Real-time measurements of aminoglycoside effects on protein synthesis in live cells

Abstract

The spread of antibiotic resistance is turning many of the currently used antibiotics less effective against common infections. To address this public health challenge, it is critical to enhance our understanding of the mechanisms of action of these compounds. Aminoglycoside drugs bind the bacterial ribosome, and decades of results from in vitro biochemical and structural approaches suggest that these drugs disrupt protein synthesis by inhibiting the ribosome's translocation on the messenger RNA, as well as by inducing miscoding errors. So far, however, we have sparse information about the dynamic effects of these compounds on protein synthesis inside the cell. In the present study, we measured the effect of the aminoglycosides apramycin, gentamicin, and paromomycin on ongoing protein synthesis directly in live Escherichia coli cells by tracking the binding of dye-labeled transfer RNAs to ribosomes. Our results suggest that the drugs slow down translation elongation two- to fourfold in general, and the number of elongation cycles per initiation event seems to decrease to the same extent. Hence, our results imply that none of the drugs used in this study cause severe inhibition of translocation.

Keywords: antibiotics; single-molecule tracking; superresolution microscopy; tRNA; translation.

Fig. 1.

Electroporation and tracking of [Cy5]tRNA^Phe in aminoglycoside-treated E. coli cells (A) Microscopy sample preparation. (B) Representation of one [Cy5]tRNA^Phe-binding cycle. The fast diffusion state is assigned to freely diffusing tRNA and tRNA bound to EF-Tu, while the slow state represents ribosome-bound tRNA, comprising two consecutive elongation cycles. (C) Example of segmented E. coli cells and the generated trajectories from the fluorescent particle tracking, fitted using HMM, and course-grained to two diffusional states—fast and slow.

Fig. 2.

Longer and less frequent ribosome binding of [Cy5]tRNA^Phe during aminoglycoside treatment. Single-molecule tracking of Phe-[Cy5]tRNA^Phe in live E. coli cells, untreated or exposed to apramycin, gentamicin, and paromomycin at 10 and 100 µg/mL. Error bars represent bootstrap estimates of SEs. (A) HMM-estimated dwell-time of Phe-[Cy5]tRNA^Phe in the ribosome-bound state. (B) HMM-estimated occupancy of Phe-[Cy5]tRNA^Phe in the ribosome-bound state. (C) Estimated usage frequency of Phe-[Cy5]tRNA^Phe calculated from the occupancy in the ribosome-bound state divided by the corresponding dwell-time.

Fig. 3.

Measurements of the effect of aminoglycosides in the first elongation cycle. Single-molecule tracking of fMet-[Cy5]tRNA^fMet in live E. coli cells, untreated or exposed to aminoglycoside antibiotics at 10 and 100 µg/mL Error bars represent bootstrap estimates of SEs. (A) HMM-estimated dwell-time of fMet-[Cy5]tRNA^fMet in the ribosome-bound state. (B) HMM-estimated occupancy of fMet-[Cy5]tRNA^fMet in the ribosome-bound state. (C) Usage frequency of fMet-[Cy5]tRNA^fMet calculated using the dwell-times and occupancies at the ribosome-bound state.

Fig. 4.

The [Cy5]tRNA^Phe usage relative to [Cy5]tRNA^fMet usage decreases in the presence of aminoglycosides. The usage frequency of [Cy5]tRNA^Phe relative to [Cy5]tRNA^fMet from cells exposed to aminoglycosides at 10 and 100 µg/mL The error bars represent calculated SE of the ratio, propagated from bootstrapped estimates of SEs of measured parameters.

Fig. 5.

The impact of aminoglycosides in the first elongation cycle compared to an average elongation cycle. The first elongation cycle time was obtained by subtracting the subunit joining time (40 ms) calculated from [Cy5]tRNA^fMet tracking in the absence of antibiotics from the [Cy5]tRNA^fMet ribosome-bound dwell-time in cells exposed to the aminoglycosides. The average elongation time was calculated considering that the observed dwell-time of [Cy5]tRNA^Phe in the ribosome-bound state spans two elongation cycles at any point in the mRNA. Error bars represent bootstrap estimates of SEs propagated from the dwell-time error-estimates.