Spatial Distribution and Ribosome-Binding Dynamics of EF-P in Live Escherichia coli

Abstract

In vitro assays find that ribosomes form peptide bonds to proline (Pro) residues more slowly than to other residues. Ribosome profiling shows that stalling at Pro-Pro-X triplets is especially severe but is largely alleviated in Escherichia coli by the action of elongation factor EF-P. EF-P and its eukaryotic/archaeal homolog IF5A enhance the peptidyl transfer step of elongation. Here, a superresolution fluorescence localization and tracking study of EF-P-mEos2 in live E. coli provides the first in vivo information about the spatial distribution and on-off binding kinetics of EF-P. Fast imaging at 2 ms/frame helps to distinguish ribosome-bound (slowly diffusing) EF-P from free (rapidly diffusing) EF-P. Wild-type EF-P exhibits a three-peaked axial spatial distribution similar to that of ribosomes, indicating substantial binding. The mutant EF-P^K34A exhibits a homogeneous distribution, indicating little or no binding. Some 30% of EF-P copies are bound to ribosomes at a given time. Two-state modeling and copy number estimates indicate that EF-P binds to 70S ribosomes during 25 to 100% of translation cycles. The timescale of the typical diffusive search by free EF-P for a ribosome-binding site is τ_free ≈ 16 ms. The typical residence time of an EF-P on the ribosome is very short, τ_bound ≈ 7 ms. Evidently, EF-P binds to ribosomes during many or most elongation cycles, much more often than the frequency of Pro-Pro motifs. Emptying of the E site during part of the cycle is consistent with recent in vitro experiments indicating dissociation of the deacylated tRNA upon translocation.IMPORTANCE Ribosomes translate the codon sequence within mRNA into the corresponding sequence of amino acids within the nascent polypeptide chain, which in turn ultimately folds into functional protein. At each codon, bacterial ribosomes are assisted by two well-known elongation factors: EF-Tu, which aids binding of the correct aminoacyl-tRNA to the ribosome, and EF-G, which promotes tRNA translocation after formation of the new peptide bond. A third factor, EF-P, has been shown to alleviate ribosomal pausing at rare Pro-Pro motifs, which are translated very slowly without EF-P. Here, we use superresolution fluorescence imaging to study the spatial distribution and ribosome-binding dynamics of EF-P in live E. coli cells. We were surprised to learn that EF-P binds to and unbinds from translating ribosomes during at least 25% of all elongation events; it may bind during every elongation cycle.

Keywords: EF-P; binding dynamics; live E. coli; superresolution fluorescence.

FIG 1

(A) Organization of the different domains of the E. coli EF-P protein with C-terminal fusion to mEos2, a fluorophore. The key residue Lys-34 is marked with an arrow. (B) Simple two-state kinetics scheme for EF-P–mEos2 binding to and dissociation from a 70S ribosome with empty E site. Positions of Lys-34 at the N terminus and mEos2 (red star) at the C terminus are depicted schematically. The ribosome-bound species (D_slow) diffuses much more slowly than free EF-P (D_fast). Binding and dissociation rate constants are 1/τ_free and 1/τ_bound, respectively.

FIG 2

(A) Phase-contrast image of an example cell overlaid with trajectories of single EF-P–mEos2 copies. Imaging at 2 ms/frame. (B) (Top) Localization probability density heat map of 2,678 EF-P–mEos2 copies imaged at 2 ms/frame in different cells of 3.5 to 4.5 µm in length. Each location is placed on a common scale of relative axial position. Only molecules that lasted at least 7 frames contribute to the distribution. (Bottom) Distribution of axial projections on the same relative scale. (C) (Top) Localization probability density map of 32,000 ribosome copies (30S labeled by mEos2). Same imaging conditions as t hose for EF-P–mEos2. (Bottom) Distribution of axial projections on the same relative scale.

FIG 3

(A) Phase-contrast image of an example cell overlaid with trajectories of single mutant EF-P^K34A–mEos2 copies. Imaging at 2 ms/frame. (B) (Top) Localization probability density heat map of 4180 EF-P^K34A–mEos2 copies imaged at 2 ms/frame in different cells of 3.5 to 4.5 µm in length. Each location is placed on a common scale of relative axial position. Only molecules that lasted at least 7 frames contribute to the distribution. (Bottom) Distribution of axial projections on the same relative scale. For comparison, axial distribution of molecules uniformly filling a spherocylinder of 0.9-µm diameter and 4-µm length is shown in black. (C) Red, experimental probability distribution of 17,300 single-step displacements taken by EF-P^K34A–mEos2 molecules in 2 ms. Black, the best unconstrained fit to a static two-state model (without transitions). Model parameters: D_slow = 3.2 µm²/s (σ_slow = 50 nm), f_slow = 0.65, D_fast = 9.7 µm²/s (σ_fast = 90 nm), f_fast = 0.35, with χ_ν² = 2.0. The slow and fast components are shown as dashed lines as labeled.

FIG 4

(A) Black, experimental probability distribution of single-step displacements taken by EF-P–mEos2 molecules in 2 ms. Red, the best fit to a static two-state model (without transitions) with D_slow constrained. Model parameters: D_slow = 0.2 µm²/s (σ_slow = 50 nm), f_slow = 0.30, D_fast = 4.3 µm²/s (σ_fast = 75 nm), f_fast = 0.7, with a χ_ν² = 1.0. The individual slow and fast components are shown in dashed lines as labeled. (B) Black, experimental probability distribution of single-step displacements taken by ribosomes (30S labeled with mEos2) in 2 ms. Red, the best unconstrained fit to a static two-state model (without transitions). Model parameters: D_slow = 0.20 µm²/s (σ_slow = 40 nm), f_slow = 0.65, D_fast = 0.8 µm²/s (σ_fast = 75 nm), f_fast = 0.35, with χ_ν² = 2.3. The slow and fast components are shown as dashed lines as labeled.

FIG 5

Black, experimental probability distribution of the mean of six consecutive steps from each trajectory for EF-P–mEos2 molecules. Longer trajectories truncated to 6 steps. Red dashed line, best-fit two-state model with binding-unbinding kinetics. Model parameters: D_slow = 0.2 µm²/s (σ_slow = 50 nm), f_slow = 0.3, D_fast = 4.3 µm²/s (σ_fast = 75 nm), f_fast = 0.7, τ_free = 16 ms, and τ_bound = 7 ms, with χ_ν² = 1.4. Green line, for comparison, a simulated static two-state model (no transitions) using the same diffusion coefficients and fractions gave χ_ν² = 10.1.