SMN, the Spinal Muscular Atrophy Protein, Forms a Pre-Import Snrnp Complex with Snurportin1 and Importin β

Snurportin1 is primarily a cytoplasmic protein

Snurportin1 is a protein of approximately 45 kDa that is required for nuclear snRNP import in Xenopus oocytes (24). In order to better characterize the complexes that carry out snRNP import in mammals, we developed polyclonal anti-peptide antibodies against human SPN and cloned the cDNA by RT-PCR using the published sequence (24). Western blot analysis with anti-SPN serum R89 identified a single band of the expected size that was specifically competed upon preincubation with the antigenic peptide (Fig. 1B). Immunofluorescence with this antiserum revealed that the protein is localized throughout the cytoplasm, concentrating near the nuclear periphery (Fig. 1C). This localization pattern is completely consistent with snurportin’s role as a nuclear import adaptor. Equivalent results were obtained upon expression of GFP-SPN (Fig. 1C). Furthermore, incubation of the antibody with excess peptide antigen completely blocked the immunofluorescence signal (Fig. 1C). Western analysis of extracts from cells transfected with GFP-SPN revealed the presence of an additional band of the appropriate size (data not shown). Thus SPN antiserum R89 specifically recognizes both the exogenous and endogenous snurportins.

A snurportin mutant is actively transported to the nucleus

Huber et al. (24) showed that the N-terminal 65 amino acids of snurportin1 contain a domain (IBB, Fig. 1A) required for binding to importin β. Overexpression of SPNΔN65 in frog oocytes abolished import of snRNPs into the nucleus (24). Thus we were quite surprised to find that GFP-SPNΔN65 was primarily nuclear when expressed in human cells (Fig. 2). The molecular mass cutoff for proteins passively diffusing through the NPC is reportedly ∼40-60 kDa (2,40) and depends to some extent on the shape of the macromolecule. That is, proteins larger than this require assistance from importins and exportins. GFP-SPNΔN65 contains an intact TMG-binding domain but cannot bind to importin β (Fig. 1A). Therefore, it was formally possible that the construct could simply diffuse into the nucleus and bind to snRNPs. In the case of wild-type SPN, cargo release is effected upon binding to the export factor CRM1/Xpo1 (25). However, binding to CRM1/Xpo1 and recycling to the cytoplasm is disrupted in SPNΔN65, as export depends on multiple sites throughout the length of the protein (25). Only the C-terminal 75 amino acids are dispensable for this activity (Fig. 1A). Since it was conceivable that GFP-SPNΔN65 (calc. MW ∼65 kDa) was small enough to pass through the nuclear pores unassisted, we cloned a second GFP domain in-frame with GFP-SPN and GFP-SPNΔN65 to ensure that these constructs exceeded the limit for passive diffusion. As shown in Figure 2A, the localization patterns of the 2 ×GFP constructs were identical to those of their 1 × GFP counterparts. These results indicate that GFP-SPNΔN65 is actively imported into the nucleus by an alternative pathway that does not involve direct binding to importin β (11,23,24,37).

Overexpression of SPN and SPNΔN65 disassembles Cajal bodies

In the course of our localization experiments, we noticed that GFP-SPNΔN65 occasionally displayed nuclear foci (Fig. 2A). Immunofluorescence with CB marker proteins coilin and SMN revealed that these focal accumulations did not strictly correspond to CBs, although they sometimes colocalized (Fig. 2B and data not shown). Strikingly, we found that cells transfected with GFP-SPNΔN65 typically displayed fewer CBs than did neighboring untransfected cells (Fig. 2B). In fact, careful analysis revealed that even cells transfected with wild-type GFP-SPN showed a reduction in CB number (Table 1). Whereas moderate levels of GFP-SPNΔN65 expression typically resulted in CB disassembly, only high levels of GFP-SPN had this effect (Fig. 2B). Similar results were obtained in parallel experiments when SMN was used as the CB marker (Table 1). This effect was specific to SPN, as cells transfected with GFP alone showed no difference in CB numbers when compared to untransfected cells (Table 1).

As mentioned above, the typical SPN localization pattern was diffuse throughout the cytoplasm, with a relatively pronounced staining of the nuclear periphery. Thus the majority of cells did not show any obvious colocalization with the cytoplasmic fraction of SMN (data not shown). However, we sometimes observed local accumulations of both GFP-SPN and SMN in the cytoplasm (Fig. 2C), suggesting a possible interaction between the snRNP assembly and import machineries.

Snurportin1 forms a cytoplasmic complex with SMN and splicing snRNPs

The SMN complex is required for assembly of the Sm ring which, in turn, is required for cap hypermethylation (reviewed in 15). Precisely when SMN leaves the snRNP complex following Sm core assembly has not been established. In order to determine whether SMN and SPN are capable of binding to the same RNPs, we performed GST-pulldown assays with HeLa cytoplasmic extracts. As shown in Figure 3A, SMN, Gemin3 and SmB/B’ proteins were recovered by GST-SPN bound to glutathione-sepharose beads. Control experiments with GST alone showed only residual binding. Thus the SMN complex remains bound to TMG-positive Sm snRNPs. In order to prove that SPN forms a native complex with SMN, cells were transfected with GFP-tagged SPN constructs and analyzed by co-immunoprecipitation with anti-GFP antibodies. As shown in Figure 3B, SMN was coprecipitated by both GFP-SPN and GFP-SPNΔN65. Parallel northern blotting experiments (Figs 3C and D) showed that U2 snRNA was also coprecipitated with the SPN constructs. Both wild-type SPN and the IBB mutant protein have intact TMG-binding domains and, as expected, each bound equivalent amounts of U2 snRNA in the GST-pulldown experiment (Fig. 3C). However, quantification of the blot in Figure 3D showed that GFP-SPNΔN65 brought down roughly five times more U2 RNA than did GFP-SPN and 10 times more than GFP alone. GFP-SPN brought down twice as much RNA as did the GFP control. Given the nuclear localization of both GFP-SPNΔN65 and the bulk of the Sm snRNPs, this result is not surprising. Additionally, SPN binding to CRM1/Xpo1 is required for cargo release and cytoplasmic recycling. Thus the SPNΔN65 construct, which cannot bind to CRM1, likely binds ‘irreversibly’ to snRNPs, whereas the wild-type protein does not.

To test whether SMN and SPN might interact directly, we purified recombinant proteins expressed in bacteria and performed GST-pulldowns. No direct binding was detected (Fig. 3E). Therefore, the interaction between SPN and SMN is likely mediated by the snRNA. To test this hypothesis, we performed GST-pulldown assays from cell extracts incubated in the presence or absence of RNases A and T1. Figure 3F reveals that SMN was efficiently recovered in the absence of added RNase and not detectable in its presence. The blots were reprobed with anti-GST to demonstrate that roughly equivalent amounts of protein were used. As an independent control for monitoring RNase activity, extracts were incubated in the presence or absence of RNase and then assayed for binding of U2B″, a U2-specific snRNP protein, to U2 RNA (data not shown).

In order to determine if SMN and SPN complexes were restricted to a particular subcellular compartment, nuclear and cytoplasmic HeLa cell extracts were prepared and GST-pulldown assays performed with GST-SPN and GST alone. Cell fractionation was monitored using control antibodies (data not shown) against proteins that localize to the cytoplasm [glutaredoxin, (41)] and nucleus [coilin, (42)]. As shown in Figure 3G, SMN was only recovered from the cytoplasmic fractions. Taken together, these data establish that SPN and SMN co-exist in a cytoplasmic complex, bridged by snRNAs. Thus we conclude that SMN remains bound to Sm snRNPs after cap hypermethylation, forming a putative pre-import complex.

SMN and snurportin form a complex that contains importin β

The interaction of Sm snRNPs, SPN and importin β has been well established (24). Our finding that SMN and SPN form a complex together prompted the question of whether SMN is present in an import-competent snRNP complex with importin β. As a first step, we transfected cells with myc-SMN, immunoprecipitated with anti-myc antibodies, and probed a western blot with anti-importin β. Interestingly, Figure 4A shows that SMN exists in a complex with importin β in vivo. In order to show that SPN, SMN and importin β are together in the same complexes, gel filtration assays were performed using HeLa cytoplasmic extract, followed by western blot analysis. SMN runs in two peaks, the first a broad one between 600 and 1500 kDa and the other between 200 and 500 kDa (Fig. 4C). Notably, importin β, SMN and SPN comigrate in the 400-500 kDa region, suggesting that these proteins form a single complex.

As described above, several lines of evidence suggest the existence of an Sm core-directed, TMG cap-independent import mechanism that also impinges upon the importin β pathway (11 and references therein). In light of our findings, the simplest explanation for these data is that the SMN complex is the factor that recognizes both the Sm core and importin β, thus acting as the adaptor for this cap-independent pathway. We decided to test this idea further by asking whether SMN and importin β could bind to each other directly using recombinant proteins. As shown in Figure 4B, GST-SMN, but not GST alone, clearly interacted with His-tagged importin β.

Snurportin, SMN and the zinc finger protein ZPR1 form a complex

Given that SPN, importin β and SMN form what appears to be an import-competent complex, this raised the question as to whether the SMN complex is imported into the nucleus along with the newly assembled snRNPs. A recent study found that the zinc finger protein ZPR1 is required for the accumulation of SMN in the nucleus (39). ZPR1 is an essential protein in yeast (44). In mammals, the protein is involved in relaying proliferative growth signals from the cell surface to the nucleus (45,46). Although ZPR1 is localized to the cytoplasm in quiescent cells, the protein is translocated to the nucleus upon mitogen stimulation (44-46). Truncation of the C-terminus of ZPR1 abolished the interaction with SMN and was dominant negative for SMN localization to nuclear bodies (39). Therefore, we wanted to determine whether ZPR1 might also be complexed with SPN. Toward that end, we performed GST-pulldowns and co-immunoprecipitations. HeLa cell lysate was passed over GST and GST-SPN beads and then probed with anti-ZPR1 antibodies (Fig. 5A), revealing an interaction between the two proteins. Similarly, lysates from cells transfected with GFP-SPN were immunoprecipitated with anti-GFP antibodies and probed for the presence of ZPR1. As shown in Figure 5B, GFP-SPN but not GFP alone brought down ZPR1 protein. The interaction between SPN and ZPR1 was sensitive to RNase digestion (Fig. 5C) and was primarily restricted to the cytoplasm (Fig. 5D). Furthermore, gel fractionation studies showed that ZPR1 also comigrated with SPN, SMN and importin β (data not shown), suggesting that these factors are present in the same complexes. Thus the fact that ZPR1, which is involved in regulating the accumulation of SMN in the nucleus (39), is also complexed with SPN and importin β suggests that SMN could indeed accompany newly assembled snRNPs across the NPC and perhaps help to target them to CBs (47,48).

DNA constructs

Snurportin1 (SPN) and SPNΔN65 were amplified from a PCR-ready human brain cDNA library (Clontech Laboratories, Palo Alto, CA, USA) and cloned into pCR 2.1 (Invitrogen, Carlsbad, CA, USA). Primers containing EcoRI and BamHI restriction sites were used to PCR amplify SPN from pCR 2.1. SPN was subsequently cloned into EcoRI/BamHI-cut pEGFP-C3 (Clontech Laboratories). 2 × EGFP was generated by PCR amplification of EGFP from pEGFP-C3 using primers containing BglII and HindIII restriction sites followed by cloning into BglII/HindIII-digested pEGFP-C3. SPN and SPNΔN65 were then excised from pEGFP-C3-SPN and pEGFP-C3-SPNΔN65, respectively, with EcoRI and BamHI and cloned into EcoRI/BamHI-digested 2 × EGFP-C3, resulting in 2 × EGFP-SPN and 2 × EGFP-SPN1-DN65. Glutathione S-transferase (GST) fusions of SPN and SPNΔN65 were generated by PCR amplifying SPN from 2 × EGFP-C3 SPN with EcoRI forward and BamHI reverse primers. This fragment was subsequently cloned into a BamHI/EcoRI-digested pGex 2T vector (Amersham Pharmacia, Piscataway, NJ, USA).

Antibodies

A polyclonal SPN antibody was generated (Sigma Genosys, The Woodlands, TX, USA) using a synthesized peptide consisting of CGMESEEENKKDDEE residues corresponding to amino acids 73-85 of human SPN. Other antibodies utilized in this study include: mouse monoclonal anti-SMN (7B10; 52); mouse monoclonal anti-ZPR (LG1; 39); mouse monoclonal anti-Gemin 3 (11G9; 43); mouse monoclonal anti-Sm (Y12; 64); rabbit polyclonal anti-importin beta (gift of K. Weis); mouse monoclonal anti-GST (Santa Cruz, CA, USA); rabbit polyclonal and mouse monoclonal anti-GFP (Roche, Indiana-polis, USA); mouse monoclonal anti-myc (9E10; Santa Cruz); mouse monoclonal anti-T7 (Novagen, Madison, WI, USA); rabbit polyclonal anti-coilin (R288; 65); mouse monoclonal anti-U2B″ (gift of H. Salz); and rabbit polyclonal antiglutaredoxin (gift of J. Mieyal). Secondary antibodies utilized were goat anti-mouse and goat anti-rabbit conjugated horseradish peroxidase (Pierce Chemical, Rockford, IL, USA) as well as goat anti-rabbit conjugated fluorescein (Vector Laboratories, Burlingame, CA, USA).

Co-immunoprecipitation

HeLa cells were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) (Gibco BRL, Gaithersburg, MD, USA) with 10% fetal bovine serum (FBS) (Gibco BRL), penicillin and streptomycin. pEGFP-SPN or pEGFP-SPNΔN65 was transfected into HeLa cells using SuperFect (QIAGEN, Valencia, CA, USA). As a control, untransfected HeLa cells were also utilized. Cells were harvested 24 h after transfection, washed with phosphate-buffered saline (PBS) and resuspended in 1 μl of mRIPA buffer (50 mm Tris-Cl, pH 7.5, 150 mm NaCl, 1% NP-40, 1 mm EDTA) plus protease inhibitor cocktail tablets (Roche) to lyse cells. Resuspended cells were nutated at 4 °C for 30 min, followed by centrifugation for 5 min to pellet cellular debris. Where indicated, 5 mg of total protein from the lysate was treated with RNase A (1 mg, Sigma, St. Louis, MO, USA) and RNase T1 (5000 units, gift of H. Salz) for 1 h at 30 °C. Fifty microliters of polyclonal anti-GFP (Clontech Laboratories) were then added to 900 μl of lysate. After nutating for 1 h, 60 μl of 50% protein A Sepharose beads (Amersham Pharmacia) were added to the lysates, and incubated overnight. Beads were then washed six times with 1 μl of mRIPA, resuspended in 15 μl of 5×sodium dodecyl sulfate (SDS) loading buffer, and boiled, and proteins were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). After transfer to nitrocellulose, membranes were probed with anti-SMN (7B10) followed by incubation with goat anti-mouse conjugated horseradish peroxidase (Pierce Chemical) and chemiluminescence detection (Roche).

Pulldowns

Expression and isolation of recombinant proteins was performed using standard techniques (66). GST pulldowns were conducted by incubating HeLa cell lysate, prepared as described (66), with glutathione bead captured GST tagged proteins for 1 h in a buffer containing 450 mm Na⁺. Beads were then washed six times with 1 μl of mRIPA, resuspended in 15 μl of 5 × SDS loading buffer, and boiled, and proteins were resolved by SDS-PAGE. Following transfer to nitrocellulose, membranes were probed with appropriate primary and secondary antibodies.

Immunofluorescence

HeLa cells were grown on slides (Nunc, Rochester, NY, USA) to 80% confluency. Slides were fixed in 4% paraformaldehyde for 10 min at room temperature, followed by extraction with 0.5% Triton for 5 min. Incubation in 10% normal goat serum was followed by the addition of anti-SPN1 rabbit polyclonal antibody, R89 and detection with fluorescein-conjugated goat anti-rabbit (Molecular Probes, Eugene, OR, USA). Incubations at 30 °C for 30 min were followed with three washes for 5 min each at room temperature with 1 × PBS.

Protein fractionation

The S100 cytoplasmic fraction was generated from 500 μl of packed HeLa cells (National Cell Culture Center) using the NE-PER extraction system (Pierce Chemical). Five hundred microliters of S100 containing 10.5 mg of total protein were passed over a Superose 6HR 10/30 column and fractionated (300 μl fractions) using fast performance liquid chromatography (FPLC). Twenty-five microliters of each fraction were added to 5 μl of 5 SDS-loading buffer, boiled and subjected to SDS-PAGE. Thirty microliters of S100 fraction were loaded as a positive control. Following transfer, membranes were probed with the appropriate primary and secondary antibodies.