neg	ND	A2,29; B44; Cw16
P4	Resolved HCV	Neg	ND	A2,3; B7,60; Cw7
P5	Chronic HCV	357,169	2	A2, 11; B55,62; Cw1,9
P6	Chronic HCV	7,718	2a/c	A1,2; B37,44; Cw5,6
P7	Chronic HCV	67,051	1a	A1,2; B8,55; Cw3,7
P8	Chronic HCV	456,490	2b	A2,29; B18,40; Cw3,5
P9	Chronic HCV	33,637	2b	A2,3; B39,44; Cw7,16
P10	Chronic HCV	506,420	1b	A2,32; B40; Cw3
P11	Chronic HCV	1,213,300	3a	A2,29; B44, 50; Cw4,16
P12	Chronic HCV	996,000	1a	A1,2; B8,44; Cw1
P13	Chronic HCV	>1,000,000	1a	A2,24; B8,41; Cw7,17
P14	Chronic HCV	159,320	2b	A1,2; B7,8; Cw7

^aAlleles for which optimal epitopes were tested are in boldface.

^bNeg, negative.

^cND, not determined.

^dGT, genotype.

HLA typing.

HLA typing was performed by the Massachusetts General Hospital Tissue Typing Laboratory using standard serological and molecular techniques.

Synthetic peptides.

The peptide sequences correspond to the HCV 1a subtype (6). All peptides were synthesized as COOH-terminal free acids on a Synergy 432A Peptide Synthesizer (Applied Biosystems, Foster City, Calif.). Peptides were 20 amino acids (aa) in length, overlapping adjacent peptides by 10 aa. Fine mapping was achieved by using additional smaller peptides. All peptides were reconstituted in sterile RPMI 1640 medium containing 10% dimethyl sulfoxide (Sigma Chemical Co.) and 1 mM dithiothreitol (Sigma Chemical Co.).

Elispot assay.

Polyvinylidene plates (96-well; Millipore) were coated with 2.5 μg of recombinant human anti-gamma interferon (IFN-γ) antibody (Endogen)/ml in phosphate-buffered saline at 4°C overnight. Fresh or previously frozen peripheral blood mononuclear cells (PBMC) were added at 200,000 cells/well in 140 μl of R10 medium (RPMI 1640 [Sigma-Aldrich], 10% fetal calf serum [Sigma-Aldrich], and 10 mM HEPES buffer [Sigma-Aldrich] with 2 mM glutamine and antibiotics [50 U of penicillin-streptomycin/ml]). Peptides were added directly to the wells at a final concentration of 10 μg/ml. The plates were incubated for 18 h at 37°C in 5% CO₂. Plates were then washed, labeled with 0.25 μg of biotin-labeled anti-IFN-γ (Endogen)/ml, and developed by incubation with streptavidin-alkaline phophatase (Bio-Rad), followed by incubation with 5-bromo-4-chloro-indolylphosphate (BCIP)-nitroblue tetrazolium (NBT) (Bio-Rad) in Tris buffer (pH 9.5). The reaction was stopped by washing with tap water, and the plates were dried overnight, prior to counting on an Elispot reader (AID, Strassberg, Germany). The background was always fewer than 15 spot-forming cells (SFC)/10⁶ PBMC. Responses were considered positive if the number of spots per well minus the background was at least 25 SFC/10⁶ PBMC; phytohemagglutinin served as a positive control.

HLA class I peptide tetramer staining.

HLA class I peptide tetramers were prepared as previously described (29) and included tetramers specific for six epitopes restricted by HLA A2 (core peptide 35-44, YLLPRRGPRL; NS3 peptide 1073-1081, CINGVWCTV; NS4 peptide 1406-1415, KLVALGINAV; NS4 peptide 1807-1816, LLFNILGGWV; NS4 peptide 1851-1859, ILAGYGAGV; and NS5B peptide 2594-2603, ALYDVVTKL) (13, 28, 29). A total of 0.5 to 1 million PBMC were stained as described earlier (13). Flow cytometric analysis was performed with a Becton Dickinson FACSCalibur fluorescence-activated cell sorter, and data analysis was performed with CellQuest software. Staining was considered positive if tetramer-positive cells formed a cluster distinct from the tetramer-negative cells and the frequency of tetramer-positive cells was >0.02% of the total CD8⁺ population.

Intracellular IFN-γ staining.

Intracellular cytokine staining (ICS) for IFN-γ was performed as described previously (1). Briefly, 10⁶ PBMC were incubated with 4 μM peptide and anti-CD28 and anti-CD49d monoclonal antibodies (1 μg/ml each; Becton Dickinson) at 37°C and 5% CO₂ for 1 h before the addition of brefeldin A (1 μl/ml; Sigma-Aldrich). The cells were incubated for an additional 5 h at 37°C in 5% CO₂. PBMC were then washed and stained with surface antibodies, allophycocyanin-conjugated anti-CD8, and phycoerythrin-conjugated anti-CD3 (Becton Dickinson) at room temperature for 20 min. After the washing, the PBMC were fixed and permeabilized (Caltag, Burlingame, Calif.), and the fluorescein isothiocyanate-conjugated anti-IFN-γ monoclonal antibody (Becton Dickinson) was added. Cells were then washed and analyzed on a FACSCalibur flow cytometer, using CellQuest software (Becton Dickinson). For HLA restriction, partially matched heterologous B-cell lines (BCL) were pulsed with 10 μg of peptide for an hour, washed three times with R10, and then 2 × 10⁵ of the BCL were added to the T-cell line instead of peptide.

Bulk stimulation of PBMC.

For establishing CTL lines, cryopreserved or fresh PBMC (4 × 10⁶ to 10 × 10⁶) were stimulated with 1 μg of synthetic HCV peptide/ml and 0.5 μg of the costimulatory antibodies anti-CD28 and anti-CD49d (Becton Dickinson)/ml in R-10. Irradiated feeder cells (2 × 10⁷ allogeneic PBMC) were added to the culture in a 25-cm² culture flask (Costar, Cambridge, Mass.). Recombinant interleukin-2 (25 U/ml) was added on day 2 and twice a week thereafter.

Expansion of peptide-specific cells for tetramer staining was done by culturing 5 × 10⁶ PBMC pulsed with 10 μg of each of four peptides for 8 or 9 days in R-10 containing 25 U of interleukin-2/ml.

Cytotoxicity assay.

Autologous BCL were pulsed with 10 μg of peptide and Na₂[⁵¹Cr]O₄ (New England Nuclear, Boston, Mass.) and incubated for 1 h at 37°C in 5% CO₂. The BCL target cells were washed three times with cold R-10 medium and then incubated with effector cells at 37°C for 4 h in three replicate wells. Cellular release of [⁵¹Cr]O₄ into the supernatant was measured by using a Top Count Microplate scintillation counter (Packard Instrument Company, Meridien, Conn.), and the percent specific cytotoxicity was calculated by the following formula: percent lysis = [(experimental release − spontaneous release)/(maximum release − spontaneous release)] × 100. Results are reported as the means of triplicate values.

Statistical analysis.

Statistical analysis (Mann-Whitney rank sum test and correlation coefficient) was done by using GraphPad Prism 3.0a for Macintosh.

RESULTS

A minority of HLA A2-positive individuals has detectable HCV-specific CD8⁺ T-cell responses in PBMC toward previously described HLA A2 epitopes.

We first examined a cohort of 14 anti-HCV-positive individuals who expressed the class I allele A2 in order to determine the relative frequency of responses to reported and predicted epitopes restricted through this common HLA allele. Initially, we used 19 peptides representing all previously reported HCV-specific, HLA A2-restricted epitopes (Table 2). With this approach we clearly demonstrated responses in PBMC, but these were detected in only 4 of the 14 individuals (Fig. 1A). In the four individuals, between one and three of the epitopes were recognized, and the combined magnitude of these responses ranged between 25 and 285 SFC/10⁶ PBMC per individual (Fig. 1A). It is also noteworthy that all HLA A2-restricted responses were directed against just three different epitopes, whereas none of an additional 16 HLA A2 epitopes was recognized by any individual (Fig. 1).

TABLE 2.

Tested epitopes

HLAa	HCV protein	aa	Sequence	Reference(s)
A2	Core	35-44	YLLPRRGPRL	2, 4
A2	Core	132-140	DLMGYIPLV	4, 35, 36
A2	Core	178-187	LLALLSCLTV	2, 4
A2	E1	220-227	ILHTPGCV	35
A2	E1	257-266	QLRRHIDLLV	34
A2	E1	363-371	SMVGNWAKV	35
A2	E2	401-411	SLLAPGAKQNV	35
A2*	NS3	1073-1081	CINGVCWTV	4, 22, 29, 41, 42
A2	NS3	1169-1177	LLCPAGHAV	4
A2	NS3	1287-1296	TGAPVTYSTY	25
A2*	NS3	1406-1415	KLVALGINAV	4
A2	NS4B	1789-1797	SLMAFTAAV	4
A2	NS4B	1807-1816	LLFNILGGWV	2, 4
A2	NS4B	1851-1859	ILAGYGAGV	2
A2*	NS5A	2221-2231	SPDAELIEANL	42
A2	NS5A	2252-2260	ILDSFDPLV	4
A2	NS5B	2577-2586	RLIVFPDLGV	2
A2*	NS5B	2594-2602	ALYDVVTKL	29
A2	NS5B	2727-2735	GLQDCTMLV	2
A3*	NS5B	2510-2518	SLTPPHSAK	41, 42
A3*	NS5B	2588-2596	RVCEKMALY	22, 41
A11*	Core	1-9	MSTNPKPQK	21, 41
A11*	E2	621-628	TINYTIFK	22, 41, 42
A11*	NS3	1261-1270	TLGFGAYMSK	41, 42
A11*	NS3	1636-1643	TLTHPVTK	41, 42
A24*	NS3	1031-1039	AYSQQTRGL	26
A29	NS2	827-834	MALTLSPY	23, 41
B7*	Core	41-49	GPRLGVRAT	21, 22, 41
B7	Core	110-118	DPRRRSRNL	16
B8*	NS3	1395-1403	HSKKKCDEL	22, 41, 42
B8*	NS3	1611-1618	LIRLKPTL	42
B37*	NS2	957-964	RDWAHNGL	11
B37*	NS4B	1966-1976	SECTTPRCSGSW	29
B44	Core	88-96	NEGCGWAGW	19, 20
B50*	E2	569-578	CVIGGAGNNT	41
B57*	NS5B	2629-2637	KSKKTPMGF	41
B60*	Core	28-37	GQIVGGVYLL	18
B60*	E2	530-539	GENDTDVFVL	42
B60*	E2	654-662	LEDRDRSEL	42
B60*	NS5A	2152-2160	HEYPVGSQL	42
B60*	NS5A	2267-2275	REISVPAEIL	41

^aEpitopes for which the HLA restriction and the optimal peptides have been defined by using T-cell clones (by testing partially HLA-matched BCL and shorter and longer peptides in serial dilution) are indicated with an asterisk.

FIG.1.

Analysis of the HCV-specific CD8⁺ response against HLA A2 epitopes. (A) Magnitude and breadth of the HCV-specific CD8⁺-T-cell responses as detected with 19 different previously described HLA A2-restricted peptides in an Elispot assay. Individual epitopes tested are indicated by the numbering of the first amino acids in the HCV 1a sequence, and magnitudes are shown as IFN-γ SFC/10⁶ PBMC. Responses were detected only against three of the 19 peptides tested. (B and C) These results were confirmed by tetramer analysis as shown for subject P6 (B), and the results from the Elispot and tetramer assays correlated significantly (C).

One possible reason for the lack of detection of HLA A2-restricted CD8⁺-T-cell responses could be cells deficient in IFN-γ expression, as previously described (13, 29). In order to assess whether antigen-specific CD8⁺ T cells were present but not secreting IFN-γ, we analyzed all individuals by direct visualization of HCV-specific cells with six different HLA A2 tetramers (Fig. 1B and data not shown). Although the percentage of positive cells was found to be higher by tetramer analysis than by Elispot assay, all responses detected were confirmed, and no additional responses were seen by tetramer analysis. Furthermore, Elispot and tetramer analyses also correlated well in terms of the magnitude of the responses (Fig. 1C, R = 0.91, P < 0.001).

Detection of non-HLA A2-restricted CTL responses in HLA A2-positive persons.

The above data indicate that the minority of HCV-seropositive persons have detectable CD8 T-cell responses restricted by HLA A2. In order to determine whether CD8 T-cell responses occurred in response to epitopes other than those predicted to be presented by HLA A2, we tested also for responses to 22 previously defined non-HLA A2 epitopes that have been reported (Table 2). Between zero and nine (median, three) optimal HCV-specific CTL epitopes restricted by HLA alleles other than A2 were tested per person, depending on the HLA type (Table 3). Each epitope was tested in between one and six subjects, depending on the prevalence of the respective HLA allele in the cohort. Only three subjects had a significant response, each to just one of the peptides (Table 3). It is noteworthy that two of the three individuals with a readily detectable response against one of these non-HLA A2 epitopes had tested negative for all 19 HLA A2 peptides. The overall predictive ability of an HLA allele for specific responses was low, as had been the case for HLA A2-restricted epitopes. For example, of six persons expressing HLA B44, only one targeted a previously described B44 epitope (17). Likewise, none of five HLA B8-positive persons targeted either of two previously described HLA B8-restricted epitopes (22, 41, 42), although they recognized other epitopes. These results demonstrate that circulating HCV-specific CD8⁺-T-cell responses are detectable by Elispot assay in PBMC but are poorly predicted simply through the HLA type of the host.

TABLE 3.

Previously described epitopes (non-HLA A2) targeted by circulating CD8⁺ T cells

Subject	Peptides tested (n)	Peptides recognized
P1	Core 88, NS3 1395, NS3 1611 (3)	None
P2	NS5B 2629 (1)	None
P3	Core 88, NS2 827 (2)	Core 88
P4	Core 28, core 41, core 110, E2 530, E2 654, NS5A 2152, NS5A 2267, NS5B 2510, NS5B 2588 (9)	Core 41
P5	Core 1, E2 621, NS3 1261, NS3 1636 (4)	None
P6	Core 88, NS2 957, NS4B 1966 (3)	None
P7	None (0)	None
P8	NS2 827 (1)	None
P9	Core 88, NS5B 2510, NS5B 2588 (3)	None
P10	None (0)	None
P11	Core 88, E2 569, NS2 827 (3)	None
P12	Core 88, NS3 1395, NS3 1611 (3)	None
P13	NS3 1031, NS3 1395, NS3 1611 (3)	None
P14	Core 41, core 110, NS3 1395, NS3 1611 (4)	NS3 1395

Use of overlapping peptide libraries to confirm and extend previously identified HCV-specific CD8 T-cell responses.

The above studies indicate that CD8⁺-T-cell assays based on predicted or previously reported epitopes detect responses in peripheral blood in only a minority of HCV-infected persons. We had previously reported more broadly directed responses with a laborious CTL cloning method to identify novel epitopes (29, 42). In order to more comprehensively and directly assess responses, we developed a screening Elispot assay to examine responses against all expressed proteins that was independent of previous epitope prediction models. An Elispot assay was devised by using 301 peptides, each 20 aa in length and overlapping adjacent peptides by 10 aa, thus covering the entire 3,011-aa HCV 1a polyprotein. In order to facilitate the analysis, peptides were combined in pools of 10 peptides, with each peptide being contained in two different pools in a matrix setup. This allowed for internal controls, since every peptide was present in two different wells, and with 60 wells containing peptide, there were overall many wells not containing a recognized peptide serving as standards for a negative response in the presence of peptide.

To test the feasibility of screening with the peptide matrix, we first analyzed a subject (designated P0) who had been extensively studied previously by limiting dilution assay during resolution of acute infection and in whom a broadly directed CD8⁺-T-cell response had been characterized (29). When we analyzed PBMC from this individual more than 2 years after initial infection by using the panel of overlapping peptides, we were able to detect responses to eight 20mer peptides which contained seven of the eight optimal epitopes identified previously through limiting dilution cloning (Fig. 2). The magnitude of the responses was lower compared to the time point very early in acute infection, when the CTL clones had been established (29), but only the response that was lowest in acute infection was no longer detected. Comparison of responses to the 20mer peptides containing the targeted epitopes to the defined optimal epitopes did demonstrate greater responses for the optimal epitopes. However, for the one response that was no longer detectable, even the optimal peptide no longer elicited a response, suggesting the loss of this weak response over time (peptides 2221-2240 [20mer] and 2225-2233 [optimal 9mer], respectively, Fig. 2). Another previously described response (NS4B 1744-1754), which was never detected by Elispot assay in the acute or early phase of infection in this subject, was now detected by using the overlapping 20mer peptides and confirmed by using the optimal peptide (Fig. 2). Finally, we detected a response to an epitope previously not detected in this individual that was contained in peptide 951-970. At the time point studied, this epitope elicited the strongest response of all of the nine epitopes. Whether this response was originally not detected because of a lower frequency or because of the different detection techniques used is unclear. That this peptide was indeed recognized by CD8⁺ T cells was shown by ICS of PBMC (Fig. 3A) and the HLA restriction was demonstrated by establishing a T-cell line (Fig. 3B). Subsequently, truncated peptides were used to determine the minimal peptide required for recognition (Fig. 3C). The HLA restriction and the definition of the minimal peptide revealed that this previously undetected response was directed against an epitope also described by Giuggio et al. (11). Finally, these cells were not only able to secrete IFN-γ upon specific stimulation but were true CTLs able to kill peptide-pulsed autologous B cells in a standard chromium release assay (Fig. 3D). These data demonstrate the feasibility of screening for HCV-specific epitopes by an Elispot assay with overlapping peptides and that these responses represent CTLs. Of note, they also show the persistence of polyclonal CTL responses against HCV in a person (P0) with resolved HCV infection.

FIG. 2.

A screening Elispot with overlapping 20mer peptides can detect HCV-specific epitopes. We first tested the screening Elispot assay in an individual who had been studied previously by limiting dilution cloning. During acute infection eight different HCV-specific epitopes had been defined from that patient (29). Responses to seven of the eight epitopes were still detectable 2 years after the individual had cleared HCV infection (black bars). When compared to responses to the corresponding 20mer peptides used in the screening Elispot assay (gray bars), all seven responses were also detected. However, some of the 20mer peptides elicited responses that were almost as strong as the optimal epitope, whereas other responses were less vigorous. In addition to the previously described epitopes, a response to an additional peptide was identified (p951-970, striped bar). At the time point tested, this peptide elicited the strongest response of all recognized peptides.

FIG. 3.

Peptides detected in the screening Elispot represent CTL epitopes. (A) That stimulation with peptide 951-970 was inducing IFN-γ secretion by CD8⁺ T cells was shown by ICS with PBMC. (B) Use of a peptide-specific T-cell line and partially HLA-matched B-cell lines in the ICS demonstrated that the peptide was presented by HLA B37. (C) The same T-cell line was used to define the minimal epitope in an Elispot with serial dilutions of peptide truncations of the original peptide. (D) The T-cell line specific for peptide 951-970 was also capable of mediating specific cytotoxic activity against autologous peptide pulsed B cells.

The strength and breadth of the total HCV-specific CD8⁺ T-cell responses is not predicted by the responses against a limited number of epitopes.

In order to compare the total response with the response against the predicted epitopes, we applied the screening Elispot to the cohort of 14 individuals, resulting in the identification of CTL responses in 9 of 14 persons tested. This included not only detection of previously described HLA A2 and non-A2 epitopes (Fig. 4, white and black bars, respectively) but also a total of nine novel responses in seven individuals (Fig. 4, gray bars, corresponding to peptides listed in Table 4). All responses were confirmed to be CD8⁺ T cell dependent by CD8⁺ and CD4⁺ depletion experiments (data not shown). Four of the actual epitopes within the larger 20-aa peptides were mapped by using truncated peptides in a similar Elispot format, and their HLA restriction was determined (Table 4). One of the targeted peptides, aa 781 to 800, was recognized by three different persons, whereas six of the other newly identified immunogenic peptides were recognized by only one individual. Three of the seven individuals (i.e., P7, P9, and P12) recognizing peptides not previously reported to contain CTL epitopes showed no detectable response by the original assay with only the optimal epitopes and thus would have been erroneously deemed to have no CTLs to HCV. In the other four subjects with detectable responses against previously described epitopes, the novel responses comprised between 40 and 80% of the total magnitude of all HCV-specific CD8⁺-T-cell responses together (Fig. 4). There was no significant correlation between the overall magnitude of the novel HCV-specific CD8⁺-T-cell responses and the overall magnitude of the HLA A2 responses or the overall magnitude of the responses toward all previously defined epitopes (P = 0.97 and P = 0.87, respectively). In summary, of the 14 persons tested by using HLA A2-presented epitopes, 4 had detectable responses, whereas 9 of the 14 persons (including these 4) had detectable responses when screened by the more comprehensive method. Moreover, in two of four persons positive for an HLA A2-restricted response, this was not the dominant response.

FIG. 4.

Comprehensive analysis of the HCV-specific CD8⁺-T-cell response. Elispot responses to previously described HLA A2-restricted epitopes are shown as white bars; Elispot responses to previously described non-HLA A2-restricted epitopes are shown as black bars, and novel responses are indicated as gray bars.

TABLE 4.

Novel epitopes detected by comprehensive Elispot analysis

HCV protein	aa position	Sequencea	HLA restric- tion	Sub- ject(s)
E2	610-618	DYPYRLWHY	Cw7	P4
P7	781-800	KWVPGAVYTFYGMWP     LLLLL*	ND	P9, P11, P14
NS3	1171-1190	CPAGHAVGIFRAAVCTRGVA*	ND	P9
NS3	1435-1443	ATDAL MTGY	A1	P2
NS5A	2191-2210	ARGSPPSVASSSASQLSAPS*	ND	P12
NS5B	2568-2577	QPEKGGRKPA	B55	P7
NS5B	2898-2907	SPGEINRVAA	B55	P7

^a*, Minimal epitope and HLA restriction not determined.

In vitro stimulation reveals additional subdominant HCV-specific CTLs in a majority of individuals.

Having failed to demonstrate A2-restricted responses in circulating PBMC, despite frequent recognition of other epitopes, we next addressed whether additional HLA A2-restricted responses might be present but below the limits of detection in direct assays on fresh cells. To address this question, we stimulated PBMC in vitro from nine of the study subjects, including five who had previously been shown to have no HLA A2-restricted responses by Elispot and tetramer analyses. PBMC were stimulated in vitro in the presence of IL-2 and HCV peptides over 8 to 9 days. Four different HLA A2-restricted HCV peptides, for which tetramers were available, were used for stimulation. The results obtained by tetramer analysis with unstimulated cells were identical to the results obtained by Elispot assay, with the same four subjects having detectable responses (Table 5; a representative example is shown in Fig. 5). However, after stimulation, responses to additional epitopes that had been below the limits of detection in the direct assays were determined to be present by using this technique (Fig. 5). In the four subjects who had detectable responses to HLA A2 epitopes in the direct ex vivo assay, three had additional responses that first became detectable after in vitro expansion (Table 5). For the one individual in whom no additional response was detected (P6), the three ex vivo detectable responses were readily expanded. Of the five subjects tested with negative results directly ex vivo, three (P4, P8, and P10) tested positive after specific stimulation. One epitope (NS4 1851-1859) was never detected, either directly ex vivo or after stimulation. These results show that memory responses to HCV peptides are present in the majority of infected persons but that HLA A2-restricted responses are often weak and are usually not the dominant responses in persons with HCV infection.

TABLE 5.

Tetramer analysis of stimulated cells

Subject	% Tetramer-positive cellsa
	NS3 1073		NS4 1406		NS4 1851		NS5 2594
	Ex vivo	Stim.	Ex vivo	Stim.	Ex vivo	Stim.	Ex vivo	Stim.
P2	1.31	1.4	0.12	4	<0.02	<0.02	<0.02	11
P3	<0.02	<0.02	<0.02	<0.02	<0.02	<0.02	<0.02	<0.02
P4	<0.02	12	<0.02	9	<0.02	<0.02	<0.02	<0.02
P6	0.42	43	0.09	2.50	<0.02	<0.02	0.23	17
P8	<0.02	1.50	<0.02	<0.02	<0.02	<0.02	<0.02	<0.02
P10	<0.02	0.40	<0.02	41	<0.02	<0.02	<0.02	<0.02
P11	<0.02	2	<0.02	5.60	<0.02	<0.02	0.23	25
P12	<0.02	<0.02	<0.02	<0.02	<0.02	<0.02	<0.02	<0.02
P14	0.17	31	<0.02	6.60	<0.02	<0.02	<0.02	2.50

^aEx vivo, ex vivo assay; Stim., post-stimulation assay. The number of positive responses (indicated in boldface) for the ex vivo and poststimulation assays, respectively, were as follows: for NS3 1073, 3 and 9; for NS4 1406, 2 and 6; for NS4 1851, 0 and 0; and for NS5 2594, 2 and 4.

FIG. 5.

Tetramer analysis before and after stimulation of PBMC with peptides representing HCV epitopes. In subject P14, only one of four tetramers (NS3 1073) tested positive directly ex vivo. After 9 days of stimulation with the four respective peptides, three of the four tetramers tested positive, with frequencies of up to 30% of the CD8⁺ T cells.

DISCUSSION

Utilizing a comprehensive approach for the analysis of HCV-specific CD8⁺ T lymphocytes we showed that these responses can be detected directly ex vivo in the majority of anti-HCV-positive individuals without requiring predictive algorithms or prior knowledge of HLA type. Using 301 overlapping peptides spanning all HCV proteins, we found circulating HCV-specific CD8⁺ T cells in 9 of 14 subjects tested, whereas only 4 of these HLA A2-positive persons had detectable HLA A2-restricted responses without prior in vitro stimulation. Importantly, half of the responses detected were directed against previously uncharacterized epitopes, and the vast majority of previously reported A2 epitopes were either not targeted or were below the limits of detection in persons with clear CD8⁺-T-cell responses to other epitopes. These data indicate that HLA type does not predict dominant HCV responses in seropositive individuals, that targeted epitopes are scattered throughout all gene products, and that no epitopes or regions are preferentially targeted by the majority of infected persons.

These studies extend previous studies analyzing the HCV-specific CTL response in PBMC that have usually relied on testing a limited number of previously described or predicted class I epitopes (2, 4, 5, 14, 15, 28, 30, 33, 34, 37-39). By relying on HLA A2-restricted epitopes, many individuals are excluded from study, especially those from non-Caucasian populations in whom HLA A2 is less frequently expressed (9). Our study demonstrates that the analysis of responses restricted by a single HLA allele cannot be used as a surrogate marker for the overall HCV-specific CTL response. A response to any of the 19 previously described HLA A2-restricted HCV epitopes was seen in only 4 of 14 subjects, yet 9 of 14 had responses when the comprehensive approach with overlapping peptides was applied. Moreover, in HLA A2-positive persons with HLA A2-restricted responses, these were not always the dominant responses. In vitro simulation increased the detection of responses to some A2-restricted epitopes, indicating that weak responses were present but underscoring that these responses were minor compared to those directed at non-HLA A2 epitopes in a number of persons. This is consistent with studies in other chronic viral infections such as HIV, in which the expressed HLA alleles are poorly predictive of the targeted epitopes (3, 10, 12). These findings will be particularly important to consider in studies designed to evaluate correlation of CD8 T-cell responses to other parameters such as liver histology and/or viral load, as assessment of these relationships will require comprehensive analyses such as those described here.

These studies also allowed us to assess the relative predictive ability of algorithms designed to predict CTL epitopes by using the sensitive techniques of Elispot and tetramer analysis. Our results show that a minority of predicted HLA A2-restricted epitopes are detected directly in PBMC. At the time of our experiments, 19 HLA A2-restricted HCV-specific epitopes had been described; only 3 of 19 (NS3 1073, NS3 1406, and NS5B 2594) were recognized by Elispot analysis in any of the patients, and these 3 were repeatedly detected in more than one person. We confirmed these findings by tetramer analysis to exclude the possibility of specific cells being present but undetected due to a defects in functional properties (13, 29). While frequencies of HCV-specific cells were indeed higher when analyzed by tetramer, the results obtained by Elispot assay correlated significantly in their relative frequency. More importantly, tetramer analysis did not reveal any response that was not also detected by the Elispot assay.

A review of the literature confirmed that the three HLA A2-restricted epitopes detected in our study are the HLA A2-restricted HCV-specific epitopes most commonly detected by all different assays available, be they direct ex vivo assays such as Elispot or tetramer or assays requiring previous specific or nonspecific expansion of cells in vitro (5, 13, 14, 28, 29, 37-39). For the six epitopes tested by tetramer, all had similar scores (between 23 and 30; mean, 24.4) in the algorithm described by Rammensee et al. (32). These scores, ranging from 6 to 31 (mean, 24.4), were also not significantly different from the scores for the other epitopes tested in the Elispot assay. Furthermore, the three epitopes recognized have variable sequences, whereas the epitope in the core is highly conserved through all genotypes but was never detected directly ex vivo. In this context it is noteworthy that we detected responses in individuals infected with various genotypes, despite our peptides being derived from a genotype 1 sequence.

For one of the epitopes (NS3 1073) cross-reactivity with a flu epitope has been proposed as the mechanism leading to preferred recognition (40). Further studies will be needed to establish whether this is a general phenomenon and also whether there is a difference in function and pathogenicity and in their possible role in vaccines between the epitopes which are more readily detectable and those requiring in vitro expansion.

Using the peptide matrix approach, we were able to detect circulating HCV-specific CTL responses in 6 of 10 individuals with chronic HCV infection, and this is comparable to results obtained by cloning from liver-derived lymphocytes where such a response was detected in 45% of subjects (41). This suggests that, while the frequencies of HCV-specific cells have been shown to be higher in the liver, this assay is able to detect HCV-specific CTLs with similar sensitivity in the periphery without the need of in vitro expansion. This will facilitate future studies of HCV immunopathogenesis, since liver-derived cells are only occasionally available and the number of extracted cells is usually very low. However, more information is needed regarding how the cells in the two compartments compare in terms of phenotype and functional properties.

In conclusion, we show that the detailed analysis of the HCV-specific CD8⁺-T-cell response requires a comprehensive screening that cannot be substituted by testing a limited number of epitopes as indicators of the T-cell response. The study of HCV-specific CTLs directly ex vivo without the restriction to certain HLA alleles will be important in determining the role of CD8⁺ T cells in the control and immunopathology of HCV infection. Already the data indicate that the response is more limited than expected. This finding provides a rationale for future studies designed to measure the impact of augmented HCV-specific immunity on immune control and disease progression.