Diverse genetic approaches have enabled rapid dissection of the complex genetic and cellular events that underlie initiation and progression of motoneuron death in ALS. Clinicopathological characterization has helped drive these advances. Overall, progress in understanding ALS can usefully be divided into 2 eras encompassing investigations before and after the discovery of the role of mutations in the 43kd TAR DNA binding protein (TDP-43) in ALS.
ALS in the SOD1 Era: Protein Toxicity
The identification of mutations in the superoxide dismutase 1 (SOD1) gene in 1993 triggered the first major wave of molecular research in ALS. SOD1 is a ubiquitously expressed protein that catalyzes the detoxification of superoxide. More than 160 mutations in SOD1 associate exclusively with ALS. Nearly all are dominant missense mutations. Transgenic mice overexpressing mutant SOD1 display an ALS-like phenotype and remain a cornerstone of ALS research. Mutant SOD1 toxicity is mediated through several mechanisms, prominently including protein misfolding and oligomerization. Downstream effects of this gain-of-function toxicity include impaired mitochondrial metabolism, axonal degeneration, axonal transport failure, excitotoxicity, proteasomal disruption, and endoplasmic reticulum stress.
These insights notwithstanding, SOD1-based research has not yet led to clinically applicable therapeutic advances. That the mouse model has facilitated the definition of the pathobiology of ALS is clear. However, its very strength, the consistency of mutant SOD1-induced motoneuron degeneration and lethality, is a potential limitation in therapeutic trials because the overwhelming toxicity of SOD1 overexpression creates a nearly insurmountable therapeutic barrier. A corollary view is that using survival as the endpoint in these mouse studies may have vitiated any opportunity to see interventional benefit. Outcomes more directly related to the primary toxicities of mutant SOD1 (eg, motor endplate denervation, axonal transport failure) might be more sensitive measures of therapeutic efficacy in these mice.
Intriguingly, recent studies have implicated wild-type SOD1 toxicity in the genesis of SALS. Wild-type SOD1 in SALS can assume aberrant conformations that resemble those of mutant ALS; misfolded wild-type SOD1 can reproduce some forms of cytotoxicity induced by mutant SOD1. It has also been postulated that wild-type SOD1, once misfolded, can propagate intercellular pathology in a prion-like manner in which there is induction of toxic misfolding in otherwise normal SOD1 upon exposure to misfolded SOD1. An important implication of these studies is that therapies that mitigate the neurotoxicity of mutant SOD1 may also be effective in SALS in the absence of SOD1 gene mutations.
ALS after TDP-43: RNA Processing
Following the SOD1 discovery, genetic linkage analyses and candidate gene screening identified rare ALS-related mutations in the alsin, senataxin, dynactin, and VAPB genes.[7-11] However, the next major milestone came in 2006 with the identification of TDP-43 inclusions in ALS brains and spinal cords.[12, 13] In 2008, the discovery of TDP-43 gene mutations in ∼4% of FALS confirmed a mechanistic link between TDP-43 and ALS pathogenesis.[14, 15]
TDP-43 is a ubiquitously expressed DNA/RNA binding protein with diverse roles including gene transcription, RNA splicing, RNA shuttling and translation, and microRNA biogenesis. Although most abundant in the nucleus, TDP-43 shuttles between nuclear and cytoplasmic compartments and is transported along axons. In disease, TDP-43 is ubiquitinated, hyperphosphorylated, and cleaved to form intranuclear and cytosolic aggregates. There is an overall shift in its localization from the nucleus to the cytoplasm and axons. More than 40 dominant missense mutations have been defined in TDP-43, all except 1 in the C-terminal glycine-rich domain. Mutant TDP-43 may have an increased propensity to cleavage[15, 18] and may be more resistant to degradation than wild-type protein.[19, 20] TDP-43 tightly regulates its own expression through negative feedback exerted by binding its own 3′ untranslated region (3′-UTR). Variants in the 3′-UTR have been found in some ALS patients that predict failure of this feedback loop, suggesting that an excess of TDP-43 could contribute to disease.
Evidence from transgenic worms, flies, zebrafish, and rodents indicates that both loss and gain-of-function mechanisms may underlie TDP-43–mediated neurodegeneration.[23-27] Mouse models of TDP-43–derived ALS have been challenging. Forcing high expression of wild-type or mutant TDP-43 generates variable phenotypes including tremor and death due to gastrointestinal stasis. However, bacterial artificial chromosome transgenic mice, conditional expression approaches, and rat models more accurately recapitulate the human disease course and pathology.[28-30] Overall, it is clear that TDP-43 plays vital roles both during development and in the survival of adult motoneurons, and that subtle perturbations in TDP-43 levels are poorly tolerated.
In 2009, mutations in FUS/TLS (fused in sarcoma/translocated in liposarcoma) were identified as a cause of around 4% of FALS.[31, 32] FUS shares similar functional domains to TDP-43 and is also predominantly intranuclear. Under neuronal stress, FUS can exit and subsequently re-enter the nucleus. ALS-related FUS mutations may impair nuclear import, leading to loss of nuclear function and intracytosolic aggregation of FUS. Both TDP-43 and FUS possess prion-like domains. Insoluble TDP-43 has been shown to seed TDP-43 aggregation. This has underscored the hypothesis that, as argued for SOD1, propagation of misfolding between mutant and wild-type TDP-43 or FUS proteins may be critical in the pathogenesis, and spatial spreading, of motor neuron dysfunction in ALS.
C9orf72: Repeat Expansion Pathology
The concept that defective RNA processing represents an Achilles heel for motor neurons was further strengthened by investigations of C9orf72, a gene of unknown function. In 2006, linkage analysis of families with ALS-FTLD implicated a chromosome 9p locus.[36, 37] The same locus was significantly associated with sporadic ALS in genome-wide association studies.[38, 39] In 2011, the genetic culprit was finally identified as a massive intronic hexanucleotide expansion in C9orf72.[40, 41] C9orf72 expansions are the commonest genetic cause of ALS, accounting for up to 50% of FALS in populations of European descent, up to 25% of familial FTLD, and ∼5% of apparently sporadic ALS and FTLD; expansions are also seen in ∼0.5% of controls. Haplotype analysis indicates that a common European founder appears to be responsible for all cases. The high population frequency of the expansions has led to frenzied work to develop C9orf72 models of disease, which could provide valuable insight into sporadic disease, given that C9orf72 ALS shares 2 major features with SALS: co-occurrence of ALS and FTLD, and TDP-43 pathology, neither of which are seen in SOD1 ALS.
Four mechanistic models of C9orf72-mediated ALS have been proposed. First, haploinsufficiency is suggested by the finding of reduced levels of C9orf72 transcripts in ALS brains. Zebrafish C9orf72 knockdown also causes motoneuron degeneration. Haploinsufficiency might be anticipated if an aberrant conformation of the hexanucleotide-expanded genomic DNA impaired C9orf72 transcription and/or if expanded RNA transcripts were abnormally spliced or translated.
Second, neuropathological studies indicate that the transcribed expansion forms nuclear RNA foci. Biochemical studies and in silico modeling predict that the C9orf72 expansion can form highly stable G-quadruplex structures and hairpin loops. DNA and RNA G-quadruplexes have physiological roles, such as transcriptional and translational regulation, RNA transport, and telomere stability, which are highly relevant to neuronal biology and ageing and could be disrupted if pathological C9orf72 expansions sequester the proteins to which physiological G-quadruplexes normally bind.
Third, the hexanucleotide repeat may promiscuously bind and sequester transcription factors, by analogy with sequestration of the transcription factor muscleblind by tri- and tetranucleotide expansions in myotonic dystrophy types 1 (DM1) and 2, respectively. Multiple investigators are avidly identifying RNA binding proteins and other factors that may be sequestered in C9orf72 foci.
The fourth potential mechanism for neurotoxicity of C9orf72 expansions is a form of illegitimate protein translation termed repeat-associated, non-ATG (RAN) translation. This mechanism was first identified in DM1 and spinocerebellar ataxia 8. Two groups have recently demonstrated that RAN translation arises from C9orf72 expansions yielding RAN polypeptides that are sequestered in insoluble aggregates.[50, 51] It remains to be established whether these proteins are cytotoxic.
Axon Biology in ALS and Modifiers of the ALS Phenotype
ALS research has generally focused on pathological processes affecting the cell body, but it is clear that critical events in ALS pathogenesis implicate the neuronal periphery. The earliest pathological changes in ALS appear to occur in axons, dendrites, and synapses. Pathological studies indicate early peripheral denervation before ventral nerve root or cell body degeneration. In the central nervous system, spinal cords from ALS patients demonstrate distal corticospinal tract inflammatory changes[53, 54] and giant axonal swellings suggesting early distal axonal degeneration, findings supported by antemortem diffusion tensor imaging. TDP-43 aggregates form early within motor axons.[57, 58] Similarly, mutant-SOD1 mice demonstrate presymptomatic muscle denervation and terminal axonal degeneration before anterior horn cell loss.[52, 59]
Genetic findings also strongly implicate the axon in ALS. For example, FALS can be caused in some pedigrees by missense mutations in the gene profilin-1 (PFN1), which mediates actin polymerization. PFN1 mutations impair axonal extension and growth cone elongation, leading to an adult onset, predominantly lower motor neuron deterioration.
Genetic variants that reduce function of EphA4, an ephrin receptor, have been shown to be beneficial in ALS. Lower expression of EphA4 in peripheral lymphocytes correlated with delayed onset of ALS and a more protracted disease course. Among many functions, EphA4 signals cessation of axonal extension of distal neuronal terminals during synaptogenesis. These data strongly point to EphA4 as an important target for therapy development in ALS.
That axonal transport may be implicated in ALS has long been intimated on general grounds (ie, that normal transport is indispensable for a motor neuron, whose axon may be up to 20,000× larger than the cell body and extend to 1m in length). A role for altered axonal transport in ALS was also suggested by early studies of isolated human motor nerve axons from ALS cases. Recent evidence shows that mutant SOD1 impairs axonal transport in transgenic ALS mice and in isolated axoplasm from squid giant axons.
Given the above, it is intriguing that we now know that axons have a self-destruct mechanism, independent of apoptosis, which can be significantly delayed in vivo given the correct molecular environment. The slow Wallerian degeneration (WldS) mouse first demonstrated that axons can survive independent of the cell body for weeks after nerve transection.[65, 66] Recent discoveries that loss-of-function mutations in the Sarm1 and highwire genes have similarly potent axon-protective effects clearly suggest that axon degeneration is an active process.[67, 68] Elucidating the molecular cascade responsible for axon killing holds promise for eventually identifying therapeutic targets for ALS.
In addition to EphA4, other genes have been reported to modify the ALS phenotype (Table 1). Genetic studies implicate UNC13A (which is also a susceptibility factor) and KIFAP3 as modifiers of survival in some populations, whereas the single nucleotide polymorphism rs3011225 modifies age of onset, as does the P413L allele of the chromogranin B gene. Elegant studies by Henderson and colleagues indicate that the metalloprotease MMP9 is a determinant of ALS susceptibility; it is expressed more abundantly in motor neurons that are ALS susceptible (eg, lumbosacral motoneurons) than in those that are ALS resistant (eg, oculomotor and sphincteric motoneurons; C. Henderson, personal communication).
Table 1. Genes Implicated in Familial and Sporadic ALS
|ALS Type||Locus||Familial, n = 24||Sporadic, n = 6|
|ALS||9q||Ubiquilin 1|| |
|ALS||X||Ubiquilin 2|| |
That non-neuronal cells can modulate the phenotype in animal models of ALS is now clear; mutant SOD1 in microglia and astrocytes accelerates the disease course after onset, and mutant SOD1 in oligodendroglial precursor cells hastens death in transgenic mice.[72-75] Similar findings were recently reported for astrocytes in a TDP-43 rat. The mechanisms for these non–cell-autonomous influences are not well defined, although it is intriguing that oligodendrocytes are known to provide significant metabolic support to axons through lactate transport.