• colorectal carcinomas;
  • colorectal neoplasms;
  • hereditary nonpolyposis;
  • DNA repair


  1. Top of page
  2. Abstract
  6. Acknowledgements


Alterations of BRAF have been implicated in the carcinogenesis of colorectal tumors with microsatellite instability (MSI). These alterations were attributed to defective DNA mismatch repair, which underlies MSI. It was the objective of this study to clarify the role of BRAF in colorectal carcinoma with MSI.


After sequencing for BRAF and KRAS in 82 colorectal tumor samples with or without MSI, mismatch repair protein expression was analyzed by immunohistochemistry, and promoter methylation of hMLH1 was analyzed with a methylation-specific polymerase chain reaction. Results were correlated with the germline status of hMLH1 or hMSH2 and clinical characteristics.


BRAF was mutated more often in tumors with MSI than in tumors without MSI (27% vs. 5%; P = 0.016). The most prevalent BRAF alteration, V599E, occurred only in tumors with MSI. BRAF V599E was associated with more frequent hMLH1 promoter methylation (P = 0.07) and loss of hMLH1 (P = 0.02). The median age of patients with BRAF V599E was older compared with the median age of patients without this mutation (P = 0.001; 78 vs. 49 yrs). No BRAF alterations occurred in patients with germline mutations of mismatch repair genes. Five novel BRAF mutations were identified.


Although BRAF V599E was common in colorectal carcinomas with MSI, it was not a consequence of deficient mismatch repair. The current data showed instead that the BRAF V599E mutation was associated only with a subgroup of colorectal carcinomas with MSI that were obtained from older patients without hereditary nonpolyposis colorectal carcinoma and showed epigenetic silencing of hMLH1. These results indicated that tumors with MSI caused by epigenetic MLH1 silencing have a distinct mutational background from that of tumors with genetic loss of mismatch repair, suggesting that there are two genetically distinct entities of microsatellite-instable tumors. Cancer 2005. © 2005 American Cancer Society.

It is known that activating alterations of the protooncogene KRAS are crucial for the development of several human malignancies. In colorectal carcinoma, KRAS is mutated in approximately 50% of all patients, with hot spots in codons 12, 13, 59, and 61.1KRAS is a key component of the mitogen-activated protein (MAP) kinase (MAPK) pathway, which is a cascade of kinases mediating cellular response to a variety of extracellular signals.2 Receptor tyrosine kinases in the cellular membrane are activated by linkage of growth signals, hormones, and adhesion factors, leading to a stimulation of the membrane-bound G-protein RAS.3 RAS activates its downstream effector RAF, followed by phosphorylation and, thus, activation of MAPK-extracellular-regulated kinase (ERK) kinase (MEK), which, in turn, activates ERK in the same manner. This network is able to amplify and modulate signals according to their amplitude and kinetic.

Based on findings of the Human Cancer Genome Project, Davies et al. investigated the role of BRAF, a downstream actor of KRAS in the MAPK pathway, in different malignancies. Those authors recently reported that BRAF mutations play a key role in the oncogenesis of melanoma and colorectal carcinoma, although no detailed association with different subtypes of colorectal carcinoma was investigated.4 Carcinomas of the breast, stomach, ovary, and lung as well as sarcomas and non-Hodgkin lymphomas also exhibited BRAF alterations.4–7 The majority of the analyzed samples exhibited one specific BRAF alteration that caused the exchange of valine to glutamate, a negatively charged amino acid, at position 599 (V599E) in the kinase domain of the protein. Moreover, it has been shown that this alteration causes the growth of carcinoma cell lines independent of RAS activation.4 The P-loop of BRAF, which is engaged in the regulation of the kinases' activity, is the second target for alterations in tumor cells. To date, all reported BRAF alterations affect exons 11 (encoding the P-loop) and 15 (encoding the kinase domain).4, 8

Approximately 15% of all colorectal carcinomas exhibit a special type of genetic instability called microsatellite instability (MSI), which is caused by genetic or epigenetic inactivation of the mismatch repair (MMR) system.9–11 The MMR system corrects DNA polymerase errors, which preferentially arise during the replication of repetitive DNA sequences due to slippage of the template against the newly synthesized strand (for review see Modrich12). The phenotypical and clinical features of MSI tumors are attributed to their unique mutational spectrum, especially frameshifting events, which arise due to length alterations of repeat sequences in the coding regions of genes that affect carcinogenesis.13–15 MSI instability also is the hallmark of hereditary nonpolyposis colorectal carcinoma (HNPCC) (Online Mendelian Inheritance of Man Data Base no. 11450016, 17), a hereditary carcinoma predisposition syndrome that is caused by heterozygous germline mutations in the MMR genes.

Recent data have shown that MSI in colorectal carcinoma is associated with the V599E mutation in BRAF, whereas MMR-proficient tumor cells rarely exhibit BRAF alterations and, instead, show a higher rate of mutated KRAS.18 In that study, however, the underlying reason for the increase in the BRAF mutation rate or its association to different subtypes of MMR-deficient tumors was not assessed. It was suggested that the V599E alteration in BRAF may be an immediate result of the repair deficiency and that mutations of either BRAF or KRAS have a similar effect on disease progression.

It was the objective of the current study to investigate the correlation between the MMR status and the protein components of the MAPK pathway (KRAS or BRAF) that are targeted for mutational activation in colorectal carcinoma. We screened 45 colorectal carcinomas with MSI and 37 colorectal tumors without MSI, but with similar clinical characteristics for alterations in the hot-spot regions of BRAF and KRAS. Furthermore, we investigated the methylation status of the hMLH1 promoter, which is prone to transcriptional silencing by CpG island methylation, in 2 separate regions. Expression of the MMR genes hMLH1 and hMSH2 was analyzed by immunohistochemistry. In addition, we assessed the phenotypical features and clinical outcomes of patients who had colorectal carcinoma with and without BRAF alterations.


  1. Top of page
  2. Abstract
  6. Acknowledgements

Patients, Tumor Samples, and Detection of MSI

Forty-five colorectal tumor samples with MSI from 44 patients were assessed. All patients had histologically verified adenocarcinoma and had been referred for treatment evaluation or genetic counseling for suspected hereditary colorectal carcinoma. For controls, we assessed 37 tumor samples from patients who had colorectal carcinoma without MSI but with similar International Union Against Cancer (UICC) tumor stages at the time of diagnosis. For analysis of MSI, DNA was extracted from colorectal carcinoma tissue and corresponding normal colonic epithelium after microdissection. MSI status was determined as described previously.19

Twenty-two of 44 patients were tested for HNPCC-associated germline alterations in hMSH2 and hMLH1 as reported in part in our previous study.19 Twelve of 13 detected germline alterations could be classified as pathogenic for HNPCC, because they caused nonsense-codons (5 alterations), gross alterations (1 alteration), frameshifting events (2 alterations), or missense alterations (4 alterations) that have been reported previously as pathogenic in at least 2 patients by the International Collaborative Group on Hereditary Non-Polyposis Colorectal Cancer (available at

DNA Extraction, Amplification, and Sequencing

Paraffin-embedded tissue samples were lysed in 20 μL of a 1% Triton X-100 solution at 95 °C for 20 minutes and were used directly for polymerase chain reaction (PCR) analysis. If this procedure did not yield PCR products, then DNA was extracted from microdissected samples as follows: The sample was placed in 100 μL Buffer P (50 mM Tris-HCl, pH 8.5; 100 mM NaCl; 1 mM ethylenediamine tetraacetic acid [EDTA]; 0.5% Tween-20; 0.5% NP40; 20 mM dithiothreitol) and incubated at 90 °C for 15 minutes. After cooling to 60 °C, proteinase K (1 μL 20 mg/mL) was added, overlaid with mineral oil, and incubated at 60 °C overnight. Proteinase K was denatured by a 95 °C incubation for 5 minutes. Mineral oil was removed, and the sample was centrifuged at × 13,000 g for 10 minutes. Then, 75 μL of the supernatant were transferred to a new tube, and DNA was purified by ethanol precipitation. BRAF analysis only evaluated mutations in exons 11 and 15, which encode the mutational hot-spot regions of the P-loop and the activation segment, because no mutations have been found in other regions of the gene.4, 18KRAS analysis targeted exons 2 and 3, which encode the domains that are mutated in human carcinoma. PCRs of exons 2 and 3 of KRAS and exons 11 and 15 of BRAF were performed with primers that were described previously4 along with 2 units AmpliTaq Gold Polymerase (Applied Biosystems, Darmstadt, Germany) per reaction. Cycling conditions were as follows: 95 °C for 10 minutes; then 35 cycles at 95 °C for 1 minute, 55 °C for 30 seconds, and 72 °C for 2 minutes; and 72 °C for 10 minutes. Some DNA samples did not yield PCR products of exon 3 of KRAS and or of exon 11 of BRAF. In these samples, nested PCRs were performed. KRAS exon 3 first was amplified using the above-mentioned primers with 50 cycles; then, 0.5 μL of the PCR product was reamplified using the following primers: 5′-AAGGTGCACTGTAATAATCCAGAC-3′ (forward) and 5′-CCTTAATGTCAGCTT ATTATATTC-3′ (reverse). Similarly, BRAF exon 11 first was amplified as described above and, after dilution, was reamplified using the following primers: 5′-TTTCTTTTTCTGTTTGGCTTG-3′ (forward) and 5′-CGAACAGTGATATTTCCTTTGAT-3′ (reverse). All PCR products were sequenced using BigDye version 1.1 chemistry on an ABI 3100 sequencer (Applied Biosystems).

Methylation Analysis of the hMLH1 Promoter

Methylation of CpG islands in the promoter of hMLH1 was assessed by methylation-specific PCR after bisulfite conversion.20 DNA was extracted and ethanol was purified from 25 mm2 tumor tissue as described above. Bisulfite modification was performed as described previously.20 Methylation was assessed in 2 different regions of the hMLH1 promoter. Primer sequences and reaction conditions for the methylated and unmethylated amplification of Region A21 and Region C22 have been published previously, and PCR conditions also were used according to those publications. DNA from the cell line 293T (hMLH1 promoter methylated)23 and from peripheral blood mononuclear cells of a healthy young individual (hMLH1 promoter unmethylated) served as positive and negative controls, respectively.

Immunohistochemical Analysis

Sections (2 μm) of representative samples were cut from paraffin embedded colorectal carcinoma specimens. Normal colon mucosa served as an internal control. Sections were dewaxed through graded alcohols, and this was followed by a 30-minute incubation with 3% H2O2 for blocking endogenous peroxidase. Antigen retrieval by heating was required in a microwave oven for 20 minutes with EDTA buffer, pH 8.0. Before and between different incubation steps, the sections were washed with phosphate buffered saline, pH 7.4. Primary antibodies were diluted in phosphate buffered saline, pH 7.4, with 1% bovine serum albumin. For the detection of hMLH1 (BD PharMingen; mouse monoclonal antibody clone G168-728; dilution 1:750) and hMSH2 (Oncogene; mouse monoclonal antibody clone GB12; dilution 1:25), sections were incubated with the primary antibodies overnight at 4 °C followed by application of the EnVision-system (DakoCytomation) with horseradish peroxidase as enzyme and 3,3′diaminobenzidine tetrahydrochloride as chromogen. The sections were counterstained with Gill hematoxylin.


Differences between the groups were evaluated using the Fisher exact test and the Mann–Whitney U test. Survival analysis was performed using a Kaplan–Meier analysis. All reported P values were 2-sided, and P values < 0.05 were considered statistically significant.


  1. Top of page
  2. Abstract
  6. Acknowledgements

Mutations in KRAS and BRAF

Forty-five colorectal adenocarcinomas with MSI, including 13 tumors from patients with HNPCC-associated hMLH1 or hMSH2 germline alterations, and 37 colorectal carcinomas without MSI (control group) were screened for mutations of BRAF and KRAS. BRAF missense alterations were identified in 12 of 45 tumors (27%) with MSI (Table 1, Fig. 1). Seven of those 12 alterations (58%; 16% of all tumors with MSI) caused the previously described amino-acid exchange from glutamate to valine at residue 599 (V599E), which reportedly accounts for up to 90% of BRAF mutations in colorectal carcinoma.4, 18, 24 Furthermore, we identified four previously undescribed BRAF alterations in five MSI samples. These were S606Y and L587F (both heterozygous), S466L (homozygous), all in 1 patient each, and R602Q in 2 patients (heterozygous and homozygous) (Fig. 1). Normal tissues from these five patients were studied and showed wild type BRAF in the germline, suggesting that these were not polymorphisms.

Table 1. Clinical and Mutational Data in Colorectal Carcinomas with Microsatellite Instabilitya
Patient no.GenderAge (yrs)pTNM (grade)LocalizationUICC stageBRAF mutationKRAS mutationhMLH1 promoter methylationImmunohistochemistryHNPCC mutation
Region ARegion ChMLH1hMSH2
  • pTNM: pathologic tumor, lymph node, and metastasis status; UICC: International Union Against Cancer; HNPCC: hereditary nonpolyposis colorectal carcinoma; M: methylated signal in methylation-specific polymerase chain reaction analysis (MSPCR); U: no methylated signal in MSPCR; ND: not determined due to insufficient material (promoter methylation or immunohistochemistry) or insufficient family history for HNPCC (sequencing for HNPCC mutation); −: negative; +: positive.

  • a

    Forty-five tumors were examined in 44 patients. This table is sorted by age, which refers to age at the time of diagnosis. BRAF mutations are presented in full with the V599E mutation left-aligned and other (novel) mutations right-aligned. For KRAS mutations, only the affected codon is mentioned.

1Female22T3N0 (2)C. ascendensII 12MMNDNDhMSH2 C697F
2Female29T3NM0 (2)C. descendensII 13MM+hMSH2 R389X
3Female30T2N0 (2)RectumI  MM+hMLH1 T117M
4Female32T4N0 (2)C. descendensII  UU+
5Female37T3N0M0 (3)CoecumII  MU++ND
6Female37T3N2M1 (2)CoecumIV  MU+ND
7Male38T3N0M0 (2)RectumII  UU++ND
8Male39T2N2 (2)C. ascendensIII 13MMhMSH2 2345ΔC, hMLH1 L618T
9Male39T3N0Mx (2)C. ascendensII 69NDMNDNDhMLH1 ΔEx1–10
10Male40T2N0M0C. ascendensIIL587F MU++ND
11Female42T3N0Mx (2)C. ascendensII  MU+ND
12Female42T2N0M0 (2)CoecumI  MM+hMLH1 R659P
13Male43T3N0 (2)CoecumII  MMNDND
14Male43T2N0MxSigmaI 12NDU+hMLH1Δ1380GA
15Male45T3N0 (2)C. descendensII  MM++
16Male46T3N1Mx (3)C. transversumIII  MU++hMSH2 Q374X
17Male46T3N2Mx (2)CoecumIIIR602Q13MU++
18Male46T3N0 (2)SigmaII 12, 62UU+ND
19Male49T3N1Mx (3)RectumIII  MU+hMSH2 R621X
20Female50T3N0Mx (2)C. ascendensII  MU++ND
21Male50T3N1 (2)C. ascendensIII  MM+ND
22Female51T3N2M0 (2)C. ascendensIII  UM+hMSH2 G322D
23Female51T3N0 (2)C. descendensII 13UM++ND
24Male51T3N0 (21)RectumIIS606Y UU++hMSH2 G580X
25Male55T3N3 (4)RectumIII  MM++
26Female56T1N0 (2)C. ascendensI 56MU+hMSH2 G674D
27Male56T3N0 (2)SigmaIIS466L MM++ND
28Female56T2N0 (2)C. ascendensIV599E MM+ND
29Male58T3N0 (2)C. ascendensII  MU++ND
30Male60T3N0Mx (3)RectumII  MM++
31Male62T2N0 (21)C. ascendensI 13UU+ND
32Male64T1N0Mx (3)C. ascendensIR602Q MU+hMSH2 R680X
33Female68T2N0Mx (2)C. ascendensI  MM+
34Female69T3N0 (2)C. ascendensII 13UU++ND
35Female73T2N0 (2)C. descendensIV599E MM+ND
36Male74T2N0 (3)C. ascendensI  MM+ND
37Female74T3N1M0 (3)C. ascendensIIIV599E58MMNDND
38Female78T3N2Mx (2)C. ascendensIII  MM+ND
39Male78T4N0Mx (3)C. transversumII  MMND
   T3N0Mx (2)C. ascendensIIV599E MM 
40Female79T3N0 (3)C. ascendensII 58MM+
41Female89T3N2Mx (x)C. ascendensIII  MM+ND
42Female90T3N0 (2)CoecumIIV599E MM+ND
43Male91T3N0 (2)Coecum/C. ascendensIIV599E MM+ND
44Female93T3N0 (2)C. ascendensIIV599E MM+ND
thumbnail image

Figure 1. This overview shows the detected BRAF alterations in their sequence context. The alterations were heterozygous, except for S466L and R602Q. For most previously undescribed alterations (S466L, L587F, S601F, R602Q, and S606Y), analysis of normal tissue verified that patients carried wild type germline BRAF, excluding the possibility that the alterations identified were polymorphisms. Alterations found in tumors with microsatellite instability (MSI) are marked. The electropherograms are annotated with the nucleotide and amino acid sequences, and altered bases are printed in boldface.

Download figure to PowerPoint

No BRAF V599E alteration was present in colorectal carcinoma samples without MSI. However, we found 2 additional, novel, heterozygous BRAF missense alterations (E450K and S601F) (Fig. 1) in the control group of samples without MSI (5%). Analysis of normal tissues from the patient with the S601F alteration exhibited wild type BRAF, whereas a polymorphism could not be excluded for the E450K alteration due to insufficient material. The rate of BRAF mutations was significantly higher in colorectal carcinomas with MSI than without MSI (27% vs. 5%; P = 0.016) (Table 2), whereas KRAS was mutated to a similar extent in both colorectal carcinoma types (13 of 45 tumors with MSI; 29%; 10 of 37 tumors without MSI; 27%) (Tables 1, 2). It is noteworthy that none of the 13 tumors from HNPCC patients with germline hMLH1 or hMSH2 mutations showed the BRAF V599E mutation, whereas 2 tumors carried novel alterations (R602Q and S606Y) (Fig. 1).

Table 2. BRAF and KRAS Mutations in Colorectal Tumors
MutationTumors with MSITumors without MSI
  1. MSI: microsatellite instability.

Total no.4537
BRAF alterations: No. (%)12 (27)2 (5)
V599E7 (16) 
R602Q2 (4) 
S466L1 (2) 
P586L1 (2) 
S606Y1 (2) 
E450K 1 (3)
S601F 1 (3)
KRAS alterations: No. (%)13 (29)10 (27)

Immunohistochemical Staining

Immunohistochemical staining of hMLH1 was performed in colon carcinoma samples with MSI to determine protein expression in relation to methylation and available mutation data. Evaluation of expression was possible for 41 of 45 tumors (Table 1), whereas 4 samples were not assessed due to insufficient quality. Twenty of 41 samples (49%) showed normal hMLH1 expression, whereas hMLH1 protein was lost in 21 tumors (51%). Four of those 21 tumors were from HNPCC patients with hMLH1 germline mutations, suggesting somatic loss of the wild type hMLH1 allele. Another 13 tumors were methylated in both regions of the hMLH1 promoter (see below), suggesting protein loss by transcriptional silencing. In the residual 4 tumors, hMLH1 may have been lost due to somatic mutations. It is noteworthy that hMLH1 was lost in all tumors with BRAF V599E in which hMLH1 status was assessed (6 of 7 tumors). BRAF mutation V599E was associated with loss of hMLH1 protein (P = 0.02).

hMSH2 was absent in nine of 41 samples (22%). Five of those samples were from patients with germline hMSH2 mutations, suggesting that somatic loss of the wild type allele caused this deficiency.

hMLH1 Promoter Methylation

The promoter of hMLH1 is prone to CpG island methylation, which can lead to transcriptional silencing of hMLH1, rendering the cell defective in MMR.21 Thus, along with mutational inactivation of MMR genes, promoter methylation is supposed to contribute to the MMR deficiency of colorectal carcinomas with MSI. Traditionally, hMLH1 methylation was assessed in a distal region of the promoter, which shows a dense population of CpG dinucleotides.21 This area (referred to as Region A in Fig. 2) recently has been correlated poorly with the actual expression of hMLH1, whereas the methylation status in a promoter region more proximate to the translational start (referred to as Region C) reportedly correlated well with a lack of protein expression.25, 26 We assessed methylation in colorectal carcinoma samples with MSI in both regions of the hMLH1 promoter (Table 1) using methylation-specific PCR. For 40 tumor samples, methylation of both regions as well as immunohistochemical information was available. Region A was methylated in 32 of 40 samples (80%), whereas Region C was methylated less frequently (23 of 40 samples; 58%). Methylation of Region C in most samples (21 of 23 samples; 91%) was accompanied by methylation of Region A, suggesting that methylation of Region A precedes that of Region C. Loss of hMLH1 was associated with methylation in Region A in 18 of 20 samples (90%) and with methylation in Region C in 17 of 20 samples (85%). However, Region A also frequently was methylated, even though hMLH1 was expressed (14 of 20 samples; 70%), whereas methylation of Region C was accompanied only occasionally by protein expression (6 of 20 samples; 30%). Therefore, assessing the methylation of Region C instead of Region A yields a more specific and comparably sensitive prediction for expression of hMLH1, confirming previous studies.25, 26 Samples that are negative for hMLH1 staining despite incomplete promoter methylation may be explained by somatic mutational loss of hMLH1.

thumbnail image

Figure 2. hMLH1 promoter methylation is demonstrated in this schematic overview of the hMLH1 promoter (5′-untranslated region) with CpG islands (signified by balls with vertical lines). Promoter Region A (from − 711 to − 577; positions refer to the transcriptional start) and promoter Region C (from − 248 to − 178), in which CpG island methylation was assessed in the current study, are indicated by horizontal brackets. Positive and negative controls of the methylation-specific polymerase chain reaction (PCR) for both promoter regions are shown above the brackets. Blood DNA from a healthy individual was used as reference for unmethylated DNA (Um DNA), and DNA extracted from 293T cells, which have a methylated hMLH1 promoter,23 served as a control for methylated DNA (M DNA). PCR reactions were performed with primers specific for unmethylated (U) or methylated (M) DNA in both regions (s, specific product; ns, nonspecific product).

Download figure to PowerPoint

It is noteworthy that all 7 seven samples with the BRAF V599E mutation were methylated in both regions of the hMLH1 promoter, whereas other BRAF mutants showed no association with methylation. In tumors with MSI, BRAF mutation V599E was correlated with hMLH1 promoter methylation in both regions examined (P = 0.07).

Phenotype of Colorectal Carcinomas with MSI and the BRAF V599E Mutation

All 7 tumors with the BRAF V599E mutation showed hMLH1 promoter methylation, and no hMLH1 was detected by immunohistochemical staining. Loss of hMLH1 expression occurred more often in MSI tumors with BRAF V599E (P = 0.02). Furthermore, patient age at the time of diagnosis was older compared with patients who had MSI tumors without this mutation (P = 0.001; 78 yrs vs. 49 yrs; 95% confidence interval for the difference in the median, 17–49 yrs) (Fig. 3). This also held true BRAF V599E tumors were compared with tumors that had other BRAF alterations (P = 0.009; 95% confidence interval, 11–44 yrs) (Fig. 3). All clinical tumor stages were UICC Stage I–III, and no patients had metastasis. Six of seven tumors were localized in the right colon, whereas one tumor was localized in the left colon. None of the patients with BRAF V599E tumors revealed germline alterations of hMLH1 and hMSH2. One tumor showed a KRAS mutation in exon 3.

thumbnail image

Figure 3. These box plots correlate the age of patients who had tumors with microsatellite instability (MSI) and the BRAF mutation type (wild type, non-V599E, and V599E mutations). The thick horizontal lines indicate the median distance, and the boxes indicate the interquartile distance (values within 25–75% of the total spread). Whiskers embrace values within the 1.5-fold interquartile distance. One value outside the 1.5-fold interquartile distance is indicated by a small circle.

Download figure to PowerPoint

The patients with microsatellite-unstable tumors had a significantly better overall survival (5-yr survival: 94% for patients with MSI tumors and 56% for patients without MSI tumors; P < 0.024), confirming previous findings. No differences in survival were evident for patients who had BRAF mutations compared with patients who did not have alterations in this gene, but the older age of the patients with BRAF V599E (see below) and their relatively low number did not allow a final conclusion.

Phenotype of Colorectal Carcinomas with MSI and other BRAF Mutations

In contrast to tumors with the V599E mutation, all tumors with MSI and other BRAF mutations (n = 5 tumors) showed normal hMLH1 expression. The hMLH1 promoter showed methylation in both regions (S466L) and in only 1 region (L587F, n = 1 tumor; R602Q, n = 2 tumors) or remained unmethylated (S606Y). Two colorectal carcinomas with the R602Q and the S606Y BRAF mutations were from HNPCC patients with known hMSH2 nonsense mutations. Patient age at the time of diagnosis did not differ between patients with these novel BRAF mutations and MSI tumors without BRAF mutation (Fig. 3). Localization of the tumor was not related to a specific localization (ascending colon, n = 2 patients; sigmoid colon, n = 1 patient; rectum, n = 1 patient; cecum, n = 1 patient).


  1. Top of page
  2. Abstract
  6. Acknowledgements

In a recent analysis of BRAF and KRAS in colorectal tumors, MSI was associated with mutated BRAF, which was significantly prevalent in MMR-deficient tumors, whereas KRAS mutations occurred more often in cells with proficient MMR.18 This prompted the hypothesis that BRAF mutations are a consequence of deficient MMR, and that both tumor types, although they progress through the same biologic pathway, have different mutational targets (KRAS or BRAF), depending on the underlying genetic instability.4, 18

In the current study, we performed an extensive comparison of colorectal carcinomas with MSI (n = 45 tumors) and without MSI (n = 37 tumors) and evaluated the BRAF and KRAS mutational status as well as clinical features. In tumors with MSI, the BRAF V599E mutation was associated with hMLH1 inactivation by promoter methylation and loss of the hMLH1 protein. The V599E alteration was not found in colorectal carcinomas from patients with hMLH1 or hMSH2 germline mutations (patients with HNPCC) or in tumors without MSI.

Although the presence of the BRAF V599E mutation correlated well with MSI, a defect in MMR does not seem to underlie this mutation for three reasons: First, no HNPCC-related colorectal tumor exhibits this mutation. Second, V599 is not encoded within a repetitive DNA sequence context that would make it a likely substrate for repair (Fig. 1), whereas alterations of other tumor-promoting genes (e.g., Bax, Caspase-5, transforming growth factor-α, PTEN) in MMR-deficient tumors affect repetitive DNA sequences.13–15 Third, it has been reported that other malignancies with MSI do not to accumulate BRAF mutations.27 The finding that MMR deficiency is not a prerequisite for the V599E alteration but, rather, correlates with hMLH1 promoter methylation is supported by other recently published data.24, 28

Many reports have shown that MSI in colorectal carcinoma can originate either from genetic10, 11 or epigenetic9, 21, 29–33 inactivation of MMR. The current data indicate that these are not simply different ways of acquiring a genetic instability that finally results in otherwise identical tumors. Rather, it should be assumed not only that the underlying reason for development of MSI is different but that the mutational spectra also diverge. Therefore, both tumor types would need to be considered as different entities of colorectal carcinomas with MSI. A suggested model for development of these tumors based on the current data and the published literature is shown in Figure 4. According to this model, a germline and/or somatic elimination of MMR, presumably occurring early during tumorigenesis, underlies the “classic” MSI pathway that occurs in patients with HNPCC and also occurs sporadically (the genetic MSI type). In contrast, strong CpG island methylation (which affects not only the distal Region A of the hMLH1 promoter but also Region C, the methylation of which imparts silencing of the gene), as a primary event, causes MMR deficiency and MSI by stopping hMLH1 transcription in patients with the BRAF V599E variant (the epigenetic MSI type). The V599E alteration itself, however, is not the direct result of MMR deficiency but, more likely, arises through loss of another DNA surveillance gene, which probably also is prone to epigenetic silencing by promoter methylation, explaining the concomitant occurrence of MSI and the BRAF V599E mutation. This putative antimutator gene may inhibit the alteration of thymidine to adenine, because its loss promotes the conversion of GTG (valine) to GAG (glutamate) at codon 599 of BRAF, thereby causing activation of the protooncogene. It recently was reported that a similar situation (inactivation of the tumor suppressor gene APC through loss of the antimutator gene MYH) underlies a recessively inherited attenuated polyposis syndrome.34, 35 The CpG island methylation that seems to initiate carcinogenesis of MSI tumors of the epigenetic type may originate either from a specific defect in the methylation machinery, as suggested for the “CpG island methylator phenotype” (CIMP),36–38 or may result from age-related methylation.39 The finding that patients with the V599E mutation are significantly older than other patients with MSI tumors supports the second possibility.

thumbnail image

Figure 4. This suggested model explains the distribution of BRAF mutations in colorectal carcinoma. It has been shown previously that the microsatellite instability (MSI) in colorectal carcinomas originates either from genetic loss of mismatch repair (MMR)10, 11 or from epigenetic inactivation of the hMLH1 promoter, which also inactivates MMR.9, 21, 29–33 Genetic loss (top) can be promoted by a preexisting germline MMR mutation (in patients with hereditary nonpolyposis colorectal carcinoma [HNPCC]).10, 11 The current results showed that the epigenetic type of colorectal carcinoma with MSI (bottom) frequently is accompanied by the BRAF V599E mutation. This mutation is not a consequence of the deficient MMR (in that case, the mutation also would occur in the genetic type). Because extensive CpG island methylation precedes loss of mismatch repair in these tumors, we suggest that a putative, methylation-mediated loss of another DNA-repair activity in the same tumor concomitantly causes the BRAF V599E mutation. The current data show that BRAF V599E is associated with the epigenetic type of colorectal carcinomas with MSI, highlighting the finding that tumors of the epigenetic and genetic type differ in their mutational spectra (and, thus, form distinct tumor entities), even though both share the same underlying genetic instability.

Download figure to PowerPoint

Fifteen percent of all tumors with MSI in our study exhibited BRAF V599E and hMLH1 promoter methylation accompanied by loss of this protein and, thus, belong to the epigenetic MSI type. Previous analyses of colorectal carcinomas with MSI found BRAF V599E rates of 10%,28 29%,18 and even 45%.24 These divergent rates may reflect different patient material. Assuming the currently available data (this study; see also Deng et al.40 and Miyaki et al.41), the identification of a BRAF V599E mutation in colorectal carcinoma seems to exclude a coexisting HNPCC germline mutation. Along with clinical and molecular genetic criteria for the identification of potential carriers of germline HNPCC mutations,19, 42 screening for BRAF V599E in colorectal tumor tissue may reduce further the number of patients undergoing an expensive mutation analysis for colorectal carcinoma with MSI.

MSI tumors of both types share similar clinical features, which likely mirrors the identical genetic instability that significantly contributes to their development. However, tumors of the epigenetic MSI type also can silence genes by promoter methylation, as has been demonstrated for PTEN,43 and, as the current results suggest, by an indirect pathway involving the accumulation of further mutations. The presence of alterations that are unique to epigenetic MSI tumors may confer different sensitivity to chemotherapeutic agents. Patients with tumors of the “epigenetic type,” for example, may benefit from an inhibitor that blocks BRAF V599E. Studies addressing the chemotherapy of colorectal carcinomas with MSI should take the heterogeneity of these tumors into consideration.

In contrast to BRAF V599E, the 5 novel BRAF alterations identified in the current study showed no significant association with any tumor trait that was assessed. It was shown recently that BRAF mutations can exert their oncogenic action either directly, by elevation of the proteins' kinase activity, or by indirect activation of MEK and its downstream effector ERK through CRAF, even when the kinase activity is reduced.8 The precise effect of all novel mutations currently is under investigation.

Although BRAF alterations were observed significantly more often in colorectal tumors with MSI than in colorectal tumors without MSI, the frequency of KRAS mutations did not differ between the two tumor groups (Table 2). Moreover, 1 tumor contained both BRAF and KRAS mutations (Patient 17, Table 1), a concurrence that was not observed in a previous analysis.18 Although the oncogenic effect of the novel BRAF mutation observed in this tumor requires confirmation, these findings do not support the notion that colorectal carcinomas progress preferentially, independent of their MMR status, through mutation of BRAF (tumors with MSI) or KRAS (tumors without MSI). Rather, our current data suggest that BRAF is an additional mutational target in tumors with MSI.

In conclusion, the current results suggest that there are different pathways in colorectal carcinomas with MSI: Tumors with the classic genetic MSI type arise due to germline and/or somatic loss of MMR activity, followed by genome-wide alterations that progress tumorigenesis. In tumors with the epigenetic MSI type, methylation of Region C of the hMLH1 promoter silences MMR in older patients and results preferentially in tumors that carry the BRAF V599E mutation. Therefore, tumors with MSI can be distinguished into at least two different entities with differences found not only in the underlying reason for their deficiency (genetic or epigenetic) but also in the progression through distinct oncogenic pathways.


  1. Top of page
  2. Abstract
  6. Acknowledgements

The authors are grateful to Priv.-Doz. Dr. Eva Herrmann for her help in the statistical evaluation of the data.


  1. Top of page
  2. Abstract
  6. Acknowledgements