Introduction

Spinal muscular atrophy (SMA) refers to a group of disorders that affect the lower motor neuron, and a number of genes have been defined that cause SMA. The most common SMA is proximal SMA that maps to chromosome 5q. In contrast, the late-onset distal SMAs (DSMAs), which show considerable overlap with Charcot–Marie–Tooth disease phenotype, are mostly non-5q SMAs. For these disorders many genes have been implicated, including: GARS , DCTN1 , HSPB8 , HSPB1 , BSCL2 , SETX , HSPB3 , DYNC1H1 , REEP1 , and SLC5A7 . Dominant SMAs with proximal weakness have mutations in VAPB , TRPV4 , LMNA , and, most recently, BICD2 . The recessive non-5q SMAs are caused by mutations in IGHMP2 for DSMA1 or SMA with respiratory distress, mutations in Gle1 for severe lethal congenital contracture syndrome 1, which results in loss of motor neurons in the fetus, and lastly the X-linked SMA caused by mutations in Ube1 . Overall, there is no indication of an overlapping function that is disrupted by other causative SMA genes with the exception that, in dominant proximal SMA, the genes VAPB , TRPV4 , and BICD2 may play a role in vesicle release. However, the phenotype of these proximal dominant SMAs is quite different from proximal recessive 5q SMA. For instance, cases with the BICD2 mutation show symptoms at birth with a slow progression, upper motor neuron involvement, and sparing of axial muscle. Therefore it is possible that mutations in different SMA causative genes represent alternate paths to motor neuron destruction. Here we focus on the genetics of 5q SMA.

Proximal 5q SMA is an autosomal recessive disorder and is the most common genetic cause of infant death. Proximal 5q SMA is caused by loss or mutation of the survival motor neuron 1 gene ( SMN1 ) and retention of the copy gene SMN2 . 5q SMA has a frequency of 1/11,000 new births and carrier frequencies range from 1/47 to 1/72, depending on racial group. The SMN1 and SMN2 genes lie in a complex region of the genome that shows a very high divergence among different individuals. Thus, any one particular arrangement of the SMN1 and SMN2 genes does not reflect the arrangement of the genes across the population. Of particular relevance to SMA disease is the inherent copy number variation of SMN1 and SMN2 in the population. The SMN2 copy number varies in SMA individuals and, provided that the SMN2 gene is intact, these extra copies modify the phenotype. An inverse correlation of phenotypic severity to copy number of SMN2 is found in SMA patients. Notably, for many SMAs the mutated gene is expressed in a wide range of tissues and expression is not confined to the nervous system. This is also the case with SMN and as such the reason for selective motor neuron or motor circuit deficit is not known.

The Spinal Muscular Atrophy Gene Region

The SMA region is often diagramed as an inverted duplication of SMN1 and SMN2 . This arrangement of genes was identified in a particular yeast artificial chromosome (YAC). However, this region is somewhat more complicated and we now know the arrangement of genes is not the same in all individuals ( Fig. 6.1 ). The gene for SMA was first mapped by linkage analysis to position 5q for all SMA types. Subsequently, as in most positional cloning, closer markers and physical maps of the region between the two flanking markers were then determined. However, the more closely located markers often gave rise to multiple copies in an individual (i.e., greater than two alleles) indicating duplication of the marker. Furthermore, the variation in copy number of these markers between individuals indicated different associated genomic DNA. Some markers indicated reduced gene copy number in many type 1 patients thus suggesting a large deletion in this area. The SMA area was cloned and physical maps were produced utilizing YACs. One difficulty encountered while building these maps was that the YAC clones from this region frequently underwent deletion or some form of rearrangement. This could be easily detected by genotyping of the original cell line and the YAC clones, which most often contained one allele instead of multiple alleles. Thus maps of the SMA region produced from YAC clones with deletions were interpreted as if only one arrangement of multiple chromosomes was possible. Further studies, using pulsed field gel electrophoresis to map the SMA region, revealed marked differences between individuals as different size restriction fragments were identified with probes from the SMA region. Lastly, unequal crossing-over was found in some individuals indicating a noninverted duplication of the SMN genes. Therefore, multiple alternative structures of the SMA region have been identified and as such this area is not always composed of an inverted duplication ( Fig. 6.1 ). One consideration, which even today has not been addressed, is whether particular genomic structures are more prone to new mutations. The rate of new mutation has been reported to be 2%. However, considering the mutation rate by itself, then this should lead to an exceptionally high rate of disease alleles for an autosomal recessive disorder; yet, this is not the case. The most likely reason for this is the presence of an equal back-mutation rate of SMN2 exchange or conversion to SMN1 . Due to the complex nature of this region, it is often incorrectly represented in sequence databases and caution should be applied when extracting data from these sources.

Diagram of the arrangement of genes in the spinal muscular atrophy (SMA) region. The survival motor neuron survival motor neuron (SMN)-gene-containing region is often depicted as a 500-kb inverted duplication based on the arrangement found in the original yeast artificial chromosome (YAC) clone. However, as depicted above, the gene can be arranged in a tandem duplication or with genes absent from one of the repeat units. Based on pulsed field gel analysis and mapping there are many possible organizations of these genes and not all organizations are depicted here. BTF2p44, basic transcription factor 2 44-kDa subunit ; ψNAIP, Psi NLR family, apoptosis inhibitory protein pseudo gene ; SMN2, survival motor neuron 2 ; SERF1, small EDRK-rich factor ; SMN1, survival motor neuron 1 ; NAIP, NLR family, apoptosis inhibitory protein ; Cen, centromeric; Tel, telomeric. An example of the direction of transcription in is indicated by arrows; however, other arrangements are possible.

Genetics of 5q Spinal Muscular Atrophy and Phenotype Modification in Man

Proximal 5q SMA is an autosomal recessive disorder and is caused by mutation or loss of the SMN1 gene and retention of the SMN2 gene. The SMN1 and SMN2 genes essentially differ by a single C–T change in exon 7 of SMN2 . This nucleotide change results in disruption of a splice modulator such that the majority of the transcript from SMN2 lacks exon 7. The SMN lacking the amino acids encoded by exon 7 does not oligomerize efficiently and is rapidly degraded. This results in SMN2 not producing as much full-length SMN protein as SMN1 , and thus there is a deficiency of SMN in SMA. The reduction of SMN occurs in all tissues; however, motor neurons are particularly affected. It is clear that the amount of SMN produced is critical in determining the severity of the SMA phenotype. Indeed, quantitative assays for the determination of SMN1 and SMN2 copy number permit both the detection of carriers and the correlation of phenotypic SMA severity with copy number of SMN2 .

SMA can be divided into five categories of severity from type 0 to type 4. Type 0 or 1a, which in general have one copy of SMN2 , have the disease at birth. Type 1 SMA patients generally have two copies of SMN2 and develop SMA prior to 6 months of age and die before the age of 2 without ventilatory support. It should be noted that type 1 is often divided into subgroups 1b and 1c where 1c is slightly milder. It has been reported that 85% of type 1 patients have one or two copies of SMN2 where the remaining 15% have three copies of SMN2 . In some but not all cases, type 1 patients with three copies of SMN2 show a slightly milder progression. In type 2 SMA, onset occurs between 6 and 18 months and patients can sit unaided but never stand or walk. In type 2 patients, approximately 82% have three copies of SMN2 , 11% have two copies and 7% have four copies. In type 3 and 4 SMA, onset is between 18 months and 30 years. In these cases, three or four copies of SMN2 are most common; however, 4% of type 3 patients have two copies of SMN2 .

Not only does the copy number of SMN2 influence phenotype, but the type of SMN2 gene can also influence phenotype ( Fig. 6.2 ). For example, the variant c.859G>C in exon 7 of SMN2 results in an approximate 20% increase in full-length SMN RNA and a mild SMA phenotype. Interestingly, the variant c.859G>C in exon 7 of SMN2 does not occur in type 1 patients, is found in one copy in type 2 patients, and occurs in two copies in milder type 3b patients (the milder spectrum of type 3 patients). Clearly the output of SMN2 is critical to determination of phenotype. However one important consideration is that the ability to identify an intact SMN2 gene is not possible with the current diagnostic testing of SMN2 . Some Multiplex Ligation-dependent Probe Amplification (MLPA) assays do look at the copy number of additional exons such as exon 1; however, internal mutations of SMN2 will not be detected. Thus, with current screening procedures we cannot obtain precise information on whether these SMN2 genes are fully intact and functional ( Fig. 6.2 ). Therefore, while type 1 patients with three copies of SMN2 , and type 2 cases with four copies of SMN2 occur, it is not known if all the identified SMN2 genes are intact and producing full-length SMN.

Diagram of survival motor neuron 2 (SMN2) alleles in SMA. (A) Intact SMN2 with polymorphic regions and type indicated. (B) 5′-Truncated SMN2 allele lacking exon1-2b. (C) 3′-Truncated allele reported as polymorphic variant. The exact breakpoint is not known. The allele shown here contains exon 1 but exons 2a–6 are questionable. (D) Alternative 5′-truncated SMN2 allele with exon 1 missing. (E) Severe SMN2 missense mutation. (F) Mild SMN2 missense mutation. AG1-CA, AAA, TTT, CA, AGGCA: polymorphic markers identified in the intronic regions of SMN.

Often there is discordance of the Ag1-CA/C272 marker with the SMN2 copy number as well as results with MLPA assays. We identified Ag1-CA as a marker that showed allele loss in type 1 SMA and was in clear linkage disequilibrium with SMA. It was subsequently found that Ag1-CA lies close to exon 1 of the SMN genes in the promoter ( Fig. 6.2(A) ). The marker typing correlated to SMN2 copy number for The Ohio State University families used in linkage analysis. Indeed, when SMN2 copy number and Ag1-CA genotype were examined in German families, there was clear evidence of SMN2 genes that lack the 5′-end marker and were therefore not intact functional SMN2 genes.

Lastly, the SMN2 gene can have mutations in it just like SMN1 . This has already been reported with the pSMNG278R missense mutation detected in exon 7 of SMN2 . It is predicted that SMN2 genes will have similar mutations to SMN1 , including small deletions, duplications, and splice disrupting mutations. In this regard mild mutations could also occur in SMN2 . We have shown that mild SMN1 mutations act by complementing the full-length SMN produced by SMN2 ( Fig. 6.3 ). Therefore, a mild missense mutation occurring in an SMN2 gene is predicted to not complement as well as a wild-type SMN2 allele. Thus the presence of the mutated SMN2 gene may not give the same degree of alteration to the phenotype. We have diagramed in Fig. 6.2 potential SMN2 alleles that are likely to occur in SMA patients. Most are truncated alleles of SMN2 that have a 3′-end and thus test positive in the currently used diagnostic assays. In addition we indicated a reported polymorphic variant, which contains the 5′-end of either SMN1/2 , but not the 3′-end that is used for copy number detection of SMN2 . The latter allele can confuse scoring of intact SMN2 alleles if you use a single site for indicating dosage. To fully analyze the SMN2 alleles in SMA requires a system that gives accurate copy number of all regions of the gene along with the complete sequence of the SMN2 genes present.

SMN missense mutations in complementation. (A) Diagram of the domains of SMN and the missense mutations that are found in SMA patient alleles. (B) Illustration of mild SMA causing missense mutations and their mode of action. Homomeric complexes of missense mutations do not rescue Smn ^−/− mice or perform snRNP assembly. In the presence of SMN2 , however, small amounts of full-length SMN are produced and heteromeric complexes with SMN protein and the missense mutation are formed. These heteromeric complexes are functional, perform snRNP assembly, and rescue Smn ^−/− mice.

Although the form of SMN2 is critically important in determining the severity of the SMA phenotype there are certainly trans-acting modifiers. Evidence for trans-acting factors comes from the observation of haploidentical siblings that have a marked difference in phenotype but identical SMN2 copy number. We and others have previously identified a number of these families where one sibling has type 1 and the other has type 2 SMA, or more commonly type 2 and mild type 3 (usually 3b or milder) and finally, type 3 and unaffected. Modifiers could act to ameliorate the phenotype by acting in critical pathways downstream of reduced SMN or they can act by altering the amount of SMN produced by SMN2 . Currently we do not know the identity of the trans-acting modifier(s) or how they act. Plastin 3 ( PLS3 ), located on the X-chromosome, has been suggested to be a modifier of SMA. This was based on the observation that lymphoblasts from the milder sibling had a higher level of PLS3 expression. However, the reason for this increased expression is not clear as no DNA variant has been reported. Furthermore, no variance in X inactivation or epigenetic regulation of PLS3 has been demonstrated in these patients. Curiously, increased expression of PLS3 also occurs in males but does not modify the SMA phenotype, and female patients with high PLS3 expression with the more severe SMA phenotype in the sibling pair have been reported. One possibility is the modifier PLS3 is female dependent and not completely penetrant. An alternative explanation is that PLS3 is in fact not the critical modifier of SMA phenotype. As no DNA change in PLS3 or an activator of PLS3 expression that segregates with the mild sibling has been reported, the role of PLS3 overexpression in SMA remains controversial. Finding a DNA change that accounts for the alteration in PLS3 expression will allow a more straightforward approach to testing its importance in a larger patient population. In this regard studies on reoccurrence of colon cancer have clearly identified certain variants that alter PLS3 expression. High PLS3 levels have been found in Sezary Syndrome patients and this was associated with loss of CD26. In addition, PLS3 -positive cells showed hypomethylation of the PLS3 CpG island at sites 95–99. Interestingly, the polymorphism SNP PLS3 rs871773T allele is associated with a higher protein expression of the PLS3 gene in colon cancer and an increased risk of recurrence of colon cancer. If PLS3 does alter severity of SMA, defining the role of both PLS3 rs871773 and the hypomethylation of sites 95–99 will be important as it gives a mechanism of PLS3 activation and may even result in a DNA marker that could be followed in patient material.

SMN1 loss accounts for 95% of SMA cases. In the remaining 5% of cases the SMN1 gene contains a small mutation. Most of the mutations disrupt the SMN reading frame or disrupt a splice site in the gene and are thus similar to the loss of SMN1 . However, many missense mutations that disrupt SMN function have been reported. These mutations are diagramed Fig. 6.3(A) along with the domains they disrupt and the severity of each mutation. Insights can be gained by studying how these missense mutations function upon disruption of SMN domains. At least some of the severe missense mutations disrupt the ability of SMN to efficiently oligomerize and therefore act much like SMN lacking the sequence encoded by exon 7. This form of SMN is rapidly degraded and results in minimal amounts of SMN protein. An example of a missense mutation like this is p.SMNY272C, which results in a severe phenotype in the presence of two copies of SMN2 . Patient lymphoblasts containing the p.SMNY272C mutation have the same amount of SMN protein as type 1 SMA patients with two copies of SMN2 . This indicates that minimal amounts of SMN are produced by the p.SMNY272C allele. Other severe mutations result in unfolding of protein domains including the Tudor domain. Perhaps one of the most interesting severe missense mutations is p.SMNE134K, which appears to disrupt the binding of SmD1/HuD proteins. This mutation could be used to dissect the contributions of different SMN assembly reactions to the phenotype (see below for SMN function).

We then come to the mild SMN missense mutations. These mutations occur in similar domains ( Fig. 6.3 ) yet produce a mild SMA phenotype. This raises the question: is a homomeric complex just composed of mutant SMNs functional, or is the small amount of full-length SMN protein produced by SMN2 required to complement the missense mutations? We have found that two mutations p.SMNA2G and p.SMNA111G, both mild mutations, can rescue the survival of Smn ^−/− mice only in the presence of full-length SMN and therefore do not have function on their own. Furthermore, these two amino-terminal mutations also do not complement each other in the absence of full-length SMN. The full-length SMN protein from SMN2 oligomerizes with the SMN missense mutation protein making a functional SMN complex. In summary, heteromeric SMN complexes are functional and homomeric complexes are not ( Fig. 6.3 ). The amount of full-length SMN and the amount of missense SMN expressed then becomes critical in determining phenotype as it appears likely that only one monomer of wild-type full-length SMN per complex is required for function. While this finding was first demonstrated with p.SMNA2G and p.SMNA111G, it also applies to the mutations p.SMND44V, p.SMNT274I, and p.SMNQ282A as well, and could be a general phenomenon of all mild missense alleles. Indeed, while missense alleles are rare, no consanguineous cases with the absence of SMN2 have ever been reported.

One current conundrum in the field is the behavior of equivalent missense mutations in nonmammalian species. In Caenorhabditis elegans the equivalent of p.SMND44V (Smn.D27N) has been examined. Animals with this mutation can survive and offspring can be obtained in the absence of wild-type SMN. Homozygous mutants show various phenotypes including a motility defect. However this invertebrate finding is different than the observation in mammals where p.SMND44V does not rescue Smn ^−/− mice in the absence of full-length SMN from SMN2 . At least some of the mild SMN alleles have been analyzed in Drosophila and appear to complement the maternal SMN present in these mutants. Although some severe alleles behave in the expected manner, others such as Smn-pSMN.G210V (equivalent to p.SMNG279V, a severe allele in humans) do not. The exact reasons for these differences among species remain unclear.