Somatic Frameshift Mutations in the BAX Gene in Colon Cancers of the Microsatellite Mutator Phenotype

See allHide authors and affiliations

Science  14 Feb 1997:
Vol. 275, Issue 5302, pp. 967-969
DOI: 10.1126/science.275.5302.967


Cancers of the microsatellite mutator phenotype (MMP) show exaggerated genomic instability at simple repeat sequences. More than 50 percent (21 out of 41) of human MMP+ colon adenocarcinomas examined were found to have frameshift mutations in a tract of eight deoxyguanosines [(G)8] within BAX, a gene that promotes apoptosis. These mutations were absent in MMP tumors and were significantly less frequent in (G)8 repeats from other genes. Frameshift mutations were present in both BAX alleles in some MMP+ colon tumor cell lines and in primary tumors. These results suggest that inactivating BAX mutations are selected for during the progression of colorectal MMP+ tumors and that the wild-type BAX gene plays a suppressor role in a p53-independent pathway for colorectal carcinogenesis.

The MMP pathway for colon cancer is characterized by genomic instability that leads to the accumulation of deletion and insertion mutations at simple repeat sequences (13). The fixation of these slippage-induced replication errors as mutations (4) is associated with defects in DNA mismatch repair (5). Colorectal MMP+ tumors frequently contain frameshift mutations in the type II transforming growth factor-β (TGF-β) receptor gene (6) but are usually wild type for the p53 tumor suppressor gene (1, 7). In addition to its central role in cell growth arrest (8), p53 also plays a role in apoptosis in response to DNA damage (9). The p53 protein transactivates BAX (10), a member of the BCL2 gene family (11) that promotes apoptosis (12).

The human BAX gene contains a tract of eight consecutive deoxyguanosines in the third coding exon, spanning codons 38 to 41 (ATEmbedded Image Embedded Image Embedded Image Embedded ImageAG) (12). To determine whether this sequence is a mutational target in MMP+ tumor cells, we amplified by the polymerase chain reaction (PCR) the region containing the (G)8 tract from various MMP+ tumor cell lines. This analysis revealed band shifts suggestive of insertions and deletions of one nucleotide in some of these tumor cells (Fig. 1A). Prostate (DU145) and colon (LS180) tumor cells exhibited PCR patterns indistinguishable from those amplified from plasmids containing a BAX fragment with mutant (G)9 and (G)7 tracts (Fig. 1A, P9 and P7), respectively. DNA derived from tumor cell lines from colon (LS411N and HCT116) and from uterus (SKUT1B) generated patterns corresponding to mixtures of the wild-type (P8) and mutant alleles. The band patterns of LoVo colon cancer cells matched those of both mutant alleles.

Fig. 1.

(A) Frameshift mutations in BAX in MMP+ tumor cell lines (23). PCR products of a BAX fragment containing the (G)8 tract (24) are shown. SW620 MMP colon carcinoma (lane a); LoVo, colon (lane b); DU145, prostate (lane c); LS411N, colon (lane d); SKUT1B uterus (lane e); HCT116 colon (lane f); SW48, colon (lane g); CAL51, breast, (lane h); LS180, colon (lane i); and DLD-1, colon (lane j). Two other MMP+ cell lines from endometrial (AN3CA) and ovarian (SK-OV-3) cancers also showed a wild-type pattern. The triangles pointing up or down indicate the presence of insertions and deletions of one nucleotide, respectively. (B) Immunoblot analysis of BAX protein in MMP+ tumor cell lines (25). Lanes are the same as in (A). Human F1β adenosine triphosphatase (ATPase) was used as an internal control for the amount and stability of cell proteins. Molecular size markers are indicated at left.

Immunoblot analysis (Fig. 1B) showed that LoVo, DU145, and LS180 cells, which contain no wild-type BAX alleles, did not express detectable amounts of BAX protein. In contrast, LS411, SKUT1B, HCT116, and tumor cells without frameshift mutations expressed various amounts of BAX. These results indicate that DU145 and LS180 cells contain homozygous (or hemizygous) frameshift insertion (DU145) and deletion (LS180) mutations in BAX. LS411N, SKUT1B, and HCT116 cells harbor heterozygous mutations (or are heterogeneous cell populations) for the BAX (G)8 frameshift mutations. LoVo cells contain frameshift mutations in both BAX alleles, one by insertion and the other by deletion of a single nucleotide in the (G)8 tract.

We next analyzed the status of BAX in 41 primary MMP+ colon tumors. Tumor 132 contained an insertion and a deletion mutation, and tumor 197 a deletion mutation (Fig. 2A). Sequence analysis revealed that the band shifts were due to insertions [(G)9] or deletions [(G)7] of one nucleotide in the (G)8 tract (Fig. 2B). BAX mutations were detected in 21 of 41 (51%) primary MMP+ colorectal carcinomas (Fig. 3A), but not in 49 MMP carcinomas (Fig. 3B). Thus, frameshift BAX mutations are specifically associated with cancer of the MMP (probability P < 10−8, Fisher exact test). Moreover, deletion or insertion mutations in other (G)8 tracts were significantly less frequent in the same MMP+ tumors (13, 14).

Fig. 2.

PCR and sequence analysis of BAX mutations in MMP+ primary colon tumors. (A) PCR products of the region spanning the (G)8 tract in BAX from tumors (T) 132 and 197 and corresponding normal (N) tissue. (B) Sequences of amplified DNA fragments that showed mobility shifts (up in tumor 132 and down in tumor 197) and the normal fragment (WT) of 132 (26).

Fig. 3.

Frameshift mutations in BAX in MMP+ primary colon tumors. Representative MMP+ (A) or MMP (B) tumors (27). Normal tissue DNA is at the left and tumor tissue at the right in each case. Normal tissue DNA was not available for cases 151, 299, and 328. The mutations in tumor 43 were detected after microdissection from a paraffin block to enrich for tumor tissue.

We conclude that frameshift BAX mutations are selected for during tumorigenesis in MMP+ colon tumors (15). The mutations in both BAX alleles in primary tumors, such as tumors 43, 132, and 328, indicate that they are not due to selective pressure during in vitro culture, and that in cancer of the MMP, the occurrence of biallelic mutations is not restricted to microsatellite sequences (1). The presence in tumor cells of inactivating mutations within both alleles of mutator or suppressor cancer genes may be diagnostic of the prior existence of the MMP (14). These functional double mutational hits also provide evidence that the loss of BAX function plays a direct role in tumorigenesis. Cells lacking BAX protein may have a diminished capacity to trigger apoptosis upon receiving a death signal. This is consistent with the phenotype of mice deficient in BAX, which grow normally but eventually develop lymphoid hyperplasia (16). Heterozygous BAX mutations (15) may also contribute to tumor progression. Inactivation of one BAX allele presumably reduces the amount of wild-type BAX, which may in turn facilitate escape from apoptosis by diminishing the BAX-BCL2 ratio (17).

In contrast to p53 and other cancer genes that indirectly affect apoptosis, the only known function of BAX is to directly promote apoptosis (11, 12, 17). With the exception of two missense mutations detected in two human leukemia cell lines (18), BAX mutations have not been detected in human cancers. The presence of somatic BAX mutations in colorectal tumor cell lines and primary tumors suggests that BAX may be an important tumor suppressor gene in human cancer and could provide insight into the two distinct molecular genetic pathways for colon tumorigenesis (3, 19).

Our results may explain why colon tumors of the mutator pathway (3) typically do not contain p53 mutations (1, 7), in contrast with those of the suppressor pathway (19). Once the MMP is manifested (after the occurrence of mutator mutations in, for example, DNA mismatch repair genes), mutations at the BAX (G)8 hotspot would be more likely to occur than other frameshift or missense mutations in p53. In tumor cells with frameshift BAX mutations, transcriptional activation of BAX by wild-type p53 (11) would be irrelevant. In cancer of the MMP, the generation of thousands of DNA mismatches during every replication of each MMP+ tumor cell (20) may trigger the p53-mediated apoptotic response to DNA damage (21). But this response would be futile because the chain leading to apoptosis is broken in a downstream link. Therefore, it seems reasonable to speculate that BAX mutations eliminate the selective pressure for p53 mutations during colorectal tumorigenesis. Because some MMP+ tumors do not have mutations in either BAX or p53, and because inactivation of these genes may be insufficient to inactivate the apoptotic pathways, escape from apoptosis may occur by other mechanisms. Although they share BAX mutations, some of the tumor cell lines used in this study appear to belong to different complementation groups for apoptosis (22), thus illustrating the redundant complexity of the mechanisms regulating cell death.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.

Stay Connected to Science

Navigate This Article