Epstein-Barr virus (EBV) is a gamma herpes virus persistently infecting over 90% of the adult population worldwide. It has been aetiologically linked to a number of human malignancies, including more than 90% of post transplant lymphoproliferative disease (PTLD), 50% of Hodgkin’s lymphoma (HL), virtually all undifferentiated nasopharyngeal carcinoma (NPC), and approximately 10% of gastric carcinoma (GC). As EBV infection in healthy individuals is mainly controlled by virus specific cytotoxic T lymphocytes (CTLs), we hypothesise that engineering T cells with chimeric T cell receptors (cTCRs) specific for EBV latent membrane proteins (LMPs) will confer on these cells the ability to target and kill the malignant cells of cancers associated with Epstein-Barr virus. Thus, the aim of this project was generate these engineered T cells and to set up a severe combined immunodeficient (SCID) mouse model in which to test their effectiveness. Three EBV-infected cell lines derived from HL, NPC and GC gave rise to tumours in 11 of 12 (92%), 12 of 12 (100%) and 10 of 10 (100%) SCID mice respectively, when 1x107 cells were injected subcutaneously. Immunohistochemical analysis showed that the HL SCID tumours were CD4-, CD15-, CD20+, CD30+, consistent with a HL Reed-Sternberg cell phenotype, and NPC and GC SCID tumours expressed the epithelial cell marker cytokeratin. Furthermore, all tumours expressed EBVencoded RNAs (EBERs) and LMP1. This was identical to parent cell line expression patterns, and hence growth in vivo did not affect cell phenotype. T cells were successfully transduced with a retroviral vector encoding a CD19-specific cTCR (CD19- cTCR) with a mean transduction rate of 13%±6%. Transduced cells were cytotoxic for HL-derived L591 cells in vitro, with specific lysis of 24%±11% at an effector to target ratio of 20:1. This was significantly higher than specific lysis seen in mock transduced cells (p>0.05). At a tumour inoculation dose of 5x106, in vivo sc transfer of 5x107 CD19-cTCR transduced cells was able to prevent HL tumour development in 6 of 6 (100%) test mice, whereas 17 of 22 (77%) control mice and 2 of 3 (66%) mice treated with unmodified EBV-specific CTLs developed tumours. Moreover, iv transfer of 5x107 CD19-cTCR transduced cells mediated complete regression of HL SCID tumours in 3 out of 6 (50%) mice. Phage display selection experiments to isolate a single chain antibody fragment (scFv) specific for viral LMPs for incorporation in a cTCR were performed. Linear, biotinylated and cyclised biotinylated peptides derived from the external reverse turn loops of LMP2 were used as target antigens. Despite extensive testing, no reactive clones specific for the peptides were identified. The ability of CD19-cTCR transduced cells to specifically lyse HL cells in vitro, and clear tumour burden in vivo, supports a future role for engineered T cells in the treatment of HL. Despite the lack of success in isolating a scFv for LMP2, the use of viral antigen specific, cTCR redirected T cells remains in principle a valuable therapeutic alternative for EBV-associated malignancies. The SCID models for HL, NPC and GC will provide a useful preclinical tool for investigation of their efficacy in vivo.

Epstein-Barr virus (EBV) is a gammaherpes virus that infects >90% of the adult population worldwide. During childhood infection is generally sub-clinical, however if delayed until adolescence infectious mononucleosis (IM) may develop. The virus has also been aetiologically linked with a number of tumours including B-cell lymphoma following organ transplantation: post-transplant lymphoproliferative disease (PTLD). The symptoms of IM are caused by an expansion of immune cells in response to infection whilst in the transplant situation immunosuppressive drug therapy allows the outgrowth of the tumour. Understanding the immuno-regulatory mechanisms involved in such EBV-associated diseases is crucial for devising new treatment strategies. We undertook 3 separate studies (1-3) investigating different aspects of the immune response to EBV. In a recently reported phase II trial using allogenic, EBV-specific cytotoxic T-cell (CTL) to treat PTLD, tumour response was significantly increased with a high degree of donor/recipient HLA-allele matching suggesting that further refinement of the matching procedure may be important. In study 1 we investigated the epitope specificity and T-cell receptor (TCR) clonality of the infused CTL to identify potential areas for refinement. We found the protein specificity of the CTL to be polyclonal with dominant recognition of Epstein-Barr nuclear antigen-3 proteins and sub-dominant recognition of Latent membrane protein (LMP)-1 and LMP-2 proteins. Where possible, specificity was confirmed at the peptide level. No single TCR family was preferentially used by CTLs. The CTL epitope specificity did not differ between treatment responders and non-responders however the response was improved in those with several CTL HLA-restricted epitope matches and those infused with CTL containing polyclonal TCR families as opposed to monoclonal. CTL/recipient matching based on HLA matching alone was improved when also matched via HLA- restricted epitiope specificity. Therefore mapping CTL peptide epitope specificity prior to CTL infusions may enhance patient responses. In recent years, interest has developed in genetic variation within components of the immune system. Of particular interest are cytokine/cytokine receptor genes and genes of the human leukocyte antigen (HLA), both of which act to regulate the immune response. Variation within these genes could potentially alter the immune response leading to disease. In study 2 we investigated single nucleotide polymorphisms (SNPs) in several cytokine genes (TNF, IL-1, -6, -10) in both IM and PTLD cases and compared with relevant control groups. We found that the frequency of two TNF promoter alleles was significantly increased in PTLD patients compared to controls whilst the frequency of a TNF receptor II allele was increased in IM and EBV seropositive individuals, suggesting a role for this allele in susceptibility to EBV infection. The frequency of a second TNF receptor II allele was increased in both PTLD and IM subjects compared to controls highlighting the possible significance of TNF and its receptor in the development of EBV associated disease. In study 3 we analyzed two microsatellite markers and two SNPs located near the HLA class I locus in IM, PTLD and control subjects to further determine whether the HLA genes may affect development of EBV-associated diseases. Alleles of both microsatellite markers were significantly associated with development of IM. Specific alleles of the two SNPs were also more frequent in IM patients. Moreover IM cases possessing the associated microsatellite allele had significantly fewer lymphocytes, increased neutrophils, and displayed higher EBV titres and milder IM symptoms relative to IM cases lacking the allele. The results indicate that HLA class I polymorphisms may predispose patients to development of IM upon primary EBV infection and suggest that genetic variation in T cell responses can influence the course of EBV infection.

During latent Epstein-Barr virus (EBV) infection, the T helper cell response to the EBV latent membrane protein (LMP)1 is dominated by the production of IL-10, but by IFN- γ during acute EBV infection. The purpose of this thesis was to develop methods for the enumeration and characterisation of the IL-10 producing CD4+ T cells that respond to peptides of the EBV protein LMP1, and to investigate the possible involvement of IL-27 in the development of these cells. It was found that some human donors have very high concentrations of IL-27 within serum, which is not dependent on EBV infection status but demonstrates relatively low heritability. The addition of IL-27 to cultures of human T cells did not induce IL-10 but the production of IL-17 was inhibited. To identify and characterise LMP1 responsive cells I used CD154 as a marker of activated T cells. Having optimised the methodology, 14 donors of known EBV serostatus were tested for activated IL-10 producing cells after culturing peripheral blood mononuclear cells with LMP1 peptides. However in most cases the frequency of CD4+ lymphocytes upregulating CD154 and IL-10 in response to LMP1 peptides was below the assay’s sensitivity. When the CD154+IL-10+ CD4+ cells were stained for T helper cell subset markers they were positive for every marker and isotype control, suggesting that the cells were non-specifically binding labelled antibody. Both single positive CD154+ and single positive IL-10+CD4+ cells were also present in response to LMP1, but again typically at frequencies below the level of sensitivity for this assay. The conclusions of the work are that IL-27 responses are heterogeneous, but unlikely to play an important role in the induction of IL-10+ T cells in EBV infection. The frequency of LMP1 responsive T cells is very low (<0.08%) in most donors.

Immune responses to Epstein-Barr Virus (EBV) encoded Latent Membrane Protein 1 (LMP1) peptides in seropositive donors are dominated by the induction of IL-10 secretion. These IL-10 responses, characteristic of T regulatory 1 (Tr1) cells, were able to inhibit T-cell proliferation and interferon-γ (IFN-γ) section induced by other antigens. Furthermore, this inhibition was specific to the co-presented antigen and persisted after LMP1 peptide had been removed. Thus, it may be possible to exploit such specific induction therapeutically to inhibit pathogenic responses in immune-mediated diseases. The current study was initiated to confirm and extend these findings and then to investigate the ability of LMP1 peptides to inhibit pathogenic responses to Rh autoantigen in AIHA patients <i>in vitro</i>. Inhibitory properties of LMP1 peptides were confirmed, although most such inhibition appeared non-specific and was not associated with IL-10. Inhibition appeared to involve an effect on antigen presenting cells. When autologous red cells or RhD peptides were used as stimulating antigens in patients with AIHA, IL-17 responses were more frequent than IFN-γ secretion. Furthermore, disease activity correlated better with IL-17 responses than TH1 responses. These data therefore suggest that the pathogenesis of AIHA has a substantial Th17 component. Finally, these autoreactive responses could be inhibited by LMP1 peptides.

Epstein-Barr virus, or EBV, is a human herpesvirus that is nearly universally present in most of the human population. The infection rate in adults is quickly approaching an astonishing 95% worldwide. While the most common clinical manifestation of EBV infection is infectious mononucleosis, there are multiple malignancies strongly associated with the virus, such as nasopharyngeal carcinoma (NPC). A contemporary problem in public health is the scarcity of markers to predict the development of EBV-associated malignancies, which prevents a possible cure or a suitable treatment. Of the approximately 85 proteins that the virus expresses, EBV serological tests rely on detection of antibodies against only three antigens: BFRF3 (viral capsid antigen or VCA), BMRF1 (early antigen or EA), and BKRF1 (nuclear antigen or NA). Antibodies against two of these antigens, VCA and NA, are produced by almost all EBV carriers for their entire lifetime, and so detection of these antibodies can make it difficult to differentiate between non-cancerous EBV carriers and EBV-associated cancer patients. To address this problem, the project aimed to identify a set of markers to diagnose NPC and treat it. To test this, 3 EBV genes were cloned with a 3X FLAG-tag into a mammalian expression vector, then expressed and detected by immunoblotting with a monoclonal FLAG antibody and each of the 9 human serum samples used in this study. Our results were generally inconclusive, and further studies are warranted to explore possible diagnostic and prognostic biomarkers.

Thesis (Ph. D.)--University of Hong Kong, 2001. / Includes bibliographical references (leaves 158-191).

Thesis (M. Phil.)--University of Hong Kong, 1989.

published_or_final_version / Dentistry / Doctoral / Doctor of Philosophy

 In Epstein-Barr virus (EBV)-associated malignancies, the virus is harbored in every tumor cell and persists in a tightly latent form (latency I, II or III) expressing a limited number of viral latent proteins. Induction of EBV lytic cycle, which triggers expression of a much larger number of viral proteins, may lead to therapeutic effects against EBV-associated cancers. We previously found that suberoylanilide hydroxamic acid (SAHA), a FDA-approved histone deacetylase inhibitor, induced EBV lytic cycle and mediated enhanced cell death in EBV-positive gastric carcinoma cells (latency II). In this thesis, we sought to investigate SAHA’s induction of EBV lytic cycle and its cellular consequences in EBV-associated epithelial malignancies, with particular focus on nasopharyngeal carcinoma (NPC) due to its strong association with EBV and high prevalence in southern Chinese populations. SAHA effected strong induction of EBV lytic cycle in EBV-positive epithelial malignancies, including gastric carcinoma and NPC, as evidenced by strong expression of EBV lytic proteins, replication of viral DNA and production of infectious viral particles. Immunofluorescent staining revealed that up to 70% EBV-positive epithelial cancers expressed EBV lytic proteins following treatment with micromolar concentrations of SAHA. However, SAHA could not induce EBV lytic cycle in NK lymphoma cells (both NPC and NK lymphoma express EBV latency II pattern), indicating preferential viral lytic induction in epithelial rather than lymphoid malignancies. EBV lytic cycle induction in NPC by SAHA required activation of protein kinase C-delta (PKC-) and acetylation of non-histone protein but required neither phosphatidylinositol 3’-kinase (PI3K), MAPK/ERK kinase (MEK), c-Jun aminoterminal kinase (JNK) nor p38 stress mitogen-activated protein kinase (MAPK) signaling pathway. Conflicting observations regarding the effect of EBV lytic cycle induction on apoptosis were reported. Thus, we investigated the relationship between EBV lytic cycle induction and apoptosis in NPC following treatment with SAHA. EBV-positive NPC showed a higher percentage of apoptosis and proteolytic cleavage of PARP, caspases-3, -7 and -9 over EBV-negative NPC and greater than 85% of NPC cells co-expressed EBV immediate-early (Zta), early (BMRF1) or late (gp350/220) lytic proteins and cleaved caspase-3. Tracking of expression of these lytic proteins over time demonstrated that NPC proceeded to apoptosis following EBV lytic cycle induction, contrary to the previously reported anti-apoptotic effect of EBV lytic proteins in Burkitt lymphoma. Analyses of cleaved caspase-3 expression upon RNAi knockdown and exogenous expression of Zta further supported that EBV lytic cycle directly led to apoptosis of EBV-positive NPC cells. Interestingly, inhibition of EBV DNA replication and late lytic protein expression by phosphonoformic acid did not impact on SAHA’s induced cell death in NPC, indicating that early rather than late phase of EBV lytic cycle contributed to the apoptotic effect. Finally, in vivo effects of SAHA on EBV lytic cycle induction and tumor growth suppression were observed in NPC tumors established in nude mice. In conclusion, activation of EBV lytic cycle from latent cycle in EBV-positive epithelial malignancies including NPC by SAHA effected apoptosis and tumor growth suppression of the cancer cells and provided experimental evidence for virus-targeted therapy against EBV-positive cancers. / published_or_final_version / Paediatrics and Adolescent Medicine / Doctoral / Doctor of Philosophy

Whether certain Epstein-Barr virus (EBV) strains are associated with pathogenesis of nasopharyngeal carcinoma (NPC) is still an unresolved question. In the present study, we aimed to sequence the complete EBV genomes harbored in NPC tumor biopsies and compare against the non-NPC EBV strains to identify NPC-specific EBV variations. In the first part of the study, EBV genome contained in one primary NPC tumor biopsy was PCR-amplified and sequenced using next-generation and dideoxy-DNA sequencing. The EBV genome, designated HKNPC1 (Accession number JQ009376), was generated by reference mapping and it appears to be a uniform strain in general despite minor heterogeneity. Phylogenetic analysis with the four published EBV strains, B95-8, AG876, GD1, and GD2, indicated HKNPC1 was more closely related to the Chinese NPC strains. HKNPC1 contains 1,589 single nucleotide variations (SNVs) and 132 insertions or deletions (indels). We found 76 non-synonymous SNVs shared amongst the Chinese GD1, GD2 and HKNPC1 isolates, while another 88 nonsynonymous SNVs were shared only by the two NPC tumor-derived strains HKNPC1 and GD2. In the second part of the study, SureSelect target enrichment technology was used instead of PCR to capture EBV DNA from total DNA. The study was scaled-up to sequence EBV strains in cell lines, saliva and NPC tumor, using the MiSeq Personal Sequencer and the Genome Analyzer IIx platforms. The reads were de novo assembled to generate 17 complete EBV genomes, out of which 9 were NPC-EBV strains. Phylogenetic analysis of all available EBV strains has demonstrated that all NPC strains were type 1 EBV. Phylogeny predicted by LMP-1 gene showed clear geographical pattern of where the EBV strains were isolated. A total of 5,011 variations were identified by comparing every EBV strain against the reference. MicroRNAs and EBERs are generally well conserved across all genomes. Comparative analysis of variations between NPC and non-NPC EBV strains discovered 904 NPC-specific variations, out of which 112 appeared in more than one NPC strains. Among these recurrent variations, 39 non-synonymous substitutions and seven deletions in coding region were found. About half of these recurrent variations were located in EBNA-3A, -3B and -3C, while the rest was found in latent, tegument, capsid and packaging-related proteins and transcription factors. There were two NPC EBV strains isolated from the primary tumors which later diagnosed to have distant metastasis. Unique variations were shared in these two EBV strains in regions between IR2 and IR3, where genes such as BPLF1, BOLF1 and EBNA-3A, -3B and -3C were located, and leftward of IR3, where BBLF2/3 and BBRF1 were found. In conclusion, we have demonstrated the feasibility of target capture and next-generation sequencing in whole genome sequencing of EBV. Comparison of reference mapping and de novo assembly of EBV sequences illustrated that both are feasible approaches, though de novo assembly is preferred since the method is less dependent on the reference genome. Large-scale sequencing of NPC and non-NPC EBV strains may facilitate the discovery of previously unknown variations of biological significance and reveal the diverse role of EBV in NPC pathogenesis. words) / published_or_final_version / Paediatrics and Adolescent Medicine / Doctoral / Doctor of Philosophy