High-throughput tracking enables systematic phenotyping and drug repurposing in C. elegans disease models
- Institute of Clinical Sciences, Imperial College London, United Kingdom
- MRC London Institute of Medical Sciences, United Kingdom
Figures
Figure 1
Overview of genotype-phenotype mapping and disease model panel.
Disease modelling as genotype-phenotype mapping in humans and a model organism. Arrows show the progression from symptom identification (a phenotype outside the healthy range) to genotyping (placing a patient in genotype space by identifying a genetic variant), disease model creation (making a corresponding mutation in a model organism), and model organism phenotyping. Phenotypic drug screens identify compounds that move a disease model toward the wild-type phenotype (green arrow). These candidates can then be tested in humans (dashed green arrow). Thin lines show other symptom-gene-ortholog-phenotype connections. (B) Venn diagram showing a number of conserved genes (Ortholist 2), those involved in neuron or muscle function (Wormbase), and those associated with human genetic disorders according to the Online Mendelian Inheritance in Man (OMIM) database. (C) Sequence similarity between human and C. elegans genes and the total number of orthology programs predicting that the gene is an ortholog.
Figure 2
Diverse multidimensional behavioural phenotypes are obtained across the diverse panel of disease model mutants.
(A) Representative field of view for a single camera channel with individual skeletonised worm images of bbs-1(syb1588) (top) and N2 (bottom) imaged on the same plate viewed using Tierpsy Tracker. (B) Representative behavioural phenotypes extracted by Tierpsy, representing changes in morphology, posture, and locomotion of different disease mutant strains. Boxes show interquartile range, error bars show minimum and maximum values excluding outliers. p-values are for comparisons to wild-type N2 worms, block permutation t-test, corrected for multiple comparisons. (C) Hierarchical clustering of behavioural fingerprints. Features are Z-normalised. The top barcode shows the period of image acquisition where the behavioural feature was extracted by Tierpsy: pre-stimulation (pink), blue light (blue), and post-stimulation (green). (D) Principal component analysis of the disease model mutants and N2 reference (blue). Strains move in phenospace between pre-stimulus (circular points), blue light (crosses) and post-stimulation (squares) recordings. Strains with an aberrant blue light response are shown in red. Error bars represent the standard error of the mean. (E) The total number of statistically significant behavioural features for each strain compared to N2 (from the total set of 8289 features extracted by Tierpsy). p-values for each feature were calculated using block permutation t-tests, using n=100,000 permutations, and p<0.05 considered statistically significant after correcting for multiple comparisons using the Benjamini-Yekutieli method.
Figure 3
Disease model phenologs.
(A–D) Key behavioural features altered in loss-of-function mutant strains associated with ciliopathies: bbs-1(syb1588), bbs-2(syb1547), tub-1(syb1562) and tmem-231(syb1575), under baseline (pre-stimulus) imaging conditions. Individual points marked on the box plots are well-averaged values (three worms per well) for each feature across the independent days of tracking. Boxes show interquartile range, error bars show minimum and maximum values excluding outliers. p-values are for comparisons to wild-type N2 worms using block permutation t-tests (n=100,000 permutations, correcting for multiple comparisons using the Benjamini-Yekutieli method). (E–H) Changes in selected features in response to stimulation with a single 10 s blue light pulse (blue-shaded region). Feature values were calculated using 10 s windows centred around 5 s before, 10- s after, and 20 s after the beginning of each blue light pulse. (I) Heatmap of the entire set of 8289 behavioural features extracted by Tierpsy for the disease model strains associated with ciliopathies and N2. The ‘stim type’ barcode denotes when during image acquisition the feature was extracted: pre-stimulation (pink), blue light stimulation (blue), and post-stimulation (green). Asterisks show the location of selected features present in A-E.
Figure 4 with 1 supplement
Na+ cation leak channel (NALCN) disease model phenologs.
(A–D) Key behavioural and postural features altered in loss-of-function mutant strains associated with NALCN mutants: nca-2(syb1612), unc-77(syb1688), and unc-80(syb1531), under baseline (pre-stimulus) imaging conditions. Individual points marked on the box plots are well-averaged values (three worms per well) for each feature across the independent days of tracking.Boxes show interquartile range, error bars show minimum and maximum values excluding outliers. p-values are for comparisons to wild-type N2 worms using block permutation t-tests (n=100,000 permutations correcting for multiple comparisons using the Benjamini-Yekutieli method). (E–H) Changes in selected features in response to stimulation with a single 10 s blue light pulse (blue-shaded region). Feature values were calculated using 10 s windows centred around 5 s before, 10- s after, and 20 s after the beginning of each blue light pulse. (E) A representative ‘fainting phenotype’ for unc-80(syb1531) and nca-2(syb1612), characterised by an increase in pausing following the cessation of stimulation with blue light. (I) Heatmap of the entire set of 8289 behavioural features extracted by Tierpsy for the disease model strains associated with NALCN disease and N2. The ‘stim type’ barcode denotes when during image acquisition the feature was extracted: pre-stimulation (pink), blue light stimulation (blue), and post-stimulation (green). Asterisks show the location of selected features present in A-D.
Figure 4—figure supplement 1
Number of initial compound hits detected when analysing increasing numbers of features.
The number of compound hits (as defined in the main text of this study) identified when performing statistical analysis (as detailed in the methods section) across the hand-selected core behavioural features (3), the entire feature set (8000), and for pre-defined Tierpsy feature sets containing an increasing number of features (Tierpsy-8, Tierpsy-16, Tierpsy-256, and Tierpsy-2k). There is a loss of statistical power when analysing >256 features using a low number of replicates (n<9) and no statistically significant differences can be identified between the different drug treatments and the mutant or wild-type control after correcting for multiple comparisons.
Figure 5 with 1 supplement
Drug repurposing screening.
(A) Phenotypes of unc-80(syb1531) mutant (red star) and N2 (blue star) worms treated with 1% DMSO, and unc-80(syb1531) mutants treated with a library of 743 FDA-approved drugs at a concentration of 100 μM for 4 hr (circles). Each point represents an average of three well replicates, across three independent days of tracking (n=9 total). Blue points are the top 30 compounds that significantly improved all three of the core behavioural features, pushing the unc-80 mutant strain towards the control in phenospace. (B) Confirmation screen of the top 30 compounds identified in the initial library screen. Again unc-80(syb1531) and N2 DMSO treated controls are represented by red and blue stars, respectively, and each circular point represents unc-80(syb1531) treated with 100 μM compound for 4 hr. The 13 compounds coloured in the yellow lead to the worsening of >1000 behavioural features (see below). Liranaftate and atorvastatin both lead to a rescue of the core mutant phenotype with a low number of side effects. (C) Total number of behavioural ‘side effects’ following treatment of unc-80(syb1531) with the 30 compounds in the confirmation screen. Side effects are defined as features that are not significant between unc-80 mutants and wild-type N2 worms treated with 1% DMSO but where there is a significant difference between unc-80 mutants treated with a drug compared to N2. Red dashed line separates drug treatments that lead to a worsening of >1000 behavioural features that correspond to the points coloured in yellow in the 3D scatterplot.
Figure 5—figure supplement 1
Behavioural side effects across a reduced behavioural featureset.
Total number of behavioural ‘side effects’ within a reduced set of features extracted by Tierpsy, Tierpsy-256, following treatment of unc-80(syb1531) with the 30 compounds in the confirmation screen. Side effects are defined as features that are not significant between unc-80 mutants and wild-type N2 worms treated with 1% DMSO but where there is a significant difference between unc-80 mutants treated with a drug compared to N2. The plot is coloured as described in Figure 5.
Additional files
-
Supplementary file 1
Table indicating the unique Wormbase and Ensembl database accession numbers for every C. elegans gene in our panel of disease model mutants.
Alongside denoting which (and how many) gene orthology programs predict that the worm gene is an orthologous to a human gene, and the genetic similarity and Blast-E score for each C. elegans gene.
- https://cdn.elifesciences.org/articles/92491/elife-92491-supp1-v1.xlsx
-
Supplementary file 2
Document with a 1-page summary of each disease model created including basic genetic information, human and worm gene names, phenotypic fingerprint, and box plots for selected features comparing the disease model to wild-type worms.
- https://cdn.elifesciences.org/articles/92491/elife-92491-supp2-v1.pdf
-
Supplementary file 3
Summary of the associated human ortholog(s), predicted functional class, and key associated human disease phenotypes (Human Phenotype Ontology database) for each C. elegans disease model mutant.
- https://cdn.elifesciences.org/articles/92491/elife-92491-supp3-v1.xlsx
-
MDAR checklist
- https://cdn.elifesciences.org/articles/92491/elife-92491-mdarchecklist1-v1.docx
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
High-throughput tracking enables systematic phenotyping and drug repurposing in C. elegans disease models
eLife 12:RP92491.
https://doi.org/10.7554/eLife.92491.4
