Virus driven transformation of lymphoid cells is a major clinical consequence of infection with persistent infections such as EBV and HTLV in humans. Progress in understanding these human diseases is hindered by the lack of suitable model systems. MDV represents a natural α-herpesvirus of galliform birds capable of inducing rapid onset of tumors in susceptible birds. Losses caused by this group of viruses also represent a substantial problem in their own right; without MDV vaccination the poultry industry would be unsustainable. Indeed the ability to vaccinate against MDV tumor formation has implications for control of medically relevant tumors 
. Within this framework, we addressed the issue of T cell clonality during infection and tumor formation, dissecting the tumor, spleen and blood to identify repertoire changes in the transformed CD4+ cells and the responding CD8+ cells. With MD almost all cell lines and in vivo tumors have been characterised as CD4+, 
. In one study using the intraperitoneal infection route one of twelve cell lines was CD4- CD8α+ but this lacked expression of CD8β 
. All of the CD8+ samples in this study were prepared using anti-CD8β to avoid isolation of non-classical CD8αα T cells. We chose to examine the Vβ profiles as a measure of clonality since this receptor is clonally expressed with a single in-frame sequence present in each clone of T cells due to the process of allelic exclusion that takes place during T cell development in the thymus 
Tumor clonality is a fundamental issue in MD pathogenesis. The infectious cycle involves transfer of the virus from the lungs to initiate a cytolytic infection in B cells. This is followed by spread and lytic cycling infection largely within CD4 TCRαβ T cell population, before development of latent infection and transfer of MDV into the feather follicle epithelium from where the infectious virus is shed 
. All infected birds experience a persistent, latent infection, and susceptible birds develop tumors usually within 4 to 5 weeks. Herein resides the problem; if the transformation event is rare, how to explain the high penetrance and temporal reproducibility of the tumor phenotype, unless the “tumors” are induced as a result of polyclonal transformation. Previous studies have addressed this issue in relation to the pattern of MDV genomic integration within the host cell genome 
or by cell surface staining with CD30 as a tumor associated marker 
. These two studies reached opposing conclusions, with the restricted MDV integration profiles used to propose clonal tumors (with metastasis), contrasted with the high expression of CD30 in both TCRαβ families within a single tumor being used to propose polyclonality. Our studies using TCRVβ repertoire analysis techniques 
as a viral integration independent clonal “bar-code” to identify the repertoire of CD4+ TCRαβ+ T cells in tumor-derived cell lines and with in vivo
derived tumor samples revealed a characteristic of clonal dominance within an oligoclonal framework of tumor-capable CD4+ T cells.
All of the established tumor-derived cell lines were monoclonal (each expressing a single TCRβ CDR3 sequence), although one short-term line developed during the course of these studies was biclonal at second passage. In contrast the REV-T-transformed AVOL-1 cell line was oligoclonal after over 37 passages expressing at least three TCRVβ1 and one Vβ2 TCR CDR3 sequences. The spectratype profiles obtained with all cell lines were diagnostic in terms of the clonality of the CDR3 as defined by sequence analysis. The clonal structure of the cell lines was not influenced by the length of time in culture which suggests that monoclonality is not an artefact of in vitro
selection as a result of multiple passages. It is therefore likely that selection for dominant transformed clones had already occurred in vivo
and is retained in MDV cell lines as suggested previously 
. Furthermore, cell lines generated in this study from fresh tumors expressed a TCR identity shared with the source tumor in vivo
. Where cell lines were established most (~90%) expressed the Vβ1 family of T cell receptors with only one expressing Vβ2, a ratio consistent with the 84% bias previously reported 
The profile of most primary tumors was dominated by a single clone of transformed T cells, although biclonal dominance in individual tumor sites was not uncommon. However, sequence analysis revealed smaller secondary clones of expanded CD4+ T cells in most tumors (~10% of the CDR3 sequences) and the outgrowth of one of these during ex vivo culture indicates the tumor potential of sub-dominant CD4+ clones. Some of the very large clonal populations also failed to establish as tumor cells lines ex vivo, perhaps indicating a phenotypic variability in transformation state. Indeed, considering the very large TCR clones (40 to 100% of CD4+ CDR3 in one site) these were evenly distributed between TCRVβ1 (9 sequences) and TCRVβ2 (8 sequences) (). The bias in TCRVβ usage within cell lines may represent a cultivation artefact or reflect the biology of cells expressing different TCR family members. Nonetheless, the multi-step analysis of dominant CDR3 in the primary tumor, in sorted CD4+ cells and after establishment of lymphoblastoid cell lines ex vivo are important in confirming the capacity of the identified large clones to express a tumor capability. All four Jβ segments were present in the CDR3 of both TCRVβ1 and TCRVβ2 expressing large tumor-like clones or in cultured tumor cell lines.
Our data resolves many of the issues surrounding MD tumor clonality. Essentially, we demonstrate clonal dominance within MD tumors (broadly similar to that reported by Delacluse et al., 
) although our integration-site independent analysis using the T cell receptor CDR3 region revealed a more complex clonal framework within, and between, tumor sites in vivo
. Different tumor sites within a single individual may be dominated by shared or distinct clones, hence a single individual may experience multiple transformation events giving rise to tumors that have very different characteristics. On most occasions the dominant in vivo clone present at a particular site was the only clone represented in ex vivo
cultured cell lines grown under tumor culture conditions. However, on one occasion one of the lower frequency clones exhibited tumor-like growth patterns ex vivo (alongside the dominant clone in the original site) indicating that some of the smaller clones exist in a transformation capable state. The fact that many individuals harbour both metastatic and single-site tumor clones indicates a complex interplay between transformation and clonal competition. Indeed, with most individuals the overall tumor burden was the result of a small number of independent transformation events (i.e. more than one but fewer than 3 or 4). In contrast, with some individuals the multi-site tumors were the result of metastasis from a single tumor clone. The relationship between these “successful” tumor clones and the infected cell population deserves attention.
In a broader context, the monoclonal origin of adult T-cell leukaemia/lymphoma (ATLL) induced by the human T-lymphotropic virus type -1 (HTLV-1) associated malignancy is well documented 
. This profile is probably related to the rarity of ATLL even among HTLV-1 seropositive individuals 
reflecting the acquisition of secondary genomic lesions in persistently infected T cells. Nonetheless, the rapid onset MD tumors with clonal dominance in the context of a more complex framework of oligoclonal expansion may also reflect a circumstance common to other tumor associated persistent viruses of lymphocytes including HTLV-1. Perhaps the main differences may lie in the vigor of MDV-induced T cell replication leading to a compressed time-frame compared with other lymphotropic, tumor associated viruses.
Biological differences were also detected amongst the very large clonal CD4+ “tumors”, with some clones found in multiple sites including the blood and spleen whereas others were located in a single site, indicating phenotypic diversity based upon metastatic capability. The identification of metastatic tumor clones in the blood allowed serial analysis of blood samples from infected birds to determine the dynamics of the appearance of the tumor clone, in relation to the time of infection and onset of clinical signs. The spectratype analysis of blood samples prior to infection and in the first 10–14 days revealed a profile consistent with a polyclonal population of circulating cells. However in some cases, the ‘tumor-specific’ spectratype signature could be detected in blood 12 to 16 dpi, more than two weeks before appearance of clinical signs. The appearance of the tumor clone at detectable levels in the blood supports the proposal of an early transformation event. The level of tumor clone expansion in the blood compartment at the onset of clinical disease was extreme, and in some individuals, these were the only T cell clones detectable (e.g. within TCRVβ1 for Bird15 and 17) represented the tumor ( and ). There was also evidence for disturbance within the polyclonal repertoire in TCRVβ2 expressing cells () suggesting that the blood niche for T cells was being filled by the tumor. Hence, with a circulating TCR profile dominated by a single clone, it is of little surprise that MDV-infected birds develop immune deficiency [reviewed in 
]. These dramatic repertoire changes would have greater impact than the reported changes in cytokine production 
and would be immunologically catastrophic. Infiltration of the skin with CD4+ T cells is a consequence of MDV infection 
,  
and the high frequency tumor clones in the blood are likely to represent the relocation of MDV to the site of transmission.
In mammals, many persistent viral infections including EBV, CMV and HTLV stimulate highly focussed repertoire expansion in responding CD8+ T cells 
. Although the MDV tumors were populated by relatively small numbers of CD8+ T cells, their repertoire was highly structured and oligoclonal in nature. The CD8+ T cell clone sizes of around 25 to 50 million cells are similar to those reported during persistent viral infections in humans 
. However in the case of MD, these are developed over a much shorter period of time than considered with mammalian infections. For example, taking a conservative estimate of prolonged T cell division of 12 hours/division 
and assuming no cell death (unlikely), the latest time point for initial stimulation of the CD8+ T cell would be ~15 days prior to sampling. This calculation would place the initiation of these clones of specific CD8+ T cells at ~15 DPI, probably earlier, around the time at which latent infection was initiated. The rapid focussing and clonal expansion of the MDV-specific repertoire suggests restriction to a small selection of MDV antigens. Indeed, Omar and Schat 
examined the cytolytic response of infected birds against a panel of cell lines expressing individual genes from MDV found that in MHC B19
homozygote Line P2a
birds, the cytolytic activity was restricted to meq, gB and pp38 antigens, while the genetically-resistant line N2a
) birds also detected the ICP4 antigens. In our studies, tumor-infiltrating CD8+ T cells produce greater levels of IFNγ mRNA than CD8+ T cells derived from the spleen of uninfected birds (unpublished data, Mwangi, Peroval et al.,). The CD8+ T cell response of susceptible birds is insufficient to prevent tumor progression; our data provides a framework for comparisons with resistant or vaccinated birds which do not develop tumors. Our sequence analysis clearly detected large CD8+ T cell clones and allowed an approximation of the clone size, the application of higher throughput sequencing technologies may be useful in the future to identify smaller clonal expansions and provide more accurate estimations of clone sizes. Understanding the nature of the TCR repertoire to specific antigens after infection and vaccination can be used to improve vaccine approaches in the future. The rapid nature of focussing within the CD8+ population may reflect a combination of the minimal MHC configuration where each haplotype is dominated by presentation through a single MHC class I gene 
and the minimal TCRVβ locus with 13 Vβ segments in just two families 
The high frequency CD8+ T cell clones were found in both tumor sites and in the spleen of infected individuals, either restricted to one tumor site or present in multiple tumor sites. One of the largest CD8+ clones has a CDR3 sequence (“GSS”) of note, in that identical sequences were detected in different individuals. This type of CDR3 is known as a “public” TCR rearrangement and, although previously reported with mammals, is relatively rare 
. Upon closer examination, it was clear that the public GSS amino acid sequence for the CDR3 also represented shared nucleotide sequence in different individuals. Interestingly, the GSS sequence represents retention of a fragment of the D segment, after deletion of six nucleotides in the D and three nucleotides in the Vβ1 segment. Although not noted previously, it is clear that a CDR3 constructed by deletion (with no retained nucleotide addition) is much more likely to occur in multiple individuals than one generated by addition of nucleotides. We propose that public CDR3 sequences in other contexts (e.g. in humans) may also conform to this arrangement, representing a deletion-based junctional modification. This feature might be useful and exploitable in diverse scenarios to improve “public” responses to vaccines. The remaining CDR3 sequences positively identified as clonal expansions in CD8+ cells (or as not in CD4+ cells) all represented “private” CDR3 identities ().
In this report, we have documented the TCR Vβ repertoire changes associated with infection, tumor development and anti-tumor response that characterise MDV pathogenesis. Upon consideration of our data in the context of previous reports, we propose that the MD tumors are dominated by clonal expansion in an oligoclonal framework of minor clones of pre-cancerous cells. We propose that this type of population structure explains the penetrance and narrow temporal window that characterise MD in susceptible birds. The CDR3 analysis identified that all established MDV-transformed cell lines tested were clonal (with one bi-clonal short term culture), and that these clones represent dominant clones detected in vivo. Within birds harbouring multiple tumors there was a mixture of metastatic and site-specific tumor clones. Overall, we examined 50 tumors derived from 21 individuals, and all tumors were dominated by one or two clones with some birds harbouring a single metastatic tumor clone and others with different clones in different sites. The TCR repertoire analysis system has allowed examination of diverse areas of MD lymphoma biology and the CD8+ response against the infection. We consider that this type of approach can be used to further define MD pathogenesis and the response generated against infection and/or tumors. These types of study also have the potential to impact much more broadly, identifying strategies to vaccinate against or otherwise control viral driven lymphomas in medical and veterinary fields.