Differential effects of Vav‐promoter‐driven overexpression of BCLX and BFL1 on lymphocyte survival and B cell lymphomagenesis

Overexpression of BCLX and BFL1/A1 has been reported in various human malignancies and is associated with poor prognosis and drug resistance, identifying these prosurvival BCL2 family members as putative drug targets. We have generated transgenic mice that express human BFL1 or human BCLX protein throughout the haematopoietic system under the control of the Vav gene promoter. Haematopoiesis is normal in both the Vav‐BFL1 and Vav‐BCLX transgenic (TG) mice and susceptibility to spontaneous haematopoietic malignancies is not increased. Lymphoid cells from Vav‐BCLX TG mice exhibit increased resistance to apoptosis in vitro while most blood cell types form Vav‐BFL1 TG mice were poorly protected. Both transgenes significantly accelerated lymphomagenesis in Eμ‐MYC TG mice and, surprisingly, the Vav‐BFL1 transgene was the more potent. Unexpectedly, expression of transgenic BFL1 RNA and protein is significantly elevated in B lymphoid cells of Vav‐BFL1/Eμ‐MYC double‐transgenic compared to Vav‐BFL1 mice, even during the preleukaemic phase, providing a rationale for the potent synergy. In contrast, Vav‐BCLX expression was not notably different. These mouse models of BFL1 and BCLX overexpression in lymphomas should be useful tools for the testing the efficacy of novel human BFL1‐ and BCLX‐specific inhibitors.


Introduction
The physiological roles of BFL1/A1, an antiapoptotic member of the BCL2 family, are still poorly understood. A1 was discovered in 1993 as an early response gene in GM-CSF-treated bone marrow-derived macrophages [1,2] and later shown to be induced by antigenmediated activation in T and B cells [3,4]. In mice, A1 is produced by three independent genes (Bcl2a1-a, Bcl2a1-b and Bcl2a1-d) [5] and is mainly expressed in the haematopoietic system where it is dynamically regulated in response to antigens or inflammatory cues engaging NF-kB [6], NF-AT [7], and PU.1 [8] transcription factors. Mice that lack all functional Bcl2a1 genes do not exhibit major impairments in the development and composition of their immune system [9] or T cell-mediated immune responses [10]. The human homologue, BFL1, which is highly homologous to A1 (72% amino acid identity) is encoded by a single gene [11]. Elevated BFL1 expression has been associated with many malignancies, including acute lymphoblastic leukaemia, chronic lymphocytic leukaemia and melanoma skin cancer [12,13]. In mouse models, lentiviral transduction of bone marrow cells with Bcl2a1-a led to the development of B cell lymphomas in recipient mice [14] and cotransduction with human BFL1 and c-MYC caused acute myelogenous leukaemia [15]. Importantly, BFL1 mutants that escape ubiquitin-mediated proteasomal degradation are more stable and accelerate tumour formation in the presence of a dominant negative, truncated version of p53 DD [16], indicating the importance of BFL1 expression levels for facilitating tumour formation.
BCLX is essential during early development as Bclxdeficient mice die at embryonic day 13 due to defective erythropoiesis and massive cell death in the central nervous system [17,18]. In the haematopoietic system, BCLX deficiency leads to a loss of pre-B cells [19], impaired erythropoiesis [20] and decreased platelet life span [21]. Interestingly, although highly expressed in CD4 + CD8 + double-positive (DP) thymocytes, Bclx deletion does not substantially influence T cell development but only reduces the life span of DP thymocytes ex vivo [22,23].
The transcription factor c-MYC is a key transcriptional regulator involved in many cellular processes including metabolism, cell cycle and apoptosis [24]. Aberrant expression of MYC is associated with a significant number of human malignancies [25], including human Burkitt's lymphomas, which harbour chromosome translocations linking the MYC gene with Ig heavy (IGH) or Ig light chain (IGL) loci [26]. El-MYC transgenic mice, which model Burkitt's lymphoma to a certain degree, develop monoclonal pro-/pre-B and immature B cell lymphomas between 4 and 7 months of age [27,28]. Of note, BCL2 [29,30], MCL1 [31] or BCLX [32] transgenes significantly accelerate lymphomagenesis in El-MYC mice, indicating the importance of overcoming apoptosis for MYC-driven lymphomagenesis.
Little is known about the lymphomagenic potential of BFL1/A1. Using an shRNA-based model to knock down A1 protein expression in mice, we recently observed that MYC-induced lymphomas select against low A1 levels and that diminished A1 renders premalignant cells more susceptible to apoptosis ex vivo [33]. Studies using mice totally lacking A1 also suggest that A1 contributes to tumour cell survival in the context of MYC overexpression [34]. Moreover, the recent report of a patient with DLBCL [35] having a BFL1/ IgH translocation as well as a MYC/IgL translocation suggests that BFL1 overexpression can act as a second hit in MYC-driven B cell lymphomagenesis.
To investigate the impact of pan-haematopoietic overexpression of BFL1 and BCLX, we have generated Vav-BFL1 TG and Vav-BCLX TG mice. We found that both the Vav-BFL1 and the Vav-BCLX transgenes can accelerate El-MYC-driven lymphomagenesis and observed an unexpected interrelationship between MYC and BFL1 TG expression levels.

Enforced expression of BFL1 or BCLX does not perturb haematopoiesis in mice
The BFL1 TG and BCLX TG mice were generated by pronuclear injection of oocytes using a haematopoietic-specific transgenic vector driven by the Vav gene promoter [36]. For each transgene, independent colonies were established from three PCR-positive founders and the two lines showing detectable exogenous protein expression were chosen for further characterization (Fig. 1A,B), alongside previously derived Vav-Mcl1 TG [31] and Vav-BCL2 TG mice [37]. The Vav-BFL1 TG and Vav-BCLX TG mice were healthy, showed normal fertility and did not exhibit any premature deaths within the first year of age, unlike Vav-Mcl1 or Vav-BCL2 transgenic mice, which develop auto-immune and/or malignant disease [31,37,38].
To assess the impact of transgene expression on the haematopoietic compartment, adult mice were analysed between 8 and 12 weeks of age. First, we monitored the white blood cell (WBC) counts in the peripheral blood (PB). Different to Vav-BCL2 TG mice neither Vav-BCLX TG nor Vav-BFL1 TG mice had significantly increased WBC numbers in the PB (Fig. 1C). Furthermore, neither Vav-BFL1 nor Vav-BCLX TG strains showed aberrant cellularity in bone marrow, thymus or spleen (Fig. 1D, TG lines were pooled to simplify data presentation), while Vav-Mcl1 and Vav-BCL2 TG mice showed splenomegaly (Fig. 1E), as reported before [31,37].
Next, we examined the abundance of different lymphocyte subsets in primary and secondary lymphoid organs. Thymocyte development was normal in Vav-BFL1 and Vav-BCLX TG mice throughout all developmental stages ( Fig. 2A), in contrast to Vav-BCL2 TG mice which had decreased CD4 + CD8 + DP thymocytes and increased CD4 À CD8 À double negative (DN) and CD4 + and CD8 + single-positive (SP) cells, as reported previously [37]. Furthermore, the composition of mature CD4 + and CD8 + T cells in the periphery was similar between all genotypes analysed (data not shown).
Regarding B cell development, Vav-BFL1 and Vav-BCLX TG mice did not show any major abnormalities in the bone marrow and spleen (Fig. 2B,C), unlike Vav-BCL2 TG and Vav-Mcl1 TG mice, which showed changes expected from previous studies [31,37]. One Vav-BFL1 TG line (L3) displayed a trend towards an increase in immature B cells and a reduction in recirculating B cells (Fig. 2B), but this did not reach statistical significance nor was it seen in the second transgenic line. In the spleen, Vav-BFL1 TG mice (L1) showed a tendency towards a loss of Transitional (T) 1 B cells, similar to what could be seen for Vav-BCL2 mice (Fig. 2C), and a mild increase in follicular B cells (Fig. 2C).
We also examined the abundance of Mac-1 + Gr-1 hi granulocytes in the bone marrow and spleen. While the L1 Vav-BFL1 TG line tended to have an increased percentage of granulocytes in the bone marrow, this was not observed in the L3 line. Granulocyte numbers were also normal in Vav-BCLX and Vav-Mcl1 TG mice (Fig. 2D), whereas Vav-BCL2 TG mice had a significantly lower percentage of granulocytes in the bone marrow and spleen, as noted before [37].

Exogenous BFL1 is a weak antagonist of apoptosis
Next, we analysed the ability of the overexpressed proteins to protect cells from spontaneous and druginduced apoptosis. Equal numbers of mice from both Blood lymphocytes    lines of Vav-BFL1 TG and Vav-BCLX TG strains were analysed. No significant differences were detected and the data have therefore been pooled in Fig. 3.

1406
The previously [31,37], Vav-BCL2 and Vav-Mcl1 TG thymocytes were strongly protected from spontaneous apoptosis (Fig. 3A). Thymocytes from both Vav-BCLX TG lines also showed significantly delayed cell death, although not to the same extent, but cells from Vav-BFL1 TG mice remained as sensitive as wild-type controls (Fig. 3A). A similar picture was observed upon treatment with various apoptosis-inducing stimuli, including irradiation, staurosporine (STS) or the glucocorticoid (Glc) corticosterone. Next we analysed pre-B cells isolated by flow cytometry from bone marrow. Interestingly, Vav-BFL1 and Vav-BCLX TG pre-B cells both showed significantly delayed spontaneous apoptosis when cultured in vitro However, although the Vav-BCLX TG also protected pre-B cells from STS and Glc-induced apoptosis, Vav-BFL1 expression did not (Fig. 3C, bar graph). Neither the Vav-BFL1 nor the Vav-BCLX TG significantly enhanced the overall viability of bone marrow granulocytes in vitro, while Vav-Mcl1 and especially Vav-BCL2 TG expression did so effectively (Fig. 3D). Strikingly, none of the prosurvival proteins was able to protect granulocytes from STS-induced cell death at the time-point chosen for analysis ( Fig. 3D bar graph).
Lastly, we treated thymocytes with ABT-737, a BH3 mimetic that induces apoptosis by inhibiting BCL2, BCLX and BCLW. Confirming the functionality of the transgene, Vav-BFL1 TG thymocytes were significantly protected from ABT-737-induced apoptosis, although the protection was not as pronounced as that seen in Vav-Mcl1 TG thymocytes (Fig. 3E). As expected, thymocytes from Vav-BCLX and Vav-BCL2 mice were highly sensitive to ABT-737.
Quantitative rather than qualitative differences in between different BCL2-prosurvival proteins define the degree of protection from thymocyte apoptosis We wished to clarify whether the degree of protection correlated with the level of transgene expression rather than with qualitative functional differences between the prosurvival proteins. To assist in the quantitation, we prepared transiently transfected 293T cells expressing cDNAs encoding for an HA and streptavidin (HS)tagged version of BCL2, BCLX, Mcl1 and BFL1 and loaded different amounts of these proteins next to 20 lg total protein lysates from the TG thymocytes (Fig. 4). Since the quantity of the HS-tagged proteins were within a comparable range within the lysates from transfected 293T cells, as shown by the HA western blot, we were able to better judge the relative TG expression found in thymocytes using target-specific antibodies. This comparison made it evident that the Vav-BCL2 TG was expressed at much higher levels in the thymus when compared to the highest concentration of 293T lysate loaded. In contrast, the Vav-BCLX and Vav-Mcl1 TG thymocyte extracts showed comparable signals to those seen in the highest concentration of 293T cell lysate loaded. This suggests that the relative expression levels achieved in Vav-BCLX and Vav-Mcl1 TG thymocytes were comparable to each other but much lower than those of BCL2 in Vav-BCL2 TG thymocytes. Strikingly, the BFL1 signal from the thymocyte lysates of Vav-BFL1 TG mice was lower than the lowest signal generated from the HS-BFL1 dilution series in 293T cell extracts. We conclude that thymocytes from Vav-BCL2 TG mice express the highest    Tables S1-S3) is shown on the right. IgM À refers to CD19 + B220 + IgM À lymphoma cells, IgM + refers to CD19 + B220 + IgM + lymphoma cells, mixed refers to lymphomas containing both, CD19 + B220 + IgM À and CD19 + B220 + IgM + tumour cell types (F) Representative dot-plots of stem/progenitor cell lymphomas in the thymus. Other than B220 + CD4 À lymphoma cells B220 + CD4 + lymphoma cells do not express CD19. Bar graph: Quantification of CD19 À B220 + CD4 + stem/progenitor cell lymphomas in the thymus. Kaplan-Meier plot: Survival curve of El-MYC/Vav-BFL1 DT mice with or without B220 + CD4 + stem/progenitor cell lymphomas in their thymus. Statistical analysis for tumour-free survival was performed by using a log-rank (Mantel-Cox) test. All other statistical analyses were performed by using a one-way ANOVA with Holm-Sidak's multiple comparison test compared to El-MYC TG. *P < 0.05; **P < 0.01; ****P < 0.0001; n ≥ 9 AE SD.

1409
The amounts of transgenic protein, those from Vav-BCLX and Vav-Mcl1 TG express lower levels, albeit comparable to each other, and those from Vav-BFL1 TG mice have the lowest expression. These results partly explain the differences observed in the relative resistance conveyed by the different transgenes to apoptosis-inducing stimuli in culture.

BFL1-MYC crosstalk accelerates B-cell lymphomagenesis in mice
Since the Vav-BCL2 and Vav-Mcl1 transgenes accelerated El-MYC-driven lymphomagenesis [30,31], we aimed to test if this was also the case for Vav-BCLX TG and Vav-BFL1 TG mice. Therefore, Vav-BCLX TG mice (line A) and Vav-BFL1 TG lines 1 and 3 mice (data pooled) were intercrossed with El-MYC TG mice and monitored for acute signs of lymphomagenesis such as enlarged lymph nodes and shortness of breath. Of note, the El-MYC/Vav-BFL1 double-transgenic (DT) mice succumbed significantly faster to malignancy (median survival 53 days) than El-MYC/Vav-BCLX DT mice (median survival 67 days) and El-MYC TG mice (median survival 139 days) (Fig. 5A). This result was unexpected since we had observed comparable protective capacity of the Vav-BFL1 and Vav-BCLX transgenes in cultured pre-B cells (Fig. 3C) and significantly better protection by the Vav-BCLX transgene against glucocorticoid treatment. Sick El-MYC/Vav-BFL1 and El-MYC/Vav-BCLX DT mice had similar WBC counts and these were in both cases significantly higher than that for sick El-MYC TG mice (Fig. 5B). Spleen weights were comparable between all three genotypes (Fig. 5C) while thymus weights were significantly higher in both El-MYC/Vav-BFL1 and El-MYC/Vav-BCLX DT mice than in El-MYC TG mice (Fig. 5D). Next, we analysed the phenotype of the tumours by flow cytometric analysis (Fig 5E, Tables S1-S3). The El-MYC mice developed mainly immature B220 + CD19 + IgM À tumours (Table S1). Most El-MYC/Vav-BFL1 mice also developed pre-B lymphomas (Fig. 5E, Table S2) but a significant number (7 of 11 mice analysed) additionally developed CD19 À B220 + CD4 + progenitor cell lymphomas that have been described before for El-MYC/El-BCL2 mice [29]. In the El-MYC/Vav-BFL1 mice, these progenitor tumours were largely restricted to the thymus (Fig. 5F, Table S2). However, the onset of disease was not significantly different between El-MYC/Vav-BFL1 mice that harboured CD19 À B220 + CD4 + progenitor cell lymphomas in their thymus and those that did not (Fig. 5F, Kaplan-Meier plot).
Interestingly, El-MYC/Vav-BCLX DT tumours appeared more variable by showing IgM À , IgM + and mixed phenotypes consisting of IgM + and IgM À lymphoma cells respectively (Fig. 5E and Table S3). However, the different tumour types did not differ in the onset of disease (not shown). Next, we compared the premalignant phenotypes of El-MYC/Vav-BFL1 and El-MYC/Vav-BCLX DT mice. As El-MYC/Vav-BFL1 DT mice can succumb to tumours as early as 29 days, we analysed the mice at 2 weeks of age in order to avoid transformed cells. Importantly, we analysed both El-MYC/Vav-BFL1 lines to exclude potential side-effects caused by random transgene insertion. First, we monitored white blood cell counts. Pups from both El-MYC/Vav-BFL1 DT lines had five times higher WBC counts than El-MYC TG pups at that age (Fig. 6A). Interestingly, WBC counts from El-MYC/Vav-BCLX DT pups although significantly higher than those of El-MYC TG mice were three times lower than in El-MYC/Vav-BFL1 DT pups (Fig. 6A). Of note, the elevated WBC counts represent an early burst of pre-B cells, caused by MYC overexpression [39], and the WBC counts subsided by 4 weeks of age (data not shown). Cell counts from the bone marrow and spleen of 2-weekold mice were comparable between all the genotypes (data not shown). However, the percentage of B220 + B lymphoid cells in the bone marrow was significantly higher in El-MYC/Vav-BFL1 DT mice than in El-MYC TG and El-MYC/Vav-BCLX DT mice (Fig. 6B left graph). Most B220 + cells were immature CD19 + IgM À cells (data not shown), but a small proportion also represented CD19 À CD4 + cells and was significantly higher in El-MYC/Vav-BFL1 DT mice compared to the other genotypes ( Fig. 6B right graph). However, this population was present in comparably low numbers (0.1% of all B220 + cells) in the blood (data not shown) and the spleen (Fig. 6D right graph) of all tested genotypes. Importantly, B220 + B lymphoid cells from El-MYC/Vav-BFL1 and El-MYC/ Vav-BCLX DT bone marrows survived better in culture than those from El-MYC mice (Fig. 6C). B220 + B lymphoid cells were also increased proportionally in the spleen of both DT genotypes compared to El-MYC TG mice, although this reached significance only for El-MYC/Vav-BFL1 DT mice (Fig. 6D left  graph). Interestingly, El-MYC/Vav-BFL1 DT mice had significantly more B220 + CD19 + IgM À B lymphoid cells in the spleen than El-MYC TG mice, and El-MYC/Vav-BCLX DT mice showed significantly less (Fig. 6D middle graph) but no increase in B220 + CD19 À CD4 + cells (Fig. 6D right graph). . B220 + cells were further discriminated into CD19 À CD4 + cells (right bar). (C) Total bone marrow was cultured for 30 h and the abundance of living (Annexin V À ) B220 + IgM À immature B lymphoid cells was assessed by flow cytometry. (D) Abundance of total B220 + B lymphoid cells in the spleen was analysed by flow cytometry (left graph). B220 + cells were further discriminated into CD19 + IgM À (middle graph) and CD19 À CD4 + cells (right graph). (E) Total splenocytes were cultured for 30 h and abundance of living (Annexin V À ) B220 + IgM À immature B lymphoid cells was assessed by flow cytometry. Statistical analysis was performed by using a one-way ANOVA with Dunnett's multiple comparison test compared to El-MYC control. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; n ≥ 3 AE SD.

1411
The lymphoid cells showed increased survival in culture (Fig. 6E).

MYC overexpression increases transgenic BFL1 RNA and protein levels
We were surprised by the strong acceleration of disease in El-MYC/Vav-BFL1 DT mice and the enhanced survival capacity of premalignant El-MYC/Vav-BFL1 immature B cells as we had not observed major impact from expression of Vav-BFL1 alone. We, therefore, wondered whether Vav-BFL1 transgene expression had changed in the El-MYC TG background. Indeed, western blots revealed a striking difference in BFL1 levels in total splenocytes isolated from 2-week-old El-MYC/ Vav-BFL1 DT mice compared to those from agematched Vav-BFL1 TG mice (Fig. 7A). Importantly, both transgenic lines showed highly enhanced BFL1 expression on the El-MYC TG background. This phenomenon did not appear to be a global event since other BCL2 family members were not affected. Furthermore, we showed that the enhanced transgene expression was not Vav promoter dependent since the endogenous VAV protein was not increased by MYC overexpression, but rather mildly reduced. Importantly, the increase in BFL1 expression by MYC overexpression happened to the same extent in both independently generated Vav-BFL1 TG lines, minimizing potential transgene insertion effects. Interestingly, BFL1 mRNA was found increased by approximately ninefold in Vav-BFL1/El-MYC DT splenocytes when compared to mRNA levels found in Vav-BFL1 TG mice (Fig. 7A, bar graph). Endogenous A1 expression levels were not found substantially upregulated in total splenocytes by MYC overexpression, neither on mRNA nor on protein levels ( Fig. 7B and data not shown). Together these findings argue for enhanced BFL1 mRNA stability or reduced protein turnover. MYC protein expression was also slightly higher in El-MYC/Vav-BFL1 DT splenocytes than in El-MYC TG samples (Fig. 7A) although MYC mRNA levels were not significantly elevated (not shown), indicative of a feed-forward loop where increased cell death resistance allows cells to tolerate increased MYC protein levels. In tumour samples, we found that BFL1 expression was further elevated when compared to splenocytes from 2-week-old premalignant El-MYC/Vav-BFL1 DT mice (Fig. 7B). In order to determine BFL1 expression levels in the different tumour cell subsets, we FACS-sorted B220 + CD19 + tumour cells and B220 À CD19 À nontumour cells from the spleen and B220 + CD19 + and B220 + CD19 À CD4 + tumour cells from the thymus of diseased El-MYC/Vav-BFL1 DT mice. BFL1 protein levels were only detectable in B lymphoid tumour cells, while they were absent in non-B220 + cells (Fig. 7C). Furthermore, we could not detect any quantitative differences in the BFL1 expression between B220 + CD19 + and B220 + CD19 À CD4 + tumour populations. Comparable increases in transgenic BCLX protein expression were not observed in splenocytes from premalignant El-MYC/Vav-BCLX DT vs. premalignant Vav-BCLX TG mice, although levels increased mildly in El-MYC/Vav-BCLX tumours (Fig. 7D). Furthermore, BCLX TG mRNA was not influenced by exogenous MYC expression in 2-week-old spleen extracts (Fig. 7D, bar graph). We conclude that El-MYC overexpression positively influences expression of transgenic Vav-BFL1, at both the protein and RNA level, but not the Vav-BCLX transgene.

Discussion
The physiological importance of BFL1/A1 is still poorly understood, despite increased recent efforts. In a broad variety of immune cells, including neutrophils, granulocytes, B and T cells, A1 is rapidly inducible in response to diverse stimuli, including antigen receptor stimulation, GM-CSF, BAFF receptor or CD40-ligation [40,41], predictive of crucial roles in inflammation and immunity. Nevertheless, complete deficiency of A1 does not impair the normal development or function of the immune system nor does it influence the normal behaviour and life span of mice [9,10].
Here, we describe the generation and characterization of mice that overexpress human BFL1 or human BCLX under the control of a haematopoietic-specific vector driven by the Vav gene promoter [42]. In contrast to Vav-BCL2 [37] and Vav-Mcl1 [31] TG mice, the overall composition and cellularity of all major lymphoid organs was essentially normal in both Vav-BCLX and Vav-BFL1 mice (Figs 1 and 2). Interestingly, while the Vav-BCLX TG partially protected both thymocytes and pre-B cells from spontaneous and drug-induced apoptosis, the Vav-BFL1 TG provided only a minor survival benefit in pre-B cells or when thymocytes were treated with ABT-737 (Fig. 3). In general, the graded survival benefits observed in the different mouse models correlated directly with the level of transgenic protein expressed (Figs 3 and 4). It remains unclear why both Vav-BFL1 TG lines express such low amounts of BFL1, especially when compared to the two Vav-BCLX TG lines that were generated contemporaneously, but this likely reflects the short half-life of BFL1 [43,44].
Surprisingly, when we crossed Vav-BFL1 TG with El-MYC TG mice, expression of the Vav-BFL1 TG was boosted (Fig. 7) and lymphomagenesis was dramatically accelerated (Fig. 5). Indeed, the median survival of El-MYC/Vav-BFL1 DT lymphomas was only 53 days, comparable to that observed for El-MYC mice lacking the BH3-only protein BIM [45,46]. While the Vav-BCLX transgene also accelerated lymphomagenesis in El-MYC mice (median survival 67 days), previously described El-Bclx/El-MYC DT mice died at an age of only 6 weeks (42 days) [32]. The difference is likely to reflect the relative level of BCLX protein achieved.
Intriguing differences were observed in the tumour phenotypes. C57BL/6 El-MYC TG mice can develop either pro-/pre-B or B lymphomas [34,45], with the former dominating in our colony. Intriguingly, our El-MYC/Vav-BCLX DT mice developed mainly IgM + B cell lymphomas, like El-MYC mice lacking BIM [45] or BMF [46]. It has been reported that El-MYC-driven IgM À immature B lymphomas are more aggressive and develop faster than their IgM + counterparts [30]. This observation would be consistent with the prolonged tumour-latency observed for El-MYC/Vav-BCLX DT mice compared to El-MYC/ Vav-BFL1 DT mice that only develop IgM À tumours. However, within the group of El-MYC/Vav-BCLX DT mice, no differences in latency were observed between the different immunophenotypes (not shown).
Lastly, we are intrigued by the elevated expression of BFL1 protein and mRNA in premalignant El-MYC/Vav-BFL1 TG splenocytes compared to Vav-BFL1 TG samples (Fig. 7A,B). Intriguingly, BFL1 was upregulated to the same extent in both El-MYC/ Vav-BFL1 DT lines (Fig. 7A) and was already apparent at 2 weeks of age, arguing against clonal expansion of cells with stronger Vav-BFL1 expression during transformation. El-MYC/Vav-BCLX DT cells did not show elevated transgene expression compared to Vav-BCLX TG littermates (Fig. 7D). Since MYC influences global gene expression [47], the elevation of transgenic BFL1 expression might reflect integration of the transgene within MYC accessible sites, in both independently generated Vav-BFL1 TG lines. It might also be possible that c-MYC stabilizes the transgenic BFL1 mRNA in the context of an artificial 3 0 UTR as endogenous A1 was not found elevated (Fig. 7B). The molecular mechanism responsible for this intriguing phenomenon remains to be investigated further.
Together, our findings underline the major impact of elevated BFL1 on tumour development, an effect that might not be confined to MYC-induced lymphomas. They also emphasize the potential usefulness of the development of BFL1-specific inhibitors for cancer treatment and the mouse model described here might be perfectly suited for their preclinical testing.

Haematopoietic cell analysis and flow cytometry
Peripheral blood was analysed with a scilVet abc blood counter (Viernheim, Germany) or by flow cytometric analysis after red blood cell lysis using 0.168 M ammonium chloride in PBS. Single-cell suspensions were prepared from thymus, lymph nodes (axillary, brachial, inguinal and mesenteric), bone marrow (from femurs) and spleen, and viable cells were counted using a Neubauer counting chamber by trypan blue exclusion. Cell composition was determined by staining with cell surface marker-specific antibodies and flow cytometric analysis using a LSR mRNA isolation and quantitative real-time PCR analysis RNA was isolated using TRIzol (ThermoFisher Scientific, Inc.) and quantified with a NanoDrop 1000 Spectrophotometer (ThermoFisher Scientific, Inc.). Two hundred nanograms of RNA was reverse transcribed into cDNA using iScript cDNA Synthesis Kit (Bio-Rad) according to the manufacturer's instruction. Quantitative real-time PCR was performed with a 100th part of the cDNA in a StepO-nePlus System (ThermoFisher Scientific, Inc.) using Bimake SYBR Green (Bimake, Houston, TX, USA) according to the manufacturer's instructions and 100 nM of the following primers: Flag-BFL1 forward primer ACA AAG ACG ATG ACG ATA AAA CAG A and reverse primer AGC ACT CTG GAC GTT TTG CT; Flag-BCLX forward primer: CAA AGA CGA TGA CGA TAA AGG ATC T and reverse primer TCC AGC TGT ATC CTT TCT GGG A; Actin-beta forward primer: ACT GGG ACG ACA TGG AGA AG and reverse primer GGG GTG TTG AAG GTC TCA AA. PCR conditions were 95°C for 10 min, 40 cycles of (95°C for 15 s and 60°C for 60 s), 95°C for 15 s, 60°C for 60 s followed by a melting curve with 0.3°C increment steps up to 95°C for 15 s. Results were normalized to Actin-beta expression to be compared with the DDC t relative quantification method.

Statistical analysis
Statistical analysis was performed using GRAPHPAD PRISM Version 7.03. for Windows, GraphPad Software, La Jolla, CA, USA, www.graphpad.com. Used tests are indicated in the figure legends.
Fava for helpful discussion. We also thank M. Herold (WEHI) for the anti-A1 antibody and J. Borst (NKI) for the anti-BFL1 antiserum. This work was supported by grants from the Austrian Science Fund (FWF), Grant I 3271 (FOR-2036), the MCBO Doctoral College 'Molecular Cell Biology and Oncology' (W1101) and the ' € Osterreichische Krebshilfe Tirol'. MDH and ST are supported by a DOC-fellowship from the Austrian Academy of Science ( € OAW).

Supporting information
Additional Supporting Information may be found online in the supporting information tab for this article: Table S1. Phenotypic characterization of lymphomas arising in El-MYC TG mice. Table S2. Phenotypic characterization of lymphomas arising in El-MYC/Vav-BFL1 DT mice. Table S3. Phenotypic characterization of lymphomas arising in El-MYC/Vav-BCLX DT mice.