Neuromuscular Disease Curation Project

Tracking epic: monarch-initiative/dismech#9005

Overview

Neuromuscular disease is organized around the motor unit: lower motor neuron → peripheral nerve → neuromuscular junction (NMJ) → muscle fiber. Each compartment fails in its own characteristic way, and the classification that the Mondo neuromuscular workshop is converging on — classify by affected structure rather than by etiology, because the causes (genetic, autoimmune, toxic, infectious, metabolic, paraneoplastic, degenerative) are too varied to partition cleanly — is exactly the classification dismech should mirror in its mechanism modules.

This project has two aims:

1. Fill the module layer. Give each motor-unit compartment at least one conserved mechanism module, so that entries that currently share nothing can declare conforms_to against a common trunk.
2. Record what is genuinely unresolved. Several of the most-cited mechanisms in this field are contested or rest on model systems that do not reproduce the human phenotype. Those belong in mechanistic_hypotheses and HUMAN_MODEL_MISMATCH discussions rather than being flattened into a single confident causal chain.

Current state of the knowledge base

Counted over kb/disorders/ (69 entries matching neuromuscular naming patterns, excluding retinal/corneal dystrophies, leukodystrophies, cardiomyopathies, and rhabdomyosarcomas):

The 19 that do conform reach the module layer almost entirely through the nerve and motor neuron compartments — peripheral_axonal_degeneration (the Charcot-Marie-Tooth and inflammatory neuropathy entries), tdp43_proteinopathy / glutamate_excitotoxicity / disabled_macroautophagy (ALS), plus a handful reached incidentally (lysosomal_substrate_accumulation for Pompe, molecular_mimicry_autoimmunity for Guillain-Barré, antisense_oligonucleotide_therapy for Duchenne).

There is no muscle-side mechanism module in the repository. Of the 123 modules in kb/modules/, none models sarcolemmal integrity, excitation-contraction coupling, sarcomere assembly, skeletal-muscle excitability, myofiber necrosis, or endplate transmission. The consequence is visible in the orphan list: Spinal_Muscular_Atrophy, Myasthenia_Gravis, Congenital_Myasthenic_Syndrome, Myotonic_Dystrophy_Type_1, Nemaline_Myopathy, Central_Core_Myopathy, Malignant_Hyperthermia_of_Anesthesia, Dermatomyositis, Facioscapulohumeral_Muscular_Dystrophy, and Emery_Dreifuss_Muscular_Dystrophy all carry zero module conformance.

The one partial exception is presynaptic transmission: synaptic_vesicle_cycle already names presynaptic congenital myasthenic syndrome in scope and covers UNC13A, SNAP25, SYT2, and VAMP1. It is explicitly and deliberately presynaptic, so the postsynaptic and synaptic-basal-lamina arms of the NMJ have no home.

Proposed modules

Tier 1 gives each motor-unit compartment a trunk and mirrors the four-way Mondo split (motor neuron disorder / peripheral neuropathy / NMJ disease / myopathy). Tier 2 covers well-bounded mechanisms with several ready conformers each.

Tier 1

1. sarcolemmal_mechanical_fragility — the muscular dystrophy trunk

Loss of a dystrophin-glycoprotein complex (DGC), membrane-repair, or sarcolemma-ECM linkage component → loss of mechanical coupling between the cortical cytoskeleton and the basal lamina → contraction-induced (particularly eccentric) sarcolemmal injury with Ca²⁺ influx → calpain activation, mitochondrial Ca²⁺ overload, and myofiber necrosis → regenerative exhaustion of the satellite cell pool → fibro-fatty replacement and progressive weakness. The terminal step feeds the existing fibrotic_response module.

Ready conformers: Duchenne_Muscular_Dystrophy (which currently reaches the module layer only through its exon-skipping treatment), Autosomal_Recessive_Limb-Girdle_Muscular_Dystrophy, Limb-Girdle_Muscular_Dystrophy_Autosomal_Dominant, Ullrich_Congenital_Muscular_Dystrophy and Bethlem_Myopathy (the COL6 extracellular arm), Dystroglycanopathy (whose muscle arm is unmodeled — the entry currently conforms only to the brain-side pial_basement_membrane_radial_glial_endfoot_failure).

This is the highest-value single module in the project. It should carry the mechanical-fragility vs mechanotransduction controversy as competing mechanistic_hypotheses (see below) rather than asserting either.

2. neuromuscular_transmission_failure — the NMJ trunk

An endplate lesion (postsynaptic AChR subunit or clustering-pathway defect, synaptic basal lamina / acetylcholinesterase anchoring defect, or presynaptic release failure) → reduction in the safety factor of neuromuscular transmission → intermittent failure of endplate potentials to reach threshold → transmission block at a growing fraction of endplates → fatigable weakness.

The safety factor is the organizing quantity: normally 3–5 in adult mammals, and the thing that every CMS mechanism erodes, whether by reducing quantal content, reducing AChR number, altering endplate geometry, or desensitizing AChR through ACh overexposure (Engel et al., PMC6032286). Postsynaptic and endplate-development/maintenance defects account for roughly 77% of CMS cases, and ~40 CMS genes are now reported.

Three etiologic entry arms, matching the workshop deck's proposed NMJ split:

• genetic — congenital myasthenic syndromes
• autoimmune — myasthenia gravis (AChR / MuSK / LRP4), Lambert-Eaton
• toxin — botulinum toxin, α-neurotoxin envenomation, tick paralysis

Ready conformers: Myasthenia_Gravis, Congenital_Myasthenic_Syndrome, Congenital_Myasthenic_Syndrome_6, Botulism. Carries the acetylcholinesterase-inhibitor drug pattern (pyridostigmine target_mechanisms on the transmission node) and the 3,4-diaminopyridine presynaptic pattern.

Deliberately complementary to synaptic_vesicle_cycle, which stays the home for the presynaptic release machinery itself; this module models the endplate and the safety-factor consequence.

3. sarcolemmal_excitability_channelopathy — the muscle channelopathy trunk

The skeletal-muscle counterpart of the existing (and explicitly cardiac) cardiac_ion_channel_repolarization. Skeletal muscle ion channel variant (CLCN1 / ClC-1, SCN4A / NaV1.4, CACNA1S / CaV1.1, KCNJ2 / Kir2.1) → altered sarcolemmal and t-tubular resting potential or action-potential behavior → two mutually exclusive branches:

• hyperexcitable — reduced chloride conductance or impaired Na⁺ channel inactivation → repetitive after-discharges → myotonia (delayed relaxation)
• inexcitable — sustained depolarization → NaV1.4 inactivation → loss of action potential generation → flaccid periodic paralysis

Ready conformers: Thomsen_and_Becker_disease (CLCN1), Hypokalemic_Periodic_Paralysis, Andersen-Tawil_Syndrome (which currently sits on the cardiac side only).

This module is the direct dismech counterpart of the deck's proposal to give muscular channelopathy (MONDO:0019119) a real definition and reclassify it under myopathy.

4. denervation_reinnervation_remodeling — the motor unit trunk

Distinct from peripheral_axonal_degeneration, which models length-dependent sensory-and-motor distal axonopathy. This one models the motor unit as a unit: motor neuron or motor axon loss → NMJ denervation → collateral sprouting from surviving axons and reinnervation of orphaned fibers → motor unit enlargement with fiber-type grouping → eventual exhaustion of sprouting capacity → neurogenic atrophy and weakness.

Ready conformers: Postpoliomyelitis_Syndrome (the paradigm case — decades of compensated reinnervation followed by late failure), Progressive_Muscular_Atrophy, Monomelic_Amyotrophy, Spinal_Muscular_Atrophy, Kennedy_Disease, Amyotrophic_Lateral_Sclerosis, Madras_Motor_Neuron_Disease.

Fills the Mondo deck's motor-neuron section, which is currently a stub ("Add content from discussion on 2026-08-13").

Tier 2

5. excitation_contraction_coupling_calcium_release

RYR1 / CACNA1S / STAC3 defect at the triad → dysregulated sarcoplasmic reticulum Ca²⁺ release, with two branches: reduced or uncoupled release producing core formation and congenital weakness, and hypersensitive release under volatile anaesthetic or succinylcholine producing uncontrolled Ca²⁺ efflux, hypermetabolic crisis, and rhabdomyolysis.

Ready conformers: Central_Core_Myopathy, Multiminicore_Disease, Malignant_Hyperthermia_of_Anesthesia — three entries that today share the same gene and none of the same graph. Carries the dantrolene target_mechanisms (INHIBITS) pattern on the Ca²⁺-release node.

6. rna_repeat_spliceopathy

Expanded CUG (DM1) or CCUG (DM2) repeat transcribed but retained → nuclear ribonuclear foci → MBNL protein sequestration with reciprocal CELF1 stabilization → reversion of alternative splicing to fetal patterns across dozens of transcripts → the multisystem DM phenotype, with individual features traceable to individual mis-splicing events (CLCN1 → myotonia, which links this module to module 3; INSR → insulin resistance; BIN1 → t-tubule disorganization).

Ready conformers: Myotonic_Dystrophy_Type_1; DM2 is not yet in the KB.

Distinct from fame_pentanucleotide_repeat_rna_toxicity, which is deliberately conservative about intronic TTTCA repeat RNA and makes no spliceopathy claim, and from polyglutamine_expansion_proteotoxicity, which acts at the protein level.

7. sarcomere_thin_filament_assembly

Sarcomeric or Z-disk protein defect (ACTA1, NEB, TPM2/TPM3, TNNT1; DES, CRYAB; TTN) → impaired thin-filament or Z-disk assembly and stability → pathological protein aggregation (nemaline rods, myofibrillar inclusions) → reduced specific force → congenital hypotonia and weakness. A candidate Xogenesis module — nemaline rod bodies are a formed pathological structure with an MPATH/OGMS anchor.

Ready conformers: Nemaline_Myopathy, Myofibrillar_Myopathy, TTN_Related_Myopathy_Dominant_Negative_TTNsv, Distal_Myopathy_6_Adult-Onset_Autosomal_Dominant.

8. immune_mediated_myofiber_injury

Three mechanistically distinct arms under one autoimmune-myopathy trunk, which is why the current lumping into "myositis" loses information:

• dermatomyositis — type I interferon signature plus complement membrane-attack complex on endomysial capillaries → capillary dropout → perifascicular atrophy (a microangiopathy, not primarily a myofiber-directed attack)
• polymyositis / inclusion body myositis — myofiber MHC class I upregulation → CD8⁺ cytotoxic T cell invasion of non-necrotic fibers
• immune-mediated necrotizing myopathy — anti-HMGCR or anti-SRP autoantibodies → myofiber necrosis with sparse cellular infiltrate (includes statin-associated autoimmune myopathy, which also makes this a treatment-toxicity module)

Ready conformers: Dermatomyositis, Polymyositis, Inclusion_Body_Myositis (which already carries the degenerative arm via cellular_senescence).

9. muscle_energy_failure_rhabdomyolysis

Block in glycogenolysis/glycolysis, fatty acid oxidation, or oxidative phosphorylation → ATP deficit under a workload pattern specific to the block (brief high-intensity for the glycogenoses; prolonged exertion, fasting, or illness for FAO defects) → failure of SERCA and Na⁺/K⁺-ATPase to maintain ionic gradients → Ca²⁺ dysregulation and myofiber necrosis → rhabdomyolysis, myoglobinuria, and pigment nephropathy.

Ready conformers: Glycogen_Storage_Disease_Type_V, Glycogen_Storage_Disease_Type_VII, Carnitine_Palmitoyltransferase_II_Deficiency, Neutral_Lipid_Storage_Myopathy, MSTO1-Related_Mitochondrial_Myopathy.

The workload-specificity is the mechanistically interesting part and is what distinguishes the two exercise-intolerance phenotypes clinically.

Tier 3 — proposed, decision needed

• nuclear_envelope_mechanotransduction (LMNA/EMD). Real cross-system reach (Emery_Dreifuss_Muscular_Dystrophy, Familial_Partial_Lipodystrophy, Dilated_Cardiomyopathy), but the unifying mechanism across those tissues is genuinely contested, which may make it premature.
• dux4_reactivation (FSHD). D4Z4 contraction or SMCHD1 loss → chromatin derepression → DUX4 expression in skeletal muscle → activation of an embryonic transcriptional program → myotoxicity. Mechanistically clean and a paradigm of epigenetic derepression, but currently one conformer (Facioscapulohumeral_Muscular_Dystrophy); may be better as disease-level curation until a second conformer exists.
• alpha_dystroglycan_matriglycan_glycosylation. Probably not a separate module: the muscle consequence is loss of laminin binding, i.e. exactly the linkage failure that module 1 models, and the enzymology is already partly served by congenital_disorder_of_glycosylation. Recommend modeling it as a substitution branch inside sarcolemmal_mechanical_fragility.
• SMN / snRNP assembly. Recommend against a module. A module should recur across disorders; SMN deficiency is essentially one disease. Curate at the Spinal_Muscular_Atrophy entry and route its muscle consequence through module 4.

Knowledge gaps

These are the places where the field does not have a settled answer, and where dismech's mechanistic_hypotheses, KNOWLEDGE_GAP, and HUMAN_MODEL_MISMATCH machinery is the right way to record the state of knowledge instead of picking a side.

1. Dystrophin: mechanical fragility or mechanotransduction?

The textbook account is that dystrophin is a shock absorber and its loss makes the sarcolemma tear under eccentric contraction. The competing account is that the DGC is primarily a mechanotransduction and signaling hub, and that much of the pathology follows from disrupted signaling — notably nNOS mislocalization producing functional ischemia during exercise — rather than from frank membrane rupture (Comms Biol 2022). The distinction matters therapeutically: micro-dystrophin constructs that restore mechanical linkage but omit the nNOS-binding domain test the two accounts directly.

→ Curate as competing mechanistic_hypotheses on sarcolemmal_mechanical_fragility (mechanical_fragility_model CANONICAL vs mechanotransduction_signaling_model ALTERNATIVE), with disease-level edges opting into groups.

2. ALS: dying-back or dying-forward?

NMJ denervation is among the earliest detectable events in ALS, and 2025 work frames skeletal muscle as an active participant in degeneration rather than a passive target, with NMJ regenerative capacity as a therapeutic axis (bioRxiv 2025; State of the Art 2025). Whether the primary lesion is in the motor neuron soma with distal consequence, or at the periphery with retrograde consequence, is unresolved.

→ mechanistic_hypotheses on Amyotrophic_Lateral_Sclerosis and on denervation_reinnervation_remodeling.

3. DM1: MBNL sequestration does not account for the whole phenotype

Mouse models reproduce substantially less splicing mis-regulation than is seen in DM1 patient brain, and additional mechanisms — mRNA translation, localization, and stability, repeat-associated non-AUG translation, CELF1 hyperphosphorylation — are implicated but not quantified.

→ HUMAN_MODEL_MISMATCH discussion on rna_repeat_spliceopathy: evidence exists in the model, translational validity to the human CNS phenotype is the open question. This is precisely the distinction from a plain KNOWLEDGE_GAP.

4. Selective vulnerability is unexplained across the whole domain

Extraocular muscles are spared in Duchenne but preferentially targeted in myasthenia gravis and mitochondrial myopathy. Cranial motor units differ in vulnerability in ALS and SMA. There is a measurement trap here worth recording: in the most vulnerable muscles there is underlying loss of endplates, which can mask denervation on standard assays.

→ KNOWLEDGE_GAP discussions with proposed_experiments; the endplate-loss caveat belongs in limitations on any model link that uses endplate occupancy as a readout.

5. The safety factor is a model-system quantity

Neuromuscular transmission safety factor is the organizing concept for the whole NMJ module, and it is measured in mouse and in intercostal biopsy — almost never in a living patient, where the surrogate is repetitive nerve stimulation decrement. The surrogate has not been validated against the underlying quantity across CMS subtypes.

→ KNOWLEDGE_GAP on neuromuscular_transmission_failure, in the same spirit as the DTI-ALPS imaging-surrogate caveat in glymphatic_dysfunction.

6. Post-therapy natural history is a new and unmodeled phenotype

Nusinersen- and onasemnogene-treated SMA, and exon-skipping-treated Duchenne, are producing disease courses that no pre-therapy natural history describes. What residual deficit remains, and which compartment it lives in, is open.

→ Worth an explicit KNOWLEDGE_GAP rather than silently curating pre-therapy natural history as if it still applied.

7. NMJ aging and neuromuscular disease converge

The NMJ is a shared vulnerability of aging and disease, and biomarkers are crossing over — neurofilament light is elevated in severe sarcopenia as well as in motor neuron disease. This is a candidate link between this project and the existing aging modules (inflammaging, stem_cell_exhaustion, cellular_senescence).

8. Entry-level gaps in the knowledge base

Missing entries that would each unlock a module conformer:

9. Ontology-side gaps carried over from the Mondo workshop

Recorded here because they constrain what dismech can bind to, not because this project owns them:

• No Mondo term for toxin-induced neuromuscular junction disease (proposed in the workshop deck to cover botulism, tetanus, tick paralysis, and envenomation).
• No Mondo terms for CMS associated with UNC13A, RPH3A, LAMA5, MACF1, GMPPB.
• No acquired neuromuscular disease counterpart to hereditary neuromuscular disease (MONDO:0100546); the workshop flags that there is no semantic basis on which to maintain such a class.
• atrophic muscular disease (MONDO:0004714) and muscular channelopathy (MONDO:0019119) are proposed to move under myopathy, and myopathy (MONDO:0005336) to become a direct child of neuromuscular disease.

Proposed groupings

kb/groupings/ currently holds Motor_Neuron_Disorders and Charcot-Marie-Tooth_Diseases for this domain. Candidates to add, each with a CONFORMS_TO_MODULE necessary criterion once the corresponding module exists:

• Muscular_Dystrophies → sarcolemmal_mechanical_fragility
• Congenital_Myopathies → sarcomere_thin_filament_assembly / excitation_contraction_coupling_calcium_release
• Skeletal_Muscle_Channelopathies → sarcolemmal_excitability_channelopathy
• Neuromuscular_Junction_Disorders → neuromuscular_transmission_failure
• Idiopathic_Inflammatory_Myopathies → immune_mediated_myofiber_injury

Worklist

• [ ] Create sarcolemmal_mechanical_fragility module with competing mechanical / mechanotransduction hypotheses
• [ ] Create neuromuscular_transmission_failure module with the three etiologic arms and the safety-factor knowledge gap
• [ ] Create sarcolemmal_excitability_channelopathy module
• [ ] Create denervation_reinnervation_remodeling module
• [ ] Wire Duchenne_Muscular_Dystrophy, Myasthenia_Gravis, Congenital_Myasthenic_Syndrome, Thomsen_and_Becker_disease, and Postpoliomyelitis_Syndrome as the first conformers, one per Tier 1 module
• [ ] Create excitation_contraction_coupling_calcium_release and wire the three RYR1 entries
• [ ] Create rna_repeat_spliceopathy with the DM1 HUMAN_MODEL_MISMATCH
• [ ] Create sarcomere_thin_filament_assembly (evaluate Xogenesis anchor for nemaline rods)
• [ ] Create immune_mediated_myofiber_injury with the three-arm split
• [ ] Create muscle_energy_failure_rhabdomyolysis
• [ ] Decide Tier 3: laminopathy, DUX4, matriglycan-as-branch
• [ ] Add the six missing disorder entries listed under knowledge gap 8
• [ ] Add the five proposed groupings