mRNP architecture in translating and stress conditions reveals an ordered pathway of mRNP compaction

Abstract

Stress granules (SGs) are transient membraneless organelles of nontranslating mRNA-protein complexes (mRNPs) that form during stress. In this study, we used multiple single-molecule FISH probes for particular mRNAs to examine their SG recruitment and spatial organization. Ribosome runoff is required for SG entry, as long open reading frame (ORF) mRNAs are delayed in SG accumulation, indicating that the SG transcriptome changes over time. Moreover, mRNAs are ∼20× compacted from an expected linear length when translating and compact ∼2-fold further in a stepwise manner beginning at the 5' end during ribosome runoff. Surprisingly, the 5' and 3' ends of the examined mRNAs were separated when translating, but in nontranslating conditions the ends of long ORF mRNAs become close, suggesting that the closed-loop model of mRNPs preferentially forms on nontranslating mRNAs. Compaction of ribosome-free mRNAs is ATP independent, consistent with compaction occurring through RNA structure formation. These results suggest that translation inhibition triggers an mRNP reorganization that brings ends closer, which has implications for the regulation of mRNA stability and translation by 3' UTR elements and the poly(A) tail.

Figure 1.

mRNA recruitment to SGs is dependent on when ribosomes run off mRNAs after translation inhibition. (A) Representative smFISH images acquired for three different transcripts (AHNAK, PEG3, and NORAD) for U-2 OS cells treated with 0.5 mM NaAsO₂ with or without 10 µg/ml puromycin for 30 or 60 min. Nuclei (blue), G3BP1 (green), and individual RNA (red). Scale bar: 2 µm. (B–D) Fraction of specific RNA molecules found in SGs at 15, 30, 45, and 60 min in U-2 OS when stressed with 0.5 mM NaAsO₂ (B), 0.5 mM NaAsO₂ (C), and 10 µg/ml puromycin, and 0.5 mM NaAsO₂ with 50 µg/ml cycloheximide (D) added after cells were stressed for 30’. More than 250 RNAs were counted for each sample. The 15 and 30 min results shown in D are identical to the 15- and 30-min result shown in B.

Figure 2.

Organization of AHNAK mRNPs in nonstress and stress conditions. (A) Cartoon schematic indicating where smFISH probes bind to AHNAK mRNAs. smFISH probes binding to the 5′ ends, middle, or 3′ ends are labeled with distinct fluorophores and are false-colored red, blue, and green, respectively. (B, D, and F) Left: Representative AHNAK smFISH images of U-2 OS cells that were not stressed (B) or stressed with 0.5 mM NaAsO₂ for 60 min (D) or heat shock at 42°C for 60 min (F). Right: 3D rendering of smFISH spots by Bitplane Imaris imaging analysis software. Scale bar: 250 nm. (C, E, and G) Cumulative frequency graphs (in fractions) of smallest distances between 5′ to 3′ end smFISH spots (solid lines), 5′ end to middle smFISH spots (dash lines), and middle to 3′ end smFISH spots (dotted lines) in unstressed cells (black), 0.5 mM NaAsO₂-treated cells (green), and heat shock cells (red). More than 850 smallest distances were quantified for each sample (n = 1,189 [no stress], n = 1,062 [NaAsO₂], and n = 860 [heat shock]). The nonstress samples are identical in panels C, E, and G and in Fig. 4 B. **, P ≤ 0.01; ***, P ≤ 0.001 (Student’s two-tailed t test).

Figure 3.

Organization of DYNC1H1 mRNPs in nonstress and stress conditions. (A) Cartoon schematic indicating where smFISH probes bind to DYNC1H1 mRNAs. smFISH probes binding to the 5′ ends, middle, or 3′ ends are labeled with distinct fluorophores and are false-colored red, blue, and green, respectively. (B, D, and F) Left: Representative DYNC1H1 smFISH images of U-2 OS cells that were not stressed (B) or stressed with 0.5 mM NaAsO₂ for 60 min (D) or heat shock at 42°C for 60 min (F). Right: 3D rendering of smFISH spots by Bitplane Imaris imaging analysis software. Scale bar: 250 nm. (C, E, and G) Cumulative frequency graphs (in fractions) of smallest distances between 5′ to 3′ end smFISH spots (solid lines), 5′ end to middle smFISH spots (dash lines), and middle to 3′ end smFISH spots (dotted lines) in unstressed cells (black), 0.5 mM NaAsO₂-treated cells (green), and heat shock cells (red). More than 1,000 smallest distances were quantified for each sample (n = 1,113 [no stress], n = 1,032 [NaAsO₂], and n = 1,056 [heat shock]). The no stress samples are identical in panels C, E, and G and in Fig. 4 D. The smallest distances between the 5′ and middle and the 3′ and middle smFISH spots reach a cumulative frequency of ∼75% within 600 nm because the middle smFISH probes are not as effective and for ∼25% of DYNC1H1 mRNAs, a middle smFISH spot was not detected. ***, P ≤ 0.001 (Student’s two-tailed t test).

Figure 4.

Compaction of AHNAK and DYNC1H1 mRNPs correlates with ribosome release from mRNA. (A, C, and E) Left: Representative AHNAK and DYNC1H1 smFISH images of U-2 OS cells treated with 10 µg/ml puromycin for 60 min or treated with 0.5 mM NaAsO₂ and 1 µM emetine (eme) for 60 min. Cells were stained with smFISH probes that bind specifically to the 5′ end (false-colored red), middle (false-colored blue), and 3′ end (false-colored green) of AHNAK and DYNC1H1 mRNAs. Right: 3D rendering of smFISH spots by Bitplane Imaris imaging analysis software. Scale bar: 250 nm. (B, D, F, and G) Cumulative frequency graphs (in fractions) of smallest distances between 5′ to 3′ end smFISH spots (solid lines), 5′ end to middle smFISH spots (dash lines), and middle to 3′ end smFISH spots (dotted lines) in unstressed cells (black), 10 µg/ml puromycin-treated cells (blue), 0.5 mM NaAsO₂-treated cells (green), and 0.5 mM NaAsO₂ plus 1 µM eme-treated cells (gold). More than 990 smallest distances were quantified for each sample (n = 1,107 [AHNAK puromycin], n = 992 [DYNC1H1 puromycin], n = 1,216 [nonstress, F and G], n = 999 [NaAsO₂, F and G], and n = 1,181 [NaAsO₂ plus emetine, F and G]). Scale bar: 250 nm. (H) Cumulative frequency graph (in fractions) of smallest distances between 5′ to 3′ end smFISH spots (solid lines) inside and outside SG in U-2 OS cells stressed with 60-min 0.5 mM NaAsO₂. More than 290 smallest distances were quantified (n = 550 [inside], n = 296 [outside]). The analysis was performed with the experimental results as shown in Fig. 2 (B and C). *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001 (Student’s two-tailed t test).

Figure 5.

Compaction of the 5′ end to the middle precedes compaction of the middle to the 3′ end during NaAsO₂ stress. (A) Cartoon schematic illustrating where smFISH probes bind to AHNAK mRNAs. smFISH probes binding to the 5′ end, middle, or 3′ end are labeled with distinct fluorophores and are false-colored as red, blue, and green, respectively. (B) Representative AHNAK smFISH image of U-2 OS cells that were not stressed or stressed with 0.5 mM NaAsO₂ for 10, 20, and 30 min. Scale bar: 1 µm. (C–E) Cumulative frequency graphs (in fractions) of smallest distances between 5′ to 3′ end smFISH spots (C), 5′ end to middle smFISH spots (D), and middle to 3′ end smFISH spots (E) in unstressed U-2 OS cells or 0.5 mM NaAsO₂-treated U-2 OS cells for 5–30 min. More than 800 smallest distances were quantified for each sample (n = 1,015 [0 min], n = 954 [5 min], n = 814 [10 min], n = 952 [15 min], n = 848 [20 min], n = 915 [25 min], and n = 949 [30 min]). *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001 (Student’s one-tailed t test).

Figure 6.

Compaction of the 5′ end to the middle and compaction of the middle to the 3′ end occur simultaneously when cells are treated with puromycin. (A) Cartoon schematic illustrating where smFISH probes bind to AHNAK mRNAs. smFISH probes binding to the 5′ end, middle, or 3′ end are labeled with distinct fluorophores and are false-colored as red, blue, and green, respectively. (B) Representative AHNAK smFISH image of U-2 OS cells that were not stressed or stressed with 10 µg/ml NaAsO₂ for 4, 8, and 12 min. Scale bar: 1 µm. (C–E) Cumulative frequency graphs (in fractions) of smallest distances between 5′ to 3′ end smFISH spots (C), 5′ end to middle smFISH spots (D), and middle to 3′ end smFISH spots (E) in unstressed U-2 OS cells or puromycin-treated U-2 OS cells for 4–20 min. More than 1,000 smallest distances were quantified for each sample (n = 1,037 [0 min], n = 1,383 [4 min], n = 1,183 [8 min], n = 1,166 [12 min], n = 1,014 [16 min], and n = 1,134 [20 min]. *, P ≤ 0.05; ***, P ≤ 0.001 (Student’s one-tailed t test).

Figure 7.

Translation inhibition with puromycin, NaAsO₂, or heat shock in U-2 OS cells disproportionally shrink the distances between the 5′ and 3′ ends relative to the middle of AHNAK and DYNC1H1 mRNPs. (A) Cartoon schematic indicating the angles that were measured in B and C. (B and C) Histograms illustrating the relative frequency (fractions) of angles from middle smFISH spots to 5′ end and 3′ end smFISH spots of AHNAK (B) and DYNC1H1 (C) mRNAs in unstressed (black line), puromycin-treated (blue), NaAsO₂-treated (green), or heat shocked (red) U-2 OS cells. The histograms were generated by binning every 15°. For AHNAK, more than 850 angles were quantified for each sample (n = 1,189 [no stress], n = 1,062 [NaAsO₂], n = 860 [heat shock], and n = 1,107 [puromycin]). For DYNC1H1, more than 1,000 angles were quantified for each sample (n = 1,113 [no stress], n = 1,032 [NaAsO₂], n = 1,056 [heat shock], and n = 992 [puromycin]).

Figure 8.

Distances between SunTagx32-DYNC1H1 mRNAs 5′ and 3′ ends are larger when engaged in translation than when it is not. (A) Cartoon schematic illustrating where smFISH probes bind to SunTagx32-DYNC1H1 single-molecule translation reporter. The smFISH probes binding to the 5′ end and 3′ end are false-colored as green and red, respectively, and the scFv-sfGFP antibody is false-colored as blue. The SunTagx32-DYNC1H1 reporter and the scFv-sfGFP are stably expressed in HeLa cells. (B and C) Representative images of HeLa cells stably expressing SunTagx32-DYNC1H1 and scFv-sfGFP (false-colored gray) treated with NaAsO₂ for 60 min or puromycin for 30 min without (B) or with (C) 2DG and CCCP. Middle (C): 2DG, CCCP, and NaAsO₂ were added at the same time to the cells and incubated for 60’. Right (C): 2DG and CCCP were added first to cells for 30 min before adding puromycin to the cells and incubated for another 30’. Cells were stained with DAPI (false-colored blue). Scale bar: 5 µm. (D–F) Representative images of individual SunTagx32-DYNC1H1 reporter RNA (false-colored blue) and stained with SunTagx32-specific (false-colored green) and 3′ end DYNC1H1 smFISH probes (false-colored red) in HeLa cells unstressed (D) or treated with puromycin (E), or puromycin, 2DG, and CCCP (F). Scale bar: 250 nm. (G) Cumulative frequency graph (in fractions) of smallest distances between 5′ to 3′ end smFISH spots in unstressed HeLa cells with a corresponding scFv-sfGFP fluorescent spot (black line) and puromycin-treated or puromycin (blue-line), 2DG-, and CCCP-treated HeLa (blue dashed line). At least 100 smallest distances were quantified for each sample (n = 103 [translating], n = 100 [not translating], and n = 110 [puromycin + 2DG + CCCP]. **, P ≤ 0.01; ***, P ≤ 0.001 (Student’s two-tailed t test).

Figure 9.

Model depicting mRNP compaction and mRNA recruitment to SGs. Under nonstress conditions, mRNPs are engaged in translation. Relative to its contour length, significant compaction was observed. During the early stages of stress, ribosomes will migrate toward the 3′ end of mRNAs and mRNAs start to compact at the 5′ end, most likely because of intramolecular interactions formed. When mRNA exits translation, the mRNP compacts further and the ends are disproportionally close. One hypothesis is the closed-loop conformation is reestablished. Some of these mRNPs, preferentially long mRNAs, start to accumulate in SGs via intermolecular interactions formed with other mRNPs.