Identification of Human Herpesvirus 8-Specific Cytotoxic T-Cell Responses

Abstract

Human herpesvirus 8 (HHV-8) (or Kaposi’s sarcoma-associated herpesvirus) is implicated in the etiopathogenesis of Kaposi’s sarcoma (KS) and certain lymphoproliferations. The introduction of more effective therapies to treat human immunodeficiency virus infection has led to a decline in the incidence of KS and also in the resolution of KS in those already affected. This suggests that cellular immune responses including cytotoxic T lymphocytes (CTLs) could play a vital role in the control of HHV-8 infection and in KS pathogenesis. Here we elucidate HLA class I-restricted, HHV-8-specific cellular immune responses that could be important in the control of HHV-8 infection and subsequent tumor development. We show the presence of CTLs against HHV-8 latent (K12), lytic (K8.1), and highly variable (K1) proteins in infected individuals.

The interface between infection, immunity, and malignancy is highlighted by cancers prevalent in patients infected with human immunodeficiency virus (HIV) and in organ transplant recipients. Immunosuppressed individuals are prone to tumors caused by the gamma herpesviruses Epstein-Barr virus (EBV) in lymphomas (25) and human herpesvirus 8 (HHV-8; also called Kaposi’s sarcoma-associated herpesvirus) in Kaposi’s sarcoma (KS) and certain lymphoproliferations (2, 7, 11).

HHV-8 is the most recently identified oncogenic virus and is causally linked to KS (6, 10), the most common tumor in HIV-infected individuals, and also to primary effusion lymphoma and the immunoblastic variant of Castleman’s disease (4, 8, 29). The introduction of aggressive anti-HIV therapies has led to a decline in the incidence of KS in AIDS patients and also in the resolution of KS in those already affected (16). This suggests that cellular immune responses, compromised in AIDS but recovering after highly active antiretroviral therapy (HAART), could be important in the control of HHV-8 infection and in the development of KS.

The immune system is capable of mounting potent attacks on invading viruses and of eliminating some viral infections. Virus-specific, HLA-restricted cytotoxic T-lymphocyte (CTL) responses are critical to clear early viremia in acute HIV infection, are important in the control of opportunistic viral infections such as cytomegalovirus or herpes zoster reactivation, and play an important role in the control of human papillomavirus-induced squamous cell carcinomas and in EBV-induced lymphoproliferation.

We postulate that HHV-8 establishes a persistent infection, which is normally controlled by the immune system, and that the number of HHV-8-infected cells is under immunological control. When this immune control declines due to acquired or iatrogenic immunosuppression, the number of HHV-8-infected cells increases with the subsequent unchecked proliferation of virally infected cells and the development of HHV-8-related tumors. The human gamma herpesviruses EBV and HHV-8 establish latent infections in lymphoid cells, where the viral episomes express only a limited number of genes (the so-called latent genes), and this means that only a limited number of peptides may be recognized in association with HLA class I molecules by CTLs. In EBV infection, virus-specific CTL activity directed against peptides from latent and lytic proteins is important in the pathogenesis of EBV-associated diseases (26).

To investigate the existence of CTLs against HHV-8, we selected the products of three HHV-8 open reading frames: K1, K8.1, and K12. None of these have sequence similarity to EBV proteins, thereby excluding the possibility of cross-reactivity with EBV-specific CTLs.

K1 is at the left-hand side end of the HHV-8 genome, in a position equivalent to the gene encoding the herpesvirus saimiri transforming protein (STP), but K1 has no sequence or structural similarity to STP. K1 is oncogenic when overexpressed in rodent fibroblasts (19); however, it is not yet clear whether this protein is expressed in latency in mesenchymal cells (e.g., KS spindle or tumor cells). In effusion lymphoma cells K1 expression is restricted to the lytic cycle (18). K1 is highly variable among HHV-8 isolates (22) and is therefore presumed to be under significant biological pressure, suggesting that this protein may be important in HHV-8 pathogenesis.

K8.1 is a 228-amino-acid viral glycoprotein expressed during lytic viral replication (20, 24). K8.1 is highly immunogenic and therefore useful to measure humoral immunity against HHV-8 (24). K8.1 has no overt amino acid sequence similarity with any viral or cellular sequence currently available in databases (24). K8.1 localizes on the surfaces of cells and virions (20). The open reading frame in EBV that shares genomic position and orientation with K8.1 encodes gp350/220, which is known to bind to CR2 (CD21) on host cells (32). This suggests that K8.1 might also be involved in cell attachment (20). gp350/220 of EBV evokes powerful cellular immune responses and is indeed being investigated as an EBV vaccine (9, 25).

K12 encodes a unique viral protein expressed during latent infection (35). K12 is expressed in nearly all KS spindle cells and also in latently infected primary effusion lymphoma cells (30). K12 is transforming in vitro (21), and it may therefore play a role in HHV-8-induced cellular proliferation.

Study participants.

Study participants were selected from HIV-positive and -negative individuals attending the Genitourinary Clinic at the Kobler Centre, Chelsea and Westminster Hospital, London, United Kingdom. Control donors were laboratory workers who were at a low risk of HHV-8 infection. The study was approved by the ethical committee of the Chelsea and Westminster Hospital Trust.

HLA typing.

DNA was extracted from 200 μl of EDTA-peripheral blood, by using a QIAamp blood kit (Qiagen, Crawley, United Kingdom). HLA class I typing was performed by amplification of refractory mutation system PCR with sequence-specific primers (17). PCR was performed in 96-well PCR plates with 5 to 20 μg of DNA, 5 μl of allele sequence-specific primers (Oxford Transplant Centre, Oxford, United Kingdom), deoxynucleoside triphosphates, and Taq DNA polymerase. Products were visualized in a 1% agarose gel with ethidium bromide.

HHV-8 serological assay.

HHV-8 antibodies were detected by using an indirect immunofluorescence serological assay as described previously (12, 34). To determine anti-HHV-8 antibody titers, sera were diluted in 3% fetal calf serum in phosphate-buffered saline. Twofold dilutions were made starting at a concentration of 1:100.

Construction of recombinant modified vaccine Ankara (MVA) expressing K1, K8.1, and K12.

Total cellular DNA was extracted from the primary effusion lymphoma cell line BCP-1, which carries HHV-8 but not EBV (1, 12). The K1 open reading frame was amplified from BCP-1 DNA by PCR with the forward primer 5′-GGACGCGGCCGCGTCTTTCAGACCTTGTTGGAC-3′ and the reverse primer 5′-AATCCAGCGGCCGCGAATGTCAGTACCAATCCAC-3′. The K1 PCR product was digested with NotI restriction endonucleases (NotI restriction sites are underlined), the staggered ends were filled in with the Klenow fragment of DNA polymerase, and the blunt-ended fragments were inserted into the SmaI site of pSC11 (5).

The K12 open reading frame was amplified from BCP-1 DNA by PCR with the forward primer 5′-GCATGCGGCCGCATGGATAGAGGCTTAACGG-3′ and the reverse primer 5′-CGTAGCGGCCGCTAGCTTCAGTGCGCGC-3′. The K12 PCR product was digested with NotI restriction endonucleases (NotI restriction site is underlined) and inserted into a novel NotI site of pCS11. This NotI site in the pSC11 plasmid was created by ligating a synthetic oligonucleotide linker containing it to the SmaI-digested pSC11. The new version of the plasmid was called pSC11N.

The K8.1 open reading frame was excised from plasmid pCDNA-K8.1 (24) with BamHI and XbaI restriction endonucleases, the staggered ends were filled in with the Klenow fragment of DNA polymerase, and the NotI linker was ligated. The resulting K8.1-NotI was digested with NotI and inserted into the NotI site of pSC11N.

BHK21 cells were infected with MVA at 0.05 PFU per cell. pSC11 or pSC11N plasmids carrying K1, K8.1, or K12 genes were transfected with Perfect Lipid (Invitrogen, Groningen, The Netherlands). Total virus from the cells and supernatant were harvested 3 days later and used for the reinfection of BHK21 cultures. The plaques of recombinant MVAs (rMVAs) were identified by using X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) color selection (3) and purified by five rounds of plaque purification. Bulk stocks of the rMVAs were grown and purified by the centrifugation of cytoplasmic extracts through a 36% (wt/vol) sucrose cushion in a Beckman SW28 rotor at 13,500 rpm for 80 min. rMVA expressing Escherichia coli β-galactosidase was used as a control.

Cell monolayers were grown to approximately 70% confluency and infected with different rMVAs. Cells were harvested 16 h later, washed with phosphate-buffered saline, and resuspended in a lysis buffer for RNA extraction.

RT-PCR.

Total RNA was extracted from individual rMVA-infected cells by using an RNeasy kit according to the manufacturer’s instructions (Qiagen). For reverse transcription (RT)-PCR the Stratagene RT-PCR kit was used according to the manufacturer’s instructions. RT of 5 μg of total RNA was performed at 37°C for 1 h in a 50-μl reaction mixture containing 300 ng of random primers, 0.2 mM concentrations of deoxynucleoside triphosphates, and 50 U of Moloney murine leukemia virus reverse transcriptase. Subsequently the above-mentioned mixture was heated to 90°C for 5 min. Five microliters of the RT product was subjected to PCR amplification with sequence-specific primers for each gene. Amplification was done for 30 cycles (94°C for 30 s, 55°C for 30 s, and 72°C for 45 s) and a final cycle at 72°C for 10 min. K8.1 was amplified with the forward primer 5′-ATGAGTTCCACACAGATTCGC-3′ and the reverse primer 5′-CACTATGTAGGGTTTCTTACGCCG-3′. The sequences of the primers for K1 and K12 are described above.

CTL assays.

Peripheral blood mononuclear cells were isolated on a Ficoll-Hypaque density gradient and washed twice in RPMI 1640 (Sigma). Peripheral blood mononuclear cells were then cultured at a density of 1 × 10⁶ to 2 × 10⁶/ml in a 24-well plate (Nunc) at 37°C in 5% CO₂, together with a sonicated preparation of rMVA expressing K1, K8.1, or K12 in RPMI 1640 supplemented with 10% fetal calf serum (R10). rMVAs were used at a multiplicity of infection of 0.2. The cultures were supplemented with 10 U of recombinant human interleukin 2 per ml on day 3 and were tested for specific killing by CTLs on days 10 to 14.

Virus-infected target cells were prepared by incubating pelleted autologous EBV-transformed B-cell lines (BCLs) or HLA-matched or -mismatched BCLs (kindly provided by Alan Rickinson, University of Birmingham, Birmingham, United Kingdom) with rMVA K1, rMVA K 8.1, rMVA K12, or rMVA β-galactosidase at a multiplicity of infection of 10 in a conical-bottom tube for 90 min. Target cells were then washed, resuspended in R10, and cultured overnight to allow the expression of viral genes. Target cells were labelled with 150 μCi of ⁵¹Cr/10⁶ cells (Amersham International, Little Chalfont, United Kingdom) for 1 h before being washed three times and plated into U-bottom 96-well plates at a density of 5 × 10³ cells/well in 100 μl. Effectors were added at different concentrations in 100 μl of R10, and the mixtures were incubated for 4 h. All assays were performed in duplicate. A total of 20 μl of the supernatant was removed from each well onto a filter mat (Wallac), which was dried and then sealed with a plastic bag containing a liquid scintillation cocktail (Wallac). The amount of ⁵¹Cr was counted on a Wallac counter. The percentage of specific lysis was calculated by using the following formula: 100 × (E−M)/(T−M), where E is the amount of ⁵¹Cr released into 20 μl of supernatant from wells containing targets and effectors, M is the amount of ⁵¹Cr released into wells containing targets and the medium only, and T is the amount of ⁵¹Cr released from wells containing targets lysed by the addition of 5% Triton X-100. The amount of spontaneous release [100 × (M/T)] was always less than 35%. A positive CTL response was defined as one in which the specific lysis of target cells was more than 10% above that of the same target cell infected with β-galactosidase-expressing rMVA (13, 14).

HLA class I blocking assay.

In the inhibition experiments, the monoclonal antibody W6/32 (Dako), specific for HLA class 1-peptide complexes, was added to the target cells 30 min prior to coincubation with the effector cells. The final concentration of antibody was 1 μg/ml (23). The HLA typing, HIV and HHV-8 serological status, HHV-8 antibody titer, CD4⁺ T-cell count, and clinical status of all patients and donors are presented in Table 1.

TABLE 1.

Patients included in this study^a

Sampleb	Age (yr) of patient	HIV status	HHV-8 status	HHV-8 titer	No. of CD4 T cells	KSc	HLA type(s)	Target match	% Lysis of target expressingd
									K1	K8.1	K12	Gald
1	43	+	+	12,800	232	T1I0S1	A24, A25, B7, B39	A24	—	—	—	—
2	29	+	+	25,600	353	T0I0S0	A24, A29, B14	A24	—	—	—	—
3	29	+	+	25,600	495	T0I0S0	A24, A29, B14	A24	—	—	—	—
4	33	+	+	6,400	302	T1I0S1	A2, A25, B58	A2	—	—	—	—
5	34	+	+	100	488	T0I0S0	A3, B27	A3	—	12	21	—
6	37	+	+	12,800	121	T0I1S0	A2, A3, B7, B60	A2	—	12	—	—
7	66	−	+	100	ND	StageII**	A2, A3, B18, B58	A2	—	—	—	—
8	49	+	+	3,200	456	***	A26, A68, B7, B40	B7	—	—	24	—
9	42	+	+	51,200	291		A66, B41	Auto	—	21	—	—
10	40	+	+	400	964		A24, A25, B8, B18	B8	25	20	26	—
11	29	+	+	100	389		A32, A66, B8, B81	B8	12	40	37	—
12	44	+	+	100	199		A2, B44, B57	A2	—	11	17	—
13	44	+	+	100	232		A2, B44, B57	A2	—	—	20	—
14	38	+	+	100	263		A2, A3, B49, B62	A2	—	—	12	—
15	36	+	+	25,600	531		A2, B44	A2	17	—	—	—
16	25	+	−		357		A3, B7, B44	Auto	—	—	—	—
17	42	+	−		313		A2, A68, B7, B42	A2	—	—	—	—
18	39	+	−		158		A1, B44	B44	—	—	—	—
19	28	−	−		ND		A11, A24, B18, B44	B44	—	—	—	—
20	31	+	−		133		A3, A29, B7	A3, B7	—	—	—	—
21	27	−	−		ND		A2	A2	—	—	—	—
22	33	−	−		ND		A2	A2	—	—	—	—

^a**, post-renal transplant KS. ***, previous KS resolved after patient had started HAART. ND, not done. Auto, autologous BCLs used as the target. 

^bSamples 2 and 3 and 12 and 13 were obtained at two separate time points from single patients. 

^cKS staging was done according to the criteria of the AIDS Clinical Trials Group. T0, confined to skin, lymph nodes, and the mouth. T1, edema, ulceration, or systemic disease. I0, ≥200 CD4 cells/μl. I1, <200 CD4 cells/μl. S, systemic illness including opportunistic infection or B symptoms. 

^dLysis of target cells of <10% is considered negative (−). 

The lack of antibodies against K1, K8.1, and K12 hindered our ability to assess their expression at the protein level. The expression of rMVA K1, K8.1, and K12 was therefore assessed by RT-PCR. Following the infection of BHK21 cells, all three genes were amplified by RT-PCR with sequence-specific primers (Fig. 1); K1 migrated at 911 bp, K8.1 migrated at 790 bp, and K12 migrated at 190 bp, corresponding to the predicted molecular sizes.

FIG. 1.

RT-PCR results showing RNA transcripts for rMVA K1, K8.1, and K12 from BHK21 cells. The amplification was done in the presence (+) or absence (−) of reverse transcriptase. The band sizes for K8.1 are in agreement with the sliced products (α and β) of this transcript.

Twenty individuals (six who were HHV-8⁺ and KS⁺, seven who were HHV-8⁺ and KS⁻, and seven who were HHV-8⁻) were evaluated for HHV-8-specific CTLs, by using rMVA (K1, K8.1, and K12) and control rMVA β-galactosidase-infected target cells (Table 1). The lymphocytes were prestimulated as detailed above and used as effectors in a ⁵¹Cr release assay.

One hundred percent of HHV-8⁺ KS⁻ patients had CTL responses to K1, K8.1, or K12 (Table 1 and Fig. 2); K1, K8.1, and K12 were recognized by the immune systems of three, four, and five patients, respectively. Among the HHV-8⁺ KS⁺ patients, the CTL response rate was 33% (two of six patients). Patient 6 recognized K8.1 only weakly (12% lysis). Patient 5, who had very mild KS, recognized K8.1 and K12 (12 and 21%). Patient 8, who previously had KS that resolved during HAART, showed a CTL response to K12. Of note, K1 was not recognized by any of the patients with KS. None of the HHV-8 immunofluorescent antibody-negative patients had HHV-8-specific CTL responses. No lysis of control targets infected with rMVA expressing β-galactosidase or of HLA-mismatched target cells infected with rMVA was seen.

FIG. 2.

Recognition of rMVA-expressed HHV-8 antigens by CTLs from an HHV-8⁺ KS patient (no. 3), an HHV-8⁺ patient without KS (no. 11), and an HHV-8 negative patient (no. 17). Patient details are shown in Table 1. CTLs tested at the effector-to-target ratios of 40:1 and 10:1 are shown. b-gal, β-galactosidase.

To evaluate whether the lysis observed was HLA class I restricted, CTL lines, obtained from patients 6, 13, and 15 and directed against K8.1, K12, and K1, respectively, were analyzed in the presence or absence of anti-HLA class I (W6/32) antibody. Lytic activity was completely inhibited at an effector/target ratio of 40:1 (Fig. 3) when target cells were pretreated with the anti-HLA class I antibody.

FIG. 3.

Inhibition of CTL activity by treatment with anti-HLA class I antibody (W6/32). The targets were preincubated for 30 min at 22°C before the addition of effector cells at a concentration of 40:1.

Conclusions.

We demonstrate here that HHV-8, like other herpesviruses, is able to elicit HLA class I-restricted CTL responses. We show specific responses for three different HHV-8 proteins—K1, K8.1, and K12, which had been introduced into MVA. The CTL activity against these HHV-8 proteins is HLA class I restricted, implying that they represent CD8⁺ T cells.

A cellular immune response against both HHV-8 latent (K12) and lytic (K8.1) proteins is shown. CTLs against EBV lytic proteins appear to be important in the pathogenesis of EBV replicative lesions (31) and could be important in viral replication and also viral shedding. CTLs against latent proteins could be essential to prevent the outgrowth of virally transformed cells.

The relative structure and position of HHV-8 open reading frame K1 is comparable with the latent membrane protein of EBV and herpesvirus saimiri STP. It appears that K1 has been under significant biological pressure and is used to distinguish four major clades (A, B, C, and D) of HHV-8 that are associated with different ethnic groups in different geographical settings (15). CTL activity against this protein might therefore be important in HHV-8 transmission and pathogenesis in different geographical regions. The K1 sequence used in this study was cloned from BCP-1 cells (1) and belongs to the A clade of HHV-8, which is frequently seen in AIDS-KS patients in the West (15). The absence of CTLs against K1 in some individuals could be due to infection with different HHV-8 strains. We are presently using overlapping peptides from the different K1 isolates to delineate K1-specific CTL responses in HHV-8⁺ patients from different ethnic groups.

CTLs restricted by the HLA molecules A2, A3, B7, and B8 were all shown to recognize at least one of the HHV-8 proteins tested. HLA alleles were found to present epitopes from more than one viral protein (e.g., HLA A2- and A3-restricted epitopes were demonstrated in K8.1 and K12, and HLA B8 presented all three proteins). This suggests a broad repertoire of CTL responses to HHV-8 as seen in other viral infections. HLA A2, A3, B7, and B8 are present in more than 60% of the Caucasian population, and it should therefore be possible to identify and study HHV-8-specific CTLs by using these constructs in many individuals.

Although this is a pilot study, we were able to compare patients with and without KS; we did not demonstrate HHV-8-specific CTL responses in most patients with KS, indicating that a decline in cellular immune responses against HHV-8 may be present in HIV⁺ patients with KS and could contribute to KS pathogenesis. This would be reminiscent of the lack of EBV-specific CTLs seen in immunosuppressed patients, which correlates with the onset of EBV-driven lymphoproliferation (25, 27). One patient (patient 8) previously had KS, but this resolved during HAART, and we were able to demonstrate CTLs specific to the K12 protein in this patient. No HHV-8-specific CTL responses were seen in HHV-8 antibody-negative patients, and no responses were seen to HLA-matched or autologous target cells infected with control MVA containing β-galactosidase. Overall, we were less likely to see CTL responses in patients with a high antibody titer against HHV-8. Antibody titer correlates with viral load (28, 33). This therefore suggests that the lack of CTL activity correlates with a higher HHV-8 viral load, although we will need a higher number of patients to confirm this.

KS is a complex tumor, and various immune responses could be involved in its pathogenesis (10). The rapid resolution of KS in some HIV-positive patients started on HAART suggests that a small improvement in immunity might be important in disease control. CD4⁺ T-helper responses, not studied here, natural killer cells, and leukocyte-activated killer cells could also be involved in the control of the growth of HHV-8-positive cells. The rapid decline in the viral load of HIV itself has also been suggested to play a role in the response of KS lesions to HAART (10).

The identification of HLA class I-restricted CTLs against HHV-8 will allow us to identify viral epitopes that serve as recognition sites for HHV-8-specific CTLs and to evaluate the effectiveness of HAART in the reconstitution of HHV-8-specific CTLs. We are currently conducting a prospective study to evaluate the effects of HAART on HHV-8-specific CTL responses.