Ex Vivo Profiling of CD8⁺-T-Cell Responses to Human Cytomegalovirus Reveals Broad and Multispecific Reactivities in Healthy Virus Carriers

Abstract

Human cytomegalovirus (HCMV) can establish both nonproductive (latent) and productive (lytic) infections. Many of the proteins expressed during these phases of infection could be expected to be targets of the immune response; however, much of our understanding of the CD8⁺-T-cell response to HCMV is mainly based on the pp65 antigen. Very little is known about T-cell control over other antigens expressed during the different stages of virus infection; this imbalance in our understanding undermines the importance of these antigens in several aspects of HCMV disease pathogenesis. In the present study, an efficient and rapid strategy based on predictive bioinformatics and ex vivo functional T-cell assays was adopted to profile CD8⁺-T-cell responses to a large panel of HCMV antigens expressed during different phases of replication. These studies revealed that CD8⁺-T-cell responses to HCMV often contained multiple antigen-specific reactivities, which were not just constrained to the previously identified pp65 or IE-1 antigens. Unexpectedly, a number of viral proteins including structural, early/late antigens and HCMV-encoded immunomodulators (pp28, pp50, gH, gB, US2, US3, US6, and UL18) were also identified as potential targets for HCMV-specific CD8⁺-T-cell immunity. Based on this extensive analysis, numerous novel HCMV peptide epitopes and their HLA-restricting determinants recognized by these T cells have been defined. These observations contrast with previous findings that viral interference with the antigen-processing pathway during lytic infection would render immediate-early and early/late proteins less immunogenic. This work strongly suggests that successful HCMV-specific immune control in healthy virus carriers is dependent on a strong T-cell response towards a broad repertoire of antigens.

Human cytomegalovirus (HCMV) is a classic example of a group of herpesviruses that are found throughout all geographic locations and socioeconomic groups, and it infects between 50 and 85% of adults (3, 32). For most healthy persons who acquire primary HCMV infection after birth, there are few symptoms and no long-term health consequences. Occasionally, some adults with primary HCMV infection display infectious mononucleosis-like symptoms, with prolonged fever and mild hepatitis. After invasion and the lytic cycle of replication, the virus remains dormant by establishing a reservoir of latently infected cells from which chronic low-grade reactivation into the virus productive (lytic) cycle occurs throughout life. Although the factors controlling latency and reactivation are not completely understood, impairment of the body's cell-mediated immune system, either by drug-induced immunosuppression or infection by certain pathogens, can consistently reactivate the virus (27, 52).

HCMV infection is, therefore, important to certain high-risk groups. Major areas of concern are the following: (i) the risk of infection to the unborn baby during pregnancy, (ii) the risk of infection to people who work with children, and (iii) the risk of infection to immunocompromised persons (e.g., organ transplant patients) (8, 30, 32). The risk to the fetus appears to be almost exclusively associated with women who acquire a primary infection during pregnancy (15). In 1996 alone, more than 17,000 cases of HCMV-induced sequelae or death were estimated in Europe and the United States (31). Epidemiological studies have shown that 80 to 90% of infants who acquired congenital HCMV infection display a variable pattern of pathological sequelae within the first few years of life that may include hearing loss, vision impairment, and mental retardation. Another 5 to 10% of infants who are infected but without symptoms at birth will subsequently develop various degrees of hearing and mental or coordination defects. In addition, recent studies suggest that HCMV-seropositive individuals who have undergone coronary angioplasty develop restenosis more frequently than seronegative patients (30), although a causal relationship has yet to be shown. Thus, there is a range of clinical situations where HCMV is a significant cause of morbidity and mortality.

There is an increasing argument that a reduction in HCMV load in these individuals could provide a significant therapeutic benefit (1, 15, 38). Immunization would provide the most practical modality for achieving such a reduction in HCMV load. Attenuated viral preparations have had limited success, and a subunit approach to a HCMV vaccine is considered more desirable due to the repertoire of immunomodulator proteins or immunoevasins (36). To develop a successful immunization strategy, viral antigens that activate both protective cytotoxic-T-lymphocyte (CTL) and humoral responses need to be identified, and a formulation based on these epitopes needs to be tested to assess their immunogenicity (1, 14, 17, 33). To date most of the studies on CTL responses toward HCMV have primarily focused on pp65 (2, 7, 19, 20, 25, 26, 35, 43, 50, 51). Although there is some evidence to suggest that other HCMV antigens may also be targets for CTL control, information on these is rather limited (16, 18, 37, 41). The present study was designed to profile T-cell responses to a wide variety of HCMV antigens (pp28, pp50, pp65, pp150, pp71, gH, gB, IE-1, US2, US3, US6, US11, UL16, and UL18) that are expressed at different stages of infection and play an important role in the overall pathogenesis of HCMV-associated diseases. The viral glycoproteins gB and gH were chosen because they appear to be involved in virus attachment (45), whereas the IE-1 protein is expressed during viral replication. IE-1 is considered to be important for viral reactivation and to also play a role in HCMV-induced pathology (42). The phosphoproteins (pp28, pp50, pp65, pp71, and pp150) are all tegument proteins, while the US- and UL-designated proteins all play a role in immune modulation or evasion (reviewed in references 36 and 45). By priming the immune response to these antigens, viral replication could be controlled at the levels of attachment, replication, assembly, and reactivation from the latent phase, all of which are crucial stages in the development of HCMV disease.

Preliminary analysis of these HCMV protein sequences, using predictive algorithms, strongly suggested that these antigens contained potential CTL epitopes. Synthetic peptides were subsequently synthesized and tested for their ability to recall memory T-cell responses from seropositive donors, as measured by gamma interferon (IFN-γ) production in an enzyme-linked immunospot (ELISPOT) assay. Finally, we generated both polyclonal and clonal CTLs from seropositive donors that showed positive responses in ELISPOT assays. This confirmed cytolytic activity toward target cells that were either sensitized with synthetic peptides, infected with HCMV, or infected with recombinant vaccinia virus that encoded individual antigens. Using these approaches, we have identified a number of novel HCMV-encoded proteins as potential targets for CD8⁺-T-cell immunity. These T-cell determinants may, in the future, prove to be very important in understanding the dynamics of HCMV immune control in clinical settings and vaccine development.

MATERIALS AND METHODS

Establishment and maintenance of cell lines.

All cell lines were routinely maintained in RPMI 1640 supplemented with 2 mM l-glutamine, 100 IU of penicillin/ml, and 100 μg of streptomycin/ml plus 10% fetal calf serum (FCS) (referred to as growth medium), unless otherwise stated.

Epstein Barr virus-transformed lymphoblastoid cell lines (LCLs) were established from HCMV-seropositive donors by exogenous virus transformation of peripheral B cells using the B95.8, BL74, and QIMR-WIL virus isolates as described previously (28). Autologous LCLs, pulsed with HCMV peptides, were used to stimulate T-cell lines and clones on a weekly basis or as target cells in cytotoxicity assays (see below).

To generate phytohemagglutinin (PHA) blasts, peripheral blood mononuclear cells (PBMC) were stimulated with PHA (20 μg/ml; CSL Ltd., Melbourne, Australia). After 3 days of culture, growth medium containing supernatant from the T-cell line MLA 144 (Gibbon ape lymphoma; American Type Culture Collection [ATCC], Bethesda, Md.) and highly purified recombinant human interleukin 2 (rIL-2) was added (21). PHA blasts were propagated by twice-weekly replacement of rIL-2 and MLA 144 supernatant (no further PHA added) for up to 6 weeks.

The TAP peptide transporter-negative B × T hybrid cell line 0.174 × CEM.T2 (referred to as T2; HLA A2, B51) (40) was used to test the ability of HLA A2-predicted peptides to bind to major histocompatibility complex (MHC) molecules (see below). T2 cells transfected with other individual HLA class I antigens were also used for MHC stabilization assays. T2 cells expressing HLA B35, B7, and B27 have been described elsewhere (44, 46, 53). T2 cells expressing HLA A3, A24, and B8 were established by transfecting expression vectors encoding individual class I alleles as described previously (24). Briefly, cDNA for these HLA class I alleles was amplified using sequence-specific primers and cloned into an EGFP-N1 expression vector, which contains a green fluorescent protein tag and an antibiotic resistance gene to allow for positive selection. T2 cells were transfected with these expression vectors and cultured in growth medium supplemented with G418 antibiotic (800 μg/ml) (Life Technologies, Grand Island, N.Y.) for 3 weeks. Green fluorescent protein-positive cells were sorted using a fluorescence-activated cell sorter vantage, and the purified cells were maintained in growth medium that was supplemented with G418 antibiotic (800 μg/ml) to select for the growth of transfected cells. HLA class I expression on these transfectants was confirmed by using HLA allele-specific antibodies and flow cytometry.

The JS fibroblast line (HLA A1, A2, B8, B51) was used to test the ability of either HLA A1-, HLA A2-, or HLA B8-restricted T-cell lines or clones to recognize HCMV-infected target cells in cytotoxicity assays. Prior to HCMV infection, fibroblasts at 70 to 80% confluence were treated with IFN-γ (100 U/ml; Roche Diagnostics, Mannheim, Germany) for 24 h at 37°C in 5% CO₂. Cell culture supernatant was removed and replaced with a small volume of growth medium (5 ml) which contained the HCMV AD-169 strain at a multiplicity of infection (MOI) of 0.05:1. Virus was incubated with fibroblasts for 2 h to allow for virus adsorption. After the initial incubation, an extra 25 ml of growth medium was added and fibroblasts were incubated for a further 24 h before use as target cells in cytotoxicity assays (see below).

Epitope prediction and peptide synthesis.

Two predictive algorithms (SYFPEITHI [http://www.uni-tuebingen.de/uni/kxi/] [34] and the Bioinformatics and Molecular Analysis Section [BIMAS] HLA Peptide Binding Predictions programs [http://bimas.dcrt.nih.gov/molbio/hla_bind] [29]) were used to predict putative HLA class I-restricted CTL epitopes from within the amino acid sequences of HCMV antigens (pp28, pp50, pp65, pp150, pp71, gH, gB, IE-1, US2, US3, US6, US11, UL16, and UL18). These algorithms were used to identify potential epitopes for HLA A1, A2, A3, A24, A26, B7, B8, B27, B35, and B44 alleles. Each peptide was assigned a score on the basis of the strength of the interaction between the MHC molecule and the peptide. Peptides that ranked higher than 24 in the SYFPEITHI program predictions and peptides that scored greater than 100 from the BIMAS program predictions were synthesized using the Merrifield solid phase method (49) (Mimotopes, Melbourne, Australia). All peptides were dissolved in 10% dimethyl sulfoxide and diluted in serum-free RPMI 1640 medium for use in assays which tested for their ability to induce both the production of IFN-γ and cytotoxic activity in donor PBMC and T-cell clones and polyclonal lines in vitro.

MHC stabilization assay.

The ability of synthetic peptides to stabilize MHC molecules on the surface of the T2 cell line was measured by indirect immunofluorescence (11). T2 cells (2 × 10⁵) were incubated in serum-free AIM-V medium (Invitrogen, Melbourne, Australia) in the presence of 100 μg (final concentration) of peptide/ml for 1 h at 37°C and 5% CO₂ in a humidified atmosphere. Samples were then incubated for a further 14 to 16 h at 26°C, after which the cells were returned to 37°C for 2 to 3 h prior to immunofluorescent staining. Cells were washed free of unbound peptide with growth medium prior to the addition of primary antibody. Anti-HLA allele-specific monoclonal antibody was added to the T2 cells and incubated at 4°C for 30 min. HLA-specific antibodies used in this study were BB7.2 (HLA A2 specific; ATCC, Charlottesville, Va.), SFR8-B6 (HLA Bw6 specific; ATCC), and TU109 (HLA Bw4 specific). After being washed with growth medium, these cells were incubated with fluorescein isothiocyanate- or phycoerythrin-labeled anti-mouse immunoglobulin (Ig)-specific antibody (Silenus/Chemicon, Boronia, Victoria, Australia) at 4°C for 30 min. Finally, cells were washed and resuspended in 500 μl of cold phosphate-buffered saline (PBS) supplemented with 1% FCS. A sample of T2 cells was incubated with AIM-V medium alone (no peptide) at 26°C for 14 to 16 h and served as a negative control. The second negative control comprised a sample of T2 cells that had been cultured in growth medium without peptide at 37°C. Fluorescence intensities of the T2 cells were then measured with either a FACScan or FACScalibur flow cytometer (BD Biosciences, San Jose, Calif.). The MHC stabilization efficiency (MSE) for each peptide was calculated as the percent increase of the mean fluorescence above that of the negative controls. All peptides that induced fluorescence intensity greater than mean + 3 standard errors of the mean (SEM) of the fluorescence intensity on T2 cells in the absence of peptide at 26°C (negative control) were considered as positive binders.

ELISPOT assay.

The ELISPOT assay was used to assess whether synthetic HCMV peptides could stimulate a memory response, as measured by the production of IFN-γ, in PBMC from a large panel of seropositive donors (4). Briefly, a 96-well nitrocellulose plate (Multiscreen; Millipore, Bedford, Mass.) was coated overnight at 4°C with mouse monoclonal antibody anti-IFN-γ IgG1 (10 μg/ml; Mabtech, Nacka, Sweden). The plate was then washed six times in PBS and blocked for 1 h at 37°C with PBS supplemented with 5% FCS. The blocking solution was removed, and PBMC from healthy HCMV seropositive donors were added at a concentration of 2.5 × 10⁵ cells per well in growth medium. These cells were incubated for 18 h at 37°C in a 5% CO₂ atmosphere in the presence of synthetic peptides from HCMV antigens (10 μg/ml). After incubation, the plate was washed three times with PBS supplemented with 0.05% Tween, followed by three washes with PBS alone. Biotinylated detection antibody, anti-IFN-γ (Mabtech), was added to each well at a final concentration of 1 μg/ml in PBS. The plates were incubated at room temperature in the dark for 4 h and then washed, as described above. Streptavidin-alkaline phosphatase (Sigma, St. Louis, Mo.) was added to each well at a final concentration of 1 μg/ml in PBS and incubated at room temperature in the dark for 2 h. After a final wash with PBS, the substrate, 5-bromo-4-chloro-3-indolyl phosphate and nitroblue tetrazolium, was added to each well, and the plates were incubated for 30 min at room temperature. Cells that produced IFN-γ in response to the presence of peptide were detected as purple spots on the nitrocellulose membrane of each well. The spots were counted automatically using either a closed-circuit camera and ImagePro image analysis software (4) or automated ELISPOT reader (AID, Strassberg, Germany). The T-cell precursor frequency for each peptide was based on the total number of PBMC in the well and the number of peptide-specific spots per well, over an average of three wells. The number of peptide-specific spots was also calculated by subtracting the negative control values, which consisted of PBMC without peptide (an average of three wells), from the test wells. All peptides that induced IFN-γ response greater than mean + 3 SEM of the control wells were considered putative CD8⁺-T-cell epitopes. ELISPOT results were expressed as number of spot-forming cells (SFC)/10⁶ PBMC.

Generation of polyclonal and clonal HCMV-specific CTLs.

To generate polyclonal CTLs, 10⁶ autologous PBMC were incubated with 20 μg of peptide/ml for 1 h at 37°C. Peptide-sensitized PBMC were washed free of unbound peptide and cocultured with 2 × 10⁶ PBMC from HCMV-seropositive healthy donors for 7 days. On day 7, these lymphocytes were restimulated with peptide-sensitized (5 μg/ml), γ-irradiated (8,000 rad) autologous LCLs. After 10 days of culture in growth medium supplemented with rIL-2 (20 U/ml) and 30% MLA 144 supernatant, the cells were used as polyclonal effectors in a standard ⁵¹Cr release assay.

To generate HCMV-specific CTL clones, PBMC were initially stimulated as described above for polyclonal CTLs. After 3 days of culture in growth medium, CTL clones were generated by seeding in 0.35% agarose (12). On days 7 to 10 of culture at 37°C, CTL clones were harvested from the agarose into 96-well round-bottom microtiter plates and maintained in growth medium containing rIL2 (20 U/ml) and 30% MLA 144 supernatant. Clones were restimulated once weekly with peptide-sensitized γ-irradiated (8,000 rad) autologous LCLs. These CTL clones and the polyclonal lines described above were screened for cytotoxic activity on a panel of autologous or allogeneic target cells that were either sensitized with synthetic peptides, infected with recombinant vaccinia virus encoding individual HCMV antigens, or infected with AD-169 HCMV strain (see below).

Vaccinia virus recombinants.

Recombinant vaccinia constructs encoding HCMV antigens and a negative control vaccinia virus construct made by insertion of the pSC11 vector alone, which is negative for thymidine kinase (Vacc.TK⁻), have been previously described (9, 10, 39). LCLs were infected with recombinant vaccinia virus at an MOI of 10:1 for 1 h at 37°C, as described earlier (21, 22). After overnight infection, cells were washed with growth medium and labeled with ⁵¹Cr for use as targets in cytotoxicity assays (23).

Cytotoxicity assay.

Target cells were infected with either HCMV (AD-169 strain) or recombinant vaccinia viruses or presensitized with synthetic peptides prior to incubation with ⁵¹Cr for 90 min at 37°C. Following incubation, these cells were washed in growth medium and used as targets in a standard 5-h ⁵¹Cr-release assay (12) at an effector-target ratio of 10:1. The level of specific CTL lysis was calculated using the following formula: [(specific release − spontaneous release)/(total release − spontaneous release)] ×100. A spontaneous release of less than 10 to 15% of the total release was observed in all assays.

RESULTS

HLA class I epitope prediction and HLA binding peptides from HCMV antigens.

In the first set of experiments, putative HLA class I-restricted T-cell epitopes were identified from 14 HCMV antigens using two computer-based algorithms, which predict 8-, 9-, or 10-mer amino acid sequences likely to bind successfully to MHC class I molecules (HLA A1, A2, A3, A24, A26, B7, B8, B27, B35, and B44) based on the half-time dissociation of the interaction. Peptides that ranked higher than 24 in the SYFPEITHI program and/or peptides that scored greater than 100 from the BIMAS program (Table 1) were used to screen healthy, seropositive donors. A total of 202 sequences were identified as putative HLA class I-restricted CD8⁺-T-cell epitopes (Table 1). All peptides predicted to bind HLA A2, A3, A24, B7, B8, B27, and B35 were analyzed for their capacity to bind to HLA class I MHC molecules and stabilize their expression on the surface of T2 cells transfected with individual HLA alleles (Table 1). Representative data for the HLA A2 stabilization are shown in Fig. 1. HCMV peptides that induced an increase in the relative fluorescence intensity, above that of the negative controls plus three SEM for an individual HLA class I allele, were considered positive. Of the 202 predicted peptides, 94 peptides clearly showed positive HLA class I binding (see Table 1). A total of 55 predicted epitopes did not stabilize the expression of MHC molecules on the surface of T2 cells, since the relative increase in the fluorescence intensity was less than mean + 3 SEM of the negative control. The remaining 52 peptides could not be tested in MHC stabilization assays because T2 cells transfected with the HLA A1, A26, or B44 alleles were not available for this study. These peptides were designated not tested (NT) in Table 1 but were subsequently analyzed, where appropriate, in the functional assays described below. Predictive sequences from all antigens, except US6, included potential CD8⁺-T-cell epitopes that stabilized HLA class I molecules on T2 cells.

TABLE 1.

List of HLA class I-restricted predictive CTL epitopes from HCMV proteins and MSE

Antigen or peptide sequence	HLA	MSEa		Antigen or peptide sequence	HLA	MSEa
pp28
LLIDPTSGL	A2	+
LLVEPCARV	A2	++
LLLIVTPVV	A2	++
FLLSHDAAL	A2	++
PLREYLADL	A2	−
GLLGASMDL	A2	−
LVEPCARVY	A1/A3	NT/++
GIKHEGLVK	A3	+++
ELLAGGRVF	A3	−
RLLDLAPNY	A3	+++
ELLGRLNVY	A3	−
CRYKYLRKK	B27	−
ARVYEIKCR	B27	−
pp50
LLNCAVTKL	A2	++
QLRSVIRAL	A2	−
VTEHDTLLY	A1	NT
RGDPFDKNY	A1	NT
GLDRNSGNY	A1	NT
TLLNCAVTK	A3	+++
TVRSHCVSK	A3	+++
TRVKRNVKK	B27	−
YEQHKITSY	B44	NT
SEDSVTFEF	B44	NT
US3
YLFSLVVLV	A2	+
TLLVYLFSL	A2	++
LLFRTLLVYL	A2	+
VYLFSLVVL	A24	−
US6
LYLCCGITL	A24	−
UL16
TLFFFLLAL	A2	−
TLARDIVLV	A2	++
LYPRPPGSGL	A24	+++
LYTSRMVTNL	A24	−
YPRPPGSGL	B7	+++
DEIIGVAFTW	B44	NT
NESVVDLWLY	B44	NT
RTVPVTKLY	A1	NT
pp71
QLLIPKSFTL	A2	−
TLVIPSWHV	A2	++
LLIPKSFTL	A2	−
DLVPLTVSV	A2	+
CSDPNTYIHK	A1	NT
EYIVQIQNAF	A24	+++
AEVVARHNPY	B44	NT
pp150
GQTEPIAFV	A2	+
KMSVRETLV	A2	++
FLGARSPSL	A2	++
ALVNAVNKL	A2	++
ALVNFLRHL	A2	++
NILQKIEKI	A2	+
LIEDFDIYV	A2	++
PLIPTTAVI	A2	−
RIEENLEGV	A2	+
NVRRSWEEL	B7	++
WPRERAWAL	B7/B8	+++/++
KARDHLAVL	B7	+++/PICK>
LLYPTAVDL	A2	+++
ALDPYNEVV	A2	+++
SAIIGIYLL	A2	−
ITSLVRLVY	A1	NT
HHEYLSDLY	A1	NT
AIIGIYLLY	A1/A3	NT/+++
QTEKHELLV	A1	NT
ATDSRLLMM	A1	NT
FLDAALDFNY	A1	NT
DTQGVINIMY	A1	NT
LRENTTQCTY	A1	NT
SAIIGIYLLY	A1	NT
SLRNSTVVR	A3	+++
ALALFAAAR	A3	−
QLNRHSYLK	A3	+++
RLFPDATVP	A3	−
RLNTYALVSK	A3	+++
LVRLVYILSK	A3	+++
YLMDELRYVK	A3	−
ELYLMGSLVH	A3	−
ALTVSEHVSY	A3	++
NYLDLSALL	A24	−
SYVVTNQYL	A24	−
SYLKDSDFL	A24	−
TYALVSKDL	A24	−
SYRSFSQQL	A24	−
TYGRPIRFL	A24	−
YYVFHMPRCL	A24	−
MYMHDSDDVL	A24	−
ETFPDLFCL	A26	NT
DLTETLERY	A26	NT
SPRTHYLML	B7	+++
FPDLFCLPL	B7	+++
SPRTHYLMLL	B7	+++
MPRCLFAGPL	B7	+++
TPMLLIFGHL	B7	++
APYQRDNFIL	B7	+++
GRCQMLDRR	B27	−
RRDHSLERL	B27	−
SEALDPHAF	B44	NT
RENTTQCTY	B44	NT
DDVLFALDPY	B44	NT
IE-1
SLLSEFCRV	A2	+++
VLAELVKQI	A2	++
ILGADPLRV	A2	++

^aMSE was assessed using T2 cells transfected with individual HLA class I alleles. The levels of HLA in the presence of peptide are expressed relative to expression on T2 cells in the absence of peptide at 26°C. +++, 200 to 300%; ++, 100 to 199%; +, 51 to 99%; −, 0 to 50% (relative increase in HLA expression); NT, not tested.

FIG. 1.

MHC stabilization on T2 cells using potential HLA A2-binding peptides from HCMV antigens (pp65, pp71, pp150, IE-1, gB, pp50, pp28, gH, US3, US2, UL16, and UL18). T2 cells were initially incubated with 100 μg (final concentration) of each of the peptides/ml for 1 h at 37°C and then for 14 to 16 h at 26°C, followed by incubation at 37°C for 2 to 3 h. HLA A2 expression on these cells was analyzed by flow cytometry using the BB7.2 antibody. The dotted line indicates the mean + 3 SEM of the fluorescence intensity for HLA A2 on T2 cells incubated at 26°C without peptide, which was the cutoff for a positive result.

Ex vivo analysis of T-cell responses to HLA A2 predicted peptides.

In the second set of experiments, we used ELISPOT assays to test the peptides predicted to bind to HLA A2 MHC molecules, since the A2 allele is expressed in approximately 50% of the Caucasian population. The ELISPOT assay allowed for the rapid identification of potential T-cell epitopes, without prolonged in vitro culture, by measuring a memory response in seropositive individuals who are successful in controlling HCMV replication and disease. A panel of eight HLA A2-positive healthy virus carriers were recruited at the outset of this study (SB, SE, CJ, CP, JP, BS, WS, and JT). The full HLA class I tissue type of each of these donors is listed in Table 2. PBMC were isolated from whole blood and stimulated with the peptides predicted to bind to HLA A2 molecules, and the cells that produced IFN-γ were detected. A summary of ELISPOT results based on all HLA A2 predicted epitopes is presented in Table 3. A total of 27 CTL epitopes were identified from the panel of 68 HLA A2 predicted peptides. Of these 27 CD8⁺-T-cell epitopes, 23 have not previously been identified. These responses were not constrained to a single HCMV antigen. The majority of the HLA A2-positive donors tested in our study showed a broad range of responses to multiple antigens. Overall, the T-cell responses for HLA A2-restricted epitopes in ELISPOT assays indicated an interesting hierarchy between the different antigens of HCMV. As reported previously (18), pp65 was clearly the most dominant antigen recognized by all the healthy virus carriers. The majority of the donors recognized two or more epitopes within this antigen (see Table 3). A range of precursor frequencies for the pp65 epitopes were evident among the different donors. These ranged from 24 ± 8 SFC/10⁶ PBMC (BS) to 1,201 ± 51 SFC/10⁶ PBMC (JP). The previously defined NLVPMVATV peptide epitope from pp65 was the only epitope recognized by every HLA A2-positive donor tested in this study. Another commonly recognized epitope from pp65 was RIFAELEGV (6 of 8 donors). For pp65 epitopes the average precursor frequency was highest for NLVPMVATV, followed by MLNIPSINV, LMNGQQIFL, and RIFAELEGV, respectively. Subdominant responses to other epitopes (RLLQTGIHV and ILARNLVPM) within pp65 were also detected by two different donors.

TABLE 2.

HLA antigen (class I) typing of the HCMV-immune donors included in this study

Donor	HLA type
MB	A1	A24	B8	B58
SB	A2		B35	B57
SC	A1		B8
JDu	A31	A33	B35	B58
JDa	A28	A31	B27	B18
JG	A1	A2	B7	B37
TD	A24	A26	B15	B62
RE	A11	A24	B35	B60
SE	A2	A29	B44	B60
PH	A1	A2	B8	B44
CJ	A2	A3	B8	B60
RK	A23	A24	B27	B41
TK	A3	A25	B35	B44
PP	A1	A24	B8	B14
CP	A2	A32	B7	B50
JP	A2	A28	B8	B62
CR	A1	A3	B35	B57
BS	A2	A11	B13	B27
CS	A3	A23	B35	B44
WS	A2	A32	B27	B60
JT	A2	A32	B44	B62
JW	A23	A32	B35	B49
MW	A1	A3	B8	B35
TC	A1	AA11	B8	B35
SW	A1		B8	B62

TABLE 3.

List of HLA A2-restricted putative CTL epitopes from HCMV antigens assessed by ELISPOT assays^a

HCMV antigen	Peptide sequence	Result for HCMV-seropositive donor (SFC/106 PBMC):
		SB	JP	WS	CJ	BS	CP	SE	JT
pp28	LLIDPTSGL	270	—	77 ± 49	—	—	—	—	—
	LLVEPCARV	—	—	—	—	—	—	—	—
	LLLIVTPVV	—	—	—	—	—	—	—	—
	FLLSHDAAL	—	—	—	—	—	—	—	—
	PLREYLADL	—	—	—	—	—	—	—	—
	GLLGASMDL	—	—	—	—	—	—	—	—
pp50	LLNCAVTKL	—	—	—	—	—	—	16 ± 5	—
	QLRSVIRAL	—	—	—	—	—	—	—	—
pp65	NLVPMVATV∗	1074 ± 63	1201 ± 51	1106 ± 295	220 ± 48	171 ± 24	670 ± 42	798 ± 12	1188 ± 202
	RIFAELEGV∗	29 ± 6	178 ± 13	240 ± 36	—	39 ± 9	274 ± 23	—	172 ± 66
	ILARNLVPM	—	—	—	—	—	—	182 ± 23	—
	VIGDQYVKV	—	—	—	—	—	—	—	—
	QMWQARLTV	—	—	—	—	—	—	—	—
	ALFFFDIDL	—	—	—	—	—	—	—	—
	RLLQTGIHV∗	—	—	—	—	68 ± 27	—	—	—
	LMNGQQIFL	—	—	201 ± 102	—	28 ± 6	436 ± 117	—	—
	MLNIPSINV∗	—	—	857 ± 50	—	24 ± 8	413 ± 60	—	—
pp71	QLLIPKSFTL	—	—	—	—	—	—	—	—
	TLVIPSWHV	—	—	—	—	—	—	—	—
	LLIPKSFTL	—	—	—	—	—	—	—	—
	DLVPLTVSV	—	—	—	—	—	—	—	—
pp150	GQTEPIAFV	305 ± 17	—	—	—	—	—	—	—
	KMSVRETLV	—	—	—	—	—	—	—	—
	FLGARSPSL	—	—	—	—	—	—	—	—
	ALVNAVNKL	—	—	60 ± 16	—	—	—	—	—
	ALVNFLRHL	—	—	57 ± 3	—	—	—	—	—
	NILQKIEKI	—	—	—	—	—	—	—	—
	LIEDFDIYV	—	—	88 ± 26	—	—	40 ± 3	—	—
	PLIPTTAVI	—	—	—	—	—	—	—	—
	RIEENLEGV	—	—	—	—	—	—	—	—
gB	IILVAIAVV	—	—	—	—	—	—	—	—
	DLDEGIMVV	—	—	—	—	—	—	—	—
	QMLLALARL	—	—	—	—	—	—	—	—
	NLFPYLVSA	—	—	—	—	—	—	—	—
	AVGGAVASV	41 ± 33	—	—	—	10 ± 1	120 ± 38	—	—
	RIWCLVVCV	18 ± 7	108 ± 14	—	—	16 ± 3	162 ± 16	—	—
	YINRALAQI	—	—	—	—	—	—	—	—
	GLDDLMSGL	—	—	—	—	—	—	—	—
gH	YLMDELRYV	—	—	—	—	—	—	—	—
	LLMMSVYAL	—	—	—	—	—	—	—	—
	YLTVFTVYL	—	—	—	—	—	—	—	—
	LIFGHLPRV	—	—	—	—	—	—	11 ± 1	—
	TLTEDFFVV	—	—	—	—	—	—	—	—
	SLVRLVYIL	—	—	—	—	—	—	—	—
	LLYPTAVDL	—	—	—	—	—	—	—	—
	YLLYRMLKT	—	—	—	—	—	—	—	—
	ILFDGHDLL	—	—	—	—	—	—	—	—
	ALDPYNEVV	—	—	—	—	—	—	—	—
	LMLLKNGTV	—	13 ± 1	—	68 ± 16	14 ± 1	90 ± 29	—	—
	SAIIGIYLL	—	—	—	—	—	—	—	—
IE-1	TMYGGISLL	—	—	—	—	—	—	—	—
	VLAELVKQI	177 ± 53	—	—	20 ± 1	48 ± 10	60 ± 16	—	—
	ILDEERDKV	—	—	—	—	—	—	—	—
	LLSEFCRVL	—	—	—	—	—	—	—	—
	VLEETSVML	4,340 ± 104	—	—	—	—	302 ± 90	—	—
	YILEETSVM∗	469 ± 26	—	—	—	—	102 ± 57	—	—
	SLLSEFCRV	—	—	589 ± 67	—	—	—	—	—
	CLQNALDIL	—	—	—	—	—	—	221 ± 26	—
	ILGADPLRV	—	—	—	—	—	—	—	—
UL18	TMWCLTLFV	—	—	—	—	—	—	—	—
US2	SMMWMRFFV	—	—	—	—	32 ± 9	—	383 ± 2	—
	LLVLFIVYV	—	—	92 ± 16	—	—	22 ± 11	—	—
	TLLVLFIVYV	—	—	—	—	—	—	483 ± 7	—/PICK>
US3	YLFSLVVLV	—	—	73 ± 17	—	—	—	—	—
	LLFRTLLVYL	96	—	77 ± 38	—	—	—	—	—
	TLLVYLFSL	—	—	157 ± 78	—	—	—	—	—
US11	TLFDEPPPLV	—	—	—	—	—	—	—	—
UL16	TLFFFLLAL	—	NT	—	NT	—	NT	—	NT
	TLARDIVLV	—	NT	—	NT	—	NT	—	NT

^aData are presented as mean ± SE of SFC/10⁶ PBMC. —, no response detected; NT, not tested; ∗, previously published epitope (see references 14, 16, 19, 20, 25, 26, 35, 37, 41, 50, and 51).

IE-1 was identified as the second-most-dominant antigen, after pp65, followed by pp150, gB, gH, US2, pp28, US3, and pp50. Six of the eight HLA A2-positive donors responded to at least one peptide within IE-1 (Table 3). Interestingly, one of most dominant T-cell responses to any antigen was identified within IE-1. The VLEETSVML peptide showed a precursor frequency of 4,340 ± 104 SFC/10⁶ PBMC in the donor SB. This epitope was predicted from the HCMV AD-169 strain amino acid sequence (GenBank accession number P13202). VLEETSVML shares some sequence homology with a previously published epitope, YILEETSVM (37), from within the same region of IE-1. The YILEETSVM peptide was synthesized and included in our analyses to test if the two peptides could be recognized with equal efficiency or if they induced separate reactivities in donor T cells. In our study, the VLEETSVML epitope was recognized more efficiently than YILEETSVM by cells from two different donors, indicating that the VLEETSVML epitope was distinct from the YILEETSVM epitope. Other novel epitopes identified from IE-1 (VLAELVKQI, SLLSEFCRV and CLQNALDIL) showed comparably lower frequency. CD8⁺-T-cell frequencies to epitopes within gB, gH, pp28, pp50, US2, US3, US11, and pp150 were generally low. There were no HLA A2-restricted epitopes identified within the pp71 antigen through the use of predictive algorithms. Notably, the HCMV-encoded immunomodulator protein US2 and US3 antigens were identified for the first time as targets for T-cell responses. Four HLA A2-positive healthy virus carriers responded to peptides within these antigens. Further screening for responses restricted through other HLA class I alleles identified a number of additional potential epitopes within these antigens (see below). These observations strongly suggest that viral proteins involved in HCMV-mediated interference with antigen processing can be targeted by CD8⁺-T-cell responses.

T-cell responses to other HLA class I predictive HCMV peptides.

In the next set of experiments, we extended our study to map potential CD8⁺-T-cell epitopes restricted through other common HLA class I alleles (HLA A1, A3, A24, A26, B7, B8, B27, B35, and B44). PBMC from a panel of healthy virus carriers (Table 2) were stimulated with individual peptides, and the cells that produced IFN-γ were detected. A total of 35 novel CD8⁺-T-cell epitopes, restricted through either HLA A1, A3, A24, B7, B8, B27, B35, or B44, were identified from a panel of 134 peptides tested in these assays (Tables 4 to 12), although no epitopes were mapped for HLA A26 and B35 alleles. In addition, T-cell responses to three previously published epitopes (YSEHPTFTSQY, HLA A1 restricted, and FPTKDVAL, HLA B35 restricted) were also detected (41, 51). Comparison of the overall success rate of predictive algorithms revealed that more than 40% of the epitopes predicted for HLA A1 and B7 alleles were confirmed by ELISPOT assays (Tables 4 and 8). The success rates for HLA B8, B27, A24, B44, and A3 were 27, 20, 16.6, 11.76, and 8.3%, respectively (Tables 5, 6, and 9 to 12). As in the case of HLA A2-restricted T-cell responses, both pp65 and IE-1 were the most immunodominant antigens. Of the 15 epitopes predicted from IE-1 antigen, eight (53.33%) were confirmed as CD8-T-cell epitopes by ELISPOT assays, while for pp65, 8 of 35 (22.85%) were mapped as epitopes. Surprisingly, T-cell responses to IE-1 epitopes in some individuals constituted 5 to 10% of their total CD8⁺-T-cell population (data not shown). Other antigens, such as pp150, gH, pp28, pp50, and UL18, were also identified as potential targets for class I-restricted T-cell responses (Tables 4 to 12). Although pp28 and pp50 have been identified as potential targets of T-cell response (18), this is the first report which identifies multiple epitopes restricted through HLA A1, A2, and B27. Of particular interest was the HLA A1-restricted epitope (VTEHDTLLY) from pp50, which was consistently recognized as one of the most dominant responses in all HLA A1-positive healthy virus carriers (7 of 8) (Table 4). Furthermore, the frequency and intensity of this response was comparable to those measured for the response to the previously published NLVPMVATV epitope in HLA A2-positive donors.

TABLE 4.

List of HLA A1-restricted putative CTL epitopes from HCMV antigens assessed by ELISPOT assays^a

Antigen	Peptide sequence	Result for donor:
		MB	SC	JG	PH	PP	MW	TC	SW
pp28	LVEPCARVY	—	—	—	—	—	—	—	—
pp65	ATVQGQNLKY	NT	—	—	—	—	—	—	NT
	IRETVELRQY	—	—	—	—	—	—	—	NT
	IGDQYVKVY	—	—	—	188 ± 107	—	—	29 ± 1	NT
	TVQGQNLKY	—	—	—	403 ± 16	—	—	—	—
	YRIQGKLEY	—	—	—	—	—	—	395 ± 10	—
	SQEPMSIYVY	—	—	—	—	—	—	—	NT
	YSEHPTFTSQY∗	NT	—	—	—	—	263 ± 40	132 ± 5	1,163 ± 142
pp50	VTEHDTLLY	945 ± 142	902 ± 169	—	396 ± 41	1,341 ± 10	895 ± 55	1,977 ± 40	986 ± 142
	RGDPFDKNY	—	—	—	125 ± 14	772 ± 64	893 ± 5	1,536 ± 137	1,080 ± 200
	GLDRNSGNY	—	—	—	—	—	—	375 ± 4	439 ± 30
pp71	CSDPNTYIHK	—	—	—	—	—	—	—	—
gB	VKESPGRCY	—	—	—	—	—	—	—	NT
	LDEGIMVVY	—	—	—	—	—	—	345 ± 2	—
	ATSTGDVVY	—	—	—	—	—	—	280 ± 3	255 ± 26
	NTDFRVLEL	—	—	—	—	—	—	1,435 ± 29	50 ± 15
gH	ITSLVRLVY	—	—	—	—	—	NT	—	NT
	HHEYLSDLY	—	—	—	—	—	NT	—	NT
	AIIGIYLLY	—	—	—	—	—	NT	—	NT
	QTEKHELLV	—	—	—	—	—	NT	—	NT
	ATDSRLLMM	—	—	—	—	—	—	—	NT
	FLDAALDFNY	—	—	—	—	—	—	—	NT
	DTQGVINIMY	—	—	—	—	—	—	—	NT
	LRENTTQCTY	—	—	—	—	—	—	—	NT
	SAIIGIYLLY	—	—	—	—	—	—	—	NT
IE-1	IKEHMLKKY	NT	—	—	—	—	NT	5,640 ± 10	—
	DEEEAIVAY	NT	—	—	—	—	NT	4,297 ± 87	—
	CVETMCNEY	NT	—	—	—	—	NT	5,485 ± 19	—
UL16	RTVPVTKLY	—	—	—	—	—	—	—	—

^aData are presented as means ± SE of SFC/10⁶ PBMC. —, No response detected; NT, not tested. ∗, previously published epitopes (see references 2, 19, 20, 25, 26, 35, 36, 37, 41, 50, and 51).

TABLE 12.

List of HLA B44-restricted putative CTL epitopes from HCMV antigens assessed by ELISPOT assays^a

Antigen	Peptide sequence	Result for donor:
		CS	TK
pp50	YEQHKITSY	67 ± 43	—
	SEDSVTFEF	—	—
pp65	SEHPTFTSQY	—	—
	CEDVPSGKLF	—	—
	NEIHNPAVF	—	—
	RETVELRQY	—	—
	QEPMSIYVY	—	—
pp71	AEVVARHNPY	—	—
gH	SEALDPHAF	—	—
	RENTTQCTY	—	—
	DDVLFALDPY	—	—
US11	SESLVAKRY	—	—
	LELEIALGY	—	—
UL16	DEIIGVAFTW	—	—
	NESVVDLWLY	—	—
UL18	GEINITFIHY	—	—
	TENGSFVAGY	283 ± 43	—

^aData presented as mean number ± SE of SFC/10⁶ PBMC. —, no response detected; NT, not tested.

TABLE 8.

List of HLA B7-restricted putative CTL epitopes from HCMV antigens assessed by ELISPOT assays^a

Antigen	Peptide sequence	Result for donor:
		MB	JG
pp65	HVRVSQPSL	359 ± 211	—
	QARLTVSGL	1,168 ± 297	125 ± 5
	EPMSIYVYAL	—	—
pp150	NVRRSWEEL	1,483 ± 95	19 ± 14
	WPRERAWAL	—	23 ± 10
	KARDHLAVL	1,027 ± 116	—
	SPWAPTAPL	232 ± 31	65 ± 13
	RPSTPRAAV	358 ± 215	10 ± 4
IE-1	KARAKKDEL	283 ± 54	—
gH	SPRTHYLML	—	—
	FPDLFCLPL	—	—
	SPRTHYLMLL	—	—
	MPRCLFAGPL	—	—
	TPMLLIFGHL	—	—
	APYQRDNFIL	—	—
US11	TPRVYYQTL	—	—
	APVAGSMPEL	—	—
UL16	YPRPPGSGL	—	—

^aData are presented as mean numbers of SFC/10⁶ PBMC ± SE. —, no response detected.

TABLE 5.

List of HLA A3-restricted putative CTL epitopes from HCMV antigens assessed by ELISPOT assays^a

Antigen	Peptide sequence	Result for donor:
		MW	CS	TK
pp65	QVIGDQYVK	—	—	—
	LLLQRGPQY	—	—	—
	RVTGGGAMA	—	—	—
	GVMTRGRLK	—	—	—
pp28	LVEPCARVY	—	NT	—
	GIKHEGLVK	—	—	—
	ELLAGGRVF	—	—	—
	RLLDLAPNY	—	—	—
	ELLGRLNVY	—	—	—
pp50	TLLNCAVTK	673 ± 3	—	—
	TVRSHCVSK	1,178 ± 63	—	—
pp150	KMTATFLSK	—	—	—
	IMREFNSYK	—	—	—
gH	SLRNSTVVR	—	—	—
	ALALFAAAR	—	—	—
	QLNRHSYLK	—	—	—
	RLFPDATVP	—	—	—
	AIIGIYLLY	—	—	—
	RLNTYALVSK	—	—	—
	LVRLVYILSK	—	—	—
	YLMDELRYVK	—	—	—
	ELYLMGSLVH	—	—	—
	ALTVSEHVSY	—	—	—
IE-1	KLGGALQAK	—	—	—

^aData are presented as mean numbers of SFC/10⁶ PBMC ± SE. —, no response detected; NT, not tested.

TABLE 6.

List of HLA A24-restricted putative CTL epitopes from HCMV antigens assessed by ELISPOT assays^a

Antigen	Peptide sequence	Result for donor:
		RE	MB	RK	PP	TD
pp65	VYALPLKML∗	—	—	—	—	—
	QYDPVAALF*	—	—	—	—	—
	VYYTSAFVF	—	—	—	—	—
pp71	EYIVQIQNAF	—	—	—	—	—
gB	AYIYTTYLL	—	—	—	—	—
	SYENKTMQL	—	—	—	—	—
	AYEYVDYLF	—	—	—	—	—
	CYSRPVVIF	—	—	—	—	—
gH	NYLDLSALL	—	—	—	—	—
	SYVVTNQYL	—	—	—	—	—
	SYLKDSDFL	—	—	—	—	—
	TYALVSKDL	—	—	—	—	—
	SYRSFSQQL	—	—	—	—	—
	TYGRPIRFL	—	—	—	—	—
	YYVFHMPRCL	—	—	—	—	—
	MYMHDSDDVL	—	—	—	—	—
IE-1	QYILGADPL	—	NT	—	—	—
	KYTQTEEKF	—	NT	—	—	—
US2	VYVTVDCNL	581 ± 72	—	98 ± 52	—	—
US3	VYLFSLVVL	388 ± 59	—	63 ± 4	55 ± 16	—
US11	YYVECEPRCL	—	—	37 ± 9	—	—
US6	LYLCCGITL	—	—	120 ± 15	130 ± 49	—
UL16	LYPRPPGSGL	497 ± 158	—	85 ± 18	—	—
	LYTSRMVTNL	268 ± 102	—	67 ± 15	—	—

^aData are presented as mean numbers of SFC/10⁶ PBMC ± SE. —, no response detected; NT, not tested; ∗, previously published epitope (see references 2, 19, 20, 25, 26, 35–37, 41, 50, and 51).

TABLE 9.

List of HLA B8-restricted putative CTL epitopes from HCMV antigens assessed by ELISPOT assays^a

Antigen	Peptide sequence	Result for donor:
		SC	MW	JP	PH	TC	SW	PP
pp65	RRRHRQDAL	—	—	—	—	—	—	—
	QPKRRRHRQ	—	—	—	—	—	—	—
	VLCPKNMII	—	—	—	—	—	—	—
	LCPKSIPGL	—	—	—	—	—	—	—
pp150	WPRERAWAL	—	—	—	—	—	—	—
IE-1	KARAKKDEL	—	—	—	—	—	—	—
	VMKRRIEEI	—	—	—	—	—	—	—
	RHRIKEHML	—	—	—	—	—	—	—
	ELRRKMMYM	1,383 ± 142	74 ± 20	555 ± 24	595 ± 64	2,327 ± 5	68 ± 18	—
	QIKVRVDMV	2,960 ± 112	63 ± 37	275 ± 34	1,931 ± 619	2,042 ± 47	178 ± 2	—
	ELKRKMMYM	1,383 ± 313	1,992 ± 521	441 ± 31	431 ± 57	2,122 ± 5	685 ± 25	232 ± 39

^aData are presented as mean numbers of SFC/10⁶ PBMC ± SE. —, no response detected; NT, not tested; ∗, previously published epitope (see references 2, 19, 20, 25, 26, 35–37, 41, 50, and 51).

Analysis of HCMV-specific CD8⁺-T-cell responses using cytotoxicity assays.

To further characterize the CD8⁺-T-cell epitopes mapped in this study, a large panel of HCMV-specific polyclonal or clonal CTL lines were generated. These CTL lines were generated not only against previously recognized HCMV antigens but also against those antigens from which a number of novel epitopes have been mapped in this study. As with the ELISPOT analysis, we targeted responses that were restricted through the common Caucasian HLA class I alleles. Representative data from 23 different polyclonal or clonal CTL lines specific for CD8⁺-T-cell epitopes restricted through HLA A1, A2, B8, B27, B35, and B44 established from the donors SB, PP, CS, SE, RK, MB, TC, RE, MW, and SC are shown in Fig. 2. The donor SB was chosen to represent data pertaining to HLA A2-restricted epitopes since this donor showed one of the broadest ranges of responses across many antigens. Figure 2A shows the data from CTL lines specific for CD8⁺-T-cell epitopes restricted through the HLA A2 allele. Of the six HLA A2-restricted epitopes from gB, IE-1, pp65, and pp150, five epitopes recalled strong CTL responses, while a low level of CTL activity was observed in the polyclonal CTL line established against the GQTEPIAFV (pp150) epitope. PBMC from other HLA A2-positive donors (e.g., donor SE) that showed responses to peptides in ELISPOT assays were also used to generate T-cell lines and clones (Fig. 2A and data not shown).

FIG. 2.

Recognition of HCMV peptide epitopes by T cells from healthy seropositive donors. PBMC were cocultivated with peptide-sensitized (20 μg/ml) autologous PBMC at a ratio of 2:1 for 7 days. On day 7 these cultures were restimulated with autologous γ-irradiated Epstein-Barr virus-transformed LCLs sensitized with peptide epitopes. On day 10 these T-cell lines were used as polyclonal effectors in a standard ⁵¹Cr release assay against peptide-sensitized autologous PHA blasts. An effector-to-target ratio of 10:1 was used in these assays. Panel A shows data for HLA A2-restricted epitopes, while panel B shows data for HLA A1-, A24-, B8-, B27-, B35-, and B44-restricted epitopes. T-cell lines were established from donors SB and SE (A) and CS, SC, TC, MB, PP, RE, RK, and MW (B). Results are expressed as percent specific lysis. Complete T-cell epitope sequence, source of the antigen, and the HLA restriction for each of these epitopes is shown on the y axis. Representative data from a minimum of three different sets of experiments are shown.

CTL lines specific for other epitopes mapped within pp65, pp50, and IE-1 also showed strong cytolytic activity against peptide-sensitized autologous PHA blasts (Fig. 2B), while the CTL responses to the epitopes from pp28, US6, and UL18 were comparatively weaker. Consistent with the ELISPOT data, CTL lines specific for pp65 and IE-1 epitopes were readily generated and displayed strong cytolysis, whereas T-cell lines specific for epitopes with subdominant antigens, such as pp28, US2, US3, UL18, US11, gH, and gB, showed low to undetectable levels of CTL lysis (Fig. 2B; summarized in Table 13). Among the subdominant antigens, however, the HLA A1-restricted epitope (VTEHDTLLY) from pp50 was of particular interest. This epitope was one of the most dominant responses among the HLA A1-positive donors included in this study (Fig. 2B). Based on ELISPOT assays and/or in vitro cytotoxicity assays, we mapped a total of 63 CD8⁺-T-cell epitopes (Table 13).

TABLE 13.

Comprehensive list of HLA class I-restricted CTL epitopes from HCMV antigens mapped by ELISPOT and CTL assays

HCMV antigen	Peptide sequence	HLA restriction	ELISPOT responsea	CTL responseb
gB	RIWCLVVCV	A2	+	−
	AVGGAVASV	A2	+	+
	LDEGIMVVY	A1	++	NT
	ATSTGDVVY	A1	++	NT
	NTDFRVLEL	A1	+++	NT
gH	LIFGHLPRV	A2	+	NT
	LMLLKNGTV	A2	+	NT
IE-1	SLLSEFCRV	A2	++	−
	VLAELVKQI	A2	+	++
	VLEETSVML	A2	+++	++
	YILEETSVM	A2	++	ND
	CLQNALDIL	A2	++	NT
	KARAKKDEL	B7	++	ND
	QIKVRVDMV	B8	+++	+++
	ELRRKMMYM	B8	++	+++
	ELKRKMMYM	B8	++	+++
	IKEHMLKKY	A1	+++	NT
	DEEEAIVAY	A1	+++	NT
	CVETMCNEY	A1	+++	NT
	RRKMMYMCY	B27	+++	NT
pp28	LLIDPTSGL	A2	+	NT
	ARVYEIKCR	B27	+	+
pp50	LLNCAVTKL	A2	+	NT
	VTEHDTLLY	A1	++	+++
	YEQHKITSY	B44	+	−
	RGDPFDKNY	A1	++	NT
	GLDRNSGNY	A1	++	NT
	TLLNCAVTK	A3	++	NT
	TVRSHCVSK	A3	+++	NT
pp150	ALVNAVNKL	A2	+	NT
	ALVNFLRHL	A2	+	NT
	LIEDFDIYV	A2	+	NT
	NVRRSWEEL	B7	++	−
	KARDHLAVL	B7	+++	−
	SPWAPTAPL	B7	++	−
	RPSTPRAAV	B7	++	−
	WPRERAWAL	B7	+	NT
	GQTEPIAFV	A2	++	−
pp65	RLLQTGIHV	A2	+	−
	NLVPMVATV	A2	++	++
	LMNGQQIFL	A2	++	−
	MLNIPSINV	A2	++	−
	RIFAELECV	A2	++	−
	ILARNLVPM	A2	++	++
	HVRVSQPSL	B7	++	−
	QARLTVSGL	B7	++	−
	IGDQYVKVY	A1	++	NT
	TVQGQNLKY	A1	++	NT
	YRIQGKLEY	A1/B27	++/+++	NT
	FPTKDVAL	B35	++	+++
	YSEHPTFTSQY	A1	++	NT
US2	SMMWMRFFV	A2	++	+
	TLLVLFIVYV	A2	++	NT
	LLVLFIVYV	A2	+	NT
	VYVTVDCNL	A24	++	−
US3	YLFSLVVLV	A2	+	NT
	LLFRTLLVYL	A2	+	−
	TLLVYLFSL	A2	+	−
	VYLFSLVVL	A24	++	−
US11	YYVECEPRCL	A24	+	−
UL18	TENGSFVAGY	B44	++	−
US6	LYLCCGITL	A24	++	−
UL16	LYPRPPGSGL	A24	++	−
	LYTSRMVTNL	A24	++	−

^aELISPOT response: +, 10 to 100 SFC/10⁶ PBMC; ++, 101 to 500 SFC/10⁶ PBMC; +++, >500 SFC/10⁶ PBMC.

^bCTL response −, <10% specific lysis; +, 10 to 20% specific lysis; ++, 21 to 50% specific lysis; +++, >50% specific lysis; NT, not tested. This list also includes previously published epitopes (see references 2, 14, 16, 19, 20, 25, 26, 35, 37, 41, 50, and 51).

CTL recognition of virus-infected target cells.

Although we showed that synthetic peptides could recall memory T-cell responses, it was important to demonstrate that these CTL effectors were also able to recognize naturally processed HCMV antigens expressed in the context of heterologous gene expression from vaccinia virus or from HCMV infection itself. Clonal or polyclonal T-cell lines were chosen to cover a cross-section of strong and weak cytolytic responses from both dominant and subdominant antigens (pp28, pp50, pp65, and IE-1). HLA-matched LCLs that were infected with recombinant vaccinia vectors expressing individual HCMV antigens (pp28, pp65, and IE-1) along with HCMV-infected JS fibroblasts (HLA A1, A2, B8, and B51) were used as target cells. The reactivity of clones or polyclonal CTL lines specific for epitopesNLVPMVATV (SB 62.3; HLA A2 restricted), VLEETSVML (SB 75.20; HLA A2 restricted), and ELRRKMMYM (SC ELRR; HLA B8 restricted) is illustrated in Fig. 3A. These data demonstrate that CTL clones SB 62.3 and 75.20 recognized target cells infected with recombinant vaccinia encoding pp65 (Vacc.pp65) and IE-1 (Vacc.IE-1), respectively. The polyclonal CTL line, SC ELRR, generated from an HLA B8-positive donor, showed a comparable level of lysis of autologous LCLs infected with Vacc.IE-1 (Fig. 3A). In contrast, the HLA B27-restricted polyclonal CTL line specific for the pp28 epitope (ARVYEIKCR) and generated from donor RK did not recognize autologous LCLs infected with Vacc.pp28 (data not shown). The endogenous processing of CTL epitopes was also confirmed by assessing the CTL lysis of HCMV-infected target cells. Generally, HCMV-specific CTL lines showed lower levels of lysis towards the virus-infected target cells than lysis of peptide-sensitized or recombinant vaccinia-infected target cells. Data presented here demonstrate that CTL epitopes from pp65, pp50, and IE-1 are naturally processed within virus-infected cells (Fig. 3B), suggesting that CTL epitopes from immediate-early antigens can be endogenously processed in target cells expressing structural antigens such as pp65.

FIG. 3.

Specific lysis by HCMV-specific T cells of autologous LCLs infected with recombinant vaccinia virus encoding individual HCMV antigens (Vacc.pp65 and Vacc.IE-1) (A) or HCMV-infected HLA-matched fibroblasts (B). LCLs were infected for 12 to 14 h (MOI, 10:1) with vaccinia constructs and processed for standard ⁵¹Cr release assay. T-cell clones or polyclonal lines from donors SB (pp65 specific, SB 62.3 and IE-1 specific, SB 75.20) and SC (IE-1 specific, SC ELRR) were used as effectors in the assay (A). Fibroblasts were infected with AD-169 strain of HCMV for 14 to 16 h (MOI, 5:1) and used as targets in the CTL assay. CTL clones or polyclonal lines from donors SB (IE-1 specific, SB 75.20; pp65 specific, SB NLVP), PP (pp50 specific, PP VTEH), and TC (IE-1 specific TC, ELRR and IE-1 specific, TC-QIKV) were used as effectors in this assay (B). Results are expressed as percent specific lysis observed at an effector-to-target ratio of 5:1. Representative data from a minimum of three different sets of experiments are shown.

DISCUSSION

The present study provides, for the first time, an extensive analysis of CD8⁺-T-cell responses to 14 HCMV-encoded proteins. Using computer-based algorithms, ELISPOT, and cytotoxicity assays, we have identified a number of novel HCMV-encoded proteins as targets for CD8⁺-T-cell responses. In addition, we have described numerous peptide sequences from within these proteins that T-cell responses are directed toward. These observations have a number of important implications for enhancing the current understanding of immune regulation of a latent viral infection and for future vaccines designed to control HCMV-associated pathogenesis. In a recent review, Plotkin (31) proposed that the HCMV vaccine should combine in one single regimen all those antigens that might provide protection. The data presented here clearly indicate that in healthy immune individuals, CD8⁺-T-cell responses are directed towards multiple antigens (pp65, pp50, pp28, pp150, IE-1, US2, US3, US6, US11, UL16, UL18, gB, and gH), which may all be crucial in controlling HCMV reactivation. Moreover, more than 40% of the T-cell reactivity was directed towards antigens other than pp65 and IE-1. The immunological relevance of CD8⁺-T-cell responses to antigens such as gB, gH, pp50, pp28, US2, US3, US6, US11, UL16, and UL18, some of which have not previously been identified as potential targets for CTL response, will need to be further addressed. In particular, the importance of these CTL responses to HCMV-related pathogenesis in transplant patients needs to be reassessed. Here we have provided tools which will allow for a much broader analysis of the HCMV response in the future. For example, the identification of CD8⁺-T-cell epitopes within those antigens involved in reactivation may, in the future, be exploited for the restoration of immunity and control of virus replication in immunocompromised patients. Moreover, the unexpected detection of T-cell responses to HCMV-encoded immunomodulators such as US2, US3, and UL18 may help to clarify the mechanisms of latency and immune evasion. At present, it is unclear how the responses described here contribute towards the overall immune regulation of HCMV infection and how these responses are generated. Since the immunomodulating proteins are known to block endogenous antigen presentation by virus-infected cells, it is possible that most of the T-cell responses detected in this study were generated by cross-priming. Indeed, previous studies with other virus-encoded antigens, which are poorly processed by virus-infected cells, have shown that much of the T-cell response to these antigens is generated through the cross-priming pathway (5, 6). Previous studies have also shown that not only are the MHC proteins degraded after US2 binding but US2 itself is also degraded by proteasomes (13). Thus, it is also plausible that US2 epitopes may be processed and presented by virus-infected cells. If the CD8⁺ T cells specific for US2, US3, or UL18 can recognize virus-infected cells, it may be pertinent to include epitopes from these antigens in any future vaccine designed to control HCMV infection.

To ensure maximal coverage in the targeted community, it would be necessary to include multiple peptide epitopes in a vaccine formulation. It is estimated that the peptide epitopes defined to date, including those defined in the present study, would span between 85 and 90% of the ethnically diverse population. One approach might be to combine defined epitopes into a single formulation through the use of “polyepitope” technology (48). This technology is designed to express minimal CTL epitopes as a continuous “string of beads” which are fused together to construct a recombinant or synthetic polyepitope protein (rather than in their natural context within a protein).

The results presented here strongly suggest that broadly directed CTL responses to multiple epitopes may be essential in controlling the HCMV replication. Indeed, recent studies with other human viral infections have also indicated that individuals with broad T-cell reactivity were able to clear the virus infection more efficiently than those who displayed a more narrowly focused T-cell response to a single antigen or a limited number of epitopes (4, 47). In addition, we have significantly expanded the repertoire of candidates now available for the design of a peptide-based vaccine against HCMV. Future investigations will need to assess the levels of protection afforded by responses directed toward these epitopes. After this is achieved, the development of efficacious prophylactic and therapeutic anti-HCMV formulations could be realized.

Antigen or peptide sequenceB27/B8+++ TMYGGISLLA2++  LLSEFCRVLA2++  VLEETSVMLA2++  YILEETSVMA2++  CLQNALDILA2++  ILDEERDKVA2++  IKEHMLKKYA1NT  DEEEAIVAYA1NT  CVETMCNEYA1NT  KLGGALQAKA3+++  QYILGADPLA24+++  KYTQTEEKFA24−  KARAKKDELB7/B8−/−  VMKRRIEEIB8+  RHRIKEHMLB8−  ELRRKMMYMB8++  QIKVRVDMVB8+  ELKRKMMYMB8−  RRKMMYMCYB27+  RRIEEICMKB27− US2 LLVLFIVYVA2−  SMMWMRFFVA2++  TLLVLFIVYVA2−  VYVTVDCNLA24− gB RIWCLVVCVA2−  QMLLALARLA2+  GLDDLMSGLA2++  IILVAIAVVA2+  DLDEGIMVVA2−  NLFPYLVSAA2++  AVGGAVASVA2−  YINRALAQIA2++  CYSRPVVIFA24−  KMTATFLSKA3+++  IMREFNSYKA3+++  VKESPGRCYA1NT  LDEGIMVVYA1NT  ATSTGDVVYA1NT  NTDFRVLELA1NT  AYIYTTYLLA24−  SYENKTMQLA24−  AYEYVDYLFA24−

TABLE 7.

List of HLA A26-restricted putative CTL epitopes from HCMV antigens assessed by ELISPOT assays

Antigen	Peptide sequence	Result for donor TD
pp65	DIYRIFAEL	—a
	DVPSGKLFM	—
	YVKVYLESF	—
	DIDLLLQRG	—
	TVQGQNLKY	—
gH	ETFPDLFCL	—
	DTQGVINIMY	—
	AIIGIYLLY	—
	DLTETLERY	—

^a—, no response detected.

TABLE 10.

List of HLA B27-restricted putative CTL epitopes from HCMV antigens assessed by ELISPOT assaysr^a

Antigen	Peptide sequence	Result for donor:
		RK	JDa
pp28	CRYKYLRKK	—	—
	ARVYEIKCR	155 ± 78	—
pp50	TRVKRNVKK	—	—
pp65	RRDHSLERL	—	—
	RRRHRQDAL	—	—
	YRIQGKLEY	—	1896 ± 191
IE-1	RRKMMYMCY	—	3500 ± 284
	RRIEEICMK∗	—	—
gH	GRCQMLDRR	—	—

^aData are presented as mean numbers of SFC/10⁶ PBMC ± SE. —, no response detected; ∗, previously published epitope (see references 2, 19, 20, 25, 26, 35–37, 41, 50, and 51).

TABLE 11.

List of HLA B35-restricted putative CTL epitopes from HCMV antigens assessed by ELISPOT assays^a

Antigen	Peptide sequence	Result for donor:
		RE	JDu	CS	TK	TC	SB	JW
pp65	FPTKDVAL∗	NT	250 ± 38	NT	—	170 ± 85	897 ± 74	1405 ± 20
pp150	WPRERAWAL	—	—	—	—	—	NT	—

^aData are presented as mean numbers of SFC/10⁶ PBMC ± SE. —, no response detected; NT, not tested; ∗, previously published epitope (see references 2, 19, 20, 25, 26, 35–37, 41, 50, and 51).