A visual map of a fifteen-year experimental programme running from axonal biology through transcriptomics to an AI-driven target-prioritisation platform — SBMA and ALS4 Research Group, UCL Queen Square Institute of Neurology.
Two genome-wide screens — embryonic motor neurons (2019) and iPSC-derived motor neurons (2025) — mapped disease-associated transcriptional change and pointed to Chmp7 and the ESCRT-III trafficking machinery as an early, disease-specific signal.
2019: embryonic (E13) AR100 vs wild-type spinal-cord motor neurons, senior author, Sci Rep. 2025: dihydrotestosterone-treated SBMA patient iPSC motor neurons vs isogenic controls, co-senior author, bioRxiv preprint.
Presymptomatic (3-month) tissue unless noted. Chmp7 change was specific to SBMA: unchanged in the SOD1-G93A ALS and SETX-R2136H ALS4 mouse models.
STRING network hubs from the 2019 screen centred on EGFR, UQCRC1 (mitochondrial respiratory chain), FANCE (DNA repair) and CTNNB1 (Wnt effector), with enrichment across Wnt signalling, focal adhesion, endocytosis and oxidative phosphorylation.
Convergent pathways across species: both the 2019 mouse and 2025 human-iPSC screens show early dysregulation of spliceosome, cell-cycle and mitochondrial pathways — evidence of a developmental component to SBMA that precedes overt degeneration.
Cross-species validation: CHMP7 was also reduced in 2 of 3 SBMA patient iPSC motor neuron lines relative to controls (1.47-fold, 1.32-fold, both p<0.01) — the same trafficking signal seen first in mouse.
Longitudinal physiology in the AR100 mouse (Gray et al., 2020) showed hindlimb muscle force failing well before any detectable motor neuron loss — establishing muscle as a primary, early site of SBMA pathology rather than a downstream bystander.
Contractile slowing (time-to-peak, half-relaxation) also becomes significant from 12–18 months. EDL fatigue index reaches ~45.7% at 6 months. Motor neuron loss is not detectable until late-stage disease.
Muscle transcriptomics (with Leonnette Annan): 508 genes differentially regulated between AR20 and AR100 muscle at the presymptomatic stage (p<0.05, FDR<0.05), 188 alternatively spliced — enriched for extracellular matrix, proteasome, mitochondrial function, focal adhesion and PI3K–Akt signalling.
A distinct molecular route: the muscle pathway signature is different from, but complementary to, the motor-neuron signature above — muscle and spinal cord appear to degenerate through separate mechanisms in SBMA.
Three mechanistic studies, each testing a candidate driver of motor neuron death and, where the hypothesis held, a corresponding intervention.
Biochemical and live-transport assays in AR100 motor neurons found no impairment of axonal transport, unlike other polyQ and motor neuron diseases.
Mechanism ruled outOral arimoclomol after symptom onset delayed progression, preserved functional motor units, reduced muscle atrophy and raised Vegf. Arimoclomol subsequently entered ALS clinical trials.
Therapeutic effect shownCa²⁺ depletion and ER stress found from the presymptomatic stage onward; salubrinal reduced ER-stress-associated apoptosis in vitro via caspase-12.
Candidate target identifiedMyoNeuroTwin carries the disease pathways identified above — mitochondrial dysfunction, DNA-damage-repair failure, proteostasis, immune dysregulation — from wet-lab discovery into a computational decision framework for target prioritisation.