Multi-Color Single-Molecule Imaging Uncovers Extensive Heterogeneity in mRNA Decoding

Abstract

mRNA translation is a key step in decoding genetic information. Genetic decoding is surprisingly heterogeneous because multiple distinct polypeptides can be synthesized from a single mRNA sequence. To study translational heterogeneity, we developed the MoonTag, a fluorescence labeling system to visualize translation of single mRNAs. When combined with the orthogonal SunTag system, the MoonTag enables dual readouts of translation, greatly expanding the possibilities to interrogate complex translational heterogeneity. By placing MoonTag and SunTag sequences in different translation reading frames, each driven by distinct translation start sites, start site selection of individual ribosomes can be visualized in real time. We find that start site selection is largely stochastic but that the probability of using a particular start site differs among mRNA molecules and can be dynamically regulated over time. This study provides key insights into translation start site selection heterogeneity and provides a powerful toolbox to visualize complex translation dynamics.

Graphical abstract

Figure 1

Development of the MoonTag, a Fluorescence Labeling System to Visualize Translation of Single mRNAs (A) Schematic representation of the nanobody peptide labeling system. (B and C) Representative images of HEK293T cells transfected with the indicated constructs. (D) The indicated constructs were transfected in U2OS cells stably expressing the GFP-tagged MoonTag nanobody. The GFP:mCherry fluorescence intensity ratio on mitochondria was quantified. Each dot represents a single cell, and lines indicates the average. (E) Schematic of the translation reporter (top panel) and nascent polypeptide labeling strategy using the MoonTag system (bottom panel). (F) Representative image of U2OS cells expressing MoonTag-Nb-GFP and PCP-mCherry-CAAX and the MoonTag translation reporter shown in (E). (G and H). Schematic of reporters (top) and representative images of Moon/Sun cells expressing the indicated reporters (bottom). In (G), the arrowhead and arrow indicate SunTag and MoonTag translation, respectively. In (H) the arrowhead indicates mRNA with 3′ UTR translation. (I and J) Moon/Sun cells were transfected with the reporter indicated in (H), and MoonTag and SunTag intensities on single mRNAs were tracked over time. (I) Boxplot indicating the frequency of 3′ UTR translation (percentage of ribosomes) calculated for each mRNA. The dashed line represents the median value, the box indicates the 25%–75% range, and whiskers indicate the 5%–95% range. (J) Example dual-color intensity traces of two mRNAs with a MoonTag (blue) and SunTag signal (green). The number of experimental repeats and mRNAs analyzed per experiment are listed in Table S1. See also Videos S1, S2, and S3. Scale bars: 5 μm (B), 10 μm (C), and 1 μm (F–H).

Figure S1

Development and Characterization of the MoonTag System, Related to Figure 1 (A-C) U2OS cells stably expressing MoonTag nanobody-GFP were transfected with 12xMoonTag-H2B-mCherry (A), 12xMoonTag-CAAX and mCherry-CAAX (B), or 12xMoonTag-Mito-mCherry and MoonTag nanobody-Halo^JF646 (C). Representative cells are shown. Scale bars, 10μm (A, B), and 5μm (C). (D) Moon/Sun cells expressing the reporter indicated in Figure 1H. Correlation between the main coding sequence translation initiation rate and 3′ UTR translation frequency on single mRNAs is shown. Every dot represents a single mRNA and line depicts moving average over 15 mRNAs. Number of experimental repeats and mRNAs analyzed per experiment are listed in Table S1.

Figure 2

The MashTag: A Reading Frame Sensor to Visualize Translation Start Site Selection (A) Schematic of the MashTag design (B, C, and F–H) Schematic of MashTag translation reporters (top panels). Circles with triangles and squares represent start and stop sites, respectively. The colors of the filled circles indicate the reading frame (blue, MoonTag reading frame; green, SunTag reading frame). For simplicity, 24xPP7 sites in the 3′ UTR are not depicted in (F–H). (B and C) Representative images of mRNAs in Moon/Sun cells expressing the indicated translation reporters prior to (top images) and after (bottom images) 5 min incubation with the translation inhibitor puromycin. (D) Moon/Sun cells expressing the MashTag translation reporter shown in (B). The asterisk indicates an mRNA with OOF translation. Arrowheads indicate the OOF SunTag signal. Time is indicated in minutes. (E) Schematic of OOF translation on the MashTag reporter. (F) Theoretical intensity trace of a single ribosome translating a MashTag reporter mRNA. (G and H) Fluorescence intensities of single ribosomes translating the reporter mRNA in the SunTag frame, either when the SunTag is in the main frame (SunStart reporter, G) or is OOF (MoonStart reporter, H). Intensity traces are aligned at the last time point that contains a SunTag signal (i.e., just before translation termination). Solid lines indicate experimentally derived values; shaded areas surrounding solid lines indicate SEM. Dashed lines in (F–H) indicate the expected single-ribosome intensity trace of the SunTag reading frame. The number of experimental repeats and mRNAs analyzed per experiment are listed in Table S1. See also Video S4. Scale bars, 1 μm.

Figure S2

Quantifying Translation Dynamics Using the MashTag Systems, Related to Figure 2 (A-F) Indicated reporters were expressed in Moon/Sun cells. A-B) Representative images of mRNAs in Moon/Sun cells expressing the indicated translation reporters. (C) Distribution of mCherry intensities of mRNAs associated with MoonTag signal only (blue bars) or mRNAs associated with both MoonTag and SunTag signal (orange bars). (D) Normalized SunTag intensity on mRNAs after harringtonine treatment. Grey lines depict selected single mRNA intensity traces and the black line shows the average of all mRNAs. Red line indicates harringtonine addition. E) Distribution of the intensity of mature proteins expressed from the SunTag frame. Mature proteins were tethered to the plasma membrane through a CAAX motif. Note that identical imaging settings were used to measure mature protein intensities plotted in (E) and translation site fluorescence intensity traces. (F) Distribution of the intensity of mature proteins expressed in the SunTag frame. SunTag proteins were expressed either from the main reading frame (green) or as OOF translation protein products (orange). Mature proteins are tethered to the membrane through a CAAX domain encoded in the SunTag frame. P-values are based on a two-tailed Student’s t-test. Number of experimental repeats and mRNAs analyzed per experiment are listed in Table S1. Scale bars, 1 μm.

Figure 3

A Computational Pipeline to Quantitatively Interpret Fluorescence Signals (A) Schematic illustrating how the fluorescence originating from multiple ribosomes translating an mRNA together generates a fluorescence intensity trace. Each color represents a single translating ribosome. Triangles indicate the moment of translation initiation. (B and C) Schematics of translation reporters (top). For simplicity, 24xPP7 sites in the 3′ UTR are not depicted. The black start or stop sites in (C) indicate that only a single reading frame contains MoonTag or SunTag peptides. (B) An example dual-color intensity trace of a single MoonStart-MashTag mRNA with a MoonTag (top panel) and SunTag (center panel) signal. Dashed lines indicate experimentally observed intensities, and solid lines display the optimal fit. Colored triangles below the x-axes of the top and middle graphs represent translation initiation events. The bottom panel shows ribosome occupancy per reading frame over time as determined by RiboFitter. (C) Boxplots indicating the relative percentage of ribosomes translating the SunTag frame on single mRNAs of the reporter mRNAs indicated above. The dashed line represents the median value, the box indicates the 25%–75% range, and whiskers indicate the 5%–95% range. The number of experimental repeats and mRNAs analyzed per experiment are listed in Table S1.

Figure S3

Normalizing and Quantifying MashTag Fluorescence, Related to Figure 3 (A-F) Schematics of translation reporters (top). For simplicity, 24xPP7 sites in the 3′UTR are not depicted. Moon/Sun cells were transfected with indicated reporters. (A) Correlation between total cell MoonTag intensity and average MoonTag translation signal on mRNAs. (B) Correlation between total cell SunTag intensity and average SunTag translation signal on mRNAs. (C) Distribution of the intensity of the MoonTag translation signal of individual mRNAs. (D) An example intensity track (dashed line) and fit (solid line) are shown for the MoonTag signal after one iteration (top) or 1000 iterations (bottom) of fit optimization. Colored triangles below the x-axes represent translation initiation events. Each row of triangles illustrates an independent round of fitting. Corresponding root mean squared error (RMSE) values for each round of fitting are shown. (E-F) RMSEs after indicated number of iterations of fit optimization are shown for three representative mRNAs. Number of experimental repeats and mRNAs analyzed per experiment are listed in Table S1.

Figure 4

Heterogeneity in Translation Start Site Selection among Different mRNA Molecules (A–F) The indicated reporters were transfected into Moon/Sun cells, and MoonTag and SunTag intensities on single mRNAs were tracked over time. For simplicity, 24xPP7 sites in the 3′ UTR are not depicted. (A) Boxplot indicating the relative percentage of ribosomes translating the SunTag frame on single mRNAs. The dashed line represents the median value, the box indicates the 25%–75% range, and whiskers indicate the 5%–95% range. (B) Example graphs of four representative mRNAs in which the number of ribosomes in each reading frame is plotted over time of the reporter indicated in (A). The percentages of SunTag ribosomes on each mRNA are shown (% OOF). (C) p-values for enrichment of ribosomes translating either the SunTag or MoonTag frame on individual mRNAs. Every dot represents a single mRNA (left graph). The color of the dot indicates the reading frame that is enriched. Also shown are example traces of single mRNAs that show enrichment of ribosomes translating either the SunTag or MoonTag frame (right graphs). (D) Correlation between overall translation initiation rate and relative SunTag frame translation frequency for individual mRNAs of the reporter indicated in (A). Every dot represents a single mRNA, and the line depicts the moving average over 15 mRNAs. (E) Linear regression analysis of MoonTag and SunTag intensities for the indicated reporter mRNAs (left graph). An example trace of one mRNA is shown (right bottom graphs) with the indicated R² value. (F) Sliding window analysis (see Figures S4H and S4I for details) of initiation events in MoonTag and SunTag reading frames on mRNAs of the reporter indicated in (A). Every dot depicts the strongest p-value of a single mRNA (left graph). Example traces show the number of ribosomes in each reading frame over time (top right graphs) and corresponding sliding window p values (bottom right graphs). The number of experimental repeats and mRNAs analyzed per experiment are listed in Table S1.

Figure S4

Heterogeneity in Translation Start Site Selection among Different mRNA Molecules Expressed from a Single Genomic Locus, Related to Figure 4 The reporter indicated in Figure 4A was expressed in Moon/Sun cells (A-D, F-J). (A) The median frequency of ribosomes translating the SunTag reading frame as determined by either the variable (histogram) or constant (red dashed line) AUC fitting approach (See STAR Methods). B) Northern blot with probes against the BGH polyadenylation sequence. RNA was extracted from either parental Moon/Sun cells or from a polyclonal Moon/Sun cell line in which the targeting construct [P2A-Puro-BGH]-[TetOn-MoonStart-MashTag reporter] was integrated into the AAVS1 safe harbor locus (in the PPP1R12C gene). The indicated band represents an mRNA encoding the 5′ part of PPP1R12C fused to P2A-puro-BGH. 28S rRNA is shown as a loading control (bottom). C, D, F, J) Comparison between MoonStart-MashTag mRNAs either expressed from transiently transfected plasmids (replotted from Figures 4A, 4C, 4E, and 4F) or expressed from the AAVS1 genomic locus. (C) Boxplot indicates the relative percentage of ribosomes translating the SunTag frame on single mRNAs. Dashed line represents median value, box indicates 25%–75% range, and whiskers indicated 5%–95% range. P-value is based on two-tailed Mann-Whitney test. (D) P-values for enrichment of ribosomes translating either the SunTag or MoonTag frame on individual mRNAs. Every dot represents a single mRNA. The color of the dot indicates the reading frame that is enriched. Dashed red line indicates p = 0.01. (E) Example intensity traces of MoonTag and SunTag on an mRNA of the Moon-SunTag reporter, which illustrates why the linear regression analysis might underrepresent fluorescence intensity correlation over time between MoonTag and SunTag signals; 1) the MoonTag signal appears slightly earlier than the SunTag signal on the Moon-SunTag reporter (left bottom graph) due to its upstream position in the reporter; 2) a poor correlation is obtained if both signals remain largely constant over time (right bottom graph). R² values are shown for each graph, as determined by linear regression analysis. (F) Linear regression analysis of MoonTag and SunTag intensities. Each dot represents a single mRNA molecule. (G) Cells were treated with 200 μg/ml cycloheximide for 1 min and imaged at indicated time-intervals. MoonTag fluorescence intensity was measured over time. Mean and SD of the intensity was calculated for 6 consecutive time points and the mean/SD is plotted. (H) Schematic of sliding window analysis approach. First, ribosomes were fit to raw intensity traces, and the time of each translation initiation event was determined (indicated by triangles under the x-axis of top graph), as described in Figure 3B (top). Initiation events in both MoonTag and SunTag frames were then merged onto a single time-line. For each consecutive set of initiation events (window length of 10 initiation events is shown), a p-value was calculated using a binomial test, which represents the probability of observing the ratio between MoonTag and SunTag initiation events within that window, based on the MoonTag and SunTag translation initiation rates of the entire mRNA trace (middle). The p-value for each window of 10 consecutive initiation events of an mRNA was plotted and the strongest p-value per mRNA was determined (bottom). The mRNA shown here was also used as an example in Figure 4F. (I) Example graphs showing the number of ribosomes in each reading frame over time for an example mRNAs (left panel) and corresponding sliding window p-value graphs (right) for sliding windows with indicated number of initiation events per window. (J) Sliding window analysis of initiation events in MoonTag and SunTag reading frames. Every dot depicts the strongest p-value of a single mRNA (left graph). Example traces of the number of ribosomes in each reading frame over time (right graphs) with corresponding p-values (colored dots). Dashed red line indicates p = 0.05. Number of experimental repeats and mRNAs analyzed per experiment are listed in Table S1.

Figure 5

Alternative Start Site Selection Contributes to OOF Translation (A) Schematic of different possible translation paths of individual ribosomes on a MashTag mRNA. (B–D) For simplicity, the reporter schematics only indicate the 5′ region of the mRNA. The indicated reporters were transfected into Moon/Sun cells, and MoonTag and SunTag intensities were tracked over time on single mRNAs. The boxplots indicate the relative percentage of ribosomes translating the SunTag frame on single mRNAs. The p-values are based on a two-tailed Mann-Whitney test: ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. For comparison, data shown in dark blue is are re-plotted from Figure 4A. The dashed line represents the median value, the box indicates the 25%–75% range, and whiskers indicate the 5%–95% range. The number of experimental repeats and mRNAs analyzed per experiment are listed in Table S1.

Figure S5

Translation Initiation Dynamics on 5′ UTRs, Related to Figure 5 (A-C) Schematics of translation reporters (right). For simplicity, only the 5′ region of the mRNA is shown. Indicated reporter mRNAs were expressed in Moon/Sun cells. Boxplots represent the overall translation initiation rates (i.e., initiation rates of MoonTag and SunTag frames combined) (A, C) or percentage of ribosomes translating the SunTag frame (B) for single mRNAs. P-values are based on two-tailed Mann-Whitney tests; ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.001; ^∗∗∗∗p < 0.0001. Dashed line represents median value, box indicates 25%–75% range, and whiskers indicated 5%–95% range. Number of experimental repeats and mRNAs analyzed per experiment are listed in Table S1.

Figure 6

A Single-Molecule uORF Sensor Based on the MashTag (A, B, and F) Schematics of translation reporters (top). For simplicity, 24xPP7 sites in the 3′ UTR are not depicted. (A) Schematic of different possible ORFs that can be translated on a uORF-MashTag mRNA. (B) Fraction of ribosomes undergoing each translation path. The thickness of the lines reflects the relative usage frequency. Solid lines indicate translation, dashed black lines indicate ribosome scanning, and the dashed gray line indicates ribosome dissociation from the mRNA. Colored numbers at branchpoints indicate the relative fraction of ribosomes that follow each path. The red vertical line indicates non-canonical start sites in any of the three frames in the MashTag. (C–G) MashTag reporters were transfected into Moon/Sun cells, and MoonTag and SunTag intensities on single mRNAs were tracked over time. (C) Cells were either untreated (top) or treated with puromycin for 5 min (bottom); representative images are shown. Scale bar, 1 μm. (D) Example graphs of single mRNAs of the number of ribosomes translating either the SunTag or MoonTag frame over time in cells expressing the reporter indicated in (A). (E) p-values for enrichment of ribosomes translating either the SunTag or MoonTag frame on individual mRNAs. Every dot represents a single mRNA (left graph). The color of the dot indicates the reading frame that is enriched. Also shown are example traces of single mRNAs that show enrichment of ribosomes translating either the SunTag or MoonTag (right graphs). p-values are indicated for example traces. (F) Linear regression analysis of MoonTag and SunTag intensities for the indicated reporter mRNAs (left graph). For comparison, data shown in brown are replotted from Figure 4E. Example graphs of two mRNAs are shown with indicated R² values (right graphs). (G) Sliding window analysis of initiation events in MoonTag and SunTag reading frames on mRNAs of the reporter indicated in (A). Every dot depicts the strongest p-value of a single mRNA (left graph). Example traces are shown of the number of ribosomes in each reading frame over time (top right graphs) and corresponding sliding window p-values (bottom right graphs). The number of experimental repeats and mRNAs analyzed per experiment are listed in Table S1.

Figure S6

Translation Initiation Dynamics on a uORF-Containing mRNA, Related to Figure 6 (A-D) Schematics of translation reporters (top). For simplicity, 24xPP7 sites in the 3′UTR are not depicted. Indicated reporter mRNAs were expressed in Moon/Sun cells. (A) Boxplots of relative initiation frequency in the MoonTag frame (left) or SunTag frame (right) (relative to the sum of the MoonTag and SunTag frame) on single mRNAs. (B-D) Boxplots of translation initiation rates in the MoonTag frame (B), SunTag frame (C), or MoonTag and SunTag frame combined (D) for single mRNAs. For comparison, data indicated in dark blue in D is replotted from Figure S5C. P-values are based on two-tailed Mann-Whitney tests: ^∗∗∗p < 0.001; ^∗∗∗∗p < 0.0001. Dashed line represents median value, box indicates 25%–75% range, and whiskers indicated 5%–95% range. Number of experimental repeats and mRNAs analyzed per experiment are listed in Table S1.