Fig 1.
Translation studies with single-molecule resolution.
A) Imaging single-molecule translation dynamics is achieved by the measurement of fluorescence spots that are produced when nascent proteins display epitopes that are recognized by antibody fragments bound to fluorescent probes. The gene construct encodes a 10X FLAG SM tag followed by a protein of interest (POI) and the 24X MS2 tag in the 3’ UTR region. B) Microscopy image showing translation at single-molecule resolution; red spots represent single mRNA, and green spots represent nascent proteins. Below is a representative trace showing the intensity fluctuation dynamics of a single-transcript translating FLAG-10X-KDM5B. C) Simulated time courses representing the characteristic single-molecule fluctuation dynamics. A representative trace is selected and highlighted. At the bottom of the figure is given the normalized auto-covariance function (G(τ)) calculated from simulated time courses. The time at which G(τ) hits zero represents the dwell-time (τFCS). D) Harringtonine inhibits the translation initiation step by binding to the ribosomal 60S subunit. The plot shows the average fluorescence after Harringtonine treatment. Without new initiation events, the fluctuations diminish causing the intensity to drop to zero at time τROA. E) FRAP causes a rapid drop in the fluorescent intensity and a subsequent recovery that is proportional to the time needed by ribosomes to produce new nascent proteins with non-photobleached probes. The bottom plot shows the temporal dynamics of FRAP, where it can be observed by the abrupt decrease in intensity and a recovery time (τFRAP) correlated with the gene length. All simulations correspond to KDM5B for 100 spots and a frame rate of 1 FPS. Error bars represent the standard error of the mean.
Fig 2.
Modeling single-molecule translation.
A) Translation is divided into three main processes: initiation, elongation, and termination. The ribosome footprint represents the physical space occluded by the ribosome, enforcing that no two ribosomes can occupy the same space and time. B) Kymographs represent ribosome movement as a function of time (y-axis) and position (x-axis). Each line represents a single ribosome trajectory. The average slope is proportional to the effective ribosome elongation rate. The plot to the right shows the relationship between ribosome movement and fluorescence intensity, and the plot below shows the ribosome loading at each codon position, calculated as the time-average of ribosome occupancy at the corresponding codon. C) Comparison of the average elongation time (top) and the mean (middle) or variance (bottom) of fluorescence intensity as calculated using the simplified model (Eqs 18 to 21), a linear moments-based model (Eqs 9 to 17), and a full stochastic model (Eqs 1 to 5). Gray area represents previously reported parameter values for ribosome initiation. Panels B and C correspond to simulations for the β-actin. Asterisks represent the specific parameter combination used for Table 1.
Table 1.
Comparing model dynamics.
Fig 3.
Comparing experimental methodologies to estimate ribosome elongation rates.
Elongation rate estimate experiments were simulated for 2,647 human genes, using (A) Fluorescence Correlation Spectroscopy (FCS), (B) Run-Off Assays (ROA), and (C) Fluorescence Recovery After Photobleaching (FRAP). Top panels show the distributions of estimated for long genes (> 1000 codons, n = 658, purple), medium length genes (500 − 1000 codons, n = 1719, blue), and short genes (< 500 codons, n = 270, orange) using 100 mRNA spots for 300 frames at 1/3 FPS. The true elongation rate is denoted by a vertical dashed line. Bottom panels show the RMSE in elongation rate estimation as a function of the number of mRNA spots and the sampling rate. Red boxes highlight all experimental designs that yield a RMSE < 2.0. Asterisks represent the frame rate and number of repetitions used in panel (A). The ‘true’ elongation rate was set at
, and the initiation rate was fixed at ki = 0.03 sec−1 for all simulations.
Fig 4.
Fitting single-molecule data with the full stochastic model.
Experimental data show the fluctuation dynamics of gene constructs encoding an N-terminal 10X FLAG ‘Spaghetti Monster’ SM-tag (green) followed by a protein of interest and finally a 24X MS2 tag (red) in the 3’ UTR region. Three proteins were studied: A) H2B (orange), B) β-actin (blue) and C) KDM5B (violet). Middle figures show the simulated (colors) and measured (black) probability distributions for an mRNA to have a fluorescence intensity corresponding to i units of mature proteins (ump). Right images show the normalized auto-covariance function (G) calculated from experimentally measured (black error bars) and computationally simulated (colors) autocorrelation functions. Error bars in the experimental data and shadow bars in the simulated auto-covariance plots represent the standard errors of the mean. Elongation and initiation rates were obtained by parameter optimization, using the Hooke and Jeeves Algorithm ([29]). Optimized parameters and their uncertainties (see Methods) are provided in Eq 29.
Fig 5.
Ribosome dynamics under experimentally reported initiation and elongation rates.
A) Simulated mean number of codons between ribosomes for the β-actin gene as function of initiation and elongation constants. In the plot, previous literature initiation and elongation values are highlighted by the squares [6–10], and values estimated in this study are denoted by asterisks. B) Simulated number of collisions per ribosome as a function of initiation and elongation constants. C) Top panel, kymograph showing the ribosomal dynamics for SunTag-24X-Kif18b using experimentally determined parameters ki = 1/100 sec−1 and = 3.1 aa/sec [10]. Center panel, kymograph showing the ribosomal dynamics for FLAG-10X-KDM5B using experimentally determined parameters ki = 1/30 sec−1 and
= 10 aa/sec [6]. Bottom panel, kymograph showing the ribosomal dynamics for SunTag-56X-Ki67 using experimentally determined parameters ki = 1/13 sec−1 and
= 13.2 aa/sec [9]. White lines in kymographs represent single ribosome positions, and green spots represent ribosome collisions.
Fig 6.
RNA Sequence to NAscent Protein Simulation (rSNAPsim).
A) rSNAPsim is divided into four upper tabs and three lower tabs. Upper tabs allow the user to select and adjust sequences and then run simulations under varying conditions. Sequence selector allows the user to load a raw text file or GenBank file for their simulation needs. An option to pull from GenBank via accession number is available. All simulation parameters are also set on this tab. B) After a file is loaded, rSNAPsim allows the user to change the tRNA copy numbers and codon types under the Codon Adjustment tab. Post simulation, the lower tabs display simulation information such as average intensity over time of N simulations. C) Screen-shot of a kymograph. The kymograph tab allows the user to create their kymographs with varying display options. The Stochastic Simulation tab shows the time course data from the selected simulations. The Fluorescence Correlation Spectroscopy tab displays and compared simulated and experimental single-molecule translation dynamics, the auto-covariance function, and biophysical parameters, such as the elongation constant or ribosomal density. All functionality in the GUI is also available in a command-line module for Python included with rSNAPsim.