Expanded polyalanine tracts function as nuclear export signals and promote protein mislocalization via eEF1A1 factor

Polyalanine (poly(A)) diseases are caused by the expansion of translated GCN triplet nucleotide sequences encoding poly(A) tracts in proteins. To date, nine human disorders have been found to be associated with poly(A) tract expansions, including congenital central hypoventilation syndrome and oculopharyngeal muscular dystrophy. Previous studies have demonstrated that unexpanded wild-type poly(A)-containing proteins localize to the cell nucleus, whereas expanded poly(A)-containing proteins primarily localize to the cytoplasm. Because most of these poly(A) disease proteins are transcription factors, this mislocalization causes cellular transcriptional dysregulation leading to cellular dysfunction. Correcting this faulty localization could potentially point to strategies to treat the aforementioned disorders, so there is a pressing need to identify the mechanisms underlying the mislocalization of expanded poly(A) protein. Here, we performed a glutathione S-transferase pulldown assay followed by mass spectrometry and identified eukaryotic translation elongation factor 1 α1 (eEF1A1) as an interacting partner with expanded poly(A)-containing proteins. Strikingly, knockdown of eEF1A1 expression partially corrected the mislocalization of the expanded poly(A) proteins in the cytoplasm and restored their functions in the nucleus. We further demonstrated that the expanded poly(A) domain itself can serve as a nuclear export signal. Taken together, this study demonstrates that eEF1A1 regulates the subcellular location of expanded poly(A) proteins and is therefore a potential therapeutic target for combating the pathogenesis of poly(A) diseases.

Polyalanine (poly(A)) diseases are caused by the expansion of GCN (where N refers to any of the four bases) trinucleotide repeats in the disease genes encoding the poly(A) amino acid tract in the disease proteins (1,2). Expansion of alanine tracts has been shown to cause several human pathological conditions, including congenital malformations and/or mental retardation (1,3). Up to the present, nine disease-associated proteins with alanine tract expansions have been identified (4). Eight of the nine poly(A) disease proteins are transcription factors with nuclear localization (1,5). Nuclear localization is essential for transcription factors to bind to DNA and activate target gene transcription (6). It has been reported that poly(A) disease proteins, including HOXD13, HOXA13, RUNX2, SOX3, FOXL2, ARX, and PHOX2B, are mislocalized in the cytoplasm when overexpressed in mammalian cell lines (7)(8)(9)(10)(11). Furthermore, the degree of cytoplasmic mislocalization increases as the length of the alanine tract increases (10,12). Polyalanine expansion in FOXL2 gives rise to its accumulation in the cytoplasm in a length-dependent manner. Expanded FOXL2 containing 19 alanine residues is found both in the nucleus and the cytoplasm. For FOXL2 with 37 alanine residues, all of the cells display cytoplasmic localization (9). Moreover, the degree of transcriptional activity of poly(A) disease protein decreases as the alanine tract increases (9). For instance, PHOX2B is a gene associated with congenital central hypoventilation syndrome (13). The majority of patients with central hypoventilation syndrome have a poly(A) repeat expansion mutation in PHOX2B protein (13). Luciferase assay results reveal an inverse correlation between trans-activation activity and the length of the poly(A) tract (7,14). Similarly, FOXL2, which encodes a transcription factor, is the causative gene in blepharophimosis, ptosis, and epicanthus inversus syndrome (15). The wild-type FOXL2 protein has a highly conserved 14 residue C-terminal alanine tract (16). Luciferase assay and quantitative reverse transcription-polymerase chain reaction (RT-PCR) of the target genes suggest that the longer a poly(A) tract appended to the disease protein is, the lower the trans-activation activity of FOXL2 (9). These studies suggest that poly(A) tract expansion causes protein cytoplasmic localization, impairs the trans-activation activity of these poly(A) disease proteins, and results in a loss-of-function phenotype.
Eukaryotic translation elongation factor 1 ␣ (eEF1A) 3 is in charge of the delivery of aminoacylation-tRNA (aa-tRNA) to the A site of ribosome during translation elongation (17). Aside from the canonical role in translation, eEF1A is involved in the nuclear export of proteins (18). Translation elongation factor EF-1 ␣ (Tef1) is the yeast orthologue of human eEF1A, and Tef1 was implicated in the tRNA nuclear export process in Saccharomyces cerevisiae (19). In mammalian cells, eEF1A was also reported to mediate the nuclear export of transcription-dependent nuclear export motif (TD-NEM) containing proteins, including the poly(A)-binding protein 1 (PABP1) and the von Hippel-Lindau (VHL) tumor suppressor protein (18) and SNAG-containing proteins, such as Snail1 (20). The above evidence clearly demonstrates that eEF1A plays a role in protein nuclear export. In this study, we showed that eEF1A1 regulates the localization of protein that carries a novel nuclear export signal (NES) in a poly(A) tract of disease proteins that are implicated in poly(A) disorders. Our findings provide a mechanistic explanation for how transcription factor dysregulation is triggered in poly(A) disease pathogenesis.

Continuous expanded poly(A) tract alters the localization of nuclear proteins
To determine the changes in subcellular localization of proteins with an expanded poly(A) tract, the Rev nuclear export assay (21) was employed. The wild-type HIV-1 viral Rev protein carries a functional nucleolus localization signal and NES, which direct Rev protein transports between the nucleus and cytoplasm through the nuclear pore complex (22). The nuclear export reporter protein, Rev(1.4)-EGFP, is a fusion protein that carries a mutant version of Rev (Rev(1.4)) and EGFP. The Rev(1.4) mutant protein lacks nuclear export activity and thus localizes primarily to the nucleus (21). To address the effect of poly(A) tracts on the subcellular localization of Rev(1.4), we fused Rev(1.4)-EGFP with either an unexpanded (A6) or expanded (A37) poly(A) repeat (Table 1) and subsequently transfected HEK293 cells with these constructs. The subcellular localization of proteins was determined and quantified according to Ref. 21. The nuclear export reporter protein, Rev(1.4)-EGFP, without a poly(A) tract (control) was found to predominantly localize to the nuclear compartment (Fig. 1, A  and B). The Rev(1.4)-NES-EGFP protein was used as the positive control to show that the insertion of a functional NES isolated from the HIV-1 Rev protein (21) caused cytoplasmic localization of the nuclear export reporter protein (Fig. 1, A and  B). Similar to the Rev(1.4)-EGFP control, the Rev(1.4)-A6-EGFP protein was found mainly in the nuclear compartment (Fig. 1, A and B), which suggests that the unexpanded poly(A) tract conferred no detectable effect on the subcellular localization of the Rev(1.4)-EGFP reporter. In contrast, the Rev(1.4)-A37-EGFP protein was found to reside in both the nuclear and cytoplasmic compartments (Fig. 1, A and B). This result suggests that the expanded poly(A) tract alters the localization of nuclear proteins, directing them to the cytoplasm. We next sought to investigate whether the continuity of the expanded poly(A) tract would be essential for the mislocalization. We generated two expanded polyAPA variant constructs, pRev(1.4)-A10PA9PA9PA9-EGFP and pRev(1.4)-A7PA6PA6 PA6PA6PA6-EGFP, which contain three or five proline residues within the expanded 37 poly(A) tract (Table 1). As a result, the continuity of the expanded poly(A) tract is disrupted, and the long poly(A) tracts are split into shorter alanine stretches. In contrast to Rev(1.4)-A37-EGFP which showed both nuclear and cytoplasmic localization, Rev(1.4)-A10PA9PA9PA9-EGFP and Rev(1.4)-A7PA6PA6PA6PA6PA6-EGFP proteins were predominantly found in the nuclear compartment ( Fig. 1, C and  D). This indicates that the continuity of expanded poly(A) tract is crucial for its effect of cytoplasmic mislocalization.
Next, we seek to determine the minimal length of poly(A) tract that would lead to cytoplasmic localization of nuclear proteins. We fused Rev(1.4)-EGFP with expanded poly(A) tract of different lengths, including A22, A25, A28, A31, and A34 (Table 1), and analyzed the localization of EGFP in HEK293 cells. The A22, A25, and A28 proteins resulted in 96, 98, and 98% of the cells with pure nuclear localization, respectively, which did not appear to be different from A6 (Fig. 1, E and F). But when the expanded poly(A) tract reached 31 repeats, we started noticing a difference, because A31 resulted in 74% of the cells with pure nuclear localization, and 26% of them showed both nuclear and cytoplasmic localization (Fig. 1, E and F). When the poly(A) tract reached 34 repeats, the majority of cells possessed a nuclear ϩ cytoplasmic localization pattern, with 78% of the cells having EGFP in both compartments and only 22% cells with pure nuclear localization (Fig. 1, E and F). From these results, we concluded that the minimal length of expanded poly(A) tract that would lead to a detectable increase of cytoplasmic localization is somewhere between 28 and 31 alanine repeats, whereas the threshold repeat length that would lead to nuclear ϩ cytoplasmic localization in the majority of cells lies between 31 and 34 repeats.

Eukaryotic translation elongation factor 1 ␣1 is a novel interacting partner of protein containing expanded poly(A) tract
To identify cellular protein-interacting partners of expanded poly(A) proteins, a GST-pulldown assay followed by mass spectrometry was performed. The HEK293 cell lysate was allowed to interact with purified bacterially expressed GST-A7 or -A37 fusion proteins in vitro. Mass spectrometric analysis of the eluents unveiled eEF1A1 as an expanded poly(A) protein (GST-A37)-interacting partner (supplemental Fig. S1). To confirm this interaction, we repeated the in vitro GST-pulldown experiment in HEK293 cells and detected eEF1A1 by means of Western blotting. The endogenous eEF1A protein was only detected in eluent of the GST-A37 capture and not the GST and GST-A7 controls ( Fig. 2A). We also showed that expanded poly(A) protein interacted with Myc-eEF1A1 by a GST-pulldown experiment in HEK293 cells (Fig. 2B). Taken together, our results indicate that eEF1A1 interacts specifically with protein carrying an expanded poly(A) tract but not with the unexpanded poly(A) tract.
The eFF1A1 protein can be structurally divided into three domains (Fig. 2C); domain I functions as a GTP-binding domain, domain II is involved in tRNA binding, and domain III is respon-  3 and A7(PA6) 5 ) localized in the nuclear compartment. Cell nuclei were stained with Hoechst 33342. Scale bar, 10 m. D, statistical analysis of C. E, subcellular localization of Rev(1.4)-poly(A)-EGFP protein with A22, A25, A28, A31, and A34. A22, A25, and A28 showed nuclear localization. A31 resulted in 26% of cells with both nuclear and cytoplasmic localization, whereas A34 resulted in up to 78% of cells with both nuclear and cytoplasmic localization. Scale bar, 10 m. F, statistical analysis (E). N, nuclear; C, cytoplasmic; C ϩ N, cytoplasmic and nuclear localization. Three independent transfection experiments were performed. At least 100 transfected cells were analyzed in each experiment. Error bars, S.E.

eEF1A1 modulates localization of expanded poly(A) proteins
sible for actin binding (23). To gain further insights into the expanded poly(A)/eEF1A1 interaction, we performed a GST-pulldown assay to determine which domain(s) in eEF1A1 is responsible for interacting with the expanded poly(A) tract. The result showed that eEF1A1 domain III and domain IIϩIII, but not domain I, II, and IϩII was able to interact with GST-A37 (Fig. 2D). This indicates that domain III in eEF1A1 is responsible for interacting with expanded poly(A) tract-containing protein. To further map out the expanded poly(A)-interacting region in eEF1A1, we transfected HEK293 cells with a series of Myc-tagged eEF1A1 deletion constructs (Fig. 2E) and used cell lysates to perform a GST-pulldown experiment. The results indicate that the region encompassing amino acids 429 -449 of eEF1A1 is essential for the protein to interact with expanded poly(A) tracts (Fig. 2F).
To determine whether the interruption of continuity of the expanded poly(A) tract would affect the interaction between eEF1A1 and expanded poly(A) tract-containing proteins, we performed a co-immunoprecipitation experiment in HEK293   (Table 1) and Myc-tagged eEF1A1. We found that Myc-eEF1A1 could be co-immunoprecipitated with Rev(1.4)-A37-EGFP but not the other Rev(1.4)-polyAPA-EGFP variants, including Rev(1.4)-A10PA9PA9PA9-EGFP and Rev(1.4)-A7PA6PA6PA6PA6PA6-EGFP (Fig. 2G). This result indicates that the continuity of the expanded poly(A) tract is an essential determinant for its interaction with eEF1A1.

eEF1A1 regulates the cytoplasmic localization of expanded poly(A) tract-containing protein
eEF1A is a GTP-binding protein that associates with protein synthesis (23). Besides its canonical role in protein translation, eEF1A has also been reported to be involved in nuclear export of proteins that carry the TD-NEM, such as the VHL tumor suppressor in mammalian cells (18). We thus speculated that eEF1A1 is involved in regulating the localization of expanded poly(A) tract-containing proteins.
To investigate the effect of eEF1A1 on the subcellular localization of poly(A)-containing protein, we analyzed the subcellular localization of Rev(1.4)-poly(A)-EGFP in HEK293 cells with or without knockdown of eEF1A1 using confocal microscopy. The nucleo-cytoplasmic distribution of protein was quantified by measuring the fluorescence intensity of the EGFP signals in the nuclear compartment and the cytoplasmic compartment and expressed as the nuclear/cytoplasmic (N/C) ratio. The result showed that knockdown of eEF1A1 expression caused an enrichment of the Rev(1.4)-A37-EGFP protein in the nuclear compartment but had no effect on the subcellular localization of the Rev(1.4)-A6-EGFP control protein (Fig. 3, A and  B). An increase in the percentage of cells with an N/C ratio Ͼ1 indicated that more EGFP signals were detected in the nucleus. For the reporter proteins that were tagged with a functional NES from the HIV-1 Rev, the knockdown of eEF1A1 has no effect on changing the N/C ratio, suggesting that eEF1A1 is not involved in the nuclear export mediated by the classical Rev NES (Fig. 3, A and B). The efficiency of the eEF1A1 siRNA was analyzed by immunoblotting. The results showed that eEF1A1 protein levels were largely reduced after eEF1A1 siRNA treatment (Fig. 3C). Our results also showed that overexpression of eEF1A1 enhanced the cytoplasmic localization of expanded poly(A) protein, whereas overexpression of eEF1A1 ⌬429 -449 had no such effect (Fig. 3, D-F). These data indicate that eEF1A1 regulates the subcellular localization of expanded poly(A) tract-containing protein.
During eukaryotic protein translation elongation, eEF1A is responsible for the enzymatic delivery of aa-tRNA to the A site of the ribosome, and eukaryotic translation elongation factor 2 (eEF2) is responsible for translocating the peptidyl-tRNA out of the A site to the P site (23). To eliminate the possibility that the eEF1A1 knockdown-mediated nuclear enrichment of expanded poly(A) proteins results from a blockage of general protein translation elongation, we examined the effect of eEF2 knockdown on the nucleo-cytoplasmic distribution of expanded poly(A) tract-containing protein. We found that the knockdown of eEF2 expression had no effect on the N/C ratio of Rev(1.4)-poly(A)-EGFP protein (Fig. 3, G-I). This suggests that the role of eEF1A1 in modulating the localization of expanded poly(A) protein is not related to its protein translation function.

Knockdown of eEF1A1 expression resulted in a nuclear enrichment of expanded poly(A)-containing SoxN
To relate the mislocalization of expanded poly(A) domain to poly(A) toxicity, we extended our study to an in vivo poly(A) disease transgenic fly model. The human sex-determining region Y-box 3 (SOX3) protein is a transcription factor that is expressed in the brain and plays a role in neural development (1). A mutation in SOX3 causing the addition of an extra 7-11 alanine residues to normal 15-alanine tract induces X-linked hypopituitarism, a prevalent poly(A) disease exhibiting mental retardation, short stature, and growth hormone deficiency (11,24,25). The SoxNeuro (SoxN) protein is the Drosophila ortholog of SOX3, and it plays a critical role in embryogenesis, including the formation of neuroblasts (26,27) and axonal patterning (28).
We first expressed A12-SoxN-EGFP and A34-SoxN-EGFP fusion proteins to determine the subcellular localization of overexpressed poly(A)-SoxN in SK-N-MC cells. The A12-SoxN-EGFP protein was found to be localized in the nucleus, whereas A34-SoxN-EGFP predominantly localized in the cytoplasm and formed protein aggregates (Fig. 4A). Next, we observed that the knockdown of eEF1A1 expression did not affect the distribution of the unexpanded A12-SoxN-EGFP. In contrast, the knockdown of eEF1A1 caused the nuclear enrichment of A34-SoxN-EGFP (Fig. 4A). The fluorescence intensity analysis of the two fluorescence patterns was performed along the line passed the whole cell. A significant amount of A12-SoxN-EGFP signal was found to overlap with nuclear stain in cells treated with both control siRNA and eEF1A1 siRNA. On the contrary, significantly less overlapping was observed for A34-SoxN-EGFP in cells treated with control siRNA. However, upon eEF1A1 knockdown, the overlapping was partially restored (Fig. 4A). To further investigate the effect of eEF1A1 knockdown biochemically, the nucleocytoplasmic fractionation assay was performed. The nuclear A34-SoxN protein levels rather than the A12-SoxN protein levels were found to be increased upon eEF1A1 knockdown (Fig. 4B). These results demonstrated that eEF1A1 mediates expanded poly(A) domain-containing SoxN nucleocytoplasmic distribution but does not alter wild-type SoxN localization.

Knockdown of Ef1a48D expression in Drosophila attenuated the toxicity induced by expanded poly(A) SoxN
To gain insights into the effect of poly(A) repeats in vivo, we established transgenic fly lines, UAS-A12-SoxN and UAS-A34-SoxN, which consist of the SoxN gene with a 12-alanine tract or a 34-alanine tract. Gmr-GAL4 driver was used to overexpress A12-SoxN or A34-SoxN in the fly eye, and the external eye phenotypes were examined. The overexpression of A34-SoxN resulted in abnormal eyes, characterized by a rough surface due to the disturbance of ommatidia architecture, whereas the overexpression of A12-SoxN did not display any effect on the external eye morphology, nor did the GAL4 driver control (  4C). Upon Ef1a48D knockdown by dsRNA, decreased levels of Ef1a48D expression were detected by RT-PCR. However, this decrease did not alter the transcript levels of A12-SoxN and A34-SoxN (Fig. 4D). Therefore, our results support the notion that Ef1a48D specifically modulates the toxicity induced by expanded poly(A)-containing SoxN without reducing its transcript levels.
In conclusion, the subcellular localization of expanded poly(A)-containing SoxN is tightly correlated to degeneration in poly(A) diseases. Expanded poly(A) tracts exert their toxicity by inducing the mislocalization of the disease proteins from the nucleus to the cytoplasm. Ef1a48D modifies expanded poly(A)containing SoxN toxicity by mediating its mislocalization.

Knockdown of eEF1A1 partially restores the transcription function of poly(A)-containing protein
Eight of nine poly(A) disease genes encode for transcription factors, and such nuclear localization is essential for transcription factors to function. To determine the effect of expanded poly(A) tracts on the transcriptional activity of transcription factors, we performed an unconventional luciferase reporter assay to examine the regulation of gene expression mediated by the GAL4 transcription activator fused with either an unex-

eEF1A1 modulates localization of expanded poly(A) proteins
panded or expanded poly(A) tract. The conventional luciferase assay measures the enzymatic activity of luciferase through detecting the luminescence of its substrate. Because the translational machinery may be affected by the knockdown of eEF1A1, we assayed for the firefly luciferase transcript instead in our experiments. The pFR-Luc plasmid contains a promoter with five tandem repeats of the GAL4 binding sites that control the expression of the firefly luciferase gene (29,30). When constructs containing GAL4-poly(A) were co-transfected with the pFR-Luc reporter plasmid, the transcription of the luciferase gene would be induced. The expression level of luciferase thus reflects the transcriptional activity of GAL4-poly(A). At the same time, phRL-TK, which is Renilla co-reporter vector, was also co-transfected to serve as an internal control for normalization of differences in transfection efficiency (31). The result showed that GAL4-A37-FLAG caused a significant decrease in firefly luciferase expression level compared with the GAL4-A6-FLAG control (Fig. 5). We detected the localization of GAL4poly(A)-FLAG proteins. The GAL4-A6-FLAG was found diffusely localized in both nucleus and cytoplasm, whereas GAL4-A37-FLAG was found to be mainly localized in the cytoplasmic component (supplemental Fig. S2). The subcellular localization result could explain why expanded GAL4-A37-FLAG protein demonstrated a reduced transcriptional activity. We have shown above that the knockdown of eEF1A1 expression partially corrected the cytoplasmic mislocalization of expanded poly(A) tract-containing protein (Fig. 3, A and B). Here, we speculated that eEF1A1 knockdown would restore the transcriptional activity of the expanded poly(A)-containing GAL4. As expected, the reduction of firefly luciferase transcription level in GAL4-A37-FLAG-transfected cells was partially restored upon eEF1A1 knockdown (Fig. 5).

Expanded poly(A) tract possesses nuclear export activity mediated by nuclear eEF1A1
We hypothesized that the expanded poly(A) tract may serve as a novel nuclear export signal that causes the cytoplasmic localization of poly(A) disease proteins. To determine whether the Rev(1.4)-A37-EGFP protein detected in the cytoplasm was exported out from the nuclei, we performed the fluorescence loss in photobleaching (FLIP) nuclear export assay in living cells (18). In brief, a small region of interest (ROI) of the cytoplasmic compartment of the target cell was repeatedly photobleached using a high-power 488-nm laser and followed by an immediate postbleach capture using a low-power 488-nm laser. The Rev(1.4) possesses only nuclear import activities but lacks nuclear export activities. Thus, if the expanded poly(A) tract mediates nuclear export activities, then the nuclear export of the reporter protein would result in the dilution of the remaining fluorescence in the nuclear compartment. The nuclear export activity of the target protein was determined by comparing the relative fluorescence loss of protein in the nuclear compartment over time. After photobleaching of the cytoplasmic compartment, a statistically significant decrease in the relative nuclear fluorescence intensity of the Rev(1.4)-A37-EGFP protein was observed when compared with the Rev(1.4)-A6-EGFP control (Fig. 6, A and B). This result demonstrates nuclear export activity of Rev(1.4)-A37-EGFP protein from the nuclear compartment to the cytoplasmic compartment.
To further obtain a direct visualization of a nuclear export event, we next performed the photoactivatable GFP (PA-GFP) nuclear export assay. PA-GFP (32) is a photoactivatable variant form of EGFP (33) that has a single amino acid substitution at the 203 position from threonine to histidine (T203H) in EGFP. PA-GFP proteins only emit green fluorescence after photoactivation at 405 nm. The photoactivation property of PA-GFP allows the capturing of nucleocytoplasmic trafficking of proteins in living cells (34). In the PA-GFP nuclear export assay, the nucleolus was selected as the ROI for photoactivation using the 405-nm laser. Because the newly synthesized protein in the cytoplasm was not photoactivated, the cytoplasmic fluores-    (Fig. 6C). In contrast, the nuclear photoactivated Rev(1.4)-A37-PA-GFP protein was found in the cytoplasm starting from 1 min post-photoactivation (Fig. 6C). This strongly suggests that the photoactivated Rev(1.4)-A37-PA-GFP protein was exported out from the nuclear compartment to the cytoplasmic compartment. These results clearly demonstrate that the expanded poly(A) tract possesses NES activity.
If eEF1A1 is a mediator of expanded poly(A) tract-induced nuclear export, it should be present in the nucleus to interact with the poly(A) tract. However, it has been previously reported in a number of studies that eEF1A is exclusively localized to the cytoplasm (35,36); thus, there is a need to verify the presence of eEF1A1 in the nucleus. We performed Western blotting analysis on cytoplasmic and nuclear fractions

eEF1A1 modulates localization of expanded poly(A) proteins
extracted from HEK293 and SK-N-MC cells to determine the presence of nuclear eEF1A1. Despite the majority of eEF1A1 being in the cytoplasm, there was a detectable amount of eEF1A1 in the nucleus (Fig. 6D). This result is consistent with the findings of a recent study, which demonstrated that eEF1A is capable of shuttling between the nucleus and the cytoplasm (20).
Because eEF1A1 has been detected in both the nucleus and the cytoplasm, we would like to determine which compartment's eEF1A1 mediates the cytoplasmic localization of poly(A)-containing protein. Hence, we co-expressed eEF1A1-NLS or eEF1A1-NES in HEK293 cells with Rev(1.4)-A6-EGFP or Rev(1.4)-A37-EGFP. eEF1A1-NLS was found to localize to the nucleus, whereas eEF1A1-NES localized to the cytoplasm (Fig. 6E). To our surprise, both eEF1A1-NLS and eEF1A1-NES resulted in increased cytoplasmic localization of Rev(1.4)-A37-EGFP (Fig. 6, E and F). Thus, our results suggest that nuclear eEF1A1 probably mediates nuclear export of poly(A)-containing proteins, whereas cytoplasmic eEF1A1 may also mediate cytoplasmic retention of poly(A)-containing proteins.
In summary, we demonstrated that the expanded poly(A) tract can serve as a novel NES, leading to nuclear export of poly(A)-containing nuclear proteins. Nuclear eEF1A1 is probably involved in regulating this nuclear export process, whereas cytoplasmic eEF1A1 may also participate in the cytoplasmic retention of poly(A)-containing proteins. These results suggest that eEF1A1 plays dual roles in both the nucleus and the cytoplasm to mediate the cytoplasmic mislocalization of proteins with an expanded poly(A) tract. Because the cytoplasmic mislocalization of poly(A)-containing proteins is a major cause of poly(A) diseases, eEF1A1 may be a potent therapeutic target in treating these diseases.

Discussion
The main finding of this study is the demonstration of the continuous expanded poly(A) tract as a novel NES and the regulation of poly(A) protein localization by eEF1A1, which provides new insights into the underlying mechanisms of poly(A) diseases. Our study identifies eEF1A1 as an interacting partner of the expanded poly(A) domain, and it serves as a modulator of the poly(A)-induced mislocalization of transcription factors. The depletion of eEF1A1 mitigates the nuclear loss of poly(A)containing transcription factors and partially restores their transcriptional activities.
Previous studies have noted that the poly(A) disease symptoms are linked to loss of function of the poly(A) disease proteins (15,(37)(38)(39). These wild type transcription factors localize in the nucleus, whereas the expanded poly(A) mutant proteins are mislocalized in the cytoplasm (7)(8)(9)(10)(11). Furthermore, the mislocalization of poly(A) disease proteins leads to loss of function of these poly(A) disease proteins and reduces its transcriptional activity (7,9,40,41). In addition, the expanded poly(A) tract is sufficient to confer cytoplasmic localization to a YFP reporter protein (12). Therefore, it is believed that the expanded alanine tract per se contributes to the cytoplasmic localization of poly(A) disease proteins. Altogether, current studies suggest that the pathogenesis of poly(A) diseases is closely related to the cytoplasmic mislocalization of poly(A) disease proteins. Therefore, the aim of our study is to uncover the mechanisms that underlie this mislocalization.
In this study, we performed a GST-pulldown assay and mass spectrometry analysis and identified eEF1A1 as an expanded poly(A) protein-interacting partner that binds specifically to the expanded poly(A) sequences (Fig. 2, A and B). At the structural level, human eEF1A1 is divided into three domains, referred to as domains I, II, and III (42), and binds GTP, tRNA, and actin, respectively (23). eEF1A1 binds to exportin 5 in a complex with aa-tRNA (35), which involves domain II (23). But unlike the eEF1A1/exportin 5 interaction, expanded poly(A) tract interacts only with domain III of eEF1A1 (Fig. 2D), indicating that tRNA is not required in the interaction between eEF1A1 and expanded poly(A) domain. Our data further show that amino acids 429 -449 of human eEF1A1 are essential for interaction with the expanded poly(A) domain (Fig. 2F). Our results also demonstrate that the continuity of the expanded poly(A) tract is essential to the poly(A)-containing protein's interaction with eEF1A1 (Fig. 2G), and this interaction is a major determinant of the poly(A)-containing protein's mislocalization in the cytoplasm (Fig. 3, D and E).
The canonical role of eEF1A is protein synthesis in the cytoplasm (43), but many studies have reported its involvement in nuclear export functions as well. Previous mammalian studies have demonstrated eEF1A as a novel component of the nuclear protein export machinery, participating in the nuclear export of selected RNAs and proteins, such as tRNAs (17), SNAG-containing proteins (20), the VHL tumor suppressor, and PABP1 (18). However, it has been postulated that the nuclear export function of eEF1A may be executed in the cytoplasm instead of the nucleus (17,18,44). This is because eEF1A has been found to be exclusively localized to the cytoplasm in immunocytochemistry experiments (35,36). Interestingly, a recent study demonstrated that eEF1A is localized in the nucleus when it is in a complex with Snail1 but localized in the cytoplasm when it is alone (20). Thus, eEF1A can be shuttled between the nucleus and the cytoplasm, depending on what proteins it is interacting with. Our Western blotting results show that despite the majority of eEF1A1 being cytoplasmic, there is still a detectable amount of nuclear eEF1A1 in HEK293 and SK-N-MC cells (Fig.  6D), which is likely to mediate nuclear export of poly(A)-containing proteins. This is supported by the fact that the overexpression of nuclear eEF1A1 resulted in increased cytoplasmic localization of expanded poly(A)-containing Rev(1.4)-EGFP protein (Fig. 6, E and F). However, overexpression of cytoplasmic eEF1A1 also had similar effects (Fig. 6, E and F); thus cannot exclude the possibility that cytoplasmic eEF1A1 may contribute to the cytoplasmic retention of expanded poly(A)containing proteins as well.
Apart from its canonical role in translation, eEF1A is associated with protein degradation as well (23). Thus, we sought to determine whether eEF1A1 could be involved in the degradation of expanded poly(A)-containing proteins. However, we found that the knockdown of eEF1A1 expression did not affect the transcript and protein levels of poly(A)-containing proteins (supplemental Fig. S3, A and B), indicating that its effects on poly(A)-containing proteins are unlikely to be associated with protein degradation. APRIL 7, 2017 • VOLUME 292 • NUMBER 14

JOURNAL OF BIOLOGICAL CHEMISTRY 5793
Our findings in this study demonstrate that the expanded poly(A) tract can function as a novel NES. We used the Rev(1.4)-EGFP reporter protein to investigate the nuclear export property of expanded poly(A) domain. Our results show that the insertion of a 37-alanine tract causes cytoplasmic localization of the Rev(1.4)-EGFP protein (Fig. 1, A and B). This cytoplasmic localization is in line with previous poly(A) disease studies (7, 9 -11, 24, 40, 45). For instance, it has been reported that the expanded poly(A) disease proteins, including HOXD13, HOXA13, RUNX2, and SOX3, are cytoplasmically localized when overexpressed in COS-1 cells (10). In addition, it has also been reported that YFP-poly(A) protein of the 29-alanine tract is predominantly localized in the cytoplasm (12). Our FLIP nuclear export assay then demonstrates that the expanded poly(A) tract mediates nuclear export of Rev(1.4)-A37-EGFP protein (Fig. 6, A and B). The PA-GFP nuclear export assay allows a continuous monitoring of nucleocytoplasmic trafficking and illustrates the nuclear export event of expanded poly(A) protein in living cells (Fig. 6C). The data from these different nuclear export assays all support the notion that the expanded poly(A) tract possesses nuclear export activity. Therefore, although we cannot exclude the possibility of nuclear import impairment and/or cytoplasmic retention, we have at least demonstrated its function as a novel NES, mediating nuclear export.
The classical NES is a hydrophobic leucine-rich sequence, whose nuclear export activity is mediated by chromosome region maintenance 1 (CRM1), also known as exportin 1 (XPO1) (46,47). The knockdown of eEF1A1 does not alter classical NES-mediated nuclear export (Fig. 3, A-C), suggesting that eEF1A1 and CRM1 operate independently and that eEF1A1 is not a general regulator for protein nuclear export. From our results, although the eEF1A1 protein level was reduced about 90% after eEF1A1 siRNA treatment (Fig. 3C), only about 40% of nuclear export of Rev(1.4)-A37-EGFP was suppressed. There are at least two possible interpretations to explain this result. First, eEF1A1 may be only a factor involved in the nuclear export but not the actual receptor. Genome-wide siRNA screening may further be performed to identify components of the expanded poly(A) tract nuclear export pathway(s). Second, multiple receptors may contribute to the nuclear export of the expanded poly(A) tract. It is not uncommon for more than one receptor to regulate the nuclear export of a particular type of protein. For instance, the nuclear export of peroxisome proliferator-activated receptor is regulated by calreticulin and XPO1 (48). One of the major isoforms of thyroid hormone receptor, TR␣1, harbors multiple NESs and exports from the nucleus via an XPO1-dependent pathway in which calreticulin plays a role as a chaperone (49). Both XPO1-dependent and XPO1-independent pathways contribute to the nuclear export of constitutive androstane receptor, although it harbors only an XPO1-independent NES (50 -52). Although we did not perform a genome-wide siRNA screen, we carried out a screen using a candidate gene siRNA knockdown approach to identify other molecules that may regulate the nuclear export of poly(A)-containing proteins (supplemental Fig. S4, A and B).
The candidates in our screen include multiple exportins, such as XPO1, XPO4, XPO5, XPO6, XPO7, and XPOT. It has been proposed in previous studies that XPO5 may mediate the export of eEF1A protein from the nucleus (36). However, our candidate gene siRNA knockdown data show that the knockdown of XPO5 and related genes does not alter the subcellular localization of the Rev(1.4)-A37-EGFP protein (supplemental Fig. S4, A and B). This result is consistent with a previous study indicating that eEF1A mediates nuclear export of TD-NEMcontaining proteins, such as VHL tumor suppressor protein, via an XPO5-independent pathway (18).
Our results show that overexpression of eEF1A1 enhances the cytoplasmic localization of expanded poly(A) protein, whereas overexpression of eEF1A1 ⌬429 -449 has no such effect (Fig. 3, D-F). This effect on poly(A)-containing protein localization is not due to the enhancement of general protein translation, because we do not observe changes in the levels of protein translation when eEF1A1 is overexpressed (supplemental Fig. S5), which is consistent with the findings of a previous study (53). In fact, even the overexpression of eEF1A1 ⌬429 -449 has no detectable effect on general protein translation (supplemental Fig. S5).
Whereas eEF1A1 is responsible for the enzymatic delivery of aa-tRNA to the A site of the ribosome, eEF2 is responsible for translocating the peptidyl-tRNA out of the A site to P site (23). Our results show that the knockdown of eEF2 does not affect the subcellular localization of Rev(1.4)-A37-EGFP protein (Fig.  3, G-I), which indicates that the knockdown of eEF1A1 expression suppresses the cytoplasmic mislocalization of expanded poly(A) tract specifically and that the cytoplasmic localization of expanded poly(A) protein was not due to the blockage of general protein translation.
Consistent with these findings, expanded poly(A) protein expression does not perturb the cellular function of eEF1A1 in protein translation (supplemental Fig. S6); nor does it affect the nuclear export function of eEF1A1, because VHL is still capable of exiting the nucleus when Rev(1.4)-A37-EGFP is being expressed (supplemental Fig. S6). Taken together, the canonical translation functions and nuclear export functions of eEF1A1 are unaffected by its interaction with expanded poly(A) proteins.
Protein misfolding and aggregation are also common characteristics of poly(A) expansion proteins (54). Expression of EGFP fusion with 19 alanine residues in COS-7 cells forms aggregates and induces cell death (54). The A37-EGFP protein forms cytoplasmic aggregates in the eye imaginal disc of Drosophila third instar larvae and causes smaller and rough eye phenotype (55). These studies suggest that aggregate formation in cytoplasm also contributes to poly(A) pathogenesis (10, 54 -58). In line with the aforementioned findings, we find that overexpression of A34-SoxN-EGFP results in the formation of aggregates in the cytoplasm in SK-N-MC cells (Fig. 4A). Because protein misfolding is a common feature among poly(A) expansion proteins, we wondered whether it underlies the effect of how eEF1A1 regulates the subcellular localization of poly(A) expansion proteins. However, our results show that the knockdown of eEF1A1 does not induce changes in the transcript level of binding immunoglobulin protein (BiP), a chaperone protein that regulates protein folding (supplemental Fig.  S3A). This indicates that the regulation of subcellular localiza-

eEF1A1 modulates localization of expanded poly(A) proteins
tion of poly(A) expansion proteins by eEF1A1 is not related to protein misfolding.
To correlate the nuclear export activity of the expanded poly(A) domain to poly(A) toxicity in an in vivo model, we assessed the effect of Ef1a48D on eye degeneration of poly(A)-SoxN transgenic flies. Our data show expanded poly(A) induced toxicity in the fly eye, which is consistent with the previous reports that A37 is toxic in cell culture (59), and its expression in the fly eye results in eye degeneration (55). Our results demonstrate that the knockdown of Ef1a48D expression attenuates the expanded poly(A)-SoxN-induced toxicity and eye degeneration in vivo (Fig. 4C). This suggests that Ef1a48D is a genetic modifier of expanded poly(A)-SoxN-induced toxicity and degeneration, possibly by limiting the cytoplasmic localization of poly(A)-SoxN.
Nuclear localization is essential for transcription factors to function. In our study, we combined a transcription factor, GAL4, with an unexpanded or expanded poly(A) tract to investigate the transcriptional activities of poly(A)-containing protein. Our results suggest that the transcription activity of GAL4-A37-FLAG decreases significantly compared with GAL4-A6-FLAG control (Fig. 5). Intriguingly, siRNA-silencing eEF1A1 rescues the transcriptional activity of GAL4-A37-FLAG (Fig.  5), which indicates that the knockdown of eEF1A1 partially restores the cellular function of the poly(A)-containing transcription factors. It has recently been reported that the expanded poly(A) PHOX2B disease protein adopts a cytoplasmic localization, and it impairs the wild-type PHOX2B protein function by heterodimerizing with the wild-type protein, which inhibits the nuclear import of the wild-type protein (60). These reports are consistent with our findings, and both studies provide evidence that the nuclear localization is essential for poly(A) proteins to function as transcription factors. In our working model of poly(A) disease pathogenesis, the expanded poly(A) domain serves as a nuclear export signal and causes cytoplasmic localization of poly(A) disease-associated transcription factors. This in turn reduces their cellular transcriptional activities, resulting in gene dysregulation and lossof-function toxicity (Fig. 7). Simultaneously, cytoplasmic expanded poly(A)-containing proteins form aggregates and may exert the gain-of-function toxic effect. eEF1A1 interacts with expanded poly(A) domain and is a mediator for the cytoplasmic translocation of expanded poly(A)-containing proteins. Depletion of eEF1A1 suppresses the cytoplasmic translocation of expanded poly(A) tract-containing proteins, which partially restores the cellular transcriptional activities of expanded poly(A) transcription factors in the nucleus. Therefore, the knockdown of eEF1A1 expression can partially rescue toxicities of poly(A) caused by cytoplasmic mislocalization of the disease proteins (Fig. 7).
In conclusion, expanded poly(A) tracts possess NES activity and contribute to the cytoplasmic localization of poly(A)-containing proteins. Our study uncovers the role of eEF1A1 in modulating expanded poly(A) protein subcellular localization and identifies it as a novel therapeutic target for poly(A) diseases.

FLIP nuclear export assay
HEK293 cells were seeded onto a 35-mm -Dish (ibidi) at a density of 1.8 ϫ 10 5 . Twenty-four hours later, the cells were transfected with 1 g of pRev(1.4)-poly(A)-EGFP constructs for 12 h. The cells were subsequently stained with ER-Tracker TM Red (1:1,000; Life Technologies) at 37°C for 15 min. Then cells were maintained in DMEM without phenol red (Life Technologies). In the FLIP experiment, the subcellular localization of EGFP fusion proteins was monitored using 10% power of a 488-nm laser at 4 s/pixel. All of the prebleach, photobleaching, and postbleach steps were controlled automatically by the time controller module of Olympus Fluoview version 2.0c software. In the prebleach step, a single image was taken with 10% power of the 488-nm laser at 4 s/pixel to record the initial subcellular localization of the EGFP fusion proteins. Then square ROIs of the cytoplasmic region of transfected cells were repeatedly bleached five times with 100% power of the 488-nm laser at 200 s/pixel, and this was followed by an immediate capture of a postbleach image with 10% power of the 488-nm laser at 4 s/pixel. Quantification of fluorescence intensity was performed using ImageJ version 1.45s software (National Institutes of Health, Bethesda, MD).

Photoactivatable GFP nuclear export assay
HEK293 cells were prepared as described under "FLIP nuclear export assay." Cells were transfected with pRev(1.4)poly(A)-PA-GFP constructs for 48 h before the PA-GFP assay. The preactivation, photoactivation, and postactivation steps were controlled automatically by the time controller module of Olympus Fluoview version 2.0c software. At preactivation, a single image was taken with 10% 488-nm laser at 4 s/pixel to show that no fluorescence was observed before photoactivation. Then circular ROIs at the nucleolus of transfected cells were selected, and 20% power of the 405-nm laser at 100 s/pixel was used to photoactivate the selected ROIs. The subcellular localization of activated Rev(1.4)-poly(A)-PA-GFP proteins were monitored for 15 min at 1-min intervals using 10% power of the 488-nm laser at 4 s/pixel.

GST-pulldown assay
Expression of GST-poly(A) fusion proteins in Escherichia coli strain BL21 (DE3) was induced by 0.1 mM isopropyl ␤-D-1thiogalactopyranoside (Shanghai Biocolor Bioscience and Technology). A bacterial pellet containing recombinant protein was collected by centrifugation at 4,000 ϫ g at 4°C for 10 min. Then the pellet was resuspended in 10 ml of GST binding buffer (pH 7.4) containing 50 mM Tris-HCl, 1 mM EDTA, 50 mM NaCl, and 10% glycerol and sonicated using a Sonifier 450 sonicator (Branson). The protein sample was centrifuged at 16,000 ϫ g at 4°C for 15 min. A volume of 1 ml of supernatant was incubated with 50 l of glutathione-Sepharose 4 Fast Flow (GE Healthcare) at 4°C for 2 h with rotation. The glutathione-Sepharose 4 Fast Flow beads coupled with GST-poly(A) proteins were washed three times with GST binding buffer. Next, HEK293 total lysate was incubated with glutathione-Sepharose 4 Fast Flow beads coupled with GST-poly(A) proteins at 4°C overnight with rotation. The glutathione-Sepharose 4 Fast Flow beads coupled with GST fusion proteins and interacting proteins were washed five times with 1 ml of GST washing buffer (pH 7.4) containing 50 mM Tris, 1 mM EDTA, 500 mM eEF1A1 modulates localization of expanded poly(A) proteins NaCl, 10% glycerol, and 1% Nonidet P-40. Finally, 60 l of 2ϫ SDS sample buffer was added, and the interacting proteins were eluted by boiling at 99°C for 10 min. The protein samples were analyzed by SDS-PAGE and followed by Coomassie Blue staining to detect GST-poly(A) expression or by immunoblotting to detect eEF1A1 interacted with GST-A37.

Co-immunoprecipitation
Immunoprecipitation was essentially performed as described previously (63). In brief, the HEK293 cells were lysed using 1 ml of co-immunoprecipitation (co-IP) buffer containing 10 mM HEPES (pH 7.5), 5 mM MgCl 2 , 142.5 mM KCl, 1 mM EDTA, 10% glycerol, and 1% Triton X-100 supplemented with 10 l of protease inhibitor mixture (Sigma) on ice for 10 min. The cell lysates were placed at 4°C for 1 h with rotation and followed by centrifugation at 16,100 ϫ g, 4°C for 30 min. The supernatant was collected, and 40 l of it was saved as "input." For each sample, 450 l of supernatant was added to the 20 l of anti-c-Myc-agarose affinity gel antibody produced in rabbits (Sigma), 450 l of which was added to 20 l of protein A-agarose fast flow beads (Sigma) added with 0.1 g/ml rabbit IgG (Vector Laboratories). The binding step was performed at 4°C overnight with rotation. The agarose beads were washed five times with co-IP buffer and then resuspended with 30 l of 2ϫ SDS sample buffer and boiled at 99°C for 10 min. Immunoprecipitates were analyzed by Western blotting.

Nucleocytoplasmic fractionation
The cells were lysed with fractionation buffer containing 10 mM Tris-HCl (pH 7.4), 50 mM NaCl, 3 mM MgCl 2 , and 0.5% Nonidet P-40. The cell lysate was centrifuged at 16,000 ϫ g for 10 min at 4°C. The supernatant was collected as cytoplasmic fraction. The pellet was washed three times with fractionation buffer. Each time the samples were centrifuged at 500 ϫ g for 5 min at 4°C. Finally, the pellet was resuspended with resuspension buffer containing 20 mM Tris-HCl (pH 8.0) and 2% SDS and saved as the nuclear fraction.

External eye phenotypic detection of adult flies
The adult fly eyes were observed under a stereomicroscope (SZX-12, Olympus, Tokyo, Japan). External eye images were captured by SPOT Insight CCD camera controlled by the SPOT Advanced software (version 5.1; Diagnostic Instruments Inc.).

Statistical analyses
Statistical analyses were performed using a two-tailed unpaired Student's t test, except in Fig. 1G. All experiments were performed at least three times. Results are presented as means Ϯ S.E. A p value of less than 0.05 was considered statistically significant.