Estimation of mutation rates in cultured mammalian cells

Abstract

The factors that affect reliable estimations of mutation rates (μ) in cultured mammalian somatic cell populations by fluctuation analysis are studied experimentally and statistically. We analyze the differential effect of the final cell population size in each culture ( N t) and the number of parallel cultures ( C) on the variation in the rate estimates ( μ ) inferred from the P 0 method. The analysis can be made after the derivation of the variance of μ , which is a measure of variation of μ for a given combination of N t and C in a number of repeat experiments. The variance of μ is inversely proportional to C and to the square of N t. N t determines the probability of occurrence of mutation in a cell culture. By influencing the size of P 0, N t also determines whether a rate estimate is obtainable from the experiment. Since P o is estimated from the fraction of cultures containing no mutation in a set of C cultures, C becomes a determining factor for the accuracy of μ . The rate estimated from P ̂ 0 is biased, but the bias is in general 2 orders of magnitude smaller than μ . By the selection of an appropriate combination of N t and C for the experiment, this bias can be reduced even further. Based on the notion of comparing two proportions, we propose a test statistic and have applied it to experimental results for a test of equality of mutation rates in different cell lines. This development places the comparison of mutation rates on a statistical basis.