Abstract
Spinal muscular atrophy is caused by reduced levels of SMN resulting from the loss of SMN1 and reliance on SMN2 for the production of SMN. Loss of SMN entirely is embryonic lethal in mammals. There are several SMN missense mutations found in humans. These alleles do not show partial function in the absence of wild-type SMN and cannot rescue a null Smn allele in mice. However, these human SMN missense allele transgenes can rescue a null Smn allele when SMN2 is present. We find that the N- and C-terminal regions constitute two independent domains of SMN that can be separated genetically and undergo intragenic complementation. These SMN protein heteromers restore snRNP assembly of Sm proteins onto snRNA and completely rescue both survival of Smn null mice and motor neuron electrophysiology demonstrating that the essential functional unit of SMN is the oligomer.
Introduction
Spinal muscular atrophy (SMA) has an incidence of approximately 1/10 000 and without therapeutic intervention is the leading genetic cause of infant death (1–4). SMA is caused by a deletion or mutation in the survival motor neuron 1 gene (SMN1) and retention of the SMN2 gene (5,6). SMN1 and SMN2 essentially differ by a single critical nucleotide that alters a splicing regulator resulting in the lack of exon 7 from the majority of SMN2 transcripts (7–11). The loss of exon 7 results in a SMN protein that does not oligomerize efficiently and is rapidly degraded (12–14). Some full-length SMN is produced by SMN2 but the low amount of SMN is insufficient and results in SMA (15,16). The complete absence of SMN in mice and humans is embryonic lethal (6,17,18). The severity of SMA is inversely correlated with SMN2 copy number with more functional copies of SMN2 resulting in a milder phenotype (18–21). Some of the mutations that give rise to SMA are small insertions, deletions and occasionally missense mutations in SMN1 (6,22–25). Missense mutations can be classified as mild or severe by whether they result in a mild or severe SMA phenotype in the presence of 1 or 2 copies of SMN2 (6). For example, SMNA2G and SMNT274I give are associated with mild type 3 SMA when 1 copy of SMN2 is present (25).
Diagram of the human SMN1 gene indicating the location of SMN missense mutations. Protein domains are indicated relative to their coding exons. The position of amino acids is indicated below each exon. SMNA111G is located in the Tudor domain but is not predicted to disrupt Sm binding. SMNT274I is located in the YG box; however, it does not disrupt oligomerization. Q282A is not located in any particular protein domain.
SMN is believed to function as an oligomeric protein that is required for snRNP assembly (13). Mutations in the C-terminal domain lie on the inner surface of the YG box and block oligomerization (26,27). The Tudor domain is reported to bind the arginine and glycine (RG) tail of Sm proteins and severe mutations disrupt this ability (23,28). Mutations in either the YG box or the Tudor domain disrupt the ability of these alleles to perform snRNP assembly (13). The mild mutation SMNT274I lies on the outer surface of the YG box and as such does not disrupt oligomerization (26). The YG box promotes the formation of oligomers composed of two tetramers. SMN functions in the assembly of the Sm ring on snRNAs and the Sm/Lsm ring on U7 (13,29–31). It has also been suggested that SMN functions in the assembly of RNP complexes that are important for mRNA transport and translation in the axon, although no direct biochemical assay of this activity exists (32,33).
Previously we examined human SMN missense mutation transgenes in the presence and absence of SMN2 in mice. In all cases, the mild missense mutations SMNA2G, SMND44V, SMNA111G, SMNT274I and SMNQ282A did not rescue Smn null mice which lack wild-type SMN (34–36). Conversely, all of the mutations were able to rescue Smn null mice when SMN2 was present. This indicates that homomeric complexes of each mild SMN mutation by themselves, in the absence of full-length wild-type SMN protein, are not functional. Instead, it is the heteromer formed between monomers harboring a SMN missense mutation and the wild-type SMN monomers produced by SMN2 that results in a functional complex. In fact, no SMN missense mutations have been reported in humans in the absence of SMN2. As the SMN2 gene has been found to be absent in 10–15% of the normal population, one would expect to find these SMA patients if the missense alleles had partial function (20).
In the current study, we show that the mild N-terminal mutation SMNA111G can undergo intragenic complementation with the C-terminal mutations SMNT274I or SMNQ282A to form a fully functional SMN complex. Although these SMN missense alleles have no function in the absence of wild-type full-length SMN protein; here we show that these N- and C-terminal mutant SMN alleles can complement and completely rescue weight gain and survival of Smn null mice lacking wild-type SMN protein. Additionally, normal levels of snRNP assembly were measured in spinal cord tissue extracts in these mice. Finally, motor neuron function is completely recovered in adult animals as normal compound muscle action potentials (CMAP) and motor unit number estimations (MUNE) were observed. These findings support the conclusion that the functional unit of the SMN complex is the oligomer. Furthermore, a mutant domain in one oligomer juxtaposed to a wild-type domain in another oligomer results in a functional molecule. The complete rescue of both survival and motor neuron electrophysiology in mice suggests that snRNP assembly activity is the critical function of the SMN complex in SMA.
Results
Survival of Smn null mice containing two missense mutations
We have previously reported the creation of a series of transgenic mice lines expressing mild SMN missense alleles in the N-terminus (SMNA2G, SMND44V, SMNA111G) as well as C-terminus (SMNT274I and SMNQ282A) (34–36) (Fig. 1). In every case, the mild SMN missense mutations were not able to rescue Smn null mice (Smn−/−). Thus, we have established that mild and severe SMN alleles are not functional in mice in the absence of full-length wild-type human SMN protein (34–36). Conversely, we found that all mild SMN missense alleles could rescue the weight gain and survival of Smn null mice when the SMN2 gene was present. These findings suggest that some amount of wild-type SMN protein must be present for mild missense alleles to function in the mouse.
In this study, we sought to determine if an N-terminal and a C-terminal mild missense mutation could undergo intragenic complementation and create a functional SMN complex in the absence of full-length wild-type SMN. We crossed mice containing the mouse Smn knockout allele and the N-terminal SMNA111G (Smn+/−; SMNA111G+/+) or the C-terminal SMNQ282A (Smn+/−; SMNQ282A+/+) alleles. The mice were then intercrossed so that alleles were homozygous (Smn−/−; SMNA111G+/+; SMNQ282A+/+). Specific probes were used with droplet digital polymerase chain reaction (ddPCR) to confirm homozygosity of the missense alleles. The same method was used to generate mice that were homozygous for the N-terminal SMNA111G and the C-terminal SMNT274I alleles in the Smn null background (Smn −/−; SMNA111G+/+; SMNT274I+/+). Remarkably, both combinations of N- and C-terminal missense mutations resulted in the complete rescue of the weight and survival of Smn null mice (Fig. 2). There was no statistically significant difference in weight between the Smn −/−; SMNA111G+/+; SMNQ282A+/+ mice and Smn heterozygous littermate controls (Smn+/−; SMNA111G+/+; SMNQ282A+/+) (P = 0.46) (Fig. 2A). The survival of Smn −/−; SMNA111G+/+; SMNQ282A+/+ mice was identical to Smn heterozygous littermate controls (Fig. 2B). Mice lived for more than 250 days with no hindlimb weakness that is typically observed for Smn−/−; SMN2+/+; SMNΔ7+/+ mice. When mice were observed at 250 days, it was not possible to tell which mice were rescued (Smn−/−) and which were heterozygous (Smn+/−) by phenotype. This contrasts the ∆7SMA animals (Smn−/−; SMN2+/+; SMN∆7), which live for 14 days and display overt weakness of the hindlimbs (37). Similarly, the Smn−/−; SMNA111G+/+; SMNT274I+/+ mice were no different in weight (P = 0.83) (Fig. 2C) or survival (Fig. 2D) when compared to Smn heterozygous littermates (Smn+/−; SMNA111G+/+; SMNT274I+/+). The Smn−/−; SMNA111G+/+; SMNQ282A+/+ and Smn−/−; SMNA111G +/+; SMNT274I+/+ lines were maintained as viable, fertile breeding pairs that produce offspring that are all null for Smn. At 100 days of age, mice were weighed prior to electrophysiological assessment. Again, we found no difference in weight between the Smn−/−; SMNA111G+/+; SMNQ282A+/+ (25.4 ± 1.1 g, n = 5, P = 0.53) and Smn−/−; SMNA111G +/+; SMNT274I+/+ (28.5 ± 2.2 g, n = 5, P = 0.07) lines and their Smn heterozygous littermate controls (23.1 ± 1.7 g, n = 6).
The N- and C-terminal mild missense mutations completely rescue the weight and survival of Smn null mice. (A, B) SMNA111G and SMNQ282A complement to rescue the weight and survival of Smn null mice (Smn−/−; SMNA111G+/+; SMNQ282A+/+ n = 8). (C, D) Similarly, SMNA111G and SMNT27I complement to rescue weight and survival of Smn null mice (Smn−/−; SMNA111G+/+; SMNT274I+/+ n = 17). The weight of mice for each line is not statistically different from control littermates (Smn+/−; SMNA111G+/+; SMNQ282A+/+ P = 0.46) or (Smn+/−; SMNA111G+/+; SMNT274I+/+ P = 0.83). (B, D) Survival is completely rescued with mice living more than 250 days. The line representing either A111G/T274I or A111G/Q282A is located on top of the line representing the controls in each of the graphs. For comparison, the ∆7SMA mice survive an average of 14.1 ± 0.1 days (Smn−/−; SMN2+/+; SMNΔ7+/+ n = 86). Statistical test: compare growth curve function of the R-Package (Statmod) and Kaplan–Meier with a log-rank test.
Previously we demonstrated that two N-terminal mild missense mutations, SMNA111G with SMNA2G in the absence of SMN2 do not rescue Smn null mice (34,35). Here we tested the two C-terminal missense mutations, SMNT274I together with SMNQ282A, and found that this combination was not able to rescue Smn null mice (Table 1). In crosses that were homozygous for each missense allele and heterozygous for the mouse Smn allele (Smn+/−; SMNQ282A+/+; SMNT274I+/+), no rescue mice were observed on the day of birth. Smaller litters were observed for Smn+/−; SMNQ282A+/+; SMNT274I+/+ crosses (6.3 ± 0.5 pups, n = 12 litters) when compared to the SMN ∆7 model (7.2 ± 0.1 pups, n = 200 litters), and FVB/N mice (Stock No: 001800) (7.3 pups per litter) (38). We did not determine if the mice were viable at any embryonic stage of development.
The two mild C-terminal missense mutations SMNT274I and SMNQ282A do not rescue Smn null mice. We measured the survival of SMNT274I+/+; SMNQ282A+/+ mice in the presence and absence of Smn. A total of 18 crosses resulting in 75 viable pups were genotyped on the day of birth. No mice with the Smn−/−; SMNT274I+/+; SMNQ282A+/+ genotype were obtained (χ2 = 21.5, P < 0.0001). Statistical analysis was performed with Fisher’s Exact chi-square test with 5 degrees of freedom
| C-terminal mutations SMNT274I/SMNQ282A do not rescue Smn null mice | ||
|---|---|---|
| Genotype | Expected | Observed |
| Smn +/−; SMNT274I+/+; SMNQ282A+/+ | 37.5 | 51 |
| Smn +/+; SMNT274I+/+; SMNQ282A+/+ | 18.75 | 24 |
| Smn −/−; SMNT274I+/+; SMNQ282A+/+ | 18.75 | 0 |
| n = 75 | ||
| C-terminal mutations SMNT274I/SMNQ282A do not rescue Smn null mice | ||
|---|---|---|
| Genotype | Expected | Observed |
| Smn +/−; SMNT274I+/+; SMNQ282A+/+ | 37.5 | 51 |
| Smn +/+; SMNT274I+/+; SMNQ282A+/+ | 18.75 | 24 |
| Smn −/−; SMNT274I+/+; SMNQ282A+/+ | 18.75 | 0 |
| n = 75 | ||
The two mild C-terminal missense mutations SMNT274I and SMNQ282A do not rescue Smn null mice. We measured the survival of SMNT274I+/+; SMNQ282A+/+ mice in the presence and absence of Smn. A total of 18 crosses resulting in 75 viable pups were genotyped on the day of birth. No mice with the Smn−/−; SMNT274I+/+; SMNQ282A+/+ genotype were obtained (χ2 = 21.5, P < 0.0001). Statistical analysis was performed with Fisher’s Exact chi-square test with 5 degrees of freedom
| C-terminal mutations SMNT274I/SMNQ282A do not rescue Smn null mice | ||
|---|---|---|
| Genotype | Expected | Observed |
| Smn +/−; SMNT274I+/+; SMNQ282A+/+ | 37.5 | 51 |
| Smn +/+; SMNT274I+/+; SMNQ282A+/+ | 18.75 | 24 |
| Smn −/−; SMNT274I+/+; SMNQ282A+/+ | 18.75 | 0 |
| n = 75 | ||
| C-terminal mutations SMNT274I/SMNQ282A do not rescue Smn null mice | ||
|---|---|---|
| Genotype | Expected | Observed |
| Smn +/−; SMNT274I+/+; SMNQ282A+/+ | 37.5 | 51 |
| Smn +/+; SMNT274I+/+; SMNQ282A+/+ | 18.75 | 24 |
| Smn −/−; SMNT274I+/+; SMNQ282A+/+ | 18.75 | 0 |
| n = 75 | ||
Analysis of SMN mRNA and protein expression in Smn null mice containing two missense mutations
We performed RT-PCR on spinal cord tissue collected at postnatal day 5 and used ddPCR to determine the amount of human SMN mRNA produced. Both transgenes were homozygous and were on the Smn −/− background (Smn−/−; SMNA111G +/+; SMNT274I+/+ and Smn−/−; SMNA111G+/+; SMNQ282A+/+) The amount of transcript was compared to a ∆7SMA mouse (Smn−/−; SMN2+/+; SMN∆7) as a Smn−/− mouse is not viable without SMN2. The quantification of SMN mRNA expression for each of the mouse lines used in this study is shown in Figure 3A. The primer/probe set used detects only full-length SMN mRNA and does not detect mouse Smn or SMN∆7 mRNA. Expression is measured relative to the housekeeping gene YWHAZ12. The Smn−/−; SMNA111G+/+; SMNQ282A+/+ mice had a ~ 15-fold increase in SMN expression (353.8 ± 50.7, n = 3) and the Smn−/−; SMNA111G+/+; SMNT274I+/+mice (304.3 ± 27.1, n = 3) had ~ 13-fold increase in SMN expression when compared to ∆7SMA mice (24.3 ± 6.9, n = 3). Thus, there was a marked increase in the expression of full-length human SMN mRNA when compared to SMN∆7 SMA mice with two copies of SMN2.
Expression of SMN cDNA and protein in the spinal cord in Smn null mice containing two missense mutations at postnatal day 5. (A) Full-length SMN cDNA expression is significantly increased in both Smn−/−; SMNA111G+/+; SMNQ282A+/+ and Smn−/−; SMNA111G+/+; SMNT274I+/+ lines as compared to ∆7SMA mice. Spinal cord tissue from Smn−/−; SMNA111G+/+; SMNQ282A+/+ mice displayed a ~ 15-fold increase in SMN expression and Smn−/−; SMNA111G+/+; SMNT274I+/+ had an ~ 13-fold increase when compared to the expression from two copies of SMN2 in ∆7SMA mice (n = 3 mice for each group). SMN cDNA expression was measured relative to YWHAZ12 expression. Probes were specific to human FL-SMN. (B) Quantification of total SMN protein in spinal cord tissue measured by ELISA indicates a ~ 1.5 fold increase in protein expressed in the N- and C- terminal mild SMN missense mutations in Smn null mice (n = 4 mice in each group) as compared to the total SMN protein in wild-type Smn ∆7 control (Smn+/+; SMN2+/+; SMNΔ7+/+) spinal cord tissue (n = 3). (Error bars = SEM, ***P < 0.001, **P < 0.01). Statistical test: one-way ANOVA
The combined SMN levels in Smn−/−; SMNA111G+/+; SMNQ282A+/+ and Smn−/−; SMNA111G+/+; SMNT274I+/+ mice was measured by enzyme-linked immunosorbent assay (ELISA) (36) (Fig. 3B). There was a greater than 1.5-fold increase in total SMN protein expression in the spinal cord of Smn−/−; SMNA111G+/+; SMNQ282A+/+ (42.0 ± 1.2 ng/mg, n = 4) and in Smn−/−; SMNA111G+/+; SMNT274I+/+ mice (45.5 ± 1.2 ng/mg, n = 4) when compared to wild-type Smn ∆7 controls (Smn+/+; SMN2+/+; SMNΔ7+/+) (24.8 ± 1.5 ng/mg, n = 3). Thus, the amount of SMN protein in spinal cord tissue expressed by the N- and C- terminal mild SMN missense mutations in Smn null mice is greater than that of wild-type Smn ∆7 control mice (Smn+/+; SMN2+/+; SMNΔ7+/+).
Analysis of snRNP assembly in Smn null mice containing two missense mutations
The most well characterized function of the SMN complex is snRNP assembly which can be readily assayed in spinal cord tissue from mice. We determined the snRNP assembly activity in Smn null mice containing the mild missense mutation transgenes. Spinal cord tissue from Smn−/−; SMNA111G+/+; SMNQ282A+/+ and Smn−/−; SMNA111G+/+; SMNT274I+/+ mice were assessed at 5 days of age (Fig. 4). Representative western blots of spinal cord tissue show the amount of protein isolated for snRNP analysis (Fig. 4A, D). The amount of SMN protein used in the assay was measured relative to the PRMT5 control and the protein measured in wild-type Smn ∆7 control mice (Smn+/+; SMN2+/+; SMNΔ7+/+) was set to 1 (Fig. 4B, E). We found that in the spinal cord the Smn−/−; SMNA111G+/+; SMNQ282A+/+ mice (0.96 ± 0.04, n = 4), which are no different from wild-type Smn ∆7 controls (Smn+/+; SMN2+/+; SMNΔ7+/+, n = 4), produced 5 times more protein than ∆7SMA mice (Smn−/−; SMN2+/+; SMNΔ7+/+) (0.21 ± 0.01, n = 3) (Fig. 4A). Furthermore, snRNP assembly in Smn−/−; SMNA111G+/+; SMNQ282A+/+ mice (0.73 ± 0.13, n = 4) was 15-fold higher than that of ∆7SMA mice (0.05 ± 0.02, n = 4) and was nearly restored to that of wild-type Smn ∆7 controls (Smn+/+; SMN2+/+; SMNΔ7+/+) (n = 3) (Fig. 4B). The activity of snRNP assembly in the wild-type Smn ∆7 control mice (Smn+/+; SMN2+/+; SMNΔ7+/+) was set at 1 and all other values were calculated accordingly. Similarly, Smn−/−; SMNA111G+/+; SMNT274I+/+mice (0.98 ± 0.08, n = 3) produced 5-fold more protein than ∆7SMA mice (0.21 ± 0.01, n = 3) (Fig. 4C) and snRNP assembly was also increased by more than 15-fold in Smn−/−; SMNA111G+/+; SMNT274I+/+mice (0.86 ± 0.15, n = 3) when compared to ∆7SMA mice (0.05 ± 0.01, n = 3) (Fig. 4D). Thus, the N- and C-terminal mild SMN missense mutations complement to rescue protein expression and snRNP assembly in Smn null mice.
The N- and C-terminal mild SMN mutations complement to rescue protein expression and snRNP assembly in Smn null mice. (A, D) Representative western blots from total spinal cord isolated at postnatal day 5. The amount of total SMN protein (human and mouse) was measured relative to PRMT5. (B, E) SMN protein isolated for snRNP analysis from spinal cord. The protein isolated wild-type Smn ∆7 controls (Smn+/+; SMN2+/+; SMNΔ7+/+) was set to 1 and all other concentrations adjusted accordingly. Both the Smn−/−; SMNA111G+/+; SMNQ282A+/+ and Smn−/−; SMNA111G+/+; SMNT274I+/+ mice produced 5-fold more SMN protein than ∆7SMA mice (Smn−/−; SMN2+/+; SMNΔ7+/+). (C) snRNP assembly in Smn−/−; SMNA111G+/+; SMNQ282A+/+ mice (n = 4) and (F) and Smn−/−; SMNA111G+/+; SMNT274I+/+ mice (n = 4) was more than 15-fold higher than that of ∆7SMA mice (n = 4), and nearly restored to the levels measured in wild-type Smn ∆7 controls (Smn+/+; SMN2+/+; SMNΔ7+/+) (n = 3). SnRNP assembly activity in wild-type Smn ∆7 controls (Smn+/+; SMN2+/+; SMNΔ7+/+) was set to 1 and all other levels adjusted accordingly. (Error bars = SEM, **P < 0.01). Statistical test: one-way ANOVA
Electrophysiological measurements of the motor unit in Smn null mice containing two missense mutations are normal
We have previously shown that the CMAP and MUNE are decreased in SMA mice (39). These are the same measurements used in the clinic to assess the motor neuron function of SMA patients (40). We measured CMAP and MUNE in our mice at 100 days of age when the neuromuscular system is fully mature. We found that the neuromuscular junction is fully functional in both Smn−/−; SMNA111G+/+; SMNQ282A+/+ and Smn−/−; SMNA111G+/+; SMNT274I+/+ mice as both CMAP and MUNE measurements were no different from heterozygous Smn ∆7 control animals (Smn+/−; SMN2+/+; ∆7SMN+/+) (Fig. 5). There was no statistical difference between mice with complementing alleles Smn−/−; SMNA111G+/+; SMNQ282A+/+ (39.0 ± 1.1 mV, n = 5), Smn−/−; SMNA111G+/+; SMNT274I+/+(39.3 ± 2.0 mV, n = 5) or heterozygous Smn ∆7 control animals (Smn+/−; SMN2+/+; ∆7SMN+/+) (40.9 ± 1.9 mV, n = 6) as determined by one-way ANOVA analysis (P = 0.72) (Fig. 5A). MUNE values were also completely rescued indicating the presence of a fully functional motor unit; there was no statistical difference between Smn−/−; SMNA111G+/+; SMNQ282A+/+ (323.8 ± 16.5, n = 5), Smn−/−; SMNA111G+/+; SMNT274I+/+(313.8 ± 23.5, n = 5) or heterozygous Smn ∆7 control animals (Smn+/−; SMN2+/+; ∆7SMN+/+) (346.8 ± 19.3, n = 6) as determined by one-way ANOVA analysis (P = 0.49) (Fig. 5B).
Motor unit function at postnatal day 100 is completely rescued by complementation of the N- and C-terminal mild SMN missense mutations in Smn null mice. There is no significant difference in the measures of (A) compound muscle action potential (CMAP) (P = 0.72) or (B) Motor unit number estimation (MUNE) (P = 0.49) in either Smn−/−; SMNA111G+/+; SMNQ282A+/+ (n = 5) or and Smn−/−; SMNA111G+/+; SMNT274I+/+ (n = 5) Smn null mice when compared to wild-type Smn ∆7 controls (Smn+/+; SMN2+/+; SMNΔ7+/+) (n = 6). SMA mice (Smn−/−; SMN2+/+; SMNΔ7+/+) could not be used as a control because they only survive for 14 days. (Error bars = SEM). Statistical test: one-way ANOVA
Discussion
SMA is caused by deletion or mutation of the SMN1 gene and retention of the SMN2 gene (5,6). The most common mutation in SMN1 is a large deletion that renders the gene non-functional (5,23,41). Smaller deletions that either disrupt the reading frame or create a premature stop codon have been identified in SMN1 (5,6). Rarely, missense mutations in the SMN1 gene also occur. SMN mutations with a single altered amino acid can provide insight into the function of different protein domains (6). The SMN1 missense mutations can be divided into two groups: mild and severe. This is based on the severity of the SMA phenotype in individuals with two copies of SMN2 and a missense mutation. Typically an individual with a homozygous SMN1 deletion and two copies of SMN2 has type 1 SMA. The presence of a mild missense mutation, with two copies of SMN2, results in type 3 SMA. In contrast, a severe missense mutation results in type 1 SMA in the presence of two copies of SMN2 (6). In SMA patients, missense mutations in SMN1 are always found with a deletion in the other SMN1 allele in the presence of two or more copies of SMN2. This means that a small amount of wild-type full-length SMN, produced by SMN2, is always present in humans which can oligomerize with protein produced by the single SMN missense allele. The Smn null background in the mouse allows us the unique opportunity to assay the function of the SMN missense alleles, including survival, motor neuron function and snRNP assembly, in the complete absence of wild-type SMN.
SMN is an oligomeric protein thought to be comprised of two tetramers to create an octomer (27). Many of the missense mutations have been examined in vitro using biochemical assays to study the oligomerization of SMN monomers. Several severe missense mutations in the C-terminus, including SMNY272C, SMNG279V and M263R, have been reported to disrupt SMN oligomerization (12,26,27). In fact, an individual with the SMNY272C mutation presented with SMN levels that were the same as a typical type 1 SMA patient, revealing the destabilizing effect of this mutation that results in SMN degradation (14,16). The YG box in exon 6 has been shown to be essential for SMN oligomerization. A network of tyrosine-glycine packing between helices drives the formation of SMN oligomers. However, the presence of a severe SMN missense allele disrupts oligomerization and results in monomeric SMN (27). Yet, not all mutations in the YG box disrupt oligomerization. Interestingly, the mild SMN missense mutation SMNT274I is predicted to be on the outside of the YG box, and as such should not disrupt oligomerization of SMN (27).
Previously, we have shown that mild missense alleles do not rescue Smn null mice unless full-length SMN produced by SMN2 is present. These studies demonstrated that homomeric complexes of missense mutations are not functional in the absence of full-length SMN (34–36). However, we predicted that if two missense alleles with mutations in different domains could oligomerize, then a functional SMN complex would result. Specifically, the non-mutated portions of each protein would compensate for the mutated amino acid to form a functional oligomeric complex. Here we have demonstrated that this in fact is the case. The N-terminal mutation SMNA111G can complement the C-terminal mutation SMNT274I to form a functional SMN complex that rescues weight, survival and electrophysiological measurements in the Smn null mouse in the complete absence of wild-type SMN. This phenomenon is known as intragenic complementation (Fig. 6). Similarly, the C-terminal mutation SMNQ282A can also complement SMNA111G and again rescue weight, survival and electrophysiological measurements in the absence of wild-type SMN protein. We would predict that in humans, two mild missense mutations, in the absence of wild-type SMN, could give rise to a normal phenotype. These individuals would not be detected as they would not develop SMA.
Intragenic complementation. The N- and C- terminal mild SMN missense mutations Smn−/−; SMNA111G+/+; SMNQ282A+/+ or Smn−/−; SMNA111G+/+; SMNT274I+/+ form a heteromeric complex that is fully functional in the absence of wild-type SMN protein. Both the snRNP assembly, the essential biochemical function of SMN, and the survival of Smn null mice was rescued in the presence of these complexes. However, heteromeric complexes comprised of two C-terminal mild SMN mutations, Smn−/−; SMNQ282A+/+; SMNT274I+/+ did not produce viable mice in the absence of wild-type SMN. None of the SMN missense mutations in this study affect SMN oligomerization. As such the oligomers depicted here have equal ability to efficiently oligomerize. We propose that the N- and C-terminal domains of SMN possess different essential functions that are not reconstituted in oligomers containing two C-terminal missense mutations.
Furthermore, we have shown that complementation does not occur with all mild missense mutations. The two C-terminal mutations, SMNT274I and SMNQ282A, cannot form a functional complex as we were unable to recover any Smn null mice. In this case, we predict that both missense alleles disrupt the same function, and therefore are incapable of complementing one another. It is remarkable that two different missense alleles, which have no function on their own, can oligomerize to create a functional complex in the complete absence of wild-type full-length SMN. This study is the first to demonstrate in vivo that the functional unit of SMN is the oligomeric complex.
The best characterized functions of SMN are the assembly of the UsnRNA for the splicing snRNPs and the assembly of the U7 Sm/Lsm ring for the processing of maturing histone mRNA. snRNP and Sm protein assembly can be directly measured in mouse spinal cord tissue. In addition the reaction components needed to assemble the Sm ring onto snRNA in vivo have been defined (42). Many additional functions of SMN have been proposed, such as local translation and axonal transport, yet the role of SMN has not been mechanistically defined. At this time quantifying the axonal transport of RNAs in the motor neuron in vivo is not feasible. The altered transport of mRNA has been demonstrated however when using transfected reporter constructs with specific modified RNAs in cell culture (33).
Here we show that the snRNP assembly is completely restored when two missense mutations are combined. Although it is possible that a combination of two mutations could restore multiple functions to SMN it seems unlikely that a Tudor domain, which is essential for Sm transfer, and the mutations in the C-terminus of SMN could combine to assemble an essentially completely different set of proteins. It is possible that the rescue of snRNP assembly could in turn rescue the splicing of downstream components required for axonal transport. This would indicate that SMN’s role in transport is dependent upon restoration of snRNP assembly. We have also shown normal electrophysiological activity as measured by MUNE and CMAP in rescued mice. These measurements are of critical importance as this assessment provides the functional read out of the motor neuron. We suggest that Sm assembly is the critical function of SMN both for rescue of lethality and rescue of the SMA phenotype. Furthermore, the complete rescue of both survival and motor neuron electrophysiology in mice is consistent with the loss of a central SMN role in snRNP assembly as critical to SMA pathogenesis. This does not, however, rule out the potential pathogenic importance of additional functions for the reconstituted SMN heteromer in, for example, maintenance presynaptic NMJ mRNA pools.
Intergenic complementation has been previously demonstrated in oligomeric enzymes when the interaction between two mutant protein subunits forms an active site. Typically examples of intragenic complementation, which are well documented in C elegans, involve the reconstitution of a catalytic site (43). However there is no known catalytic site associated with SMN per se. These studies demonstrate an unique example of intragenic complementation where the independent functions of the protein complex are not yet defined.
There are various human conditions that result from intragenic complementation in multimeric enzymes such as Propionic Acidemia, a metabolic disorder characterized by deficiency of propionyl CoA carboxylase (44,45), and Argininosuccinic Aciduria (ASA), an autosomal recessive disorder in which a mutation in the tetrameric protein argininosuccinate lyase (ASL) causes the accumulation of argininosuccinic acid in the blood and urine (46). In both cases, the interaction of two mutant protein subunits creates a normal active site in the heteromeric protein (47). In addition, the ASL tetramer is similar to SMN in that the homomeric mutant complexes are not functional. It is the creation of the active site in the heteromeric tetramer that results in ASL activity (48). The SMN complex can also be viewed as an enzyme, as the assembly of the Sm proteins onto snRNA, which occurs in an ATP-dependent manner, is its best characterized function (6,13).
Another intriguing example of intergenic complementation occurs in the neuromuscular disease Limb Girdle Muscular Dystrophy (LGMD2A) (49). In this case, the activity of the muscle protease calpain (CAPN3) can be reconstituted by combining two mutated autolytic fragments. One mutated autolytic fragment, the G222R mutation, lies in the N-terminal protease core domain. As a result, this mutation causes decreased protease activity and a severe phenotype when homozygous. The other mutation, R748Q, that lies in the C-terminal penta-EF hand (PEF)(L) dimerization domain, results in accelerated autolysis and again a severe phenotype (49–51). Together these two N- and C-terminal mutations are able to reconstitute CAPN3 activity and permit autolytic cleavage. This situation is quite similar to the functional SMN complex formed with N- and C-terminal missense mutations in which the measurement of snRNP activity demonstrates the recovery of function. Thus, there is precedence for the occurrence of complementation in oligomeric proteins. Yet the amount of recovery of SMN function is nearly to normal levels upon complementation unlike the restoration of only one of the four active sites in ASL. It appears that having a wild-type domain opposite a mutant domain in the SMN oligomer results in remarkable recovery of activity. This could also be due to the fact that the SMN oligomer is an octomer, composed of two tetramers, that loads Sm proteins onto a single snRNA at a time (27). Thus any one of the monomers containing the wild-type domain can perform that domain’s function and as such the resultant oligomeric protein retains activity.
The complementation of C-terminal and N-terminal SMN missense mutations indicates that the essential function of SMN is performed by the oligomer. Furthermore, these findings predict that two independent domains of SMN are both required for the function of SMN and also act independently to result in a high degree of functional recovery. We suggest that SMN’s function in assembling the Sm ring is the essential activity and this imparts the ability to rescue both the survival and SMA phenotypes in mice, The identification of a suppressor of a non-functional SMN missense allele that restores motor neuron function and snRNP assembly in the mouse could further confirm the critical role of SMN in SMA. For example, a suppressor in a Sm protein could restore the ability of SMN to rescue survival of the mouse but not the function of the motor neuron. This would determine if SMN plays a different role other than snRNP assembly in the motor neuron. Alternatively both survival and motor neuron function could be rescued showing that SMN’s role in assembling the Sm ring is essential for both activities.
Even in the current era of therapeutics for SMA, it is still unclear why low levels of ubiquitously expressed SMN cause motor neuron loss (52–54). Utilizing mutations previously identified in patients can help further our understanding the function of the SMN complex. Here we have demonstrated that the oligomer is the essential functional unit of SMN, both for the rescue of survival and for the development of SMA. In addition, the SMN protein has at least two independent domains that can be genetically separated. Although SMN is often proposed to have many functions, only two have direct biochemical assays: assembly of snRNPs for splicing and assembly of the U7 snRNP for processing of histones (6). Here we show that snRNP assembly is fully restored upon complementation, suggesting that instead of the multiplicity of proposed functions, SMN’s role in snRNP assembly is essential to both cell survival and function of the motor neuron.
Materials and Methods
Breeding and genotyping of transgenic mice
Mice containing SMN2+/+, Smn+/− and human SMN missense allele transgenes (36) were crossed to mice containing only the Smn knockout allele (Smn+/−, Stock No: 006214). They were backcrossed until SMN2 was eliminated, as confirmed by genotyping tail DNA as previously described (55). Pups were tattooed and tail snips collected on the day of birth. Tails were dissociated in 180 μl 50 mM NaOH at 95°C for 10 min and neutralized with 20 μl of 1 M Tris–HCl pH 8.0 as previously described (56). The Smn knockout allele was genotyped in a multiplex PCR reaction with SmnF 5′GATGATTCTGACATTTGGGATG, SmnR 5′ACCGTTCTTTAGAGCATGCTATG and BgalR 5′AACAAACGGCGGATTGAC. The mouse Smn wild-type allele yields a 325 bp product while the mouse knockout allele is 410 bp. Mice containing the Smn knockout allele (Smn+/−) and the SMN missense allele were intercrossed to generate males and females that were homozygous for each SMN missense allele. The mouse lines have been deposited in Jackson Laboratories: Smn−/−; SMNA111G+/+; SMNQ282A+/+, Stock No: 031909 and Smn−/−; SMNA111G+/+; SMNT274I+/+, Stock No: 031906.
Each SMN missense allele was genotyped with a specific LNA probe using ddPCR on the BioRad QX-200. A non-extending blocking oligo was added to each reaction to prevent non-specific amplification of the wild-type allele or the other missense allele present in the mouse. SMNA111G was detected with A111G_FP 5′TGCCATTTGGTCAGAAGACG, A111G_RP 5′CTCCTCTCTATTTCCATATCCAGTG, A111G-FAM 5′FAM-ATGGT+A+C+CTGGGTAA and A111G_block-PHO 5′TTAAAATCAATTGAAGCAATGGTAGCTGGGTAAATG-PHO. The + symbol precedes the locked nucleotide. The reaction was incubated at 50°C for 2 min followed by 55 cycles of amplification (95°C 30s, 59°C 1 min, 72°C 30s). SMNT274I and SMNQ282A use the same primers and blocker but each have their own probe for identification as follows: T274I/Q282A-FP 5′GTATGTTAATTTCATGGTACATGAGTG, T274I/Q282A-RP 5′CCTTAATTTAAGGAATGTGAGCAC, T274I-Q282A-block-PHO 5′TGATTTTGTCTGAAACCCATATAATAGCCAGTATGATA-PHO with T274I_FAM 5′FAM-ATAGCC+A+A+TATGATA to detect SMNT274I or Q282A-FAM 5′FAM-TGATT+T+G+C+TCTGAAAC to detect SMNQ282A. Each ddPCR reaction was multiplexed with Smn-intron1-FP 5′CTGTGTGACTGTGAGGGGATGTG, Smn-intron1-RP 5′CCTGTGAACATCTTCATCCTGACCTAA and Smn-intron1-HEX 5′HEX-AGGCTGGCTGAAGCAAGGCAACCAGATA as a two-copy control. The primers and probe bind to a region of Smn that is present in Smn+/+ and Smn−/− mice. All DNA samples were digested with BglII to linearize the DNA and separate the tandem insertions of each SMN missense transgene. This allows for an accurate copy number determination indicating homozygosity and heterozygosity of the SMN missense transgenes.
Weight and survival measurements of Smn null mice containing two missense mutations
Mice were weighed and observed daily starting on the day of birth. Mice were weaned at 21–28 days of age and then weighed weekly. In this study, all procedures were performed in accordance with our Ohio State University Laboratory Animal Resource approved protocol (2008A0089) following all Institutional Animal Care and Use Committee (IACUC) guidelines. In this study n represents biological replicates.
Analysis of SMN mRNA expression in Smn null mice containing two missense mutations by RT-PCR
Spinal cord tissue was harvested on postnatal day 5 as previously described (57). RNA was isolated using TRIzol reagent (Invitrogen), purified with the RNeasy kit (Qiagen) and converted to cDNA using pdN6 primers (Sigma), RNaseOUT (Invitrogen), and AMV-RT enzyme (NEB) according to the manufacturer’s instructions. mRNA expression was measured in a multiplex ddPCR reaction using the BioRad QX200 system. Primers and probes detected full-length SMN expression: SMN-FL-FP 5′CAAAAAGAAGGAAGGTGCTCA, SMN-FL-RP 5′TCCAGATCTGTCTGATCGTTTC and SMN-FL-FAM probe 5′FAM-TTAAGGAGAAATGCTGGCATAGAGCAGCAC. SMN expression was normalized to the YWHAZ Tyrosine 3-Monooxygenase housekeeping gene: ywhaz-FP 5′AGGGTGACTACTACCGTTACTT, ywhaz-RP 5′TGCTTCTTGGTATGCTTGCT and ywhaz-HEX probe 5′HEX-AGGTTGCTGCTGGTGATGACAAGA.
Analysis of SMN protein expression in Smn null mice containing two missense mutations by ELISA
SMN protein was measured using an ELISA at PharmOptima (Portage, MI) as previously described (36,58,59). Briefly, using electrochemiluminescence (ECL) immunoassay based on Meso Scale Discovery technology, the mouse monoclonal antibody 2B1 (60) is used to capture and a rabbit polyclonal anti-SMN antibody (Protein Tech, Cat. No. 11708-1-AP) labeled with a SULFO-TAG™ is used for detection. A standard curve using recombinant SMN protein (Enzo Life Sciences, Cat. No. ADI-NBP-201-050) is generated for calibration and plates are read with the Meso Scale 6000 sector imager. Three biological replicates were assessed for each group.
Analysis of snRNP assembly in Smn null mice containing two missense mutations
In vitro snRNP assembly of tissue extracts was performed as previously described (43). Briefly, Protein A DynaBeads (Invitrogen 100.01D) were washed and conjugated to Y12 antibody (LS Bio B8621) for 1 h at 4°C. The beads were washed and resuspended in RSB-500-0.1% NP-40-heparin buffer. snRNP assembly was performed for 1 h at 30°C using biotinylated U4 or U4∆Sm snRNA. The reactions were immunoprecipitated with the Y12-conjugated dynabeads for 1 h at 30°C, washed and assembled snRNPs were detected with 1:10 000 dilution of NeutrAvidin-HRP, (Invitrogen A2664) for 1 h at 30°C. The beads were washed, resuspended in SuperSignal Fempto ELISA substrate (Pierce 37 075), and luminescence was measured on Tecan Infinite F200 using I-Control software.
Electrophysiology of motor neurons in Smn null mice containing two missense mutations
Electrophysiological measurements of CMAP, and MUNE were assessed at 100 days of age as described previously (39). Smn−/−, SMNA111G+/+; SMNQ282A+/+ and Smn−/−, SMNA111G+/+; SMNT274I+/+ mice (n = 5 in each group) were compared to heterozygous ∆7 control animals (Smn+/−; SMN2+/+; ∆7SMN+/+) (n = 6).
Statistical analyses
Significance of weight data was determined with the Compare Growth Curve function of the R-Package (Statmod) (61,62). Kaplan–Meier survival curves were performed with Sigma Plot and significance was determined with the log-rank test. Statistical analyses were performed with Prism as previously described for MUNE and CMAP (39). Chi-square analysis was performed using the Fisher Exact Probability Test using the Freeman–Halton extension (63) on a 2 × 3 contingency table with 5 degrees of freedom at vassarstats.net. All error bars are represented as SEM. In these studies, ‘n’ represents the number of biological replicates in each assay.
Acknowledgements
We wish to thank Phillip Zaworski at PharmOptima for the ELISA studies. We also thank Dr Louise Simard and Dr John McPherson for critical reading of the manuscript.
Conflict of Interest statement. None declared.
Funding
Cure SMA; the Marshall Heritage Foundation; The National Institute of Neurological Disorders and Stroke [R01 NS0385650]; and Miracles for Madison.
Authors' contributions
V.L.M., D.J.B., and A.H.M.B. designed research; V.L.M., K.M.C., W.D.A., S.I.D., C.C.I., A.M-L., E.W. and A.P. performed research; D.J.B. contributed new reagents/analytic tools; V.L.M., K.M.C., C.C.I., and A.H.M.B. analyzed data; and V.L.M. and A.H.M.B. wrote the paper.
