Part 1 · The disease
TDP-43 and what goes wrong
In most people with ALS, a protein called TDP-43 leaves the nucleus of motor neurons and forms clumps in the cell body. Follow it from its normal job to the damage, then switch to SOD1-ALS or FUS-ALS to see the forms that work differently.
Schematic, not to scale. Shares in the bars are drawn, not measured.
TDP-43 domain map, amino-acid positions from Prasad 2019. 294,430
How common is TDP-43 pathology? A 2022 Lancet seminar puts it at about 97% of ALS and about 50% of frontotemporal dementia (FTD). 180 A 2013 review says "over 90%" of ALS. 294 Both summarise the same autopsy evidence rather than one counted cohort. In one autopsy series of 111 cases, every sporadic case (59), every case with dementia (26) and every SOD1-negative familial case (11) had TDP-43-positive inclusions, and none of the 15 SOD1 cases did. 309
Three pathologies side by side
| What the inclusions contain | TDP-43 inclusions? | Share of ALS | |
|---|---|---|---|
| TDP-43 proteinopathy | TDP-43 that is hyperphosphorylated, ubiquitinated and cut into C-terminal fragments; found only in affected brain and spinal cord regions. 220,392 | Yes nucleus often cleared | About 97% of ALS 180; "over 90%" 294. Includes sporadic and C9orf72 ALS. 73,309 |
| SOD1-ALS | Misfolded SOD1 in ubiquitin-positive inclusions. 309,492 | No 309 | In a meta-analysis: Europe 14.8% of familial and 1.2% of sporadic ALS; Asia 30.0% and 1.5%. 554 |
| FUS-ALS | FUS-positive cytoplasmic inclusions. 279,449 | No, though "some debate remains" 449 | Europe 2.8% of familial and 0.3% of sporadic ALS; Asia 6.4% and 0.9%. 554 |
How the pathology spreads: Brettschneider stages
In 76 ALS autopsy cases, phosphorylated TDP-43 (pTDP-43) appeared in four stages, each adding regions to the ones before. The 10 C9orf72-expansion cases followed the same order with more pathology in each region. Oligodendroglia carried pTDP-43 aggregates at every stage. 73
Established The staged pattern itself, seen in autopsy tissue.
Hypothesis That the pathology travels from cell to cell along axonal pathways, as the authors propose. 73
Schematic side view; region positions are approximate.
Other proposed mechanisms
Many processes go wrong in ALS motor neurons, and it is not settled which come first. A 2022 review groups them as impaired RNA metabolism, altered protein handling and autophagy, cytoskeleton and transport defects, and mitochondrial dysfunction. 180 The labels rate how strong the evidence is that each one helps cause ALS in people: Established repeatedly shown in people with ALS, Early evidence shown in models or limited human data, Hypothesis proposed and still being tested.
Cryptic exons from TDP-43 loss
STMN2 and UNC13A cryptic exons are found in tissue from people with ALS, and the TDP-43 repression of cryptic exons is impaired in ALS-FTD tissue. Forcing cryptic exons back out prevented cell death in models. 80,268,295,307,328
Glutamate excitotoxicity
High-affinity glutamate uptake was 59% lower in spinal cord and 70% lower in motor cortex from 13 people with ALS, and not lower in unaffected regions. The astrocyte transporter EAAT2 (GLT-1) was down by 71% in motor cortex. The authors note this could be a cause or a consequence. 457,458 Uncleared glutamate drives repeated firing and calcium entry. 492
Riluzole's rationale. It reduces glutamate release and blocks the persistent sodium current. 144,504 In the first trial, 74% on riluzole and 58% on placebo were alive at 12 months. 57 The US label says its mechanism in ALS is unknown. 178 See Riluzole, edaravone and others.
Oxidative stress
Markers of oxidative damage are raised in ALS spinal cord and cerebrospinal fluid, but antioxidant trials have been disappointing. 40 Mutant SOD1 does not harm through lost enzyme activity, so SOD1-ALS is not simply a shortage of an antioxidant. 492
Edaravone's rationale. It is an antioxidant that scavenges peroxyl radicals. 529 In a selected early-stage group, ALSFRS-R fell 5.01 points with edaravone and 7.50 with placebo over 24 weeks 325; how far this applies more widely is contested. 180 The US label says its mechanism in ALS is unknown. 176
Mitochondria and energy
Spinal motor neurons rely heavily on mitochondria to buffer calcium and have little calcium-binding protein of their own, which is one proposed reason they are vulnerable. 11,404 Excess glutamate stresses mitochondria and the endoplasmic reticulum. CHCHD10, a gene for a mitochondrial protein, is a rare ALS gene. 180,492
Axonal transport
Motor axons can be over a metre long. SOD1 mutants slow transport in both directions months before neurons die, and TDP-43 mutations impair transport of RNA granules in flies and in motor neurons grown from patients' cells. Evidence is strong in models and indirect in people. 492 TUBA4A and PFN1 mutations disrupt the cytoskeleton. 180
Protein handling and autophagy
TDP-43 pathology is said to impair the clearing of damaged proteins. 180 TBK1, which regulates autophagy and inflammation, is an ALS gene. 492 Clumps are not the whole story: removing large SOD1 aggregates did not change disease in mice. 492
Transport into the nucleus
The C9orf72 repeat RNA binds RanGAP1. RanGAP was misplaced and nuclear import impaired in flies, patient-derived neurons and patient brain, and drugs aimed at the repeat corrected it. 550 Hardened TDP-43 droplets trap import factors too. 195
Glia: damage from neighbouring cells
In SOD1 mice, mutant SOD1 in motor neurons set the time of onset, while lowering it in microglia slowed later progression. 67 Mutant SOD1 astrocytes killed motor neurons in culture. 355 Oligodendrocytes degenerated before onset, and grey-matter demyelination was seen in patients. 254 Inflammatory cells are present in ALS, but in mice glial activation followed axon damage, so its causal weight is debated. 47,185
Open questions
Are stress granules the seed?
For: most TDP-43 inclusions made in cultured cells overlapped stress granules, drugs that block stress granules reduced them, and inclusions in patient tissue carried stress granule markers. 298
Against: cytoplasmic TDP-43 formed long-lived droplets without conventional stress granules, and these droplets hardened, blocked nuclear import and killed cells. 195
Misfolded SOD1 in sporadic ALS?
For: oxidised normal SOD1 shared a misfolded shape (the C4F6 epitope) with mutant SOD1, was found in spinal motor neurons in some sporadic cases, and slowed axonal transport. 69
Against: seven antibodies for misfolded SOD1 found it below detection in sporadic and non-SOD1 inherited ALS. 132 In cell models, misfolded SOD1 can seed more misfolding and pass between cells. 353
Lost function or toxic clumps?
Cells with TDP-43 clumps usually also lose nuclear TDP-43, so both are likely involved. 294 But some mutant TDP-43 mice lost lower motor neurons with neither clumps nor nuclear loss, only changes in splicing. 294