Persistent viruses have developed potent strategies to overcome host immune defenses. In particular, viral interference with Ag presentation by HLA class I molecules can effectively impair the host’s CTL function. Here we provide evidence for a novel aspect of differential splicing on endogenous processing of a latent viral transcript resulting in dominant protein isoforms from which the CTL determinant has been deleted. Consequently, virus-infected cells expressing these isoforms were poorly recognized by CTLs. Molecular analysis revealed that this splicing significantly reduced expression of the viral transcript encoding the relevant epitope to levels below the threshold required for CTL recognition. The importance of splicing was further reinforced by the observation of efficient CTL recognition of target cells expressing a truncated viral transcript that abolished differential splicing. Thus, differential splicing, which is a common mechanism of gene regulation in many pathogens, may unexpectedly interfere with immune recognition.
The success of persistent viruses lies in their capacity for evading the host defense mechanisms by either mutating target Ags or interfering with components of the host immune response. In particular, viral interference with HLA-restricted CTL response is a potent immune evasion strategy resulting in reduction or loss of CTL function (1). In the case of persistent herpes virus infections, such as EBV, many of the target epitopes for CTLs are included within some of the latent viral Ags (2). However, it is tempting to speculate that general regulatory mechanisms of gene expression may mask potential T cell determinants within other viral Ags. Of particular interest is one of the EBV-encoded BamHI A transcripts (referred to as RK-BARF0) (3) which is ubiquitously expressed in all EBV-associated malignancies such as Burkitt’s lymphoma (BL)3 and nasopharyngeal carcinoma (NPC) (4), making this Ag a potential target for CTL control. Indeed, we have recently reported that RK-BARF0 encodes the HLA-A*0201-restricted CTL epitope LLWAARPRL, which stimulates CTLs from healthy virus carriers, albeit at a low CTL precursor frequency (5). Although these RK-BARF0-specific CTLs effectively recognized target cells exogenously sensitized with the LLWAARPRL synthetic peptide, EBV-infected B cells were not lysed in a standard 51Cr release assay. This lack of endogenous presentation of this epitope opens the possibility that EBV utilizes a mechanism to mask expression of ubiquitously expressed viral proteins that might threaten viral persistence.
Materials and Methods
Both the EBV-positive BL cell lines (MutuI c59 and MutuIII c62) and lymphoblastoid cell lines (LCLs) (NK-Wil and SB-B95.8) transformed with EBV strains QIMR-Wil (6) or B95.8 (7) are HLA-A2 positive. DG75 is an EBV-negative, HLA-A2-positive BL cell line. The EBV-positive NPC cell lines C15 (HLA-A2 negative) and C17 (HLA-A2 positive) were a gift from P. Busson (Institute Gustave Roussy, Villejuif, France). B cell lines were maintained in RPMI 1640 with 10% FCS (growth medium) and are described elsewhere (8).
Generation of plasmids and transfected cells
Plasmid pE-BARF0 containing the BARF0 open reading frame (ORF) in the expression vector EBO-pLPP was reported recently (5). The Flag epitope-tagged cDNA sequence of RK-BARF0 (a gift from N. Raab-Traub, University of North Carolina, Chapel Hill, NC) was also cloned into the vector EBO-pLPP. Conditions for cell transfection have been reported recently (5), and polyclonal cell cultures were maintained with 600 or 150 μg hygromycin B (Boehringer Mannheim, Indianapolis, IN) per ml of growth medium for DG75 or LCL transfectants, respectively.
Establishment of the RK-BARF0-specific CTL clone C1
CTL clones were generated from PBMC of donor NK (EBV/HLA-A*0201 positive) as described previously (9). Briefly, donor cells were stimulated with γ-irradiated autologous LCLs which were stably transfected with plasmid pE-BARF0 expressing the BARF0 ORF. T cell clones were grown in soft agar and expanded by cocultivation with the stimulator cells, and CTL clone RK-BARF0 C1 was selected. The specificity of this clone toward the HLA-A2-restricted peptide LLWAARPRL was defined with a standard 51Cr release assay with PHA blasts which were exogenously sensitized with synthetic peptides and in a cold target inhibition assay. FACScan analysis demonstrated that clone C1 was >98% CD3/CD8 positive. T cell cultures were maintained in growth medium supplemented with recombinant IL2 as described previously (9).
Target cells, either untreated, plasmid transfected, or peptide sensitized, were incubated with 51Cr for 90 min followed by the addition of effector cells in a standard 5-h 51Cr release assay (9).
RT-PCR analysis and sequencing
A detailed description for the preparation, DNase I treatment, and oligo(dT) primer-based reverse transcription of total cellular RNA as well as for the PCR conditions used was outlined recently (5, 8). The following PCR primers were used to amplify EBV sequences with 32 or 35 PCR cycles (the position within the EBV strain B95.8 (7) is given in parentheses): BARF0-F, 5′-GCCCGAGGAGCTGTAGACC (160308); LLW-F, 5′-TGTCCAGCGCTCTGGTCG (160586); BARF2-R, 5′-CCACGGCAAC CCTTCCAC (160812). The β2-microglobulin primers β2-M5′ (5′-CCCCCACTGAAAAAGATGAG) and β2-M3′ (5′-TCACTCAATCCAAATGCGGC) as well as the β-actin primers 0016 (5′-CACAGAGCCTCGCCTTTG) and 0017 (5′-TGGATAGCAACGTACATG) were used in 23 PCR cycles. The PCR protocol was as follows: 5 min denaturation at 95°C; cycles of 1 min at 95°C, 30 s at 60°C, 1 min at 72°C; followed by 5 min extension at 72°C. The PCR products were separated and visualized by electrophoresis on a 2.5% agarose gel containing ethidium bromide. The gel was photographed under UV light, and the film was analyzed on a Computing Densitometer 300 B system (Molecular Dynamics, Sunnyvale, CA). The isolated cDNA products were cloned with the pGEM-T vector system (Promega, Madison, WI) and sequenced with the PRISM Reader DyeDeoxy terminator cycle sequencing kit (Applied Biosystems, Foster City, CA) and M13 primers.
Results and Discussion
Differential splicing silences the endogenous presentation of a viral CTL epitope
CTL lines that recognized the HLA class I-restricted epitope LLWAARPRL encoded in the RK-BARF0 gene did not kill EBV-positive BL cells or LCLs that expressed RK-BARF0 (5). In this study, we explored the possibility that increased expression of the RK-BARF0 protein in EBV-infected B cells might restore RK-BARF0-specific CTL recognition. Thus, the HLA-A2-positive LCL NK was stably transfected with the plasmid pE-F-RKBARF0 (referred to as NK(E-F-RKBARF0)) which encodes the full length RK-BARF0 sequence, N-terminally tagged with the Flag epitope (Fig. 1,A). Surprisingly, neither NK(E-F-RKBARF0) nor the vector control (NK(E)) transfectants were recognized by the RK-BARF0-specific CTL clone C1 in a standard 51Cr release assay (Fig. 1,B). Similar results were also obtained using other HLA-A2-positive LCLs and EBV-positive BL cell lines as targets and polyclonal effector CTL lines specific for the LLWAARPRL epitope. This lack of killing was not due to impaired HLA-A2 expression by the NK LCL transfectants given that exogenous loading of LLWAARPRL peptide onto these target cells induced CTL-mediated killing (Fig. 1,B). Moreover, FACScan analysis with the use of the HLA-A2-specific mAb MA2.1 demonstrated high levels of HLA-A2 expression on the NK cell transfectants (data not shown). To check expression of the Flag-tagged RK-BARF0 protein in the cell transfectants, immunoblot analysis was used incorporating an anti-Flag mAb (Fig. 1 C). The expected full length 32-kDa RK-BARF0 protein was detected in the NK(E-F-RKBARF0) cells but not in the control cells (NK(E)). However unexpectedly, the majority of Flag-tagged proteins were truncated 16- to 20-kDa proteins.
One possible explanation for both the immunoblot result and the lack of killing of RK-BARF0 transfectants may relate to posttranscriptional processing of this transcript. To address this issue, RT-PCR was performed with the use of RNA from the NK(E-F-RKBARF0) cell line and BARF0-specific primers (BARF0-F and BARF2-R) (Fig. 2,B). This analysis revealed a surprising splicing pattern of the RK-BARF0 mRNA. The major products were splice cDNAs of 350 and 400/410 bp, whereas the full length 700-bp cDNA (which includes the LLWAARPRL epitope) was virtually undetectable (Fig. 2,A, lane 6). Sequencing of the RT-PCR products demonstrated that the differently spliced transcripts encoded three protein isoforms (150, 190, and 177 aa) the molecular masses of which correlated well with the observed mass of the 16- to 20-kDa truncated Flag proteins detected in Fig. 1,C. This observation was reinforced by a similar splicing pattern in a panel of EBV-positive BL (MutuI, MutuIII), LCL (NK, SB), and NPC (C15) cell lines (Fig. 2,A). Cloning and sequencing of the isolated cDNA products demonstrated three distinct splicing events which used a common 5′-splice site and three different 3′-splice sites with virtual perfect homology to the mammalian consensus splice sequences (summarized in Fig. 2 B). These splice variants encoded two truncated RK-BARF0 proteins (S#1, S#2) and a frame-shifted chimeric protein product (S#3), all of which lacked the LLWAARPRL epitope sequence. These data demonstrated that differential splicing significantly reduced the full length transcript that included the CTL epitope.
Abolition of differential splicing restores endogenous presentation and recognition of a viral CTL epitope
A logical prediction of the data was that removal of the 5′-splice site would abolish differential splicing of the RK-BARF0 transcript and result in increased cDNA encoding the LLWAARPRL epitope. To test this hypothesis, an EBV-negative, HLA-A2-positive BL cell line (DG75) was stably transfected with plasmids pE-BARF0 (expressing the BARF0 ORF without the 5′-splice site), pE-F-RKBARF0, or a control vector (Fig. 3,A). RT-PCR analysis of RNA from these transfectants using primers BARF0-F and BARF-2R showed strong differential splicing in the cell line containing plasmid pE-F-RKBARF0 (Fig. 3,B, lane 3). In contrast, RT-PCR analysis using a primer combination flanking the LLWAARPRL epitope (LLW-F and BARF2-R) revealed a single 227-bp cDNA product in BARF0-expressing cells (DG75(E-BARF0) and DG75(E-F-RKBARF0)) but not from the control cells (Fig. 3 B). Densitometric analysis (standardized to a 131- or 498-bp cDNA fragment of the housekeeping gene β2-microglobulin or β-actin, respectively) indicated that cells transfected with the splice-deficient BARF0 construct expressed ∼4-fold more LLWAARPRL epitope-encoding mRNA than cells expressing full length RK-BARF0.
To test whether these mRNA levels corresponded to an increase in endogenous presentation, these DG75 transfectants were analyzed for CTL-mediated killing using the RK-BARF0-specific CTL clone C1 (Fig. 3,C). Only the DG75(E-BARF0) transfectants expressing the splice-deficient BARF0 transcript were susceptible to lysis, whereas cells transfected with pE-F-RKBARF0 or the vector control were not recognized. Similarly, only the EBV-positive NK LCL stably expressing the splice-deficient BARF0 transcript (NK(E-BARF0)), but not the transfectants NK(E-F-RKBARF0) and NK(E), were killed by the RK-BARF0-specific CTL clone (Fig. 3 C).
The immunoblot analysis of the DG75(E-F-RKBARF0) and NK(E-F-RKBARF0) transfectants using an anti-Flag mAb revealed that the majority of the Flag-tagged proteins were 16–20 kDa in mass which correlated with the predicted molecular mass of the spliced cDNA variants lacking the LLWAARPRL epitope. In contrast, the full length RK-BARF0 (32kDa) protein was ∼9 fold lower expressed as indicated by densitometric analysis (Fig. 1 C and data not shown). Although the level of full length RK-BARF0 was below the threshold of T cell recognition, it was apparently sufficient to induce and maintain a specific CTL response in seropositive but not seronegative individuals (5). Expression of endogenous RK-BARF0 protein in the EBV-positive NK LCL transfectants was difficult to monitor as the polyclonal rabbit anti-RKBARF0 serum previously reported (10) showed cross-reactivity with cellular B cell proteins which were of a mass similar to that of the 32-kDa RK-BARF0 protein (our manuscript in preparation). Taken together, these results clearly demonstrated that differential splicing directly influences endogenous presentation of this viral CTL epitope in both transfected and EBV-infected target cells and that deletion of the unique 5′-splice site restores CTL killing.
In EBV-positive BL and NPC cells, there are different splice variants of the BamHI A transcript that share the BARF0 ORF at their 3′-terminal ends, giving rise to many different predicted protein products (3, 4). Our data in Fig. 2 showed that the BARF0 ORF was differently spliced in LCLs, BL, and NPC cells established with five virus isolates, indicating an evolutionary conservation of this splicing pattern. Thus, it seems that EBV takes advantage of a common cellular mechanism of gene regulation, differential splicing, which excises immunogenic fragments from the BamHI A protein products, thereby silencing T cell determinants. Recent reports from melanoma research have demonstrated that malignant cells, but not their normal counterparts, can express functional CTL epitopes (that are normally hidden within introns) which render them susceptible to melanoma-specific CTLs (11, 12).Although in this case differential splicing created new CTL epitopes, our data demonstrate a contrary function that results in silencing of a functional CTL epitope. Thus, differential splicing, as seen in many viruses and parasites, may not only have a role in regulating gene expression and generating protein isoforms but also contribute to immune evasion of pathogens and malignancies. Given the universal usage of differential splicing in eukaryotes, it is also likely that this process could be extended to the discrimination of self and nonself Ag and contribute to autoimmunity.
We thank the members of the EBV Unit at Queensland Institute of Medical Research, particularly for technical help provided by M. Buck, S. Cross, and L. Morrison; the critical manuscript reading of D. Moss; and P. Busson and N. Raab-Traub for providing cells and plasmids.
This work was supported by grants from the National Health and Medical Research Council, the Queensland Cancer Fund, and the University of Queensland, Australia. R.K. is supported by an R. D. Wright Fellowship from the National Health and Medical Research Council.
Abbreviations used in this paper: BL, Burkitt’s lymphoma; NPC, nasopharyngeal carcinoma; LCL, lymphoblastoid cell line; ORF, open reading frame.