Transportin-SR, a Nuclear Import Receptor for SR Proteins

A fragment corresponding to human ASF/SF2 RS domain (amino acids 198–248) was PCR amplified and inserted into BamHI and XhoI sites in either pGEX-5X-1 (Pharmacia Biotech) or pMal-c2 (New England Biolabs). This fragment was also cloned between EcoRI and SalI sites of pLexA (Clontech, Inc.) for the construction of the yeast two-hybrid library screening bait plasmid. The plasmid encoding human SC35 RS domain (amino acids 90–222) was made by ligation of the BamHI-XhoI fragment of SC35 RS domain amplified by PCR into pGEX-5X-1. The full-length of TRN-SR was amplified by PCR with Pfu DNA polymerase (Stratagene) and inserted as an EcoRI-XhoI fragment into pET28A (Novagen). Proteins were overexpressed in the BL21(DE3) Escherichia coli strain and were purified by methods that the manufacturers recommend. Glutathione-S-transferase (GST)-SV-40 T NLS, GST-IBB, and GST-M9 proteins were purified as described (Pollard et al., 1996). His-tagged RanQ69L (GTP form) was purified as described previously (Siomi et al., 1997).

Nuclear import assays were performed as described (Pollard et al., 1996). Rabbit reticulocyte lysate (Promega) was used as a cytosol source and prepared as described previously (Adam et al., 1990). The transport substrates were added at a concentration of 50 μg/ml. For competition experiments with maltose-binding protein (MBP) fusion proteins, competitors (1 mg/ml) were added to the complete transport mix except transport substrate, incubated on ice for 15 min, and then combined with transport substrate and nuclei. For the import experiments with recombinant receptor protein, recombinant His-tagged TRN-SR and His-tagged RanGDP were added at concentrations of 120 and 40 μg/ml, respectively.

The HeLa MATCHMAKER LexA cDNA library, yeast strains, and cloning vectors were obtained from Clontech, Inc. All library screening and yeast manipulations were carried out as recommended by the manufacturer. Saccharomyces cerevisiae strain EGY48 was transformed simultaneously with pLexA-ASF/SF2 RS and the HeLa cell cDNA library. 2 × 10⁶ transformants were plated onto 20 150-mm plates of X-gal–synthetic medium lacking histidine, uracil, tryptophan, and leucine. 32 Leu⁺ growers that had shown blue color on those plates were isolated. Insert cDNAs were amplified by PCR on these yeast cells using the Advantage-HF™ PCR kit (Clontech, Inc.) and sequenced.

The PCR fragment from clone 1-1 was used as a hybridization probe to screen the λ phage HeLa cell cDNA library (Clontech, Inc.). Several clones were isolated, and the clone that had the longest insert was sequenced and thus determined as the full-length coding sequence of TRN-SR.

TRN-SR was produced by in vitro transcription-translation of His-TRN-SR, using a TNT kit (Promega) in rabbit reticulocyte lysate in the presence of [³⁵S]methionine (Amersham) according to the procedure that the manufacturer recommends. Purified recombinant GST and GST fusion proteins (5 μg each) were immobilized on 50 μl of glutathione-Sepharose (Pharmacia) in PBS for at least 1 h at 4°C. The resin was washed with 500 μl of binding buffer (50 mM Tris-HCl, 400 mM NaCl, 5 mM MgOAc, 2 μg/ml of leupeptin, 2 μg/ml of pepstatin, 1% aprotinin, and 0.05% [wt/vol] digitonin; Calbiochem). In vitro translated TRN-SR was added and incubated with these immobilized proteins for 1 h at 4°C. For the experiments to check the effect of exogenous Ran protein, His-tagged RanQ69L (GTP form) was added at a concentration of 2 μM. The resin was washed with 500 μl of binding buffer five times and the bound fraction was eluted by boiling in SDS-PAGE sample buffer. The bound fraction was then analyzed by SDS-PAGE and visualized by fluorography.

The binding experiments with recombinant proteins were done essentially as described above except 20 μg of His- and T7-tagged recombinant TRN-SR was used. Binders were analyzed by 12.5% SDS-PAGE, and detected by an anti-T7 monoclonal antibody (Novagen) and ECL system (Amersham).

Purified SR proteins were kindly provided by Dr. Akila Mayeda prepared from HeLa cells as described previously (Zahler et al., 1992). 10 μg of proteins was analyzed by SDS-PAGE and transferred to nitrocellulose membrane. Far Western blotting was performed as described previously (Siomi et al., 1997) by using either TRN-SR or TRN1 produced by a TNT kit (Promega) in rabbit reticulocyte lysate in the presence of [³⁵S]methionine (Amersham).

To characterize the import pathway for SR proteins, we carried out in vitro nuclear import assays in digitonin-permeabilized HeLa cells (Adam et al., 1990). As a substrate we used recombinant GST fused to amino acids 198–248 of ASF/SF2 which corresponds to the RS domain (GST-ASF/SF2 RS) of this protein (Ge et al., 1991; Krainer et al., 1991). Efficient nuclear import of GST-ASF/SF2 RS was observed in the presence of cytosol and an ATP-regenerating system (Fig. 1). As no import was detected without addition of cytosol (Fig. 1), this indicates that nuclear import of ASF/SF2 requires additional soluble factor(s). Efficient nuclear import of GST-ASF/SF2 RS was observed in the presence of an ATP-regenerating system and was reduced by incubation with apyrase (Fig. 1), suggesting a role for NTPs in this process. The import of GST-ASF/SF2 RS was strongly inhibited by RanQ69L, a Ran mutant that cannot hydrolyze GTP at a significant rate (Klebe et al., 1995) (Fig. 1), suggesting a role for RanGTP. GST-ASF/ SF2 RS import also has several characteristics of nuclear import that occur through NPCs. Both WGA and an importin β dominant-negative mutant (Impβ ΔN44), reagents which block active nuclear import through NPCs (Forbes, 1992; Görlich et al., 1996b; Kutay et al., 1997a), completely abolished GST-ASF/SF2 RS nuclear import (Fig. 1).

To determine whether a specific and saturable factor(s) participates in RS domain–mediated nuclear import, we tested the effect of excess RS domain on the import of classical NLS, M9, and RS domain–bearing proteins. For these experiments we prepared an MBP fusion of the ASF/SF2 RS domain, termed MBP-RS, as a competitor. Nuclear import assays were carried out in the presence of a 20-fold molar excess of either MBP or MBP-RS. MBP itself had no effect on nuclear import; however, MBP-RS strongly inhibited GST-ASF/SF2 RS import, whereas import of other substrates, GST-SV-40 T NLS, GST-IBB, and GST-M9, was unaffected or only slightly reduced (Fig. 2). These results suggest that a specific nuclear import receptor, distinct from importin β and TRN1, mediates RS domain nuclear import.

To identify candidate mediator(s) of SR protein nuclear import, we carried out a yeast two-hybrid screening on a HeLa cell cDNA library using the COOH-terminal 51– amino acid region of ASF/SF2 as bait. This fragment contains the RS domain and is sufficient for complete nuclear localization of myc-tagged pyruvate kinase in HeLa cells (data not shown). Several positive interacting clones were isolated and characterized. One of these, clone 1-1, was isolated four times out of 16 clones, and its deduced amino acid sequence showed significant similarity to that of a putative importin β/transportin-related nuclear transport receptors.

The 1-1 DNA insert was subcloned and used for hybridization screening of a λ phage HeLa cDNA library. A 3-kb clone that appears to contain the entire coding region was obtained. The predicted amino acid sequence of this protein, which we termed TRN-SR, because it turned out, like TRN1, to be a transport receptor of pre-mRNA/mRNA-binding proteins, is shown in Fig. 3. TRN-SR is a 975– amino acid protein with a calculated molecular mass of 109,838 D and an estimated pI of 5.29. The amino-terminal domain of TRN-SR shows significant sequence similarity to other importin β/transportin family members, including a region required for RanGTP binding (Görlich et al., 1997). The sequence of the original 1-1 clone isolated from the yeast two-hybrid screening starts at amino acid 590 of the TRN-SR sequence and contains the entire COOH-terminal domain. A BLAST homology search with full-length TRN-SR revealed three proteins that bear significant homology to TRN-SR in other species (Fig. 3). The most similar of these, AF025464 of Caenorhabditis elegans, is 26% identical and 45% similar to TRN-SR. Another apparent homologue is AL022304 of Schizosaccharomyces pombe that is 25% identical and 46% similar, although this clone does not appear to contain the full-length protein sequence. These two sequences are the two closest orthologues of TRN-SR present in available databases. Of previously characterized proteins, the most significant similarity is found with the S. cerevisiae protein Mtr10p (Kadowaki et al., 1994) which has been shown recently to be a nuclear import receptor for Npl3p (Pemberton et al., 1997; Senger et al., 1998). Npl3p is an hnRNP protein in yeast (Bossie et al., 1992; Russell and Tollervey, 1992; Wilson et al., 1994). The amino acid sequences of TRN-SR and Mtr10p are 21% identical and 42% similar.

To confirm that TRN-SR binds specifically to SR proteins, we carried out in vitro binding experiments using TRN-SR produced by transcription-translation in rabbit reticulocyte lysate. In the same experiments we also tested another RS domain, that of the SR splicing factor SC35 (amino acids 90–222; Fu and Maniatis, 1992). TRN-SR binds to the RS domains of both ASF/SF2 and SC35, but not to IBB or to hnRNP A1 M9 (Fig. 4 A). RanQ69L abolishes the binding of TRN-SR to RS domains (Fig. 4 A), consistent with the possibility that it is a nuclear import receptor for these proteins. Since rabbit reticulocyte lysate contains many proteins, the binding of TRN-SR detected in Fig. 4 A could be indirect. To examine whether TRN-SR can bind to the RS domains directly, we carried out binding assays using purified recombinant TRN-SR. As shown in Fig. 4 B, bacterially produced TRN-SR binds to both GST-ASF/SF2 RS and GST-SC35 RS directly, but not to GST alone. These results strongly suggest that TRN-SR is a specific import receptor for SR proteins.

To determine if TRN-SR is the nuclear import receptor of SR proteins, recombinant TRN-SR was used in in vitro nuclear import assays using either GST-ASF/SF2 RS or GST-SC35 RS as a substrate. Neither GST-ASF/SF2 RS nor GST-SC35 RS by itself accumulated in the nucleus (Fig. 5). However, in the presence of ATP, an ATP-regenerating system and RanGDP, TRN-SR efficiently imported GST-ASF/SF2 RS and GST-SC35 RS into the nucleus (Fig. 5). Thus, TRN-SR is a nuclear import receptor for ASF/SF2, SC35, and likely for other RS domain–containing proteins.

Mammalian cells contain several SR proteins in addition to ASF/SF2 (Zahler et al., 1992; Fu, 1995). Since TRN-SR binds the RS domains of both ASF/SF2 and SC35 (Fig. 4), we examined whether it can also bind other SR proteins. The SR protein fraction was purified from HeLa nuclear extracts (Zahler et al., 1992), resolved by SDS-PAGE, and immobilized on a nitrocellulose membrane. By Western blotting with the anti-RS domain antibody mAb104, these purified SR proteins show the typical pattern reported previously (Fig. 6; Zahler et al., 1992). The capacity of TRN-SR to bind these proteins was determined by far Western blotting using ³⁵S-labeled TRN-SR produced in rabbit reticulocyte lysate (Siomi et al., 1997). TRN-SR bound several of these proteins, whereas TRN1 did not (Fig. 6). In addition to proteins of ∼33 kD, that likely correspond to ASF/SF2 and SC35, proteins of ∼20, 46, and 55 kD also bound specifically to TRN-SR. This observed profile is similar to that detected by Western blotting with mAb104 (Fig. 6), suggesting that TRN-SR is a common nuclear import receptor for many of the SR proteins.

In this report we have identified a novel receptor, TRN-SR, as a nuclear import receptor for SR proteins. Of the known proteins currently present in the sequence databases, we note the considerable amino acid sequence homology of TRN-SR with the S. cerevisiae Mtr10p (Fig. 3). Mtr10p has been shown to be a nuclear import receptor for the yeast pre-mRNA/mRNA-binding protein Npl3 (Pemberton et al., 1997; Senger et al., 1998). Npl3p, which is also referred to as Nop3p and Nab1p, is an hnRNP protein that contains within its carboxyl terminus an RGG-box within which are several serine-arginine (SR) dipeptides (Bossie et al., 1992; Russell and Tollervey, 1992; Wilson et al., 1994; Siebel and Guthrie, 1996; Pemberton et al., 1997). The NLS of Npl3p has not been precisely delineated but is contained in this region of the protein (Senger et al., 1998). The SR dipeptides of Npl3p may be important for Mtr10p recognition, although this has not been determined. Two additional proteins, one in C. elegans and one in S. pombe, show similarity to TRN-SR (Fig. 3). Several candidate SR proteins are found in the C. elegans database, and one SR protein has been recently cloned from S. pombe (Gross et al., 1998). Therefore, these TRN-SR homologues may be the import receptors of SR proteins in these organisms.

TRN-SR binds to the RS domain of ASF/SF2 and of SC35, and these interactions are disrupted by RanQ69L (Fig. 4 A). Furthermore, TRN-SR also binds other proteins enriched in an SR protein fraction (Fig. 6). These results strongly suggest that TRN-SR is a general nuclear import receptor for SR proteins. However, we note that no binding of TRN-SR to SRp75 was detected by far Western blotting, although this protein is abundant in the fraction we tested (Fig. 6). The reason for this is unknown, but it is possible that SRp75 may have a different receptor. There are additional SR proteins, including pre-mRNA splicing factors such as 9G8, U170K, U2AF35, and 65 (Fu, 1995), as well as two large SR proteins (Blencowe et al., 1998), and it remains to be determined whether TRN-SR also mediates the nuclear import of these proteins.

Several abundant hnRNP proteins, including hnRNP A1, A2, and F, are imported by TRN1 (Pollard et al., 1996; Siomi et al., 1997). Thus, in mammalian cells there are at least two nuclear import pathways for pre-mRNA/ mRNA-binding proteins, one mediated by TRN1 and one by TRN-SR. The relative amounts of hnRNP proteins and SR proteins are important for alternative pre-mRNA splicing. For example, the ratio between hnRNP A1 and ASF/SF2 affects 5′ splice site selection (Mayeda and Krainer, 1992; Zahler et al., 1993; Cáceres et al., 1994; Yang et al., 1994). As both of these proteins shuttle between the nucleus and the cytoplasm (Piñol-Roma and Dreyfuss, 1992; Alzhanova-Ericsson et al., 1996; Cáceres et al., 1998), it is conceivable that their relative amounts in the nucleus may be controlled by regulating their rates of nuclear import. Thus, by modifying either the transportins themselves or the respective NLSs, M9 and RS, splice site selection could be modulated. Indeed, several protein kinases have been reported to phosphorylate serine residues in the RS domains of SR proteins (Gui et al., 1994a,b; Colwill et al., 1996; Rossi et al., 1996; Kuroyanagi et al., 1998; Okamoto et al., 1998; Wang et al., 1998). While overexpression of some of these SR protein kinases causes disruption of nuclear speckles (Gui et al., 1994a; Kuroyanagi et al., 1998; Wang et al., 1998), they do not disrupt the nuclear localization of SR proteins. However, overexpression of one SR protein kinase, Clk/Sty kinase, does cause cytoplasmic accumulation of ASF/SF2 in HeLa cells (Cáceres et al., 1998). More recently it was reported that overexpression of kinase-inactive mutant of SR protein kinase-2 causes cytoplasmic accumulation of ASF/SF2 (Koizumi et al., 1999). It will be interesting to determine the effect of RS domain phosphorylation on the SR proteins–TRN-SR interaction.

The physiological function of the shuttling of SR proteins is not known. Both hnRNP A1/A2 proteins and SR proteins are associated with the same mRNAs as they are exported to the cytoplasm (Alzhanova-Ericsson, 1996; Visa et al., 1996) and it is thus possible that they both play a role in mRNA export. Nuclear export signals in the shuttling SR proteins have not been identified yet. The identification of nuclear export signals in shuttling SR proteins, if such exist, and of export receptors for them are issues of considerable interest that remain to be clarified.

We thank Dr. James Manley for the ASF/SF2 cDNA; Dr. Xian-dong Fu for SC35 cDNA; Dr. Dirk Görlich for importin β ΔN44 protein and RanQ69L expression plasmids; and Dr. Akila Mayeda for SR proteins. We also thank Lili Wan for help on yeast two-hybrid library screening, and other members of our laboratory, especially Drs. Sara Nakielny, Haruhiko Siomi, Lili Wan, and Robert Perkinson for critical reading and comments on the manuscript.

This work was supported by a grant from the National Institutes of Health (G. Dreyfuss), and by a long-term fellowship from Human Frontier Science Program Organization (N. Kataoka). G. Dreyfuss is an Investigator of the Howard Hughes Medical Institute.

GST

glutathione-S-transferase

hnRNP

heterogeneous nuclear RNP

IBB

importin β–binding domain

MBP

maltose-binding protein

NLS

nuclear localization signal

NPC

nuclear pore complex

RS domain

arginine/serine rich domain

TRN1

transportin 1

TRN-SR

transportin-SR