Functional Heterogeneity and High Frequencies of Cytomegalovirus-Specific CD8⁺ T Lymphocytes in Healthy Seropositive Donors

Abstract

Human cytomegalovirus (HCMV) infection is largely asymptomatic in the immunocompetent host, but remains a major cause of morbidity in immunosuppressed individuals. Using the recently described technique of staining antigen-specific CD8⁺ T cells with peptide-HLA tetrameric complexes, we have demonstrated high levels of antigen-specific cells specific for HCMV peptides and show that this may exceed 4% of CD8⁺ T cells in immunocompetent donors. Moreover, by staining with tetramers in combination with antibodies to cell surface markers and intracellular cytokines, we demonstrate functional heterogeneity of HCMV-specific populations. A substantial proportion of these are effector cytotoxic T lymphocytes, as demonstrated by their ability to lyse peptide-pulsed targets in “fresh” killing assays. These data suggest that the immune response to HCMV is periodically boosted by a low level of HCMV replication and that sustained immunological surveillance contributes to the maintenance of host-pathogen homeostasis. These observations should improve our understanding of the immunobiology of persistent viral infection.

Human cytomegalovirus (HCMV) is a complex DNA virus from the beta subgroup of the herpesvirus family (4). Primary infection is usually localized to epithelial cells of the salivary glands, and because HCMV employs several highly specialized strategies which interfere with the immune system, it is rarely, if ever, eliminated from the infected host. Instead the virus disseminates throughout the host and establishes latency in a number of cell types. Recently, CD33⁺ progenitors expressing markers of dendritic and myeloid lineage have been identified as major reservoirs of latent HCMV (20, 41). A small proportion of these cells harbor CMV-associated latent transcripts, but they fail to support the replication of productive virus (15). Instead, differentiated macrophages are permissive to viral replication both in vitro and in vivo.

The immune system plays a crucial role in the control of HCMV replication in the chronically infected host. In addition, CD8⁺ T cells and NK cells are essential to the control of murine cytomegalovirus (MCMV) replication (33). For HCMV, there is also evidence that CD8⁺ cytotoxic T lymphocytes (CTLs) represent a fundamental component of protective immunity. In allogeneic bone marrow transplant (BMT) recipients, the delayed regeneration of HCMV-specific CD8⁺ T lymphocytes following transplantation has been correlated with HCMV disease (24, 36). Moreover, the adoptive transfer of HCMV-specific CD8⁺ T-cell clones has led to the restoration of virus-specific immunity in BMT patients (44). However, relatively little is known about the magnitude and specificity of the HCMV-specific CD8⁺ immune response in immunocompetent individuals (2, 3, 35).

We performed a detailed analysis to quantitate, determine the phenotype, and analyze the function of HCMV-specific CD8⁺ CTLs in healthy HCMV-seropositive individuals. We utilized HLA-A*0201- and HLA-B*0702-restricted epitopes derived from the lower matrix protein, pp65, which have been shown by limiting dilution analyses (LDA) to represent immunodominant epitopes (49). Soluble tetrameric major histocompatibility complex (MHC) class I molecules were used to quantitate and determine the phenotype of HCMV-specific CD8⁺ T lymphocytes (1). In addition, we compared the frequencies of HCMV-specific CD8⁺ T lymphocytes obtained by using tetramers with that derived by LDA, since the latter may underestimate precursor frequencies in persistent viral infections (6, 30). In three donors, tetramer-positive T lymphocytes were isolated, and cytotoxicity assays were performed with discrete phenotypic subsets to identify the phenotype of cells which mediate immediate lytic activity.

MATERIALS AND METHODS

Subjects.

Peripheral blood samples were collected from laboratory donors and anonymous donations from the National Blood Service, Oxford, United Kingdom. Seventeen healthy donors were studied, and HCMV seropositive status was confirmed by an immunoglobulin G enzyme-linked immunosorbent assay. None of the donors had a history of HCMV-associated disease, and the precise dates of primary infection are unknown. HLA typing was performed by PCR amplification with allele-specific primers (5).

Isolation of PBMCs.

Peripheral blood mononuclear cells (PBMCs) were isolated from heparinized venous blood by Ficoll-Hypaque density gradient centrifugation. Cells were washed in RPMI supplemented with 10% fetal calf serum (R/10), 100 IU of penicillin per liter, and 100 mg of streptomycin per liter prior to cellular or flow cytometric analysis.

Generation of MHC class I tetramers.

The HLA-B*0702 construct containing the BirA substrate peptide (BSP) was constructed by PCR amplification with an HLA-B*0702 plasmid (gift from Linda Barber) as a template.

In brief, the 5′ ATTGCAGACATATGGGCTCCCACTCCATGAC and 3′ TGTGATAAGCTTAATGCCATTCAATTTTCTGTGCTTCAAAAATATCA TTCAGGGATCCTGGCTCCCATCTCAGGGTGAGG primers yielded a truncated PCR product, modified at the C terminus by the addition of a glycine-serine linker and the BSP. The construct was cloned into the expression vector derivative, pGMT7. The HLA-A*0201-BSP construct was a gift from Bent Jakobsen. Both expression plasmids were transformed into Escherichia coli strain BL21(DE3)pLysS, and protein expression and purification were carried out according to standard protocols (1, 10). Thirty milligrams of soluble MHC class I heavy chain was refolded in vitro in the presence of 10 mg of beta-2 microglobulin (β2M; kind gift from D. Garboczi) and 10 mg of relevant peptide. The peptide epitopes included the HLA-A*0201-restricted epitope NLVPMVATV (amino acids 495 to 503 of the lower matrix protein pp65) and the HLA-B*0702 epitope TPRVTGGGAM (amino acids 417 to 426 of the lower matrix protein pp65). Refolded MHC class I complexes were buffer exchanged into a mixture of 100 mM Tris (pH 7.5), 20 mM NaCl, and 5 mM MgCl₂ prior to biotinylation in the presence of BirA holoenzyme. Biotinylated MHC-peptide complexes were purified by fast-performance liquid chromatography and ion-exchange chromatography prior to addition of avidin-phycoerythrin (PE) conjugate (1).

Triple and four-color staining of peripheral blood-derived T lymphocytes.

PBMCs (10⁶) were costained with 0.5 μl of fluorescein isothiocyanate (FITC)-conjugated anti-CD3 (Dako, Carpinteria, Calif.), 0.5 μg of anti-CD8-Tricolor (Caltag, South San Francisco, Calif.), and 0.5 μl of PE-conjugated tetrameric complex. A total of 700,000 events were collected per sample by using CELLQUEST software. The percentage of tetramer-staining cells was calculated within the CD3⁺ CD8⁺ gate.

For analysis of the membrane phenotype, 10⁶ PBMCs were incubated with 0.5 μg of tetramer, 0.5 μl of Tricolor-conjugated anti-CD8, and 0.5 μg of one of a panel of FITC-conjugated antibodies for 30 min at 4°C. The panel comprised anti-CD45RA, anti-HLA-DR, anti-CD7, anti-CD57, anti-CD27 (Becton Dickinson, CA), anti-CD11b, anti-CD28 (Immunotech, France), anti-CD45RO, anti-CD38, and anti-CD62L (Dako) monoclonal antibodies (MAbs). Following incubation, cells were washed in phosphate-buffered saline (PBS) supplemented with 0.1% bovine serum albumin (BSA) and fixed in PBS–5% formaldehyde. CD8^bright cells were gated to estimate the proportion of tetramer-positive cells expressing a given cell surface antigen.

For four-color fluorescence-activated cell sorter (FACS) analysis, CD45RA-FITC, CD8-APC and CD45RO-PerCP (Becton Dickinson) were used in conjunction with tetrameric complexes.

Intracellular staining of tetramer-positive cells.

PBMCs were stained with tetrameric complexes for 15 min at 37°C and subsequently incubated at 37°C in R10 with or without 10 μM peptide for 6 h. Brefeldin A, at a final concentration of 10 μg/ml, was added to samples during the second hour of incubation. Cells were then washed in PBS and fixed and permeabilized in FACS permeabilization buffer for 10 min. After washing, staining was performed for 30 min at room temperature in the dark by using a panel of FITC or antigen-presenting cell (APC)-conjugated MAbs: isotype control (Dako), anti-human gamma interferon (hIFN-γ) (Pharmingen), anti-hMIP-1β (R&D systems), anti-human tumor necrosis factor alpha (hTNF-α) (Becton Dickinson), and anti-human perforin (Pharmingen). Cells were washed and fixed in 5% formaldehyde.

Generation of HCMV-specific CD8⁺ T lymphocytes in LDAs.

LDA microcultures were set up and maintained as previously described (48). Replicate microcultures (n = 27) of purified CD8⁺ T cells were set up in 96-well plates in which the number of responder T cells per well was progressively reduced over an appropriate range of dilutions in RPMI 1640 supplemented with HCMV-seronegative 10% human AB serum (R/HuAB). Human interleukin 2 (IL-2) was added to 5 IU/ml. Cultures were refed with R/HuAB supplemented with 5 IU of IL-2 per ml on days 5 and 10. Autologous PBMCs served as stimulator cells; these were pulsed for 1 h with pp65 peptide, irradiated (2,400 rad), and added at 5 × 10⁴ cells per well. On day 14, by using split-well analysis, the cells in each well were resuspended and divided into aliquots that were assayed simultaneously for cytotoxicity against radiolabelled target cells in 4-h ⁵¹Cr-release assays. Target cells comprised 4 × 10³ cells/well of autologous and MHC mismatched lymphoblastoid B-cell lines (L-BCLs) that were pulsed or not with the HLA-A*0201-restricted NLVPMVATV peptide and the HLA-B*0702-restricted TPRVTGGGAM peptide (40 mg/ml) for 1 h or infected for 18 h with vaccinia virus pp65 or control vaccinia virus T7 (multiplicity of infection of 10). The LDA results were analyzed as previously described (49). Bulk peptide-specific CD8⁺ T-lymphocyte lines were generated as previously described (48).

Purification of tetramer-reactive CD8⁺ T lymphocytes.

Cells (4 × 10⁷) were stained with HLA-A*0201^495–503 tetramer, anti-CD8 MAb, and specific phenotypic markers for 30 min at 4°C. Samples were washed twice in PBS prior to FACSorting. Purified cells were sorted into tubes containing R/HuAB, where they were maintained prior to analysis.

T-cell cytotoxicity assays.

Autologous L-BCLs were used as target cells in standard ⁵¹Cr-release CTL assays. Target cells (5 × 10³) were pulsed with peptide at a final concentration of 5 μM. Effector cells were added to the peptide-pulsed targets at various effector/target (E:T) ratios. For fresh killing, assay plates were incubated for 15 h before harvest. All samples were established in duplicate, and specific ⁵¹Cr release was calculated as ([experimental release − spontaneous release]/[maximum release − spontaneous release]) × 100%.

Culture of CD45RA^high and CD45RO^high populations of CD8⁺ tetramer-positive cells.

PBMCs were stained and sorted as described previously. A total of 20,000 sorted cells were incubated with 20,000 irradiated, autologous L-BCLs, which were pulsed with 50 μM peptide. After 3 days, Lymphocult T (Biotest AG) was added, and at day 14, a cytotoxicity assay was performed.

Staining of tetramer-positive T cells with TCR V region-specific MAbs.

Tetramer-positive T cells were stained with T-cell receptor (TCR) V region-specific MAbs according to previously described protocols (51). In brief, 10⁶ PBMCs were stained with a panel of TCR-specific antibodies for 30 min at 4°C, followed by incubation with a second layer of rabbit anti-mouse antibody conjugated to FITC for 20 min at 4°C. Cells were then stained with anti-CD8-Tricolor and tetrameric complexes for 40 min at 4°C, washed, and fixed prior to analysis.

Cloning and sequencing of TCR sequences.

TCR amplification and cloning were performed according to previously described protocols. In brief, RNA was isolated from PBMCs and transcribed into cDNA by using oligo(dT). A 5′ TCRBV14 primer (TCTCGAAAAGAGAAGAGGAAT) and a 3′ TCRBC primer (CGTTTGTCGTCGACCTCCTTCCCATTCACC) were used to amplify the TCRBV14 gene under standard PCR conditions (29). TCR transcripts were cloned by blunt-end ligation into pGEM-T Easy vector (Promega) and transformed into JM109 competent cells. A total of 20 ampicillin-resistant colonies were selected and sequenced by the dideoxy chain termination reaction.

RESULTS

HCMV peptide-containing tetramers are specific for HCMV-specific CD8⁺ T lymphocytes.

Two tetrameric complexes containing HCMV epitopes derived from the matrix protein pp65 were generated. Both tetramers specifically stained the appropriate T-cell lines and clones (Fig. 1). A total of seven PBMC samples from HCMV-seronegative donors were stained, and in all instances, the binding of tetramers was minimal (less than 0.01%) (Fig. 2).

FIG. 1.

Staining of CD8⁺ T-lymphocyte cultures with HCMV-specific HLA-peptide tetramers. Staining of CD8⁺ T-lymphocyte cultures with the tetrameric complex HLA-A*0201/pp65^495–503. (a) A polyclonal CD8⁺ T-lymphocyte line specific for HCMV pp65^495–503 is stained with a high degree of fluorescence by the tetramer. A control HLA-A*0201-restricted CTL line generated against influenza virus matrix peptide 58–66 does not show staining with this tetramer. (b) Tetramer HLA-B*0702/pp65^417–426 binds specifically to a CTL clone grown on the pp65^417–426 peptide, but fails to recognize a nonspecific HLA-B*0702-restricted CTL clone.

FIG. 2.

Representative data of tetramer staining of PBMCs from HCMV-seropositive and -seronegative donors. PBMCs from HCMV-seropositive and -seronegative donors were tested for their ability to bind the tetrameric complexes used in this investigation. Representative dot plots where tetrameric staining is represented along the x axis and CD8 staining is represented along the y axis are summarized. Both HLA-B*0702/pp65^417–426 and HLA-A*0201/pp65^495–503 tetramers display their ability to stain HCMV-specific CD8⁺ T lymphocytes in seropositive donors, yet display minimal binding to CD8⁺ T cells from HCMV-seronegative individuals. Plots a to c represent HLA-A*0201⁺ CMV-seropositive donors, and these samples were stained with A*0201/pp65^495–503 tetrameric complexes. Plots d to f are representative HLA-B*0702⁺ CMV-seropositive donors, and these samples were stained with the B*0702/pp65^417–426 tetramer. Plots g and h are HLA-A*0201⁺ CMV-seronegative donors which were stained with the A*0201/pp65^495–503 tetramer. Plot i is an HLA-B*0702⁺ CMV-seronegative sample which was stained with the B*0702/pp65^417–426 tetramer.

Frequency of HLA-A*0201- and HLA-B*0702-restricted HCMV-specific CD8⁺ cells in healthy HCMV-seropositive donors.

Tetrameric HLA-peptide complexes were used to determine the frequency of HCMV-specific CD8⁺ T lymphocytes in PBMC samples from HCMV-seropositive donors. Donors were grouped on the basis of genetic homozygosity or heterozygosity for HLA-A*0201 or HLA-B*0702 alleles (Table 1). HCMV-specific CD8⁺ T lymphocytes were detectable in all HCMV-seropositive donors. Within the HLA-A*0201 group, between 0.22 and 4.44% of CD8⁺ T cells were stained with HLA-A*0201/pp65^495–503 tetramer, with an average value of 0.75%. There was large variation between donors in the number of tetramer-reactive CD8⁺ T cells detected and in donor 13, nearly 1 in 20 of peripheral CD8⁺ T lymphocytes are specific for the HCMV peptide pp65^495–503. In the HLA-A*0201 homozygous group, the mean percentage of tetramer-positive CD8⁺ T cells was 1.24% (range, 0.22 to 4.44%). In contrast, in donors heterozygous for HLA-A*0201 expression, the percentage was lower, with a mean of 0.17% (P < 0.1).

TABLE 1.

Quantitation of HCMV-specific CD8⁺ T lymphocytes in healthy donors by fluorescent staining with tetramer^a

Donor	Haplotype for specific MHC alleleb	% CD8+ tetramer-reactive cells
A*0201+		A*0201/pp65495–503
1	Heterozygous	0.12
2	Heterozygous	0.14
3	Heterozygous	0.06
4	Heterozygous	0.03
5	Heterozygous	0.31
6	Homozygous	0.41
7	Homozygous	0.24
8	Homozygous	0.22
9	Homozygous	0.24
10	Homozygous	0.41
11	Homozygous	2.03
12	Homozygous	1.30
13	Homozygous	4.44
B*0702+		B*0702/pp65417–426
6	Heterozygous	1.13
14	Heterozygous	4.43
15	Heterozygous	0.13
16	Heterozygous	1.40
17	Heterozygous	1.63
18	Heterozygous	5.02

^aPBMCs were costained with PE-conjugated MHC tetrameric complexes in conjunction with Tricolor-conjugated anti-CD8 MAb. Results are expressed as a percentage of total CD8⁺ T cells. 

^bDonor samples were separated according to homozygosity or heterozygosity for A*0201 or B*0702 alleles. 

Within HCMV-seropositive HLA-B*0702 individuals, the B*0702/pp65^417–426 tetrameric complex detected an average of 1.85% of CD8⁺ T lymphocytes (range, 0.13 and 5.0%). All of the HLA-B*0702 donors were heterozygous at this allele, and in comparison with donors who were heterozygous for HLA*A0201, the pp65^417–426 epitope evokes a larger CD8⁺ immune response than the HLA-A*0201-restricted pp65^495–503 epitope (P < 0.05). This is also suggested by analysis of donor 6, who is both HLA-A*0201 and HLA-B*0702 positive and in whom the frequency of CD8⁺ cells specific for the pp65^417–426 epitope is over twice that calculated for pp65^495–503.

The frequency of tetramer-positive cells was determined on more than one occasion for a number of donors. Blood samples were obtained from donor 3 on three occasions and from donors 13, 16, and 17 on four occasions over a period of 12 months. In all instances, fluctuations in tetramer staining were less than 8%.

Comparison of HCMV-specific CD8⁺ T-lymphocyte frequencies by using LDA and staining with tetrameric complexes.

LDA is a valuable technique for estimating precursor frequencies of antigen-specific T lymphocytes, but may significantly underestimate the true frequency in persistent viral infection (6, 30). The frequency of CD8⁺ T lymphocytes obtained by LDA was compared with that obtained by tetramer staining in five HCMV-seropositive donors (Table 2) and shows that the two techniques are not easily comparable. Interestingly, the discrepancy between the two approaches is greater in donors in whom there is a higher absolute value of HCMV-specific CD8⁺ T lymphocytes, as determined by tetramer staining (correlation coefficient, 0.95; P < 0.05). In donor 13, the frequency of HLA-A*0201/pp65^495–503-specific cells as calculated by LDA was 29 times lower than that estimated by using tetramers. It is clear that large populations of effector CD8⁺ cells may not be detected by LDA, and LDA profiles may not give a true reflection of the total number of antigen-specific CD8⁺ T cells.

TABLE 2.

Comparison of HCMV-specific CD8⁺ T-cell frequencies as determined by staining with fluorescent tetramers or by LDA^a

Donor	% Tetramer frequency (a)	% LDA frequency (b)	a/b ratiob
3	0.06	0.06	1
4	0.03	0.15	0.2
13	4.4	0.15	29
16	1.4	0.28	5
17	1.6	0.1	16

^aFrequencies of HCMV-specific T lymphocytes, as calculated by fluorescent staining and LDA, are reported as a percentage of CD8⁺ T cells. 

^bThe a/b ratio (tetramer staining/LDA values) denotes a correlation index between the individual techniques (i.e., values approaching 1 indicate close correlation). 

Phenotypic profiles of HCMV-specific CD8⁺ T lymphocytes.

Triple-staining flow cytometric analysis was performed to characterize the phenotype of HCMV-specific CD8⁺ T cells (Table 3). The results reveal a heterogeneous phenotype both within and between different individuals. Tetramer-positive cells were found within both the CD45RA and CD45RO subsets, and many cells were positive for both markers (Fig. 3). Activation markers such as HLA-DR, CD11b, CD57, and CD38 were expressed on a minority of cells, but with considerable differences between individuals. Many CTLs lacked expression of CD27 and CD28, and the percentage of cells lacking these markers was correlated between different individuals.

TABLE 3.

Phenotypic analysis of peripheral blood-derived CD8⁺ HCMV-specific T lymphocytes

Donor	% of cellsa
	CD45RA			CD450			HLA-DR	CD28	CD11b	CD57	CD7	CD27	CD38	CD62L
	High	Medium	Low	High	Medium	Low
1	29	50	21	29	46	25	13	63	44	16	85	67	30	63
2	29	50	21	24	53	24	0	60	22	33	71	25		75
5	25	10	65	63	12	25	29	48	30	41	57	75	38	62
6-A2	51	36	13	NDb	ND	ND	15	33	66	43	40	27	5	20
6-B7	29	61	11	13	62	25	6	31	77	57	76	24	11	ND
7	57	24	17	17	19	65	7	25	23	46	82	23	0	71
8	33	48	19	22	45	33	29	70	46	21	57	65	43	46
9	20	49	31	41	31	28	ND	ND	ND	ND	ND	ND	ND	ND
10	59	31	10	ND	ND	ND	23	92	72	24	98	95	69	95
11	13	17	70	81	12	7	8	18	84	93	94	19	12	33
12	70	24	6	6	20	74	33	17	50	51	96	31	19	28
13	14	68	18	18	69	13	3	21	8	26	96	17	3	28
14	30	33	38	17	31	52	3	64	38	23	93	68	39	56
15	37	52	11	9	59	32	8	54	41	25	76	44	17	41
16	18	51	31	33	54	14	6	77	94	78	96	74	18	20
17	22	43	35	36	46	18	20	24	84	82	98	21	12	50
Meanc	34	40	26	29	40	31	14	46	52	44	83	45	23	49

^aResults are expressed as percentage of tetramer-reactive CD8⁺ T cells. 

^bND, not determined. 

^cMean staining as calculated for individual phenotypic markers. 

FIG. 3.

Dual expression of CD45RA and CD45RO isoforms on tetramer-reactive CD8⁺ T lymphocytes from donor 13. Four-color FACS analysis, using a tetrameric complex in conjunction with anti-CD8, anti-CD45RA and anti-CD45RO MAbs, was performed with freshly isolated PBMCs from donor 13 in an attempt to estimate the proportion of cells which were positive for both the RA and RO isoforms of CD45. CD8⁺ tetramer-positive and CD8⁺ tetramer-negative cells were gated, and the percentages of cells coexpressing CD45RA and CD45RO within these subsets were estimated.

Assessment of the functional profile of tetramer-positive CD8⁺ T cells by using intracellular cytokine staining.

Cytokine and chemokine expression of tetramer-binding cells was assessed by intracellular staining in both the resting state and following activation by cognate peptide. Detection of intracellular cytokines by MAb staining of permeabilized T cells generally requires cell activation in the presence of reagents which inhibit protein migration through the endoplasmic reticulum (32). However, T-cell activation by either mitogens or specific peptide leads to down-regulation of the CD3-TCR complex, making it difficult to detect tetramer-binding cells. This problem was overcome by staining the cells with tetramer prior to activation and intracellular staining (Fig. 4a). By this approach, expression of MIP-1β, IFN-γ, perforin, and TNF-α was determined at rest and following activation in both the tetramer-positive population and the remaining CD8⁺ cells in PBMCs (Fig. 4b). The majority of unstimulated tetramer-positive CD8⁺ T cells showed low levels of expression of MIP-1β, IFN-γ, and TNF-α, all of which were significantly enhanced following specific activation. The levels of intracellular cytokine and chemokine were significantly higher in the tetramer-positive population than in the control CD8⁺ population (data not shown). However, a small population of tetramer-positive cells remained negative for each of these markers even after prolonged stimulation with high-dose peptide: this was most marked for TNF-α and IFN-γ. Perforin was detected in over 50% of the tetramer-positive cells, and expression was slightly decreased after stimulation.

FIG. 4.

Intracellular cytokine and chemokine in tetramer-reactive CD8⁺ cells. Freshly isolated PBMCs from donor 16 were permeabilized and stained with MAbs to CD69, IFN-γ, MIP-1β, TNF-α, and perforin either immediately following isolation or 6 h postactivation with cognate peptide. Both nonstimulated and peptide-stimulated CD8⁺ tetramer-positive cells were gated separately (a), and the proportions of cells within the given subsets expressing cytokines and chemokines (b) are summarized.

Cytotoxic activity of tetramer-specific T lymphocytes.

Tetramer-positive cells were assayed for their cytotoxic activity against peptide-pulsed autologous L-BCLs. Donors 13, 16, and 17 were selected because of high levels of tetramer-binding CD8⁺ T lymphocytes in peripheral blood (Table 2). Initially, the cytotoxicity of freshly isolated total PBMC populations was measured. HCMV peptide-specific lysis was evident in two of the three donors (Fig. 5a and b), but did not reach significance in the third (data not shown). Significant cytotoxicity was observed at E:T ratios of 50:1 in donor 13 (Fig. 5a) and 100:1 in donor 17 (Fig. 5b), in agreement with the percentages of tetramer-reactive cells detected in both donors. Failure to detect “fresh” cytotoxicity in donor 16 despite a number of circulating tetramer-positive PBMCs equivalent to that in donor 17 suggests that the activation status of the circulating tetramer-positive cells may be important in tests of rapid effector function. In this regard, it may be relevant that donor 16 has a relatively high percentage of CD27⁺ and CD28⁺ cells.

FIG. 5.

Fresh cytotoxicity of PBMC fractions from HCMV-positive donors. Autologous ⁵¹Cr-labelled L-BCLs from donors 13 (a) and 17 (b) were incubated with antigenic peptide at a final peptide concentration of 5 μM and were used as targets in a 15-h ⁵¹Cr-release assay. Freshly isolated PBMCs were used as a source of effector cells. Cytolytic activity was assayed at various E:T ratios.

In order to define the fresh cytotoxic potential of discrete subgroups of tetramer-positive PBMCs, CD28⁻ and CD28⁺ subsets from donors 13, 16, and 17 were sorted by FACS and incubated with HCMV peptide-pulsed targets. Tetramer-positive CD57⁺ and CD57⁻ cells from donor 13 were also purified to assess their ability to lyse target cells. Although tetramer-positive CD8⁺ T cells could be sorted from donor 13, this was not possible from donors 16 and 17 due to the low numbers of triple-stained CD8⁺ T lymphocytes, and so unfractionated CD8⁺ cells were used instead. Cytotoxicity was concentrated in the CD28⁻ subset in all three donors, although the low level of tetramer-positive CD8⁺ T lymphocytes from donors 16 and 17 led to a low level of killing. (Fig. 6a). This level was considered positive, since nonspecific killing was very low (less than 10% of killing observed on peptide-pulsed targets). In this investigation, the direct cell sorting of CD57⁻ and CD57⁺ HCMV-specific T lymphocytes from donor 13 demonstrated that both were able to exhibit HCMV-specific cytotoxicity in vitro (Fig. 6b).

FIG. 6.

(a) Cytotoxicity of CD28⁻ and CD28⁺ CD8⁺ subsets. HLA-A*0201/pp65^495–503-specific CD8⁺ T lymphocytes were sorted directly from PBMCs of donor 13 on the basis of the presence or absence of CD28 expression. Bulk CD8⁺ populations of CD28⁻ and CD28⁺ CD8⁺ T cells were sorted by FACS from donors 16 and 17. Sorted populations were incubated with peptide-pulsed L-BCL targets, and cytotoxicity was assayed after 15 h. The E:T ratios were 0.5:1 for donor 13 and 2:1 for donors 16 and 17. The final peptide concentration was 5 μM. The percentage of specific lysis represents CTL lysis of pulsed targets − lysis of nonpulsed targets. (b) Cytotoxicity of CD57⁻ and CD57⁺ CD8⁺ subsets. PBMCs from donor 13 were stained with A*0201/pp65^495–503 tetrameric complex in conjunction with anti-CD8 and anti-CD57 MAbs and CD8⁺ tetramer-positive populations were sorted by FACS on the basis of the presence or absence of CD57. Sorted populations were incubated with peptide-pulsed L-BCL targets, and cytotoxicity was assayed after 15 h. The E:T ratio was 0.5:1.

Expanded populations of HCMV-specific CD8⁺ T lymphocytes are oligoclonal.

In donor 13, in whom 4.4% of all CD8⁺ T cells were specific for the HLA-A*0201/pp65^495–503 epitope, over 70% of the tetramer-positive cells stained with an antibody to TCRBV14 (Fig. 7a). In HLA-A*0201/pp65^495–503-specific CD8⁺ T-lymphocyte cultures, the TCRBV14⁺ population was expanded to 90% after 2 weeks (Fig. 7b). PCR and sequence analysis confirmed that the TCRBV14 CTL line was clonal (data not shown). In donor 16, the majority of the HLA-B*0702/pp65^417–426-positive CD8⁺ T cells propagated in a CTL line also costained with the TCRBV14-specific MAb, although the clonality of this population was not determined. We failed to identify the TCR restriction of the HLA-B*0702/pp65^417–426-binding T cells from donor 17 by using this panel of MAbs.

FIG. 7.

Staining of HLA-A*0201/pp65^495–503-specific CTLs from donor 13 with a MAb specific for TCRBV14. Freshly isolated PBMCs and 12-day-old CTL lines were costained with anti-CD8-Tricolor, anti-TCR-FITC, and PE-conjugated HLA-A*0201/pp65^495–503 tetramer. CD8⁺ lymphocytes were gated, and the proportions of tetramer-reactive cells expressing TCRBV14 both in fresh PBMCs (a) and the CTL line (b) were assessed. Control antibodies included anti-BV16 and anti-BV17 TCR MAbs.

DISCUSSION

In this study, we describe the frequency and functional properties of HCMV-specific CD8⁺ T lymphocytes in immunocompetent individuals. It is well established that HCMV reactivation or reinfection in the immunocompromised host can lead to a number of clinical manifestations, and there is strong evidence to suggest that the recruitment of CD8⁺ T lymphocytes is paramount to the control of disease. These findings imply that CD8⁺ T cells may also be critical in suppressing replication of HCMV in the immunocompetent individual, but little is known about the extent to which CD8⁺ T lymphocytes are recruited. A recent report emphasized the dominant role of CD8⁺ T cells in controlling reactivation of MCMV (33).

HCMV employs a number of highly specialized strategies to avoid cellular immune recognition (11, 12, 46). Consequently, the targets for the HCMV-specific cellular immune response may be limited. To date, the majority of known CD8⁺ T-cell epitopes are those derived from the structural tegument protein /pp65 (26, 37, 49), for which a number of immunodominant epitopes for CD8⁺ T lymphocytes have now been defined (49). We have exploited tetrameric MHC class I-peptide complexes to study /pp65-specific CD8⁺ T lymphocytes in healthy seropositive individuals. These reagents have provided a novel approach for the direct visualization of antigen-specific CD8⁺ T lymphocytes in a number of human and murine immune responses (1, 6, 30, 31).

The most striking observation in this investigation is the very high frequency of HCMV-specific CD8⁺ T lymphocytes in the blood of HCMV-seropositive individuals. HCMV-specific CD8⁺ T cells were detected in all donors, with average values of 0.75 and 1.85% for the HLA*A0201- and HLA*B702-restricted peptides, respectively (range, 0.03 to 5%). Strikingly, in three donors, almost 5% of peripheral blood T lymphocytes were specific for a single peptide from HCMV. These values are among the highest recorded in chronic viral infection and are surprising given the accepted principle that HCMV becomes latent after primary infection. Similar findings have been reported by using the technique of peptide stimulation followed by intracellular cytokine assay (18, 19). In addition, we recently reported on a patient who suffered simultaneous coinfection with HCMV, human immunodeficiency virus (HIV), and Epstein-Barr virus (EBV). In prospective studies, the CTL response to HCMV clearly dominated over those against EBV and HIV (17). It is unclear how the high level of virus-specific CTLs is maintained. Analysis of tetramer-reactive CD8⁺ T cells in HIV-seropositive individuals has shown that following introduction of highly active antiretroviral therapy, there is a marked decline in the frequency of HIV-specific CD8⁺ cells, and this supports a model in which replicating virus is required to support a high antigen-specific CTL level (7, 14, 31). These data imply that the large numbers of circulating HCMV-specific CD8⁺ T lymphocytes may be boosted by periodic low-grade HCMV replication in the latently infected host. The identification of replicating HCMV in immunocompetent donors is difficult, but recent studies using sensitive PCR monitoring for HCMV reactivation demonstrated viral shedding in all healthy seropositive individuals at some stage over a 6-month period of study (42). The notion that both latent and lytic infections can coexist in the immunocompetent host has been documented for other latent viral infections, including EBV (52). Myeloid and dendritic cell precursors represent an important reservoir of latent HCMV infection, and APCs such as dendritic cells may represent a site of periodic HCMV reactivation which could recruit large numbers of HCMV-specific CD8⁺ T lymphocytes.

We were only able to study CD8⁺ T-cell responses to two immunodominant peptides from HCMV, and the total T-cell response to all immunogenic HCMV peptides must be even greater. Such a finding would be compatible with the observation that individuals who are HCMV seropositive have a higher absolute CD3⁺ CD8⁺ lymphocyte count than those who are seronegative (13). Our results, together with those obtained with intracellular cytokine assays (18) and tetramer staining of CMV-specific CTLs in patients with HIV infection (17), suggest that large populations of HCMV-specific CD8⁺ T lymphocytes explain this increment in the lymphocyte count. At present, we have no explanation for the variation in the frequency of HCMV-specific CD8⁺ T cells between different donors. However, given the extensive genetic polymorphism in the HLA system, there will be variation between different donors in the HCMV peptides that are selected for presentation to the immune system. Differences in endogenous HCMV viral load may also contribute to this variation. In HIV infection, a negative correlation between plasma viral load and tetramer-reactive CD8⁺ T lymphocytes has been observed (31). This suggests that the stable levels of CMV-specific CTLs seen in individual donors over time indicate relatively constant levels of CMV viremia.

Previous estimations of the frequency of HCMV-specific T lymphocytes have been calculated by LDA. While this technique is robust and reproducible, it seems likely that the results may significantly underestimate the true frequency of circulating CD8⁺ T lymphocytes (2). Identification of CD8⁺ cells by LDA requires that T lymphocytes retain the ability to proliferate in vitro. However, effector T cells exhibit poor proliferative potential (23, 25, 28) and are therefore likely to be excluded or underestimated in this type of analysis. LDA of five donors in this study indicated a discrepancy between the number of CD8⁺ cells determined by tetramer staining and the value obtained by LDA. Although one donor showed a higher CD8⁺ T-lymphocyte frequency by LDA, this technique underestimated the frequency of antigen specificity in the others by a factor of between 5 and 29 times. The degree of underestimation of CD8⁺ T-cell frequency by LDA was positively correlated to the absolute number of tetramer-positive cells detected in the blood. The largest discrepancy between the two techniques was in donor 13, in whom a high percentage of tetramer-staining T lymphocytes had an effector-like phenotype (16), a population known to have poor replicative potential (28). The findings suggest that LDA measurement may be more appropriate as an indicator of the number of antigen-specific memory cells capable of significant proliferation. In individuals who develop large expansions of effector CD8⁺ T lymphocytes, LDA becomes unreliable as a measure of the complete antigen-specific response.

The application of HLA-peptide tetramers allowed us to make a detailed study of the phenotype of CMV-specific CTLs. The data demonstrate considerable heterogeneity in the proportion of the different phenotypic subsets both within and between individuals. CD45RA^high and CD45RO^high populations were seen in all individuals, reinforcing the evidence that CD45RA is not a reliable marker of naive lymphocytes within the CD8⁺ population (27, 50). A population of CD8⁺ CD45RA⁺ effector CTLs has been described and may represent reversion from an initially CD45RO^high cohort (16). Sorted CD45RA^high and CD45RO^high populations retained the ability to proliferate and affect cytotoxicity when cultured in the presence of HCMV peptide in vitro (Fig. 8). A third population of CD45RA^dim CD45RO^dim tetramer-positive cells that coexpressed both isoforms of CD45 (Fig. 9) was evident in all donors by four-color FACS analyses (Fig. 3). CD57 is a carbohydrate epitope expressed on a subset of CD8⁺ and CD4⁺ T cells, and the percentage and absolute number of CD57⁺ T cells are known to be increased in HCMV-seropositive individuals (22, 38, 45). Although the function of CD57⁺ is unknown, it has been associated with terminally differentiated effector cells (8). Our results demonstrate that CD57⁺ is expressed on a mean of 44% of HCMV-specific CD8⁺ T cells (range, 16 to 93%), and these data are similar to those of Kern et al., who used intracellular cytokine expression to identify CMV-specific CTLs in both immunocompetent and immunosuppressed donors (18). Lack of expression of the costimulatory molecules CD27 and CD28 on CD8⁺ T lymphocytes has also been associated with an effector cell phenotype (16, 18, 43, 47). Expression of these molecules on HCMV-specific CD8⁺ T lymphocytes was variable between donors, although within individuals, the percentage of cells expressing CD27 and CD28 was tightly correlated (0.87; P < 0.001), and both were absent from large proportions of CTLs in a number of donors. Lack of expression of the lymph node homing receptor CD62L is associated with a memory CD8⁺ phenotype and has been documented on tetramer-staining CD8⁺ T lymphocytes in both in HIV (1) and EBV (40). We found that the expression of this molecule was highly variable between different donors and did not appear to closely correlate with the expression of other surface markers. Most donors, in particular those with the highest frequencies of tetramer-reactive lymphocytes, displayed a large number of cells lacking CD27, CD28, and CD62L, a phenotype associated with effector function (16). The proposed functions of CD7 include cellular activation and adhesion (21, 39). CD7 was expressed on the vast majority of tetramer-reactive T lymphocytes, but in some donors, a significant proportion of CD7⁻ CD8⁺ T lymphocytes was detected.

FIG. 8.

Cytolytic ability of sorted CD45RO^high and CD45RO^high tetramer-reactive cells propagated in vitro. PBMCs from donor 13 were stained with tetrameric complex, anti-CD8 MAb, and anti-CD45RA and anti-CD45RO antibodies, and populations of tetramer-reactive CD45RA^high and CD45RO^high cells were sorted by FACS and cultured in the presence of peptide-pulsed autologous L-BCLs. Their ability to lyse peptide-pulsed targets by a standard ⁵¹Cr-release assay was assessed following 14 days of culture. Both populations lysed target cells with similar efficiencies at E:T ratios of 10:1.

FIG. 9.

Segregation of tetramer-reactive cells expressing low, medium, and high levels of CD45RA and CD45RO isoforms. PBMCs were triple stained with anti-CD8 MAb, tetrameric complex, and either CD45RA or CD45RO isoform antibodies. CD8^high populations were gated, and the percentages of tetramer-reactive cells expressing the CD45 isoforms were estimated. A typical dot plot of tetrameric staining versus CD45RA and CD45RO antigen expression from donor 17 is summarized.

It is important to correlate the quantitative and phenotypic data described above with the qualitative function of the tetramer-reactive T lymphocytes. Initially we were not able to study the intracellular expression profile of tetramer-positive cells following activation by peptide stimulation due to down-regulation of TCR expression upon activation. However, a novel approach which involved the staining of T-cell populations with tetramer prior to peptide stimulation allowed direct comparison of intracellular expression levels in tetramer-positive and control CD8⁺ T-cell populations. Optimization of this approach on HCMV-specific CTL clones and lines showed that staining of CTLs with tetramer prior to the addition of peptide did not have any demonstrable influence on the pattern or level of cytokine expression compared to that documented after peptide alone (data not shown). Levels of detection prior to activation were low, but following activation, there was a significant increase in expression of MIP-1β, TNF-α, IFN-γ, and perforin. The vast majority of tetramer-positive cells expressed both cytokines and chemokines following peptide stimulation, but perforin levels were slightly reduced. The loss of perforin upon peptide stimulation is likely to reflect perforin release from CTLs after engagement with peptide-pulsed targets. We have observed the same phenomenon in antigen-activated HCMV-specific CTL clones (unpublished data) and EBV-specific T cells (M. Callan, personal communication). Collectively, these data reinforce the picture of functional heterogeneity among CMV-specific CTLs.

The cytotoxic activity of CTLs remains the ultimate assay of their function. With freshly isolated PBMCs, HCMV peptide-dependent lysis was detected in two of three donors with large numbers of tetramer-positive cells. Interestingly, such fresh killing was greatest in those donors with large numbers of circulating CD27⁻ CD28⁻ CD8⁺ T cells (16). When the E:T ratio for fresh killing by PBMCs is corrected for the percentage of HCMV-specific CD8⁺ T lymphocytes detected by tetramers, cell-mediated lysis is comparable to what would be expected from an antigen-specific CTL clone; this implies that in these donors, the majority of the tetramer-reactive T lymphocytes are able to exert effector function in vitro. In view of the heterogeneous phenotypic profile of tetramer-positive CTLs, the cytolytic properties of distinct subgroups were assessed. Cell-mediated cytotoxicity was clearly restricted to tetramer-reactive cells which lacked CD28 expression, but was detected in both CD57⁻ and CD57⁺ subsets.

The human CD8⁺ T-cell repertoire is frequently associated with oligoclonal expansions which accumulate with age (34), and CMV /pp65-specific CTLs have previously been shown to have restricted TCR expression (47). We investigated the TCR expression of tetramer-positive cells by costaining with tetramers and antibodies to TCR V regions (51). In two donors, HCMV-specific CD8⁺ T cells predominantly expressed the TCRBV14 segment, despite the fact that the peptides were different in each case. PCR sequencing of the TCRBV14 transcripts confirmed an oligoclonal population in donor 13. The reasons for the dramatic focusing of the immune response to HCMV are not yet clear, but preferential expansion of the TCRBV14 clone was observed with short-term in vitro culture.

We have demonstrated that in individuals who are HCMV seropositive, there is a considerable CD8⁺ T-cell response mounted against the virus even when there is no history of symptomatic HCMV infection. Many of these CD8⁺ T lymphocytes are effector cells, as judged by their ability to lyse peptide-pulsed targets in fresh cytotoxic T-cell assays, their cell membrane phenotype, and their production of both cytokines and chemokines rapidly upon antigenic stimulation. These data suggest that in the setting of latent infection, HCMV may periodically reactivate and recruit cytotoxic CD8⁺ T lymphocytes. Reactivation may occur when the normal homeostatic mechanisms are perturbed—for instance, at a site of inflammation or infection or possibly during reactivation of other herpesviruses. Because macrophages are known to support the reactivation of HCMV and these cells are likely to be abundant at the site of inflammation, it is plausible that they give rise to HCMV progeny.

CMV remains a significant clinical problem in immunosuppressed individuals, and protocols such as adoptive transfer (44) or vaccination have been attempted in order to induce protective immunity. Our data suggest that large numbers of effector CD8⁺ T lymphocytes may be required for adequate protection against viral reactivation. HLA-peptide tetramers are a powerful approach to monitoring such immunity and may be valuable as a tool in T-cell adoptive transfer, for instance transferring HCMV-specific CD8⁺ T lymphocytes directly from stem cell transplant donors to recipients (9).