Spinal muscular atrophy (SMA) is caused by loss of the survival motor neuron gene (SMN1) and retention of the SMN2 gene. The copy number of SMN2 affects the amount of SMN protein produced and the severity of the SMA phenotype. While loss of mouse Smn is embryonic lethal, two copies of SMN2 prevents this embryonic lethality resulting in a mouse with severe SMA that dies 5 days after birth. Here we show that expression of full-length SMN under the prion promoter (PrP) rescues severe SMA mice. The PrP results in high levels of SMN in neurons at embryonic day 15. Mice homozygous for PrP-SMN with two copies of SMN2 and lacking mouse Smn survive for an average of 210 days and lumbar motor neuron root counts in these mice were normal. Expression of SMN solely in skeletal muscle using the human skeletal actin (HSA) promoter resulted in no improvement of the SMA phenotype or extension of survival. One HSA line displaying nerve expression of SMN did affect the SMA phenotype with mice living for an average of 160 days. Thus, we conclude that expression of full-length SMN in neurons can correct the severe SMA phenotype in mice. Furthermore, a small increase of SMN in neurons has a substantial impact on survival of SMA mice while high SMN levels in mature skeletal muscle alone has no impact.
Spinal muscular atrophy (SMA) is an autosomal recessive disorder characterized by loss of α motor neurons in the spinal cord and is the leading hereditary cause of infant mortality (1,2). SMA is caused by loss or mutation of the telomeric copy of survival motor neuron gene (SMN1) and retention of the SMN2 gene (3,4). The SMN2 gene differs from the SMN1 gene by a single nucleotide change in exon 7 (C to T transition) that does not change the amino acid sequence but does affect the incorporation of exon 7 into the final transcript (3,5,6). This change in exon 7 incorporation is due to disruption of an SF2/ASF binding site in an exon splice enhancer (7), and the creation of an exon splice silencer (8). Thus, the SMN1 gene produces full-length transcript whereas the majority of the SMN2 transcripts lack exon 7 (3,9,10). The protein product lacking exon 7 (SMNΔ7) oligomerizes less efficiently (11), appears unstable and is rapidly degraded (12). In severe SMA patients with limited SMN2 copy number the SMN protein level is reduced in all tissues examined (13,14), whereas in mild SMA patients SMN reduction is more apparent in the spinal cord (13).
In mice there is one survival motor neuron (Smn) gene that is equivalent to SMN1 (15,16). Complete loss of this gene results in embryonic lethality as would be expected given SMN’s essential role in snRNP assembly (17–19). Introduction of two copies of SMN2 rescues the embryonic lethal phenotype resulting in mice with severe SMA, whereas eight copies of SMN2 result in rescue (20–22). Thus, SMN2 can produce the functional SMN protein required for all cells including motor neurons.
Expression of SMNΔ7 under the SMN promoter in severe SMA mice modulates the phenotype with mice living 14 days on an average. This indicates that increased expression of SMNΔ7 is not detrimental (23). Expression of a mild SMA missense mutation (A2G) in severe SMA mice (SMN2+/+; Smn−/−; SMNA2G+/−) results in a mild SMA phenotype with mice living greater than 1 year (24). Furthermore, when this transgene is homozygous (SMN2+/+; Smn−/−; SMNA2G+/+) the mice have no obvious motor impairment or EMG abnormality. However this transgene, in the absence of SMN2 (Smn−/−; SMNA2G+/−), cannot rescue embryonic lethality of the Smn knockout indicating that the presence of small amounts of full-length SMN is critical to obtain phenotypic modification (24). In all of these cases the SMN mutants are expressed under the SMN promoter resulting in ubiquitous expression. Thus, one cannot determine if two copies of SMN2 is sufficient for survival in a particular cell type. The use of mice with Smn exon 7 flanked by lox sites and expression of tissue-specific Cre recombinase drivers to remove functional SMN from specific cell types, clearly indicates that full-length SMN is essential for the survival of neurons, muscle and liver (25–27). It is not clear whether motor neurons require more snRNPs and are thus uniquely sensitive to reduced SMN levels (28). Or whether sufficient amounts of SMN are produced by two copies of SMN2 for snRNP biogenesis but not for an essential function of SMN in motor neurons.
The 38 kD SMN protein is ubiquitously expressed and localizes to both the cytoplasm and nucleus (29,30). In the nucleus SMN localizes in dot like structures termed gems that overlap, or are in close proximity to, coiled bodies (29,30). SMN participates in assembly of RNA protein complexes (18,19,28,31). In particular SMN is important for snRNP assembly (18,19,28,32). In neurons SMN also localized within axons and growth cones (33–37). Motor neurons cultured from severe SMA mice (SMN2+/+; Smn−/−) show decreased axonal length, reduced growth cone size and a reduced level of β-actin mRNA and protein at the growth cone (38). More recently cultured sensory neurons have also been reported to have this phenotype but whether this extends to other neuronal types is not clear (39). In zebrafish, knockdown of SMN leads to specific axonal defects with truncated, branched axons that do reach the targeted muscle but display incorrect patterning (40,41). These zebrafish show no obvious defects in other neurons including the sensory neurons, and the axonal phenotype is cell autonomous (40,41).
Currently it is not clear if an increased level of neuronal is sufficient to rescue the SMA phenotype, or if high levels of SMN are also required in muscle. In Drosophila, a null mutation in Smn is partially rescued by maternal SMN expression permitting development to the larval stage (42). As maternal SMN will eventually be depleted in all tissues, it would be expected that replacement of SMN in all tissues would be required to rescue this mutant. Interestingly, expression in only mesodermal tissue, including muscle and muscle progenitor cells, had a major impact on survival of the fly allowing development to the pupal stage. Expression of SMN in both mesodermal and nervous tissue resulted in complete rescue (42). More recently, Rajendra et al. (43) have reported a Drosophila Smn allele that removes SMN from thoracic muscle and, in particular, the indirect flight muscle. Similar to findings in the mouse (26), removal of SMN from muscle fibers in the fly results in destruction of that muscle implying SMN is an essential protein in all tissues (44). In addition, they report an association of SMN with myofibrils (43). These studies suggest the importance of SMN in muscle and imply that high levels of SMN expression in muscle will impact the SMA phenotype. To investigate whether this is the case in mammals, we have expressed SMN in skeletal muscle fibers under the human skeletal actin (HSA) promoter (45–47) and in neurons under the prion promoter (PrP) in SMA mice (48). We demonstrate that full-length SMN by itself can correct the SMA phenotype in mice. Furthermore, high expression of SMN in muscle alone does not alter the SMA phenotype whereas a small increase in SMN expression in neurons has a major impact on the SMA phenotype and survival in the severe SMA mouse. Thus, we conclude that increased level of SMN in neurons is required to correct SMA.
Generation of HSA-SMN and prion-SMN transgenic mice
To determine where SMN is required to rescue the SMA phenotype in severe SMA mice, we expressed SMN in muscle or neurons. SMN cDNA was placed under the control of the HSA promoter (HSA-SMN) to obtain high levels of SMN in skeletal muscle fibers (45–47) (Fig. 1). This HSA promoter has been shown to drive expression from the early stages of muscle fiber formation at embryonic day 9.5 (46,47). Additionally, the SMN cDNA was placed under the control of the PrP (PrP-SMN) to express SMN at high levels in neurons (48). This mouse PrP has been used previously for the preparation of various transgenes that require high expression in neurons (48). This promoter does result in expression in astrocytes but not in macrophages. High expression is found in the brain while lower levels are detected in cardiac muscle, skeletal muscle and kidney. Liver and spleen did not show significant expression of the transgenes reported (48). The transgenes here showed expression completely consistent with previous reports of this promoter (48). Analysis of a bovine PrP fragment has shown expression in neurons, lymphocytes and keratinocytes (49). The SMN constructs driven by the promoters described were microinjected into fertilized mouse oocytes (FVB/N) to generate transgenic mice.
Five HSA-SMN lines were generated (HSA63, HSA65, HSA66, HSA67 and HSA69). All of the lines obtained expressed very high levels of SMN in muscle. Lines HSA63 and HSA69 were analyzed in more detail. Line HSA63 had five copies of the HSA-SMN transgene whereas line HSA69 had 11 copies as determined by real-time PCR.
Five PrP-SMN lines were obtained (PrP13, PrP86, PrP90, PrP92 and PrP94). Lines PrP13, PrP86 and PrP90 showed low expression of SMN while lines PrP92 and PrP94 gave high expression. Low expressing line PrP13 and high expressing line PrP92 were analyzed in more detail. Line PrP13 contains one copy of the PrP-SMN transgene while line PrP92 contains eight to nine copies as determined by real-time PCR. None of the prion or HSA transgenes affected the phenotype of normal or carrier littermate mice despite a marked increase of SMN indicating that over-expression of SMN is not pathologic.
Expression analysis of HSA-SMN and PrP-SMN
To determine the protein expression of HSA63-SMN, HSA69-SMN, PrP13-SMN and PrP92-SMN transgenes in SMA animals, we characterized these lines by western blot analysis. Figure 2 shows the SMN protein levels in 5-day-old mice (PND05) of the various genotypes. In these experiments all SMA mice are homozygous for SMN2, lack mouse Smn (Smn−/−) and are homozygous for either the HSA-SMN or PrP-SMN transgenes. The levels of SMN in the various transgenic mice were compared with carrier mice (SMN2+/+; Smn+/−) and SMA mice (SMN2+/+; Smn−/−), as well as high copy SMN2 mice (line 566 containing 16 copies of SMN2, SMN2+/+; Smn−/−) (21). As shown in Figure 2 and summarized in Table 1, HSA69-SMN produces extremely high levels of SMN in skeletal muscle but not other tissues such as brain, heart, spinal cord or liver. Line HSA63 produces very high amounts of SMN in muscle and heart as well as low levels of SMN in spinal cord (Fig. 2B–D). The level of SMN protein in spinal cord is increased from SMA mice (SMN2+/+; Smn−/−) but does not reach the level found in carrier mice (SMN2+/+; Smn+/−). We estimate from quantification of the western blots that the amount of SMN produced by the HSA63-SMN transgene in the spinal cord is about a 3-fold increase over the level found in severe SMA mice (SMN2+/+; Smn−/−) and approximately one-fifth of SMA carrier levels (SMN2+/+; Smn+/−). We found it difficult to reliably quantify the increase in SMN in heterozygous HSA63-SMN spinal cords from SMA mice (SMN2+/+; Smn−/−). Note that the level of SMN in cardiac muscle of HSA63-SMN SMA mice is well over double the level found in carrier mice (SMN2+/+; Smn+/−) or high copy SMN2 mice (Fig. 2C). Thus HSA69-SMN produces very high levels of SMN specifically in muscle while HSA63-SMN has high levels of SMN in muscle as well as leaky expression in spinal cord, brain, heart and liver.
|Line||Copy no.||Protein expression||Mean lifespan in days|
|92||8–9||+++||+++||−||+||+||150 ± 100 (n= 21)||210 ± 97 (n= 46)|
|13||1||−||−||−||−||−||7.0 ± 0.7 (n= 36)||Maximum 28|
|63||5||+||+||+++||+++||+||6.6 ± 0.8 (n= 29)||160 ± 33.9 (n= 20)|
|69||11||−||−||−||+++||−||ND||3.5 ± 0.5 (n= 15)|
|SMA||+||−||−||−||−||4.6 ± 0.4 (n= 33)|
|High copy||+++||+++||++||++||++||Normal (21)|
|Line||Copy no.||Protein expression||Mean lifespan in days|
|92||8–9||+++||+++||−||+||+||150 ± 100 (n= 21)||210 ± 97 (n= 46)|
|13||1||−||−||−||−||−||7.0 ± 0.7 (n= 36)||Maximum 28|
|63||5||+||+||+++||+++||+||6.6 ± 0.8 (n= 29)||160 ± 33.9 (n= 20)|
|69||11||−||−||−||+++||−||ND||3.5 ± 0.5 (n= 15)|
|SMA||+||−||−||−||−||4.6 ± 0.4 (n= 33)|
|High copy||+++||+++||++||++||++||Normal (21)|
For protein expression data PrP92, PrP13, HSA63, HSA69, animals are all homozygous for the HSA-SMN or PrP-SMN transgene, SMN2 and do not contain mouse Smn. Protein expression determined by western blot analysis on tissue from PND05 animals.
Copy no. refers to the number of copies of the PrP or HSA transgenes.
`−' not detectable, ‘+' low, ‘++' medium, ‘+++' high expression, ND = not determined, n = number of mice.
For mean lifespan in days heterozygous and homozygous refer to the transgene, all animals are homozygous for SMN2 and do not contain mouse Smn.
In the case of the PrP13-SMN, very low levels of SMN below detection levels of western blots were produced (Fig. 2). PrP92-SMN on the other hand, produced large amounts of SMN in brain and spinal cord easily restoring SMN levels (Fig. 2A and B). Indeed, levels of SMN in neural tissue of PrP92-SMN SMA mice were greater than that of high copy SMN2 mice and carrier mice (SMN2+/+; Smn+/−). Additionally, PrP92-SMN expression was found in other tissues, most notably muscle, but SMN levels were only slightly increased and still below levels seen in carrier mice (SMN2+/+; Smn+/−) (Fig. 2D and E). The expression profile of these PrP92-SMN SMA mice is similar to that reported using the same PrP promoter to drive mutant SOD1 constructs (48). In summary, PrP92-SMN produces very high levels of SMN in spinal cord while detection of PrP13-SMN was not possible by western blot analysis.
To further characterize the expression of the prion and HSA transgenes, reverse transcription–polymerase chain reaction (RT–PCR) was performed. HSA69-SMN had detectable expression by RT–PCR in skeletal muscle but not in other tissues therefore this line is in fact muscle specific (data not shown). Analysis of fetal mouse tissue by RT–PCR indicated that the transgene had detectable expression by embryonic day 10.5 days consistent with previous analysis of this promoter (46,47). In addition to muscle, HSA63-SMN had detectable SMN expression in all tissues tested including spinal cord and brain. PrP13-SMN, PrP90-SMN and PrP92-SMN had detectable transgenic derived SMN in all tissues tested including spinal cord, brain and muscle. The level of PrP13-SMN mRNA was detectable but the increase over SMN2 levels in spinal cord was not measurable in a reliable manner by real-time PCR, indicating very low levels of expression. Line PrP90 gave similar results (data not shown). Characterization of PrP92-SMN indicated high levels of PrP-SMN derived mRNA in neural tissue. In adult animals there was an 8-fold-increase over SMN2 mRNA levels in spinal cord tissue (Fig. 2F). PrP92-SMN expression was also found in muscle tissue (Fig. 2F). In summary, expression analysis by RT–PCR was consistent with results obtained from western blot analysis.
Timing of PrP92-SMN expression in embryos
Expression of SMN in the PrP92-SMN line was examined in fetal tissue to determine the timing of high-level SMN expression of this transgene. As shown in Figure 3, expression was slightly increased at embryonic day 13 (e13) at 1.3-fold over a SMA mouse (SMN2+/+; Smn−/−) (Fig. 3A and C). At e15 the expression of SMN in the PrP92-SMN SMA embryo was considerably increased to 4-fold over the SMA mouse (Fig. 3B and D). SMN levels however do not reach that of carrier mice at e15. Expression of PrP92-SMN remains high postnatally (Fig. 2A–E) and continues to increase in adult animals (Fig. 2F). This indicates that PrP92-SMN expression is only slightly increased at e13 and that expression increases rapidly before e15 in PrP92-SMN SMA mice.
Survival of SMA mice with HSA-SMN or PrP-SMN transgenes
To determine the effect of HSA-SMN and PrP-SMN on the phenotype of SMA mice we interbred FVB/N SMA carrier mice (SMN2+/+; Smn+/−) with the transgenic mice. Severe SMA mice (SMN2+/+; Smn−/−) without the HSA-SMN or PrP-SMN transgenes in this experiment have a mean lifespan of 4.6 ± 0.4 days (n =33 mice), which is consistent with what has been previously reported (21). The HSA-SMN lines with high expression in muscle were first examined in heterozygous crosses. Lines HSA66, HSA67 and HSA69 showed no effect on survival of SMA mice. Line HSA69 was bred to homozygosity and no alteration in the survival of SMA animals was detected. As shown in Figure 4A, SMA mice homozygous for HSA69-SMN (SMN2+/+; Smn−/−; HSA69-SMN+/+) survived an average of 3.5 ± 0.5 days (n =15). Thus HSA69, with high expression of HSA-SMN in muscle alone, showed no effect on the survival of SMA animals.
In contrast, HSA63 did have an impact on survival of SMA animals. SMA mice heterozygous for the HSA63-SMN transgene (SMN2+/+; Smn−/−; HSA63-SMN+/−) had a mean survival of 6.6 ± 0.8 days (n =29) with some living past 15 days (Fig. 4A). This was a statistically significant increase in survival compared with severe SMA mice (SMN2+/+; Smn−/−) (P < 0.003). SMA mice when homozygous for the HSA63-SMN transgene survived an average of 160 ± 33.9 days (n =20) and some lived for more than a year (Fig. 4B). Additionally, the tails of these mice became necrotic resulting in very short tails by 21 days of age (Fig. 4D). Thus, the HSA63-SMN transgene when homozygous significantly impacts the survival of SMA animals (P < 0.000001).
SMA mice heterozygous for low expressing PrP13-SMN and PrP90-SMN transgenes had a marginal, albeit significant, effect on survival of severe SMA animals increasing the average lifespan to 7.0 ± 0.7 days (n =36) and 7.5 ± 0.7 days (n =17) respectively (Fig. 4A). Making either of these transgenes homozygous further extended survival with some animals living to 28 days (data not shown). In contrast, SMA mice containing the high expressing PrP92-SMN transgene survived a mean of 150 ± 100 days (n =21) when heterozygous, and 210 ± 97 days (n =46) when homozygous with many mice living over a year (P < 0.000001) (Fig. 4B). As previously reported, Smn+/− mice have an average survival of 736 days (24). On an average SMA PrP92 homozygous mice had a shorter lifespan than Smn+/− mice, yet many mice have lived for 2 years and thus this line can have a normal lifespan. Why some SMA PrP92 homozygous mice died earlier than others is not clear and will require further study. SMA mice homozygous for the PrP92-SMN transgene have tails that are about one-third shorter than normal mice (Fig. 4D). The PrP92-SMN transgene in either the heterozygous or homozygous state prevented the weight loss usually seen in severe SMA mice over the first 5 days of life (Fig. 4C). In addition no hind limb weakness or clasping of hind limbs was detected. Thus, high expression of SMN in neurons had a major impact on both the SMA phenotype and survival of SMA animals.
SMN expression in motor neurons of PrP92-SMN animals
To confirm that PrP92-SMN is expressed in motor neurons, the expression of SMN in PrP92-SMN SMA mice (SMN2+/+; Smn−/−; PrP92-SMN+/+) was examined by immunostaining of cross-sections of spinal cord. As shown in Figure 5, gems were clearly detected in the PrP92-SMN SMA line. In addition, this transgene showed cytoplasmic aggregates of SMN. SMN expression was detected in large neurons in the ventral horn indicating expression in motor neurons.
Muscle analysis of PrP92-SMN SMA animals
The muscle morphology of PrP92-SMN SMA mice (SMN2+/+; Smn−/−; PrP92-SMN+/+) was examined histologically (Fig. 6). There was no overt phenotype and no increase in either necrotic or angulated fibers in PrP92-SMN SMA (SMN2+/+; Smn−/−; PrP92-SMN+/+) mice compared with normal muscle. The average area of muscle fibers was similar between PrP92-SMN SMA mice (SMN2+/+; Smn−/−; PrP92-SMN+/+) (2023 ± 650 µm2) and carrier mice (SMN2+/+; Smn+/−) (2451 ± 848 μm2). However there was a difference in distribution with PrP92-SMN SMA animals displaying a greater number of smaller fibers, in the 1300–2900 µm2 fiber size range, than controls (Fig. 6). There was no difference in the number of central nuclei in corrected SMA animals compared with control animals. The HSA69-SMN SMA animals showed no correction of muscle fiber size as compared with typical SMA animals. HSA63-SMN SMA animals displayed smaller fibers (1820 ± 554 µm2) with a fiber size distribution similar to that of the PrP92-SMN SMA animals. From this we conclude, while the PrP92-SMN transgene rescues the SMA phenotype and survival in the severe SMA mouse, changes in muscle fiber size distribution are present.
Rescue of motor neurons in PrP92-SMN SMA mice
To examine the rescue of PrP92-SMN SMA motor neurons, we counted the axons in the L4 ventral root of PrP92-SMN SMA and SMNA2G mice at 1 year of age. We have previously shown that addition of the SMNA2G transgene results in mild SMA mice that have motor neuron loss (24). Examination of PrP92-SMN SMA mice (SMN2+/+; Smn−/−; PrP92-SMN+/+) revealed normal lumbar root counts whereas the SMNA2G mice (SMN2+/+; Smn−/−; SMNA2G) (mild SMA) of the same age have a clearly significant reduction in root counts (Fig. 7). There was no alteration in the mean diameter of the roots in the groups tested. Thus, high expression of SMN under the PrP rescues the motor neuron loss of SMA mice.
Requirement of SMN in muscle
SMA is caused by reduced levels of SMN, which leads to loss of functional motor neurons (1,13,14). SMN is however ubiquitously expressed and performs an essential function in all cells, namely snRNP assembly (18,19,28,31,32). This process is developmentally regulated and different tissues do appear to have different snRNP assembly properties (50). There is currently considerable debate over whether disruption of snRNP biogenesis or an axonal function of SMN leads to SMA (51). Loss of all SMN from a specific tissue results in destruction of that tissue (25–27). Previously, others have shown that loss of functional SMN in mature muscle fibers by HSA-Cre excision of SMN exon 7 leads to a cycle of muscle degeneration and regeneration resulting in a dystrophic phenotype (26). This study illustrated the essential role of SMN for survival of muscle tissue. However, SMA is caused not by the elimination of SMN but by low levels of SMN (3,13,14). Therefore, we asked whether increasing SMN expression in muscle in the severe mouse model of SMA (21) could alter the SMA phenotype. Here we show that replacement of SMN in neurons rescues the SMA phenotype and significantly increases survival of the animal. The question of whether SMN expression in neurons alone, or expression in both neurons and muscle gives rescue cannot be fully answered with the transgenes discussed here as the PrP promoter expresses in muscle, albeit at low levels. Conversely, these experiments clearly show that replacement of high levels of SMN in muscle fibers alone has no effect on the SMA phenotype or the survival of the animal.
A series of conceptually similar experiments using the UAS-GAL4 overexpression system have been performed in Drosophila. Mutations in the Drosophila Smn gene have been recently identified. Smn73Ao and SmnB are missense mutations (G202S and S201F) in the C-terminal oligomerization domain of fly Smn (42). These mutations disrupt the oligomerization of SMN and are equivalent to a series of human exon 6 mutations (Y272C) that disrupt oligomerization and occur in severe SMA patients (11,13,52). These alleles therefore are null for SMN function (42). In Drosophila these mutations when homozygous are larval lethal due to the presence of maternal SMN in the embryo (42). Ubiquitous expression of Smn (1032-GAL4 driver) rescues larval lethality in the fly (42). Expression of Smn in mesodermal tissue (how24B-GAL4) resulted in partial rescue while pan-neural expression in mature neurons (elav-GAL4) was less effective. Expression of Smn in both muscle and nerve allowed many flies to develop normally. However, as indicated by the authors it is not clear if the expression of Smn using the UAS-GAL4 system is strictly limited to the specific tissue of interest (42). Thus it is possible that some leakage of Smn into other tissues could give sufficient Smn levels for survival. While the results in fly seem to contradict our studies, there are two major differences between the study in Drosophila and the results presented here. First, in our current study full-length SMN is continuously produced in all tissues by the SMN2 transgene resulting in low levels of SMN protein throughout life. The fly studies use a null allele of Smn therefore no functional Smn is present after maternal embryonic Smn stores have been depleted. In mice once maternal SMN is depleted the embryo undergoes massive apoptosis but this occurs very early in development (17) and the effect of extending SMN expression to later development time points and then depleting SMN is unknown. It would be of interest in Drosophila to use a hypomorphic allele producing limited quantities of functional Smn in all tissues throughout development and determine the phenotype. To date such a mutation has not been reported. In addition, expression of Smn in mature neurons using elav-GAL4 may be predicted to result in high levels of Smn too late in development to achieve complete rescue (53). Second, in the study presented here HSA-SMN is expressed in muscle fibers (54) while in the fly Smn was expressed more broadly in all mesodermal tissue including muscle precursors and satellite cells. Previously, a reduced capacity of satellite cells or myoblasts to grow and repopulate muscle was observed when functional SMN levels were reduced in Smn+/− animals (55,56). High SMN levels may be specifically important to muscle satellite cells and affect muscle repair. Similar disruption of the ability of satellite cells to repair injured muscle is seen with MyoD1 null mutants (57). However it seems unlikely that alteration of repair in muscle would have a major impact on survival of SMA animals although reduced SMN levels in satellite cells may be responsible for the altered fiber size observed in SMA mice corrected with PrP92-SMN.
The HSA63-SMN transgene, when homozygous, did have a major impact on the SMA phenotype, which we believe is due to leaky SMN expression in neurons. When heterozygous, the HSA63-SMN line had minimal influence on survival of SMA mice despite very high levels of SMN expression in heart and muscle. In contrast, the expression of IGHMBP2 in muscle and heart in a mouse model (nmd) of the human disorder SMA with respiratory distress (SMARD) did have a significant effect on survival of the animals (58). Complete correction of the phenotype in nmd mice required expression of IGHMBP2 in both muscle and neurons (58). Expression in neurons only did not have a major impact on survival (59). Thus, it would appear that the mechanism behind SMA and SMARD (nmd) are quite different even though both are motor neuron disorders.
The role of SMN in neurons
The PrP92-SMN transgene expresses SMN at high levels in neural tissue at embryonic day 15. Testing with RT–PCR revealed some expression of PrP92-SMN in all tissues but PrP92-SMN expression was highest in neurons. This promoter has been previously used to drive mutant SOD1 in mice resulting in mice with amyotrophic lateral sclerosis (48). The expression pattern reported for the SOD1 transgene is similar to that of the PrP92-SMN (48). PrP92-SMN expression in SMA mice resulted in rescue of the SMA phenotype and a significant extension of survival from 4.6 to 210 days. Additionally, motor neuron root counts were normal in these mice. In contrast, SMNA2G mice at 1 year of age show reduced motor neuron root counts. Thus PrP92-SMN corrects the SMA phenotype and loss of motor neurons. The PrP13-SMN and PrP90-SMN showed a marginal increase in survival in either the heterozygous or homozygous state (4.6 to 7.0 days). These transgenes do produce some SMN in spinal cord as indicated by RT–PCR but only at extremely low levels below what is readily detectable on a western blot (Fig. 2B). As would be expected with these very low levels of SMN only a minimal increase in survival was obtained. The levels of SMN in the spinal cord of HSA63-SMN SMA mice are clearly above that found in PrP13-SMN SMA mice (3-fold greater) and thus can explain why the HSA63-SMN transgene rescues and PrP13-SMN does not (Fig. 2B). Rescue by the HSA63-SMN transgene when homozygous, but not heterozygous, indicates that a relatively small increase in the amount of SMN produced by SMN2 has a major impact on the SMA phenotype and survival in these mice. Interestingly, this situation is close to what might be predicted in humans in that two copies of SMN2 result in severe SMA, four copies of SMN2 result in a mild SMA phenotype (60–63) and five copies of SMN2 has been found in a phenotypically normal individual lacking SMN1 (63). Thus, a 3-fold increase in SMN2 expression makes a major impact on survival.
Alternate isoforms of SMN
Recently an alternative isoform of SMN, termed axonal-SMN (a-SMN), has been reported (64). In this isoform the a-SMN protein is encoded by SMN1 exons 1–3 and part of intron 3 (64). The PrP92-SMN transgene used in our studies only produces full-length SMN. Therefore, a-SMN protein would not be produced by the PrP92-SMN transgene. Axonal-SMN was reported to be produced from the SMN1 gene and not the SMN2 gene (64). As the transgene PrP92-SMN fully corrects the SMA phenotype and is not capable of producing a-SMN, and the mouse null allele used in these studies would also not produce a-SMN, we conclude that a-SMN does not play a significant role in SMA. It is possible that SMN2 does produce a-SMN but if so two copies of SMN2 would produce sufficient a-SMN. The antibodies used in our studies detect SMN epitopes in exon 2a and 2b and thus should detect a-SMN yet we have not observed a-SMN protein in either embryonic or adult mouse tissue samples. Moreover, in SMA patients where SMN1 is deleted, the copy number of SMN2 has a major impact on phenotype. In fact, in both man and mouse sufficient SMN2 copies can completely ameliorate the SMA phenotype (21,63). Finally, human patient SMN missense mutations occur throughout the SMN gene including exon 6 and exon 7 (65) and these would not disrupt the a-SMN protein, thus there is no indication that a-SMN has a role in SMA (66).
In summary, expression of full-length SMN in motor neurons is critical for correction of SMA. High levels of SMN in muscle fibers alone have no impact on the phenotype of SMA mice. However, a relatively small increase of SMN in neurons in addition to the muscle has a major impact on survival of SMA mice. In summary, we show here that increased SMN protein levels in neurons is necessary for prevention of a disease phenotype in a mouse model of SMA.
MATERIALS AND METHODS
Generation of transgenic mice
The HSA promoter vector has been previously described (54) and consists of the HSA promoter (45) a splice acceptor site, a small intron and two SV40 polyadenylation signals. The previously isolated SMN cDNA containing exons 1–8 (23) was cloned into the NotI site in the HSA vector and the ligated vector and insert were sequenced. The inset was released with ClaI and PvuII, gel purified, and injected into fertilized FVB/N mouse oocytes. The transgene was detected as previously described (23) using HSA FP ccaggtagggcaggagttgggagg or HSA FP gagccgagagtagcagttgt and SMN RP agaatcatcgctctggcctgtgcc.
The mouse PrP vector has been previously described (48,67) and contains the mouse PrP promoter and PrP exons 1, 2 and 3. The SMN cDNA cassette was excised from the HSA vector with XhoI and PvuII band isolated, end filled and cloned into the end filled XhoI site in PrP exon 2. The resulting construct was sequenced, linearized by digestion with PvuI, gel purified and injected into fertilized FVB/N mouse oocytes. The prion-SMN transgene was detected with PrP FP ggactcgtgagtatatttcag and SMN RP agaatcatcgctctggcctgtgcc primers or SMN RP agtagatcggacagattttgct.
Transgenic founders were bred to SMN2; Smn+/− mice (21,23). PrP-SMN+/−; SMN2+/−; Smn+/− or HSA-SMN+/−;SMN2+/−; Smn+/− mice were interbred resulting in mice homozygous for the transgene of interest as well as SMN2 and Smn. To identify mice that were homozygous for the PrP-SMN or HSA-SMN and SMN2 transgenes, mice carrying the transgenes were bred to non-transgenic FVB/N mice and animals that transmitted the transgene to all offspring selected. In addition, real-time PCR as described below was used to select for homozygous animals (23). The SMN2 transgene and mouse knockout allele were detected as previously described (23).
Determination of transgene copy number
Real-time PCR for copy number determination was performed on genomic DNA as previously described (23). The control primers for reference in copy number determination include Beta-globin FP aaacaagagcaaactaagtaagatgcat and Beta-globin RP tgagtgcatggatgaatctcct as well as mouse Smn FP cccagaccagttttgccttg, and mouse Smn RP caatgcctacagcccagaca. The genomic SMN2 transgene was detected with SMN2_intron1 FP acccttcttccggccca and SMN2_intron1 RP cttgcggatgtggttcgc. SMN cDNA was detected with SMNexon4.5F actgggaccaggaaagccaggt and SMNexon6R agccagtatgatagccactcatg. PCR was performed with SYBR Green Jump Start Taq Ready mix (Sigma-Aldrich) and the following cycling conditions for all reactions were as follows: 10 min 95°; 40 cycles of 95° 30 s, 60° 30 s, 72° 30 s; 95° 1 min ramp to 55° for 30 s.
The copy number of the SMN2 transgene that was used in this study [line 89, Tg(SMN2)89Ahmb] was confirmed using primers that flank the insertion site of this transgene. The primers are as follows: TG89 Border (Grm7) Fwd (591): ctgacctacagggatgagg, Tg89_Grm7_negative (597): cccaggtggtttatagactcaga and TG89 Border (SMN) Rev (592): ggtctgttctacagccacagc.
RT–PCR and real time RT–PCR
Total RNA was isolated from mouse tissue as previously described (23) and cDNA prepared using Thermoscript RT polymerase according to the manufacturer’s instructions (Invitrogen). PrP-SMN was amplified with PrP FP gtcggatcagcagaccgattctg and SMN RP agaatcatcgctctggcctgtgcc. HSA-SMN was amplified with HSA FP gagccgagagtagcagttgt and the same SMN RP agaatcatcgctctggcctgtgcc. The actin control was amplified with actin FP cagcttctttgcagctcctt and actin RP cacgatggaggggaatacag. For total human SMN (from SMN2 and HSA-SMN or PrP-SMN transgenes) SMN exon1 FP aggaggattccgtgctgttc and SMNexon 2a RP tgcttaaatgaagccacagc were used.
The level of SMN mRNA expression was determined by real time RT–PCR on adult hindbrain, forebrain, spinal cord and muscle tissue. Total RNA was isolated from mouse tissue and treated with RNase-free DNase (Ambion) as previously described (23). cDNA was produced as previously described using Thermoscript-RT polymerase. For total human SMN mRNA (SMN2 and PrP-SMN transgenes) primers SMNexon1 FP aggaggattccgtgctgttc and SMNexon2a RP tgcttaaatgaagccacagc were used. The actin control was amplified with actin FP cagcttctttgcagctcctt and actin RP cacgatggaggggaatacag. PCR was performed with SYBR Green Jump Start Taq Ready mix (Sigma-Aldrich) and the following cycling conditions for all reactions were as follows: 10 min 95°; 40 cycles of 95° 30 s, 60° 30 s, 72° 30 s; 95° 1 min ramp to 55° for 30 s. The amount of SMN PCR product was normalized to actin. The fold-increase of normalized SMN PCR product was determined relative to the amount of normalized SMN produced by SMN2 from animals containing two copies of SMN2 but lacking PrP-SMN to obtain relative fold-increase of SMN expression due to the transgene.
Western blot analysis
Western blot analysis on adult and neonatal tissue was performed as previously described (23). Embryonic hindbrain and spinal cord tissue was isolated by dissection and prepared for western analysis using the same procedures for adult tissue. Antibodies used to detect SMN were a mix of mouse monoclonal SMN antibodies to exons 2a and 2b, MANSMA2, MANSMA7, MANSMA13, MANSMA19 (1:1000) (www.rjah.nhs.uk/cind) followed by horseradish peroxidase conjugated secondary antibody (1:10 000) (Jackson Immunoresearch) and ECL chemiluminescence detection (GE Healthcare). The blots were stripped and re-probed with β-actin antibody (1:6000) (Jackson Immunoresearch) or β-tubulin antibody (1:6000) (Jackson Immunoresearch) as controls for protein loading. Quantification of western blots was performed using a Shimadzu 9000 CS scanner as previously described (14).
Histology and immunofluorescence
Hematoxylin and Eosin staining was performed on 12 µm frozen transverse sections of gastrocnemius muscle from 4 month old mice as previously described (23). Myofiber area was calculated using a SPOT-RT slider digital camera and SPOT software (Diagnostic Instruments, Inc). Immunofluorescence was performed on 8 µm frozen unfixed lumbar spinal cord sections of PND04 mice. Samples were postfixed in 4% paraformaldehyde and incubated in 10% tween 20, 0.4% goat serum, PBS with a rabbit polyclonal SMN antibody (1:500) (37) and goat anti-rabbit conjugated Alexa Fluor 594 secondary antibody (1:1000) (Molecular Probes). Tissues were mounted in Vectashield mounting medium containing DAPI (Vector labs). Images were obtained with confocal microscopy using a Leica TCS_SL scanning confocal microscope system using an inverted Leica DMIRE2 microscope, PMT detectors and a 63× HCX Plan Apo CS oil, NA = 1.4 objective. Image acquisition and scale bars were produced with the Leica Confocal Software v2.61 and subsequential image processing was performed with Adobe Photoshop CS2.
L4 ventral root axon counting
The L4 ventral root of 4% paraformaldehyde perfused animals (1 year and 6 month old animals) was identified by dissection upward from the sciatic nerve. Mid-ventral root segments were post-fixed in glutaraldehyde, epon embedded, and 1 µm cross-sections were stained with toluidine blue. Myelinated axons counts and axonal diameters (diameter of circle of equivalent area) were determined according to standard sterologic protocols (68). Statistical comparison of group L4 ventral root axon counts was by two-tailed t-test.
Survival (Kaplan–Meier) and statistical analysis was performed with SPSS v 14.0 (SPSS, Inc.). Significance of survival analysis was determined with the log-rank test. The muscle fiber size distribution was analyzed using SPSS v14 specifically testing the median with the Wilcoxon Mann–Whitney and two sample Kolmogorov–Simirnov test.
These studies were funded by NIH grant NS038650 (to A.H.M.B.), NS041649, and The Ohio State Neuroscience Center Core, P30-NS045758. V.L.M. was funded by FSMA and U.R.M. by a MDA transition award. We are also grateful to Miracles for Madison fund support. This work was supported in part by Muscular Dystrophy Grant, no. 3572 (C.J.D.) and Families of Spinal Muscular Atrophy grant no. DON-04-05 (C.J.D.) and the Medical Research Junior Board Foundation (MRJBF) at Children’s Memorial Hospital. C.J.D. is supported by an American Academy of Neurology/SMA Foundation Young Investigator award.
Dr Jill Rafael-Fortney for providing access to the SPOT system for analysis of microscope images and Dr Christine E. Beattie for critical reading of the manuscript.
Conflict of Interest statement. None declared.