Gemin4: A Novel Component of the Smn Complex That Is Found in Both Gems and Nucleoli

Spinal muscular atrophy (SMA) is a common autosomal recessive disease that is the leading hereditary cause of infant mortality. SMA is characterized by degeneration of motor neurons of the anterior horn of the spinal cord resulting in muscular weakness and atrophy (Pearn 1980; for review see Melki 1999). SMA results from deletions or mutations in the survival of motor neurons gene (SMN), which is duplicated as an inverted repeat on human chromosome 5 at 5q13 (Brzustowicz et al. 1990; Melki et al. 1990, Melki et al. 1994). The telomeric copy of the SMN gene (SMN1) is deleted or mutated in >98% of SMA patients (Lefebvre et al. 1995; for review see Burghes 1997). The SMN protein is expressed in all tissues of mammalian organisms with particularly high levels expressed in motor neurons (Coovert et al. 1997; Lefebvre et al. 1997). In contrast, individuals affected by the most severe form of SMA, Werdnig-Hoffman syndrome or SMA type I, have barely detectable levels of SMN in motor neurons (Coovert et al. 1997; Lefebvre et al. 1997).

The SMN protein is part of a multiprotein complex and two other proteins of the complex, Gemin2 (formerly SIP1) and Gemin3 (for component of gems 2 and 3, respectively) thus far have been characterized (Liu et al. 1997; Charroux et al. 1999). SMN, Gemin2, and Gemin3 localize in the cytoplasm and the nucleus of somatic cells. In the nucleus, they are concentrated in bodies called gems, which are similar in size and number to coiled bodies (CBs) and are often associated with them (Liu and Dreyfuss 1996; Liu et al. 1997; Charroux et al. 1999). In addition to SMN, Gemin2 and Gemin3, the large cytoplasmic complex of which they are part also contains Sm proteins that are components of spliceosomal small nuclear ribonucleoprotein (snRNPs; Liu et al. 1997; Charroux et al. 1999). SMN interacts directly with the Sm proteins and with Gemin2 and Gemin3 (Liu et al. 1997; Charroux et al. 1999; Pellizzoni et al. 1999). Antibody microinjection experiments in Xenopus oocytes revealed that Gemin2 has a critical role in the assembly of snRNPs (Fischer et al. 1997), a process which takes place in the cytoplasm where the Sm proteins combine with snRNAs that were exported from the nucleus (Mattaj and De Robertis 1985; Mattaj 1988; Luhrmann et al. 1990). Once properly assembled and modified, the snRNPs recruit the necessary nuclear import receptors and translocate into the nucleus where they function in pre-mRNA splicing (Mattaj 1986, Mattaj 1988; Luhrmann et al. 1990; Neuman de Vegvar and Dahlberg 1990; Zieve and Sauterer 1990). Transfections of a dominant negative form of SMN (SMNΔN27) revealed that SMN also plays a critical role in the cytoplasmic assembly of snRNPs (Pellizzoni et al. 1998). In the nucleus, overexpression of the SMNΔN27 protein causes a striking rearrangement of the snRNPs, which accumulate with the mutant SMNΔN27 in enlarged gem/coiled body structures (Pellizzoni et al. 1998). SMN has been further shown to be required for pre-mRNA splicing, likely for the regeneration or recycling of snRNPs (Pellizzoni et al. 1998). Of the known components of the SMN complex, the recently described DEAD box protein Gemin3 is the most likely candidate to have the capacity to perform such functions (Charroux et al. 1999). Interestingly, SMN mutants found in SMA patients lack the snRNP regeneration activity likely because of their defective interaction with the Sm proteins as well as with Gemin3 (Pellizzoni et al. 1998, Pellizzoni et al. 1999; Charroux et al. 1999).

Here, we report the amino acid sequencing by nanoelectrospray mass spectrometry (Wilm et al. 1996) and molecular cloning of a novel component of the SMN complex designated Gemin4 (available from GenBank/EMBL/DDBJ under accession number AF173856). Several lines of evidence suggest that Gemin4 participates in the functions of the SMN complex. Together with SMN, Gemin2 and Gemin3, Gemin4 can be isolated in a complex with the spliceosomal snRNP proteins. The presence of Gemin4 in the SMN complex in vivo is a result of its direct interaction with Gemin3 but not with SMN. Gemin4 also interacts directly with several of the spliceosomal snRNP core Sm proteins, including SmB, SmD1-3, and SmE, and is associated with U snRNAs in the cytoplasm of Xenopus oocytes. Gemin4 is a novel protein and shows no significant homology to any other protein found in the databases. Its tight association with Gemin3 suggests that it may act as a cofactor of the putative ATPase and/or helicase activity of Gemin3. We have produced mAbs to Gemin4, and show by immunofluorescence microscopy that it colocalizes with SMN in gems. Interestingly, unlike other gems proteins described so far, Gemin4 is also detected in the nucleolus, suggesting that it may have additional functions in ribosome biogenesis.

Proteins were labeled with [³⁵S]methionine by an in vitro coupled transcription-translation reaction (Promega Biotech). His-tagged Gemin4 (amino acids 611–1,058) and His-tagged SmB fusion proteins were expressed from pET28a in Escherichia coli strain BL21(DE3) and purified on nickel columns according to the manufacturer's recommendations. GST, GST-Gemin3, and GST-Gemin4 fusion proteins were expressed from pGEX-5X-3 (Pharmacia) in E. coli strain BL21 and purified using glutathione-Sepharose (Pharmacia) according the manufacturer's protocol.

Anti-Gemin4 antibody 22C10 was prepared by immunizing Balb/C mice with a His-tagged COOH-terminal fragment of Gemin4. Hybridoma production, screening, and ascites fluid production were performed as previously described (Choi and Dreyfuss 1984).

Immunoprecipitations of in vitro translated proteins were carried out in the presence of 1% Empigen BB buffer as previously described (Choi and Dreyfuss 1984). Coimmunoprecipitations were carried out using total HeLa lysate in the presence of 0.5% Triton X-100 as previously described (Pinol-Roma et al. 1988). For immunoblotting, proteins were resolved on 12.5% SDS–polyacrylamide gels and transferred to nitrocellulose membranes (Schleider and Schuell, Inc.) using a BioTrans Transblot apparatus (model B; Gelman Science) according to the manufacturer's instructions. The membranes were incubated in blocking solution (PBS 5% nonfat milk) for at least 1 h at room temperature, rinsed with cold PBS, and incubated in blocking solution with primary antibody for at least 1 h at room temperature. Membranes were washed three times in PBS containing 0.05% NP-40, followed by incubation with peroxidase-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories, Inc.), which were visualized by an ECL Western blotting kit (Amersham) after three additional washes with PBS containing 0.05% NP-40.

HeLa cells were cultured in DME supplemented with 10% FBS (both from GIBCO BRL).

Immunofluorescence staining was carried out essentially as previously described (Choi and Dreyfuss 1984). Double-labeled immunofluorescence experiments were performed by separate, sequential incubations of each primary antibody diluted in PBS containing 3% BSA, followed by the specific secondary antibody coupled to either FITC or Texas red. All incubations were carried out at room temperature for 1 h. Laser confocal fluorescence microscopy was performed with a Leica TCS 4D confocal microscope. Images from each channel were recorded separately and, where indicated, the files were merged. Antibodies used in these experiments were as follows: mouse IgG1 monoclonal anti-Gemin4 (22C10; this work); mouse IgG1 monoclonal anti-SMN (2B1; Liu and Dreyfuss 1996); rabbit polyserum anti-p80 coilin (R288; Andrade et al. 1991); mouse IgG3 monoclonal anti-Sm (Y12; Lerner et al. 1981); human autoimmune antibody against fibrillarin 1881 (Reimer et al. 1987); and rabbit affinity-purified anti-SMN exon 7 epitope antibody (Liu et al. 1997).

Purified GST or GST fusion proteins (2 μg) bound to 25 μl of glutathione-Sepharose beads were incubated with 10⁶ cpm of in vitro translated protein in 1 ml of binding buffer (50 mM Tris-HCl, pH 7.5, 200 mM NaCl, 2 mM EDTA, 0.1% NP-40, 2 μg/ml leupeptin and pepstatin A, and 0.5% aprotinin). After incubation for 1 h at 4°C, the resin was washed five times with 1 ml of binding buffer. The bound fraction was eluted by boiling in SDS-PAGE sample buffer and run on SDS-PAGE. The gels were fixed for 30 min and the signal was enhanced by treatment with Amplify solution (Amersham). For direct in vitro binding, we used 5 μg of purified His-tagged SmB and a binding buffer (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 2 mM EDTA, 0.05% NP-40, 2 μg/ml leupeptin and pepstatin A, and 0.5% aprotinin). The binding experiment was performed as described above, except that bound His-tagged SmB proteins were immunodetected by Western blot using a rabbit polyclonal anti-His tag antibody (Santa Cruz Biotechnology).

Total HeLa cell extract prepared in RSB-100 was loaded on a Superose 6 HR 10/30 column (Pharmacia). The column was washed with RSB-100 at a flow rate 0.5 ml/min. Fractions (1 min) were collected, and 1:20 of each fraction was resolved on SDS-PAGE followed by Western blotting.

Injections were carried out as previously described (Fischer et al., 1993). In brief, oocytes were incubated for 3 h in modified Barth's solution containing 0.2% collagenase type II (Sigma Chemical Co.). Defolliculated stage V and VI oocytes were collected and usually used the day after for microinjection. In a typical injection experiment, 30 nl of ³²P-labeled RNA (10⁶ cpm/μl; total concentration of 0.7 μM) was injected into the cytoplasm. For immunoprecipitation of RNA–protein complexes (Fisher et al., 1993), the injected oocytes were homogenized in 300 μl of ice-cold PBS, pH 7.4. The insoluble fraction was pelleted by centrifugation, and the clear supernatant was transferred into a new 1.5-ml Eppendorf tube containing antibodies bound to protein G–Sepharose (Pharmacia). This mixture was incubated with constant shaking for 1 h at 4°C and subsequently washed five times with 1 ml of ice-cold PBS. Bound RNAs were isolated by phenol extraction for 1 h, precipitated with ethanol, and analyzed by denaturing gel electrophoresis.

In vitro transcription of ³²P-labeled RNAs was carried out as described in Fisher et al. (1993) from plasmids encoding for U1, U2, U4, and U5 snRNAs. The plasmid encoding the chicken δ-crystallin pre-mRNA was previously described (Pellizzoni et al. 1998). The plasmid encoding the chicken δ-crystallin mRNA was constructed by elution of the chicken δ-crystallin mRNA from polyacrylamide gel followed by reverse transcriptase–PCR and subcloning into pSP65 (pSP1415m; Promega Corp). The plasmids used for in vitro transcription and translation of SMN, Gemin2, Gemin3, and the Sm proteins have been previously described (Charroux et al. 1999; Pellizzoni et al. 1999).

Immunoprecipitations from [³⁵S]methionine-labeled HeLa cell lysates with anti-SMN mAbs revealed the presence of several protein components in the SMN complex (Liu et al. 1997). Proteins that coimmunopurify with anti-SMN antibodies (Fig. 1) include Sm proteins, Gemin2 (Liu et al. 1997), and the recently described DEAD box protein Gemin3 (Charroux et al. 1999). In addition to these proteins, there are bands of 175, 97, 95, 60, and 50 kD that coimmunopurified with SMN (Fig. 1). In this paper, we describe the molecular cloning and characterization of the 97-kD polypeptide. The p97 band was digested in gel by trypsin, and the resulting peptides were sequenced by nanoelectrospray mass spectrometry as described previously (Shevchenko et al. 1996; Wilm et al. 1996). Searching the databases with several peptides from this band (using the peptide sequence tag algorithm) we identified a human expressed sequence tag sequence (clone no. R55454) (Shevchenko et al. 1996). Several additional cDNA clones were obtained by hybridization screening of a human leukemia 5′-STRETCH PLUS cDNA library using the expressed sequence tag clone as a probe. We isolated 20 independent partial cDNA clones with insert sizes ranging from 0.8 to 3.1 kb, all of which contained overlapping regions of the same open reading frame (ORF). A cDNA clone containing the longest ORF was constructed, and conceptual translation of its nucleotide sequence revealed a potential initiator methionine preceded by an in-frame stop codon. This cDNA encodes a putative protein of 1,058 amino acids with a calculated molecular mass of 119.9 kD and an estimated pI of 5.68. Database searches with the full-length clone did not reveal significant homology to any other protein or any recognizable motifs. Thus, this is a full-length cDNA clone that encodes a novel component of the SMN complex (see below), which we termed Gemin4 for component of gems number 4.

To investigate the interaction of Gemin4 with SMN and to characterize Gemin4 further, we generated mAbs by immunizing mice with a bacterially produced, purified recombinant His-tagged Gemin4 fragment (from amino acids 611 to 1,058). One hybridoma, 22C10, was selected for additional studies. Several lines of evidence demonstrate that this hybridoma indeed produced mAb that recognizes Gemin4 specifically. First, 22C10 immunoprecipitated Gemin4 produced by in vitro transcription and translation from the Gemin4 cDNA but not similarly produced hnRNP A1 or SMN proteins (Fig. 2 A). Second, on an immunoblot of total HeLa lysate, 22C10 recognized a single protein of ∼97 kD (Fig. 2 B). Finally, 22C10 specifically immunoprecipitated a single protein of ∼97 kD from [³⁵S]methionine-labeled HeLa and mouse 3T3 cell lysates (Fig. 2 C), suggesting that Gemin4, like SMN, is conserved in vertebrates.

Indirect immunofluorescence laser confocal microscopy using mAb 22C10 was performed on HeLa cells to determine the subcellular localization of Gemin4. Fig. 3A and Fig. C, shows that anti-Gemin4 antibodies display intense staining of prominent discrete bodies in the nucleus, and these are also readily discernible by differential interference contrast (Fig. 3 D). In addition, the 22C10 antibody stained nucleoli (see below). To determine if the discrete nuclear structures stained by 22C10 are gems or CBs, we performed double label immunofluorescence experiments using antibodies against Gemin4 and either p80-coilin, as a marker of CBs (Andrade et al. 1991), or SMN, as a marker of gems (Liu and Dreyfuss 1996). In many cell lines, the antibody staining of gems and CBs entirely overlap, however, in the HeLa PV strain used here these two bodies are frequently found separate from each other (Liu and Dreyfuss 1996; Matera and Frey 1998). In a recent report, Carvalho et al. 1999 have also observed that gems are often detected separately from CBs. The nuclear structures that contain Gemin4 are clearly distinct from CBs but are completely colocalized with SMN in gems (Fig. 3B, Fig. C, and Fig. E). The colocalization of Gemin4 with SMN strongly supports the conclusion that these two proteins exist as a complex in the cell. Thus, Gemin4 is the fourth constituent of gems described so far. Like staining with antibodies to SMN, Gemin2 and Gemin3, 22C10 shows a diffuse cytoplasmic staining for Gemin4 (Fig. 3 A).

To further study the nucleolar staining of Gemin4, we performed double label immunofluorescence experiments using the Gemin4 antibody 22C10 and the fibrillarin antibody 1881 as a marker of the nucleolus (Raska et al. 1990). The staining, shown in Fig. 3 F, indicates that Gemin4 and fibrillarin do not colocalize, but both show distinct nucleolar patterns. This demonstrates that Gemin4 is present in the nucleolus but excluded from the dense fibrillar compartment of the nucleolus (Ochs and Smetana 1991). Thus, Gemin4 represents the first component of the SMN complex that localizes both in gems and nucleoli (see Discussion).

To characterize the Gemin4-containing complex, we tested for its presence in the SMN complex in vivo by coimmunoprecipitation and Western blotting experiments. The anti-Gemin4 mAb 22C10 was used for immunoprecipitation from HeLa cell total extracts, and these were resolved by SDS-PAGE, immunoblotted, and probed with the anti-SMN antibody 2B1 (Liu and Dreyfuss 1996). As shown in Fig. 4 A (lane 22C10 IP), 2B1 readily detects SMN in the 22C10 immunoprecipitates, indicating that SMN is coimmunoprecipitated with Gemin4. Because SMN forms a stable complex with Gemin2 in vivo and in vitro (Liu et al. 1997), we also investigated whether Gemin2 could be coimmunoprecipitated with Gemin4. As shown in Fig. 4 A, the anti-Gemin2 mAb 2S7 clearly detects Gemin2 in the anti-Gemin4 immunoprecipitates (lane 22C10 IP). We also examined whether Gemin4 can be coimmunoprecipitated with Gemin3, which is a recently identified novel component of the SMN complex (Charroux et al. 1999). Fig. 4 A shows that, like SMN and Gemin2, Gemin4 is present in the anti-Gemin3 (lane 11G9 IP) immunoprecipitate (Charroux et al. 1999). In a reciprocal experiment, the Gemin4 protein could also be coimmunoprecipitated by the anti-SMN, the anti-Gemin2, and the anti-Gemin3 mAbs (Fig. 4 B). No SMN, Gemin2, Gemin3, or Gemin4 proteins were detected in the control nonimmune (SP2/0) immunoprecipitate (data not shown and lane SP2/O, Fig. 4 B). These results suggest that SMN, Gemin2, Gemin3, and Gemin4 are associated in vivo in one or more complexes that can be immunoprecipitated by either anti-SMN, anti-Gemin2, anti-Gemin3, or anti-Gemin4 antibodies.

Further support for the existence in vivo of a single complex that contains SMN, Gemin2, Gemin3, and Gemin4 was obtained from gel filtration experiments. Total HeLa extract was fractionated on a Superose 6 HR 10/30 high performance gel filtration column; fractions were resolved by SDS-PAGE, blotted, and probed with anti-SMN, anti-Gemin2, anti-Gemin3, and anti-Gemin4 antibodies. SMN, Gemin2, Gemin3, and Gemin4 comigrated and showed a peak at ∼800 kD, indicating that they are components of a large macromolecular complex (Fig. 4 B). Gemin4 is also detected in a second complex of ∼550 kD that lacks SMN and Gemin2 but does contain a faster migrating form of Gemin3. Thus, there appear to be at least two different Gemin4-containing complexes.

A second peak containing SMN and Gemin2, but lacking Gemin3 and Gemin4, is observed at lower molecular mass fractions that correspond to ∼100 kD. The SMN protein present in these fractions was not detectable with a rabbit polyclonal antibody specific to the peptide sequence encoded by exon 7 of SMN (Liu et al. 1997; Fig. 4 B). Thus, this smaller SMN–Gemin2 complex contains the oligomerization-deficient form of SMN lacking amino acids encoded by exon 7, which is most likely produced by the SMN2 gene (see Discussion; Gennarelli et al. 1995; Lefebvre et al. 1995; Pellizzoni et al. 1999).

To investigate the interactions of Gemin4, we performed in vitro protein binding assays using Gemin4 and several constituents of the SMN complex. For this assay, purified GST or GST-Gemin4 fusion immobilized on gluthathione-Sepharose were incubated with [³⁵S]methionine-labeled SMN, Gemin2, Gemin3, and Gemin4 produced by in vitro transcription and translation in rabbit reticulocyte lysate. As shown in Fig. 5 A, full-length Gemin3 is the only protein of the SMN complex that binds specifically to immobilized GST-Gemin4 (Fig. 5 A). No binding to GST alone was observed (data not shown).

To further investigate this interaction, Gemin3 was expressed as a fusion protein with GST, and Gemin4 was produced and labeled with [³⁵S]methionine in rabbit reticulocyte lysate. GST-Gemin3 fusion was incubated with labeled Gemin4 in the presence of increasing salt concentrations. As shown in Fig. 5 B, full-length Gemin4 bound to GST-Gemin3, and this binding appears to be avid because it is not disrupted at 750 mM NaCl. No binding of Gemin4 to GST alone was detected (data not shown).

To further characterize the Gemin4 complex, immunoprecipitations using anti-Gemin4 mAbs from [³⁵S]methionine-labeled HeLa cells were carried out, and the immunoprecipitated proteins were analyzed by SDS-PAGE. As references for these immunoprecipitations, we also performed immunoprecipitation with the anti-Sm mAb Y12 (Lerner and Steitz 1979; Lerner et al. 1981) and with the anti-SMN mAb 2B1. As shown in Fig. 6 A, several proteins are coimmunoprecipitated with anti-Gemin4 antibodies, and the pattern is very similar to the one obtained with anti-SMN antibodies. Besides SMN, Gemin2, Gemin3, and Gemin4, the Sm core proteins B/B′, D1-3, E, F, and G proteins were also coimmunoprecipitated with the 22C10 anti-Gemin4 antibody. Additional proteins coimmunoprecipitated specifically with anti-SMN (at 175, 95, 60, and 50 kD) or anti-Gemin4 (at 80 kD) mAbs. Note that the U1-specific A protein, but not the U1-specific C protein, coimmunoprecipitates with SMN and Gemin4. This indicates that the SMN complexes are not associated with mature snRNP particles.

The in vivo association of Gemin4 with the Sm proteins was confirmed by a coimmunoprecipitation and Western blot experiment. The anti-Sm mAb Y12 was used for immunoprecipitation from HeLa cell total extracts, and the immunoprecipitate was analyzed by SDS-PAGE and an immunoblot with the anti-Gemin4 antibody 22C10. As shown in Fig. 6 B, 22C10 readily detects Gemin4 in the Y12 immunoprecipitate, demonstrating that Gemin4 is associated with the Sm proteins in vivo. To investigate whether Gemin4 interacts with Sm proteins directly, purified recombinant GST-Gemin4 protein was used for binding assays with in vitro produced [³⁵S]methionine-labeled Sm proteins. Fig. 6 C shows that the Sm proteins B, D1, D2, D3, and E bind to GST-Gemin4, whereas there is no detectable binding to GST alone (data not shown). To determine if Gemin4 is able to interact directly with the Sm proteins and not via other components of the SMN complex that may be present in the rabbit reticulocyte lysate, recombinant His-tagged SmB protein was produced and incubated with GST or GST-Gemin4. Bound proteins were resolved by SDS-PAGE, and analyzed by Western blotting with a rabbit polyclonal antibody specific to the 6xHis-tag. As shown in Fig. 6 D, SmB binds specifically to Gemin4 but not to GST. We conclude that Gemin4 interacts directly with SmB.

To determine whether Gemin4 was associated with snRNAs in vivo, we used the Xenopus oocyte that provides a particularly advantageous system in which to study spliceosomal snRNP biogenesis by use of microinjection (Mattaj and De Robertis 1985; Mattaj 1986). We first wished to determine whether Gemin4 is present in Xenopus cells and whether it can be recognized by 22C10 antibody. The 22C10 mAb showed both gems and nucleolar staining on Xenopus XL177 somatic cells, strongly suggesting that Gemin4 is conserved in Xenopus (data not shown). However, immunoblotting with the anti-human Gemin4 mAb 22C10 on Xenopus tissue culture cells or on Xenopus oocyte lysates did not detect any protein (data not shown). To determine if Gemin4 is associated with U snRNAs in vivo, various ³²P-labeled RNAs including chicken δ-crystallin mRNA, chicken δ-crystallin pre-mRNA, and the spliceosomal snRNAs U1 and U5 were produced by in vitro transcription and a mixture of these RNAs was microinjected into the cytoplasm of oocytes. After 3 h, immunoprecipitations were carried out with anti-Gemin4 (22C10) and, as a positive control, anti-Gemin2 (2E17) antibodies (Fischer et al. 1997). Fig. 7 A shows that only U1 and U5 snRNAs are efficiently and specifically precipitated, indicating that they associate with Gemin4. A similar, but less efficient, immunoprecipitation of U1 and U5 snRNAs was observed with the anti-Gemin2 antibody. We further asked whether the other spliceosomal U snRNAs, U2 and U4, were associated with Gemin4 as well, and whether the association of Gemin4 with mature U snRNPs was also observed in the nucleus. To do so, a mixture of ³²P-labeled U snRNAs were injected into the cytoplasm of oocytes followed by an 18-h incubation (Fig. 7 B). After this incubation period, ∼50% of the injected snRNA was imported into the nucleus while the rest remained in the cytoplasm. Immunoprecipitations from the nuclear and cytoplasmic fractions were carried out with either anti-Gemin4 antibody, anti-Sm antibody as a positive control, or SP2/O as a negative control. The coimmunoprecipitated RNAs were analyzed by gel electrophoresis. As previously reported, U1, U2, U4, and U5 were efficiently immunoprecipitated by Y12 in approximately equal amounts from the nucleus and the cytoplasm (Fig. 7 B; Mattaj 1986; Fisher and Luhrmann, 1990). In contrast, Gemin4 associated more efficiently with U1 and U5 than U2 and U4, and this association was only observed in the cytoplasm (Fig. 7 B).

We conclude that the Gemin4, like SMN and Gemin2, is stably associated with U1 and U5 snRNAs in the cytoplasm of Xenopus oocytes but not after these snRNAs have been assembled into snRNPs and imported into the nucleus. Thus, like SMN and Gemin2 (Fisher et al., 1997), Gemin4 likely dissociates from the spliceosomal snRNPs either immediately before nuclear entry or shortly thereafter.

Using a biochemical approach to characterize additional components of the SMN complex, we have identified a novel protein termed Gemin4. Gemin4 does not contain any known or recognizable protein motifs. Several lines of evidence suggest that Gemin4, together with SMN, Gemin2, and Gemin3, function as a complex in vivo. SMN, Gemin2, Gemin3, and Gemin4 can be coimmunoprecipitated, and are present in a large complex that also contains the spliceosomal snRNP core Sm proteins. Like SMN and Gemin3, Gemin4 interacts directly with several snRNP Sm core proteins, including B/B′, D1-D3, and E and is associated with U1 and U5 snRNAs in the cytoplasm of Xenopus oocytes, where snRNP assembly takes place (Liu et al. 1997; Fischer et al. 1997; Charroux et al. 1999; Pellizzoni et al. 1999). We have previously shown that the SMN complex plays a critical role in spliceosomal snRNP assembly in the cytoplasm, and is required for pre-mRNA splicing in the nucleus (Fischer et al. 1997; Pellizzoni et al. 1998). Thus, it is likely that Gemin4 also plays an important role in these processes. Note, that so far, we have not been able to detect any effect of anti-Gemin4 antibody 22C10 on snRNP assembly and/or snRNP import into the nucleus upon microinjection into Xenopus oocytes (data not shown).

Unlike Gemin2 and Gemin3, Gemin4 does not interact with SMN directly and its presence in the SMN complex is probably the result of its direct and stable interaction with Gemin3. The observation that Gemin4 and the DEAD box protein Gemin3 interact with each other directly and avidly suggests that they function together. Previous studies have shown that the RNA helicase activity of the translation initiation factor eIF4A, also a DEAD box protein, is dependent on the presence of a second initiation factor, eIF4B. Interestingly, in a series of preliminary experiments we have so far not been able to detect RNA helicase or RNA-dependent ATPase activity for recombinant Gemin3 (Charroux et al. 1999). It is possible that such activity will only manifest itself when Gemin3 is associated with other proteins such as Gemin4.

Gel filtration experiments revealed the presence of two SMN complexes in HeLa cells. The high molecular mass complex is the most abundant and contains all the components of the SMN complex thus far identified (SMN, Gemin2, Gemin3, and Gemin4) and likely represents an active form of the complex (see below). The large size of this complex is likely the result of the capacity of SMN to form large oligomers (Pellizzoni et al. 1999). The second complex probably represents a monomeric, COOH-terminal–truncated form of SMN (SMNΔEx7) associated with Gemin2 (Pellizzoni et al. 1999). While SMN1 produces only full-length mRNA, SMN2 mainly produces an alternatively spliced form of SMN mRNA lacking exon 7 (Gennarelli et al. 1995; Lefebvre et al. 1995). Exon 7 skipping is due to the presence of a single nucleotide change in the SMN2 gene compared with SMN1 (Lorson et al. 1999), and the ratio of alternatively spliced versus full-length SMN2 mRNA correlates with the severity of SMA (Gavrilov et al. 1998). Nevertheless, no evidence for the presence of SMN protein lacking exon 7–encoded amino acids in vivo has been reported. The absence of the amino acid sequence encoded by exon7 is thought to generate a nonfunctional SMN protein that lacks the capacity to oligomerize and, thus, cannot interact with Sm proteins (Burghes 1997; Lorson et al. 1998; Pellizzoni et al. 1999). The absence of Gemin3 and Gemin4 from the SMNΔEx7–Gemin2 complex probably results from the defective interaction of SMNΔEx7 with Gemin3 (Charroux et al. 1999). We have shown that SMNΔEx7 is a nonfunctional protein that is incapable, unlike wild-type SMN, of regenerating splicing extracts in vitro (Pellizzoni et al. 1998). Thus, the loss of function of SMNΔEx7 is likely due to its defective interaction with the Sm proteins as well as with the Gemin3–Gemin4 complex (Charroux et al. 1999; Pellizzoni et al. 1999). Therefore, we suggest that the SMNΔEx7–Gemin2 complex represents an inactive form of the SMN complex, whereas the high molecular mass complex containing SMN, Gemin2, Gemin3, and Gemin4 represents the active complex that can bind substrates such as the Sm proteins and carry out the functions of the complex (Fig. 8; Fischer et al. 1997, Pellizzoni et al. 1998, Pellizzoni et al. 1999).

Higher eukaryotic nuclei contain numerous morphologically distinct substructures or nuclear bodies (for reviews see Singer and Green 1997; Lamond and Earnshaw 1998; Matera 1999). SMN, Gemin2, Gemin3, and Gemin4, in addition to their general localization in the cytoplasm, are found in the nucleus, where they are concentrated in gems (Liu and Dreyfuss 1996; Liu et al. 1997). Gems are similar in size and number to CBs, and these two bodies are often found either entirely merged or in close proximity (Liu and Dreyfuss 1996; Liu et al. 1997; Matera and Frey 1998). CBs are highly enriched in snRNPs and snoRNPs and, together with gems, appear to be involved in snRNP biogenesis and metabolism (Gall et al. 1995; Lamond and Earnshaw 1998; Pellizzoni et al. 1998; Matera 1999; Sleeman and Lamond 1999). In addition to its localization in gems, Gemin4 is in the nucleoli and, thus, represents the first component of the SMN complex present in this nuclear compartment. Using a different polyclonal anti-SMN antibody from 2B1, others have observed a strong nucleolar immunolocalization of SMN in mouse and human CNS tissues (Francis et al. 1998). However, SMN has not been detected in nucleoli of HeLa cells. Interestingly, SMN interacts with fibrillarin (Liu and Dreyfuss 1996), a common component of small nucleolar RNPs (snoRNPs). Fibrillarin is found in the nucleoli and in CBs and may be the snoRNP's functional equivalent of the core Sm proteins of spliceosomal snRNPs (Tyc and Steitz 1989; Maxwell and Fournier 1995; Smith and Steitz 1997). Because of their close association with nucleoli, it has been suggested that CBs participate in snoRNP biogenesis and/or metabolism (Raska et al. 1990). Given the close association between CBs and gems and the interaction between SMN and fibrillarin, it is possible that the SMN complex also plays a role in snoRNP assembly and ribosomal RNA metabolism. Therefore, Gemin4 may function to connect SMN to both UsnRNA biogenesis and rRNA biogenesis.

We thank Dr. Eng M. Tan (Scripps Research Institute, La Jolla, CA) for the anti-p80 coilin and anti-fibrillarin antibodies and Dr. Joan Steitz (Yale University School of Medicine, New Haven, CT) for the anti-Sm antibody. The plasmid encoding the chicken δ-crystallin mRNA was a gift from Dr. Naoyuki Kataoka (our laboratory). We are grateful to members of our laboratory for stimulating discussions and, in particular, Drs. Westley Friesen, Zissimos Mourelatos, and Sara Nakielny for helpful discussions and critical comments on this manuscript.

This work was supported by a grant from the National Institutes of Health. G. Dreyfuss is an Investigator of the Howard Hughes Medical Institute.