Figure 2: Selection strength, not mutation rate, is responsible for a lower rate of adaptation in immune-compromised mice.

(a) Higher growth rate of E. coli in immune-compromised mice. The doubling time of E. coli colonizing the gut of WT or Rag2^−/− is estimated by the amount of rRNA (see Supplementary Fig. 3). Each symbol corresponds to an average of two independent quantifications made for each mouse (n=6) during days 1 and 3 of colonization. The doubling time is significantly lower in Rag2^−/− than in WT mice, indicating that E. coli divides faster in the gut of Rag2^−/− than in WT animals (Mann–Whitney U-test, W=111.5, P<10^-3). ***P<0.001. (b) Similar frequency of spontaneous mutation in immune-competent and immune-compromised mice: the log₁₀ frequency of spontaneous resistance (μ) in WT (open circles) and in Rag2^−/− (full circles) animals is proportional to the mutation rate at mutation-selection balance. Symbols correspond to measurements of mutation frequency over a period of 15 days for rifampicin (n=6), nalidixic acid (n=6) and furazolidone resistance (n=11 and 10). We estimated an average log₁₀ μ_WT∼−7.56 and μ_Rag2^−/−∼−7.53 for rifampicin (Mann–Whitney U-test, W=284.5, P=0.95), μ_WT∼−7.76 and μ_Rag2^−/−∼−7.77 for nalidixic acid (Mann–Whitney U-test, W=202, P=0.65) and μ_WT∼−5.65 and μ_Rag2^−/−∼−5.49 for furazolidone resistances (Mann–Whitney U-test, W=1613, P=0.10). The frequency of spontaneous transpositions corresponds to the fraction of mutations conferring furazolidone resistance, which are caused by insertion of transposable elements: μ_{TE WT}∼−5.99 and μ_{TE Rag2}^−/−∼−5.75 (Mann–Whitney U-test, W=83, P=0.09). NS, not significant; P>0.05. (c) Mean selective advantage of an emerging gat allele is lower in Rag2^−/− than in WT mice. Results of short-term competitive fitness experiments (n=10) between a gatZ mutant and the ancestral E. coli. Rag2^−/− (right panel) or WT (left panel) mice from the same litter are represented by symbols of the same shape (triangles, squares, diamonds or circles). The selective advantage of gatZ along the 3 days of competition is inferred from the slope of the linear regression of ln(gatZ/anc), (dashed lines, shown for each mouse), which corresponds to the selection coefficient.