Alternative Splicing of Staufen2 Creates the Nuclear Export Signal for CRM1 (Exportin 1)*

Mammalian Staufen2 (Stau2), a brain-specific double-stranded RNA-binding protein, is involved in the localization of mRNA in neurons. To gain insights into the function of Stau2, the subcellular localization of Stau2 isoforms fused to the green fluorescence protein was examined. Fluorescence microscopic analysis showed that Stau2 functions as a nucleocytoplasmic shuttle protein. The nuclear export of the 62-kDa isoform of Stau2 (Stau2 62 ) is mediated by the double-stranded RNA-bind-ing domain 3 (RBD3) because a mutation to RBD3 led to nuclear accumulation. On the other hand, the shorter isoform of Stau2, Stau2 59 , is exported from the nucleus by two distinct pathways, one of which is RBD3-medi-ated and the other of which is CRM1 (exportin 1)-de-pendent. The nuclear export signal recognized by CRM1 was found to be located in the N-terminal region of Stau2 59 . These results suggest that Stau2 may carry a variety of RNAs out of the nucleus, using the two export pathways. The present study addresses the issue of why plural Stau2 isoforms are expressed in neurons. The localization of mRNA to defined subcellular regions en-ables the spatial and temporal control of gene expression (1–4). In neurons, the dendritic of certain mRNAs and the subsequent local synthesis of proteins are to play a role in neuronal plasticity. Mammalian Staufen2

The localization of mRNA to defined subcellular regions enables the spatial and temporal control of gene expression (1)(2)(3)(4). In neurons, the dendritic transport of certain mRNAs and the subsequent local synthesis of proteins are thought to play a role in neuronal plasticity. Mammalian Staufen2 (Stau2), 1 a dsRNA-binding protein, was shown to be a homolog of Droshophila melanogaster Staufen (dmStau) (5)(6)(7). It has been shown that dmStau anchors bicoid mRNA at the anterior of the oocyte and oskar mRNA to the posterior during early development (8,9). Mammalian Stau2 is mainly expressed in the brain and is involved in mRNA transport in neurons (6,7). Stau2 exists in the somatodendritic compartment of cultured hippocampal neurons. In dendrites, Stau2 associates with RNA granules that contain other RNA-binding proteins, ribosomal subunits, translation factors, and motor proteins (10,11). RNA granules are transported in dendrites via microtubules (12). Therefore, RNA granules are thought to play an important role in the targeting of RNA in neurons, functioning as cellular trafficking machinery (13), although the formation process by which this macromolecular complex is formed is currently unknown. In addition, it is known that Stau2 has at least three splicing variants with molecular masses of 62, 59, and 52 kDa (7). However, the issue of whether these isoforms are functionally different has not been resolved.
Because of this, the questions arise as to why such Stau2 isoforms are present in cells and whether each isoform functions divergently. To answer these questions, we investigated the intracellular dynamic behavior of two isoforms of Stau2. In this report, we show that Stau2 functions as a nucleocytoplasmic shuttle protein. In addition, we demonstrate that these two isoforms are distinctly exported from the nucleus, suggesting that each isoform may carry different sets of RNAs out of the nucleus.

EXPERIMENTAL PROCEDURES
Cloning of Rat Staufen2 59 -Using information on the expressed sequence tags, the rat Staufen2 59 was cloned from a rat brain cDNA library (Clontech) by PCR using the following primers: forward, 5Ј-CCGGACCATGCTTCAGATAAATCAGATGTTTTCGGTGC-3Ј, and reverse, 5Ј-ATGGATCCTAGATGACCGACTTTGATTTCTTGCAGTCC-TG-3Ј. The PCR product was digested with BamHI, cloned into the pEGFP-C1 vector (Clontech), and sequenced.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AY684789 and AY684790.
Cell Cultures and Transfection-HeLa cells were incubated in Dulbecco's modified Eagle's medium (Sigma) supplemented with 10% heatinactivated fetal bovine serum at 37°C in a 5% CO 2 atmosphere. Expression vectors were transfected using the Effectene transfection reagent (Qiagen) using the protocol recommended by the supplier. Leptomycin B was added to medium to a final concentration of 10 nM.
Primary Culture of Neurons and Transfection-The hippocampus was dissected from embryonic day 18 Sprague-Dawley rats and digested with 0.25% trypsin in Hanks' buffered salt solution. Neurons were suspended in neurobasal medium (Invitrogen) supplemented with B27 and 0.5 mM L-glutamine and then plated at 3.0 ϫ 10 5 /35-mm plate on polyethylenimine-coated tissue culture dishes at 37°C in a 5% CO 2 atmosphere. On day 10 after plating, neurons were transfected with expression vectors using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. At 24 h after transfection, neurons were fixed with 4% paraformaldehyde in PBS for 15 min at room temperature and then permeabilized with 0.2% Triton X-100 in PBS for 10 min at room temperature. The neurons were blocked with 3% skim milk in PBS for 1 h at room temperature followed by the same solution containing mouse monoclonal anti-MAP2 antibody (Sigma) for 1 h at room temperature. After washing three times with PBS, the neurons were incubated with Alexa 568-labeled goat anti-mouse IgG (Molecular Probes) in the blocking solution for 1 h at room temperature. The neurons were then washed five times with PBS and counterstained with 1 g/ml Hoechst 33342.
RNA Interference-Short interfering RNA for depletion of CRM1 (siCRM1) was prepared as described previously (see also Ref. 17). Sense, UGUGGUGAAUUGCUUAUACd(TT), and antisense, GUAUA-AGCAAUUCACCACAd(TT), were obtained from Takara. Transfection of HeLa cells with siCRM1 or mock (transfection reagents only) was performed in 60-mm dishes using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. The cells were harvested at 48 h after transfection and lysed. Then Western blot analysis was performed using a rabbit polyclonal anti-CRM1 antibody (Santa Cruz Biotechnology) or a mouse monoclonal anti-GAPDH antibody (Ambion).
Fluorescence Microscopy-All living and fixed cells were examined using a Zeiss Axiovert 200 fluorescent microscope (Carl Zeiss).

Stau2 Shuttles between the Nucleus and
Cytoplasm-To better understand the dynamic behavior of Stau2 in cells, we initially investigated the subcellular localization of full-length Stau2 62 by transiently expressing a GFP fusion construct. As shown in Fig. 1, the full-length Stau2 62 was located primarily in the cytoplasm. However, when a variety of Stau2 62 deletion mutant constructs, fused to GFP, were transiently expressed in HeLa cells, we found that some mutants such as Stau2 62 -(243-317) and Stau2 62 -(273-373) were clearly localized in the nucleus (Fig. 1), indicating the possibility that Stau2 62 has the capability of entering the nucleus. Furthermore, the nuclear accumulation of both the Stau2 62 -(243-317) and Stau2 62 -(273-373) mutants suggests that the overlapping domain between RBD3 and RBD4 may mediate nuclear localization. In addition, the localization of Stau2 62 -(318 -571), Stau2 62 -(318 -412), and Stau2 62 -(413-571) suggests that the domain just following RBD4 may also be involved in nuclear localization.
To exclude the possibility that these mutants might have passively diffused into the nucleus followed by the retention, they were fused to MBP (maltose-binding protein)-MBP-GFP (2ϫMBP-GFP). As shown in Fig. 2, 2ϫMBP-GFP-Stau2 62 -(243-317) was localized predominantly in the nucleus, indicating that Stau2 62 -(243-317) contains a nuclear localization signal. In contrast, 2ϫMBP-GFP-Stau2 62 -(318 -412) remained in the cytoplasm. These results indicate that the domain between RBD3 and RBD4 mediates the nuclear import of Stau2 but not the downstream domain of the RBD4.
On the other hand, the mutant Stau2 62 -(208 -272), which is equivalent to the RBD3, was predominantly localized in the cytoplasm, indicating that the RBD3 domain mediates the nuclear export of Stau2 62 . These results indicate that Stau2 62 is capable of shuttling between the nucleus and cytoplasm.
Mutation of RBD3 Prevents the Nuclear Export of Stau2 62 -To confirm that the RBD3 actually mediates the nuclear export of Stau2 62 , we introduced a specific point mutation into the RBD3 that results in a reduction in its RNA binding activity. Based on previous findings on the structure of the doublestranded RNA-binding domain (14,15), an RBD3 mutant of Stau2 62 (Stau2 62 RBD3*) was created by replacing phenylalanine 239 with alanine, which has been reported to disrupt the precise folding of the domain (Fig. 3A). HeLa cells were transfected with GFP-Stau2 62 RBD3*, and its subcellular localization was examined 20 -24 h after transfection. As shown in Fig.  3B, the full-length Stau2 62 was localized predominantly in the cytoplasm. In contrast, Stau2 62 RBD3* accumulated in the nucleus, indicating that Stau2 62 is exported from the nucleus via the RNA binding activity of the RBD3.
To address whether this export is dependent on CRM1 (exportin 1), we treated HeLa cells transiently expressing GFP-Stau2 62 with leptomycin B (LMB), a specific inhibitor of the CRM1-dependent protein export (16). No nuclear accumulation of Stau2 62 was observed (Fig. 3C), indicating that CRM1 is not involved in the RBD3-mediated nuclear export of Stau2 62 .
Stau2 59 Is Exported from the Nucleus by Two Distinct Pathways-To determine whether Stau2 59 , the shorter isoform of Stau2, behaves like Stau2 62 , we examined its subcellular localization by transiently expressing the GFP-fused Stau2 59 in HeLa cells. As shown in Fig. 4A, like the full-length Stau2 62 , Stau2 59 was also localized predominantly in the cytoplasm. Therefore, to test whether the nuclear export of Stau2 59 is mediated by RBD3, we created the RBD3-mutated Stau2 59 (Stau2 59 RBD3*). Since the secondary structures of Stau2 62 and Stau2 59 RBD3 are identical, the mutation was introduced in the same way as that for Stau2 62 . HeLa cells were transfected with GFP-Stau2 59 RBD3*, and its localization was observed (Fig. 4A). Unexpectedly, in contrast to Stau2 62 RBD3*, Stau2 59 RBD3* showed no nuclear accumulation.
From these findings, we speculate that Stau2 59 is exported via an alternate export pathway, for example, a CRM1-dependent one. Therefore, HeLa cells transfected with GFP-Stau2 59 RBD3* were treated with LMB. As shown in Fig. 4B, LMB treatment resulted in the nuclear accumulation of Stau2 59 RBD3*. In contrast, wild type full-length Stau2 59 was localized predominantly in the cytoplasm, indicating that the RBD3-mediated nuclear export pathway, which is not sensitive to LMB, is also functional for Stau2 59 . Thus, Stau2 59 appears to be exported from the nucleus by two distinct pathways; one is mediated by the RBD3, which is CRM1-independent, and the other is CRM1-dependent. To confirm this, we performed RNA interference targeting CRM1. A short interfering RNA duplex targeted to residues 90 -108 of the human CRM1 open reading frame (siCRM1) was prepared according to the previous report (17) and transfected to HeLa cells. At 48 h after transfection, the cells were harvested, and the expression level of endogenous CRM1 was analyzed by Western blotting. As shown in Fig. 4C, this treatment specifically reduced the expression level of CRM1. To test the effect of the down-regulation of CRM1 on the cellular localization of Stau2 59 , HeLa cells were transfected with siCRM1 together with GFP-Stau2 59 or GFP-Stau2 59 RBD3*. As shown in Fig. 4D, whereas the down-regulation of CRM1 resulted in the predominant nuclear localization of GFP-Stau2 59 RBD3*, GFP-Stau2 59 was primarily localized in the cytoplasm, which was consistent with the results obtained by using LMB.
Identification of a CRM1-dependent Nuclear Export Signal of Stau2 59 at the N-terminal Domain-Since the primary structure of Stau2 59 is different from that of Stau2 62 only in the N-terminal domain (Fig. 5A), we hypothesized that the Nterminal domain of Stau2 59 mediates the CRM1-dependent nuclear export. For this, the deletion mutant Stau2 59 -(1-105) was constructed as a GFP fusion protein and transfected into HeLa cells. As shown in Fig. 5B, GFP-Stau2 59 -(1-105) was localized predominantly in the cytoplasm. In addition, LMB treatment promoted the nuclear localization of GFP-Stau2 59 -(1-105). On the other hand, the localization of GFP-Stau2 62 -(1-137) was similar to that of GFP alone and was not affected by LMB treatment. These results indicate that the N-terminal domain of Stau2 59 mediates the CRM1-dependent nuclear export.
It is well known that nuclear export signals (NESs) that are recognized by CRM1 are frequently composed of a stretch of characteristically spaced hydrophobic amino acids such as leucine and isoleucine, as was first reported for the human immunodeficiency virus type 1 Rev (18) and protein kinase inhibitor (19) proteins. To test whether the hydrophobic amino acid-rich sequence INQMFSVQLSL in the N-terminal of Stau2 59 functions as the NES that is recognized by CRM1, HeLa cells were transfected with GFP-INQMFSVQLSL, and its localization was examined (Fig. 6A). GFP-INQMFSVQLSL was localized predominantly in the cytoplasm, and LMB treatment led to its nuclear localization, indicating that the sequence INQMFSVQLSL is recognized by CRM1.
To confirm that INQMFSVQLSL functions as the CRM1-dependent NES of Stau2 59 , two point mutants of Stau2 59 RBD3* FIG. 4. Stau2 59 is exported by two distinct pathways, a RBD3mediated and a CRM1-dependent one. A, HeLa cells were transfected with GFP-Stau2 59 or GFP-Stau2 59 RBD3*. GFP fusion proteins were visualized by fluorescence microscopy. Scale bar, 10 m. B, HeLa cells were transfected with the indicated constructs, and the cells were then treated with 10 nM LMB for 4 h. GFP fusion proteins were visualized by fluorescence microscopy. Scale bar, 10 m. C, HeLa cells were mock-treated (mock) or transfected with short interfering RNA for depletion of CRM1 (indicated by siCRM1). At 48 h after transfection, the cells were harvested, and Western blot analysis was performed using an anti-CRM1 antibody or an anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH) antibody. D, HeLa cells were transfected with GFP-Stau2 59 or GFP-Stau2 59 RBD3* together with (siCRM1) or without (mock) siCRM1. At 48 h after transfection, GFP fusion proteins were visualized by fluorescence microscopy. Scale bar, 10 m. were first constructed by replacing isoleucine 4 and leucine 12 with alanine, respectively, because Stau2 59 RBD3* is convenient for assessing the CRM1-dependent pathway. HeLa cells were transfected with GFP-Stau2 59 RBD3*/I4A, GFP-Stau2 59 RBD3*/L12A, or GFP-Stau2 59 RBD3*/L2A as a control (Fig. 6B). Mutations I4A and L12A abolished the CRM1-dependent nuclear export activity of GFP-Stau2 59 RBD3*, although L2A mutation did not.
To test whether these point mutations influence the localization of Stau2 59 , which is also exported by the RBD3-dependent export pathway, GFP-Stau2 59 I4A and GFP-Stau2 59 L12A were transfected into HeLa cells (Fig. 6C). As in the case in which HeLa cells transiently expressing GFP-Stau2 59 were treated with LMB, GFP-Stau2 59 I4A and GFP-Stau2 59 L12A were localized predominantly in the cytoplasm, which confirms that Stau2 59 is also exported via a CRM1-independent pathway.
Stau2 Export Mutants Show the Same Localization in Hippocampal Neurons as in HeLa Cells-It is known that Stau2 is mainly expressed in the brain and is involved in the dendritic transport of mRNA in neurons (7). To test whether the same nuclear export pathways for Stau2 exist in neurons as in HeLa cells, rat hippocampal neurons were transfected with GFPfused constructs of Stau2 export mutants (Fig. 7). Although GFP-Stau2 62 was localized predominantly in the somatoden-dritic domain, GFP-Stau2 62 RBD3* accumulated in the nucleus, indicating that Stau2 62 is, in fact, exported from the nucleus only in an RBD3-mediated manner in neurons.
Furthermore, Stau2 59 export mutants showed a subcellular localization in neurons that was similar to that in HeLa cells. That is, GFP-Stau2 59 RBD3* and GFP-Stau2 59 I4A were predominantly localized in the cytoplasm in a similar fashion as GFP-Stau2 59 , whereas GFP-Stau2 62 RBD3*/I4A accumulated in the nucleus. In addition, LMB treatment of neurons resulted in the nuclear accumulation of GFP-Stau2 59 RBD3* (data not shown). These results indicate that Stau2 59 is exported from the nucleus via two distinct pathways, an RBD3-mediated one and a CRM1-dependent one in neurons, as in HeLa cells. DISCUSSION Using Stau2 deletion mutant constructs, we report here that Stau2 is a nucleocytoplasmic shuttle protein. In addition, the findings show that the region between RBD3 and RBD4 functions as the nuclear localization signal of Stau2, consistent with the recent report by Macchi et al. (20), although the import factor that recognizes the nuclear localization signal remains to be determined.
We next analyzed the nuclear export of Stau2 in more detail. First, we showed that the nuclear export of Stau2 62 is mediated by RBD3. This is consistent with a recent report by Macchi et al. (20), who demonstrated that the RBD3 of Stau2 62 interacts with exportin 5 in an RNA-dependent manner in vitro. Moreover, they showed that the down-regulation of exportin 5 by RNA interference resulted in the nuclear accumulation of Stau2 62 , indicating that exportin 5 is the nuclear export factor of Stau2 62 . Furthermore, Brownawell and Macara (21) reported that the RBD of ILF3, Spnr, Staufen1, and protein kinase R bind to exportin 5, suggesting that exportin 5 participates in the nuclear export of dsRNA-binding proteins.
Interestingly, we report, for the first time, that a functional difference exists among the Stau2 isoforms. That is, Stau2 62 is exported from the nucleus only in a RBD3-mediated manner, whereas Stau2 59 is exported by two distinct pathways, a RBDmediated one and a CRM1-dependent one. In addition, we identified the NES sequence at the N-terminal end of Stau2 59 , indicating that alternative splicing creates a new functional domain for nuclear export. It is reasonable to speculate that another Stau2 isoform, Stau2 52 , is exported by the same two pathways because Stau2 52 contains the same RBD3 and the same NES sequence recognized by CRM1 as Stau2 59 .
The question arises as to the nature of the biological significance of the export of Stau2 59 from the nucleus in a CRM1-dependent manner. Exportin 5 preferentially recognizes RNAs containing a minihelix motif, a structure that comprises a double-stranded stem of over 14 nucleotides with a base-paired 5Ј end and a 3-8-nucleotide protruding 3Ј end (22,23). The RBD3 of Stau2 is a major dsRNA binding determinant (7). Therefore, when Stau2 62 is exported from the nucleus by exportin 5, Stau2 62 appears to bind to and, therefore, to carry restricted minihelix-RNAs from the nucleus. On the other hand, it is likely that Stau2 59 as well as Stau2 52 , which are exported by CRM1, are able to transport other sets of dsRNAs that are unable to bind to exportin 5 or bind to the RBD3 in the absence of exportin 5. Considering this, along with the data that the expression of Stau2 59 is more abundant than Stau2 62 in the brain of adult rodents (7,11), we propose the possibility that the CRM1-dependent nuclear export of Stau2 59 plays an important role in transporting various RNAs out of the nucleus. However, the possibility that Stau2 may be exported from the nucleus without binding any RNA cannot be excluded.