Modulation of Global Transcriptional Regulatory Networks as a Strategy for Increasing Kanamycin Resistance of the Translational Elongation Factor-G Mutants in Escherichia coli

Abstract

Evolve and resequence experiments have provided us a tool to understand bacterial adaptation to antibiotics. In our previous work, we used short-term evolution to isolate mutants resistant to the ribosome targeting antibiotic kanamycin, and reported that Escherichia coli develops low cost resistance to kanamycin via different point mutations in the translation Elongation Factor-G (EF-G). Furthermore, we had shown that the resistance of EF-G mutants could be increased by second site mutations in the genes rpoD/cpxA/topA/cyaA Mutations in three of these genes had been discovered in earlier screens for aminoglycoside resistance. In this work, we expand our understanding of these second site mutations, the goal being to understand how these mutations affect the activities of the mutated gene products to confer resistance. We show that the mutation in cpxA most likely results in an active Cpx stress response. Further evolution of an EF-G mutant in a higher concentration of kanamycin than what was used in our previous experiments identified the cpxA locus as a primary target for a significant increase in resistance. The mutation in cyaA results in a loss of catalytic activity and probably results in resistance via altered CRP function. Despite a reduction in cAMP levels, the CyaA^N600Y mutant has a transcriptome indicative of increased CRP activity, pointing to an unknown role for CyaA and / or cAMP in gene expression. From the transcriptomes of double and single mutants, we describe the epistasis between the mutation in EF-G and these second site mutations. We show that the large scale transcriptomic changes in the topoisomerase I (FusA^A608E-TopA^S180L) mutant likely result from increased negative supercoiling in the cell. Finally, genes with known roles in aminoglycoside resistance were present among the misregulated genes in the mutants.

Keywords: aminoglycosides; antibiotic resistance; gene regulatory networks; kanamycin; transcription factors.

Figure 1

Kanamycin resistance of mutants. (A) Boxplots showing distributions of MICs of kanamycin of the wild type (WT) and various mutants. The number of replicates is mentioned over each boxplot. All mutants, except RpoD^L261Q, CpxA^F218Y, and CyaA^N600Y, are significantly more resistant than the wild type (Welch two sample t-test, P < 10⁻⁷). Although the medians of the FusA^A608E-RpoD^L261Q/CpxA^F218Y/TopA^S180L/CyaA^N600Y double mutants tend to be higher than that of the FusA^P610T mutant, this difference is not statistically significant (P > 0.09). The difference between the medians of these double mutants and the FusA^A608E mutant are statistically significant (P < 0.02), except in the case of the FusA^A608E-TopA^S180L mutant (P = 0.061) (B) Growth of mutants in 8-kan. (C) Growth of the RpoD^L261Q, CpxA^F218Y, and CyaA^N600Y mutants in 1.5-kan. In (B) and (C), error bars represent SD of eight replicates. Some part of this data had been generated in our previous work (Mogre et al. 2014), with the inclusion of more replicates and data of single mutants.

Figure 2

Activation of the Cpx response results in resistance. Growth curves in 8-kan (A) and 0-kan (B). The labels for the x and y axis are common. Plotted are the means from eight replicates with error bars representing SD. (C) Fold changes in MICs of populations evolved in 15-kan over the MICs of populations evolved in 0-kan. Two replicate populations grown in either 0-kan or 15-kan are shown. MICs of evolving populations at the end of growth (24 hr) after each batch transfer were determined and are represented by P0-P5. The MIC of the 0-kan 1 population at P0 was used to calculate fold changes. Error bars represent SD of four replicates. (D) Heatmaps showing the abundance of variants revealed by sequencing of both control and evolved populations. The color represents the percentage of reads supporting mutations, and is an approximate proxy for the abundance of the mutation. The list of variants was trimmed such that only mutations present in >20% of the reads in at least one sample were retained. (E) Heatmaps showing abundance of all low-frequency cpxA variants. The list of variants was trimmed to include only mutations in cpxA. The mutations shown in (D) are not shown here. The color scale is as shown in (D).

Figure 3

Inactivation of adenylate cyclase results in resistance. (A) Boxplots showing the distribution of estimates of cellular cAMP concentrations of strains in the exponential and stationary phase. The difference between the wildtype and the other mutants are significant (P <1  ×  10⁻²). P values for other relevant comparisons are mentioned in the plot. The FusA^A608E-CyaA^N600Y mutant is referred to as fusA-cyaA and the CyaA^N600Y mutant as cyaA. (B and C) Growth curves in 8-kan (B) and 0-kan (C). The labels for the x and y axis are common. Plotted are the means from eight replicates with error bars representing SD. In the 8-kan growth curves, the huge error bars in some of the strains are produced when a few replicates start growing, possibly due to acquisition of some resistance conferring mutation, and thus this error cannot be eliminated. (D and E) Scatter plots comparing log₂ fold-changes of genes in the CyaA^N600Y mutant with those in the ΔcyaA (D)/Δcrp (E) knockout strains. The mutant is referred to by its gene name for brevity. The time-points for cell harvesting for RNA extraction of the ΔcyaA/Δcrp strains were similar to that of the mutants. The Spearman correlation coefficient and its P value are mentioned. (F) Barplots showing the number of targets of CRP present among the upregulated and downregulated genes in the CyaA^N600Y, ΔcyaA and Δcrp strains. The numbers of positive (+), negative (−) and dual targets (+−) of CRP present among the upregulated (red) and downregulated (blue) genes are shown in the stacked barplots. Similar results are seen with the FusA^A608E-CyaA^N600Y mutant and are shown in Figure S9 in File S2.

Figure 4

Evidence for supercoiling changes in the FusA^A608E-TopA^S180L mutant. (A) Gel picture showing mobility of pUC18 topoisomers on agarose gel containing 2.5 µg/ml chloroquine. Positions of negatively supercoiled and relaxed forms of the plasmid are indicated by a schematic. (B) Scatter plot showing correlation of log₂ fold changes of genes in the FusA^A608E-TopA^S180L mutant with microarray derived gene-expression ratios obtained by inhibiting DNA gyrase function using 20 µg/ml novobiocin (data obtained from Peter et al. 2004). For a detailed comparison with the Peter et al. (2004) dataset, refer to Figure S13 in File S2.

Figure 5

Summary of differentially expressed genes across mutants. (A) Numbers of upregulated and downregulated genes in the mutants. Mutants are referred to by their gene names for brevity. (B) Heatmap showing the matrix of Spearman correlations among mutants. Fold changes of all genes were used to derive these correlations. (C) Heatmap showing enriched GO terms among the upregulated and downregulated genes in the mutants. Many GO terms have been combined to give this simplified picture.

Figure 6

Effect of second site mutations on gene expression and dependence on EF-G. Scatter plots comparing log₂ fold changes of differentially expressed genes among mutants. Mutants are referred to by their gene names for brevity. Gray zones indicate the region between the log₂ fold changes of +1 and −1 (corresponding to fold changes of 2 and 0.5), and thus highlights the region of low/no fold change. Red points show genes upregulated and blue points show genes downregulated in the relevant second site single mutant. (A–C) Fold changes of genes differentially expressed in the CyaA^N600Y mutant were compared with the FusA^A608E-CyaA^N600Y and FusA^A608E mutants. (D–F) Fold changes of genes differentially expressed in the CpxA^F218Y mutant were compared with the FusA^A608E-CpxA^F218Y and FusA^A608E mutants. (G–I) Fold changes of genes differentially expressed in the RpoD^L261Q mutant were compared with the FusA^A608E-RpoD^L261Q and FusA^A608E mutants.