Table 1b.

Influence of PCR cycle numbers upon the rates (%) of mutagenesis using the nucleotides analogues method

Number of cycle	P1 (225 bp)	P2 (255 bp)
5	3.8 ± 0.3	3.7 ± 0.3
10	7.8 ± 0.4	7.7 ± 1.3
15	8.2 ± 0.3	7.8 ± 0.5
20	10.3 ± 1.9	9.0 ± 0.6
25	13.8 ± 1.3	11.8 ± 2.8
30	16.9 ± 0.6	13.2 ± 1.3

The mutagenesis efficiencies arising from the random incorporation of nucleotide analogues (0.5 mM dPTP; method 2) were compared to those achieved by decreasing the fidelity of the Taq polymerase (method 1). For this comparison, the 5 overlapping hypervariable domains of gp38 fragments (P1 to P5) encoding the C terminus receptor-recognition domain of this tail fiber protein (Table 1a) and the first 2 (P1 and P2) of the 15 overlapping fragments constituting the gp37 gene (Table 1b) were independently subjected to a PCR reaction made error-prone by either of the 2 methods. The resulting products were then sequenced (n = 4) and the frequencies of mutated sites recorded in terms of percentages. Using 15 error-prone PCR cycles, the random incorporation of nucleotide analogues produces much higher levels of mutagenesis than the destabilization of Taq fidelity (averages of 7.7% ± 2.6% versus 1.6% ± 1.5%, respectively, Table 1a). However, the results obtained with the latter approach appeared considerably more stochastic (standard error as large as the mean) and thus difficult to control. While neither method appeared particularly dependent on the lengths of DNA fragments (Table 1a), mutagenesis by random incorporation of nucleotide analogues is highly sensitive to the number of error-prone PCR cycles (Table 1b), thus allowing control over the levels of mutagenesis through the modulation of both the concentrations of nucleotide analogues in the reaction and the number of error-prone PCR cycles.