Compound heterozygotes in mutyh gene have been shown to be associated with familial colorectal carcinoma in human. In this work, we studied nine bacterially expressed mutant MUTYH proteins for their DNA glycosylase and binding activities. In vitro assay using synthetic DNA substrates revealed that missense mutants (R260Q and G382D) are partially active in glycosylase activity (rate constants k2, 1.28 ± 0.122 min−1 and 1.033 ± 0.0979 min−1, respectively, compare to 4.507 ± 0.5812 min−1 of WT) and DNA binding activity (26% and 36% of the substrates as complex I and II with R260Q and G382D, respectively.) whereas mutants (Y90X, Y165C, R231H, P281L, Q377X, E466X and 1103delC) are unable to generate any detectable cleavage products from the substrates containing A:GO mismatch or to bind to the substrates (See , and ).
Previously, Al-Tassan et al
. showed that the E. coli
mutant Y82C (analog to human Y165C) exhibits barely detectable glycosylase activity, whereas the mutant G253D (analogous to the human G382D variant) cleaves adenine from an A:GO mismatch almost as efficiently as WT protein.15
And yet, in another study, the murine mutant G365D protein (also corresponding to the human G382D MUTYH variant) was found to be fully active in removing A from a A:GO pair.13
Wooden et al.
characterized bacterially expressed GST-tag mutants (Y165C and G382D) and found completely inactive in glycosylase activity.26
When taken together, these previous data and our current data suggests strongly that frameshift mutants are completely defective in enzymatic activities due to loss of the C-terminal domain. Our study also shows that a mutation anywhere in the catalytic domain can impair enzyme activity (even at a distance from residue 222). This allows for the prediction of activity of clinical mutations in MYH based on sequence data.
Importantly, our DNA binding results are consistent with the relative glycosylase activity of mutant MUTYH proteins. Frameshift mutants (Y90X, Q377X, E466X and 1103delC) were unable to remove A from the substrates (although they possess intact catalytic domains), probably because of defective DNA-binding activity. Why the DNA-binding activity of the full-length (535aa) missense MUTYH variants (Y165C, R231H and P281L) was severely defective is unclear at this point and this deserves further study.
We observed that, in spite of similar substrates binding activity (74% and 68%, WT and Q324H, respectively), the SNP Q324H is only 64% active at saturation in excising A (k2
= 2.971 ± 02172 min−1
) from the substrates compare to the WT (k2
= 4.507 ± 0.5812 min−1
). In contrast, Shinmura et al
. found Q324H to be fully active as WT.29
Interestingly, Yuan et al
. has just reported that SNP Q324H is strongly associated with familial colorectal cancer among African-Americans.30
Therefore, on the basis of our results, we concluded in addition to the germ-line mutations, SNPs should also be studied for their possible involvement in MAP.
The MUTYH crystal structure has not yet been solved. However, the observed biochemical activity of the MUTYH variants in this study may be explained by the study of the amino acid sequence of the E. coli
MutY protein, which is 41% identical with MUTYH.7,26–28
MutY has a catalytic domain consisting of helix-hairpin-helix (HhH), pseudo HhH and iron-sulfur cluster [4Fe-4S], and a characteristic C-terminal domain, that is not found in other HhH-superfamily BER proteins.31–33
Positions of the amino acid substitution of the mutants studied in this work are shown in . All the missense mutants, except G382D, would lie within this catalytic domain. These amino acid residues are highly conserved among the human, E. coli
, murine and S. pombe
MutY homologs. While the C-terminal domain of E. coli
MutY protein is not needed for its adenine removal activity, it strongly interacts with the GO-containing strand to flip the target adenine out of the helix. In this way, A is buried into a pocket formed between 6-helix barrel module and [4Fe-4S] module and subsequently β-glycosidic bond between A and pentose sugar moiety is hydrolyzed by the protein’s glycosylase activity.31,32
The mutation in the G382D variant lies within the C-terminal domain which is required for GO recognition. In support of this theory, it has been shown previously that the removal of this domain from MutY protein drastically reduces its adenine incision activity from A:GO pair, but not from A:G mismatch.33
Therefore, the inactivity of our frameshift variants can be explained by the fact that the Y90X mutant loses both catalytic domain and the C-terminal domain whereas the other mutants (Q377X, E466X and 1103delC) lost only the catalytic domain (). Indeed, it was previously shown that the C-terminal domain truncated MutY is not only defective in removing A but also in binding substrate containing A:GO.33,34
Alignment of MUTYH and MutY amino acids sequences. Un-shaded, catalytic domain; shaded, C-terminal domain; red letter, conserved amino acid residue.
A mutation in the catalytic domain might render the protein partially or fully inactive where as the lost of C-terminal domain could impaired binding to the substrate containing A:GO. Protein products of SNPs are generally active as WT, but we observed that the SNP Q324H is 36 % less active than WT. This finding is consistent with the recent observation by Yuan et al.
who have reported that the Q342H variant SNP is strongly associated with familial colorectal cancer among African-American population.30
The second SNP V22M was found to be as active as the WT (k2=
4.288 ± 0.7831 and 4.507 ± 0.5812 min−1
, respectively). In summary, we have characterized in vitro
a large series of frameshift, missense mutants and SNP of MUTYH in one laboratory setting that are clinically associated with increased CRC risk. The results are summarized in Table 2. The two missense variants (R260Q and G382D) were partially active in DNA binding and BER activities, while three missense variants (Y165C, R231H and P281L) and all four frameshift variants (Y90X Q377X, E466X and 1103delC) were dysfunctional in both activities.
Adding further complexity is that there may be cross-talk between repair pathways in preventing colorectal carcinogenesis. In human cells, three BER proteins (OGG1, MTH1 and MUTYH) and three MMR proteins (MSH2, MSH6 and MLH1) guard genome from mutagenic DNA base lesion 8-oxoG.35
Their functions have been well studied both in vitro
and in vivo
and indeed MMR and MYH proteins interact biochemically.12,13,29,36–40
As such, mutations in the MUTYH catalytic domain that renders the protein fully or partially inactive could abrogate normal interactions between MYH and MMR proteins. In this study, we further confirmed that even mutations outside the catalytic domain can deactivate protein function in a partial or complete manner. This finding could explain the observed accumulation of G:C to T:A transversions in APC gene which are associated with CRC.
MUTYH protein is a second defence against deleterious 8-oxoG lesions that escape initial OGG1-based surveillance and repair. Our data suggests that mutations and SNPs in MUTYH could give rise to defective proteins that might affect genomic stability. At present, there are no reports of dysfunctional OGG1 or MTH1 as determinants of colorectal carcinogenesis. Given mutations within MMR genes are also linked to hereditary non-polyposis colorectal cancer (HNPCC),41,42
it will be of interest to closely study the interaction of wild type and mutant MUTYH proteins and MMR proteins in biochemical and cellular system to further delineate interactions between BER and MMR in colorectal carcinogenesis.