Extreme exercise in males is linked to mTOR signalling and onset of amyotrophic lateral sclerosis

Abstract

Amyotrophic lateral sclerosis (ALS) is thought to be caused by interaction between genetic and environmental factors leading to motor neuron (MN) degeneration. Physical exercise has been linked to ALS but controversy remains. A key question is to determine which individuals might be at risk of exercise-associated ALS, because unnecessary avoidance of exercise could be harmful. We implemented complementary strategies including Mendelian randomization (MR) and multiple questionnaire-based measures of physical exercise in different cohorts. We include a prospective study involving UK Biobank participants where we could test for a relationship between exercise and the timing of future ALS symptom onset. To interrogate the molecular basis of our observations we performed a genetic association study of 'extreme' exercise, equivalent to >6 h of strenuous exercise or >12 h of any leisure-time exercise per week. Our data suggest that the link between increased physical exercise and ALS is particularly important for males who perform the most activity; with no evidence of a link in females. We determined that extreme exercise in males is associated with loss-of-function genetic variants within a number of mammalian target of rapamycin (mTOR) signalling genes that are also differentially expressed in ALS spinal cord. Activity-induced mTOR signalling has been shown to selectively benefit MN. Therefore, our findings could imply that moderate exercise is neuroprotective via enhanced mTOR signalling, but extreme exercise in men is associated with neurotoxicity and ALS via a failure of this mechanism. There was no significant overlap between genes associated with extreme exercise and those associated with ALS risk, consistent with a true gene-environment interaction rather than a shared genetic basis. We are not yet able to make individual-level recommendations regarding exercise and risk of ALS, but our conclusions should provide focus for future investigation.

Keywords: amyotrophic lateral sclerosis (ALS); exercise; gene–environment interaction; mammalian target of rapamycin (mTOR) signalling.

Figure 1

Approach to identify those at risk of exercise-associated ALS. We hypothesized that exercise-associated ALS may be determined by frequent or strenuous exercise in a sex-specific manner. To investigate this, we implemented a set of three complementary strategies including two-sample Mendelian randomization (MR) utilizing genetic liability to exercise; and multiple different questionnaire measures of exercise, including a prospective study of UK Biobank (UKB) participants where we could test for a relationship between exercise and the timing of future ALS symptom onset. Finally, to gain molecular-level insight, we examined rare genetic variants associated with an extreme exercise phenotype in males and females separately. ALS also has a rare variant architecture and one consequence of this final analysis, is that we could determine whether a common set of rare genetic variants predisposes to both extreme exercise and ALS. Created in BioRender. Cooper-Knock, J. (2025) https://BioRender.com/j09u606. ALS = amyotrophic lateral sclerosis; GWAS = genome-wide association study.

Figure 2

Two-sample Mendelian randomization (MR) tests for a causal effect of frequent or strenuous leisure-time exercise (SSOE) on risk of ALS. Scatter plots demonstrate a significant association of exercise with risk of ALS, including in robust MR tests, for (A) males, but not for (B) females. Each point represents the effect size (beta) and standard errors for each SNP–outcome relationship. See the ‘Materials and methods’ section for details of SSOE. ALS = amyotrophic lateral sclerosis; SNP = single nucleotide polymorphism; SSOE = strenuous sport and other exercise.

Figure 3

Higher level of adulthood leisure-time physical activity is associated with younger age of ALS symptom onset in males but not in females. Historical physical activity was quantified using the HAPAQ questionnaire. Plots show total leisure-time physical activity energy expenditure (PAEE) for each individual measured in kJ/min, against age of symptom onset for ALS patients (A and B) and age of study enrolment for controls (C and D). Males (A and C) are plotted separately from females (B and D). For a difference in average leisure-time activity of 100 kJ/min, equivalent to very strenuous exercise in an 80 kg male, there is a 2-fold increase in risk of ALS symptom onset at any given time. A linear regression line is calculated by OLS; shading represents a confidence interval calculated from the standard error of the regression line at each point. ‘r’ indicates the Pearson correlation coefficient. HAPAQ = historical adulthood physical activity questionnaire; OLS = ordinary least squares.

Figure 4

Prospectively measured physical activity is linked to age of symptom onset of ALS in males. Utilizing the UKB enrolment questionnaire, total physical activity is quantified in MET minutes per week for participants who later developed ALS. As shown by Kaplan–Meier curves, for males (A and C) but not for females (B and D), increased physical activity is associated with earlier age of ALS symptom onset (A and B) and shorter time from questionnaire to symptom onset (C and D). Lower and higher levels of physical activity were defined by comparing quartiles of MET minutes per week. We performed a sensitivity analysis whereby we sequentially removed overlapping groups of 20 participants with progressively increasing MET minutes per week of physical activity and repeated a Cox regression test of the association between physical activity and the timing of future ALS symptom onset. The regression coefficient for the effect of physical activity (y-axis) is plotted against the mean physical activity of removed participants (x-axis) with 95% confidence intervals and line of best fit for age of ALS symptom onset (E), and the time interval from questionnaire to symptom onset (F). The regression coefficient is negatively correlated with the mean value of physical activity performed by the removed subset, suggesting that the observed association between physical activity and ALS symptom onset is disproportionately driven by individuals who perform the highest levels of physical activity. See the ‘Materials and methods’ section for further details. ALS = amyotrophic lateral sclerosis; UKB = UK Biobank; MET = metabolic equivalent value.

Figure 5

Rare genetic variant burden testing links extreme exercise to loss-of-function mutations within mTOR signalling genes, which are also differentially expressed in ALS spinal cord. (A) We explored alternative but non-exclusive models for the link between genetic background, extreme exercise and ALS risk. Hypothetically, genetic risk could be linked to exercise and ALS independently (top) or genetic drivers of extreme exercise could act on ALS risk only through performed exercise (bottom). We did not identify overlap between genes linked to extreme exercise and ALS risk genes suggesting the latter model is correct. Bulk RNA sequencing of ALS (n = 154) and control (n = 54) spinal cord and analysis of gene expression was previously performed. Transcript expression is quantified as TPM and plotted against disease status for cervical and lumbar spinal cord segments. Box plots show median and 95% confidence interval of expression. Results are shown for four mTOR signalling genes—(B) NCKAP1L, (C) GOLPH3, (D) LAMTOR2 and (E) RFFL—which are also enriched with loss-of-function mutations in males who perform extreme exercise. TPM indicates transcripts per kilobase million or the number of read counts per length of transcript per million reads mapped. (F) Previous studies have suggested that NMJ activation can be neuroprotective via activation of mTOR signalling. We suggest a model whereby exercise, via activation of mTOR signalling can ‘normally’ protect MN from disease (F, top). However, genetic inhibition of mTOR signalling prevents this protection while simultaneously predisposing (green arrow) to extreme exercise which is now rendered neurotoxic (F, bottom). Created in BioRender. Cooper-Knock, J. (2025) https://BioRender.com/zxcqdn5. mTOR = mammalian target of rapamycin; ALS = amyotrophic lateral sclerosis; TPM = transcripts per kilobase million; NMJ = neuromuscular junction; MN = motor neuron.