Fig. 6: Compound prediction algorithm identifying neuroprotective/neurotoxic compounds.
a, Flowchart depicting the in silico drug screening. We computed and correlated the conserved KEGG pathway enrichments for NeuronAge and all compounds from the CMAP dataset that are measured on the neuronal cell line NEU. To obtain a manageable list, we filtered for compounds that were measured at least twice, show consistent correlations in all measurements, have a stronger correlation at the 24-h time point compared to the 6-h time point, have information in PubCHEM and have at least an absolute correlation value of 0.25. b, The top anti-NeuronAge and pro-NeuronAge compounds after the filtering steps ranked according to their Pearson correlation. Previously published neuroprotective (blue) or neurotoxic (orange) compounds are indicated. Resveratrol, which has mixed evidence in the literature, is shown in both colors. Four compounds were selected for experimental validation of our predictions and are highlighted by black arrows and their structural formula is given. c, Fraction plots displaying the fraction of nematodes at day 1 of adulthood expressing neuronal volume markers in different neurons categorized as ‘healthy’, ‘mildly damaged’ and ‘severely damaged’ that were treated with 2.5 mM syringic acid (SA), 10 nM vanoxerine (VX), 25 nM WAY-100635 (WAY) or 250 µM (S)-(−)-Bay K8644 for 24 h. Three to four cohorts were analyzed, consisting of 10–25 individual nematodes. Data shown are the mean ± s.d. Statistical analysis was performed using a CLM with a logit link to account for the ordered categorical nature of neuronal damage scores (healthy < mild < severe). P values were adjusted for multiple testing using the Benjamini–Hochberg method. d, Fraction plot displaying the fraction size of healthy (blue), mildly damaged (orange) and severely damaged (red) URY neurons (JKM10 strain) on day 2 of adulthood after 48-h compound treatment with 25 nM WAY-100635 (WAY), 2.5 mM SA or water control. Data shown are the mean + s.d. of five independent cohorts of 10–25 nematodes each. Statistical analysis was performed using a CLM with a logit link to account for the ordered categorical nature of neuronal damage scores (healthy < mild < severe). Pairwise comparisons between conditions were performed using the ‘emmeans’ package with Tukey’s adjustment for multiple testing. e, Swarm plot displaying the avoidance index of compound-treated JKM10 (URYp::GFP) nematodes (25 nM WAY-100635 (WAY), 2.5 mM SA or water control for 48 h) on day 2 of adulthood. Data shown are the mean ± s.d. from five independent cohorts of 50–75 nematodes. Two-way ANOVA + Tukey post hoc test were used to test for significant differences.
