The oxidized deoxynucleoside triphosphate pool is a significant contributor to genetic instability in mismatch repair-deficient cells

Abstract

Oxidation is a common form of DNA damage to which purines are particularly susceptible. We previously reported that oxidized dGTP is potentially an important source of DNA 8-oxodGMP in mammalian cells and that the incorporated lesions are removed by DNA mismatch repair (MMR). MMR deficiency is associated with a mutator phenotype and widespread microsatellite instability (MSI). Here, we identify oxidized deoxynucleoside triphosphates (dNTPs) as an important cofactor in this genetic instability. The high spontaneous hprt mutation rate of MMR-defective msh2(-/-) mouse embryonic fibroblasts was attenuated by expression of the hMTH1 protein, which degrades oxidized purine dNTPs. A high level of hMTH1 abolished their mutator phenotype and restored the hprt mutation rate to normal. Molecular analysis of hprt mutants showed that the presence of hMTH1 reduced the incidence of mutations in all classes, including frameshifts, and also implicated incorporated 2-oxodAMP in the mutator phenotype. In hMSH6-deficient DLD-1 human colorectal carcinoma cells, overexpression of hMTH1 markedly attenuated the spontaneous mutation rate and reduced MSI. It also reduced the incidence of -G and -A frameshifts in the hMLH1-defective DU145 human prostatic cancer cell line. Our findings indicate that incorporation of oxidized purines from the dNTP pool may contribute significantly to the extreme genetic instability of MMR-defective human tumors.

FIG. 1.

Expression of hMTH1 in msh2-defective MEFs. (A) Western blot of msh2^−/− MEFs. Clone 2 and 5 extracts were probed with antibody against hMTH1. The DLD1 human colorectal carcinoma cell line is shown for comparison. (B) Steady-state levels of DNA 8-oxoG in untransfected and hMTH1-transfected msh2^−/− MEFs. DNA was extracted from exponentially growing cells, and 8-oxoG was determined by HPLC/EC as described in Materials and Methods. Values are the mean ± standard deviation of several independent determinations (n = 7 for msh2^−/−; n = 5 for clones 2 and 5). Data for untransfected msh2^−/− MEFs (unfilled bar), clone 2 (hatched bar), and clone 5 (filled bar) are shown.

FIG. 2.

Locations of hprt mutations in msh2^−/− MEFs and in clone 5. Changes found in msh2^−/− MEFs and in clone 5 are shown above and below the sequence, respectively. Deletions are indicated by Δ and are single bases unless otherwise indicated. +g is a single-base insertion. The asterisk indicates a double mutant.

FIG. 3.

hMTH1 overexpression and the mutator phenotype of MMR-defective human cells. (A) Western blotting and steady-state levels of DNA 8-oxoG in untransfected DLD-1 (− and unfilled bar) and hMTH1-overexpressing clone 2A (+ and filled bar). Values for 8-oxoG determinations are the mean ± standard deviation of three independent experiments. (B) HPRT mutation rates in untransfected DLD-1 (unfilled bars) and clone 2A (filled bars). Two independent determinations are shown. (C) MSI at BAT26, BAT25, and SMT15. DNAs from independent subclones of untransfected DLD-1 and DU145 and their hMTH1-overexpressing clones (DLD-1/clone 2A and DU145/clone 1) were amplified by PCR. The total number and the number of unstable clones are shown together with the percentage of unstable clones. Mutation rates were calculated as follows: number of unstable clones/total number of clones × number of generations. The number of clones with −1/−2 changes (loss) or +1/+2 changes (gain) is indicated. (D) Examples of MSI at BAT26 and SMT15 in DU145. The values designate different allele sizes. The alterations are shown next to the appropriate panel. wt, wild type.

FIG. 4.

Oxidized purines and mutations in msh2^−/− MEFs expressing hMTH1. The major mutational types that are modulated by hMTH1 expression (fold decrease in rate indicated) are shown. It is proposed that mutations arise via oxidized purines that are incorporated into the daughter DNA strand (dotted line, arrowed) during DNA replication. In some cases, the same mutations can be derived from replication of template oxidized base incorporated during a previous round of replication. These are shown where appropriate. Base pairs involving oxidized bases shown boxed are those that have been previously identified by studies using purified DNA polymerases or inferred from in vitro replication studies. Template strands are designated by solid lines. AT → TA transversions and AT → GC transitions are considered to arise from incorporation of 2-oxodAMP opposite the correct base and miscoding of oxidized A (A*) present in the parental strand in the second round of replication (28). Incorporation of 2-oxodAMP opposite a G (24, 38) or miscoding of 8-oxoG when present in the template strand might both lead to GC → TA transversions (46, 12). GC → CG transversions derive from G*  ·  G mismatches. DNA polymerase eta is indeed able to direct, at low efficiency, the incorporation of guanine opposite a template 8-oxoG (23, 48). Incorporation of 8-oxodGMP opposite an A by replicative DNA polymerases (40) is the most plausible mismatch originating AT → CG transversions.

FIG. 5.

Model of frameshift formation in the hprt gene following 8-oxodGMP incorporation. In human cells, the origin of replication has been mapped in the first intron and the transcribed strand containing G₆ is replicated by the lagging-strand polymerase (13). The location of the origin in mouse cells is unknown. The leading and lagging strands are indicated by arrowed, dotted lines. In mouse cells, −1 frameshifts might arise via an IDL of the C-containing template strand favored by an 8-oxoG in the lagging daughter DNA strand. The inversion in the types of frameshifts in human cells (+G) could be related either to the incorporation of the oxidized purine by a leading-strand DNA polymerase or to the presence of an 8-oxoG in the template strand favoring +1 frameshifts via unstacking of the pyrimidine-containing primer strand during lagging-strand synthesis.