are hallmarks of 238 Wirth et al. Annu. Rev. Genom. Hum. Genet. 2020.21:231-261. Downloaded from www.annualreviews.org Access provided by 2600:1700:3c30:a3d0:89dc:b3d1:ac73:5085 on 01/23/22. For personal use only. SMA (83, 150). This finding fits well with the three SMA protective modifiers, PLS3, NCALD, and CHP1, which bind Ca2+ and are either Ca2+ sensors or regulators (71, 84, 142). Calcium is essential for many processes at the NMJ level, including exocytosis and endocytosis. Interestingly, all three SMA protective modifiers are able to rescue impaired endocytosis in SMA cells and animal models (40, 71, 84, 142). Cytoskeletal Dynamics, Endocytosis, Mitochondrial Function, and Local Translation Although SMN is a housekeeping protein, reduced levels of 20–40%, as found in SMA individuals (68, 101), primarily affect the NMJs innervated by alpha motor neurons. Therefore, cellular pathways other than snRNP biogenesis and splicing might be tissue-specifically impaired in alpha motor neurons (62 and references therein). Whether a motor neuron–specific splicing defect is responsible for reduced SMN levels is still an enigma. Moreover, it remains unknown why some spinal motor neurons and certain regions of the spinal cord are more vulnerable than others. In general, it seems that larger neurons are more vulnerable, as they might rely on higher SMN amounts. Indeed, the gene name “survival of motor neuron” has been perfectly chosen. One of the earliest findings in SMA pathology was defective NMJ development and maturation accompanied by neurofilament accumulation (29). Studies have also reported synaptic vesicle formation and transmission defects; these defects depend heavily on actin dynamics, which is disturbed in SMA (1, 20, 40, 71, 145). F-actin, which is reduced in SMA, is essential for all types of endocytosis, a crucial process in neurotransmission. Indeed, reduced SMN levels impair endocytosis at the NMJ level and can be restored or ameliorated by the SMA modifiers PLS3, NCALD, and CHP1 (71, 84, 142). Transport and local translation of mRNAs such as ACTB, GAP43, and CARM1, which are crucial for axonal development and function, are reduced in SMA (3, 98, 145, 152). Since late endosomes are associated with local translation and mitochondrial function in axons (30), and these processes are also affected in SMA (2, 13, 122), one can envisage that SMN reduction disturbs the supply of specific mRNAs in the axon, impairing a plethora of processes connected to proper NMJ function. GENETIC TESTING AND PHENOTYPE–GENOTYPE CORRELATION The mutation spectrum in SMA patients is remarkably unique, with 96% showing a homozygous absence of SMN1 and 4% carrying point mutations (180). Due to the complex genomic structure, gene conversion and de novo rearrangements occur quite frequently. Besides the inverse correlation between disease severity and SMN2 copy number, additional variants within the SMN2 gene or independent modifiers, such as PLS3 or NCALD, can further influence disease severity (129, 138, 142). Genetic Testing The gold standard genetic testing for SMA is multiplex ligation-dependent probe amplification of SMN1 and SMN2 (10, 118). This method allows the identification of SMA patients with a homozygous SMN1 deletion; SMA patients with one SMN1 copy, who might be compound heterozygous for a second, subtle SMN1 variant; the exact number of SMN2 copies; and healthy heterozygous carriers. It fails to differentiate between individuals with two SMN1 copies on each chromosome 5 (cis version) and individuals with one SMN1 gene on each chromosome 5 (trans www.annualreviews.org • 25 Years of Spinal Muscular Atrophy Research 239 Annu. Rev. Genom. Hum. Genet. 2020.21:231-261. Downloaded from www.annualreviews.org Access provided by 2600:1700:3c30:a3d0:89dc:b3d1:ac73:5085 on 01/23/22. For personal use only. version). It also does not identify subtle mutations in SMN1 (6% of SMA patients). Therefore, an individual with two SMN1 copies can still be an SMA carrier (false-negative rate of ∼5%) (180). The existence of two SMN genes in the human genome hinders the search for variants in SMN1. There are two ways to identify