Figure 4. High frequency of A3 editing recovered by standard PCR.

A) Frequency analysis of edited genomes as a function of the number of edits per sequence for 95°C derived PCR products. The 268 bp sequences were derived from the first round product. The sequences in insert to Figure 4A were stripped down to the size of the inner 167 bp locus and reanalyzed to allow comparison with the 3DPCR products. The numbers above the columns indicate the combined numbers of sequences across the four samples. B) Frequency distribution of edited sequences for the 3DPCR products obtained at 88.7°C. In order to calculate the bias resulting from PCR close to the denaturation temperature, let's assume that the frequency of clones with 1–17 edited cytidines reflected sub optimal amplification, while the profile in Figure 4A is close to the true distribution. By summing the number of clones with 1–17 and ≥18 edits for Figures 4A and 4B the estimated number with between 1–17 edits in Figure 4B is n, where 79/n = 7/300; n = 3386. As the number of clones in Figure 4B with 1–17 edits is 284, 3DPCR underestimates the true frequency by a factor of 3386/284, or ∼12. C) Bulk dinucleotide analysis of the 95°C sequences harbouring 1, 2–4 and ≥5 G→A transitions reveals the CpC hallmark of A3G editing. D & E) Clonal analysis reveals that editing was due to an APOBEC deaminase, approximately half being due to A3G. The smaller number of sequences used (n = 40) in these figures means that the values of 55% and 45% are less robust than for Figure 3C & 3D. F) 3DPCR amplification across a 85–93°C gradient using either Taq or Pfu DNA polymerase, the latter fails to amplify DNA containing dU, the product of A3 deamination. Asterisks indicate the PCR products cloned and sequenced. G) Mutation matrices of Pfu and Taq amplified HBV hypermutants given as percentages.