Survival Motor Neuron (SMN) Protein Interacts with Transcription Corepressor mSin3A*

Spinal muscular atrophy (SMA) is the leading genetic cause of infant mortality. SMA results from loss of survival motor neuron ( SMN ) expression and subsequent death of motor neuron cells. To study SMN-associated proteins that may be involved in transcriptional regula-tion, we carried out immunoprecipitation experiments and found that the transcription corepressor mSin3A associates with SMN protein. Deletional analysis local-ized the mSin3A-interacting domain to the exon 6 region of SMN. When targeted to a promoter, wild-type SMN was able to repress transcription of a downstream luciferase reporter gene. This repression was relieved by treatment with the histone deacetylase inhibitor trichostatin A in a dose-dependent manner, and deletion of exon 6 abolished the ability of SMN to repress the reporter gene. Analysis of SMN missense mutations within the exon 6 region implicated the SMA-associated mutation Y272C with impairment of the mSin3A-interaction. Gel filtration experiments revealed that wild-type SMN, via the exon 6 region, forms protein supra-complexes exceeding 40,000 kDa in size, whereas the Y272C mutation may affect higher order protein assembly, as the mutant SMN was normalized to the internal (cid:1) -galactosidase or Renilla luciferase control and then compared with the luciferase activity from cells transfected with the pCS2-Gal4-MT empty vector. Antibodies— rabbit polyclonal unrelated rabbit polyclonal 2.5 m M MgCl 2 ) using a fast protein liquid chromatography system (Amersham Biosciences). The eluent was collected at 0.5 ml/fraction, separated by SDS-PAGE, and blotted with an anti-SMN or anti-Myc antibody to detect fractions containing endogenous and Myc-epitope-tagged SMN proteins. The molecular sizes of protein complexes in the collection were calculated according to a chromatogram of standard markers provided by the manufacturer.

Spinal muscular atrophy (SMA) is the leading genetic cause of infant mortality. SMA results from loss of survival motor neuron (SMN) expression and subsequent death of motor neuron cells. To study SMN-associated proteins that may be involved in transcriptional regulation, we carried out immunoprecipitation experiments and found that the transcription corepressor mSin3A associates with SMN protein. Deletional analysis localized the mSin3A-interacting domain to the exon 6 region of SMN. When targeted to a promoter, wild-type SMN was able to repress transcription of a downstream luciferase reporter gene. This repression was relieved by treatment with the histone deacetylase inhibitor trichostatin A in a dose-dependent manner, and deletion of exon 6 abolished the ability of SMN to repress the reporter gene. Analysis of SMN missense mutations within the exon 6 region implicated the SMA-associated mutation Y272C with impairment of the mSin3A-interaction. Gel filtration experiments revealed that wild-type SMN, via the exon 6 region, forms protein supra-complexes exceeding 40,000 kDa in size, whereas the Y272C mutation may affect higher order protein assembly, as the mutant SMN was more abundant in smaller complexes. Together, these findings provide a potential mechanism by which lack of fully functional SMN protein is detrimental to motor neuron survival.
Spinal muscular atrophy (SMA) 1 is a common human autosomal recessive neurodegenerative disease that leads to the death of spinal cord motor neurons. SMA occurs with a frequency of 1 in 10,000 individuals and is the most common genetic cause of infant mortality (1,2). SMA is classically subdivided into three types based on the age of onset and clinical severity (3). Type I SMA is characterized by severe muscular problems in infancy, whereas type II and type III SMA are characterized by minor muscle weakness in adult-hood. Independent of the age of onset, the clinical features are muscle weakness and hypotonia. SMA arises from deletion and/or mutation in the telomeric copy of the survival of motor neurons (SMN) gene (1). A near identical centromeric copy of the SMN gene primarily generates an alternatively spliced product encoding an unstable protein lacking sequence derived from exon 7 (4,5).
SMN protein is perhaps best known for its involvement in mRNA biogenesis by playing a role in the assembly and regeneration of small nuclear ribonucleoproteins (snRNPs) and spliceosomes (6,7). In this study, we report the first evidence that SMN also interacts with the transcription corepressor mSin3. Our deletion analysis reveals that the mSin3-interacting domain is encoded by exon 6 of SMN. In mammalian cells, mSin3 is expressed as the highly related mSin3A and mSin3B proteins (8). mSin3 associates with histone deacetylases (HDACs), methyltransferases and other factors to regulate the accessibility of chromatin (9 -12). In light of findings from our group and others, it is possible that a subset of SMN protein may be involved in the repression of genes critical to motor neuron survival.
Plasmids-Myc-epitope-tagged expression plasmid pCS2-MT-SMN was generated by cloning full-length SMN cDNA in-frame into the EcoRI-XbaI sites of the pCS2-MT vector. SMN deletion mutants lacking exons 1, 5, 6, or 7 were generated by polymerase chain reaction and cloned in-frame into the same EcoRI-XbaI sites of pCS2-MT vector. Point mutations S262I, Y272C, T274I, G275S, and G279V were introduced into full-length SMN cDNA by the Gene Editor TM site-directed mutagenesis kit (Promega). For generation of pCS2-Gal4-MT-SMN, the Gal4 DNA-binding domain was cloned in-frame into the unique ClaI site within the pCS2-MT-SMN vector.
Transfection and Luciferase Assay-Two duplicate wells of 65% confluent 293 cells in 6-well plates were transfected with 500 ng of firefly luciferase reporter pG5-SP1-Luc (9), pG5-E1b-Luc, (13) or pG4 -14D-Luc (14), 100 ng of pCS2-Gal4-MT-SMN, and 10 ng of pCMV-␤-gal (or 30 ng of pRL-SV40), plus 30 l of DOTAP transfection reagent (Roche Applied Science). 48 h later, cells were assayed for firefly luciferase activity on a TD-20e luminometer (Turner Designs, Sunnyvale, CA). To assess inhibition of histone deacetylases, cells were treated with trichostatin A (TSA) at different concentrations 1 day before the luciferase assay. Transfection was repeated at least three times; the luciferase activity was normalized to the internal ␤-galactosidase or Renilla luciferase control and then compared with the luciferase activity from cells transfected with the pCS2-Gal4-MT empty vector.
Immunoprecipitation and Western Blotting-1 ϫ 10 6 293 cells were lysed in 0.4 ml buffer A (10 mM Tris, pH 7.4, 100 mM NaCl, 2.5 mM MgCl 2 , 0.5% Triton X-100, 10 mM dithiothreitol) supplemented with protease inhibitor and phosphatase inhibitor mixture (Sigma). 25 l of the respective antibodies were incubated with 40 l of protein A/G agarose (Santa Cruz Biotechnology) for 30 min at 4°C in 0.2 ml of buffer A. The antibody-protein A/G-agarose complex was then incubated with 0.25 ml of fresh cell lysate for 6 h at 4°C on a rotating wheel. After one 10-min wash with buffer A and two washes with RIPA buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS), 40 l of 1ϫ SDS sample buffer was added to the agarose beads and denatured at 96°C for 5 min. 10 l of the sample was loaded onto an SDS-10% polyacrylamide gel. After separation and transfer to a polyvinylidene difluoride PVDF membrane, the protein samples were blotted with anti-SMN, anti-Myc, anti-HDAC2, or anti-mSin3A antibodies. Protein bands were visualized using the ECL Western blotting analysis system (Amersham Biosciences).
Gel Filtration-0.4 ml of frozen lysate from 293 cells expressing wild-type or mutant SMN protein was thawed, loaded onto a Superose 6 10/300 GL high performance column and size-fractionated with the elution buffer (10 mM Tris, pH 7.4, 100 mM NaCl, 2.5 mM MgCl 2 ) using a fast protein liquid chromatography system (Amersham Biosciences). The eluent was collected at 0.5 ml/fraction, separated by SDS-PAGE, and blotted with an anti-SMN or anti-Myc antibody to detect fractions containing endogenous and Myc-epitope-tagged SMN proteins. The molecular sizes of protein complexes in the collection were calculated according to a chromatogram of standard markers provided by the manufacturer.

RESULTS
Endogenous SMN Interacts with mSin3/HDAC-A schematic of SMN protein is shown ( Fig. 1A) with the indicated localizations of exons and SMA-associated point mutations. SMN is a ubiquitously expressed protein (15) and is abundant in 293 cells (Fig. 1B, lane 1, upper panel). To determine whether SMN protein is associated with mSin3 in vivo, 293 whole-cell lysate was immunoprecipitated with the rabbit K-20 anti-mSin3A antibody (Fig. 1B, lane 2) or with the rabbit AK-12 anti-mSin3B antibody (Fig. 1B, lane 3). When the immunoprecipitates were blotted with a mouse monoclonal anti-SMN antibody, it was found that both anti-mSin3A and anti-mSin3B antibodies were able to coimmunoprecipitate endogenous SMN protein. To ensure that this SMN-mSin3 interaction is specific, an unrelated rabbit polyclonal Y-11 anti-HA antibody was used in the same experiment and failed to coimmunoprecipitate SMN (Fig. 1B, lane  4).
Because mSin3 is known to complex with other histone modification enzymes (9,14,16), we carried out experiments to address whether SMN can also associate with mSin3-interacting proteins such as HDACs. Lysate from 293 cells (Fig. 1C, lane 1) was immunoprecipitated with a goat polyclonal anti-HDAC2 (Fig. 1C, lane 2) or a control goat IgG (Fig. 1C, lane 3). The immunoprecipitates were then blotted with anti-SMN (Fig. 1C, top panel), anti-mSin3 (Fig. 1C, middle panel), or anti-HDAC2 (Fig. 1C, bottom panel). Western blotting revealed that the anti-HDAC2 antibody coimmunoprecipitated both mSin3A and SMN proteins, whereas the control goat IgG did not bring down either of them in the experiments. These findings suggest that a subset of SMN forms stable intracellular complexes with mSin3/HDAC. SMN-mSin3A Interaction Requires the Exon 6 Region of SMN Protein-In the next experiment, we sought to map the SMN subdomain(s) responsible for interacting with mSin3. Full-length SMN and SMN deletion mutants lacking exons 1, 5, 6, or 7 were cloned into the mammalian expression vector, pCS2-MT. A schematic representation of SMN deletion mutants is shown ( Fig. 2A). After transfection of 293 cells with these SMN deletion constructs, the resultant lysates were incubated with K-20 antibody which can coimmunoprecipitate endogenous mSin3A and endogenous SMN (Fig. 1B). Myctagged full-length SMN and its deletion mutants were expressed in 293 cells, as confirmed by Western blotting (Fig. 2B,  lanes 2-6, upper panel). As expected, Myc-tagged full-length SMN was shown to associate with mSin3A (Fig. 2B, lane 2 Wild-type SMN Represses Transcription in an HDACdependent Manner-As a well documented transcription corepressor, mSin3A is known to exert its repressive effects by means of association with histone deacetylase, histone methyltransferase, as well as other chromatin remodeling factors (9 -12). Because our data showed that SMN protein could interact with mSin3/HDAC, we tested whether directing SMN to a particular promoter would inhibit transcription through recruitment of mSin3A and its associated HDAC activity. Because SMN has not been reported to bind to DNA in a sequence-specific manner, we constructed the expression plasmid pCS2-Gal4-MT-SMN in which the Gal4 DNA-binding domain is fused to the Myc-SMN protein. 293 cells were then transfected with pCS2-Gal4-MT-SMN plus a luciferase reporter containing five consensus Gal4-binding sites and an SP1 site in the promoter region (Fig 3A). When compared with the pCS2-Gal4-MT vector, expression of Gal4-Myc-SMN resulted in repression of the reporter gene activity by more than 70% (Fig.  3B, compare columns 1 and 2). Furthermore, this repressive function of Gal4-Myc-SMN seems to require HDAC activity because the repression was significantly relieved by treatment of the cotransfected cells with TSA, a well known in- hibitor of HDACs (17,18), at a concentration of 125 ng/ml or higher (Fig. 3B, columns 3 through 6).
To determine the minimal region(s) required for transcriptional repression, the same Gal4 DNA-binding domain was fused in-frame to SMN deletion mutants and then cotransfected with the pG5-SP1-Luc reporter. SMN mutants lacking exons 1, 5, or 7 repressed transcription from the reporter gene as strongly as the wild-type SMN. In contrast, SMN mutant lacking exon 6 no longer possessed such repressive activity (Fig. 4, top panel). To rule out the possibility that the observed differences were due to inconsistency in transfection and/or protein expression, 293 lysates expressing these Gal4-Myc-SMN mutants were separated by SDS-PAGE for protein analysis. Western blotting showed that the Gal4-fused SMN proteins were expressed at comparable levels in all these samples (Fig. 4, bottom panel), thus repression of the reporter gene by Gal4-fused SMN is unlikely to be an artifact. Instead, these results suggest that targeting of SMN to a promoter can result in recruitment of mSin3 by means of the exon 6 subdomain of SMN, and that subsequent repression of the reporter gene is mediated, at least in part, through TSA-sensitive histone deacetylases.
SMA-associated Mutation Y272C May Affect SMN-mSin3A Interaction-It is known that homozygous deletion of the telomeric SMN is responsible for SMA in 94% of cases. About 4% of patients with SMA retain at least one allele of the telomeric SMN (1). In these cases, intragenic mutations are thought to be the cause of SMA. At the present time, several SMA-related missense mutations are known to be within or immediately adjacent to the exon 6 region. These mutations include S262I, Y272C, T274I, G275S, and G279V (Fig. 1A). Among these mutations, Y272C is the most frequently occurring one that may be associated with type I SMA (19).
To investigate the binding of these mutants to mSin3A, we performed site-directed mutagenesis to introduce these point mutations into wild-type SMN. 293 cells were then transfected with the resultant missense mutation constructs, and the lysates were subjected to immunoprecipitation with the K-20 anti-mSin3A antibody. Western blotting showed that the expression levels of Myc-SMN missense mutants were comparable for all constructs (Fig. 5A, lane 1-7, upper panel). The amount of Myc-SMN-Y272C that was immunoprecipitated with K-20 antibody was lower than that of Myc-SMN protein (Fig. 5A, compare lanes 1 and 3, middle panel). However, the decrease of Myc-SMN-Y272C in the immunoprecipitate was not as complete as Myc-SMN-⌬6 (Fig. 5A, compare  lanes 3 and 7, middle panel). There was no significant change in binding of other missense mutants to mSin3A (Fig. 5, lanes  2, 4, 5, 6, middle panel). Taken together, these data suggest that the SMA-associated Y272C missense mutation may have an adverse effect on SMN-mSin3A interaction, whereas other SMN missense mutations have no appreciable effects on the binding under our experimental conditions.
To investigate whether SMN missense mutations also affect repression of the luciferase reporter gene, we constructed expression plasmids by fusing Gal4 DNA-binding domain to these SMN missense mutants. Compared with the wild-type protein, these SMN missense mutations showed minimal effects on luciferase activity and, therefore, did not seem to cause derepression of the reporter gene (Fig. 5B). To examine whether potential derepression by the Y272C missense mutation could be obtained with a different promoter construct, we tested another luciferase reporter, pG4 -14D-Luc, which contains four consensus Gal4-binding sites upstream of the 14D promoter (14). When pG4 -14D-Luc was co-transfected into 293 cells, 10-fold repression of the luciferase activity by Gal4-Myc-SMN was observed (Fig. 5C, compare the first two columns). However, none of the five SMN missense mutations had any effect on derepression (Fig. 5C, columns 2-7).
To investigate whether repression by Gal4-Myc-SMN can work on a TATA-containing promoter, the pG5-E1b-Luc reporter construct (13) was used in co-transfection with pCS2-Gal4-MT-SMN. This reporter construct contained five consensus Gal4-binding sites upstream of the E1b promoter. Sequence analysis revealed the presence of a TATA box Ϫ30 bp from the transcription initiation site. Although Gal4-Myc-SMN was able to repress expression of the luciferase reporter gene driven by this TATA-containing promoter, the five SMN missense mutations did not show appreciable derepression as compared with the wild-type SMN (Fig. 5D).
Exon 6 Region of SMN Mediates Higher Order Complex Formation-These findings suggest that either SMN mutants do not derepress gene expression or our luciferase reporter approach is unable to assess the impairment of SMN missense mutations on potential gene repression. Previously, a cell viability and proliferation assay also failed to differentiate SMArelated missense mutants from wild-type SMN (20). If these SMA-related missense mutations indeed contribute to the pathological process of motor neuron death, they would be expected to lack at least some of the functions of wild-type SMN protein.
Interestingly, a previous in vitro study reported that the exon 6 of SMN encodes a self-oligomerization region, and we showed in this study that the same exon 6 region is important for interaction with mSin3 in vivo. These findings suggest that the exon 6 region of SMN may be involved in the formation of higher order protein complexes inside cells, and SMA-related missense mutations may be deficient in such a function. To test this hypothesis, lysate from 293 cells expressing Myc-epitopetagged SMN, SMN-⌬6, or SMN-⌬7 was size-fractionated through a Superose column; all of the fractions were then analyzed by Western blotting. In 293 cells, endogenous SMN protein existed as multi-protein complexes exceeding 40,000 kDa (Fig. 6A, top panel), and wild-type Myc-SMN also formed similar higher order protein complexes (Fig. 6A, second panel). In contrast, exon 6-deletion stripped the mutant Myc-SMN-⌬6 of most interacting proteins, leading to a dramatic reduction in the size of the Myc-SMN-⌬6 complexes to around 200 kDa (Fig.  6A, third panel). The elution profile of Myc-SMN-⌬7 was not affected by exon 7 deletion and was similar to that of Myc-SMN (Fig. 6A, bottom panel), indicating that exon 7 region of SMN is not critical to higher order protein assembly.
When lysates from 293 cells expressing Myc-SMN missense mutations were size-fractionated through the same Superose column, the elution profiles of Myc-tagged S262I, T274I, G275S, and G279V mutants all showed major complexes exceeding 40,000 kDa in size (Fig. 6B). The Myc-SMN-Y272C missense mutant retained some ability to form protein complexes exceeding 40,000 kDa. However, there are more abundant smaller Y272C-containing complexes, which peaked around 700 kDa (Fig. 6B). In comparison with the wild-type SMN, we suspect that protein supra-complexes containing the Y272C missense mutation may be less stable and easier to dissociate into smaller protein complexes under our experimental conditions. DISCUSSION Based on results from this and other studies, it seems that SMN can be a component of large protein complexes that are found both in cytoplasmic and nuclear compartments (6). Cytoplasmic SMN plays an essential role in spliceosomal snRNP biogenesis and is required for the transport of the snRNP complex into the nucleus (21). In the nucleus, SMN protein is required for regenerating an active splicing complex (22). SMN protein has been reported to associate with the nuclear tran-scription activator E2 of papillomavirus (23), the DEAD box protein dp103 (also called GEMIN 3) (24,25), the SIP-1 (SMNinteracting protein-1) (26), the small nucleolar RNA-associated protein fibrillarin (27), GAR1 (28), the zinc-finger protein ZPR1 (29), the Ewing's sarcoma protein EWS (30), and RNA polymerase II (31). Finally, there is evidence that SMN may also be involved in pro-and anti-apoptotic pathways (32,33).
SMN is an essential gene in diverse organisms. A reduction of the SMN protein because of deletion and/or mutation in the telomeric SMN causes SMA. Studies on SMA have suggested that motor neuron cells, for reasons still not clear, are more sensitive than other cell types to an insufficient amount of SMN protein. It is not known which function(s) carried out by SMN is especially critical to the survival of motor neurons.
In this work, we have studied protein interactions between SMN and mSin3, a transcription corepressor involved in histone deacetylation and chromatin remodeling. We have found that interaction of SMN with mSin3 is dependent upon an intact exon 6 region of SMN. Interestingly, an earlier study identified a small, modular 30 amino acid sequence between amino acids 249 -278, which corresponds to the region encoded by exon 6 of SMN, as a domain that mediates SMN selfoligomerization (34). In our studies, we mapped the same region as the one that mediates SMN-mSin3A interaction as well as reporter gene repression. In addition, we have shown that the exon 6 region is critical to higher order complex formation by SMN protein. These results indicate that, aside from selfoligomerization, this exon 6-encoded domain is involved in mediating the interactions between SMN and other proteins within cells.
The five missense mutations tested in this study are within or immediately adjacent to exon 6 of SMN. The Y272C missense mutation is found preferentially in patients with type I SMA, and this same missense mutation was reported to reduce the ability of SMN to self-oligomerize (34). Our findings from this study suggest that SMN-mSin3A interaction may be impaired by the Y272C missense mutation; size-fractionation of Y272C lysate showed an increase in smaller Y272C-containing complexes as compared with wild-type SMN. Though derepression of different luciferase reporters by the Y272C construct was not demonstrated, this may reflect the difficulty in correlating the SMA-associated missense mutations with aberrant cellular function using non-motor neuron cells. It should be noted that a previous study (20) also encountered a similar difficulty in differentiating SMN missense mutants from the wild-type protein in a proliferation assay of an SMN-depleted chicken pre-B cell line. In the same study, the effect of exon 6 deletion was dramatically different, as the SMN-⌬6 mutant was no longer able to support the survival of the SMN-depleted cells.
The telomeric SMN gene differs from the centromeric SMN gene by a single nucleotide, which results in expression of centromeric SMN transcripts that typically lack exon 7. The exon 7-deficient protein product of the centromeric SMN gene is unstable and rapidly degraded (1,4). In our transient transfection studies, we successfully over-expressed SMN mutant that harbors deletion of the exon 7. It is likely that this is because of the stabilizing effect of the Myc-epitope tag on the SMN moiety.
Recent studies have demonstrated that the transcription coactivator CBP is trapped by polyglutamine repeats and that the resultant CBP (a histone acetyltransferase)-depletion may be linked to the neurodegenerative phenotype of Huntington's disease (35). On the other hand, our results suggest that SMN protein may functionally interact with the transcription corepressor mSin3 and its associated histone deacetylases; SMN-⌬6 deletion completely abolishes this interaction, whereas the Y272C missense mutation may impair it. In addition to 293 cells, we have carried out the same experiments in other types of cells and obtained similar results (data not shown). Although further experiments are needed, we postulate that SMN may be involved in transcriptional repression of critical genes in motor neuron cells of the spinal cord, possibly by means of mSin3-associated histone modification enzymes.