Janus kinase inhibitors are potential therapeutics for amyotrophic lateral sclerosis

Amyotrophic lateral sclerosis (ALS) is an archetypal multifactorial neurodegenerative disease whose primary cause in the majority of cases is still unknown. ALS is characterised by the loss of lower and upper motor neurons resulting in progressive failure of the neuromuscular system, with death usually occurring approximately 3–5 years after diagnosis due to the failure of the respiratory muscles [1]. A significant number of patients also show disturbed frontal and temporal lobe function [2] with some (approximately 5%) developing overt frontotemporal dementia. The disease process likely starts a significant period of time before symptoms first appear, suggesting that the functional changes identified in patients will include both disease processes and potentially beneficial compensatory adaptations.

ALS disease prevalence is approximately 9 per 100,000 persons [3], of which 5%–10% of individuals have familial ALS (fALS) defined by their family history. The most commonly associated genes in fALS are C9ORF72, TARDBP, FUS and SOD1. Recent genetic analyses have revealed numerous mutations in sporadic ALS cases, which reveal possible underlying genetic drivers of the disease [4]. Pathways implicated by such risk genes include apoptosis arising from mitochondrial dysfunction, autophagy with loss of protein homeostasis, inflammation (peripheral and central), impaired intracellular trafficking and excitotoxicity. These dysfunctional pathways operate in many cell types including the upper (corticospinal) and lower (spinal) motor neurons themselves, associated glial and Schwann cells, skeletal muscle fibres and their associated progenitors as well as cells of the immune/ inflammatory system. Considering the wide range of pathways and cell types affected, efficacious therapeutics for ALS may need to target pleiotropic regulators able to modulate a broad range of defective cellular processes, particularly as the originating causes are not fully understood.

Drug repurposing is the use of already approved drugs for new indications. Given the long and expensive drug development process, drug repurposing provides a means of accelerating therapeutic discovery at a lower financial risk. This is mainly because the toxicological and pharmacokinetic risks are known at the outset, and the molecular optimisation processes (i.e., preclinical drug discovery) have been completed. Importantly, for a rapidly progressing disease, repurposing offers the possibility of treating patients currently with the disease, an unrealistic process if starting from scratch, given the 10–15-year timescale to bring a new chemical entity to market. In addition, rare diseases such as ALS may not provide a realistic opportunity to recover the enormous costs of bringing a new drug to market ($2-3B) [6], shifting the risk-reward profile unfavourably for large pharmaceutical companies. This may explain why most of the current ALS clinical trials are investigating drugs from small pharmaceutical and biotechnology companies [4].

Only four drugs have been approved by the FDA for ALS: riluzole, edaravone, and the combination therapy AMX0035 and tofersen. Riluzole may act by reducing glutamate-mediated cytotoxicity in the motor cortex and spinal cord and has been shown to increase survival by as much as 19 months [5]. Edaravone is an antioxidant which may delay progression without affecting survival. AMX0035 alleviates mitochondrial and endoplasmic reticulum stress while tofersen is approved for treating patients with SOD1 mutations. Neither edaravone nor AMX0035 (which was approved by the FDA based solely on Phase 2 data) has been approved in Europe. There is therefore a significant and well recognised need for new disease-modifying treatments capable of reducing the rate of progression and increasing survival of individuals afflicted by ALS.

Recently, computer-enhanced methods for repurposing approved drugs have been developed, particularly in response to the recent COVID-19 pandemic [7, 8]. These include AI approaches [9] and unbiased laboratory methods [10] to identify the protein–protein interaction networks operating in the disease state. Zeng et al. [11] compiled a knowledge graph (KG) integrating scientific literature and drug properties to identify approved drugs which could be used to treat COVID-19. In this context, the BenevolentAI KG was used in early 2020 to identify the Janus kinase (JAK) inhibitor baricitinib as a combined anti-inflammatory and anti-viral agent for the treatment of hospitalised patients with COVID-19 [12‐14]. Subsequently, following the ACTT2 [15] and COV-BARRIER [16] clinical trials, this agent was approved for use in this patient population and has been used worldwide.

The BenevolentAI KG includes the contents of more than 85 diverse and independent structured biomedical databases [17, 18]. These structured data sources are further enhanced by natural language processing (NLP) pipelines, which integrate data from more than 35 million scientific papers. This KG was built to broadly cover complex disease biology across therapeutic areas and help scientists to develop data-driven hypotheses. Although the KG and associated machine learning models were predominantly built for predicting novel targets for disease, the technology can also support the repurposing of known treatments for alternative indications.

To conduct an unbiased literature-wide search of the disease mechanisms identified in ALS and their potential regulators, we used the BenevolentAI KG to identify 500 genes most specifically associated with ALS using a metric focused on the probability of the co-occurrence of two words within the literature. Since ALS causes skeletal muscle atrophy, we also identified 500 most associated genes with muscle atrophy. Pathway enrichment using these genes revealed four pathway groups as likely to be fundamental to ALS disease processes: cytokine-mediated inflammation, autophagy, apoptosis, and FOXO signalling pathways. Other mechanisms implicated to a lesser degree included TLR4-mediated inflammation and glutamate-mediated neurotransmission (Table 1).

However, of particular interest were the ALS-associated and muscle atrophy-associated genes which are common to these four processes. These were STAT3 (encoding signal transducer and activator of transcription 3, STAT3), AKT1 (AKT serine/threonine kinase 1), HSP90AA1 (heat shock protein 90 alpha family class A member 1), UBC (ubiquitin C) and FASLG (Fas ligand) (Fig. 1), indicating that STAT3 signalling is fundamental to the disease.

In addition to inflammation, autophagy, apoptosis and the FOXO pathways, JAK/STAT signalling is also heavily implicated in the characteristic pathophysiology of ALS, i.e., TDP43 protein aggregates, mitochondrial dysfunction, skeletal muscle denervation and excitotoxicity (Table 2). These disease processes are widely believed to be major contributors to ALS disease especially mitochondrial dysfunction, which results in deficits in ATP supply, protein aggregation with 95% of patients having TDP43 aggregates, and muscle denervation and atrophy which eventually causes death. The discovery that all these processes involve JAK/STAT signalling suggests that this pathway could be a therapeutic target.

In addition to the complexities inherent in modifying multiple disease processes, ALS treatment also requires that effective drugs have beneficial effects in a wide range of cell types where modulation of pleiotropic signalling pathways could have detrimental or even counter-productive effects. It is therefore important to assess the effect of drugs modifying JAK/STAT signalling in the cell types affected by the disease. The cell types and processes involved in ALS include glial cells, T cells (particularly NK cells), macrophages as well as the upper and lower motor neurons (Fig. 2).

In the following discussion the reader is referred to some recent excellent reviews for a comprehensive description of ALS [4, 19, 20], and we focus here on the potential role of STAT3 in the disease.

STAT transcription factors, primarily activated by JAKs, play a central role in transducing diverse immune signalling pathways. Regulation of STAT3 through phosphorylation, methylation and acetylation impacts various cellular functions, including gene transcription control, mitochondrial activity, calcium buffering, and apoptosis inhibition. The canonical STAT3 pathway mediates many of the actions of cytokines and is induced by JAK-mediated tyrosine phosphorylation (Y705), followed by STAT3pY705 dimerisation, nuclear import and the regulation of gene expression. This activation of the JAK/STAT pathway predominates in the post-mortem spinal cords of ALS patients where interleukin (IL)-6, IL-2, interferon (IFN) γ, IFNα and the unfolded protein response (UPR) pathways are all upregulated. Since JAK/STAT signalling is an integral part of these pathways, it is likely that JAK inhibitors will have a significant effect on their activity [see Fig. 1f of reference 21]. The non-canonical pathway is activated by serine/threonine kinase (e.g. extracellular signal related kinase [ERK], CDK5)-mediated phosphorylation of S727. The pY705/pS727 dual phosphorylation is required for full nuclear gene transcription activity [22]. The pS727 form performs distinct functions in the mitochondria as part of Complex 1 of the Electron Transport Chain (ETC) and is required for full ETC function while the cytoplasmic unphosphorylated form regulates microtubule stability [23].

ALS-associated dysfunction of mitochondrial proteins has been identified in SOD1, VCP, CHCHD10 and TBK1 subtypes of ALS, indicating that mitochondrial dysfunction may be causative in ALS. Motor neuron mitochondria in patients and a range of models bearing ALS mutant genes are morphologically abnormal with fragments and aggregates throughout the axons [37]. The accumulation of these defective (but probably ROS-producing) mitochondria with abnormal cristae [42] is observed in patients and in those models with mutated or overexpressed SOD1, C9ORF72, FUS and TDP43. There are downstream consequences of this mitochondrial damage, including cGAS/STING-mediated inflammation, fragmentation, aberrant mitochondrial transport within neurons, reduced oxidative phosphorylation and ATP generation, reduced calcium buffering capacity, increased ROS generation and activation of the cellular apoptotic pathways. Since these mechanisms reinforce each other, therapeutic agents will need to inhibit multiple processes to effectively restore mitochondrial function. This could potentially be achieved through inhibition of the JAK/STAT signalling coupled with increased mitoSTAT activity as discussed below.

The mitochondrial defects in ALS are associated with the accumulation of protein aggregates and with increased inflammation. The latter may be partly a consequence of inflammaging (i.e., the gradual increase of inflammation seen with ageing [43]) in this age-associated disease, with a significantly higher degree of inflammation than in age-matched controls [32]. Increased levels of proinflammatory cytokines have been reported in ALS plasma [32] and CSF [33], indicating that these cytokines could stimulate upper and lower motor neurons as well as glial cells in both the motor cortex and spinal cord. Increased STAT3 activation has been shown in neurons and glia in post-mortem ALS samples [44], and in SOD1 mutant mouse motor neurons [45]. Systemic cytokine/STAT3 signalling in ALS could therefore explain the mitochondrial dysfunction observed in many cell types including neurons, glia, muscle fibres and blood cells [46‐49]. Similarly, canonical STAT3 activation and/or mitoSTAT3 depletion could contribute to poor glucose metabolism in motor neurons, poor mitochondrial quality control, reduced ETC activity and poor calcium buffering as outlined below.

One of the characteristic aspects of ALS is the accumulation of protein aggregates and stress granules in the affected motor neurons which, as mentioned above, cause mitochondrial dysfunction, with increased ROS production. The most commonly aggregated protein is TDP43, the product of the TARDBP gene. This protein is involved in RNA splicing and is usually found in the nucleus, although it also engages in shuttling between the nucleus and the cytoplasm. In approximately 95% of ALS cases, cytoplasmic ubiquitinated TDP43 aggregates are observed [79], mostly in the absence of TARDBP mutations. TDP43 aggregates are seen in both upper and lower motor neurons, as well as other apparently unaffected neurons [80], but are not seen with SOD1 mutations, which instead reveal SOD1 aggregates in the mitochondria. Similar cytoplasmic FUS aggregates are seen in the presence of FUS mutations. In all three cases, it appears that these cytoplasmic aggregates have toxic properties, together with dysfunctions of RNA splicing and nucleocytoplasmic transport as added complications. These observations clearly suggest a failure of autophagy and proteostasis. Many mutated genes associated with ALS are linked to the proteasome and autophagy, including C9ORF72, OPTN, VCP, SIGMAR1, PARK9, SQSTM1, CHMP2B, UBQLN2 and FIG4, indicating that dysfunctional proteostasis may also induce ALS.

The TDP43 aggregates are associated with nuclear TDP43 depletion and accordingly likely with the dysregulation of RNA metabolism affecting thousands of genes [81]. The cytoplasmic aggregates so formed may also sequester RNA transport components essential for appropriate export of RNA from the nucleus [82] and along the axon [83]. The stress granules may be causative in the generation of the toxic protein aggregates [84], so restoration of normal autophagic clearance of stress granules [85] would seem a necessary requirement in therapy.

There is increasing evidence that neuroinflammation plays a key role in neurodegenerative diseases including ALS. Consistent with this, high levels of MCP-1, IL-15, IL-17 and TNFα are seen in the CSF of ALS patients [33] and many of the ALS-associated proteins are abundantly expressed in immune cells [79]. There is also evidence linking TDP43 to inflammatory nuclear factor NF-kappa-B (NFκB) pathways [99], as well as evidence showing that inflammation induces the cytoplasmic localisation of TDP43 [100], illustrating the inter-related nature of the disease processes of ALS.

It has long been suggested that the failure of the neuromuscular junction (NMJ) is one of the early processes in ALS, with a probable consequent loss of trophic support between nerve and muscle, resulting in nerve “die back” and muscle atrophy. Since an ALS-like syndrome can be induced by muscle-specific expression of hSOD1, such aggregates may contribute to the denervation of the muscle, perhaps through loss of proteins such as rapsyn required for the maintenance of the post-synaptic component of NMJ [98]. The resultant denervation may then result in motor neuronal injury and progression of the disease in the spinal cord.

Other conditions in which the NMJ fails include sarcopenia and cancer cachexia where this failure precedes muscle atrophy [132]. Since the establishment of a functional NMJ requires the elimination of less active synapses on the muscle fibre, the denervation may induce dieback of synapses and axons, which is mediated by similar mechanisms as synaptic pruning. Such synaptic pruning in the CNS has recently been shown to be mediated by JAK2/STAT1 signalling in the corpus callosum [133], suggesting that increased cytokine/gp130 signalling (e.g. IL-6) could be responsible for the denervation seen in ALS.

Increased IL-6 activity is reported in ALS muscles, where increased synthesis of this myokine in fibro-adipogenic progenitors results in increased IL-6 production and STAT3 activation. Inhibition of IL-6/STAT3 prevents atrophy in the SOD1 transgenic mice [134, 135] and in cancer cachexia models [136, 137]. Sartori et al. recently showed that in models of cancer-induced cachectic muscle wasting, the NMJ disassembly occurs before atrophy and is a consequence of STAT3-induced noggin expression [132]. The resultant inhibition of transforming growth factor beta family-mediated Mothers against decapentaplegic homolog (SMAD)1/5/8 signalling causes NMJ failure. This effect is synergistic with activin A- or myostatin-stimulated SMAD2/3 which also inhibits SMAD1/5/8 signalling, while increased SMAD1/5/8 signalling prevents the denervation. Similarly, Ma et al. [138] showed that JAK/STAT3 signalling mediates both IFN- and TNF-induced cachectic muscle wasting. Finally, the atrogenes responsible for muscle atrophy are induced by a combination of STAT3 and FOXO3 [139, 140]. It is therefore clear that STAT3 activation, mainly due to an increase in the myokine IL-6, can promote NMJ disassembly, which in turn further increases IL-6/STAT3 signalling [141], creating a positive feedback loop, resulting in muscular atrophy. In addition, this or related cytokines may mediate both the synapse elimination and the axonal dieback via JAK2/STAT1 [133].

Since respiratory muscle failure is a major cause of ALS-associated mortality, it is important to consider the effect of the JAK/STAT pathway on respiratory muscles. Mechanical ventilation of the lungs causes significant diaphragm dysfunction characterised by atrophy and a reduction in force which can result in ventilator dependency [142]. The mechanisms implicated are similar to those seen in ALS, i.e., apoptosis, autophagy, ROS and ROS-induced mitochondrial dysfunction, coupled with aberrant calcium handling and resultant calpain activation [143]. In both human cases and rodent models, mechanical ventilation is associated with high levels of STAT3 activation which induce mitochondrial dysfunction and ROS generation, both of which are suppressed by JAK inhibition [144]. The JAK inhibitor tofacitinib is currently being tested in a clinical trial of ventilator-induced diaphragm atrophy [145] (NCT03681275).

The reduced ability of ALS A1-like astrocytes to clear synaptic glutamate [106] resulting in increased stimulation of excitatory glutamate receptor probably exacerbates the failure of the mitochondrial/ER mechanism to buffer cytoplasmic calcium in motor neurons. The combination of increased excitation and reduced calcium buffering will then contribute to the decline of these neurons. It has been suggested that motor neurons, due to their large size and plethora of excitatory calcium-permeable AMPA receptors, are particularly susceptible to excitotoxicity. Supporting evidence comes from the partial efficacy of riluzole which may reduce synaptic glutamate release by inactivation of Na⁺ channels [151], and from the detection of cortical hyperexcitability in ALS patients [4]. Calcium-permeable AMPA receptors are highly expressed in motor neurons [78, 152], putting extra strain on the calcium buffering capacity of these cells. Indeed, inhibiting the Q/R editing of the GluA2 subunit of the AMPA receptor through adenosine deaminase ADAR2 deficiency induces an ALS-like phenotype, reinforcing the idea that calcium mishandling contributes to this disease [153]. In addition, it appears that the CaV1.3 voltage-sensitive Ca²⁺ channel and unedited AMPAR GluA2 subunits are more highly expressed in ALS-sensitive motor neurons than in the more resistant oculomotor neurons [154]. Inhibiting JAK/STAT signalling will increase synaptic glutamate clearance by astrocytes [106], reduce cytokine-induced cytoplasmic calcium [155], and increase the calcium-buffering capacity (through restoration of the mitochondrial ETC), thereby reducing the excitotoxic stress on motor neurons.

Hyperexcitability of cortical neurons in ALS has been widely described as an aspect of the ALS disease, with a reduction in threshold potential resulting in increased excitability and reduced inhibition [156]. This hyperexcitability is associated with reduced GABA and increased glutamate/glutamine levels in the motor cortices of ALS patients [157]. Interestingly, in epilepsy models, inhibition of JAK/STAT signalling prevents the loss of GABAergic neurons [158], while in brain injury models, elevated STAT3 inhibits the expression of GABA receptor subunits [159] and STAT3 has been implicated in mediating the development of epilepsy-associated hyperexcitability [160]. The hyperexcitability may also be a consequence of JAK2-mediated inhibition of the Na⁺/K⁺-ATPase [161], which is the major driver of the membrane potential, suggesting that under conditions of high cytokine action (when JAK2 is active) the membrane of the motor neurons may become partially depolarised and so hyper-responsive to depolarising stimuli. The alpha3 subunit of the Na⁺/K⁺-ATPase is heavily implicated in ALS disease progression, particularly in the SOD1 models, since SOD1 aggregates bind this subunit [162], and prevention of such binding increases the Na⁺/K⁺-ATPase activity and extends survival presumably through reduced excitability [162]. These observations suggest that JAK/STAT inhibitors could reduce the hyperexcitability of upper motor neurons through maintenance of the cortical GABAergic inhibition and the polarisation of neuronal plasma membranes. However, increasing motor neuron excitability with AMPA receptor agonists can also extend survival [163], so it is not clear what role excitability plays in the disease [164]. It is tempting to speculate that these contradictory data may be a consequence of different stages of disease with hyperexcitability in early stages and a compensatory reduced excitability in later stages.

ER stress and UPR are activated by many ALS-related proteins, including SOD1, C9ORF72, FUS and TDP43, suggesting that this pathway may be a source of potential therapeutic targets. All 3 major pathways of the UPR interact with JAK/STAT signalling. Inhibition of JAK/STAT signalling directly inhibits the serine/threonine-protein kinase/endoribonuclease IRE1a and eukaryotic translation initiation factor 2-alpha kinase 3 (EIF2AK3 or PERK) arms of the UPR through blockade of type 1 IFN receptors [165] and PERK [166], respectively, while also reducing activating transcription factor 6 expression in the third arm [167]. Improvements in mitoSTAT3 provision to the mitochondria through JAK/STAT inhibition could also reduce ER stress [168].

In addition, the generation of misfolded dipeptide repeat proteins through repeat-associated non-AUG (RAN) translation of the RNA hexanucleotide repeat expansion in the C9ORF72 mutation results in activation of ER stress and the integrated stress response. The widely used antidiabetic drug metformin has recently been shown to be effective in a model of C9ORF72-ALS [90] through inhibition of the ER stress-related IFN-induced, double-stranded RNA-activated protein kinase (PKR). However, the integrated stress response is also activated through PERK which signals through JAK/STAT, suggesting that a combination of a JAK inhibitor with metformin would be most effective in patients with dipeptide repeat and other misfolded proteins such as TDP43. The combination of improved protein aggregate clearance and inhibition of the downstream consequences of the UPR by JAK/STAT inhibitors will significantly reduce the effects of the ER stress response in motor neurons.

JAK inhibitors have been approved for the treatment of resistant rheumatoid arthritis (tofacitinib, baricitinib, filgotinib, upadacitinib) and myelofibrosis (ruxolitinib, fedratinib) [169]. Due to the involvement of upper motor neurons in ALS, it would be beneficial for the selected JAK inhibitors to be at least partially CNS-penetrant. In this context, the relatively low plasma-protein-binding of tofacitinib and baricitinib (approximately 50%) makes them good candidates, but tofacitinib has very low brain preparation (FDA Pharmacology Review). In contrast, baricitinib at therapeutic doses crosses the barrier in non-human primates (CSF concentration approximately 20% of total plasma concentration) [170], despite being a substrate of the CNS efflux pumps P glycoprotein (PGP) and a breast cancer resistance protein. Interestingly, progressive multiple atrophy is a form of motor neuron disease largely restricted to lower motor neurons which are exposed outside the blood–brain barrier, suggesting that high brain penetrance may not be required for this form of the disease. Other properties required for a JAK inhibitor treatment of ALS include inhibition of IL-6/gp130 (i.e. JAK1/2/Tyk2) and IFNα and β (JAK1/TYK2) signalling. JAK3 inhibition may also be beneficial through a reduction in NK cell and other pro-inflammatory cascades.

Although it would be better to use a fully brain-penetrant JAK inhibitor to reduce the dose required, there are no approved inhibitors of JAK1 and JAK2 with better brain access than baricitinib to our knowledge [171].

The side-effect profiles of the JAK inhibitors are variable. For tofacitinib and baricitinib, opportunistic and herpes zoster infections are the major adverse events in the context of prolonged dosing.

Long-term treatment (months to years) with JAK inhibitors is associated with side effects in a small number of patients, including venous thromboembolisms and Herpes infections [172]. In the clinical trials assessed by the EMA for the registration of baricitinib which included 4214 patient years, the most significant side-effect was a small increase of upper respiratory tract infections (similar to that observed with methotrexate). The incidence of Herpes zoster infections was low (3.2 per 100 patient-years), and similar to placebo [173].

However, in a 4-year study comparing tofacitinib and anti-TNF in patients aged over 50 years with rheumatoid arthritis in the presence of methotrexate, the JAK inhibitor showed an association with increased risks of cardiovascular issues (1.33-fold), cancers (1.48 fold) and herpes zoster reactivation (threefold) [174]. As a result, tofacitinib, baricitinib and upadicitinib have black box warnings for the risk of thrombosis and cancer [175]. In an analysis of 9 clinical trials in which some patients were dosed with baricitinib for up to 9 years, there were no significant differences for baricitinib versus placebo in the incidence of death, malignancy, serious infection, or major adverse cardiovascular events. The incidence per 100 patient years was 0.5 for venous thromboembolism, 0.3 for deep vein thrombosis, 0.2 for pulmonary embolism, and 1.0 for cancer [176]. Baricitinib has also been taken for brief periods by over one million seriously ill hospitalised patients with COVID-19. In meta-analyses, few, if any, drug-induced serious adverse events were reported in this patient population; in contrast, the drug showed significant mortality benefits even when added to other treatments such as steroids and tocilizumab [177]. In addition, the incidence of secondary infections with COVID-19 was reduced by 50% by baricitinib in the ACTT-2 clinical trial [178].

Baricitinib is approved in 2 mg and 4 mg oral formulations, has a half-life of 8–9 h and reaches a plasma Cmax (4 mg dose) of 112–135 nM, with 50% plasma protein bound, and equal plasma/blood equilibration implying free access to the inside of cells. The IC50 at JAK1/2 is 5.8 nM, Tyk2 53 nM and > 500 nM at JAK3. When dosing 4 mg qd the trough concentration is predicted to be 15 nM after a single dose with perhaps 2–3 nM in CSF. The effect of baricitinib on spinal cord inflammation in patients with ALS or Alzheimer’s disease is currently the subject of a clinical trial also examining the drug exposure in the CNS (NCT05189106).

JAK/STAT3 signalling is central to many of the disease processes associated with ALS. JAK inhibitors are therefore predicted to have widespread beneficial effects in ALS, correcting many pathophysiological processes including: mitochondrial dysfunction (abnormal morphology, reduced MMP, poor calcium buffering, reduced ATP generation), autophagy (thereby correcting protein aggregate accumulation and dysfunctional RNA metabolism), and neuroinflammation mediated by astrocytes, microglia and NK T cells, while simultaneously reducing the risk of excitotoxicity, ER stress and cytoplasmic calcium overload. Since it appears that motor neuron damage may result from both direct effects on the nerve (e.g. mediated by glial cells) and muscle atrophy-induced denervation, it is important that JAK inhibitors are predicted to protect both the neuron and the muscle through multiple mechanisms, including those with self-perpetuating positive feedback mechanisms (NMJ denervation and mitochondrial dysfunction). There are FDA- and EMA-approved JAK inhibitors which have significant, but tolerable side effects, which are likely to have a positive risk–benefit ratio given the progressive and lethal nature of ALS. Finally, this approach to identifying inhibitors capable of simultaneously inhibiting the multiple disease mechanisms operating in ALS could become a model for identifying therapeutics in other multi-factorial diseases, including other neurodegenerative diseases.