BRAF inactivation drives aneuploidy by deregulating CRAF

Abstract

Aspartate 594 is the third most common BRAF residue mutated in human cancer. Mutants of this residue are kinase inactive and the mechanism(s) by which they contribute to cancer has remained perplexing. Using a conditional knock-in mouse model, we show that the ^D594ABraf mutant does not drive tumor development per se but is able to induce aneuploidy in murine splenocytes and MEFs and contributes to immortalization through the propagation of aneuploid cells. ^D594ABraf lacks kinase activity but induces the related gene product Craf as well as the Mek/Erk pathway. Here we show that the aneuploid phenotype is dependent on Craf. Treatment with the Mek inhibitor U0126 did not attenuate the emergence of aneuploidy but prevented the growth of aneuploid cells. These results provide a previously unidentified link between Craf and chromosomal stability, with important implications for our understanding of the development of cancers with driver mutations that hyperactivate Craf

Introduction

The serine/threonine-specific kinase BRAF is a key intracellular signaling component of the RAS/RAF/MEK/ERK cascade whose principal role is to control the output of the downstream ERK kinases, the ultimate effectors of this pathway (1). Somatic mutations in both RAS and BRAF proto-oncogenes are frequently detected in human cancers (2,3), and the consequent deregulation of ERK activity is thought to subserve multiple aspects of the transformed phenotype (4). A substitution of glutamic acid for valine at residue 600 (V600E) is the most frequent BRAF mutation detected in human cancer, and melanomas represent the human cancer type with the highest levels of BRAF mutation (5). The ^V600EBRAF mutant possesses greater intrinsic kinase activity than ^WTBRAF and represents a class of high activity mutants that drives tumour development through hyperactivation of downstream signaling pathways (5,6). However, four BRAF mutants detected in human cancers (G466E, G466V, G596R and D594V) have lower activity than ^WTBRAF and represent an impaired activity class that must contribute to the cancer phenotype through different mechanisms (6). Three of these four mutants (G466E, G466V and G596R) have been shown to activate MEK-ERK by activating the BRAF homologue CRAF (6,7). The mechanism of activation of CRAF is RAS-independent but involves 14-3-3 mediated hetero-oligomerisation and transphosphorylation of the CRAF activation segment by the residual kinase activity of BRAF (7).

The remaining impaired activity mutant, D594V, affects the functionally conserved DFG motif of the activation segment, a highly conserved motif in protein kinases that is required to chelate Mg²⁺ and stabilize ATP binding to the catalytic site. Mutations of this residue are inactive in other protein kinases (8) and mutation in BRAF generates a kinase with activity as low as that of a catalytically inactive ^K483MBRAF mutant (6). Although initial studies showed that D594 mutants do not activate the MEK/ERK pathway through CRAF when over-expressed, our more recent studies using mice that express ^D594ABraf endogenously have shown that this mutant can promote the formation of aggressive tumours when co-expressed in melanocytes with endogenously expressed oncogenic ^G12DKras and it appears to do this through formation of a heterodimer with Craf (9). These data suggest ^D594BRAF mutations are cooperating rather than driver oncogenes, a result consistent with the fact that there is a significant enrichment for the coincident of ^D594BRAF and RAS mutations in human tumours (9).

Here we have further analysed the role of ^D594ABraf in tumour development by constitutive expression of this mutant in mice. We show that ^D594ABraf has dominant activity in vivo and is able to promote aneuploidy in splenic myeloid cells and cultured MEFs. Aneuploidy is tolerated and contributes to immortalisation of these cells. Although ^D594ABraf has impaired activity it leads to deregulation of Craf and the downstream Mek-Erk pathway and we show that the aneuploidy can be reversed by Craf downregulation. Mek inhibition does not reverse the aneuploidy but prevents the growth of aneuploid cells. These results have important implications for mechanisms underlying development of aneuploidy in cancer cells.

Materials and Methods

Mice

All animal experiments were carried out under UK Home Office License authority. Braf^+/LSL-D594A mice were generated in the same way as Braf^+/LSL-V600E mice except Braf exon 15 contained the A1781C mutation (10). Genotyping of Braf^+/LSL-D594A and all Cre strains was performed using the primer system previously reported (10). Craf^+/D486A mice (11) and immortomice (12) were genotyped as described. Tissues were processed for histology and immunostained as described (10).

Cells and transfections

All cells were maintained in DMEM with 10% FCS and penicillin/streptomycin. MEFs were isolated as reported (10) and treated with adriamycin at 0.25μg/ml for 24 hrs or with sorafenib at 0.2-5μg/ml for 2 hrs. For splenocyte cultures, spleens were passed through a 100μm nylonmesh and suspended in red blood cell lysis buffer (150mM NH₄Cl, 1mM KHCO₃, 0.1mM EDTA, pH7.4). Cells were washed in PBS and cultured +/− 0.2-2μM sorafenib or 0.25-2.5μM U0126.

Splenocyte analysis

For marker analysis, splenocytes were collected and stained as described (13). Cells stained with FITC-conjugated antibodies were incubated with 2μg/ml propidium iodide (PI) without fixation to detect dead cells. For DNA ploidy analysis, cells stained with antibodies were fixed in 90% methanol for 30min on ice, washed, resuspended in PBS containing 100U/ml DNAse-free RNase (Sigma), and incubated for 1 hour at 37°C. 50μg/ml PI was added, and DNA ploidy was analyzed using FACS. For cell proliferation analysis, mice were injected with bromodeoxyuridine (BrdU) (BD Biocsciences; 2mg/mouse i.p.), splenocytes were collected 24hrs after and stained with FITC-conjugated antibodies followed by BrdU staining using APC BrdU Flow Kit (BD Biosciences) according to the manufacturer's instructions.

Chromosome counting and anaphase analysis

Metaphase spreads were prepared by standard procedures and stained with DAPI or Giemsa solution. Each spread was photographed and chromosome number was manually counted. For anaphase analysis, cells grown on coverslips were fixed with 4% paraformaldehyde, stained with DAPI, and analysed using a fluorescent microscope (Leica DM5000B).

Immunoblotting and kinase assays

Protein lysates were prepared by previously published methods (10, 14), except for the lysates to detect p53 that were prepared in 1x SDS sample buffer (62.5mM Tris-HCl pH6.8, 2% (w/v) SDS, 10% (v/v) glycerol, 5% (v/v) β-mercaptoethanol). Antibodies used were: phospho-Erk1/2 (Cell Signaling #9101), Erk2 (Santa Cruz #SC-1647), Craf (BD Biosciences #610152), phospho-Mek1/2 (Cell Signaling #9154), Mek1/2 (Cell Signaling #9122), myc-tag (Cell Signaling #2276), p53 (Cell Signaling #2524), p19^ARF (Abcam #ab80), and actin (Sigma #A2103). Raf kinase activity was measured as described (6, 14). Primary antibodies used for immunoprecipitations were Braf (Santa Cruz #SC-5284) and Craf (Santa Cruz #SC-227).

Statistics

Comparison between any two groups was performed by unpaired student's t test for normally distributed data or Welch's t test for non-normally distributed data, except for the frequency of flow cytometry-detectable aneuploidy in spleens for which chi-square test was employed.

Results

Generation of mice constitutively expressing endogenous ^D594ABraf

We recently reported the generation of a mouse strain (LSL-Braf^D594A) that allows inducible expression of ^D594ABraf off the endogenous Braf gene (9). We induced constitutive expression of ^D594ABraf from one allele of Braf by intercrossing Braf^+/LSL-D594A heterozygous mice with mice heterozygous for the CMV-Cre allele (CMV-Cre^+/o) (15). On a C57BL6 background, no live Braf^+/Lox-D594A;CMV-Cre^+/o offspring were obtained and embryos died between E14.5 and birth (Table S1 & Fig. 1A). However, on a C57BL6/MF1 mixed background Braf^+/Lox-D594A;CMV-Cre^+/o mice were born alive at a frequency of 13% and these adult mice survived for at least 6 months (Table S1 & Fig. 1B). Such early lethality is not evident for mice with a heterozygous knockout of Braf (16), confirming that ^D594ABraf must possess dominant activity. The impaired kinase activity of ^D594ABraf was confirmed by an in vitro kinase assay of MEFs derived from intercrossing Braf^+/Lox-D594A mice (Fig. 1C).

Figure 1.

Survival of Braf^+/Lox-D594A;CMV-Cre^+/o mice on different genetic backgrounds. (A) An embryonic lethal phenotype on a C57BL6 background. The percentage survival of live Braf^+/Lox-D594A;CMV-Cre^+/o embryos/animals in offspring derived from Braf^+/LSL-D594A × CMV-Cre^+/o matings is plotted. (B) Shortened life span on a mixed C57BL6/MF1 background. The percentage survival of Braf^+/Lox-D594A;CMV-Cre^+/o animals derived from Braf^+/LSL-D594A × CMV-Cre^+/o matings from birth onwards is shown. (C) In vitro Braf kinase assays of lysates from Braf^{Lox-D594A/Lox-D594A} MEFs obtained by inter-crossing live Braf^+/Lox-D594A mice. Cells were treated with 0.5% FCS for 20 hours +/− 10% FCS for 10 min. Data represent mean +/−SD.

^D594ABraf induces splenomegaly with predominant amplification of monocytes

Braf^+/Lox-DA;CMV-Cre^+/o mice born alive developed splenomegaly after 3 months of age at 100% penetrance (Fig. 2A). In the mutant spleens, disorganized follicular structures with increased atypical monocytic cells were observed (Fig. 2B). Cell surface marker analysis demonstrated a relative increase of CD11b-positive myeloid cells (Fig. 2C), consistent with the myeloid origin of the atypical cells. We also found a relative increase of CD41-positive megakaryocytes in the mutant spleens (Supplementary Fig. S1). However, in vivo BrdU incorporation and Ki67 immunostaining revealed that accelerated proliferation of both myeloid and lymphoid lineages contributed to the splenomegaly (Fig. 2D). Although histological and cytological characteristics observed in the mutant spleen resembled extramedullary hematopoiesis caused by myeloproliferative neoplasms, we found no significant alteration in the cellularity or differential cell count of the mutant bone marrow (Supplementary Fig. S2), indicating that ^D594ABraf-induced deregulation of hematopoiesis is dependent on the splenic microenvironment. Interestingly, some mutant mice displayed myelofibrosis (Supplementary Fig. S2), but the incidence was only 50% even in the aged animals (3 out of 6 mutants at 9-12 months of age). Since splenomegaly was observed at complete penetrance within 3 months after birth, it is unlikely that myelofibrosis is a direct cause of the spleen phenotype.

Figure 2.

^D594ABraf-expressing mice develop splenomegaly. (A) Spleen weights. Data represent mean +/− SD. Representative photographs of spleens from Braf^+/Lox-D594A and Braf^+/+ mice are shown. (B) H&E staining of spleen sections (upper panels) and Giemsa staining of touch preparation of spleen (lower panels). (C) Flow cytometric analysis of cell surface CD11b expression of splenocytes (upper panels) and quantitation of the percentages (middle bar chart) and absolute numbers (lower bar chart) of CD11b+, B220+ and CD4/8+ splenocytes. Data represent mean +/−SD (n=5). (D) Quantitation of the percentages of dead (PI+) and proliferating (BrdU-incorporating) cells in CD11b+, B220+ and CD4/8+ splenocytes (upper bar chart), and representative Ki67 staining of spleen sections from Braf^+/Lox-D594A and Braf^+/+ mice (lower photographs). Data represent mean +/−SD (n=3).

Braf^+/Lox-D594A without the CMV-Cre transgene also developed splenomegaly (Supplementary Fig. S3), indicating that Cre recombinase-induced genomic damage does not contribute to this phenotype. In addition, splenomegaly developed 24 weeks after tamoxifen treatment of adult Braf^+/LSL-D594A mice expressing the CreER™ transgene (17), indicating that ^D594ABraf expression throughout development is not required for this phenotype (Supplementary Fig. S3). Apart from splenomegaly, no other abnormality was consistently observed, although ^D594ABraf-expressing mice did appear to be predisposed to other occasional pathologies including strokes, genital warts and ovarian cysts (Supplementary Table S2).

Aneuploidization of ^D594ABraf-expressing splenocytes

To characterise the mutant splenocytes, we performed DNA ploidy analysis, initially to evaluate their cell-cycle status. Unexpectedly, this identified aneuploid cells in the ^D594ABraf-expressing samples, regardless of the presence of the CMV-Cre transgene (Fig. 3A). Co-staining with lineage markers identified the aneuploid cells exclusively in the CD11b+ myeloid population (Fig. 3A). We cultured splenocytes and found that, at 5-7 days, ^D594ABraf-expressing splenocytes developed cellular foci comprised of CD11b+ cells whereas ^WTBraf-expressing splenocytes did not survive the culture (Fig. 3B). DNA ploidy analysis after 7 days in culture demonstrated that the modal ploidy is predominantly diploid although tetraploid cells - that may be precursors for aneuploid cells - were also present (Fig. 3C). Metaphase analysis of cells after 7 days in culture also demonstrated that the modal chromosome number was 40 although ~50% of metaphases were aneuploid, including some polyploid cells (Fig. 3D; Supplementary Fig. S4). Significantly, the frequency of diploid metaphases decreased to 25-40% after 11 days in culture (Fig. 3D). We failed to obtain proper metaphases from control cultures, because they proliferated poorly (Fig. 3B). To circumvent this problem, we cultured splenocytes from mice expressing the SV40 T antigen (12) and found that they developed cellular foci comparable to ^D594ABraf-expressing splenocytes but ~70% of metaphases maintained diploid chromosome number at day 11 (Fig. 3D). Thus, aneuploidization induced by ^D594ABraf in this cell type is more drastic than that induced by large T antigen-mediated pRb/p53 inactivation. Furthermore, the fact that mutant splenic myeloid cells have a growth advantage and develop aneuploidy in culture suggests that the effect is cell autonomous.

Figure 3.

Aneuploidization of ^D594ABraf-expressing splenocytes. (A) DNA ploidy analysis. Braf^+/Lox-DA splenocytes frequently display aneuploid peaks (arrows) in the CD11b+ population. The bar chart (right) shows this is statistically significant. (B) In vitro culture of Braf^+/Lox-DA splenocytes. Cell growth was quantitated by counting total viable cells/well after plating 4×10⁶ primary splenocytes on day 0 (left panel). Representative photographs of cultured splenocytes on day 14 (middle panels). Flow cytometric analysis of CD11b expression and Giemsa staining of the cultured splenocytes (right panels). Atypical monocytic cells (arrow heads) similar to those in primary spleens, and phagocytes with cytoplasmic projections (arrows) are observed. (C) DNA ploidy analysis of cultured Braf^+/Lox-DA splenocytes on day 7 demonstrates a tetraploid population in high forward/side scatter cells. (D) Metaphase spread analysis of Braf^+/Lox-D594A compared Braf^+/+ (+T) cultured splenocytes. Data are pooled from 3 independent cultures in which chromosome numbers were analysed in at least 135 metaphases.

^D594ABraf-expressing MEFs escape cellular senescence accompanied by aneuploidization

To investigate if the aneuploid phenotype occurs in other cell types, we derived primary MEFs and cultured them according to a 3T3 protocol. While Braf^+/Lox-D594A;CMV-Cre^+/o MEFs grew similarly to controls in early passages, they escaped senescence much earlier (Fig. 4A and Supplementary Fig. S5). This early immortalization did not require p53 functional loss or p19^ARF deletion as commonly observed during immortalization of wild-type MEFs (Fig. 4B). Braf^+/Lox-D594A;CMV-Cre^+/o MEFs consistently revealed a tetraploid/aneuploid phenotype when immortalized as assessed by population-doubling based measurements (Fig. 4C). In contrast, control lines only acquired a tetraploid phenotype when accompanied by p53 mutation and, even then, their kinetics of tetraploidization was variable (Fig. 4C). Aneuploidization of ^D594ABraf-expressing MEFs suggests that chromosome missegregation could be involved in this process and, indeed, a high proportion of cells displayed abnormal anaphases (Fig. 4D). Common abnormalities observed were chromosome bridges and lagging chromosomes, though a few cells with misaligned chromosomes were also observed. Mitotic traversal times were prolonged in 2/3 mutant lines tested, which also showed the kinetics of nocodazole-induced mitotic arrest/slippage similar to controls (Supplementary Fig. S6), indicating that a defective spindle assembly checkpoint (SAC) is not a common cause for chromosome missegregation in ^D594ABRaf-expressing cells.

Figure 4.

^D594ABraf-expressing MEFs. (A) 3T3 cultures of primary MEFs. Representative photographs at passage 19 are shown in the lower panels. (B) p19^ARF and p53 expression in response to adriamycin treatment in MEF lines at passage 40. (C) Tetraploidization of cultured MEFs. (left panels) The percentages of 2N cells (y-axis) during 3T3 culture plotted against population doubling (x-axis). (right panels) Metaphase spread analysis (n=150 for each line). (D) Abnormal anaphase progression. The frequency of abnormal anaphase (lagging chromosomes, chromosome bridges, or misaligned chromosomes) was quantified. Data represent mean +/−SD (n=3), and the total number of anaphases examined (n) is indicated. Representative photographs demonstrating misaligned chromosomes (left, white arrows), a chromosome bridge (right, red arrow) and a lagging chromosome (right, white arrow head) are indicated.

^D594ABraf induces Erk phosphorylation and Craf activation

In MEFs, phospho-Erk levels were relatively low in both mutant and control cells at early passages, but a progressive increase of the signal was observed from passage 3 to 7, which was more prominent in the mutant lines (Fig. 5A). Moreover, once immortalized, Erk phosphorylation levels were consistently higher in the ^D594ABraf-expressing cells (Fig. 5A). A similar increase in Erk phosphorylation was observed in the spleen of Braf^+/Lox-D594A mice at 6 and 35 weeks of age compared to controls (Fig. 5B). Robust Erk activation in the spleen was also observed 1 week after the induction of the Braf^+/LSL-D594A allele by CreER™ activation (Fig. 5B). Using immunohistochemical staining, clusters of phospho-Erk-positive cells were detected in the red pulp of the ^D594ABraf-expressing spleen where mainly myeloid and erythroid cells are localized; the white pulp (lymphoid follicles) was completely phospho-Erk negative (Fig. 5C). Most of the signal was observed in relatively large-sized cells with oval-shaped nuclei and low N/C ratio, consistent with the morphology of atypical monocytic cells (Fig. 2B). A modest increase in ERK phosphorylation was also observed in wild-type MEFs transduced with retroviral vectors expressing D594 mutants (Supplementary Fig. S7). Moreover, co-transfection of vectors expressing human ^WTBRAF or CRAF with ^D594ABRAF dramatically enhanced the ERK phosphorylation induced by either alone, demonstrating that ^D594ABRAF can cooperate with ^WTBRAF or CRAF to promote ERK phosphorylation, at least when overexpressed in HEK293T cells.

Figure 5.

Erk phosphorylation and Raf activity. (A) Erk phosphorylation in primary MEFs at passage 3-7 (upper panels) and immortalized MEFs in steady-state, serum-starved, or serum-starved then stimulated with 20% FCS for 10 min (lower panels). (B) Erk phosphorylation in spleens at 6 or 35 weeks of age (upper blot) and from Braf^+/LSL-DA mice +/−CreER transgene 1 week after injection with tamoxifen (lower blot). (C) Immunostaining of Braf^+/Lox-D594A spleen with phospho-ERK antibody, displaying positive staining in red pulp (boxed area of top panel enlarged in lower panel), but not in white pulp (*). (D) In vitro kinase assays of Braf/Craf in spleen (left) and immortalized MEFs (right). Data represent mean +/−SD of four samples of each genotype.

To further elucidate roles for ^WTBraf and Craf in the ^D594ABraf response we performed in vitro kinase assays. Surprisingly, Braf activity was extremely low in both Braf^+/Lox-D594A and Braf^+/+ spleens and there was a significant increase in Craf activity in Braf^+/Lox-D594A spleens compared to controls (Fig. 5D). In MEFs, Braf activity was decreased by ~40% suggesting that, in contrast to the overexpression (Supplementary Fig. S7), ^D594ABRaf does not have a major effect on transactivation of endogenous ^WTBraf when expressed at physiological levels. In a similar manner to the spleen, Craf activity was significantly increased in Braf^+/Lox-D594A MEFs (Fig. 5D).

Craf is required for the aneuploid phenotype

We crossed Braf^+/Lox-D594A mice with mice constitutively expressing a kinase inactive version of Craf (Craf^+/D486A) in which Craf expression is reduced by ~50% (11). Since Craf and Braf are both located on mouse chromosome 6 it was difficult to obtain Craf^D486A/D486A homozygotes that also contained the Braf^Lox-D594A allele and so Craf^+/D486A heterozygotes were analysed. In the Braf^+/Lox-D594A;Craf^+/D486A spleens, Craf expression was reduced by ~50% (Fig. 6A) and correlated with a significantly decreased spleen weight compared to Braf^+/Lox-D594A;Craf^+/+ littermates (Fig. 6B). Importantly, none of the Braf^+/Lox-D594A;Craf^+/DA splenocytes demonstrated aneuploid peaks (n=8; Fig. 6C). We also analysed primary MEFs from Braf^+/Lox-D594A;Craf^+/D486A mice and found that reduction of Craf protein expression by ~50% (Fig. 6D) significantly reduced the proportion of aneuploid cells and increased the proportion of euploid cells (Fig. 6D).

Figure 6.

Craf is required for ^D594ABraf-induced aneuploidy. (A-C) Spleen analysis. Decreased Craf expression (A) and weight (B), and normal DNA ploidy (C) in Braf^+/Lox-D594A;Craf^+/D486A spleen are indicated. (D) Primary MEF analysis. Metaphase spreads from MEFs at passage 4 are categorized as indicated. Data represent mean +/− SD. Decreased Craf expression in Braf^+/Lox-D594A;Craf^+/D486A MEFs is shown on the right immunoblot.

To further investigate the roles of Craf and Erk, we applied the inhibitor sorafenib or the Mek inhibitor UO126 to Braf^+/Lox-D594A in vitro spleen cultures. Although sorafenib is known to inhibit Braf, Craf and several other kinases in the low nM range (18), Braf is inert in the spleen (Fig. 5C) and our analysis has shown that cultured splenocytes do not express other kinases known to be targeted by this drug including Pdgfrb, Flt-3 and Flk-1 (Supplementary Fig. S8). Cell surface expression of c-Kit was detected in a small population of mainly small-sized monocytic cells, but not in the vast majority of cultured splenocytes (Supplementary Fig. S8). Thus, potential effects of sorafenib on Braf and other kinases are likely to be minimal.

Sorafenib and UO126 demonstrated different effects on the cells. Firstly, while UO126 treatment suppressed growth, sorafenib potentiated growth (Fig. 7A). Secondly, although sorafenib efficiently inhibited MEK phosphorylation, its effect on ERK phosphorylation was relatively modest whereas UO126 treatment profoundly attenuated ERK phosphorylation (Fig. 7B). Finally, sorafenib effectively suppressed splenocyte aneuploidization as assessed by the percentage of cells retaining diploid chromosomes whereas UO126 did not improve the percentage of diploid cells (Fig. 7C). In particular, when treated with 0.25μM UO126, a significant increase of polyploid metaphases was observed, even though Erk phosphorylation at this concentration was comparable to sorafenib treatment. We also analysed splenocytes from Braf^+/LSL-V600E;Mx1-Cre^+/o mice (10) but found that ~80% of the cultured cells maintained diploid chromosomes (Fig. 7D). Thus, robust activation of the Erk pathway by the high activity ^V600EBraf mutant is not sufficient for inducing aneuploidy.

Figure 7.

MEK/ERK activation is required for growth of aneuploid cells but does not induce aneuploidy. (A-C) Inhibitor treatments of Braf^+/Lox-D594A splenocytes. The yield of cells on day 11 of SF/UO126-treated cultures (normalized to DMSO controls) with representative photographs is shown in (A), Mek/Erk phosphorylation in cultured splenocytes treated for 24 hrs in (B), and metaphase spread analysis of SF/UO126-treated cultures in (C). (D) Metaphase spread analysis of cultured splenocytes from Braf^+/Lox-D594A, Braf^+/Lox-V600E and Braf^+/+ (+T) mice. In (C) and (D), metaphases on day 11 are categorized as diploid (40 chromosomes), hypodiploid (20-39), hyperdiploid (41-60) and polyploid (61-200), and data represent mean +/− SD. Asterisks indicate statistically significant differences compared to controls in (C) or between Braf^+/Lox-D594A and Braf^+/Lox-V600E splenocytes in (D) (** p<0.01, * p<0.05, student-t test).

Rok-α is not significantly altered

Craf is known to regulate several effectors in a Mek-independent manner including Rok-α which physically interacts with Craf in an active conformation (19). Rok-α is regulated by the small GTPase RhoA which is essential for cytokinesis (20) and thus it is a potential candidate effector of the aneuploid/tetraploid phenotype. We found a weak interaction between Craf and Rok-α in Braf^+/+ MEFs as previously reported (21, 22), but this was not altered in asynchronized Braf^+/Lox-D594A MEFs, and no significant difference in Rok-α kinase activity was observed (Supplementary Fig. S9). In mitotic cells, the Craf/Rok-α interaction was slightly attenuated in Braf^+/Lox-D594A MEFs, and there was a marginal increase in Rok-α kinase activity (Supplementary Fig. S9). However, since the basal Craf/Rok-α interaction was weak, further attenuation of the interaction by ^D594ABraf did not significantly impact on Rok-α activity.

Discussion

Here we have shown that expression of a kinase inactive Braf mutant in mice stimulates Craf activation and this drives aneuploidy in the spleen and MEFs. In both cell types, aneuploidy is tolerated and propagates immortalisation, a key cancer hallmark. Using genetic and inhibitor approaches we show that the aneuploid phenotype is dependent on the Braf homologue Craf. Although ^D594ABraf is also able to hyperactivate the Mek/Erk pathway, our data argue that this pathway is required for the growth/survival of aneuploid cells but not the emergence of aneuploidy.

We observed that ^D594ABraf has dominant activity in vivo even when expressed from only one allele of the Braf gene. However, although ^D594ABraf causes embryonic lethality and aneuploidisation of splenic myeloid cells and MEFs on distinct genetic backgrounds, no other consistent pathology was detected suggesting that this mutant does not have dominant activity in most adult or embryonic tissues. The reasons for this are not entirely clear since Braf and Craf are both widely expressed in the adult and embryo (our unpublished data). However, we have found that the spleen is unique because Braf is inactive in this tissue while Craf activity levels are relatively high regardless of the presence or absence of ^D594ABraf (Fig. 5D). Thus, one reason why abnormalities are not detected in other tissues could be because the remaining ^WTBRaf activity competes with or suppresses the hyperactivated Craf induced by ^D594ABraf. Interestingly, we have previously demonstrated inactivation of BRAF in RAS mutant melanoma cells (23) and hyperactivation of CRAF can be induced by further inhibition of intrinsic BRAF kinase activity (9). Although it is currently unknown how the initial BRAF inactivation can be achieved in this context, we assume that similar mechanisms might occur when wild-type RAS is activated by upstream signals. Therefore, it is possible that Ras.GTP levels are maintained at high levels in the spleen, retaining Braf in an inert condition, which then sensitizes the splenic myeloid cells to ^D594ABraf-induced Craf hyperactivation. Consistent with this assumption, Braf kinase activity in myeloid cells has been shown to be unresponsive to cytokine stimuli (24), suggesting that cytokine-induced Ras activation in this cell type maintains Braf in an inert condition, rather than evoking its activation. With regard to Braf^+/Lox-DA MEFs, the remaining ^WTBraf remains active (Fig. 5D) and so a different mechanism must account for the hyperactivation of Craf in this cell type, the basis of which is currently unknown.

We found that hyperactivation of the Mek/Erk pathway by expression of endogenous ^V600EBraf in the spleen does not induce aneuploidy and that U0126 treatment does not inhibit aneuploidy. In this respect our results differ from those of Guadagno and colleagues (25) who have shown that ectopic expression of ^V600EBRAF in melanoma cell lines, primary melanocytes and mammary epithelial cells induces aneuploidy, and that mitotic abnormalities of melanoma cell lines with the ^V600EBRAF mutation can be suppressed by U0126 treatment. The reasons for the differences between these results and ours are not currently clear but may reflect differences in the cell type under study. In humans, ^D594BRAF mutations are more frequent in colorectal cancers (CRCs) than any other cancers (26). The high activity mutant ^V600EBRAF is associated with microsatellite unstable CRCs (27) in which genomic instability arises as a result of defects in DNA mismatch repair. However, intriguingly, ^D594BRAF mutations are found in microsatellite-stable CRCs (28), which are likely to involve chromosomal instability because of the mutually exclusive nature of these two forms of genomic instability commonly observed in CRCs (29). These clinical observations are consistent with our finding that the aneuploid phenotype is induced by ^D594ABraf, but not by ^V600EBraf.

Aneuploidy is a common characteristic in human cancer (30), but its role in tumourigenesis remains controversial. Several lines of evidence indicate that aneuploidy negatively affects cellular growth (31) and inhibits tumourigenesis in certain contexts (32). Consistent with this observation, sorafenib-treated Braf^+/Lox-D594A splenocytes that maintain a diploid karyotype grow faster in vitro even though they have attenuated Mek/Erk activity (Fig. 7A,B). However, we have found that profound Mek inhibition using UO126 suppresses the growth and/or survival of aneuploid splenocytes (Fig. 7A,B). Thus, although our genetic and pharmacological data argue that Craf activation drives aneuploidy in a Mek/Erk-independent manner, the enhanced Erk signaling observed in aneuploid Braf^+/Lox-D594A cells might be necessary to overcome the growth-inhibitory effect of aneuploidy and facilitate the emergence of aneuploid cells with a growth advantage. The immortalisation phenotype accompanied by aneuploidisation in the mutant MEFs may also be explained by this mechanism. It is notable that ^D594ABraf-expressing MEFs with the aneuploid phenotype grow similarly to wild-type MEFs whereas aneuploid MEFs with a deregulated SAC show early-onset senescence (33, 34) or accelerated cell death (35, 36), suggesting that the growth-inhibitory effect of aneuploidy in MEFs is compensated by ^D594ABraf.

The mechanisms by which Craf induces aneuploidy are currently not clear. Recent studies have demonstrated that inefficient resolution of merotelic kinetochore-microtubule attachment during mitosis could be a widespread cause for cancer aneuploidy (37, 38). Mitotic cells with unresolved merotelic attachment evade SAC detection and proceed into anaphase with lagging chromosomes (39), which could result in either the generation of near-diploid aneuploid progenies or tetraploidization, depending on whether the cells successfully undergo subsequent cytokinesis. Our observations in mutant MEFs are mostly consistent with this scenario, suggesting that ^D594ABraf-induced Craf activation might disturb the process to resolve the merotelic attachment. One candidate effector protein is Aurora B since Craf has previously been shown to regulate this chromosome passenger (40) and Aurora B is known to prevent merotelic attachments through microtubule-depolymerizing kinesins (41). Although Rok-α represented a good candidate effector for the aneuploid phenotype downstream of Craf, its binding to Craf and its activity were not significantly altered in ^D594ABraf-expressing MEFs (Supplementary Fig. S9) and so we are focusing on utilizing proteomic screens to comprehensively elucidate other downstream effectors as well as examining Aurora B.

Although impaired activity BRAF mutations that transactivate CRAF are detected at a low frequency in human cancer, our data have important implications for a wider number of cancers that are driven by oncogenes acting upstream of CRAF including EGFR and RAS mutant tumours. Clearly a key question to address is whether these oncogenes also promote chromosomal instability through CRAF and, if so, whether this can be attenuated by CRAF inhibition. Impaired activity BRAF mutants are frequently coincident with oncogenic RAS mutations in human cancers (26) and in these - albeit rare - cancers, we may expect the hyper-elevated CRAF induced by the combination of both oncogenes to enhance the aneuploidy response compared to mutation of either oncogene alone. Such a situation is likely to be highly detrimental to the individual and, indeed, this mechanism may well account for the highly aggressive melanomas we observed following the combined expression of ^D594ABraf and ^G12DKras in melanocytes (9). To confirm this, we are currently evaluating the aneuploidy status of melanoma cells derived from tumours driven by ^D594ABraf and ^G12DKras compared to tumours driven by oncogenic Kras alone.