APOBEC3A and APOBEC3B Preferentially Deaminate the Lagging Strand Template during DNA Replication

Abstract

APOBEC family cytidine deaminases have recently been implicated as powerful mutators of cancer genomes. How APOBECs, which are ssDNA-specific enzymes, gain access to chromosomal DNA is unclear. To ascertain the chromosomal ssDNA substrates of the APOBECs, we expressed APOBEC3A and APOBEC3B, the two most probable APOBECs mediating cancer mutagenesis, in a yeast model system. We demonstrate, using mutation reporters and whole genome sequencing, that APOBEC3A- and APOBEC3B-induced mutagenesis primarily results from the deamination of the lagging strand template during DNA replication. Moreover, our results indicate that both genetic deficiencies in replication fork-stabilizing proteins and chemical induction of replication stress greatly augment the mutagenesis of APOBEC3A and APOBEC3B. Taken together, these results strongly indicate that ssDNA formed during DNA lagging strand synthesis is a major substrate for APOBECs and may be the principal substrate in human cancers experiencing replication stress.

Figure 1. APOBEC3A and 3B induce strand-bias mutations around ARS216

(A) Frequencies of canavanine-resistance (Can^R) induced in WT and ung1Δ yeast following transformation with vector control plasmid, A3A expression plasmid, or A3B expression plasmid. Can^R frequency was determined in isogenic strains with the CAN1 gene located either 16 kB 5′ of the origin of replication, ARS216, or 7 kB 3′ of ARS216. Horizontal bars and numeric values indicate the median frequency of six or seven independent replicates. Statistical significance was determined by two-tailed non-parametric Mann-Whitney Rank test. See also Figure S1. (B) Sequence logo describing the favored trinucleotide sequences mutated by A3A (top) and A3B (bottom) in ung1Δ yeast. See also Table S1. (C) Specific deamination of ssDNA formed in a transcription bubble would result in C to T transitions (red ball) within CAN1 regardless of whether the gene is positioned 5′ or 3′ of ARS216. In contrast, deamination of ssDNA formed during lagging strand synthesis would result in G to A substitutions (green ball) 5′ and C to T substitutions 3′ or the origin. (D) The number of G to A (green) and C to T (red) substitutions induced in the CAN1 gene (located 5′ and 3′ of ARS216) by A3A and A3B in ung1Δ and UNG1 yeast as determined by sequencing independent Can^R isolates. P-values were determined using a one-tailed g-test goodness-of-fit comparing the ratio of C to T substitutions to G to A substitutions to an expected 1:1 ratio. See also Table S1.

Figure 2. Genome-wide strand-bias of A3A and A3B mutations between neighboring origins of replication

(A) Distribution of C to T (red) and G to A (green) mutations induced by A3A (top) and A3B (bottom). Yeast chromosomes are displayed in gray. (B) The relative abundance of A3A- and A3B-induced C to T mutations (red) and G to A mutations (green) in yeast genes +/− 500 nucleotides transcribed on the bottom strand (B) or the top strand (T). P-values were determined using a two-tailed chi-square test comparing the numbers of each mutation type to the number of A3A- and A3B-targeted TC (red) or GA (green) dinucleotides occurring in these regions. (C) The relative abundance of A3A- and A3B-induced C to T mutations (red) and G to A mutations (green) according to the fractional distance between neighboring replication origins. Lines represent linear trend lines fitting the fractional abundance of C to T (red) and G to A (green) mutations across the entire fractional distance between neighboring origins. A statistical significance of p<0.0001 was determined by comparing the number of each mutation type in each decile bin to the corresponding abundance of TC or GA dinucleotides, by chi-square test. See also Table S2.

Figure 3. Deletion of replication fork stability factors exacerbates A3A- and A3B-induced mutagenesis around ARS216

(A) Frequencies of Can^R induced in wild type, rfa1-t33, tof1Δ, ung1Δ, rfa1-t33 ung1Δ, and tof1Δ ung1Δ yeast following transformation with vector control plasmid, A3A expression plasmid, or A3B expression plasmid. Horizontal bars and numeric values indicate the median frequency of six or seven independent replicates. P-values were determined by two-tailed non-parametric Mann-Whitney Rank test. (B) The number of G to A (green) and C to T (red) substitutions induced in the CAN1 gene (located 5′ and 3′ of ARS216) by A3A and A3B in ung1Δ yeast in the genetic backgrounds mentioned in (A) as determined by sequencing independent Can^R isolates. P-values were determined using a two-tailed Fisher's Exact test comparing the ratio of C to T substitutions to G to A substitutions between indicated genotypes. See also Table S1.

Figure 4. Chemically-induced replication stress augments A3A- and A3B-induced mutagenesis

(A) HU treatment causes a dose-dependent decrease in cell growth in ung1Δ yeast. Shown are medians and ranges for 18 independent replicates at each HU dose. P-values were determined by two-tailed non-parametric Mann-Whitney Rank test comparing untreated yeast to yeast treated with 50 mM HU. (B) Frequencies of Can^R induced in ung1Δ yeast treated with 0, 12.5, 25, or 50 mM HU following transformation with vector control plasmid, A3A expression plasmid, or A3B expression plasmid. Horizontal bars and numeric values indicate the median frequency of six or seven independent replicates. P-values were determined by two-tailed non-parametric Mann-Whitney Rank test. (C) The number of G to A (green) and C to T (red) substitutions induced in the CAN1 gene by A3A and A3B in untreated or 50mM HU treated ung1Δ yeast as determined by sequencing independent Can^R isolates. P-values were determined using a two-tailed Fisher's Exact test comparing the ratio of C to T substitutions to G to A substitutions between indicated treatments. See also Table S1.