Exportin 5 is a RanGTP-dependent dsRNA-binding protein that mediates nuclear export of pre-miRNAs

Abstract

microRNAs (miRNAs) are widespread among eukaryotes, and studies in several systems have revealed that miRNAs can regulate expression of specific genes. Primary miRNA transcripts are initially processed to approximately 70-nucleotide (nt) stem-loop structures (pre-miRNAs), exported to the cytoplasm, further processed to yield approximately 22-nt dsRNAs, and finally incorporated into ribonucleoprotein particles, which are thought to be the active species. Here we study nuclear export of pre-miRNAs and show that the process is saturable and thus carrier-mediated. Export is sensitive to depletion of nuclear RanGTP and, according to this criterion, mediated by a RanGTP-dependent exportin. An unbiased affinity chromatography approach with immobilized pre-miRNAs identified exportin 5 as the pre-miRNA-specific export carrier. We have cloned exportin 5 from Xenopus and demonstrate that antibodies raised against the Xenopus receptor specifically block pre-miRNA export from nuclei of Xenopus oocytes. We further show that exportin 5 interacts with double-stranded RNA in a sequence-independent manner.

FIGURE 1.

Microinjected pre-miR-31 is actively exported from nuclei of Xenopus laevis oocytes. Here, 15 nL of a mixture of 10 nM pre-miR-31 and U6Δss injection control were coinjected into nuclei of stage IV–V oocytes. If indicated, RNA mixtures also included 2 μM RanGAP or 500 nM or 2 μM unlabeled pre-miR-31 (competitor). Oocytes were separated into (N) nuclear and (C) cytoplasmic fractions 30 min after injection; RNA was extracted and analyzed on a 10% denaturing PAGE gel. The dried gel was quantified by a phoshorimager, and export was expressed as the cytoplasmic:nuclear (C:N) ratio.

FIGURE 2.

Exportin 5 forms export complexes with pre-miRNAs as well as with tRNA. (A) To identify a potential nuclear export receptor for pre-miRNAs, we first removed competing, endogenous RNAs from a HeLa cell lysate (see Materials and Methods) and bound the resulting extract (Start) to immobilized tRNA, indicated pre-miRNAs, or double-stranded DNA. Where indicated, 2 μM RanQ69L (GTP-form) had been included. Analysis was by SDS-PAGE, followed by Coomassie staining or immunoblotting with indicated antibodies. The load of bound material corresponds to 20 times the input. Note that Exp5 bound to tRNA and to any of the pre-miRNAs specifically and in an RanGTP-dependent manner. In contrast, exportin-t formed export complexes with tRNA only. None of the exportins bound immobilized dsDNA, which served as a control for nonspecific ionic interactions. (B) Escherichia coli lysate from bacteria expressing human Exp5 was depleted of endogenous RNA and subjected to binding to either immobilized tRNA or various pre-miRNAs in the absence or presence of RanQ69L GTP (2 μM). Binding assays were performed and analyzed as described in A.

FIGURE 2.

Exportin 5 forms export complexes with pre-miRNAs as well as with tRNA. (A) To identify a potential nuclear export receptor for pre-miRNAs, we first removed competing, endogenous RNAs from a HeLa cell lysate (see Materials and Methods) and bound the resulting extract (Start) to immobilized tRNA, indicated pre-miRNAs, or double-stranded DNA. Where indicated, 2 μM RanQ69L (GTP-form) had been included. Analysis was by SDS-PAGE, followed by Coomassie staining or immunoblotting with indicated antibodies. The load of bound material corresponds to 20 times the input. Note that Exp5 bound to tRNA and to any of the pre-miRNAs specifically and in an RanGTP-dependent manner. In contrast, exportin-t formed export complexes with tRNA only. None of the exportins bound immobilized dsDNA, which served as a control for nonspecific ionic interactions. (B) Escherichia coli lysate from bacteria expressing human Exp5 was depleted of endogenous RNA and subjected to binding to either immobilized tRNA or various pre-miRNAs in the absence or presence of RanQ69L GTP (2 μM). Binding assays were performed and analyzed as described in A.

FIGURE 3.

tRNA and pre-miRNA appear to bind to overlapping sites on Exp5. Binding of Exp5 from HeLa extract to immobilized tRNA and pre-miRNA was performed in the presence of RanQ69L (GTP) as described in Figure 2A ▶. Where indicated, 0.4 μM or 4 μM nonimmobilized pre-miR-31 had been added. The soluble pre-miRNA competed not only binding of Exp5 to immobilized pre-miRNA, but also to tRNA, which points to overlapping binding sites for tRNA and pre-miRNA on Exp5. The interaction between Exp-t and tRNA was not competed by pre-miRNA.

FIGURE 4.

Exportin 5 binds double-stranded RNA in a sequence-independent manner. Immobilized RNA homopolymers (30 nt) were used as ssRNA or annealed with the complementary homopolymer (dsRNA) to retrieve interacting proteins from cytosolic HeLa extract depleted of endogenous RNA. HeLa extract was fractionated and binding assays were performed as described for Figure 2 ▶. Formation of export complexes was analyzed by SDS-PAGE, followed by Coomassie staining. Western blotting confirmed that Exp5 (but not Exp-t) could be retrieved with dsRNA independent of nucleotide composition and sequence, provided RanQ69L GTP had been present.

FIGURE 5.

Antibodies against exportin 5 block export of pre-miR-31 from nuclei of Xenopus oocytes. (A) Here, 15 nL of a mixture of radiolabeled 10 nM pre-miR-31 and U6Δss nuclear injection control were coinjected into nuclei of stage IV–V oocytes. Both RNA mixtures contained 1 μM unlabeled pre-miR-31, 1 Unit/μL Superasin (Ambion), and either 2 mg/mL anti-Xenopus-exportin-5 antibody (αExp5) or the same concentration of a control antibody (anti-Drosophila-exportin-5, which does not cross-react with the Xenopus protein). Oocytes were separated into (N) nuclear and (C) cytoplasmic fractions 30 min after injection; RNA was extracted and analyzed on a 10% denaturing PAGE gel. (B) In this panel, 15 nL of a mixture of radiolabeled 10 nM U6Δss injection control, tRNA^Phe, and 7 nM U1ΔSm were coinjected into nuclei of stage IV–V oocytes. Injection mixtures contained 1 μM unlabeled tRNA^Phe, 1 Unit/μL Superasin (Ambion), and either 2 mg/mL affinity-purified anti-Xenopus-exportin-5 or control antibodies as indicated. Oocytes were separated into nuclear and cytoplasmic fractions either immediately (0 min), 45, 90, or 180 min after injection and processed as in A. Each timepoint in both panels represents the average of five oocytes.