Deep Molecular Characterization of Milder Spinal Muscular Atrophy Patients Carrying the c.859G>C Variant in SMN2

Spinal muscular atrophy (SMA) is a severe neuromuscular disorder caused by biallelic loss or pathogenic variants in the SMN1 gene. Copy number and modifier intragenic variants in SMN2, an almost identical paralog gene of SMN1, are known to influence the amount of complete SMN proteins. Therefore, SMN2 is considered the main phenotypic modifier of SMA, although genotype–phenotype correlation is not absolute. We present eleven unrelated SMA patients with milder phenotypes carrying the c.859G>C-positive modifier variant in SMN2. All were studied by a specific NGS method to allow a deep characterization of the entire SMN region. Analysis of two homozygous cases for the variant allowed us to identify a specific haplotype, Smn2-859C.1, in association with c.859G>C. Two other cases with the c.859G>C variant in their two SMN2 copies showed a second haplotype, Smn2-859C.2, in cis with Smn2-859C.1, assembling a more complex allele. We also identified a previously unreported variant in intron 2a exclusively linked to the Smn2-859C.1 haplotype (c.154-1141G>A), further suggesting that this region has been ancestrally conserved. The deep molecular characterization of SMN2 in our cohort highlights the importance of testing c.859G>C, as well as accurately assessing the SMN2 region in SMA patients to gain insight into the complex genotype–phenotype correlations and improve prognostic outcomes.


Introduction
Spinal muscular atrophy (SMA) is a neuromuscular disorder characterized by the degeneration and loss of alpha motor neurons in the spinal cord anterior horns, leading to progressive atrophy of proximal muscles, weakness, respiratory failure, and even death. It is the second most common recessive genetic disease of infancy and early childhood with an incidence around 1:11,000 live births and a carrier frequency of 1:51 worldwide [1,2].
SMA patients are mainly classified into five clinical groups on the basis of age of onset, achieved motor milestones, and clinical severity. Type 0 or congenital, the most severe, appears prenatally, and the patient's life expectancy is very short, usually a few weeks or months. Patients with type I, or Werdnig-Hoffmann disease (onset within the first six months of life), are never able to sit unsupported and generally do not survive beyond the age of two years. In the intermediate SMA type II (onset between 6 and 18 months of life), children acquire the ability to sit unsupported, but they never walk unaided and usually reach adolescence. Type III patients (Kugelberg-Welander disease) walk independently for a long time but eventually become wheelchair-bound. They can be further subdivided into type IIIa and IIIb depending on the age of disease onset (before or after three years of age). Finally, patients with SMA type IV present an adult onset and milder disease course [3][4][5]. It is important to bear in mind that current SMA therapies can modify the trajectory of SMA patients; therefore, this classification is mainly applied on clinical data prior to treatment [6,7].
At the molecular level, SMA is caused by the loss or mutation of both copies of the survival of motor neuron 1 (SMN1) gene, which encodes the survival motor neuron protein (SMN). In most cases, the disease is due to the homozygous absence of SMN1 (95%), although pathogenic point variants have also been described [8][9][10].
Adjacent to SMN1, in a more centromeric position, lies SMN2, an almost identical paralog gene generated by a segmental duplication [11]. The fact that SMN2 is present in humans and not in any other species suggests that the duplication of SMN1 occurred recently in time. Consequently, the homology between both genes is extremely high, differing only in 16 positions called paralogous sequence variants (PSVs) [12,13]. This makes the region highly unstable, which leads to genomic instability predisposing to gene deletions, duplications, and conversions between both genes. Indeed, SMN1 and SMN2 genes can be present in multiple copies in the general population, both in cis and trans configuration [14].
Theoretically, the SMN2 gene encodes the same protein as SMN1, but one of the PSVs, a silent transition in exon 7, alters the splicing pattern in most SMN2 pre-mRNA transcripts. This causes the skipping of exon 7, resulting in a non-functional protein (SMN-∆7) instead of the full-length protein [13]. As SMN-∆7 is highly unstable and rapidly degraded, it is unable to compensate the absence or deficiency of SMN1 in SMA patients [15]. It has been reported that each copy of SMN2 can only produce about 10-15% of functional SMN proteins [16][17][18], being the number of SMN2 copies the main modifier of SMA disease described to date.
Concretely, an inverse correlation between the number of SMN2 copies and the severity of the phenotype has been widely reported, given that the higher the number of SMN2 copies producing SMN functional protein, the milder the SMA phenotype [19][20][21]. Nevertheless, this correlation is not absolute, since discordant patients have been described in the literature, further classified as better-than-expected or worse-than-expected phenotypes according to their SMN2 copy number [19,21]. It is known that the presence of the c.859G>C and c.835-44A>G (-44G) variants, located in exon 7 and intron 6 of SMN2, respectively, explains some of the better-than-expected discordant phenotypes. These SNVs, considered positive modifiers of SMA disease, increase the inclusion of exon 7 and therefore generate greater amounts of functional SMN protein [21][22][23][24][25].
The full characterization of SMN2, including dosage and structure, will be more relevant in the current scenario where new therapies for SMA are being implemented. It is well known that SMN2 dosage is the main modifier of SMA, but it seems that this could be just the tip of the iceberg of a much more complicated framework. Indeed, all differences between SMN1 and SMN2 can be revealed by specific NGS studies [12]. It is also possible that these findings may relate to phenotype variability or to SMN2-specific treatment response [20].
In this work, we performed an in-depth characterization of the SMN region in eleven SMA patients carrying the c.859G>C modifier variant in the SMN2 gene (SMN2 859C ) and presenting a milder phenotype. By defining the genetic background of SMN2 859C , we discovered the existence of a common haplotype alongside the SMN2 gene in linkage disequilibrium with the variant and a second less common haplotype harboring two SMN2 859C copies in cis.

Clinical and Molecular Characterization of Patients
All SMA patients described in this study (ten males and one female) presented a biallelic absence of SMN1 as the determinant of SMA and shared the presence of at least one copy of SMN2 859C . Seven of these individuals carried two SMN2 copies, including five with the c.859G>C modifier variant in their two SMN2 genes (patients 1 to 5) and two with the variant in only one SMN2 (patients 6 and 7). The other four patients presented three SMN2 copies, and the variant was only present in one of their SMN2 alleles (patients 8 to 11). A summary of the clinical and molecular data of the patients is shown in Table 1. Our cohort comprised SMA patients of Spanish, Italian, Danish, and Chilean origins, and the majority were classified as SMA type IIIb (8/11) and the remaining patients as SMA type IIIa (2/11). The remaining case (Patient 6 in Table 1) was classified as type II based on his age of onset, which was prior to 18 months. Currently, at three years of age, he has not yet achieved independent ambulation. * In addition, presents a partial SMN gene comprising only exons 1 to 6 (SMN1/2∆7-8) (see Figure 1).

Haplotype Characterization by Deep Sequencing of SMN2 Genes
NGS data confirmed the biallelic absence of the entire SMN1 gene in all patients, since specific nucleotides of SMN2 were found in homozygous state (AB ratio of 100%) in all PSV positions. Similarly, the NGS results corroborated the SMN2 copy number previously assigned by MLPA via the AB ratio analysis of all the different variants detected in the SMN region of each patient. In patients with two SMN2 copies (except for patient 5), all variants were detected with an approximate allele frequency of 50% or 100%, whereas in patients with three SMN2 copies, variants were found at a frequency of around 33%, 66%, or 100% (data not shown, available upon request). Patient 5 was a special case where variants were observed at a frequency of around 33-66-100% in the 5 region and around 50-100% frequency in the 3 region. This phenomenon was due to the presence of two complete SMN2 genes and a partial SMN gene comprising exons 1 to 6 (SMN1/2∆7-8) (see Table 2). In addition, the AB analysis of all patients confirmed the copy number of the c.859G>C modifier in each case.
Overall, our 11 patients represented 16 alleles with the c.859G>C variant, including five cases with two SMN2 859C and the remainder with just one allele with the variant ( Table 2).

Establishment of Two Haplotypes Associated with the c.859G>C Modifier Variant
We initially performed an in-depth analysis of the complete SMN2 region in patients 1 and 2, who carried two SMN2 859C genes and had consanguineous parents. The studies revealed that both patients were completely homozygous for the entire studied region and identical between them. Thus, we were able to determine the specific SMN2 sequence associated with the c.859G>C modifier in their alleles, establishing a haplotype called Smn2-859C.1 (Table 2). Similarly, sequencing results in patient 3 revealed an almost identical sequence to Smn2-859C.1 in his two SMN2 genes, with the exception of one rare variant (69356349-A-G) with an allele frequency of~50%. In contrast, patients 4 and 5, who also presented two SMN2 859C copies, showed several variants in only one of their SMN2 859C along the studied region. Nonetheless, it was possible to infer that one of their SMN2 genes matched the sequence of the Smn2-859C.1 haplotype. Interestingly, in both patients, it was possible to assume a second haplotype associated with the c.859G>C variant that we defined as Smn2-859C.2 (Table 2). Applying this preliminary information, the Smn2-859C.1 haplotype was also inferred in one of the SMN2 copies of the remaining patients (patients 6 to 11), with few discrepant positions in patients 7, 10, and 11 (see Table 2).
To explore deeper into the structure of the SMN2 genes, co-segregation studies from patients with two SMN2 859C were carried out through MLPA together with NGS or allelespecific PCR. These investigations showed that patient 2 carried his two Smn2-859C.1 haplotypes in trans, inheriting one from each progenitor ( Figure 1B). Patient 1 s co-segregation was incomplete, as a sample from his father was not available, but this family was consanguineous, and the mother only presented one Smn2-859C.1 haplotype. Therefore, we could assume that his father also presented one Smn2-859C.1 haplotype, and he should harbor both Smn2-859C.1 haplotypes in trans ( Figure 1A). In contrast, the co-segregation study in patient 3 revealed that both Smn2-859C.1 haplotypes were in cis, forming a complex allele inherited from the mother ( Figure 1C). Co-segregation in patients 4 and 5 indicated that the two SMN2 859C genes (Smn2-859C.1 and Smn2-859C.2 haplotypes) were located in cis. Specifically, patient 4 inherited this complex allele from his father and a null allele (without SMN1 and SMN2) from his mother ( Figure 1D), while patient 5 inherited the complex allele from his mother and the other allele with a partial non-functional SMN1/2∆7-8 gene from his father ( Figure 1E).   T T T T T T  T  T  T  T  T  T  T  T  T  T  T  T  T  T  T T  T  T  69343570-G-T  T  T  T T T T T T T T T  T  T  T  G  T  T  T  T  T  T  T  T  T  T  T T T Figure 1. Pedigree representation of cases with two SMN2 859C copies. According to segregation studies, a cis or trans configuration was defined in each patient. (A) In patient 1, the father's sample was not available, and SMN2 configuration was inferred based on the results from the mother and patient. Given that the father did not present symptoms, we can assume that he carries at least one SMN1 gene. In addition, being a consanguineous family, we assumed that Smn2-859.C1 was transmitted by both parents (untested inferred alleles are represented by a dashed line). (B) In patient 2, the Smn2-859.C1 haplotype was inherited from both parents, in agreement with the consanguinity in the family. (C) Patient 3 had two copies of SMN2 with Smn2-859.C1 in cis, inherited from his mother. (D) Patient 4 also had two copies of SMN2 in cis, one with Smn2-859.C1 and the other with Smn2-859.C2 haplotype, forming a complex allele inherited from the father. (E) Patient 5 inherited the complex allele from his mother and the other allele with a partial non-functional SMN1/2Δ7-8 gene from his father.  [26][27][28]. In silico analysis of this deep intronic variant using the software SpliceAI [29], Alamut Visual Figure 1. Pedigree representation of cases with two SMN2 859C copies. According to segregation studies, a cis or trans configuration was defined in each patient. (A) In patient 1, the father's sample was not available, and SMN2 configuration was inferred based on the results from the mother and patient. Given that the father did not present symptoms, we can assume that he carries at least one SMN1 gene. In addition, being a consanguineous family, we assumed that Smn2-859.C1 was transmitted by both parents (untested inferred alleles are represented by a dashed line). (B) In patient 2, the Smn2-859.C1 haplotype was inherited from both parents, in agreement with the consanguinity in the family. (C) Patient 3 had two copies of SMN2 with Smn2-859.C1 in cis, inherited from his mother. (D) Patient 4 also had two copies of SMN2 in cis, one with Smn2-859.C1 and the other with Smn2-859.C2 haplotype, forming a complex allele inherited from the father. (E) Patient 5 inherited the complex allele from his mother and the other allele with a partial non-functional SMN1/2∆7-8 gene from his father.

Difference between Haplotypes and Detection of a Novel Variant Exclusively
All together, these results indicated that the Smn2-859C.1 haplotype was consistent in our cohort, since all patients presented it in association with the c.859G>C variant, either as a single allele or as part of a more complex allele formed by the Smn2-859C.1 and Smn2-859C.2 haplotypes in cis. Based on our 11 SMA patients, we have not observed any clinical difference between Smn2-859C.1 and Smn2-859C.2 haplotypes, although we only found two cases carrying the Smn2-859C.2 haplotype. Interestingly, we noticed the presence of a novel variant, c.154-1141G>A (69360651-G-A, hg19/GRCh37), located in intron 2a (Table 2). This variant was detected in all patients with the Smn2-859C.1 haplotype but absent in the Smn2-859C.2 haplotype. Moreover, this variant was not detected in the 338 SMA patients without the c.859G>C variant. The c.154-1141G>A change has not been reported in the general population according to gnomAD, ISB Kaviar3, and Bravo (as of 18 July 2022) [26][27][28]. In silico analysis of this deep intronic variant using the software SpliceAI [29], Alamut Visual Software version 2.11 (SOPHiA GENETICS), and ESRseq [30] did not predict an effect on the splicing process.

Discussion
Here, we present 11 patients with a clinical and molecular diagnosis of SMA caused by the biallelic absence of SMN1 and with a milder phenotype explained by at least one SMN2 859C gene, given that 10 out of 11 patients were walkers. We identified a specific sequence, named Smn2-859C.1, present in all patients from our cohort in linkage disequilibrium with the c.859G>C variant. In addition, two cases showed a more complex allele, assembled by Smn2-859C.1 and Smn2-859C.2 in cis.
In order to study the genetic origin of the c.859G>C variant, we expanded our cohort of Spanish cases with patients from Denmark, Italy, and Chile. We applied NGS methodologies exclusively focused on the SMN region to determine the exact sequence of SMN2 associated with the c.859G>C variant in each patient [12]. By studying the patients with two SMN2 859C , we were able to determine two haplotypes associated with the variant, Smn2-859C.1 and Smn2-859C.2. The Smn2-859C.1 haplotype, with minor modifications, was present in all 11 patients (14/16 SMN2 859C alleles), either in cis or trans configuration, while the Smn2-859C.2 haplotype was only found in two patients (2/16 SMN2 859C alleles), always in cis configuration with the Smn2-859C.1 haplotype (Figure 2). Notably, no patient was found to harbor the c.859G>C variant in association with any other haplotype, regardless of their ethnic lineage, which points towards a common ancestral origin in all cases. three SMN2 copies and the variant in more than one allele and, in fact, no patient with this genotype has been described in the literature either. This could be due to the fact that patients with three SMN2 copies showing SMA type II or III are not currently tested for the variant. Another reason could be that cases with this genotype perhaps do not manifest clear disease symptoms due to the higher production of SMN protein and therefore may never be diagnosed. Similarly, it has been previously speculated that some individuals with zero SMN1 and four or five SMN2 copies may present minimal symptoms or be asymptomatic throughout their lives, remaining undetected [35]. This corroborates the importance of implementing detection of the c.859G>C-positive modifier as part of the genetic diagnosis routine in SMA. 859G>C. An additive effect on SMA phenotype is observed depending on whether the c.859G>C variant is found in one or both SMN2 copies. SMN2 gene is represented as a rectangle, and the presence of the c.859G>C variant in exon 7 is indicated by an asterisk. Not all SMN2 genotypes represented in this figure were detected in this study (see Figure 1 for more details).
At this level of analysis and based on the clinical information available for each patient, we did not observe categorical phenotypic differences between the Smn2-859C.1 or Smn2-859C.2 haplotypes, nor the cis or trans configuration of the Smn2-859C.1 haplotype, since all cases with two SMN2 859C copies presented a milder phenotype (IIIb). Interestingly, patient 2, with the exact same sequence and configuration as patient 1, also developed type IIIb SMA, but his onset was noted earlier in comparison with patient 1 and the remaining cases with two SMN2 859C . At present, we are unable to explain this minor disparity considering all the studies performed in SMN2. Thus, this fact suggests disease onset could also be conditioned by as yet unknown factors, other than SMN2 structure. 859G>C. An additive effect on SMA phenotype is observed depending on whether the c.859G>C variant is found in one or both SMN2 copies. SMN2 gene is represented as a rectangle, and the presence of the c.859G>C variant in exon 7 is indicated by an asterisk. Not all SMN2 genotypes represented in this figure were detected in this study (see Figure 1 for more details).
The c.859G>C variant has been previously reported to increase the inclusion of SMN2 exon 7 by 20%, which leads to the generation of higher amounts of functional proteins than the wild-type SMN2 gene [22,25]. Patients carrying this variant developed milder SMA phenotypes compared with those with the same SMN2 copy number but without the variant [23]. In our case, the deep characterization of the entire SMN region supports that SMN2 859C is, at first sight, primarily responsible for the milder phenotype in our patients.
To date, together with our six newly described patients, a total of 44 patients carrying c.859G>C have been reported worldwide, including a patient recently detected by newborn screening [21][22][23]25,[31][32][33][34]. In general population databases, the c.859G>C variant is reported at a frequency of approximately 0.3% with 132 homozygotes detected [26]. However, it is possible that the data are not accurate given the high homology between SMN1 and SMN2 and their copy number variability, which poses a challenge in the analysis and proper annotation of the SMN region with non-specific NGS techniques, such as exome or genome. Nevertheless, it is possible to estimate the frequency of this variant in the SMA population based on previous studies. According to the data of Calucho et al. (2018) [19], the allelic frequency of the c.859G>C variant is 1.04% (13/1250 alleles) in a series of 625 Spanish SMA patients. In fact, in this cohort, approximately 25% of better-than-expected cases with two SMN2 copies carried the variant [19]. Although c.859G>C appears to be relatively uncommon, at present it is not routinely tested in SMA patients, deserving more studies to clearly establish its incidence.
Concerning clinical classification, patients with two SMN2 genes usually debut in the first six months of life and are classified as SMA type I [19]. In our series, patients with two SMN2 copies and the positive modifier presented at least type II or type III disease (Table 1). Furthermore, an additive effect was observed since patients with the c.859G>C change in both SMN2 genes had a better phenotype than patients carrying the variant only in one SMN2, confirming previous observations [23] (Figure 2). For instance, patient 3 (with two SMN2 859C copies) developed the first SMA symptoms at 12 years of age, being classified as type IIIb, whereas patient 7 (with the variant in one of his SMN2 genes) had manifestations at 18 months of life with a clinical diagnosis of type IIIa. Regarding cases with three SMN2 copies, all patients presented the c.859G>C variant in only one of their alleles, developing a type III phenotype. Interestingly, we did not find any patient with three SMN2 copies and the variant in more than one allele and, in fact, no patient with this genotype has been described in the literature either. This could be due to the fact that patients with three SMN2 copies showing SMA type II or III are not currently tested for the variant. Another reason could be that cases with this genotype perhaps do not manifest clear disease symptoms due to the higher production of SMN protein and therefore may never be diagnosed. Similarly, it has been previously speculated that some individuals with zero SMN1 and four or five SMN2 copies may present minimal symptoms or be asymptomatic throughout their lives, remaining undetected [35]. This corroborates the importance of implementing detection of the c.859G>C-positive modifier as part of the genetic diagnosis routine in SMA.
At this level of analysis and based on the clinical information available for each patient, we did not observe categorical phenotypic differences between the Smn2-859C.1 or Smn2-859C.2 haplotypes, nor the cis or trans configuration of the Smn2-859C.1 haplotype, since all cases with two SMN2 859C copies presented a milder phenotype (IIIb). Interestingly, patient 2, with the exact same sequence and configuration as patient 1, also developed type IIIb SMA, but his onset was noted earlier in comparison with patient 1 and the remaining cases with two SMN2 859C . At present, we are unable to explain this minor disparity considering all the studies performed in SMN2. Thus, this fact suggests disease onset could also be conditioned by as yet unknown factors, other than SMN2 structure.
As mentioned above, the Smn2-859C.2 haplotype was detected in cis configuration with respect to Smn2-859C.1, assembling a complex allele containing two different SMN2 859C genes (Figure 1). These two haplotypes differ in several positions, but an identical block of at least 8848 bp around c.859G>C is present in both (Table 2 and Figure 3B). This observation, together with the fact that we also detected an allele formed by two Smn2-859C.1 haplotypes in cis, points towards a possible origin of the complex allele through homologous recombination, implicating a double cross-over event [36]. In this event, two alleles would be involved ( Figure 3A): allele A, consisting of two SMN2 genes with the Smn2-859C.1 haplotype, and allele B, formed by at least one SMN2 with an unknown haplotype containing part of Smn2-859C.2 but without the c.859G>C variant. In the double homologous recombination process, allele A would maintain both c.859G>C variants as well as gain the part of the Smn2-859C.2 haplotype from allele B, generating the complex allele that we detected in our patients ( Figure 3B).  Table 2).
Finally, it should be noted that the Smn2-859C.1 haplotype contains the novel variant c.154-1141G>A, located in intron 2a. According to our results, this variant is in linkage disequilibrium with the c.859G>C modifier given that, in our larger cohort of 349 SMA patients, it was only detected in those carrying the c.859G>C variant, and it was not found in population databases. This observation suggests that the sequence between this variant and the c.859G>C modifier has been ancestrally conserved. In silico splicing tools did not predict any specific effect of this deep intronic variant. However, we could not rule out some influence of this change, given the limitations of splicing predictors; thus, it deserves further investigation.
Our NGS approach to characterize these patients revealed new information that could be relevant for the different functions and/or alterations of SMN2. It is important to consider whether the function and expression of SMN2 is not only modified depending on the cis or trans configuration of SMN2 859C but also on the presence of the Smn2-859C.1 or Smn2-859C.2 haplotype. Long regulators, cis-or trans-acting elements, may distinctively influence its function and/or expression according to the topography of the region.  Table 2).

Study Participants
Finally, it should be noted that the Smn2-859C.1 haplotype contains the novel variant c.154-1141G>A, located in intron 2a. According to our results, this variant is in linkage disequilibrium with the c.859G>C modifier given that, in our larger cohort of 349 SMA patients, it was only detected in those carrying the c.859G>C variant, and it was not found in population databases. This observation suggests that the sequence between this variant and the c.859G>C modifier has been ancestrally conserved. In silico splicing tools did not predict any specific effect of this deep intronic variant. However, we could not rule out some influence of this change, given the limitations of splicing predictors; thus, it deserves further investigation.
Our NGS approach to characterize these patients revealed new information that could be relevant for the different functions and/or alterations of SMN2. It is important to consider whether the function and expression of SMN2 is not only modified depending on the cis or trans configuration of SMN2 859C but also on the presence of the Smn2-859C.1 or Smn2-859C.2 haplotype. Long regulators, cis-or trans-acting elements, may distinctively influence its function and/or expression according to the topography of the region.

Study Participants
We studied eleven unrelated SMA patients from different international centers with the presence of at least one SMN2 859C gene. Patients were classified into SMA type according to age of onset, clinical severity, and achieved motor milestones, prior to receiving any modifying therapies. Criteria for correlating phenotype with SMN2 dosage were type I (non-sitters) with two SMN2 copies, type II (sitters) with three SMN2 copies, and type III (walkers) patients with three-four SMN2 copies [19]. Based on this model, our patients with two SMN2 copies were considered discordant, as none presented a type I SMA phenotype ( Table 1).
All patients were selected from a larger cohort of 349 SMA patients, undergoing an NGS study of the SMN region [12], based on the presence of the c.859G>C variant. Four of the patients were previously described as carriers of this variant (patients 1, 8, and 11 [23] and patient 7 [12]).
DNA samples were extracted from peripheral blood using standard methods. Ethics approval was granted by the Clinical Research Ethics Committee of Hospital Vall d'Hebron (Comité de Ética de Investigación con Medicamentos del Hospital Universitari Vall d'Hebron (PR(AG)229/2018)). Written informed consent was obtained from all participants or their parents/legal caregivers.

SMN2 Genotyping and Haplotype Characterization
All patients were genetically confirmed as SMA cases via previously described methods that also included testing SMN2 modifier variants [10,23,37]. A detailed molecular characterization of SMN2 was carried out in all patients by a specific NGS sequencing method [12].
In addition, to detect the presence of the c.859G>C variant in some progenitor samples, two specific PCRs were designed to amplify exons 7 and 8 of genes SMN1 and SMN2. The allele-specific PCR technique [38] was used to amplify both genes separately to ascertain in which gene the variant was present. Standard Sanger sequencing was performed with the PCR products, allowing us to detect the c.859G>C variant. These primers are also designed to study the c.835-44A>G variant. Primer sequences and PCR conditions are provided in Table S1.

Conclusions
This series of patients with milder phenotypes demonstrates the relevance of testing the c.859G>C variant in all SMA patients, with special consideration in cases with two or three SMN2 copies in the context of neonatal screening. Indeed, the presence of this rare variant in an asymptomatic neonate may help to predict a better phenotype by natural history per se, regardless of the therapeutic option chosen. This is crucial in order to evaluate the effects of the approved therapies to unmask long-term benefits in treated patients. Given that not all discordant cases can be explained by this positive variant, it is necessary to further analyze the SMN2 region by NGS to detect other reported candidate variants [24] and the presence of hybrid SMN1-SMN2 structures [20], as well as to unravel novel phenotypic modifier variants. In the current therapeutic context, genetic studies in patients confirmed with biallelic SMN1 absence or pathogenic variants should consider not only testing for SMN2 copies but also investigating SMN2 variants and structures as part of the integral characterization of patients receiving expensive and sometimes lifelong therapies. Informed Consent Statement: Written informed consent was obtained from all participants or their parents/legal caregivers. Data Availability Statement: All data and scripts used to generate the analyses of this paper are available upon request, unless the type of request compromises ethical standards or legal requirements.