The life of an mRNA in space and time

Abstract

Nuclear transcribed genes produce mRNA transcripts destined to travel from the site of transcription to the cytoplasm for protein translation. Certain transcripts can be further localized to specific cytoplasmic regions. We examined the life cycle of a transcribed beta-actin mRNA throughout gene expression and localization, in a cell system that allows the in vivo detection of the gene locus, the transcribed mRNAs and the cytoplasmic beta-actin protein that integrates into the actin cytoskeleton. Quantification showed that RNA polymerase II elongation progressed at a rate of 3.3 kb/minute and that transactivator binding to the promoter was transient (40 seconds), and demonstrated the unique spatial structure of the coding and non-coding regions of the integrated gene within the transcription site. The rates of gene induction were measured during interphase and after mitosis, demonstrating that daughter cells were not synchronized in respect to transcription initiation of the studied gene. Comparison of the spatial and temporal kinetics of nucleoplasmic and cytoplasmic mRNA transport showed that the beta-actin-localization response initiates from the existing cytoplasmic mRNA pool and not from the newly synthesized transcripts arising after gene induction. It was also demonstrated that mechanisms of random movement were predominant in mediating the efficient translocation of mRNA in the eukaryotic cell.

Fig. 1.

Cell system for following β-actin gene expression in vivo. (A) Schematic of the gene construct. The 5′ end contains a series of 256 lacO repeats that bind RFP-LacI and mark the site of integration and transcription. Transcriptional induction from the minimal CMV promoter is achieved by the binding of reverse tetracycline transcriptional activator (rtTA or Tet-On) to TREs in the presence of dox. The transcribed mRNA contains the coding sequence for CFP–chicken-β-actin, 24 MS2 repeats that are bound by the YFP-MS2 fusion protein and the 3′-UTR of chicken β-actin, which contains the zipcode. At the 3′ end, a portion of a rabbit β-globin exon-intron-exon module is followed by a cleavage-polyA signal. (B) The selected cell clone contains a small gene integration site marked by RFP-LacI (red dots, white arrows). The nucleus in uninduced cells is in blue (top). In induced cells following dox-induced transcriptional activation, CFP-actin bundles are detected (bottom). Scale bar: 10 μm. (C) Western blot of CFP-actin in cells induced with dox for 1-5 hours showed protein accumulation, compared with no change in the levels of human β-actin or tubulin. (D) TRITC-phalloidin staining shows the integration of CFP-actin into filamentous actin. Scale bar: 20 μm.

Fig. 2.

Actively transcribing sequences surround the lacO sequences at the gene locus. (A) RNA-FISH with an MS2-Cy3 probe (red) shows the accumulation of β-actin–MS2 mRNA following transcriptional induction. Top and middle rows: 1 and 3 hour inductions show a transcription site and some cellular mRNAs. Bottom row: longer induction times show accumulation of cytoplasmic mRNAs (o.n. = overnight). Scale bars: 10 μm. (B) The transcriptionally active sequences encircle the non-coding lacO repeats, as seen by the recruitment of endogenous RNA Pol II (green) to the transcription site (RFP-LacI, red). (C) The nascent transcripts (MS2-Cy3 FISH, green), (D) the endogenous SF2/ASF splicing factor (green) and (E) the rtTA transactivator (green) all surround the gene locus (CFP- or RFP-LacI). Enlarged transcription sites are shown: red and green merges show active transcription (green) surrounding the locus (red). Green alone shows the circular structure of the active sequences. Scale bars: 5 μm.

Fig. 3.

The dynamics of rtTA association with the active gene. (A) rtTA-YFP (yellow) is recruited to actively transcribing sites (red, MS2 FISH). The box in merge shows the transcription site. (B) Frames from a FRAP experiment showing the rtTA-YFP signal at the transcription site before and after photobleaching (arrow). (C) FRAP recovery curves of rtTA-YFP at the transcription site (red dots and line fit) and freely diffusing in the nucleoplasm (blue dots and line fit). (D) Scheme of the model and differential equations describing the entrance and exit kinetics of rtTA-YFP molecules at the transcription site. Scale bars: 5 μm.

Fig. 4.

Kinetics of mRNA transcription. (A) Cells expressing YFP-MS2-NLS were imaged for 115 minutes [every 5 minutes, starting 5 minutes after dox induction (t=0)]. The induction of the transcription site was followed. Scale bar: 20 μm. (See also supplementary material Movie 1.) (B) The signal at the transcription site in A was quantified over time (dots). There is an initial linear increase after the induction of transcription that slows over time (following the form Ae^−Bt + C), showing the time required to reach a steady state. The fit (line) shows a deviation after 80 minutes. (C) Active transcription sites were photobleached and the recovery of fluorescence was recorded (supplementary material Movie 3). The recovery is proportional to the polymerase rates. The FRAP recovery curves and fits of photobleached transcription sites in the β-actin–MS2 cells (blue) are compared with those of a similar experiment on another gene (red) (Darzacq et al., 2007).

Fig. 5.

Transcription induction in daughter cells is not synchronized. DIC images (top row) show the division of the imaged cell (dox was given at t=0). Fluorescence images of the YFP–MS2-expressing daughter cells are shown from 260 minutes after dox induction (sum projection of three slices centered on the plane in focus) (supplementary material Movie 2). Arrows and boxes highlight the activated transcription sites. For each nucleus, we measured the maximum (active transcription site) and median (free YFP-MS2) pixel value and plotted their ratio over time in both cells. For presentation and noise reduction, data were smoothed by averaging every four time points, beginning from the time point after cell division. Left daughter cell – red, right daughter cell – blue line. Scale bar: 5 μm.

Fig. 6.

Kinetics of mRNA transport. (A) Cells expressing NES-YFP-MS2-NLS showed transcription site induction followed by the appearance of tagged mRNPs (original frames of supplementary material Movie 5 were deconvolved to enhance the mRNP signal). Cells were imaged for a total of 5.5 hours, every 10 minutes (dox induction at t=0 minutes). An active transcription site was seen at t=40 minutes. Approximately 10 minutes later, transcripts were detected in the nucleoplasm (50 minutes) and ~10 minutes after that in the cytoplasm (60 minutes). Scale bar: 20 μm. (B) Graph representing the number of mRNPs counted in the 3D volumes of the nucleus (red) and the cytoplasm (green) throughout each frame of the movie. The plot begins at t=40 minutes, the initial time point of transcription site activation. The site turns off at 260 minutes. (C) Three frames (t=50, 80, 170 minutes) representing the increase in mRNP levels in the cell over time and the distribution of mRNPs towards the periphery of the cytoplasm. These three frames are projections of all the mRNPs in the 3D volume of the cell, presented in one frame. Therefore, the mRNA signal is pronounced in the region surrounding the nucleus, where the cytoplasmic volume is large.

Fig. 7.

Cytoplasmic distribution of β-actin mRNAs. (A) RNA-FISH with a Cy3-MS2 probe against chicken β-actin–MS2 mRNA (red) and a Cy5–β-actin probe against human β-actin mRNA (green) showed spatial separation between the mRNAs. Both mRNA species were found to decorate CFP-actin bundles. In the enlarged area (bottom right), mRNPs (red and green balls) are situated along some of the CFP-actin bundles. Scale bar: 10 μm. (B) Human β-actin mRNA is mostly centrally localized in the cytoplasm of untreated U2OS cells, as seen by RNA-FISH (Cy3–β-actin probe), even when the cell is well spread (yellow arrows show cell edges). Following serum induction (20 or 30 minutes) of serum-starved cells, human β-actin mRNA is detected at the leading edge (white arrows). Scale bar: 20 μm. (C) 200 cells with prominent leading edges were analyzed by DIC with a Cy3 probe and counted at each time point. Positively localizing cells had a strong endogenous mRNA signal at the leading edge, in comparison to cells devoid of signal in this region. The percentage of cells with localized mRNA following serum induction is plotted.

Fig. 8.

Localization of chicken β-actin–MS2 mRNA. (A) RNA-FISH with a Cy3-MS2 probe against the β-actin–MS2 mRNA showed that the transcripts were centrally distributed in serum-starved cells (top) and localized to the leading edge after 20-30 minutes of serum stimulation (middle and bottom, yellow arrows). Cells were imaged in the CFP channel for CFP-actin. Hoechst nuclear staining was also detected in this channel as a strong signal in the center of the cell. The enlarged inset shows a series of mRNAs decorating CFP-actin fibers (green arrows). (B) Following overnight dox induction, adhering cells were seeded and dox was added one hour later. Cells were fixed 2 or 5 hours after plating. RNA-FISH with a Cy3-MS2 probe against the β-actin–MS2 mRNA (red) and a Cy5–β-actin probe against the human β-actin mRNA (green) showed that the mRNAs had a similar pattern of localization to the cell periphery. Scale bars: 10 μm.

Fig. 9.

Localization of YFP–MS2-labeled mRNPs in living cells. Cells expressing NES-YFP-MS2-NLS were induced with dox overnight. After raising the FCS levels to 20%, cells were imaged every 10 minutes. (A) The appearance of the β-actin–MS2 mRNA at the cell front was observed after addition of serum. (B) The same cell as in A with a different color representation, showing the appearance of mRNPs at the leading edge (supplementary material Movie 9). The color scale indicates levels of image intensity from 0-255. (C) Example of another cell showing mRNP localization. Scale bar: 5 μm. (D) 3D representation of the signal of YFP–MS2-labeled β-actin–MS2 mRNA in A at times 0 and 90 minutes, showing the increase in peripheral localization. The relative increase in signal intensity is shown on the color scale.

Fig. 10.

Diffusion kinetics of nuclear versus cytoplasmic mRNPs. (A) β-actin–MS2-labeled mRNPs were detected either in the nucleoplasm (nucleus with mRNPs shown in a color-inverted representation) or in the cytoplasm of dox-induced cells. Scale bar: 5 μm. (B) mRNPs (pseudo-colored yellow, center of mass in green) were tracked from frame to frame (red track). Scale bar: 0.5 μm. (C) Diffusion coefficients for nuclear and cytoplasmic mRNPs were calculated. (D) Cytoplasmic mRNPs (n=17 cells) were photobleached (FRAP) in an ROI and the diffusion coefficient was calculated from the fit of the FRAP curve (D=0.13 μm²/second). Dotted lines show the curve behavior if D=0.22 μm²/second (top) or D=0.09 μm²/second (bottom). (E) FRAP recovery rates were similar in untreated (blue dots) cells versus cells treated with nocodazole (red dots, n=19 cells) or cytochalasin D (grey dots, n=22).