Adaptation to antibiotics by large genomic duplications is favoured in lon-deficient E. coli.

A. Coverage depth plots and mutations at the folA and mgrB loci of 5 independently evolved trimethoprim-resistant isolates derived from wild type (WTMPR1-5) or Δlon E. coli (LTMPR1-5). Resistant clones were isolated after ∼25 generations of evolution at 300 ng/mL of trimethoprim. Coverage depth plots show the number of reads from Illumina short-read sequencing (y-axis) mapped to each genomic coordinate of the reference genome (x-axis). Coverage of 1x, 2x and 3x are marked by dotted lines for reference. Mutations in mgrB and folA, and trimethoprim IC50 values (mean ± S.D. from 3 independent measurements) are provided for each isolate below the appropriate coverage depth plot. B. Cartoon (not to scale) showing the duplicated genomic stretch and flanking IS-elements in trimethoprim-resistant isolates. Blue arrows represent IS1, while red arrows represent IS186. The positions of folA and acrAB genes, implicated in trimethoprim resistance in earlier studies, relative to the duplicated stretch are shown. Fold increase in the number of reads corresponding to the folA gene are indicated for each isolate. C. Summary of the prevalence of GDA mutations detected in wild type or E. coli Δlon after ∼25 generations of evolution in control media (no antibiotic) or in media supplemented with sub-MIC antibiotics. Each circle represents one of the 5 replicates in the evolution experiment. Light grey circles represent replicates in which no resistance was detected and hence were not sequenced. Dark grey circles indicate that the whole population or a resistant isolate was sequenced but no GDA mutation was detected. Red circles indicate that a GDA mutation was identified.

Overexpression of DHFR in trimethoprim-resistant isolates with genomic duplication.

A. DHFR protein expression in trimethoprim-resistant isolates WTMPR4 and LTMPR1, compared with respective ancestors and mgrB-knockout strains. DHFR protein was detected by immunoblotting using an anti-DHFR polyclonal antibody. FtsZ was used as loading control. A representative immunoblot from 3 biological replicates is shown. Quantitation was performed by calculating band intensities using image analysis. Fold change values of DHFR levels over controls (wild type or E. coli Δlon, set to 1) is shown as mean ± S.D. from triplicate measurements. B. Expression level of duplicated genes in LTMPR1 and WTMPR4, determined by RNA-sequencing, expressed as fold over E. coli ΔlonΔmgrB and E. coli ΔmgrB respectively. Each point represents fold change value for a single gene. Mean of the scatter is shown as a black line, and its value is provided along with standard deviation above the plot. C, D. Correlation between expression levels of genes in LTMPR1 (C) and WTMPR4 (D) with E. coli ΔlonΔmgrB and E. coli ΔmgrB respectively. Gene expression levels were estimated using RNA-sequencing and fold changes were calculated using wild type E. coli as reference for WTMPR4 and E. coli ΔmgrB, and E. coli Δlon as reference for LTMPR1 and E. coli ΔlonΔmgrB. Each point represents log2(fold change) value for a single gene. The resulting scatter was fit to a simple linear regression and the obtained R2 values are provided. E. Relative fitness (w) of indicated E. coli strains calculated using competition experiments with E. coli ΔlacZ as reference. For strains derived from E. coli Δlon, E. coli ΔlonΔlacZ was used as reference. Neutrality of ΔlacZ was established by competition with the unmarked ancestral strains, i.e. wild type (WT) and E. coli Δlon. No change in relative fitness compared to the reference strain (w = 1) is indicated by a dotted line. Mean ± S.D. from three independent measurements is plotted.

Evolutionary fate of folA-duplication in the absence of drug pressure.

A. Trimethoprim IC50 values of 3 evolving lineages derived from LTMPR1 (A, B, C) over 252 generations of evolution in antibiotic-free medium. Mean ± S.D. from three measurements is plotted at each time point. IC50 values of E. coli Δlon and ΔlonΔmgrB are shown as dotted lines for reference. B. Colony formation of LTMPR1, WTMPR4 and WTMPR5 evolved in antibiotic-free medium at different generations. Fraction of the population capable of forming colonies at 1, 5 and 10 μg/mL trimethoprim was calculated across 3 replicate lines. Mean value from the three lines is plotted. The results of similar experiments performed on ancestors (0 generations) are also provided for reference. C. DHFR expression during evolution in the absence of trimethoprim in LTMPR1 lineages A, B and C, measured by immunoblotting using anti-DHFR polyclonal antibody. FtsZ was used as a loading control. Quantitation was performed by calculating band intensities using image analysis. DHFR expression at each time point was normalized to the ancestor (i.e. 0 generations, set to 1). Mean of three independent measurements is shown below each lane. D. Copy number of the ancestral GDA encompassing folA (GDA(folA)) and GDA-2 at different time points of evolution in the absence of trimethoprim. For GDA(folA), copy number was determined by dividing number of reads from an Illumina sequencing experiment corresponding to folA by the average number of reads mapping to the rest of the genome. E. Point mutations in folA, rpoS and mgrB, and the “GDA-2” genomic duplication in 6 randomly picked colonies from each of the LTMPR1 lines at 252 generations of evolution are shown schematically using the appropriate symbols. Asterix (*/**) marks the genotypes that were carried forward for further analyses. F, G. Point mutations in folA, mgrB and rpoS in isolates derived from 252 generations of evolution of WTMPR4 (F) and WTMPR5 (G) in no antibiotic. From each of the evolving lines 2 random isolates were picked for genome sequencing. Various point mutants at the 3 gene loci of interest are represented by appropriate symbols as shown.

Fitness enhancement of LTMPR1 in drug-free media results from lower RpoS activity.

A. Relative fitness (w) of isolates with the indicated genotypes derived from LTMPR1 evolution in antibiotic-free media (see also Figure 3) calculated using a competitive growth assay using E. coli ΔlonΔlacZ as the reference strain. Mean ± S.D. from three independent measurements is plotted. No change in relative fitness compared to the reference (w = 1) is shown as a dotted line for reference. B. Expression level of 168 known targets of RpoS in E. coli Δlon compared to wild type expressed as log2(fold change) values determined by RNA sequencing. Dotted lines at 1.5-fold higher and lower than wild type represent cut-offs used to identify overexpressed and under-expressed genes. The pie chart shows the fraction of RpoS targets that were overexpressed in E. coli Δlon by at least 1.5-fold in green. C. Expression levels of 41 RpoS targets in E. coli Δlon compared to wild type are shown as a heat map in the first vertical. The expression levels of these genes in LTMPR1, E. coli ΔlonΔmgrB or LTMPR1-derived isolates compared to E. coli Δlon are provided in the subsequent verticals. D. Relative fitness of E. coli ΔlonΔrpoS in antibiotic-free media calculated using a competitive growth assay using E. coli ΔlonΔlacZ as the reference strain. Neutrality of the ΔlacZ genetic marker was verified by competition between E. coli Δlon and E. coli ΔlonΔlacZ. Mean ± S.D. from three independent measurements is plotted. No change in relative fitness compared to the reference (w = 1) is shown as a dotted line for reference. E. Model for the effects of Lon deficiency on RpoS levels and bacterial fitness. The roles of mutations within RpoS or the GDA-2 mutations in compensating for the defects of RpoS overproduction are shown.

Evolutionary fate of folA-encompassing GDA in LTMPR1 at sustained trimethoprim pressure.

A. Trimethoprim IC50 values of 3 evolving lineages starting from LTMPR1 (A, B, C) over 252 generations of evolution at 300 ng/mL trimethoprim. Mean ± S.D. from three measurements is plotted at each time point. IC50 values of E. coli Δlon and ΔlonΔmgrB are shown as dotted lines for reference. B. Colony formation of LTMPR1, WTMPR4 and WTMPR5 evolved in 300 ng/mL trimethoprim at indicated time points. Fraction of the population capable of forming colonies at 1, 5 and 10 μg/mL trimethoprim was calculated across 3 replicate lines. Mean value from the three lines is plotted. The results of similar experiments performed on ancestors (0 generations) are also provided for reference. C. DHFR expression in LTMPR1 lineages A, B and C evolved in 300 ng/mL trimethoprim, measured by immunoblotting using anti-DHFR polyclonal antibody. FtsZ was used as a loading control. Quantitation was performed by calculating band intensities using image analysis. DHFR expression at each time point was normalized to the ancestor (i.e. 0 generations, set to 1). Mean of three independent measurements is shown below each lane. D. Copy number of folA (GDA(folA)) at different time points of evolution of LTMPR1 in trimethoprim is plotted on the left y-axis. Copy number was determined by dividing number of reads from an Illumina sequencing experiment corresponding to folA by the average number of reads mapping to the rest of the genome. Frequency of various folA alleles in the evolving populations at each of the time points is also plotted on the right y-axis. E. Point mutations in folA, rpoS and mgrB in 6 randomly picked trimethoprim-resistant colonies from each of the LTMPR1 lines at 252 generations of evolution are shown diagrammatically. Genotypes that were carried forward for further analysis are marked (* or #). F, G. Point mutations in folA, mgrB and rpoS in isolates derived from 252 generations of evolution of WTMPR4 (F) and WTMPR5 (G) in 300 ng/mL of trimethoprim. From each of the evolving lineages 2 random isolates were picked for genome sequencing. Various point mutants at the 3 gene loci are represented by appropriate symbols as shown.

Invasion of GDA-dominant populations by point mutations under trimethoprim pressure.

A. Upper panel. Competition between isolates with point mutations in folA and an isolate that harboured a folA duplication derived from the same time point of LTMPR1 evolution (upper panel). The genotypes of the isolates used are shown diagrammatically (see also Figure 5, Supplementary File 4). The initial mixing ratio was 1000:1 in favour of the GDA mutant. Point mutants were marked genetically with ΔlacZ::Cat. Neutrality of the ΔlacZ::Cat marker was verified by competing marked and unmarked point mutants (lower panel). Percentage of the population constituted by the marked point mutants over ∼105 generations of serial transfer in media supplemented with trimethoprim (300 ng/mL) is plotted. Traces for individual mutants are shown in grey. Average of the 4 point mutants is shown in red. B. Trimethoprim IC50 values of the LTMPR1 ancestor or evolved isolates from 252 generations in trimethoprim with indicated genotypes (see also Figure 5, Supplementary File 4) are plotted as bars. Mean ± S.D. from three independent measurements are plotted. Doubling times relative to LTMPR1 (set to 1) in the presence of 300 ng/mL of trimethoprim are provided above the graph as mean ± S.D. from three independent measurements.

Evolutionary fate of folA-encompassing GDA in LTMPR1 at increasing trimethoprim pressure.

A. Trimethoprim IC50 values of 3 evolving lineages starting from LTMPR1 (A, B, C) over 252 generations of evolution are plotted on the left Y-axis. Mean ± S.D. from three measurements is plotted at each time point. Trimethoprim concentrations used during evolution are plotted in the right Y-axis. B. Colony formation of LTMPR1, WTMPR4 and WTMPR5 evolved in increasing trimethoprim at indicated time points. Fraction of the population capable of forming colonies at 1, 5 and 10 μg/mL trimethoprim was calculated across 3 replicate lines. Mean value from the three lines is plotted. The results of similar experiments performed on ancestors (0 generations) are also provided for reference. C. DHFR expression in LTMPR1 lineages A, B and C evolved in increasing trimethoprim, measured by immunoblotting using anti-DHFR polyclonal antibody. FtsZ was used as a loading control. Quantitation was performed by calculating band intensities using image analysis. DHFR expression at each time point was normalized to the ancestor (i.e. 0 generations, set to 1). Mean of three independent measurements is shown below each lane. D. Copy number of folA (GDA(folA)) at different time points of LTMPR1 evolution in increasing trimethoprim is plotted on the left y-axis. Copy number was determined by dividing number of reads from an Illumina sequencing experiment corresponding to folA by the average number of reads mapping to the rest of the genome. Frequency of various folA alleles in the evolving LTMPR1 populations at each of the time points is shown on the right y-axis. E. Point mutations in folA, rpoS and mgrB in 6 randomly-picked trimethoprim-resistant colonies from each of the LTMPR1 lines at 252 generations of evolution are shown schematically. F, G. Point mutations in folA, mgrB and rpoS in isolates derived from 252 generations of evolution of WTMPR4 (F) and WTMPR5 (G) in increasing trimethoprim. From each of the evolving lineages 2 random isolates were picked for genome sequencing. Various point mutants at the 3 gene loci are represented by appropriate symbols as shown.

Proteostatic pressure facilitates maintenance of folA duplication

A. Trimethoprim-IC50 values of resistant isolates with the indicated genotypes (see also Figure 7) derived from LTMPR1 evolution in increasing antibiotic pressure. Mean ± S.D. from three independent measurements is plotted. B. DHFR expression level in LTMPR1-derived trimethoprim resistant isolates with indicated genotypes assessed by immunoblotting using anti-DHFR polyclonal antibody. FtsZ was used as a loading control. Quantitation was performed by calculating band intensities using image analysis. DHFR expression was normalized to the ancestor (LTMPR1). Mean of three independent measurements is provided. C. Cartoon representation of the structure of E. coli DHFR (PDB: 7DFR) bound to folate. Residues Trp30, Tyr151, Phe153 and Ile155 which form hydrophobic interactions and are required for proteolytic stability are shown as sticks and coloured by element (C: green, O:red, N:blue). Distances of less than or equal to 4 Å, which indicates possible interactions, are shown as dotted yellow traces. Folate bound in the active site of DHFR is shown as sticks. D. Trimethoprim IC50 values of E. coli wild type or E. coli Δlon heterologously expressing DHFR (wt) or its mutants. Mean ± S.D. from three independent measurements is plotted. E. Expression level of plasmid-borne DHFR (wt) or its mutants in wild type or Δlon E. coli assessed by immunoblotting using anti-DHFR polyclonal antibody. FtsZ was used as a loading control. Quantitation was performed by calculating band intensities using image analysis. Expression of mutants was normalized to wild type DHFR (set to 1). Mean of three independent measurements is shown.

Model of expression demand induced selection of folA gene duplication, followed by replacement by phenotypically equivalent point mutants.

Expression demand for DHFR is generated by trimethoprim pressure, which results in the selection of folA gene duplications. Gene duplication is unstable and reverses to a single copy of wild type folA when drug-pressure is withdrawn. Under drug pressure, folA gene duplication is maintained until a point mutation that confers resistance arises in one of the copies of folA (promoter or coding region). Upon acquisition of a point mutation, reversal of gene duplication ensues unless additional expression demand is generated by the action of proteostatic machinery on unstable drug-resistant DHFR mutants.

List of strains and plasmids used in the study

Stability of heterologously expressed InsB and InsL transposases in wild type or Δlon E. coli measured using a chloramphenicol (CMP) chase assay. Levels of the plasmid- expressed transposase were measured using immunoblotting with an anti-His monoclonal/polyclonal antibody in lysates prepared from cells at indicated time points following treatment with 50 μg/mL of CMP. A representative image from three independent experiments is shown. Band intensities were quantitated by image analyses and normalized to 0’ (set to 1). The fraction of protein remaining is plotted in blue for wild type and red for E. coli Δlon. Values from 3 independent experiments are plotted as mean ± S.D.

Coverage plots and flanking IS elements for E. coli Δlon populations evolving in no antibiotic or spectinomycin that showed GDA mutations. For no-antibiotic, only one of the 5 replicates is shown as similar coverage plots were obtained for all 5 replicates.

Gene expression changes associated with folA-encompassing duplication.

Heat map comparing the top 50 up/downregulated genes in LTMPR1 and WTMPR4 with E. coli ΔlonΔmgrB and E. coli ΔmgrB respectively. Log2(Fold change) values were calculated based on RNA-sequencing with respect to wild type or Δlon E. coli and represented on a green/red scale as indicated. Gene names are shown adjacent to the appropriate cell. Genes are ranked according to their fold change in LTMPR1 and WTMPR4.

Coverage depth plots for population sequencing at 105, 196 and 252 generations of the 3 lineages of LTMPR1 (A, B, C) evolving in trimethoprim-free media. Coverage depth plots show the number of reads from Illumina sequencing (y-axis) mapped to each genomic coordinate of the reference genome (x-axis). Coverage of 1x, 2x and 3x are marked by dotted lines for reference.

Coverage depth plots for population sequencing at 105, 196 and 252 generations of the 3 lineages of LTMPR1 (A, B, C) evolving in trimethoprim-supplemented (300 ng/mL) media. Coverage depth plots show the number of reads from Illumina sequencing (y-axis mapped) to each genomic coordinate of the reference genome (x-axis). Coverage of 1x, 2x and 3x are marked by dotted lines for reference.

Coverage plots for Line C Isolate 5 and Line B Isolate 2. Line B Isolate 3 showed 5x amplification of a shorter genomic stretch encompassing folA which is shown diagrammatically next to the coverage plot. Line C Isolate 5 showed an expanded GDA encompassing folA compared to the LTMPR1 ancestor, the coordinates for which are below the coverage plot.

Coverage depth plots for population sequencing at 105, 196 and 252 generations of the 3 lineages of LTMPR1 (A, B, C) evolving in increasing trimethoprim concentrations. Coverage depth plots show the number of reads from Illumina sequencing (y-axis) mapped to each genomic coordinate of the reference genome (x-axis). Coverage of 1x, 2x and 3x are marked by dotted lines for reference.