Gamma-herpesvirus latency requires T cell evasion during episome maintenance

Abstract

The gamma-herpesviruses persist as latent episomes in a dynamic lymphocyte pool. Their consequent need to express a viral episome maintenance protein presents a potential immune target. The glycine-alanine repeat of the Epstein-Barr virus episome maintenance protein, EBNA-1, limits EBNA-1 epitope presentation to CD8(+) T lymphocytes (CTLs). However, CTL recognition occurs in vitro, so the significance of such evasion for viral fitness is unclear. We used the murine gamma-herpesvirus-68 (MHV-68) to define the in vivo contribution of cis-acting CTL evasion to host colonisation. Although the ORF73 episome maintenance protein of MHV-68 lacks a glycine-alanine repeat, it was equivalent to EBNA-1 in conferring limited presentation on linked epitopes. This was associated with reduced protein synthesis and reduced protein degradation. We bypassed the cis-acting evasion of ORF73 by using an internal ribosome entry site to express in trans-a CTL target from the same mRNA. This led to a severe, MHC class I-restricted and CTL-dependent reduction in viral latency. Thus, despite MHV-68 encoding at least two trans-acting CTL evasion proteins, cis-acting evasion during episome maintenance was essential for normal host colonisation.

Figure 1. Inhibition of MHC Class I–Restricted Epitope Presentation by Physical Linkage to ORF73

(A) The SIINFEKL epitope of OVA was introduced into ORF73 near either its N-terminus (ORF73-NC) or its C-terminus (ORF73-SC). Both ORF73 derivatives were cloned into the pcDNA3 expression vector and compared with OVA in the same vector for their capacity to stimulate the SIINFEKL-specific T cell hybridoma B3Z after transfection into L929-K^b cells. After 48 h, beta-galactosidase production was assayed by cell lysis in the presence of chlorophenol-red-beta-D-galactoside and reading absorbance at 595 nm. nil, vector only. (B) L929-K^b cells were co-transfected with OVA plus the plasmid indicated. C1–C4 are control plasmids, expressing MHV-68 ORFs 19, 30, 31, and 35, respectively. K3 degrades MHC class I heavy chains and m152 is a murine cytomegalovirus gene that retains MHC class I molecules in the endoplasmic reticulum. Net absorbance = A₅₉₅ with co-transfection − A₅₉₅ with untransfected cells (<0.02). (C) Hybrids of OVA and ORF73-SC were made to identify regions of ORF73 that inhibited SIINFEKL presentation. Responses are expressed as 100(A₅₉₅ with plasmid − A₅₉₅ with untransfected)/(A₅₉₅ with OVA transfection − A₅₉₅ with untransfected). nil, vector only. Mean ± standard deviation (SD) values of triplicate cultures are shown. Each graph is representative of at least three separate experiments. In at least one experiment, equal transfection efficiency was confirmed by co-transfecting a GFP expression plasmid and checking fluorescence under ultraviolet illumination. (D) ORF73 was fused to the C-terminus of the OVA coding sequence in pcDNA3. C-terminal deletions were then made as shown. Each construct was transfected into L929-H2-K^b cells. The shaded area in (D–G) highlights a region of ORF73 that appeared to be important for inhibiting epitope presentation. (E) N-terminal ORF73 truncations were generated by PCR and fused in frame to amino acid 325 of OVA. Each construct was transfected into L929-H2-K^b cells and assayed for SIINFEKL presentation as in (D). (F) Progressive truncations of ORF73-SN were assayed for their capacity to present the SIINFEKL epitope to B3Z cells after transfection into L929-K^b cells. Selective presentation from the ORF73-SN-PstI construct was confirmed in multiple experiments, including independent plasmid preparations. (G) PCR-generated C-terminal truncations of ORF73-SN were assayed for SIINFEKL presentation after transfection of L929-H2-K^b cells. Deletions across the area identified as important for inhibiting epitope presentation in (D–E) again improved epitope presentation.

Figure 2. Inhibition of Epitope Presentation by ORF73 Fusion to OVA Correlates with Reduced Translation of the Fusion Protein

(A) N-terminal ORF73 truncations equivalent to those in Figure 1E were fused in frame to the C-terminus of OVA amino acids 41–325, thereby removing both the OVA signal sequence and the ORF73 nuclear localisation signal. (B) Serial dilutions of an expression plasmid containing each fusion gene were transfected into L929-H2-K^b cells as in Figure 1. SIINFEKL presentation was assayed using beta-galactosidase production from the B3Z hybridoma. (C) Equivalent transfected cells were immunoblotted with an anti-OVA rabbit serum (OVA). Fusion products are indicated by arrowheads where visible. Parallel immunoblots for neomycin phosphotransferase II (NPT), which is expressed from a different promoter of the same plasmid (pcDNA3), were used to control for transfection efficiency. The endogenous neomycin phosphotransferase II expressed by 293T cells was not visible at this exposure. One of three equivalent experiments is shown. (D) Forty-eight hours after transfection with the constructs indicated, 293T cells were pulse-labelled (P) for 30 min with ³⁵S-cysteine/methionine, followed by a 2-h chase (C) with excess unlabelled cysteine/methionine. OVA derivatives were then immunoprecipitated with an OVA-specific rabbit serum and resolved by SDS-PAGE. The specific bands corresponding to each fusion protein are indicated by arrowheads. The graph shows densitometry readings for each band. (E) 293T cells transfected with selected fusion proteins were labelled for a variable period (15–120 min) as indicated. OVA derivatives were then immunoprecipitated and analysed as in (D). Arrowheads show the predicted position of the relevant fusion proteins for the 120-min label samples. (F) Either SOVA-ORF73A or SOVA was transfected into 293T cells. Forty-eight hours later the cells were pulse-labelled (P) for 15 min with ³⁵S-cysteine/methionine, followed by a 15-min (C1), 45-min (C2), and 105-min (C3) chase with excess unlabelled cysteine/methionine. This was done in the presence or absence of 100 μM lactacystin. The graph shows densitometry readings for each specific band.

Figure 3. Modification of the MHV-68 Genome to Overcome cis-Acting Immune Evasion by ORF73

(A) An IRES element was inserted just downstream of ORF73, between its stop codon and that of M11. This allowed either three tandem CD8⁺ T cell epitopes (EPI) or GFP to be translated from the ORF73 mRNA. (B) DNA from BAC-cloned viral genomes (BAC) or virus-infected cells (VIR) was digested with NcoI, electrophoresed, transferred to nylon membranes, and blotted with a probe corresponding to the BamHI-G genomic fragment shown in (A). The predicted bands for WT virus were 1,021 bp, 3,121 bp, and 4,630 bp. The IRES-GFP insert introduced an NcoI site such that the WT 3,121-bp band was cut into 2,975-bp and 1,466-bp fragments. The NcoI site was lost from the IRES-EPI insert, such that the WT 3,121-bp band became a 3,861-bp band. (C) BHK-21 cells were infected (0.01 PFU/cell) with WT, GFP, or EPI viruses as indicated. Plaque titres of cell cultures are shown with time after infection. (D) H2^b MEF-1 cells or L929-K^b cells were left uninfected (UI) or infected for 2 h with MHV-68 expressing either OVA under a strong lytic promoter (OVA) or the SIINFEKL epitope of OVA as part of the ORF73-IRES-EPI construct (EPI). B3Z cells were then added, and 18 h later their beta-galactosidase response was assayed using chlorophenol-red-beta-D-galactoside substrate. Mean ± SD values of triplicate cultures are shown. The data are from one or two equivalent experiments. (E) A20-syndecan-1 cells were infected (20 PFU/cell) with GFP⁻ WT virus, WT virus with an HCMV IE1 promoter-driven GFP expression cassette (HCMV IE1-GFP), or with the ORF73-IRES-GFP virus. The numbers indicate the percentage of total cells in the gated region (GFP⁺). Expression from the HCMV IE1 promoter is probably limited to lytic infection, whereas ORF73 is expressed in latency.

Figure 4. Replication of the IRES-EPI Virus In Vivo

(A) Six days after intranasal infection with WT or EPI viruses as indicated, infectious virus in lungs was titred by plaque assay (left panel) and infectious plus latent virus in spleens was titred by infectious centre assay (right panel). Each point shows an individual mouse. Pre-formed, infectious virus was undetectable in equivalent, freeze-thawed spleen samples, so the infectious centres represent latent virus. (B) By 14 d post-infection, infectious centre titres were much lower with the EPI virus than with WT. The GFP control virus is shown for comparison. This difference was preserved at day 19 post-infection, indicating that the EPI virus was not merely delayed in host colonisation. (C) DNA was extracted from spleens and its viral genome content quantitated by real-time PCR. Genome loads broadly reflected the infectious centre titres, indicating that the viral load was reduced rather than the efficiency of ex vivo reactivation. (D) As a further measure of host colonisation, we measured B cell activation (CD69 expression on CD19⁺ B cells) at 14 d post-infection and CD8⁺ T cell activation (loss of CD62L expression) at 19 d post-infection. We also measured the day 19 expansion of the Vbeta4⁺CD8⁺ T cell subset that is characteristic of MHV-68-associated infectious mononucleosis. All these measures correlated closely with the viral latent load in lymphoid tissue and were markedly reduced with the EPI virus compared to WT or GFP. GFP expression was undetectable in ex vivo B cells after infection with the GFP virus (data not shown). (E) C57BL/6J mice were infected intranasally with WT virus, the EPI mutant, an independently derived EPI mutant (EPI-IND), or a revertant of the EPI virus (EPI-REV). Splenic infectious centres were then measured 13 and 17 d post-infection. The dashed line shows the lower limit of assay sensitivity.

Figure 5. Antigen-Specific Immune Responses to the IRES-EPI Virus

(A) CD8⁺ and CD4⁺ T cell responses were measured by interferon-gamma ELISPOT assay 13 d post-infection. The response to virus-exposed targets (VIR) is mediated by CD4⁺ T cells; the response to the p56, p79, and SIINFEKL (OVA) peptides is mediated by CD8⁺ T cells [62]. The mean number of spots with untreated targets was subtracted from the number of spots with each specific target. There was a response to the OVA peptide in the IRES-epitope construct, but not to the ASNENMETM peptide (NP). Mean ± SD values of five mice per group are shown. (B) Total and MHV-68 virion-specific serum IgG responses were measured by ELISA at 18 d post-infection. “Naive” indicates age-matched, uninfected controls. Mean ± SD absorbance values of four mouse sera per group are shown. (C) Spleen cells were stimulated for 5 h in the presence of Brefeldin A plus the peptide indicated and then stained for cell-surface CD8 and intracellular interferon-gamma. The percentage of interferon-gamma⁺ CD8⁺ cells without peptide was subtracted from the value with peptide to give the specific response. Mean ± SD values of five mice per group are shown.

Figure 6. Normal EPI Virus Replication in Non-H2^b Mice

(A) Infectious centre titres in individual spleens were determined 14 d after intranasal infection of C57BL/6J (H2^b) or BALB/c (H2^d) mice with WT or EPI virus. (B) CD69 expression on splenic B cells was measured by flow cytometry 14 d post-infection. B cells from uninfected mice were less than 5% CD69⁺. (C) The weights of individual spleens are shown 14 d post-infection, with spleens of age-matched, uninfected mice (UI) for comparison. (D) Viral genome loads in individual mice were determined by real-time PCR at 14 d post-infection. (E) Viral tRNA expression in infected germinal centres was visualised by in situ hybridization with a digoxigenin-labelled riboprobe specific for tRNAs 1–4. Representative follicles of at least five sections per mouse and three mice per group are shown. tRNA⁺ follicles were abundant with WT virus and with the EPI virus in BALB/c mice, but were not seen with the EPI virus in C57BL/6J mice. (F) BALB/c or C57BL/6J mice were infected intranasally with MHV-68 expressing OVA from an intergenic expression cassette (OVA) or with WT virus. The extent of lymphoid colonisation was determined by infectious centre assay of spleens 12 and 15 d post-infection. Mean ± SEM titres of five mice per group are shown. In contrast to the EPI virus, the OVA virus showed no defect in host colonisation.

Figure 7. Rescue of the EPI Virus by CD8⁺ T Cell Depletion

Mice were left undepleted (UD) or depleted of CD8⁺ T cells (CD8⁻) by an initial intravenous injection of mAb YTS169 2 d before infection, followed by intraperitoneal injections of the same antibody every 2–3 d up to the time of sampling. Infection was by intranasal inoculation of either WT or EPI viruses. (A) Depletion was 95%–99% complete as assessed by flow cytometry of spleen cells. CD69 expression on splenic B cells was measured 13 d post-infection. (B) Genome loads were measured 13 d post-infection by real-time PCR. Each point shows an individual mouse. (C) The infectious centre titres of individual mice at 13 d post-infection are shown for one of two equivalent experiments. The titres of pre-formed, infectious virus in freeze-thawed spleens were less than 5% of the infectious centre titres, so even after CD8⁺ T cell depletion, the infectious centre assay essentially measured latent virus. By 13 d post-infection, the lungs of both immunocompetent and CD8⁺ T cell–depleted mice were clear of infectious virus. (D) In situ hybridization for viral tRNA expression in splenic germinal centres is shown 13 d after infection of CD8⁺ T cell–depleted or undepleted mice, infected with either EPI or WT virus. Spleens of two representative mice are shown in each case.