Impact of synonymous mutations in the blaTEM-3 gene on gene expression and Escherichia coli fitness

TL;DR

Synonymous mutations in a gene for antibiotic resistance affect bacterial fitness and resistance, possibly through epigenetic or recombination mechanisms.

Contribution

Demonstrates that synonymous mutations in a β-lactamase gene influence Escherichia coli fitness and resistance, beyond transcriptional effects.

Findings

Synonymous mutations in blaTEM-3 alter bacterial fitness and resistance in antibiotic presence.

Cells with synonymous mutations outcompete others under antibiotic stress but are outcompeted in its absence.

Transcriptional and mRNA structural differences are not responsible for the observed effects.

Abstract

Synonymous mutations (SMs) have a significant effect on the expression of the β-lactamase TEM-3 and the resistance phenotype conferred. Cell-free transcription did not reproduce this effect. The amount of mRNA obtained was identical for blaTEM-3 transcripts with or without SMs, and the transcripts were indistinguishable in terms of secondary structure and thermal de- and renaturation, as shown by circular dichroism experiments. In asymmetric competition assays, cells expressing the gene with SMs replaced cells expressing the gene without SMs after five days of passaging in a low concentration of ceftazidime. However, without antibiotic present, cells expressing the gene without SMs replaced cells expressing the gene with SMs. These experiments indicate that either post-transcriptional or in cellulo processes are responsible for the differences in expression levels and fitness. Possible mechanisms to be tested in future studies include epigenetic effects (different degrees of methylation) and the removal of the gene in the absence of β-lactam by recombination. Although the phagemid used in this study has the ampicillin resistance gene removed and replaced with a chloramphenicol resistance gene, the 5′-terminal portion of the blaTEM-1a gene is still present and might account for deletion by recombination. The question would then become why recombination is observed more in the phagemids with the gene without SMs.

Most studies focusing on the evolution of antibiotic resistance focus on nonsynonymous mutations (NMs) in genes encoding proteins that cause resistance, such as β-lactamases. NMs result in amino acid changes that are often responsible for an extended substrate spectrum, insensitivity to inhibitors, or suppression of destabilizing effects introduced by other NMs. Although these mutations are very important, comparisons of natural sequences as well as directed evolution experiments have also identified synonymous mutations (SMs), but their impact is rarely studied. Here, we provide evidence that SMs in a β-lactamase gene alter the fitness levels conferred to Escherichia coli cells. Our results suggest that transcription efficiency is not affected, but that epigenetic factors could affect the stability of the gene inside the bacteria. Translation and protein folding efficiency could also not be ruled out thus far. Studying SMs will improve our understanding of how antibiotic resistance evolves and help us combat it.

Figures

(A) Agarose gel electrophoresis of transcripts obtained after cell-free transcription of the -sm and +sm genes (left side), RNA standard in the middle, and transcripts after storage at −20°C for six months and several freeze-thaw cycles (right side). (B) Circular dichroism (CD) scans of the -sm and +sm transcripts at 25°C and 100°C. (C) Thermal denaturation (from 25°C to 100°C) and renaturation (from 100°C to 25°C) of the two transcripts measured by CD signal at 270 nm.

Results of asymmetric competition assays with ceftazidime (A) and without antibiotic (B). At day 0, the assays were started with either 1% -sm and 99% + sm cells or vice versa. At day 5, cells were harvested and plated on agar plates. Ten colonies were randomly isolated and identified based on their blaTEM-3 sequence. All experiments were carried out three times except the experiment with 2-week-old cells in panel B, which was done once. Labels of the lines in the graphs indicate the initial conditions. The results for 1% -sm in panel A and 1% + sm in panel B (horizontal lines) represent 6 (3 with fresh and 3 with 4-week-old cells) and 7 (3 with fresh, 1 with 2-week-old, and 3 with 4-week-old cells) experiments, respectively. Means ± standard deviations are shown. *In some of these cells, the -sm gene was deleted as explained in the text.

Full text

OBSERVATION

In a previous report about synonymous mutations (SMs) in three genes encoding TEM β-lactamase variants, SMs had the greatest effect in blaTEM-3 (1). TEM-3 expression confers the 2be phenotype, which is characterized by resistance to extended-spectrum β-lactams, such as third-generation cephalosporins and aztreonam (2, 3). Western blots indicated TEM-3 expression in Escherichia coli DH10B cells from its published gene (4), which includes four SMs relative to the blaTEM-1a gene (5) (Fig. S1 in the Supplemental Material), was more than fourfold higher than when it was expressed from an artificial gene containing only nonsynonymous mutations (NMs). The former is designated as the blaTEM-3(+sm) gene and the latter as the blaTEM-3(-sm) gene. In disc diffusion assays with E. coli cells expressing these genes, blaTEM-3(+sm) resulted in smaller diameters than the blaTEM-3(-sm) gene. The minimum inhibitory concentration (MIC) of ceftazidime was 64 times higher with the blaTEM-3(+sm) gene, while no significant difference was observed with the other antibiotics tested.

Two avenues were pursued to decipher the underlying mechanisms responsible for the different expression and resistance levels in E. coli. Firstly, cell-free transcription and analysis of the transcript by circular dichroism (CD) spectropolarimetry were carried out to discern if transcription efficiency or stability of the transcript was affected. Cell-free transcription (MEGAscript Transcription Kit, ThermoFisher) did not result in different amounts of transcript (Fig. 1A), and the transcripts appeared to degrade at equal rates as assessed by agarose gel electrophoresis of transcript after six months of storage at −20°C and several freeze-thaw cycles (Fig. 1A). CD scans of the two transcripts at 25 and 100°C, respectively, are superimposable, indicating no difference in secondary structure (Fig. 1B), and thermal denaturation measured by CD at 270 nm was identical and completely reversible (Fig. 1C). Cell-free translation from these transcripts and coupled transcription/translation was attempted but did not yield detectable amounts of TEM-3 possibly due to insufficient formation of a disulfide bond present in TEMs (6) and improper protein folding.

(A) Agarose gel electrophoresis of transcripts obtained after cell-free transcription of the -sm and +sm genes (left side), RNA standard in the middle, and transcripts after storage at −20°C for six months and several freeze-thaw cycles (right side). (B) Circular dichroism (CD) scans of the -sm and +sm transcripts at 25°C and 100°C. (C) Thermal denaturation (from 25°C to 100°C) and renaturation (from 100°C to 25°C) of the two transcripts measured by CD signal at 270 nm.

*Results of asymmetric competition assays with ceftazidime (A) and without antibiotic (B). At day 0, the assays were started with either 1% -sm and 99% + sm cells or vice versa. At day 5, cells were harvested and plated on agar plates. Ten colonies were randomly isolated and identified based on their blaTEM-3 sequence. All experiments were carried out three times except the experiment with 2-week-old cells in panel B, which was done once. Labels of the lines in the graphs indicate the initial conditions. The results for 1% -sm in panel A and 1% + sm in panel B (horizontal lines) represent 6 (3 with fresh and 3 with 4-week-old cells) and 7 (3 with fresh, 1 with 2-week-old, and 3 with 4-week-old cells) experiments, respectively. Means ± standard deviations are shown. In some of these cells, the -sm gene was deleted as explained in the text.

Another avenue is to study the cellular background in more detail. Harboring vectors with the different genes could affect growth independent of antibiotics. Growth curves of E. coli DH10B cells harboring either pBC SK(+)-blaTEM-3(-sm) or pBC SK(+)-blaTEM-3(+sm), an empty pBC SK(+) vector, or no vector, in Müller-Hinton broth (MHB) showed no significant difference in growth (data not shown). To assess the fitness conferred by either expression of blaTEM-3(-sm) or blaTEM-3(+sm) in the presence of a small concentration (31 ng/mL or 10^−9^ times MIC) of ceftazidime, an asymmetric competition (AC) assay (7) was adapted. In short, starting with mixtures of cells expressing either the -sm or +sm genes in ratios of 1:100 or 100:1, cultures were grown in MHB in the presence of 31 ng/mL ceftazidime and passaged once per day for five days (1:50 dilution). At the end of the assay, cells were diluted and plated on agar plates (200 colonies target, actual ~50–500 colonies). Ten colonies from each plate were randomly isolated, their plasmids were extracted, and the blaTEM-3 genes were sequenced. Under these conditions, cells expressing blaTEM-3(-sm) started at 1% never replaced cells expressing blaTEM-3(+sm). In contrast, cells expressing blaTEM-3(+sm) starting at 1% constituted ~23% of colonies isolated at the end of the experiment (Fig. 2A). Interestingly, when agar plates with colonies were kept at 4°C for four weeks and then used in the same assay, cells expressing blaTEM-3(+sm) constituted almost all (~97%) of the colonies isolated. To test if the greater fitness of cells expressing blaTEM-3(+sm) was due to ceftazidime resistance, the same experiment was repeated without antibiotic present (Fig. 2B). This resulted in the reverse outcome. Cells expressing the +sm gene at an initial 1% never replaced cells expressing the -sm gene. With freshly transformed cells, about 37% of colonies picked at the end of the experiment that started with 1% -sm cells contained the plasmid encoding the -sm gene. With cells from agar plates kept at 4°C for two and four weeks, but otherwise the same conditions, this percentage increased to 70% and 100%, respectively. In the absence of antibiotics, the increased expression level of TEM-3 from the blaTEM-3(+sm) seems to lower fitness. Some of the cells also contained genes truncated after 210 nucleotides with an a109c NM, encoding a fragment with Q39—as in TEM-1—rather than K39, which is found in TEM-3. Upon closer examination, these truncated genes seem to be relics of the partially removed blaTEM-1a gene that occurred when pBC SK(+) was generated from pBluescript II (Agilent Technologies) and the blaTEM-3(-sm) gene was deleted, likely through homologous recombination.

To investigate this phenomenon further, Epsilometer tests (E-tests) (8) were performed. The results with old cells (Fig. S2A) resemble previous findings from disc diffusion assays (1), and the results for the -sm cells were comparable to the negative controls harboring the same plasmid without blaTEM-3 genes. Cells from the agar plates used in these tests were isolated, their plasmids extracted, and the blaTEM-3 genes sequenced. All the -sm plasmids had the -sm gene removed, as described above for the AC assay, while the +sm plasmids were intact. The experiment was repeated with freshly transformed cells selected for chloramphenicol resistance encoded on pBC SK(+) plus carbenicillin to ensure full-length blaTEM-3 genes. In these E-tests, the -sm cells were comparable to the +sm cells (Fig. S2B). Sequencing of the plasmids harbored by these cells confirmed that the full-length blaTEM-3 genes were still present. Further studies on why blaTEM-3(-sm) is apparently more prone to deletion than blaTEM-3(+sm) or blaTEM-1a are underway.

These results confirm that SMs in the blaTEM-3 gene increase antibiotic resistance (1). Cell-free transcription demonstrates that neither transcription efficiency nor transcript stability is responsible for this effect. The facts that the transcripts are indistinguishable in terms of secondary structure and thermal stability, as assessed by CD, and that no rare codons are used (1) suggest that translation is also unlikely to be a limiting factor. The -sm gene in the pBC SK(+) plasmids seems to be less stable than the +sm gene when there is no β-lactam selection. This could be due to epigenetic effects (9, 10). In this context, it is interesting to note that both blaTEM-1a and blaTEM-3(+sm) contain seven 5′-gatc-3′ sites for methylation by DNA adenine methyltransferase (Dam), while blaTEM-3(-sm) only contains six. Future studies will investigate the molecular mechanisms underlying the differential deletion of the -sm versus +sm blaTEM-3 genes, as well as the effects of SMs in cells and/or vectors that eliminate possible recombination.

Supplementary Material

Reviewer comments