Spinal muscular atrophy (SMA) is an autosomal recessive disorder caused by mutations in the SMN1 gene (MIM# 600354).1 SMN2 (MIM# 601627), an SMN1 homologue, typically differs from SMN1 by five nucleotides (in intron 6, exon 7, intron 7, and noncoding exon 8).2 Only one of these nucleotide changes (840C→T) is in the coding sequence, and it is translationally silent.2 Approximately 94% of clinically typical SMA patients lack both copies of SMN1 exon 7,3 and most carriers have only one copy of SMN1 exon 7, as determined by SMN gene dosage analysis.4,5 In addition to large deletions that include the entire SMN1 gene, loss of SMN1 exon 7 can occur by gene conversion from SMN1 to SMN2.6 SMA type III patients have, on average, more SMN2 copies than SMA type II or type I patients,7,8,9 and hence more copies of SMN2 derived by conversion from SMN1. The copy number of SMN2 correlates with longer survival and inversely with disease severity.8,9 Although gene conversion from SMN1 to SMN2 had been demonstrated, gene conversion from SMN2 to SMN1 has not. We show herein that increased SMN1 copy number is associated with decreased SMN2 copy number in the general population, which provides evidence that gene conversion from SMN2 to SMN1 occurs, and that SMN1 converted from SMN2 is present in the general population.

Materials and methods

Sample collection and DNA extraction

Using Puregene reagents (Gentra Systems, Minneapolis, MN, USA), DNA was extracted from peripheral blood specimens that were received with informed consent by the Molecular Pathology Laboratory of the Hospital of the University of Pennsylvania for SMN analyses on a clinical basis. We used results from all available individuals with no family history of SMA, (N=180)10 and 107 parents of an SMA patient; either parents with only one copy of SMN1, or parents of an SMA child who lacked SMN1 or had only one copy of SMN1 with a high clinical index of suspicion for SMN1-related SMA. Results were anonymized and used for our study. We also randomly selected and anonymized samples from 32 SMA patients with a known clinical type for our study.

SMN1 and SMN2 copy number assay (SMN gene dosage analysis)

A nonradioisotopic SMN gene dosage assay,7 a modification of the original method of McAndrew et al,4 was modified further, and validated for the determination of SMN1 and SMN2 copy numbers.11,12 SMN1 exon 7 copy number is used as a surrogate for SMN1 gene copy number. Briefly, SMN1 exon 7, SMN2 exon 7, a genomic two-copy reference (CFTR exon 4), and SMN and CFTR internal standards were coamplified in a single reaction. The PCR products were incubated with DraI, which digests only the SMN2 products, but not the SMN1 products. Copy number of SMN1 or SMN2 per cell (or more precisely, per diploid genome) was determined by quantification of the digested PCR products, followed by normalization utilizing the genomic standard, the internal standards, and five control samples with known SMN1 and SMN2 copy numbers. The quantitative accuracy of this method was validated as described.7,11,12 All test samples were analyzed in duplicate.


Distributions of SMN genotypes in unaffected individuals are shown in Table 1. SMN2 copy number and SMN1 copy number correlated inversely in the general population. SMN2 copy number was decreased to one or zero copies in 11 of 13 individuals with three or four copies of SMN1, whereas only 71 (43%) of 164 individuals with 2 SMN1 copies had one or zero SMN2 copies (P < 0.01; χ2 test).

Table 1 SMN genotype distributions in unaffected individuals

There was a notable tendency for SMA type I carriers to have fewer SMN2 copies, compared to type II and type III carriers (Table 1). While only three (1.7%) of 177 noncarriers (in the column indicated by ‘No FH’ in Table 1) with two or more SMN1 copies had three SMN2 copies, SMN2 copy number was increased to three or four copies in 19 (17%) of 109 carriers with SMN1 deletion/conversion mutations (P < 0.001; χ2 test).

The distribution of SMN genotype in 32 SMA patients is as follows (with genotype expressed as ‘(SMN1 copy number):(SMN2 copy number)’): Among the 16 type patients, three were 0:1 and 13 were 0:2; the single type I-II patient was 0:2; among the eight type II patients, one was 0:2 and seven were 0:3 and one was 0.4; and one patient with very mild adult-onset SMA (which may be called type IV) was 0:4. The SMN2 copy number correlated inversely with disease severity.


Using SMN gene dosage analysis, three or more copies of SMN1 were detected in some individuals in the general population, indicating the presence of chromosome 5s with two copies of SMN1.4,10 We meta-analyzed published data,3,4,8,10,13,14 and updated deduced SMN1 allele frequencies5 as follows: ‘zero-copy allele’ (chromosome 5 lacking SMN1 exon 7), 9.83 × 10−3; ‘one-copy allele’, 9.57 × 10−1; ‘two-copy allele’ (chromosome 5 with two copies of SMN1 exon 7), 3.27 × 10−2; and ‘1D allele’ (chromosome 5 with a small intragenic mutation in SMN1), 1.80 × 10−4.

One hypothesis to explain the presence of two copies of SMN1 on one chromosome 5 is unequal crossing over between homologous chromosomes during meiosis. Because of the presence of a large inverted repeat in the 5q13 region, and multiple smaller repeats contained therein, the SMN locus is considered highly susceptible to recombination. And, in fact, studies have shown that unequal crossing over at the SMN1 locus can cause de novo deletions of SMN1.7,15

An alternative hypothesis is that gene conversion from SMN2 to SMN1 can result in two SMN1 copies on one chromosome 5. In this scenario, SMN2 copy number would decrease after the gene conversion, in contrast to most unequal crossover events at the SMN1 locus. Our SMN1- and SMN2-copy-number data in the general population support this hypothesis. Our SMN2-copy-number data among SMA carriers and patients also support the hypo-thesis of the gene conversion from SMN1 to SMN2. Previous studies indicated that increased SMN2 copy number due to gene conversion from SMN1 to SMN2 is associated with a milder SMA phenotype.6,7,8,9

van der Steege et al16 studied three nucleotide differences in intron 6, exon 7, and exon 8, and found a hybrid SMN1/SMN2 gene (SMN1 intron 6/exon 7-SMN2 exon 8; intron 7 unknown) in one of their controls. Hahnen et al17 studied five nucleotide differences in intron 6, exon 7, intron 7, and exon 8, and found a hybrid gene (SMN1 intron 6/exon 7/intron 7-SMN2 exon 8) in one of their controls. The presence of these rare hybrid genes may support the hypothesis of gene conversions in either direction between SMN1 and SMN2, since a vast majority of hybrid genes present in SMA patients (presumably due to SMN1-to-SMN2 gene conversion6) had only SMN2 sequences except for one nucleotide in exon 8 (SMN2 intron 6/exon 7/intron 7-SMN1 exon 8).17 Other minor hybrid gene variants have also been described.17,18,19

In conclusion, our data provide population-based evidence of gene conversion from SMN2 to SMN1. Additional studies are necessary to confirm the hypothesis of gene conversion from SMN2 to SMN1.