Apparent bypass of negative selection in CD8+ tumours in CD2-myc transgenic mice.

A role for antigen stimulation in lymphoid neoplasia has been postulated and is supported by indirect evidence that suggests that the interaction of antigen with both T cells and B cells may constitute an epigenetic event that can contribute to tumour induction or tumour progression. Using myc-bearing transgenic mice that develop mainly clonal T-cell lymphomas we have investigated the possibility that endogenous antigen-mediated clonal deletion might be overridden in tumorigenesis. CD2-myc transgenic mice were backcrossed on to a CBA/Ca background to ensure Mtv-mediated deletion of V beta 11-expressing T cells in the resultant offspring. Lymphomas arising from these mice were subsequently screened for V beta 11 expression. There was a clear correlation between the age at which mice developed neoplasia and the tumour phenotype. Mice with CD4- CD8+ tumours succumbed to thymic lymphoma at a significantly younger age than mice developing CD4+ CD8+ tumours. A small number of tumours consisted of the 'forbidden' V beta 11 phenotype, showing that cells vulnerable to transformation could escape negative selection. The majority of the V beta 11-positive tumours were CD4- CD8+ and were only observed in mice showing clinical evidence of tumour development at a relatively young age. The phenotype of these cells and the age at which tumours arose suggests that T cells escaping tolerance may be susceptible to transformation. ImagesFigure 2

There is now considerable evidence to suggest that the oncogenic transformation of healthy tissue involves a series of events, ultimately resulting in the appearance of the fully malignant phenotype. To study this process in the T-cell lineage we have produced two lines (800 and 900 series) of transgenic mice that harbour the human c-myc gene linked to the CD2 dominant control region. These CD2-myc transgenic mice display a moderate incidence of T-cell lymphoma development but it is obvious from the stochastic appearance and the clonal nature of these tumours that one or more secondary events are necessary for transformation to the fully neoplastic phenotype (Stewart et al., 1993). These mice are, therefore, a useful model with which to explore those oncogenic events that can cooperate with myc in T-cell leukaemogenesis. In this study, we investigated the possibility that antigenic stimulation of T cells through the T-cell receptor (TCR) might, under certain circumstances, complement the action of oncogenes such as myc in T-cell transformation.
The potential role of antigen stimulation in the transformation or clonal expansion of lymphoid tissue is not known but there is indirect evidence to implicate antigenic activation of lymphocytes in tumorigenesis. Follicular lymphomas are often committed to the production of autoantibodies and it has been suggested that B cells responsive to the continuous presence of self antigen are susceptible to transformation (Dighiero et al., 1991). Further, Bahler and Levy (1992) have reported that there is evidence for antigen selection in the development of follicular lymphomas. Analysis of a thymic lymphoma associated with a field case of feline leukaemia virus (FeLV) infection revealed the presence of a provirus which had incorporated a fully processed TCR gene, raising the possibility that antigen stimulation or inappropriate antigen recognition may have been important in the genesis of this tumour (Fulton et al., 1987). Studies involving murine radiation leukaemia virus (RadLV) have shown that free virus particles can modulate the growth of a RadLV-induced T-cell lymphoma cell line (O'Neill, 1994) and that the VP repertoire of RadLV-induced lymphomas is restricted, indicating non-random selection (Sen-Majumdar et al., 1994).
The inbred mouse is a useful model with which to study endogenous antigen stimulation. Proviral mammary tumour virus (Mtv) sequences resident in the mouse genome comprise a polymorphic family of superantigens that react with different T-cell families defined by their VP gene usage and have, therefore, an important role in shaping the T-cell repertoire. Expression of the sag gene located in the 3' long terminal repeat (LTR) of endogenous mammary tumour proviral sequences results in the stimulation of mature T cells and in the clonal deletion of developing T cells bearing specific VP gene products (for reviews see Acha-Orbea & Palmer, 1991;Simpson,-1992;Marrack et al., 1993;Simpson et al., 1993).
We have exploited the presence of these endogenous superantigens to investigate the relationship between TCR VP restriction and lymphomagenesis using the tumour-prone CD2-myc transgenic mice. The appearance of tumours expressing a 'forbidden' VP phenotype would indicate that tumorigenesis bypasses or overrides negative selection. Here we present data to demonstrate that a small proportion of CD4-CD8+ T-cell tumours can express a 'forbidden' Vpll phenotype.

Experimental animals
The CD2-myc transgenic mice have been described previously (Stewart et al., 1993). The genetic background of these mice was a mixture of C57B16/J and CBA/Ca inbred lines. The transgenic mice were backcrossed to CBA/Ca mice and tumours arising in the offspring were analysed. This was carried out to ensure that the offspring all carried a functional I-Eo gene necessary for the deletion of autoreactive V,B T-cell families. Mice developing lymphomas were culled as soon as clinical signs of malaise were apparent.
Hybridisation of DNA and RNA Preparation of high-molecular weight DNA from mouse tissues was carried out using standard protocols (Hogan et al., 1986). DNA was digested with the appropriate restriction endonucleases, separated by agarose gel electrophoresis, Correspondence: ER Cameron, Received 6 April 1995; revised 17 July 1995; accepted 2 August 1995 transferred to nylon filters and hybridised to the appropriate radiolabelled probes at high stringency using the protocols described by Sambrook et al. (1989). Random priming (Random Priming Kit, Amersham) and radiolabelling of these probes was also carried out using methods described by Sambrook et al. (1989). Transgene sequences were identified using a human c-myc exon 3 probe (1.38 kb, CJaI-EcoRI). Cellular RNA was extracted from mouse tissues using the RNAzol B method (Biogenesis). Samples of 20 yig were separated by electrophoresis on 1% agarose gels containing 2.2 M formaldehyde, transferred on to nylon filters and hybridised using procedures described by Sambrook et al. (1989). VP1 1 expression was determined using a probe derived from a 0.4 Kb EcoRI/Sst fragment of the VP I1 variable region (kindly supplied by Julian Dyson, MRC Clinical Research Centre, Harrow, UK).

Flow cytometry
Thymus tissue was mixed with phosphate-buffered saline and minced using sterile scissors and the dead cells removed by centrifugation through a Ficoll-hypaque gradient (2000g for 10 min). Cells were directly labelled by resuspending them at 2.5 x 106 ml with a 1:40 dilution of pretitrated antibody.

Results
Small numbers of CD2-myc/CBA tumours expressed a forbidden VP3JJ phenotype CD2-myc transgenic mice harbour the human c-myc gene linked to the CD2 dominant control region and as a result these mice display a moderate incidence of spontaneous Tcell lymphoma. We sought to use these mice to investigate the influence of T-cell negative selection on tumour development. The presence of endogenous superantigens in mice results in the clonal deletion of specific VP expressing families. We wished to establish the CD2-myc mice (originally derived from C57B1/6 and CBA/Ca inbred strains) on a background that should result in the deletion of VPJl 1 cells. CBA/Ca mice carry Mtv loci 8 and 9 together with a functional I-E gene (Simpson, 1992), by backcrossing onto CBA/Ca mice it was predicted that the offspring would delete Vp11-expressing cells. To confirm that these F, mice could present endogenous superantigen a representative sample of the tumours was screened for the presence of an intact I-Ea gene (J Picard, personal communcation). All CBA/CD2myc backcrossed mice examined were found to have at least one allele with an intact I-E a-chain gene. The potential role of autoreactive cells in lymphoma development was investigated by screening tumours for VP11 1 expression.
The phenotype of tumours arising from CBA backcrossed transgenic mice was investigated using flow cytometry. All tumours analysed were CD3+ and the majority of tumours could be divided into two T-cell phenotypes. Nearly 70% of tumours were composed of cells that were predominantly CD4+ CD8+ and 20% were predominantly composed of cells of the CD4-CD8+ phenotype (Table I). These results are similar to those previously observed in the parental CD2-myc mice (Stewart et al., 1993). Backcrossing the mice did, how-ever, result in an increased incidence of tumour development in both lines. In the 800 line and 900 lines the incidence rose from 19% (22/114) to 30% (102/340) and from 3% (2/66) to 34% (25/73) respectively.
Flow cytometry analysis of VP11 usage in tumour samples of CD2-myc/CBA mice showed that 3 of the 17 CD4-CD8+ tumours analysed were largely composed of this 'forbidden' cell type and a fourth CD4-CD8+ tumour contained a major subpopulation of VP11 1-positive cells. In contast VP11 cells were only found in 2 of the 61 CD4+ CD8+ tumours.
To confirm that VPII expression was present in these samples we analysed two of these tumours (8V9 and 9V10) for V13ll mRNA by Northern blot hybridisation using a V31llspecific probe, these results confirmed that both tumours expressed the VP B11 transcript. The flow cytometry results and Northern blot results of these two tumours are shown in Figures 1 and 2 respectively. Therefore it appears that tumours representing T-cell families that would normally be deleted can be present in transgenic mice prone to thymic lymphoma development but their appearance is biased to the CD8 single positive subset.
Tumour phenotype in CD2-myc mice was highly correlated with age Transgenic CBA backcrossed mice developed spontaneous thymic lymphomas from 2 months of age. The majority of mice that developed tumours did so between 3 and 9 months of age, with only a small proportion of mice greater than 9 months of age succumbing to lymphoma development. The CD4/CD8 phenotype of the T-cell tumours was closely correlated with the age at which mice develop thymic lymphoma. Mice developing thymic lymphoma at a comparatively young age were much more likely to be CD4-CD8+ than mice developing lymphoma at an older age. The average age at which mice developed CD4-CD8+ tumours was 115 days whereas the average age of mice developing CD4+ CD8+ tumours was 174 days. The difference between the two groups is highly significant (Mann-Whitney test, P<0.001) and suggests that there is an age-related susceptibility to the development of CD4-CD8+ tumours. The over-representation of CD4-CD8+ tumours in young mice is shown in Figure 3.
The appearance of CD4-CD8+ Vpll tumours was also correlated with age. The mice that developed VP1 1-expressing tumours of the CD4-CD8+ phenotype were amongst the youngest mice to succumb to tumour development. Overall the average age at which mice develop thymic lymphoma is 161 days with a wide range from 65 to 333 days. The mice developing CD4-CD8+ V13l l-positive tumours all died between 65 and 94 days of age. The two mice that developed CD4+ CD8+ VP11 1-positive tumours died at 109 and 126 days.
In all of the tumours examined the thymus was greatly enlarged but gross pathological changes in the spleen and lymph nodes were much more variable. As a result the tumours could be broadly divided into two types: those that had only slight or no obvious splenic involvement and those that had gross involvement of the secondary lymphoid organs. There was a correlation between the phenotype of the tumours and the observed gross pathology. In those tumours in which gross involvement of secondary lymphoid organs was not obvious, 90% of tumours were CD4+ CD8+ and only 10% of tumours were CD4-CD8+ whereas there was a far higher representation of CD4-CD8+ tumours showing gross involvement of the secondary lymphoid organs. These results are shown in Table II.

Discussion
The multistep model of tumour induction is well established and it has been suggested that as many as 3-6 'hits' may be involved in leukaemogenesis (Berns, 1993). In this study we have investigated the possibility that antigen stimulation could constitute an epigenetic event with a contributory role in tumorigenesis. In the CD2-myc mouse model the myc transgene predisposes to tumour development, although these tumours arise some time after birth and are clonal or oligoclonal in origin. It is conceivable that developmentally acquired epigenetic factors such as the acquisition of selfreactivity during T-cell receptor rearrangement could accelerate tumorigenesis. In this context it is of interest that the tumours bearing potentially autoreactive cells were amongst the earliest to develop and that in general CD8 single positive tumours were more common in mice succumbing to T-cell lymphoma at an early age. Three out of 17 CD4-CD8+ tumours displayed a 'forbidden' VPil phenotype and a fourth tumour from this series had a large VPII subpopulation. Two alternative hypothesis are consistent with the appearance of these tumours and our understanding of T-cell ontogeny. First, the development of such tumours may be indicative of abortive negative selec-tion. The interaction of the T-cell receptor (TCR) with selfantigen normally results in the induction of apoptosis. If, however, the cells had sustained oncogenic changes before negative selection they might fail to undergo programmed cell death, particularly if some of these lesions inhibited apoptosis. Such an event would result in the survival of transformed cells carrying a potentially autoreactive VP gene product.
An alternative possibility is that tumours arise from that small population of autoreactive cells that appear to escape clonal deletion in the course of normal development. It is well established that clonal deletion of T-cells expressing potentially autoreactive VP genes is incomplete in neonatal mice (Smith et al., 1989;Schneider et al., 1989;Jones et al., 1990). Although a number of studies have indicated that autoreactive T-cells escaping thymic deletion are rendered anergic to a variety of stimuli (Ramsdell et al., 1989;Rammensee et al., 1989;Blackman et al., 1990;Smith et al., 1989). (Jones et al., 1990) have shown that a small population of mature (mainly CD4-CD8+) V,B-reactive cells are present in the thymus of neonatal mice and that these cells are functional as defined by their capacity to proliferate and generate interleukin 2 (IL-2). The transient presence of such cells suggests that thymocytes escaping clonal deletion are immediately rendered anergic and that for a period of time there exists a population of cells that are reactive to selfantigens. There are interesting parallels between these findings and the observations regarding tumorigenesis in the CD2-myc mice. Mice which develop tumours at a young age tend to develop CD4-CD8+ tumours, some of which express a 'forbidden' VP phenotype. It is possible, therefore, that these tumours (or at least a proportion of them) arise from this relatively transient population of mature CD4-CD8+ T cells that escape clonal deletion. The tumours might either arise from cells that have not yet been rendered anergic or from cells in which anergy has been reversed; Andreau-Sanchez et al. (1991) have shown that the anergic state of self-reactive T cells can be reversed.
Whether the tumours arise from cells before selection or from a transient population of normal cells that have escaped clonal deletion, the continued presence of endogenous superantigen makes stimulation of these cells a possibility as it is clear that single positive mature CD8 cells can be stimulated by endogenous superantigen in combination with class II MHC (Webb and Sprent, 1990).
An in vitro study has previously shown an oncogenic relationship between c-myc and antigen stimulation in a hybrid cell line (Kubota et al., 1992). A lymphoma line in which c-myc is deregulated loses its transformed phenotype and down-regulates c-myc when fused to an Mls-1 a antigen-ig f4 imontha) Figure 3 The relationship between tumour phenotype (based on surface expression of CD4 and CD8) and the age at which mice develop lymphomas. _, CD4-CD8+; M1, CD4+ CD8+.
specific non-tumorigenic cell line. However, repeated antigen stimulation can result in autonomous growth and can reverse the down-regulation of c-myc. In vivo studies involving retrovirus-induced lymphomagenesis have also suggested a role for antigen stimulation in tumour initiation and/or progression. De Heer et al. (1992) reported that 2 out of 14 retrovirus induced lymphomas arising in AKR mice expressed a 'forbidden' VP phenotype. A restricted VP phenotype has also been observed in T-cell lymphomas induced with RadLV, although the appearance of a 'forbidden' phenotype was not reported (Sen-Majumdar et al., 1994). However, there is an important difference between these studies and the results reported here. The mechanism of viral-induced lymphomagenesis is complex and it has been suggested that RadLV may exert its oncogenic action at more than one stage in the transformation process (O'Neill, 1994).
Overall these findings show that cells vulnerable to transformation can, under certain circumstances, escape the mechanisms that maintain tolerance. It is possible that double positive cells expressing a forbidden VP phenotype represent cells that have been transformed before selection. However, the identification of apparently mature T-cell tumours bearing this phenotype was more surprising (particularly as this was not an infrequent finding) and raises the possibility that antigen stimulation may play a role in tumour induction or in the clonal expansion of tumours. In this study we have only sampled a small part of the autoreactive repertoire but these findings may represent a more generic phenomenon. We intend to address this issue both by examining the tumour phenotype from mice that delete a wider range of VP families and by using exogenous superantigen.