Background and aims: KRAS and BRAF mutations occur in colorectal cancers (CRCs) and are considered mutually exclusive methods of activating the RAS/RAF/MEK/ERK pathway. This pathway is a therapeutic target and KRAS mutation may predict tumour responsiveness. The purpose of this study was to investigate the relationship between KRAS and BRAF mutations in 24 CRC cell lines and 29 advanced CRCs.
Methods: KRAS and BRAF mutations were detected using high resolution melting and sequencing. Expression of mutations was confirmed by reverse transcription- PCR (RT-PCR) and sequencing. CpG island methylator phenotype (CIMP) was tested by methylation-specific PCR.
Results: KRAS or BRAF mutation occurred in 79% of cell lines and 59% of CRCs. In the cell lines, KRAS mutations occurred in 54% of cases (with 62% in codons 12/13 and 38% in other codons). Four cell lines had a homozygous mutation. Only heterozygous BRAF mutations were detected in 29% cell lines. The V600E mutation occurred most commonly and was associated with CIMP+ status (p = 0.005). Mutations at codons 529 and 581 were also found and, in one case, BRAF and KRAS mutation co-occurred. Unexpectedly, BRAF splice variants (with a predicted kinase-dead protein) were found in 5/24 (21%) cell lines. In advanced CRCs, KRAS mutations occurred in 48% of cases (64% codons 12/13, 36% other codons) and BRAF mutations in 10% (66% V600E, 33% exon 11). A compound KRAS/BRAF mutation was not seen.
Conclusions: Disrupted Ras/Raf signalling is common in CRC. Homozygous KRAS mutations and concomitant KRAS/BRAF mutations may be indicative of a gene dosage effect. The significance of BRAF splice variants is uncertain but may represent another layer of complexity. Finally, if KRAS mutation is to be used for predictive testing, then the whole gene may need to be screened as mutations occur outside codons 12/13.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
The RAS and RAF family of genes code for proteins which form part of the Ras/Raf/MEK/ERK signalling cascade within cells.1 This cascade is involved in the transmission of extracellular signals which control a variety of biological processes such as cell growth, cell survival and migration.2 Disruption of this pathway, through gain-of-function mutations in RAS and RAF family members, is well described in several different types of cancer. In colorectal cancers (CRCs), mutations are most frequently found in the KRAS and BRAF genes.3 4
The Ras protein family (p21ras) comprises K-Ras, H-Ras and N-Ras. These are small GTPase proteins which are normally present in an inactive GDP-bound form. These proteins can be activated by extracellular signals (eg, through growth factor receptors) which result in an exchange of GDP for GTP. The active GTP–p21ras complex is then able to transmit the signal through interaction with Raf proteins. In mammals, there are three highly conserved Raf proteins (A-Raf, B-Raf and C-Raf). These have a serine/threonine kinase domain and, once activated by Ras proteins, they phosphorylate MEK (MAPK/ERK kinase) which in turn phosphorylates ERK (extracellular signal-regulated kinase). The signal is then further transmitted through phosphorylation of downstream targets (such as nuclear factor (NF)-κB).1
Ras family gene mutations occur in 15–30% of all cancers,5 and transcriptome analysis of the NCI-60 panel of cell lines has shown that this is the most commonly deregulated signalling pathway.6 In CRC, KRAS mutations are early events7 8 and are found in 30–50% of tumours.3 Mutation “hotspots” have been identified in codons 12, 13, 61 and, more recently, 146. The hotspots are evolutionarily conserved sites, and mutation results either in a protein which is irreversibly bound to GTP (for codons 12, 13 and 61)9 or in a protein which rapidly exchanges GDP for GTP leading to enhanced signalling activity (for codon 146).10 In both cases the result is a constitutively active protein.
Approximately 7% of all cancers have been shown to have mutations in BRAF, with A-RAF or C-RAF mutations occurring rarely.11 12 The frequency of reported mutations in CRC is between 5% and 22% and they all are gain-of-function mutations causing activation of the kinase domain.11 Mutations appear to be clustered in exons 11 and 15,13 with >90% of all reported mutations occurring at codon 600, usually in the form of a T→A missense transversion at nucleotide 1796 which causes a valine to glutamate (V600E) substitution.11 14 The reason for the high frequency of V600E mutation is not known, although it has been reported that this specific change has ∼500-fold greater kinase activity than the wild-type Braf and has greater intrinsic kinase activity than the other mutations.15 Also this change has been associated with enhanced methylation of the MLH1 promoter. Thus selection for highly active mutations may explain the bias towards the V600E16 change in BRAF, and the effect on promoter methylation may explain the described association with microsatellite instability.17
Although KRAS- and BRAF-encoded proteins are mutated in many of the same types of cancer, concomitant mutations are extremely rare.11 14 18 Intuitively this is because both genes undergo gain-of-function mutations and thus represent different mechanisms of activating the same pathway. More recently, however, the most active (V600E) BRAF mutation has been described together with a KRAS mutation in advanced CRCs and their lymph node metastases.19 There is some doubt, however, as to whether these mutations are occurring in the same cells.20 Kras protein is a target for farnesyltransferase inhibitors, and KRAS mutations are regarded as predictive of tumour response to treatments such as Cetuximab.21 22 23 Downstream mutations in the Ras/Raf pathway would, however, render such therapies inactive. The above considerations prompted us to examine a series of CRC cell lines and advanced CRCs in order to define more precisely the relationship between these genes.
Materials and methods
Cell lines and preparation of RNA, cDNA and DNA
Twenty-four-well established CRC cell lines were kindly donated by Sir Walter Bodmer and Professor Ian Tomlinson, and were fingerprinted for microsatellite instability and TP53 mutations to confirm identity. The cell lines were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Invitrogen, Paisley, UK) supplemented with 10% fetal calf serum (Invitrogen) and penicillin/streptomycin (Invitrogen) in 5% CO2 in a humidified atmosphere. Cells were grown to 90% confluence before extraction of nucleic acid. RNA and DNA were extracted from cells by using the QIAamp DNA Mini Kit (Qiagen, Crawley, UK) and the RNEasy Mini Kit (Qiagen), respectively, following the manufacturer’s protocols. cDNA was synthesised from 1 μg of denatured RNA by incubation at 37°C for 60 min in a mixture containing 1 μM pDN6, 200 U of M-MLV reverse transcriptase (Invitrogen), 10 mM dithiothreitol (DTT; Invitrogen) and 0.5 mM each dNTP.
In order to test the published observation of co-occurring KRAS/BRAF mutations in advanced CRCs, a series of 37 cases of advanced CRC (with known liver metastasis) were selected from the archives of the Pathology Department of Nottingham University Hospital Trust. Ethics approval was obtained for the use of anonymous patient materials in this study (REC reference C02.310). Histology slides were reviewed to confirm the presence of tumour, and 6×10 μm sections were cut into microfuge tubes. After overnight digestion with proteinase K (>600 mAU/ml) genomic DNA was extracted using the QIAamp DNA FFPE tissue kit (Qiagen) following the manufacturer’s protocol. The eluted DNA concentration and purity were measured using a Nanodrop spectrophotometer and the DNA was diluted to 20 ng/μl.
PCR and reverse transcription-PCR
For PCR, 50 ng of genomic DNA or 100 ng of cDNA was used as template. All PCRs were carried out in a final volume of 25 μl containing 1× PCR mastermix (Stratagene, Stockport, UK) and primers at a final concentration of 250 nM. PCR products were resolved on 2% agarose gels following 35 cycles of PCR.
In order to examine KRAS gene mutations, primers were designed for exons 2, 3 and 4 (Ensembl accession no. ENSG00000133703) containing codons 12, 13, 61 and 146) using Primer 3 software (http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi). Analysis of expressed mutations was undertaken by designing primers to amplify fragments spanning exons 2–4. In order to examine BRAF gene mutations, previously published primers for exons 11 and 13–15 (spanning the kinase domain) were used.11 In order to examine for expressed mutations, primers were designed to amplify a fragment spanning exon 9–18. The primer sequences are available from the authors.
High resolution melting curve analysis
Screening for mutations was undertaken on PCR products using the technique of high resolution melting curve (HRM) analysis.24 This method detects changes in DNA melting and depends on the formation of DNA heteroduplexes when a mutation is present. PCR products were mixed with the fluorescent DNA-binding dye Syto 9 (Invitrogen) and then analysed on a HR-1 high resolution machine (Idaho Technology Inc, Salt Lake City, Utah, USA). The PCR products were loaded into capillaries and heated to 90°C at 20°C/s and then cooled to 40°C at 20°C/s to favour heteroduplex formation. The samples were re-heated with the temperature being increased at 0.3°C/s, and fluorescence being measured between 65 and 95°C. The melting curves generated were analysed by custom software with normalised, temperature-shifted curves displayed as difference plots. Since this method depends on heteroduplex formation, it is possible that homozygous mutations would be missed as they would form homoduplexes only. For this purpose, cell line samples which were negative on initial screening were spiked with other negative samples and re-analysed. CRC samples were not spiked as the presence of stromal cells would allow formation of heteroduplexes.
In samples where a mutation was suspected, the PCR products underwent direct bidirectional sequencing. The PCR product was purified using a QIAquick PCR Purification Kit (Qiagen) according to the manufacturer’s instructions. The products were then directly sequenced using the original PCR primers and the Big Dye Sequencing kit (Applied Biosystems) on an ABI 373A (Applied Biosystems, Warrington, UK) sequencer.
Analysis of the sequencing data was carried out using the BLAST 2 sequences program (http://www.ncbi.nlm.nih.gov/blast/bl2seq/wblast2.cgi) by comparison against wild-type sequence for KRAS (NM_033360) and BRAF (NM_004333) and through examination of the generated chromatograms using the Sequencher program (Genecodes, Ann Arbor, Michigan, USA). Any novel mutations found were further examined for evolutionary conservation by comparing paralogues and orthologues using Clustal X and the Boxshade server.25
Mutations in BRAF are associated with sporadic onset microsatellite instability (MSI) and with the CpG island methylator (CIMP) phenotype17 whilst being excluded from herditary non-polyposis colorectal cancer (HNPCC)-associated tumours.26 Sporadic onset MSI is usually due to hypermethylation of the MLH1 promoter, and the V600E BRAF mutation has been shown to stimulate promoter methylation when forcibly expressed in colonic cells.16 We tested this association in the cell lines by methylation-specific PCR (MS-PCR) using primers designed to detect promoter methylation in five genes (RUNX3, NEUROG1, SOCS1, CACNA1G and IGF2). These genes constitute a panel which is recommended for the detection of CIMP.17
Bisulfite conversion of 500 ng of genomic DNA from each samples was carried out using the EZ-DNA Methylation-Gold kit (Zymo Research, Orange, California, USA) in accordance with the manufacturer’s instructions. Template preparation resulted in an average loss of 40% of DNA together with a reduction in the quality of the DNA. Modified template DNA was diluted 1:50 and 5 μl of this was used as template for MS-PCR. These reactions were carried out in a final volume of 25 μl using a QuantiTect SYBR Green PCR Master Mix (Qiagen) and primers at a final concentration of 200 nM. MS-PCR was performed on an Mx30005P real-time thermal cycler (Stratagene) using the following cycling conditions: 95°C for 10 min, 40 cycles of 95°C for 30 s, 54°C for 30 s and 72°C for 60 s, with a final extension step of 72°C for 10 min. PCR products were represented by fluorescent peaks with a melting pattern specific for each product. Cell lines were designated CIMP+ if promoter methylation was detected at ⩾3 loci.
KRAS mutation in cell lines
Primers were designed to amplify genomic and cDNA fragments containing the common hotspots of codons 12, 13 and 61. More recently another mutational hotspot at codon 146 has been described27 and this was thus also included in our screen. The PCR products underwent mutation screening by HRM analysis (fig 1a) followed by sequencing. Mutations were identified in 13/24 (54%) cell lines (table 1) and all mutations identified in the DNA were confirmed as being expressed.
The data generated verified the previously reported mutations in COLO205, DLD1, GP2D, HCT116, HT29, HT55, LOVO, RKO, SW1116, SW480, SW620 and SW948.11 13 Four cell lines whose status had not previously been reported also carried mutations. These included mutations in the common hotspots in C84 (G12A) and VACO10MS (Q61R), as well as mutations in the newly described hotspot in SW1222 (A146V) and C80 (A146V). In addition, a novel missense mutation in codon 117 in the cell line C125 (K117N) was detected. This residue has not been previously described as a site of KRAS mutation although it is part of the GTP-binding domain of KRAS and is highly conserved amongst Ras proteins and across species (fig 2).
Mutations of KRAS are gain-of-function mutations and thus it was of interest that four of the 13 mutations (30%) were present as homozygous mutations. The cell lines C84 (G12A), SW1222 (A146V), SW480 (G12V) and SW620 (G12V) all appeared to have homozygous mutations at the known hotspots. The significance and mechanism by which this occurs (ie, allelic loss or gene conversion) remain to be elucidated although it does imply some effect of gene dosage during tumourigenesis.
It is thought that most KRAS mutations occur in codons 12/13 although, in the cell lines, 5/13 (38%) occurred outside this hotspot.
BRAF mutation in cell lines
Given that Braf has protein kinase activity, we tested the cell lines for mutations occurring within the kinase domain (encoded by exons 11–15). All cell lines were examined and BRAF mutations were found in 7/24 (29%) cell lines. All mutations in BRAF were heterozygous and the most common mutation was the well described V600E mutation which was found in 5/24 (21%) of the cell lines. The two remaining mutations occurred at residues which have not previously been shown to be mutated—GP2D had a T529A mutations and HT55 had a N581Y mutation (fig 1b). Both of these mutations are situated in exon 13 and cause a change in conserved residues within the kinase domain of BRAF (fig 2). In order to confirm that these mutations were expressed, primers amplifying the coding sequence between exons 11 and 18 were designed. PCR and sequencing confirmed the presence of mutations in the mRNA.
Concomitant mutations of KRAS and BRAF
In total, 19/24 (79%) cell lines carried mutations in either KRAS or BRAF. One of the cell lines, GP2D, however contained mutations in both of these genes. The KRAS mutation was in the codon 12/13 hotspot and was present in heterozygous form. The BRAF mutation occurred in codon 529 which is outside the classical codon 600 hotspot. Since this study was conducted on cell lines, it represents the first confirmation that KRAS and BRAF mutations do occur in the same cells. It is therefore reasonable to presume that when dual mutations are described in tumours, they are probably both present in the tumour cells. Our data, albeit limited, would support a large study (of several tumour types) which reported the co-existence of KRAS and BRAF mutations but that the V600E mutation was never present in tumours with double mutations.28 In contrast, however, in a study of advanced CRCs, Oliviera et al did find that V600E BRAF mutation could co-exist with KRAS mutations.19
Association of KRAS and BRAF mutations with CIMP+ status
Our analysis of the cell lines using the five genes which constitute the CIMP panel identified 10/24 (41%) cell lines as being CIMP+—that is, promoter methylation in ⩾3 genes (fig 3). Comparisons of the mutational status of KRAS in the cell lines with CIMP status or MSI29 did not reveal any association with either of these parameters.
Comparisons of the mutational status of BRAF in the cell lines with MSI status found no significant association. However, comparison with CIMP status found a very tight association between the V600E mutation and CIMP+ status (table 2, p = 0.005, Fisher’s exact test). This confirms previously reported data from colorectal tumours linking CIMP+ status with this BRAF mutation.17
Splice variants of BRAF
During the testing for expression of the BRAF mutations, all PCR products were resolved on an agarose gel prior to sequencing. Some of the cell lines demonstrated bands which were smaller than the expected size, and sequencing indicated that these represented splice variants from which exons 14 and 15 had been excised (fig 1c–e). In order to confirm this, a new pair of PCR primers was designed which covered exons 13–17 and the experiments were repeated. Repeat examination confirmed the splice variants in COLO201, COLO320DM, DLD1, LOVO and SW480. Of these, COLO201 has a BRAF mutation (V600E), and detailed examination of the sequence confirmed expression of the mutant sequence thereby suggesting that the splice variants arise from the wild-type sequence.
KRAS/BRAF mutation in advanced CRCs
Since one study has previously described concomitant KRAS and BRAF mutations in metastatic tumours, we sought to validate these findings and our findings in the cell lines in a series of advanced CRCs. HRM was performed to screen for mutations, and samples with aberrant melting were validated by direct sequencing. Sequence-proven KRAS and BRAF mutations were identified in 14/29 (49%) and 3/29 (10%) cases respectively (table 3). There were three cases (one in KRAS and two in BRAF) in which consistent aberrant melting was seen but a corresponding mutation could not be identified. HRM can detect mutant alleles at a frequency of 2.5% within a wild-type population and is therefore an order of magnitude more sensitive than direct sequencing (data available from the corresponding author). Given this, the three cases may represent false-negative sequencing and—if added to the total—would give a total mutation frequency of 69%. However, as sequencing is currently the gold standard, they were excluded from the final analysis; thus, in total, 17/29 (59%) cases had disruption of the Ras/Raf signalling pathway.
Analysis of the frequency of mutation at the specific hotspots in KRAS reflected the findings in the cell lines. Of the total, 9/14 (64%) occurred in the codon 12/13 hotspot whilst 5/14 (36%) were in codon 61/146. In the case of BRAF, the mutation frequency was lower than would be expected and there were two mutations in the exon 15 hotspot (2/3, 66%) as compared with outside this hotspot (1/3, 33%).
Frequent disruption of the Ras/Raf pathway in CRC
This study has systematically examined KRAS and BRAF mutation in a series of CRC cell lines and advanced CRCs, and our two data sets showed broadly similar findings. First, we found that there is a wide spectrum of mutations disrupting the Ras/Raf pathway and that this pathway is disrupted in 79% of the cell lines and 59% of the CRCs. Mutations in KRAS were present in 54 and 55%, respectively, of the cell lines and tumours— this is higher than the frequency of ∼30% of CRCs reported in other studies.30 We believe that the frequency of KRAS mutation in CRCs is probably under-reported in the general literature since (1) many studies focus on codons 12/13 and (2) they often use a direct PCR sequencing strategy which may miss heterozygous mutations especially if there is infiltration of tumours by stromal cells. Our study confirmed previously described mutations of KRAS in some of these cell lines and, in addition, this study is the first report of mutation at codon 117 of KRAS in the C125 cell line. When a sequence variation is discovered, there is always some doubt as to whether it represents a true mutation or a low frequency polymorphism. We believe that this is a true oncogenic event since (1) this amino acid is within the GTP-binding domain of Kras and is conserved across species and within the Ras family and (2) this residue has been mutated in vitro and has been classed as a transforming mutation.9
Mutations of BRAF were found in 29% of the cell lines, which is also higher than levels found in previous studies.11 The majority of these were of the V600E type. The data confirmed some of the previous studies in CRC lines but also reported, for the first time, the presence of V600E mutations of BRAF in COLO201 and VACO5. In addition, this study reports novel BRAF mutations—that is, N581Y mutation in HT55 and T529A in GP2D (table 1). These codons are outside the classically reported BRAF mutation hotspots of exons 11 and 15 and, as with the novel KRAS mutations, the question arises of whether they are functionally contributing to tumourigenesis. Both N581 and T529 are highly conserved residues across evolution and within the Raf family of genes (fig 2). Codon 581 has been found to be mutated in melanoma,31 but mutation of codon 529 has not been previously reported.
The very strong association between the V600E mutation and CIMP+ status found in this study confirms data from other studies. Forced expression of the V600E mutation has been reported to induce MLH1 promoter methylation in a “normal colon” cell line. Other promoter loci that were tested in this study were found to be fully methylated in the parental cell line and thus the effect of V600E on these genes could not be tested. It is possible that this BRAF mutation affects several loci and, if so, it provides a possible mechanistic link to explain the association with CIMP+ status.
The frequency of BRAF mutations in the tumours was 10% and was lower than may be expected. This may be related to the advanced nature of the tumours as there is evidence that tumours with MSI tend to be of less advanced stage.32
The role of gene dosage and splice variants in the Ras/Raf pathway
This study describes disruption of the Ras/Raf pathway in a total of 79% of the cell lines studied, although in five (26%) of these cell lines the disruption was characterised by two gain-of-function mutations. In four cases there was homozygous KRAS mutation and one cell line contained synchronous heterozygous KRAS/BRAF mutations. Since this study was undertaken in cell lines, this is unequivocal evidence that both mutations are in the same cell. The data suggest that gene dosage may play a part in the disruption of this pathway. The presence of KRAS mutation in histologically normal colonic mucosa has been described33 and indicates that, in certain situations, the presence of a single KRAS mutation may not have a significant impact on the cell behaviour. Furthermore, other studies have suggested that mutations of BRAF may have differing levels of transforming activity in cells (with the V600E change reported to have the greatest effect) and thus all mutations in BRAF cannot be seen as equivalent. Elucidating the role of gene dosage and the nature of the interactions of specific BRAF and KRAS mutations is beyond the scope of this study but needs further investigation.
An unexpected discovery during this study was the identification, for the first time, of splice variants in human BRAF (fig 1). Murine Braf has been shown to undergo alternative splicing but this is of additional exons 8b and 9b.34 Five of the cell lines examined (COLO201, COLO320DM, DLD1, LOVO and SW480) expressed a short mRNA transcript from which exons 14 and 15 had been spliced out. Of these, only COLO201 had a BRAF mutation, and analysis of the sequence data showed that the full-length mutant sequence was expressed and that the spliced sequence was transcribed from the wild-type allele. The presence of these splice variants is, at first glance, completely counterintuitive as exons 14 and 15 code for part of the kinase domain. The activation segment of this domain is resident in exon 15, and the alternatively spliced transcripts would be predicted to produce truncated kinase-dead proteins. Analysis of the DNA did not identify any splice site mutations in any of these cell lines and, at this stage, we are uncertain about the mechanistic basis of these splice variants. Even more perplexing is the functional role that the truncated protein may have since loss of kinase activity, even from wild-type alleles, would be expected to be give a selective disadvantage. Cancer-derived kinase-dead BRAF mutations have been described15 and one speculative explanation is that the truncated protein has dominant-negative activity on an inhibitory part of the Ras/Raf pathway.
Clinical implications of KRAS/BRAF mutation
Anti-EGFR (epidermal growth factor receptor) chemotherapies have been shown to be very successful in some cases of advanced CRC. Resistance to anti-EGFR treatment has been largely attributed to KRAS,21 22 23 and it is likely that mutation analysis of KRAS will soon become an obligatory predictive test before commencing anti-EGFR treatment. We found that an unexpectedly high proportion of mutations (36% of tumours and 38% of cell lines) occurred outside the classical hotspot of codons 12/13. Given that most of the data associating KRAS mutation with resistance to anti-EGFR treatment have been derived from trials in which only codons 12/13 have been examined, it is recommended that only this site be tested.35 Our view is that, in the clinical setting, the whole of the KRAS gene should be examined since (1) significant numbers of mutations occur in codons 61/146 and (2) all mutations result in constitutively active Kras protein. Given that there are alternatives to anti-EGFR treatments, it would be better to use one of these than try anti-EGFR treatment when there is a chance that it may not work. This would reduce the number of patients eligible for treatment to 20–30% but would give a higher probability of response.
Our cell line data show that there is true co-occurrence of KRAS/BRAF mutations and this needs to be borne in mind both when considering use of anti-EGFR chemotherapies and of specific treatment targeting Kras protein.3 The presence of a gain-of-function mutation downstream of the therapeutic target will render that treatment ineffective. Thus, our data suggest that, when making decisions about costly and toxic treatments, both genes should be thoroughly examined. This view may soon be extended to include screening for PIK3CA mutations as these may also contribute to resistance to anti-EGFR treatment.36
In summary, this study has shown that the Ras/Raf signalling pathway is widely disrupted in CRC cell lines, and novel mutations at highly conserved residues in KRAS and BRAF have been described. Our data confirm that KRAS and BRAF mutations do occur in the same cell and that BRAF V600E mutation is associated with CIMP+ status. The BRAF splice variants show that disruption of this pathway is complex and the presence of homozygous and compound mutations suggests there may be gene dosage effects. All these factors may become considerations when designing predictive tests for therapeutic decision making.
Funding This work was funded by the University of Nottingham.
Competing interests None.
Ethics approval Ethics approval was obtained for the use of anonymous patient materials in this study (REC reference C02.310).