FoxO3 suppresses Myc-driven lymphomagenesis

This study demonstrates, for the first time, that loss of a single forkhead box class O (FoxO) transcription factor, can promote lymphomagenesis. Using two different mouse models, we show that FoxO3 has a significant tumour-suppressor function in the context of Myc-driven lymphomagenesis. Loss of FoxO3 significantly accelerated myeloid tumorigenesis in vavP-MYC10 transgenic mice and B lymphomagenesis in Eμ-myc transgenic mice. Tumour analysis indicated that the selective pressure for mutation of the p53 pathway during Eμ-myc lymphomagenesis was not altered. Frank tumours were preceded by elevated macrophage numbers in FoxO3−/− vavP-MYC10 mice but, surprisingly, pre-B-cell numbers were relatively normal in healthy young FoxO3−/−Eμ-myc mice. In vitro assays revealed enhanced survival capacity of Myc-driven cells lacking FoxO3, but no change in cell cycling was detected. The loss of FoxO3 may also be affecting other tumour-suppressive functions for which FoxO1/4 cannot fully compensate.

The evolutionarily conserved forkhead box class O (FoxO) transcription factors have important roles in cellular metabolism, stress tolerance and probably lifespan. 1,2 They are key regulators of cell proliferation and survival, a large number of their transcriptional targets being involved in apoptosis, [3][4][5] cell cycle arrest, 6,7 DNA repair, oxidative stress resistance and metabolic processes. 8,9 Regulation of FoxOs is complex, involving acetylation and ubiquitination as well as phosphorylation (reviewed in Eijkelenboom and Burgering 10 ). Under normal conditions of growth factor signalling, FoxO transcription factors are inactivated through phosphorylation by activated AKT. For FoxO3 this occurs at three conserved residues (Thr-32, Ser-253 and Ser-315), resulting in the export of FoxO3 from the nucleus into the cytoplasm 11 and proteasomal degradation. 12 During conditions of oxidative stress, nuclear translocation and activation of FoxOs occurs through Jun N-terminal kinase phosphorylation, which overrides growth factor signalling. 13 The FoxOs are highly related and to some degree can act redundantly as they bind to the same consensus sequence. 14 Of the four mammalian FoxO family members, FoxO1, FoxO3 and FoxO4 are widely expressed, 14 whereas FoxO6 is mainly expressed in adult brain tissue. 15 In haemopoietic tissues, FoxO1 and FoxO3 are the major FoxOs, with FoxO1 levels highest in lymphoid cells and FoxO3 levels highest in myeloid cells. 16 Although FoxO4 − / − mice do not have an obvious phenotype, 17 FoxO6 − / − mice show impaired memory consolidation 18 and deletion of FoxO1 results in defects in embryonic vascular development, resulting in death at embryonic day 10.5. 17,19 FoxO3-deficient mice display early onset female infertility because of depletion of ovarian follicles caused by widespread follicular activation. 17,20 Aged FoxO3 − / − mice were found to have reduced numbers of haemopoietic stem cells (HSCs), 21 implying that FoxO3 contributes to the maintenance of the HSC pool during aging by reducing oxidative stress. 22 FoxO3 primes HSCs for metabolic stress (starvation)-induced autophagy, which is necessary for HSC survival. 23 Activation of AKT, and thus inactivation of FoxOs, is frequently observed in a range of cancers, 24 raising the possibility that FoxO proteins may serve as tumour suppressors. An early study found that loss of individual FoxOs did not cause cancer predisposition, except for loss of FoxO1. 17,20 When FoxO1 was deleted using Mx-Cre (at 4 weeks of age), mice developed systemic hemangiomas, but with long latency and low frequency. 25 However, compound somatic deletion of three FoxOs, FoxO1/3/4, resulted in thymic lymphomagenesis. 25 Furthermore, a dominant-negative FoxO construct, which expresses the conserved DNA-binding domain of FoxO4 but lacks the transactivation domain and thus inhibits the activity of all FoxOs, greatly accelerated the onset of Eμ-myc lymphomas. 26 This effect is due, at least in part, to FoxO upregulation of p19 ARF expression, which promotes p53-dependent apoptosis. In fact, the dominant-negative FoxO Eμ-myc tumours arose as rapidly as those on a p53 +/ − background. 26 FoxO3 and MYC compete for binding to some promoters and appear to antagonise each other's activity. [27][28][29] FoxO3 directly regulates expression of the MYC inhibitor MXI1, whereas other Max-interacting protein (MXD) family members are regulated indirectly. 30 To determine whether loss of a single FoxO can cooperate in Myc-driven tumorigenesis, we crossed FoxO3 − / − mice 21 with vavP-MYC10 mice, which express the transgene in all haemopoietic lineages and primarily develop myeloid tumours with some T lymphomas, 31,32 and with Eμ-myc mice, which constitutively express Myc in B lymphoid cells and are susceptible to pre-B and B lymphomas. 33,34 Results Loss of FoxO3 accelerates Myc-driven tumorigenesis. vavP-MYC10 (hereafter MYC10) transgenic mice develop macrophage tumours with a median onset of approximately 300 days. Lymphomagenesis was significantly accelerated by the loss of FoxO3, all FoxO3 − / − MYC10 mice having succumbed to disease by 300 days compared with 400 days for the MYC10 mice ( Figure 1a) (median survival of FoxO3 −/− MYC10 256 days versus 295 days for MYC10; Po0.0001). Comparison of spleen weights and white blood cell (WBC) counts of tumour-bearing mice showed that the tumour burden was higher in the FoxO3 −/− MYC10 mice (Figures 1b   and c). The tumour phenotype (Figure 1d) was comparable between the cohorts, being predominantly monocyte/macrophage tumours (Mac1 + F4/80 + ) with a small proportion of T-cell tumours (Thy1 + CD3 + and variable CD4 and CD8 expression), as previously reported for MYC10 mice. 32 Tumours of all genotypes were transplantable in C57BL/6 recipients (n = 9-12 independent tumours).
Haemopoietic homeostasis is perturbed by loss of FoxO3. To determine whether loss of FoxO3 alters the preneoplastic phenotype of MYC10 mice, which could contribute to the enhanced tumorigenesis, we compared the composition of blood and haemopoietic tissues of healthy mice at 8 weeks of age ( Figure 3, Supplementary Figure 1 and Supplementary Table 1). Perhaps surprisingly, loss of FoxO3 alone had a greater impact on haemopoietic homeostasis than expression of the MYC10 transgene, which provokes only a mild preneoplastic phenotype. 32 The FoxO3 − / − mice had significantly increased T cells in the spleen and blood, but not in lymph nodes. Herold et al. 39 have also reported an increase in splenic T cells using the same FoxO3 − / − mice, although increased T-cell numbers were not observed in an earlier study using a different line. 40 Consistent with previous reports, the FoxO3 − / − mice also had increased myeloid cells in the spleen and bone marrow, 21,41,42 reduced recirculating B cells in the bone marrow 43 and elevated immature erythroid cells (Ter119 + ), presumably due to impaired maturation. 44,45 In general, the phenotype of FoxO3 − / − MYC10 mice resembled that of FoxO3 − / − mice. However, in the spleen,   Table 2). Loss of FoxO3 resulted in increased myeloid cells in the blood, spleen and bone marrow of Eμ-myc mice, as in MYC10 mice, but no significant differences were seen in T-cell numbers. As Eμ-myc and FoxO3 − / − Eμ-myc mice develop pre-B or B-cell lymphomas, the impact of FoxO3 loss on B lymphopoiesis was of particular interest. As documented previously, 46 young preneoplastic Eμ-myc mice had a greatly expanded pre-B-cell (B220 + IgM − IgD − ) population, with 15-to 20-fold increases in the spleen, blood and lymph nodes, and there was a modest further increase in the blood of FoxO3 − / − Eμ-myc mice.
Cell cycle analysis. Deregulation of c-myc promotes tumorigenesis through altered expression of genes regulating cell cycling and proliferation. 47 Loss of FoxO may be expected to amplify the cell cycle impact of myc, as FoxO3 regulates genes that promote cell cycle arrest. 6,48 We tested this possibility in preneoplastic populations, to avoid the complication of additional oncogenic mutations.
As the MYC10 and FoxO3 − / − MYC10 mice are predisposed to myeloid tumours, we analysed the cell cycle profiles of bone marrow macrophages and granulocytes from healthy young mice. No differences in the cell cycle distribution were detectable in either Mac1 + Gr1 − or Mac1 + Gr1 + cells ( Figure 5).
For the Eμ-myc and FoxO3 − / − Eμ-myc mice, preneoplastic pre-B cells were analysed immediately after isolation from the bone marrow and also during culture in simple medium lacking exogenous cytokines. However, there was no difference between Eμ-myc and FoxO3 − / − Eμ-myc pre-B-cell cycle profiles either before culture (first panel, Figure 6a) or at any of the time points analysed (4 h, 8 h, 24 h; Figure 6a). Furthermore, tumour cells isolated from Eμ-myc and FoxO3 − / − Eμ-myc mice had similar cell cycle profiles (Figure 6b).
Impact of loss of FoxO3 on apoptosis of Mycoverexpressing cells. The proliferative effect of Myc overexpression is partially counteracted by Myc-driven apoptosis. 49,50 We therefore investigated whether loss of FoxO3 inhibited apoptosis of cells overexpressing Myc. We reasoned that the impact of loss of FoxO3 may become more apparent during culture, as the deficiency in cytokine signalling would normally inactivate AKT and activate FoxO3.
As expression of the vavP-MYC transgene is highest in thymocytes, 32  CD4   Figure 3C). Apoptosis was also assessed for B lymphoid cells from Eμ-myc and FoxO3 −/− Eμ-myc mice. Frank tumours had comparable numbers of apoptotic cells (Figure 7a and Supplementary Figure 4). When pre-leukaemic pre-B cells sorted from bone marrow were cultured in the absence of cytokines, no significant survival advantage could be detected for those lacking FoxO3 (Figure 7b). However, when IL-7 was removed from IL-7-supported cultures of pre-B cells, loss of FoxO3 significantly enhanced the survival of Eμ-myc (but not WT) pre-B cells (Figure 7c). As IL-7 has a critical role in early B lymphopoiesis, this protective effect may account for the modest increase in pre-B cells in FoxO3 −/− Eμ-myc compared with Eμ-myc mice (Figure 4, Supplementary Table 2).

Discussion
We have demonstrated, for the first time, that loss of a single FoxO transcription factor, can promote lymphomagenesis. Whereas Paik et al. 25 found that complete loss of FoxO activity (both alleles of FoxO1, FoxO3 and FoxO4) is required for lymphomagenesis, we have shown using two different mouse models that loss of FoxO3 alone suffices to accelerate Mycdriven lymphomagenesis. In the absence of FoxO3, myeloid tumours arose more rapidly in vavP-MYC10 mice and pre-B and B lymphomas arose more rapidly in Eμ-myc transgenic mice (Figures 1 and 2). Unsurprisingly, the reduction in tumour latency was not as great as that observed when a dominantnegative FoxO construct was used to inhibit all FoxOs in the Eμ-myc model. 25 Deficiency in a single FoxO does not have an observable effect in all tumour models, however, as loss of FoxO3 or FoxO4 does not accelerate or increase the incidence of DMBA-induced tumorigenesis. 25 The larger population of myeloid cells in FoxO3-deficient MYC10 mice is likely to be a significant factor contributing to accelerated myeloid tumour formation in this model (Figure 3, Supplementary Figure 1 and Supplementary Table 1). In the Eμ-myc model, expansion of the pre-B-cell pool, although modest, may contribute to accelerated lymphoma development in the absence of FoxO3.
Several factors may contribute to increasing the size of the susceptible populations. The absence of FoxO3 had no discernible impact on the cycling of pre-leukaemic myeloid cells isolated from MYC10 mice ( Figure 5) or of pre-leukaemic or neoplastic B lymphoid cells taken from Eμ-myc mice ( Figure 6). Presumably other FoxOs can compensate in this role -FoxO1, in particular, was readily detected in lymphoid cells from Eμ-myc and vavP-MYC mice (Supplementary Figure 5), as reported previously for WT lymphoid cells. 16 Previous studies have reported that FoxO3-deficiency increased the viability of mast cells and thymocytes cultured in the absence of cytokines 39,52 and we were able to demonstrate enhanced viability of certain T-cell populations from  Figure 3). Similarly, lack of FoxO3 enhanced the viability of cultured Eμ-myc pre-B cells following IL-7 removal (Figure 7c). Reduced apoptosis has also been documented in dnFoxO-expressing Eμ-myc lymphoma cells. 26 Activation of p53 is a principal route for Myc-induced apoptosis. Myc upregulates p19Arf expression, probably indirectly, and p19Arf binds to and neutralises Mdm2, thereby preventing inhibition and degradation of p53 and facilitating expression of pro-apoptotic p53 targets, BH3-only proteins Puma and Noxa and, in a negative feedback loop, Mdm2. 53 Although expression of dnFoxO reduced the selective pressure to inactivate the p19Arf/p53 pathway during lymphomagenesis in Eμ-myc mice, 26 this was not the case for FoxO3 − / − Eμ-myc lymphomas (Figure 8).
In summary, we have demonstrated using two different mouse models that FoxO3 has a significant tumour-suppressor function in the context of Myc-driven lymphomagenesis. Although our data suggest a role for FoxO3-mediated inhibition of apoptosis in restraining Myc-driven tumour development, we suspect that loss of FoxO3 may also be affecting other tumour-suppressive functions for which FoxO1-/4 cannot fully compensate. Of note, the tumour-suppressive activity of p53 does not involve its capacity to promote apoptosis or cell cycle arrest. 54

Materials and Methods
Mice. Experimental protocols involving the use of mice were approved by the Walter and Eliza Hall Institute's Animal Ethics Committee. All mice were on a C57BL/6 background and were bred at the Walter and Eliza Hall Institute (WEHI). To generate FoxO3 − / − MYC10 mice, FoxO3 +/ − (ref. 21) females were crossed with MYC10 hom32 males (i.e., homozygous for the MYC10 transgene) then offspring were interbred. FoxO3 − / − Eμ-myc mice were produced by mating FoxO3 +/ − females with Eμ-myc 33 males followed by interbreeding of offspring. Analysis was performed on healthy young mice (4 or 8 weeks of age) and cohorts of mice were aged to ethical end point.
Cell cycle analysis. Pre-B cells (B220 + IgM − IgD − ) were purified from the bone marrow of 4-week-old mice by flow cytometry, or by using CD19 MicroBeads (Miltenyi Biotech, Bergisch Gladbach, Germany) and MS columns (Miltenyi Biotech) according to the manufacturer's protocols. The Nicoletti assay 56 was used for cell cycle analysis of the purified pre-B cellscells were resuspended in 0.1% sodium citrate, 0.1% Triton X and 50 μg/ml propidium iodide (PI) and incubated for 30 min on ice before FACS analysis. Alternatively, mixed cell populations, such as bone marrow, were stained with surface markers for the population of interest before fixation and permeabilisation with the Transcription Factor Staining Buffer Set (eBioscience, San Diego, CA, USA). Cells were then stained with PI/RNase staining solution (Cell Signaling Technology, Danvers, MA, USA) for 30 min before analysis on an LSR II flow cytometer (BD Biosciences). Cell cycle profiles were characterised using FlowJo using the Watson pragmatic model (see above).
Survival assays. Thymocyte populations or pre-B cells isolated by flow cytometry were cultured at 0.2-0.5x10 6 cells/ml in high-glucose Dulbecco's Modified Eagle's medium supplemented with 10% foetal calf serum (JRH Biosciences, Brooklyn, VIC, Australia), 50 μM 2-mercaptoethanol (2-ME; Sigma-Aldrich, St Louis, MO, USA) and 100 μM asparagine (Sigma-Aldrich) without additional cytokines. In addition, thymocytes were treated before culture with 1.25 Gy γ-irradiation. Cell viability was determined by staining with FITC-conjugated annexin V and PI followed by flow cytometry. Specific viability was calculated at each time point as (viability of treated cells/viability of untreated cells) x 100%.
T-cell blasts were generated by stimulation of splenocytes (2x10 6 /ml) for 3 days with 1% mIL-2 supernatant 57 and 2 μg/ml concanavalin A (Sigma-Aldrich) in cell culture medium as for survival assays above. Cells were then washed three times to remove concanavalin A and cultured with IL-2 alone for 24 h. To set up the death assay, cells were washed three times to remove IL-2 then cultured with or without IL-2 for 2 days. Viability was determined by flow cytometry as above.