Main text

The serine/threonine kinase BRAF is mutated and constitutively activated in 40–60% of cutaneous melanomas. Selective inhibitors of mutant BRAF, including dabrafenib and vemurafenib, improve the progression-free and overall survival of BRAF-mutant melanoma patients [1, 2]. Despite this activity, acquired resistance develops in most melanoma patients and multiple mechanisms of resistance have been described. Most of these resistance effectors, including alterations in MEK1, BRAF and N-RAS, promote the reactivation of the mitogen activated protein kinase pathway [3]. Alternate splicing of oncogenic BRAF is the most common driver of acquired resistance to BRAF inhibitors and is evident in approximately 30% of resistant melanomas [4,5,6]. Splice variants of BRAF encode an active kinase, but lack an intact RAS binding domain. These truncated proteins are prone to dimerization and have reduced affinity for the class I BRAF inhibitors, vemurafenib and dabrafenib [4]. Predictably, melanomas expressing BRAF splice variants display MAPK re-activation in the presence of BRAF inhibitors, and retain sensitivity to the inhibition of the downstream kinases MEK and ERK [5, 7].

One mechanism of altered BRAF splicing was recently shown to involve a C-to-G mutation in intron 8 within a predicted splicing branch point, 51 nucleotides upstream of BRAF exon 9 (i.e C > G − 51). This intron 8, variant was sufficient to promote the expression of a BRAF transcript lacking exons 4–8 (BRAF exon 4–8∆; also referred to as BRAF3–9) in a single vemurafenib resistant melanoma cell line [8]. The significance of this intronic mutation was not clinically validated, although we are aware of five patients with acquired BRAF inhibitor resistance driven by the BRAF exon 4–8∆ alternate transcript [4,5,6].

We selected three of these five BRAF inhibitor resistant patients (Patients 7, 10 and 28 [5]). These BRAFV600-mutant patients progressed within 1 year of receiving BRAF inhibitor monotherapy and analysis of the matched pre-treatment and progressing tumours revealed that only the progressing biopsies expressed the BRAF exon 4–8∆ splice variant (Additional file 1: Figure S1). We analysed the intron 8 splicing branch site in these matched pre-treatment and progressing melanomas and in all melanomas the -51– -45 branch point sequence and the exon 9 acceptor splice site were wild type (Fig. 1).

Fig. 1
figure 1

BRAF intronic splice and branch point sequences in melanomas expressing the BRAF exon 4–8∆ transcript. a Schematic diagram of the BRAF gene, showing exons (numbered boxes) and introns (black bars) and the alternate splicing event leading to the BRAF exon 4–8∆ transcript. b Sanger sequencing traces showing the sequence encompassing the junction between BRAF intron 8 and exon 9 in BRAF inhibitor resistant melanomas (Progs) expressing the BRAF exon 4–8∆ transcripts and derived from patient 7, 10 and 28. The branch point motifs, including the −51 nucleotide associated with BRAF exon 4–8∆ splicing, and splice sites are boxed

Full size image

We analysed the impact of the C > G − 51 intron 8 mutation using several algorithms, and it was not predicted to break the branch point nor alter splice factor binding (Table 1). The −51 G mutation potentially activates an intronic cryptic donor site (Table 1), but usage of this splice site would not generate the BRAF exon 4–8∆ transcript. Moreover, an alignment of 66 intronic branch point disease mutations revealed that mutations altering splicing occur predominantly at the highly conserved adenine residue (equivalent to −48 in the intron 8 branch site), with no reported disease causing mutation at the position equivalent to −51 [9]. It is also worth noting that unlike the branch point adenine, which is highly conserved in mammalian branch point sequences, the −5 branch point nucleotide (equivalent to the −51 BRAF intron 8 branch site) is degenerate in mammals [10], and the BRAF intron 8 sequence shows a cytosine to thymine bias in many non-human primates (data not shown).

Table 1 Predicted effects of −51 mutation in intron 8
Full size table

Significantly, the expression of BRAF splice variants is insufficient to confer MEK inhibitor resistance in melanoma (Additional file 2: Figure S2). Consequently, BRAFV600-mutant metastatic melanoma patients treated with combination BRAF and MEK inhibitors, rarely develop resistance due to alternate BRAF splicing [11]. In this clinical setting, therefore, the therapeutic potential of splicing modulators, such as spliceostatin A, which inhibit the formation of the BRAF exon 4–8∆ via inhibition of the SF3B1 splicing factor [8], is diminished.

BRAF inhibitor resistance mediated by alternate BRAF splicing is likely a result of cis-acting alterations that disrupt the use of allele-specific constitutive splice sites. Trans-acting mutations can also affect basal and alternate splicing machinery, but these mutations would be predicted to alter the processing of both wild type and mutant BRAF pre-mRNA. Several studies have now confirmed that the wild type BRAF allele is transcribed and processed correctly in mutant BRAF melanoma, and the shorter, variant BRAF transcripts are derived only from the mutant BRAF allele [4, 5]. Our data show that alternate BRAF exon 4–8∆ splicing in melanomas derived from patients with resistance to BRAF inhibitor monotherapy is not associated with a mutation in the −51 position in intron 8. Regardless of the precise mechanisms altering the splicing of BRAF, downstream MEK and ERK inhibition effectively inhibit the proliferation of BRAF inhibitor resistant melanoma cells expressing BRAF splice variants [5, 7].

Methods

Patients

All patients included in this study had BRAFV600 mutant metastatic melanoma, and were treated with either dabrafenib or vemurafenib [5]. All patients had a pre-treatment melanoma tissue sample obtained before commencing BRAF inhibitor and a matched progressing melanoma metastasis (Prog). Informed consent was obtained for each patient under approved Human Research Ethics Committee protocols.

BRAF sequence analysis

Sequences were amplified using Taq polymerase (Fisher Biotech) using primers shown in Additional file 3: Table S1. PCR products were purified using QIAquick PCR purification kit (Qiagen, Limburg, Netherlands) followed by Sanger sequencing on the 3730xl DNA Analyser (AGRF, Westmead, NSW, Australia). The Human Splicing Finder system [12] was used to identify and analyse BRAF splice sites and branch point sites and auxiliary splicing enhancer sequences.

Melanoma cell lines

The short-term patient-derived dabrafenib-resistant melanoma cells, WMD009 and SMU027, as well as SKMel28 parental and the dabrafenib-resistant BR4 cell line [7] were grown in in Dulbecco’s Modified Eagle Medium (DMEM) with 10% FBS and glutamine (Gibco-BRL).