Despite steady progress in our understanding of the genetic basis of SMA, the molecular functions of SMN and the development of animal models, the defect responsible for the specific degeneration of motor neurons in the spinal cord of SMA patients remains elusive. The best-characterized activity of the SMN complex is the assembly of Sm proteins onto snRNAs forming the spliceosomal snRNPs 
. These abundant nuclear RNPs function in the context of a dynamic macromolecular machine known as the spliceosome to carry out the excision of introns and ligation of exons to form mature mRNAs 
. With the goal of understanding SMA pathophysiology, here we have characterized SMN complex activity and the consequence of its deficiency on snRNP metabolism in the spinal cord of mouse models of the disease.
We demonstrated that the levels of a subset of Gemin proteins (namely Gemin2, Gemin6 and Gemin8) are significantly reduced in the spinal cord of severe SMA mice compared with normal mice, and that Gemin8 is an integral component of the SMN complex whose expression is most severely affected by decreased SMN ( and S1
). In contrast, the levels of Gemin4 and unrip as well as those of several other SMN-interacting proteins are unchanged. The SMN–dependent reduction of Gemin proteins expression has also been observed in cultured cell lines and likely results from degradation due to defective incorporation into SMN complexes 
. Importantly, the degree of reduction in Gemin2, Gemin6 and Gemin8 expression in the spinal cord of SMA mice correlates with disease severity (), suggesting that these proteins might be involved in the cellular activities of SMN whose impairment underlies SMA pathology. In this regard, it is noteworthy that the beneficial effect of treatment with trichostatin A on survival of SMNΔ7 SMA mice correlates with increased levels of functional SMN complexes in the CNS 
. Decreased levels of Gemin2, Gemin6 and Gemin8–which are required for efficient snRNP assembly activity 
–may also exacerbate the consequence of reduced SMN expression on Sm core formation in the spinal cord of SMA mice. Indeed, we demonstrated that snRNP assembly activity is dramatically reduced in spinal cord extracts from severe SMA mice (). A similar decrease in SMN levels and severe impairment of Sm core formation are also observed in extracts from the other SMA tissues analyzed here ( and ). Although quantitative differences exist in the snRNP assembly activity of distinct normal tissues and possibly also in the extent of its decrease in SMA tissues, these results indicate that reduced capacity for Sm core formation is a general biochemical deficiency in severe SMA mice. Importantly, we showed that the degree of snRNP assembly impairment in the spinal cord of SMA mice correlates with disease severity (). Collectively, our findings provide strong support to the possibility that deficiencies in snRNP biogenesis contribute to SMA pathology.
For the first time, we demonstrated here that the impairment of SMN function in Sm core formation leads to a significant decrease in the levels of a subset of snRNPs in the spinal cord as well as other tissues of severe SMA mice (, and ). As indicated also by similar levels of SmB protein in the spinal cord of normal, carrier and SMA mice (), the bulk of snRNPs is however only marginally affected in SMA tissues. This is especially remarkable considering the ten-fold decrease in snRNP assembly activity in neural tissues from SMA mice. It is also in agreement with the observation that steady-state snRNP levels are unchanged in DT40 cells with severely reduced SMN levels and in Drosophila
larvae harboring SMN null mutations 
. In light of the striking discrepancy between the degree of reduction in snRNP assembly and snRNP levels in tissues of SMA mice as well as other model systems, it is important to consider that in vitro
snRNP assembly assays measure the capacity of Sm core formation in extracts, which likely corresponds to the amount of SMN complexes bound to Sm proteins and therefore capable of snRNP assembly. However, reduced snRNP assembly capacity in vitro
as a consequence of low SMN levels may not necessarily translate into a defect of snRNP synthesis in vivo
if the SMN complexes are in excess. In order for snRNP accumulation to be affected in vivo
, the SMN complex must fail to meet the demand for Sm core formation on the snRNAs, which is dictated primarily by the rate of snRNA transcription. Thus, the physiologically critical parameter is the threshold of SMN activity required to support the level of snRNP synthesis needed in vivo
, which likely differs in distinct cells and changes also with development. Only if SMN activity decreases below this threshold insufficient snRNPs are made and steady-state snRNP levels may be affected. Although it is generally assumed that snRNP assembly is rate limiting in SMA, we found surprisingly that snRNP synthesis in vivo
is not affected in type I SMA fibroblasts (). Together with the overall modest reduction of snRNP levels in SMA tissues, these results suggest that SMN complexes competent for snRNP assembly greatly exceed the amount needed to meet the in vivo
demand for snRNP synthesis in most tissues and that decreased SMN levels have little if any impact on snRNP accumulation in SMA. An important implication is that, somewhat similar to the situation in larval lethal Drosophila mutants 
, the specific neuromuscular phenotype triggered by systemic SMN deficiency in mouse models of SMA is not caused by global depletion of spliceosomal snRNPs. Nevertheless, our finding that the levels of a subset of snRNPs are significantly decreased in tissues of severe SMA mice indicates that there are at least some cells in SMA in which SMN capacity for snRNP assembly becomes overwhelmed and insufficient snRNPs are made possibly due to a requirement for high levels of snRNP synthesis.
One of the most strikingly unexpected findings of our study is that impairment of SMN ubiquitous function specifically alters the snRNP profile of SMA tissues by decreasing the levels of a subset of snRNPs and more prominently of U11. Spliceosomal snRNPs are divided into two distinct classes depending on the type of introns that they remove 
. The vast majority of eukaryotic introns are processed by the major (or U2-dependent) spliceosome formed by U1, U2, U4/U6 and U5 snRNPs. Less than 1% of introns are processed by the minor (or U12-dependent) spliceosome comprising U11, U12, U4atac/U6atac and U5 snRNPs. Specific conserved sequence features distinguish the two types of introns and determine their commitment to either the major or the minor pre-mRNA splicing pathway 
. Bioinformatic analysis of intron sequences from the databases reveals the presence of approximately seven hundred U12-type introns in the human genome 
. Consistent with the relatively low abundance of the target introns, minor snRNAs (103
per cell) are two orders of magnitude less abundant than major snRNAs (105
per cell) 
. Although the SMN complex assembles the Sm core on both major and minor snRNAs 
, differences in the abundance, efficiency of Sm core formation and turnover rate of individual snRNAs might contribute to the observed effects of SMN deficiency on the levels of specific snRNPs. Future studies will be needed to address these issues directly. Regardless of the mechanism(s), however, our results indicate that SMN deficiency unevenly alters the physiological proportion of endogenous snRNPs in tissues of severe SMA mice, and preferentially affects the accumulation of U11 snRNP of the minor splicing pathway. These findings have important potential implications for SMA pathogenesis because the disease trigger targeted by SMN reduction may lie within genes containing introns processed by the minor splicing pathway. There is evidence that processing of U12-type introns is slower and possibly more error prone than that of U2-type introns under normal conditions and might represent a rate-limiting step in the expression of genes that contain these introns 
. Decreased levels of minor snRNPs such as U11 may further enhance this situation and have deleterious consequences on the expression of mRNAs containing U12-type introns in cells of SMA patients. It is also conceivable that although the minor splicing pathway processes a number of genes only a few may be affected by SMN reduction because of differences in the splicing efficiency of individual U12-type introns.
Dysfunctions in components of both constitutive and regulated pre-mRNA splicing have been implicated in the disease mechanisms of an increasing number of human disorders 
. In addition, there is precedent for a model where reduction of a particular splicing factor results in a tissue-specific phenotype. Myotonic dystrophy is caused by nucleotide repeat expansions in the non-coding region of specific mRNAs, which act to sequester splicing factors such as the muscle blind protein away from the pre-mRNA processing machinery leading to aberrant processing of distinct target mRNAs in different tissues 
. As a consequence of this splicing deregulation, in mouse models of myotonic dystrophy, the chloride channel 1 gene is processed incorrectly during early postnatal development of skeletal muscle and the fetal mRNA form that is not active in chloride conductance is retained 
. In turn, this affects the physiological increase in chloride channel conductance in skeletal muscle resulting in myotonia and muscle degeneration 
. It is tantalizing to speculate that the processing of one or more introns in specific genes that have a critical role in motor neuron biology might be similarly affected by reduced functionality of the U12-dependent spliceosome in SMA. It is noteworthy that U12-dependent introns are not randomly distributed in the genome but rather enriched in few gene families 
. Members of the voltage-gated ion channel gene family contain an unusually high frequency of U12-dependent introns and are especially interesting in the context of SMA pathology 
. These genes control numerous activities that are critical for neuronal function and muscle contraction, including action potential, signaling processes and synaptic transmission, and whose deficiency is responsible for several neuromuscular and neurological human disorders 
. One attractive albeit presently speculative scenario is that altered processing of some members of this gene family could cause electrophysiological disturbances that contribute to motor unit dysfunction and SMA pathophysiology. It should be pointed out however that there is no evidence of splicing defects in SMA and it is not known whether the reduction of snRNP levels reported here would indeed affect the functionality of the minor pre-mRNA splicing pathway. Future studies will be needed to address the many questions stemming from our work. In particular, whether the minor splicing pathway rather than the major splicing pathway is affected in SMA and whether defective processing of specific mRNAs containing U12-dependent introns contributes to the preferential vulnerability of motor neurons to reduced levels of SMN.