Skip to main content
. 2021 Jun 27;38(10):4493–4504. doi: 10.1093/molbev/msab190

Fig. 2.

Fig. 2.

Mutant library selection using piperacillin-tazobactam and TLS phage. Heatmaps summarizing a subset of fitness values of TolC mutations under selection of (A) Piperacillin-tazobactam (6 µg/ml) and (B) TLS Phage (2.5×108 pfu/ml). Columns within each matrix represent the TolC residues (from residues 1 to 121), and columns represent mutations to other codons causing synonymous(ø), nonsynonymous, or nonsense (X) mutations. Protein sequence of the wild-type TolC is represented by white pixels with cross on them. Mutants with low read counts (threshold=µ−1.5α) in the initial TolC library or after growth in plain media (untreated) were excluded from fitness calculations and represented by black pixels. All other calculated fitness values were colored from dark blue (increased susceptibility) to white (neutral effects) to dark red (increased resistance). Top row of each heatmap shows the effect of early stop codon mutations (X). Second rows from the top show the effect of silent mutations synonymous to native codon in TolC. Under piperacillin-tazobactam selection (A), nonsense mutations increased susceptibility, whereas under TLS phage selection (B), nonsense mutations increase resistance. (CH) Distribution of fitness effects (DFEs) for different selection agents. For every selection agent, DFEs were calculated under two different selection strengths. DFEs under antibiotic selection are narrow and centered around neutrality (s = 0) regardless of the selection strength, with tails extending to the left (increased sensitivity, insets). Under both colicin-E1 and TLS phage selections, DFEs were wide and mean fitness effects of mutations were negative, suggesting that TolC was not robust to mutations under selection to these agents. Under both colicin-E1 and phage selection, many mutations were initially present in the TolC library but went extinct after the selection (dark gray bins). Although it is not possible to calculate fitness in these cases, in order to show them on the histograms, we set their final counts to 1 (pseudocount) and calculated a fitness value such that they are still visible. Under both colicin-E1 and phage selections, DFEs were bimodal with a second peak corresponding to mutations increasing resistance. Means and SDs of light gray histograms are tabulated in supplementary table S1, Supplementary Material online.