The translation elongation factor eEF1A1 couples transcription to translation during heat shock response

1All organisms respond to environmental stress by activating genes that encode molecular chaperones and other cytoprotective heat shock proteins (HSPs) in a process known as the heat shock response (HSR). In mammalian cells, HSR is regulated at the level of transcription by the heat shock factor 1 (HSF1). In coordination with their augmented transcription, the stability, transport, and translation of HSP mRNAs (especially HSP70) are sharply increased during HSR, accounting for the robust production of HSPs. HSP70 is the best-studied example, which is among the most inducible HSPs in terms of promptness and robustness of the response. The mechanism of such a rapid and synchronized reaction to stress remains unknown, but is particularly remarkable considering that upon stress, the majority of other genes are inhibited at the level of transcription, RNA export, and translation.

2Our previous in vitro studies demonstrated that translation elongation factor eEF1A enhances the binding of HSF1 to a DNA containing an artificial HSE sequence via a mechanism that requires a non-coding RNA (HSR1). eEF1A1 has a well-defined role in protein synthesis, it delivers aminoacylated tRNAs to the A site of ribosome during translation elongation in a GTP-dependent manner. While the majority of eEF1A1 associates with the cytoplasmic cytoskeleton and translational machinery, eEF1A1 is able to enter the nucleus to mediate the export of tRNAs and SNAG-containing proteins under normal growth conditions. Also, eEF1A can directly bind mammalian mRNAs and mediate stability of viral and cellular RNAs by binding to their 3′ regions.

3eEF1A expression in mammalian cells happens in one of two isoforms, eEF1A1 and eEF1A2. These two isoforms are 98% similar at the amino acid level and share the same canonical function. Each isoform is the product of expression of a specific locus and shows a distinct expression pattern except in transformed cells, which express both. While the majority of cells express eEF1A1 isoform, adult neuronal and muscle cells express eEF1A2. In this study, we report that eEF1A1 isoform, but not the tissue-specific variant eEF1A2 is required for the HSP70 induction during heat shock. It mediates transcription-activation of HSP70 mRNA, its stability, transport, and translation, thereby synchronizing HSP70 transcriptional output to translational needs.

4To determine the role of eEF1A1 in HSR in vivo, we designed a strategy for its selective and efficient knockdown without compromising overall translation and cell viability. eEF1A is one of the most abundant cellular proteins, compromising 1–3% of total cytosolic proteins, and is present in large excess to its protein synthesis partners (molar ratios for eEF1A:eEF1B and eEF1A:ribosomes of 10:1 and 25:1 respectively). Therefore, a ∼70% knockdown of eEF1A1 did not compromise cell viability under non-stress conditions and overall translation under control or mild-heat shock conditions (Figure 1—figure supplement 1A,B). The levels of eEF1A1 and HSF1 (positive control), but not those of GAPDH, were greatly reduced after efficient transfection of human breast cancer cells (MDA-MB231), primary human fibroblasts (WI38), immortalized mouse embryonic fibroblast (MEFs), or mouse NSC34 motor spinal cord/neuroblastoma fusion cells, with specific and un-related sets of siRNAs (Figure 1A [siRNA pair A], Figure 1—figure supplement 2A [siRNA pair B], Figure 1—figure supplement 3A,B [siRNA C]). After heat shock, knock down of eEF1A1 (∼30% of control remaining) or HSF1, (<25% of control remaining), markedly reduced the induction of HSP70 (Figure 1A, Figure 1—figure supplement 2A, Figure 1—figure supplement 3A,B) and HSP27 (Figure1—figure supplement 3C), and resulted in loss of thermo-tolerance (Figure 1B). To rule out the possibility that the loss of induction of HSPs in eEF1A1-deficient cells was potentially due to overall compromised translation, de novo protein synthesis was analyzed in a pulse-chase experiment with [35S]-methionine. We observed a general decrease in protein synthesis after HS, but no significant difference between mock-transfected cells and cells depleted of either HSF1 or eEF1A1 (Figure 1—figure supplement 1A). Consistently, HSP70 and HSP27 mRNA levels were significantly decreased in eEF1A1-knocked down heat-shocked cells (Figure 1C, Figure 1—figure supplement 2B, Figure 1—figure supplement 3D,E). Impaired mRNA induction upon heat shock due to eEF1A1 deficiency was observed for several classes of HSP genes (Figure1—figure supplement 3F). It is known that HSP70 mRNA induction occurs upon chemical inhibition of translation. To verify that the action of eEF1A1 is independent of its role in translation inhibition, we tested eEF1A1 knockdown conditions in combination with inhibitors of translation elongation. eEF1A1-depleted cells treated with the HSR-stimulating inhibitors of translation, cycloheximide (CHX) or doxycycline (DOX) failed to accumulate HSP70 mRNA (Figure 1—figure supplement 3G). In contrast, knocking down eEF1A2 isoform had no effect on HSP70 mRNA induction or protein amount (Figure 1—figure supplement 4A,B). These results demonstrate that isoform 1 of eEF1A specifically controls HSPs gene expression irrespective of its role in translation.

5Analyses of the distribution of RNA polymerase II (RNAPII) along the HSP70 gene as a function of the presence of eEF1A1 supported this conclusion, arguing that eEF1A1 is an activator of transcription upon stress. Chromatin immunoprecipitation (ChIP) and run-on assays demonstrated that knock down of eEF1A1 reduced recruitment of RNAPII throughout the HSP70 gene, indicative of both impaired RNAPII initiation and elongation during heat shock (Figure 1D, Figure 1—figure supplement 2C, Figure 1—figure supplement 5A–C). Under the same conditions, there was no effect of eEF1A1 depletion on RNAPII at the GAPDH gene (Figure 1D, Figure 1—figure supplement 2C). Knock down of eEF1A1 did not change the relative occupancy of RNAPII on HSP70 and GAPDH genes under non-heat shock conditions (Figure 1D).

6We used single molecule fluorescence in situ hybridization (smFISH) and airlocalize software to assess the kinetics of HSP70 transcription (from control growth conditions to 2 hr of HS) and quantify the number of HSP70 transcription sites (TS) per cell and the transcription activity per TS in the presence or absence of eEF1A1. eEF1A1 knock down was monitored in immortalized mouse embryonic fibroblast (MEFs) from the homozygous Actb-MBS mouse strain expressing eEF1A1 fused to cherry and Flag, by the decrease in cherry detection (Figure 1E, Figure 1—figure supplement 1D, Figure 1—figure supplement 6A). At 30 min of HS, 80% of cells showed a strong nuclear FISH signal that accounted for active HSP70 transcription in at least 1 of its 4 alleles (MEFs are tetraploid) (Figure 1E, Figure 1—figure supplement 1D, Figure 1—figure supplement 6B,C). Cells knocked down for eEF1A1 showed half (pair A) or third (pair B) of the intensity of transcription per cell (Figure 1F, Figure 1—figure supplement 1E). This result is in perfect correlation with the quantification of HSP70 mRNA, nuclear run-on and ChIP experiments. This decrease of transcription intensity per cell is due to both a decrease in the intensity quantified per TS (Figure 1F, Figure1—figure supplement 1D) and a reduction in the number of TS per cell from a mean of 3 TS per cell in non-transfected cells to 2.6 TS in cells knocked down for eEF1A1 (siRNA pair A) and from 2.8 to 2.1 (siRNA pair B).

7To gain further insight in the mechanism by which eEF1A1 regulates transcription of HSP genes, we explored the activation of HSF1. Upon stress, HSF1 quickly trimerizes, acquires the ability to bind HSE, and reactivates paused RNAPII via a mechanism involving recruitment of the Mediator complex, the CTD kinase P-TEFb and the SWI/SNF chromatin remodeling complex. In MDA-MB231 cells, as in other cell lines, HSF1 is localized in the nucleus under normal growth conditions. In mammalian cells, HSE binding activity assessed by EMSA is entirely due to HSF1 (Figure 2D, Figure 2—figure supplement 1B). This binding activity was reduced in the protein extract (10 μg) from eEF1A1-knocked down cells (Figure 2A). Likewise, knock down of eEF1A1 also resulted in a significant reduction in recruitment of HSF1 to the HSP70 promoter, as determined by ChIP assays (Figure 2B). The low levels of HSF1 binding to HSP70 promoter under control conditions did not change in cells knocked down of eEF1A1. These results indicate that eEF1A1 enhances the binding of HSF1 to HSP70 promoter and suggest an interaction between eEF1A1 and HSF1 during HS.

8Consistent with the above observations, HSF1 co-immunoprecipitated with eEF1A1 upon HS (Figure 2C, Figure 2—figure supplement 1A). Moreover, eEF1A1 formed a ternary complex with HSF1 and DNA. Heat shock or arsenic treatments induce HSF1 DNA binding, as detected by EMSA. Under these conditions, anti-HSF1 and anti-eEF1A1, but not mouse IgG, super-shifted the HSF1–DNA complex (Figure 2D, Figure 2—figure supplement 1B). We noted that the super-shifts with anti-eEF1A1 and anti-HSF2 antibodies were similar. Considering that HSF2 binds HSE only when forming a hetero-trimer with HSF1, this observation indicates further that eEF1A1 associates directly with DNA-bound HSF1. Moreover, in vitro binding of recombinant eEF1A1 to a DNA fragment of HSP70 promoter suggests direct and specific, albeit a relatively weak, interaction of eEF1A1 to HSP70 promoter (Figure 2E). We further confirmed the presence of eEF1A1 at the HSP70 locus by ChIP-QPCR experiments with an eEF1A1-specific antibody (Figure 2F, Figure 2—figure supplement 1C). Occupancy of HSP70 and HSP27 promoters by eEF1A1 was significantly higher (p < 0.05) than at the GAPDH promoter. These results demonstrate that eEF1A1 binds specifically to promoters of at least two HSP genes before and after stress.

9Together, the above results demonstrate that eEF1A1 recruits HSF1 to the HSP70 promoter upon stress and triggers HSR by functioning as a transcriptional co-activator.

10Observations described above strongly suggest that eEF1A1 is in the nucleus of the cell at the TS of HSP70. To gain further insight in the cellular localization of eEF1A1 regarding HSP70 mRNA, we used fluorescence in situ hybridization (FISH). We used the MEFs expressing eEF1A1 fused to Cherry and Flag (Figure 1—figure supplement 5A), and we infected human diploid fibroblasts (TIGs) to also express recombinant eEF1A1 protein fused to Cherry and Flag (Figure 3—figure supplement 1). MEFs were subjected to different times of HS (from 0 to 60 min) before visualizing HSP70 mRNA. From 30 to 60 min of HS ∼80% of the cells showed a strong nuclear FISH signal that accounted for active HSP70 transcription (Figure 3A, Figure 1—figure supplement 5B,C). After 30 min of heat shock, we quantified the dots representing Cherry-eEF1A1 concentrated at several HSP70 TS (Figure 3A, Figure 3—figure supplement 2). Percentages of HSP70 TS with a positive signal for Cherry-eEF1A1 were quantified by airlocalize™ software. The percent of Cherry-eEF1A colocalizing with the HSP TS increased from 27% to −64% for 30 to 60 min (Figure 3C). Using probes for β-actin TS instead of those for HSP70 (Figure 3B), we did not see any co-localization with eEF1A nuclear dots. The Cherry-eEF1A dots were not a result of bleed through of the FISH signal because when HSP70 FISH was performed in MEFs that do not express Cherry-eEF1A1, nuclear dots in the red channel were rarely apparent, and when FISH was not performed the Cherry-eEF1A1 dots remained evident (Figure 3—figure supplement 3) The presence of eEF1A1 at HSP70 TS was further confirmed in a human fibroblast cell line (TIG) (Figure 3—figure supplement 1).

11Since several molecules of eEF1A need to accumulate in a discrete site to be detected by microscopy, we hypothesized that the presence of eEF1A at HSP70 TS not only depends on its interaction with HSF1, but also on the increase in transcription. ChIP analysis of heated MEF cells (expressing Flag-eEF1A1 or not) showed that eEF1A1 occupancy of HSP70 gene increases in the ORF and 3′UTR upon HS, as does RNAPIIS2 and RNAPII (Figure 3D, Figure 3—figure supplement 4A,B). When RNAPII elongation was inhibited by DRB treatment before HS, RNAPIIS2 occupancy on HSP70 gene was abolished (Figure 3D). Interestingly, the eEF1A1 occupancy inside HSP70 locus was reduced (p = 0.0585) as it was for total RNAPII (p = 0.0316) (Figure 3D, Figure 3—figure supplement 4A,B)

12This result together with the fact that eEF1A1 co-immunoprecipitates with the RNAPII (Figure 3E) suggests that eEF1A1 interacts with the transcription elongation complex during HSP70 transcription.

13Detection of eEF1A1 at the 3′ end of HSP70 gene suggests a role for this protein beyond transcription. Thus, we investigated the interaction between eEF1A1 and HSP70 mRNA in the soluble chromatin fraction of control and HS cells. HSP70 mRNA association with eEF1A was detected by RNA-IP in HS cells (Figure 4A). In RNA-IP experiments from total cell lysates, we also detected HSP70 mRNA associated with eEF1A1 and other known HSP70 mRNA-binders, poly(A)-binding protein 1 (PABP1) and nuclear pore complex (NPC) protein TPR1 (; Figure 4B).

14We then used RNA-EMSA to test for a direct interaction between eEF1A1 and HSP70 mRNA. eEF1A1 altered the mobility of the 3′UTR HSP70 mRNA probe, but not that of the 5′ portion of the HSP70 mRNA or of an unrelated RNA (200 nt β-actin ORF) (Figure 4C). These observations prompted us to investigate a role for eEF1A1 in stabilizing HSP mRNA. We monitored HSP70 and GAPDH mRNA decay after inhibition of transcription by actinomycin D (Figure 4D, Figure 4—figure supplement 1A). After 30 min of HS treatment, the rate of HSP70 mRNA decay increased substantially in eEF1A1-depleted cells, as compared to mock-transfected cells. In contrast, no change in the rate of GAPDH mRNA decay was observed.

15To roughly map the region within HSP70 3′UTR that binds eEF1A1, we generated three deletion constructs of stem-loops according to the m-fold predicted structure (Figure 4—figure supplement 1B). The deletions ΔSL1 (Δ7–34) or ΔSL3 (Δ217–226, Δ246–255) did not reduce eEF1A1 binding to the 3′UTR significantly (Figure 4—figure supplement 1B), whereas the deletions in stem-loop 2 (ΔSL2–Δ83–93, Δ151–158) abolished eEF1A1 binding, indicating that this region was the primary interacting site for eEF1A1. The Kd for the eEF1A1-HSP70 3UTR complex was estimated to be ∼100 nM (Figure 4—figure supplement 1C).

16Since we observed eEF1A1 associated specifically with the HSP70 promoter and then with the RNAPII elongation complex, we examined if HSP70 mRNA stabilization by eEF1A1 required both the cognate promoter and the 3′UTR. To test this, we placed the luciferase ORF, with or without the HSP70 3′UTR, under regulation of the SV40 or HSP70 promoter. Upon heat shock, only the presence of both the HSP70 promoter and 3′UTR resulted in a steep decrease of luciferase activity in eEF1A1-knocked down cells (Figure 4E). In contrast, when HSP70 containing its 3′UTR was expressed from an SV40 promoter, no decay was evident. This supports the model that eEF1A1-mediated mRNA stabilization resulted from the recruitment of eEF1A1 during transcription and its subsequent association with the 3′UTR.

17It is well established that export of non-HSP RNAs is inhibited during HS. We observed HSP70 mRNA in the cytoplasm of HS cells shortly after transcription activation (Figure 3A, Figure 3—figure supplement 2, second panel from the left [30 min of HS]). Since eEF1A1 has been reported to be a component of mammalian nuclear export machinery, we investigated whether it plays a role in HSP70 mRNA transport. To gain further insight in the cellular localization of HSP70 mRNA, we used FISH in the immortalized MEFs expressing eEF1A1 fused to Cherry and Flag (Figure 1—figure supplement 5A). Cells knocked down for eEF1A1 showed a stronger FISH signal in the nucleus than non-transfected cells, where most of the HSP70 mRNA was localized in the cytoplasm (Figure 5A, Figure 5—figure supplement 1A,B). Therefore, we quantified the effect of eEF1A1 in the partitioning of HSP70 mRNA between nucleus and cytosol after heat shock using RT-QPCR. After 2 hr of heat shock more than 95% of the total HSP70 mRNA has been exported to the cytoplasm of mock-transfected cells. In contrast, cells knocked down for eEF1A1 have accumulated ∼30% of the total synthesized HSP70 mRNA in the nucleus (Figure 5B). We did not find any significant difference for GAPDH mRNA accumulation in the nucleus (∼15% in mock transfected cells vs ∼12% in cells knocked down for eEF1A1).

18TPR1 is a nuclear pore complex (NPC) factor essential for HSP70 mRNA export. We postulated that eEF1A1 mediated the stress-induced binding of the HSP70 mRNA 3′UTR to TPR1. RNA-IP followed by RT-QPCR showed that eEF1A1 depletion greatly diminished the percentage of HSP70 mRNA bound to TPR1 upon heat shock (Figure 5C). Furthermore, both proteins co-immunoprecipitate (Figure 5—figure supplement 1C) during heat shock. This suggests a role for eEF1A1 in docking HSP70 mRNA to TPR1 in the NPC.

19Once in the cytoplasm, HSP70 is selectively translated. We observed that the areas in the cytoplasm, both perinuclear and the leading edges, with higher density of HSP70 mRNAs corresponded to a brighter signal from cherry-eEF1A1 (Figure 5A,D, Figure 5—figure supplement 1A, [yellow arrows]). We did not observe such a high-density accumulation of HSP70 mRNAs in cells knocked down for eEF1A1 (Figure 5A,D Figure 5—figure supplement 1A,B). This could be explained by the decreased amount of HSP70 mRNA, but also by a less efficient loading of HSP70 mRNA into polysomes. Upon heat shock, the polysome profile MDA-MB231 cells knocked down for eEF1A1 was similar to that from mock transfected cells but, we detected a smaller amount of HSP70 mRNA in the pooled fraction of light ribosomes in cells knocked down for eEF1A1 (Figure 5E, Figure 5—figure supplement 1C), suggesting that eEF1A1 helped the loading of HSP70 mRNA into polysomes.

20HSR is a tightly regulated and synchronized process that is essential for cell survival under stress. While many HSP mRNAs, such as HSP70, are present in only very low amounts in unchallenged cells, their synthesis, stability, and translation amplify dramatically upon stress. We observed that each of these steps gets impaired when eEF1A1 is partially depleted. It is the sum of the effects on HSP70 mRNA synthesis, stability, and transport that accounts for a ∼70% reduction of the HSP70 protein level and compromised thermotolerance in cells partially knocked down for eEF1A1. It is likely that the remaining eEF1A1 supports the residual HSP70 expression.

21We have monitored the mRNA expression of the HSP70 by single molecule FISH in cells expressing Cherry-eEF1A1 (Figures 3A, 5A,C, Figure 3—figure supplement 2). We have found that although the expression of HSP70 during heat shock is synchronized, not all the cells respond with the same intensity at the same time. Indeed, induction of HSP70 transcription within the TSs of the same cell is stochastic (Figure 3A, Figure 1—figure supplement 5B,C, Figure 3—figure supplement 2). Nonetheless, most cells switched from almost undetectable levels of HSP70 mRNA under non-stress conditions to a massive induction of transcription from 30 to 60 min of heat shock, when mRNA can also be detected in the cytoplasm. Although the stimulus persisted, transcription ends at 2 hr of heat-shock. At this time point the majority of HSP70 mRNA is located in the cytoplasm. Our results suggest that eEF1A1 tags HSP70 mRNA during its transcription to ensure its rapid and efficient translation in contrast to non-HSP mRNAs (Figure 6). This hypothesis is in agreement with recent publications showing that the fate of the messenger is predetermined from transcription.

22eEF1A1 delivers aminoacylated tRNAs to the A site of ribosome during translation elongation. We found that upon HS, when translation elongation is paused to protect cellular homeostasis, eEF1A1 accumulates in discrete dots in the nucleus that correspond to HSP70 TS (Figure 3A, Figure 3—figure supplement 2). eEF1A1 participates in the regulation of transcription by recruiting and activating HSF1 (Figure 2A,B). Formation of the HSF1-eEF1A1-DNA complex in vitro and in vivo requires an RNA component (Figure 2—figure supplement 1B), which is thought to function as a thermosensor. eEF1A1 also interacts with the hyperphosphorylated form of RNAPII upon heat shock (Figure 3E), suggesting that it travels with RNAPII during elongation of HSP mRNA (Figure 3D, Figure 3—figure supplement 4). Although it is possible that binding of eEF1A1 to HSP70 mRNA can be independent of transcription elongation, the association of eEF1A1 with RNAPII (Figure 3E) may facilitate relocation of eEF1A1 from the promoter to the 3′UTR of HSP mRNA (Figure 3D, Figure 3—figure supplement 4D) and thus may be crucial for the stability and transport of HSP mRNAs (Figures 4 and 5).

23The stability of HSP70 mRNA sharply increases upon stress via a highly conserved mechanism involving cis-acting AU rich elements (ARE) in the 3′UTR. Heat shock inhibits ARE-facilitated mRNA decay by stabilizing HSP70-AUF1 (ARE/poly(U)-binding/degradation factor 1) and PABP1 binding to the 3′UTR. Also, PKR (double-stranded RNA-dependent protein kinase) plays a significant role in stabilizing HSP70 mRNA upon stress. eEF1A1 binds directly to the 3′UTR of HSP70 mRNA during heat shock (Figure 4C, Figure 4—figure supplement 1B). Knock down of eEF1A1 results in selective destabilization of HSP70 mRNA (Figure 4D, figure 4—figure supplement 1B). The interaction of eEF1A1 with trans-acting factors, such as AUF1 and PABP1, as well as the phosphorylation and/or methylation status of eEF1A1 in response to different stimuli is likely to determine the stability of specific mRNAs and the interaction of eEF1A1 with HSP70 mRNA. It is also possible that PKR influences the stability of HSP70 mRNA via phosphorylation of eEF1A1.

24Once HSP mRNAs have been properly processed they must be promptly exported from the nucleus to ribosomes in the cytoplasm. HSF1-mediated recruitment of the NPC protein TPR1 stimulates export of HSP70 mRNA from the nucleus. Inhibition of the HSF1–TPR1 interaction suppresses HSP70 mRNA export without decreasing mRNA stability. Since eEF1A1 interacts with TPR1 during heat shock (Figure 5—figure supplement 1C) and is required for TPR1 binding to HSP70 mRNA (Figure 5C) and export of HSP70 mRNA from the nucleus (Figure 5A,B), the transport mechanism appears to rely on eEF1A1-mediated activation of HSF1 and formation of the ternary complex of TPR1-HSF1-eEF1A1 with HSP mRNA.

25Decreased transport correlated with poor loading of HSP mRNAs into active ribosomes (Figure 5E, Figure 5—figure supplement 1D), whereas both, polysome profiles and general protein synthesis, were unaffected in cells knocked down for eEF1A1 (Figure 1A—figure supplement 1C), suggesting a selective effect of eEF1A1 on HSP70 translation. It is likely that eEF1A1 modulates the expression of many genes controlled by HSF1. Our results with HSP27 and other HSPs support this hypothesis (Figure 1—figure supplement 2C,D,F, Figure 5—figure supplement 1D).

26Deregulation of HSR is associated with numerous pathologies ranging from cancer to neurodegenerative conditions. A large effort has been made to develop pharmacological approaches to either inhibit or stimulate HSR to, respectively, kill rogue cells or protect normal cells. The induction of HSP70 in transgenic mouse models of spinal bulbar muscular atrophy (SBMA) and amyotrophic lateral sclerosis (ALS) ameliorated disease symptoms and prolonged lifespan. The present findings illuminate eEF1A1 as a promising tool for the treatment of these diseases. eEF1A has two isoforms, 1 and 2, expressed from two different genes that are 96% similar at the protein level. In contrast to most other cell types, mature motor neurons express only eEF1A2 instead of eEF1A1 (Figure 1—figure supplement 3C; ). Because eEF1A2 does not support HSR (Figure 1—figure supplement 3A,B), our results provide an explanation of why motor neurons do not induce HSPs upon stress, even though they express high levels of HSF1. Our findings also suggest a novel therapeutic strategy for these neurodegenerative diseases by enabling eEF1A1 expression in motor neurons.

27In summary, the isoform 1 of translation elongation factor eEF1A has been established as a key component of HSR that controls each stage of the HSP70 induction from transcription activation to mRNA stabilization, nuclear transport, and translation (Figure 6). This remarkable feature of eEF1A1 makes it an attractive target for the treatment of many pathological conditions associated with HSR deregulation.

28293T, TIGs, MDA-MB231, and HeLa cells were obtained from ATCC (ATCC, Manassas, VA, USA); NSC34 cells were obtained from Cedarlane (Cederlane, Burlington, NC, USA). MEFs from the homozygous Actb-MBS mouse strain have been previously described. All cell lines were cultured in DMEM supplemented with 5% FBS and P/S at 37°C in an atmosphere of 5% CO2.

29To induce HSR, cells grown in tissue-culture flasks were submerged in a water bath at 42°C or 43°C (for cells of human or mouse origin respectively) for the indicated time intervals. Translation was inhibited by cycloheximide (CHX, 50 μg/ml) or doxycycline (DOX, 5 μg/ml) (Sigma, Saint Louis, MO, USA). Arsenite stress was induced with 0.1 mM arsenite (Sigma, Saint Louis, MO, USA). Transcription was inhibited with 5 μg/ml of actinomycin D (Sigma, Saint Louis, MO, USA) or 100 μM of DRB (Cayman, Ann Harbor, Michigan, USA).

30DsiRNA for eEF1A1 (NM_001402.5–Homo sapiens, NM_010106.2–Mus musculus), eEF1A2 (NM_001958.2–H. sapiens), and HSF1 (NM_005526.2–H. sapiens) were designed by IDT (IDT, Coralville, Iowa, USA) (http://www.idtdna.com/catalog/DicerSubstrate/Page1.aspx). Two DsiRNAs were selected for each gene: NM_001402.5 (1 and 10 [pair A] and 3 and 5 [pair B]) NM_010106.2 (1 and 6 [pair A] and 3 and 5 [pair B]) NM_005526.2 (1 and 2) and NM_001958.2 (1 and 2). Inhibition of the desired gene was obtained 48 hr after the second co-transfection of each DsiRNA at a final concentration of 0.6 nM with the recommended amount of trifectin transfection reagent according to the manufacturer's instructions (or 10 nM of each siRNA transfected with SilenFect [Biorad, UK)]). Consecutive co-transfections were performed at 48 hr intervals in antibiotic free media. DS scrambled negative (IDT), which did not target any gene (Si:NT), was used as a negative control.

31Transfection of expression plasmids was performed with lipofectamine 2000 (Invitrogen Life Technologies, Van Allen Way, Carlsbad, CA, USA) according to the manufacturer's instructions.

32Renilla and firefly luciferase expression plasmid were pGL-3 control (Promega, Madison, WI, USA). SV40 promoter on Luciferase-pGL3 was substituted by HSP70 promoter fragment obtained after digestion of p1730R (Enzo Life Sciences, Farmigdale, NY, USA) with XhoI and HindIII. The HSP70 (HSPA1A) (NM_005345.2) 3′UTR sequence was obtained by PCR with specific primers (Supplementary file 1A), from total HeLa cell DNA, cloned in pJet 1.2/blunt (Fermentas Thermosicentific, Pittburg, PA, USA) and re-cloned with XbaI in pGL3-control.

33eEF1A1 cDNA was obtained by RT, followed by PCR with specific primers (Supplementary file 1A), from total HeLa cell RNA. eEF1A1 cDNA was cloned with BamHI and ClaI into pHAGE-Ubc-RIG lentiviral vector (a gift from Gustavo Mostolslavsky) to achieve the expression of eEF1A fused to cherry and Flag in its N terminus (phage-ubc-cherry-flag-eEF1A1).

3430 μg of Phage-ubc-cherry-flag-eEF1A1 was transfected together with rev, gag/pol, tat and vsg plasmid using lipofectamine2000 (Life Technologies, Van Allen Way, Carlsbad, CA, USA) in 293T cells growing at 80% of confluence. Lentivirus were collected every 24 hr for 3 days after transfection, combined, and concentrated to 1 ml with lenti-X concentrator (Clontech, Mountain View, CA, USA). 2 × 106 MEFs or TIGs were infected with 300 ml of lentivirus and 4 days after infection cells were sorted for cherry expression in the flow cytometry core facility at the Albert Einstein College of Medicine.

35In vitro transcription and labeling of the 5′UTR and 3′UTR of HSP70 (NM_005345.2) mRNA, and a 200nt region of the ORF of β-actin (NM_00,101.3) RNA was performed with [32P] CTP (MP Biomedicals, Santa Ana, CA, USA) and the MEGAscript T7 kit (Ambion, Grand Island, NY, USA) from PCR products (Supplementary file 1A). PCR was performed using cDNA templates previously cloned in the pJem1.2/blunt vector from CloneJet PCR Kit (Fermentas Thermosicentific, Pittburg, PA, USA). PCR conditions for Takara polymerase were 32 × (95:15″, 60:15″, and 72:45″). PCR products were purified with the minElute PCR purification kit (Qiagen, Valencia, CA, USA). The sequences of all cDNA products were verified by sequencing (Macrogen, New York, NY, USA) from the T7 promoter (pJem1.2, Fermentas Thermosicentific, Pittburg, PA, USA) and internal regions using specific primers.

36Specific RNA probes for EMSA were synthesized in vitro and radiolabeled as described above. eEF1A1 protein was purified as follows: 50 ml of the HeLa S-100 lysate was applied to a 80-ml Q Sepharose column whose outlet was connected to a 40-ml CM Sepharose column. Both columns were equilibrated with 7.5/0 buffer (50 mM Tris–HCl, pH 7.5, 0.5 mM EDTA, 2 mM DTT, 10% glycerol) at a flow rate of 1 ml/min. The column was washed with 7.5/0 buffer until most of the unbound protein appeared in the flow-through (as determined at A280). The Q Sepharose column was then disconnected, and the CM Sepharose column was developed with 10 column volumes of a 50–650 mM gradient of KCl in 7.5/0 buffer at a flow rate of 1 ml/min, while fractions of 4 ml were collected. Fractions containing eEF1A were identified by immunoblotting, pooled, dialyzed against buffer containing 50 mM Tris–HCl, pH 7.5, 100 mM KCl, 0.5 mM EDTA, 2 mM DTT, 10% glycerol, and concentrated with Millipore Centrifugal Filter units (3 kDa MWCO). The concentrated protein was flash-frozen in liquid nitrogen and stored at −80°C. RNA EMSA was accomplished according to the protocol described by. Briefly, RNA probes were in vitro transcribed, end labeled, and gel purified. Indicated concentrations of purified eEF1A1 were incubated with 20 kcpm of HSPA1A 3′UTR RNA (∼0.1 pmol/µl) in the presence of 50 µM GTP for 30 min at room temperature.

37EMSA and RNAse treatments were performed as described previously. For supershift experiments, 10 μg of cell extracts were incubated with 1 μg of the specific antibody (HSF1 and HSF2 [Enzo Life Sciences, Farmigdale, NY, USA], eEF1A [Millipore, Billerica, MA, USA], IgG [Abcam, Cambridge, MA, USA]) for 1 hr prior to incubation with radiolabel HSE.

38Total cell RNA was extracted with the RNAeasy mini kit (Qiagen, Valencia, CA, USA). 2 μg of RNA were treated with turbo DNAse (Ambion, Grand Island, NY, USA) and reverse transcribed with random primers or oligo dT using MLV-RT (Promega, Madison, WI, USA). 5 μl of a 1:15 dilution of cDNA were used for QPCR with specific primers (Supplementary file 1A) and Power SYBR Green PCR master mix 2× (Applied Biosystems, Forest City, CA, USA) for 40 cycles in a 7300 real-time PCR system (Applied Biosystems, Forest City, CA, USA) according to the manufacturer's instructions. HSP70 Ct was normalized to GAPDH Ct for each condition and this value was related to the control value.

39Takara polymerase (TaKara, Moonachie, NJ, USA) was used for PCR following the instructions of the manufacturer.

40For ChIP experiments Real-time QPCR was performed in a Stratagene Mx3005p with Brilliant II SYBR Green kits (Stratagene, Netherlands) according to the manufacturer's instructions ans specific primers (Supplementary file 1A). Data were computed as described.

41Mock-transfected or eEF1A1 knocked down MDA-MB-231 cells were heat shocked for 45 min at 43°C and allowed to recover for 45 min at 37°C. At this time cells were treated with 100 μg/ml of cycloheximide for 15 min and collected for polysome purification using the protocol, centrifuge and ISCO fraction collector described by without modifications. Total RNA or RNA collected from polysome fractions was reverse transcribed and quantified as described above.

42Cells were labeled with 50 μCi of [35S]-methionine per ml (Easytag Express Protein Labeling Mix, Dupont/NEN) as described for both control and heat-shock conditions.

43Cell viability was measured by the MTT assay (Promega), and the OD was measured in an Infinite M200 96 well plate reader (Tecan) 24 hr after the second round of siRNA transfection. Cell death was quantified by flow cytometry (Becton Dickinson FACScalibur) after cells were stained with propidium iodide buffer (PI) (Life Technologies, Van Allen Way Carlsbad, CA, USA). Data were analyzed with Sumit software.

44Cells were washed twice in 1× PBS, snap-frozen in liquid nitrogen and resuspended in RIPA buffer (50 mM Tris–HCl, pH 7.5, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 1 mM EDTA, 150 mM NaCl, 1× protease inhibitor cocktail [Roche, Bransburg, NJ, USA] and 1× phosphostop [Roche, Bransburg, NJ, USA]) for 10 min at 4°C. 10 μg of protein were resolved on 4–12% SDS PAGE (Life Technologies, Van Allen Way Carlsbad, CA, USA) and transferred to a Nitrobound nitrocellulose membrane 0.45-μm pore size (Fisher). The membrane was blocked in 1× PBS-0.05% Tween 20 and 5% nonfat dry milk for 1 hr at room temperature and then incubated overnight at 4°C with specific antibodies (HSP70, HSF1, HSP27 (Enzo Life Sciences, Farmigdale, NY, USA), eEF1A1 (Millipore, Billerica, MA, USA), eEF1A2 (a gift from Jonathan Lee and GAPDH (Sigma, Sigma, Grand Island, NY, USA) and RNAPII (Santa Cruz, Dallas, Texas, USA)). After three washes in 1× PBS—0.05% Tween 20 membranes were incubated with horseradish peroxidase-conjugated goat-anti-mouse or goat-anti-rabbit antibody (GE-Healthcare, GE-Heathcare, Ho-Ho-Kus, NJ, USA), washed three times, and exposed to Kodak films using the ECL chemiluminescence system (GE-Heathcare, Ho-Ho-Kus, NJ, USA). Alternatively, IRD-ye and VRD-ye secondary antibodies were used before detection and quantification with the Odyssey infrared imaging system (LI-COR bioscience, Lincoln, Nebraska, USA).

45Cells were lysed in nonidet lysis buffer (25 mM Hepes, pH 7.9, 100 mM NaCl, 5 mM EDTA, 0.5% nonidet P40, 1× complete [Roche, Bransburg, NJ, USA], and phosphatase inhibitors [Roche, Bransburg, NJ, USA]) for 1 hr at 4°C. Lysates were clarified of non-protein components by centrifugation for 5 min at 14,000×g and pre-cleared with 25 μl of Dynabeads protein G (Life Technologies, Van Allen Way Carlsbad, CA, USA) washed three times with B150 buffer (20 mM Tris, pH7.4, 1 mM EDTA, 10% glycerol, 150 mM NaCl and 0.1% Triton) for 1 hr at 4°C. 1 mg of pre-cleared protein was IP overnight at 4°C with 5 μg of anti-eEF1A1 (Millipore, Billerica, MA, USA) or normal rabbit or mouse IgG (Abcam, Cambridge, MA, USA) followed by a 1-hr incubation with 50 μl of Dynabeads protein G (Life Technologies, Van Allen Way Carlsbad, CA, USA). After five washes in B150, the proteins were resolved by SDS-PAGE and analyzed by immunoblotting.

46RNA-IP experiments were performed as described previously. For each RNA-IP 5 μg of specific antibody: eEF1A (Millipore, Billerica, MA, USA), TPR1 (Santa Cruz Dallas, Texas, USA), PABP (Abcam, Cambridge, MA, USA), or IgG (Abcam, Cambridge, MA, USA) was used. Samples were analyzed by RT-QPCR or RT-PCR with specific primers (Supplementary file 1A).

47ChIP analysis were performed as described by Antibodies were: rabbit or mouse IgG (Sigma, Grand Island, NY, USA), RNAPII (2 μg of Antibody for 20 μg of relative amount of chromatin A260) (Santa Cruz, Dallas, Texas, USA), RNAPII S2 (2 μg of Antibody for 20 μg of relative amount of chromatin A260) (Abcam, Cambridge, MA, USA), eEF1A (4 μg of Antibody for 80 μg of relative amount of chromatin A260) (Designed in Nudler lab and produced in rabbits and purified by Prosci incorporated), HSF1 (2 μg of Antibody for 20 μg of relative amount of chromatin A260) (Enzo life sciences, Farmigdale, NY, USA), or Flag (2 μg of Antibody for 80 μg of relative amount of chromatin A260) (Sigma, Sigma, Grand Island, NY, USA). Pellets were resuspended in 40 μl of water and 1 μl analyzed by QPCR with specific primers (Supplementary file 1A). Percentage input for each data point was calculated by comparing the Ct value of sample to a standard curve generated from Ct of a 5-point serial dilution of input chromatin. Each immunoprecipitation was performed and assayed at least three times from independent samples.

48Transcriptional run-on assays were performed as described in using nuclei prepared from primary fibroblasts knocked down of eEF1A1 and heat shocked for 20 min at 43°C; radioactively α-32P-labeled RNA was prepared and hybridized to single-stranded oligos immobilized on nitrocellulose filters and visualized by PhosphorImager, quantitated by ImageJ.

49We performed RNA FISH on cultured cells to a protocol modified from previously published protocols. A set of Stellaris FISH probes single-labeled (quasar 670) oligonucleotides were designed to selectively bind to human HSPA1A or mouse HSPA1A (Supplementary file 1B) (Biosearch technologies, Petaluma, CA, USA) or MS2.

50We grew MEFs on coverslips in DMEM 10% FBS 1% pen-strep, then fixed them in 4% paraformaldehyde for 15 min at room temperature before washing and storing in PBSM (PBS supplemented with 5 mM MgCl2) at 4°C. Before hybridization, we permeabilized the cells 10 min in 0.5% triton X-100 in PBS, then washed them in PBS 10 min and incubated 60 min at 37°C in prehybridization solution (20% formamide, 2× SSC, 2 mg ml−1 BSA, and 200 mM vanadyl ribonucleoside complex). We then hybridized the probes to the cells for 4 hr in hybridization solution (10% dextran sulfate, 2× SSC, 10% formamide, 2 mg ml−1 BSA, 0.2 mg/ml E. coli tRNA, 200 mM vanadyl ribonucleoside complex, and 2U of RNAsin) supplemented with 10 ng DNA probe (for MS2) or 1.25 mM (for human and mouse HSPA1A) per coverslip. We washed coverslips twice 20 min at 37°C with prehybridization solution, then 10 min at room temperature in 2× SSC, and 10 min at room temperature in PBSM. We counterstained DNA with DAPI (0.5 mg l−1 in PBS). After a final wash in PBS, we mounted coverslips mounted on slides using ProLong gold reagent (Life Technologies, Van Allen Way Carlsbad, CA, USA).

51Images were acquired on an Olympus BX61 epifluorescence microscope with PlanApo 60×, 1.4NA, and UPlanApo 100×, 1.35NA, oil-immersion objectives (Olympus). An X-Cite 120 PC (Lumen Dynamics, Canada) light source was used for illumination, with filter sets 31,000 (DAPI, CMAC), 41007a (Cy3) and 41,008 (Cy5) (Chroma Technology, Bellows Falls, VT, USA). We acquired data using 21 optical sections with a z-step size of 0.2 μm using a CoolSNAP HQ camera (Photometrics) with a 6.4 μm-pixel size CCD. MetaMorph (Molecular Devices) software was used for instrument control as well as image acquisition.

52We automatically quantified the position and intensity of transcription sites within the nucleus of MEFs using software described in detail in. Briefly, the two-part algorithm begins by identifying the approximate position of transcription sites by applying a spatial bandpass filter to the image. The software identifies clustered pixels above a user-determined threshold as transcription sites. Then, the software fits a two-dimensional Gaussian, which approximates a point spread function, to determine the spatial position and intensity of each transcription site, after subtracting a local estimate of the fluorescent background. After the center position of TS spots in each channel were identified in 3-dimensional space, colocalization events were assigned when the distances between the closest spot of different colors were within two pixels.

53Values were expressed as standard error mean (SEM). Statistical significance was assessed by paired t test or ANOVA test.