Oncogenic RAF1 rearrangement and a novel BRAF mutation as alternatives to KIAA1549:BRAF fusion in activating the MAPK pathway in pilocytic astrocytoma

Abstract

Pilocytic astrocytomas (PAs), WHO malignancy grade I, are the most frequently occurring central nervous system tumour in 5-19 year-olds. Recent reports have highlighted the importance of MAPK pathway activation in PAs, particularly through a tandem duplication leading to an oncogenic BRAF fusion gene. Here we report two alternative mechanisms resulting in MAPK activation in PAs. Firstly, in striking similarity to the common BRAF fusion, a tandem duplication at 3p25 was observed, which produces an in-frame oncogenic fusion between SRGAP3 and RAF1. This fusion includes the Raf1 kinase domain, and shows elevated kinase activity when compared with wild-type Raf1. Secondly, a novel 3bp insertion at codon 598 in BRAF mimics the hotspot V600E mutation to produce a transforming, constitutively active BRaf kinase. Whilst these two alterations are not common, they bring the number of cases with an identified ‘hit’ on the Ras/Raf signalling pathway to 36 from our series of 44 (82%), confirming its central importance to the development of pilocytic astrocytomas.

Central nervous system (CNS) tumours account for ~25% of childhood malignancies (Office for National Statistics, 2004), and pilocytic astrocytoma (PA) is the most common CNS tumour in under-19 year-olds (Central Brain Tumour Registry of the United States, 2005). The World Health Organisation grades PAs as malignancy grade I (Louis et al., 2007), and they carry a more favourable prognosis than most other gliomas. They are not usually widely infiltrating and malignant progression is extremely rare, although the latter may be exacerbated by radiotherapy (Tomlinson et al., 1994). However, recurrence is seen in up to 19% of cases (Dirven et al., 1997), and both the primary tumor and surgery are associated with considerable morbidity.

Pilocytic astrocytomas classically present as relatively circumscribed tumours, with histological hallmarks including bipolar tumour cells and eosinophilic Rosenthal fibres. However, the varying histology of PAs can make diagnosis challenging, since areas may resemble oligodendroglial or higher-grade astrocytic tumours. In addition, necrosis and vascular proliferation do not necessarily imply the unfavourable prognosis that they would in other gliomas (Giannini and Scheithauer, 1997; Louis et al., 2007).

Recent reports have indicated a central role for the mitogen activated protein kinase (MAPK) pathway in the tumorigenesis of pilocytic astrocytomas. In particular, we have recently described a tandem duplication at 7q34 producing a transforming BRAF fusion gene in 29/44 tumours (66%), and V600E point mutation of BRAF in two further cases (Jones et al., 2008). Three further cases occurred in patients with clinically diagnosed neurofibromatosis type 1, an autosomal dominant condition caused by mutation of the NF1 gene encoding neurofibromin that is particularly associated with optic pathway gliomas (Listernick et al., 1999). Neurofibromin acts as a Ras GTP-ase activating protein (RasGAP), and reduced levels lead to Ras overactivity (Le and Parada, 2007). The clinicopathological details of the cases in our series, including MAPK pathway alterations, are provided in Supplementary Table 1.

In light of this high frequency of activating alterations in the Ras/Raf signalling pathway, copy number analysis and mutational status of the 10 cases in our series without an identified MAPK pathway alteration were re-examined. We have previously reported copy number gain at 3p25 in a single tumour (PA20), and noted RAF1 as a possible target (Jones et al., 2006). Tandem duplication at 7q34 leading to a fusion between KIAA1549 and BRAF occurs in the majority of PAs (Jones et al., 2008). Thus, the possibility of a RAF1 fusion in PA20 was investigated. The clones bordering the region of gain, RP11-334L22 and RP11-163D23, lie within or overlap the genes SRGAP3 (SLIT-ROBO Rho GTPase-activating protein 3, also known as WRP or MEGAP) and RAF1, respectively (See Fig. 1). RT-PCR with primers in SRGAP3 exon 10 and RAF1 exon 10 gave a product with cDNA from PA20, but not in tumours without 3p25 gain, or in normal brain (see Fig. 1b). Sequencing analysis confirmed an in-frame fusion transcript with SRGAP3 exons 1-12 joined to RAF1 exons 10-17 (Fig. 1c). The srGAP3 protein contains an amino-terminal Fes/CIP4 homology (FCH) domain, a Rho GTPase-activating protein (RhoGAP) domain, and a Src homology 3 (SH3) domain. The gene has been associated with 3p deletion syndrome and severe mental retardation (Endris et al., 2002). The same study showed high expression in fetal and adult brain. It plays a role in neuronal development, and interacts with the cytoskeleton (Soderling et al., 2007; Yang et al., 2006).

Figure 1. Identification and characterisation of a novel RAF1 fusion gene.

(a) An array-comparative genomic hybridisation (aCGH) plot of PA20 showing gain between clones RP11-334L22 and RP11-163D23 at 3p25. The array has been described previously (Jones et al., 2006). Raw data have been deposited with the Gene Expression Omnibus (GEO), accession no. GSE11263. (b) RT-PCR with primers in SRGAP3 exon 8 (PC5536) and RAF1 exon 10 (PC5537) give a 230bp product with cDNA from PA20, but not with cDNA from PA44 (lacking 3p25 gain) or normal brain (NB. Ambion, Austin, TX). A control PCR with primers in RAF1 exons 7 and 10 (PC5630 and PC5537, respectively) gave a 330bp product from wild-type RAF1 in all three samples. M; 100bp ladder (Invitrogen, Paisley, UK), –ve; no template control. (c) A sequence trace confirming a fusion between SRGAP3 exon 12 and RAF1 exon 10. RT-PCR product was cycle-sequenced with PC5536 using BigDye v3.1 chemistry and run on a 3100-Avant Genetic Analyser (Applied Biosystems, Cheshire, UK). Primer sequences and PCR conditions are given in Supplementary Table 2. The fusion transcript sequence has been deposited in the EMBL Nucleotide Sequence Database, accession no. FM209427. (d) Schematic representation of the tandem duplication rearrangement, and the SRGAP3:RAF1 fusion protein. FCH; Fes/CIP4 homology domain, KD; kinase domain. (e) NIH3T3 cells were stably retrovirally transduced as previously described (Jones et al., 2008) with either empty pBabe-puro vector (EV), wild-type RAF1 (WT), or the SRGAP3:RAF1 fusion gene (S:R). Protein was extracted from cells lysed in RIPA buffer supplemented with Complete protease inhibitor cocktail (Roche Diagnostics, Burgess Hill, UK), run on a 4-12% NuPage gel and transferred to a 45μm PVDF membrane (Invitrogen). Membranes were blocked for 1hr with 5% milk in TBS/0.1% Tween-20 and probed with either anti-phospho-MEK1/2 (top panel, rabbit monoclonal, Cell Signalling Technology, Danvers, MA) or anti-MEK1/2 loading control (bottom panel, rabbit polyclonal, Cell Signalling Technology). Goat HRP-conjugated anti-Rabbit Ig secondary antibody was used. Blots were visualised with ECL+ (GE Healthcare, Little Chalfont, UK) using a Fujifilm LAS-4000 imager (Fujifilm UK Ltd, Bedford, UK). Expression from the transduced constructs was demonstrated by probing with anti-HA (see Supplementary Figure 1). Cells expressing the srGAP3:Raf1 fusion protein (S:R) show elevated endogenous phospho-MEK levels compared with those expressing Raf1^WT (WT) or transduced with empty vector (EV). Two independent transductions gave similar results. -ve; lysis buffer only. (f) The same stably transduced cells were assayed for growth in soft agarose as previously described (Jones et al., 2008). Three wells per construct were plated using 10,000 cells/well. Representative fields are shown following 14 days of growth. Cells expressing the srGAP3:Raf1 fusion protein demonstrated a transformed phenotype as evidenced by their anchorage-independent growth. Scale bars = 250μm. (g) Colonies greater than 0.1mm were scored for each well. The mean score for each construct is given with error bars indicating +/− 1SD. ***; p<0.0005, two-tailed Student’s t-Test.

RAF1 was the first identified human homolog of the v-raf gene from a transforming retrovirus (Bonner et al., 1984; Rapp et al., 1983). The gene has since been extensively studied, and is involved in a variety of cellular processes including cell proliferation, survival and migration (Zebisch and Troppmair, 2006).

The rearrangement and the protein resulting from the SRGAP3:RAF1 fusion are shown schematically in figure 1d. The fusion transcript retains the complete, in-frame coding sequence for the Raf1 kinase domain and also encodes the srGAP3 FCH domain, which may play a role in subcellular localisation (Icking et al., 2006), but lacks the RhoGAP and SH3 domains. The transcript sequence has been deposited in the EMBL Nucleotide Sequence Database, accession number FM209427. NIH3T3 cells were retrovirally transduced with either wild-type RAF1 or the SRGAP3:RAF1 fusion gene. Hyperactivity of the fusion protein was demonstrated by elevated phosphorylation of endogenous MEK (a substrate for Raf kinase) when compared with cells expressing wild-type Raf1 (Fig. 1e). The transforming activity of the fusion protein was demonstrated by its ability to confer an anchorage-independent growth phenotype (Fig. 1f).

While mutations in BRAF are seen in ~8% of all human neoplasia (Davies et al., 2002), point mutations in RAF1 are incredibly rare. Raf1 has a lower basal activity than BRaf due to a lack of charged residues in the N-region of its kinase domain, meaning that synergistic alterations are needed to produce elevated kinase activity (Emuss et al., 2005). However, constitutive, transforming activation of Raf1 kinase by amino-terminal truncation has been demonstrated (Stanton et al., 1989), and genetic rearrangements substituting the Raf1 N’-terminus but retaining the kinase domain have been reported in screens for transforming sequences from rat and human tumours (Tahira et al., 1987). However, most of these events were thought to have occurred during transfection of DNA for screening, with only one study describing a RAF1 fusion gene from a primary cancer (Shimizu et al., 1986). Thus the present report is only the second description of oncogenic Raf1 activation due to genetic rearrangement in a primary tumour. Since activation of Raf1 in this way requires only one (albeit more complex) genetic alteration compared with the two synergistic point mutations necessary to produce elevated kinase activity, rearrangements involving RAF1 may prove to be more common than somatic mutations.

A further novel mechanism of MAPK pathway activation was observed in sample PA12. This consists of a trinucleotide insertion in BRAF at either base 1795 (c.1795_1796insCTA, shown in bold below) or 1796 (c.1796_1797insTAC, underlined below). It is not possible to distinguish between these two events, since both would result in the observed coding sequence change from GCTACAGTG to GCTACTACAGTG, leading to an additional threonine residue in close proximity to the mutational hotspot valine at position 600 (p.A598_T599insT). Sequence analysis of constitutional DNA confirmed this as a somatic mutation (see Fig. 2a). This mutation constitutively activates BRaf kinase activity to a similar degree as the V600E mutation, and is transforming in NIH3T3 fibroblasts as demonstrated by an anchorage independent growth assay (Fig. 2b,c). A trinucleotide insertion encoding a valine at this location has previously been identified in papillary thyroid carcinoma, and this was also shown to confer transforming activity (Moretti et al., 2006). Insertions of between 1 and 6 amino acids between codons 599 and 600 have also been described in a total of five skin and thyroid tumours (See http://www.sanger.ac.uk/perl/genetics/CGP/cosmic). However, the threonine insertion reported here is unique.

Figure 2. Identification and functional characterisation of a novel BRAF mutation.

(a) Sample PA12 shows a complex sequence trace due to a 3bp insertion in BRAF exon 15. This exon was then amplified from PA12 tumour and blood genomic DNA with primers PC4580 and PC4581, and cloned into a TOPO-TA vector (Invitrogen). Fourteen colonies for each were assessed by cycle-sequencing as described above using a T3 primer. Nine of the tumour colonies showed the trinucleotide insertion (bottom right), which was not seen in any of the blood DNA clones (top right). (b) Kinase activity of wild-type (WT), V600E and ins598T (insT) BRaf was assayed with an in vitro kinase assay kit as previously described (Jones et al., 2008). BRaf^ins598T showed constitutive kinase activity at a similar level to BRaf^V600E, both of which were significantly more active than BRaf^WT. Each protein was assayed in triplicate. Two independent transfections showed similar results. (c) NIH3T3 cells were retrovirally transduced with either pBabe-puro vector (EV), wild-type BRAF (WT), BRAF^V600E (V600E), or BRAF^ins598T (insT). Expression from the transduced constructs was demonstrated by western blotting with anti-HA (as described above, see Supplementary Figure 2). Stably transduced cells were then assayed for growth in soft agarose as previously described (Jones et al., 2008). Three wells per construct were plated using 10,000 cells/well. Representative fields are shown following 11 days of growth. Cells expressing mutant BRaf demonstrated a transformed phenotype as evidenced by their anchorage-independent growth. Scale bars = 250μm. (d) Colonies greater than 0.1mm were scored for each well. The mean score for each construct is given with error bars indicating +/− 1SD. *; p<0.05, **; p<0.005, ***; p<0.0005, two-tailed Student’s t-Test.

A small number of activating point mutations in K-Ras have been reported in PAs (Janzarik et al., 2007; Maltzman et al., 1997; Sharma et al., 2005). However, as previously described, screening of our series did not identify any mutations in exons 2 or 3 of H-, K- or N-Ras (Jones et al., 2008).

These data bring the number of tumours with a BRAF alteration in our PA series to 32/44 (73%), and the total with any MAPK pathway alteration to 36/44 (82%). An outline of the pathway and a summary of the alterations identified to date are given in figure 3.

Figure 3. An overview of MAPK pathway alterations in PAs.

The frequency of each of the activating alterations in the Ras/Raf pathway observed in our tumour series is indicated (n=44). RTK; receptor tyrosine kinase, TF; transcription factor.

The similarity of the rearrangements producing the SRGAP3:RAF1 fusion and the common KIAA1549:BRAF fusion is striking, and suggests that tandem duplication may be a more common mechanism of producing fusion oncogenes than previously thought. In addition, our results further confirm the pivotal role of Ras/Raf signalling to the tumourigenesis of pilocytic astrocytomas, and underline the potential of this pathway as a therapeutic target.