Detection of a Mitochondrial Fragmentation and Integrated Stress Response Using the Cell Painting Assay

Mitochondria are cellular powerhouses and are crucial for cell function. However, they are vulnerable to internal and external perturbagens that may impair mitochondrial function and eventually lead to cell death. In particular, small molecules may impact mitochondrial function, and therefore, their influence on mitochondrial homeostasis is at best assessed early on in the characterization of biologically active small molecules and drug discovery. We demonstrate that unbiased morphological profiling by means of the cell painting assay (CPA) can detect mitochondrial stress coupled with the induction of an integrated stress response. This activity is common for compounds addressing different targets, is not shared by direct inhibitors of the electron transport chain, and enables prediction of mitochondrial stress induction for small molecules that are profiled using CPA.


Figure S1 (
Figure S1 (related to Figure 1): CPA profiles for deferoxamine (DFO) and phenanthroline.(A and B) Comparison of the profiles for DFO (A) and phenanthroline (B) at different concentrations.The top line profile is set as a reference profile (100 % biological similarity, BioSim) to which the following profiles are compared.Blue color: decreased feature; red color: increased feature.(C) Cluster biosimilarity heatmap for DFO.Percent values are displayed.(D) Dose-dependent change in dyerelated CPA features at different concentrations DFO.(E) Z scores for top 30 altered features with high Z scores for ciclopirox at 30 µM as determined in CPA.Cpd: compound; BioSim: biosimilarity; Ind: induction; Conc: concentration.

Figure S2 (related to Figure 2 )
Figure S2 (related to Figure 2): Profile comparison for ciclopirox (CPX) and biosimilar compounds considering the features related to each stain only.

Figure S3 (related to Figure 2 )
Figure S3 (related to Figure 2): Profile analysis for inhibitors of the mitochondrial electron transport chain (ETC).(A) Profile cross-similarity for ETC inhibitors.(B) Comparison of the profile of ciclopirox (CPX) at 30 µM to the profiles of ETC inhibitors.(C) Comparison of the profiles for ciclopirox (CPX) at 30 µM to the profiles of ETC inhibitors considering only MitoTracker-related features.For B and C: the top line profile is set as a reference profile (100 % biological similarity, BioSim) to which the following profiles are compared.Blue color: decreased feature; red color: increased feature.Cpd: compound; BioSim: biosimilarity; Ind: induction; Conc: concentration.

Figure S4 (
Figure S4 (related to Figure 2): CPA profile for SB525334 and compound 2. (A and B) Cluster biosimilarity heatmap for SB525334 (A) and compound 2 (B) Values are biosimilarity in %. (C) Comparison of the profile of ciclopirox (CPX) at 30 µM to the profile of compound 2 at 10 µM.The top line profile is set as a reference profile (100 % biological similarity, BioSim) to which the following profile is compared.Blue color: decreased feature; red color: increased feature.Cpd: compound; BioSim: biosimilarity; Ind: induction; Conc: concentration.L/CH: Lysosomotropism/cholesterol homeostasis; PYR: pyrimidine.(D) Quantification of the MitoTracker staining in CPA.Z scores for selected cellsrelated features are shown.

Figure S5 :
Figure S5: Influence of the compounds on cell growth.U-2OS cells were treated with the compounds for 48 h in presence of propidium iodide (PI) and CellEvent™ Caspase-3/7 Green to detect cell toxicity and apoptosis, respectively.Images were acquired every hour using the IncuCyte Zoom imaging system.Image-based analysis was performed to quantify cell growth by means of cell confluence as a readout, or cell toxicity and apoptosis by means of PI and caspase 3/7 activity-related fluorescence.(n =3; mean values ± SD). (A) Ciclopirox.(B) GSK-J4.(C) SB525334.(D) Compound 2.

Figure S6 :
Figure S6: Influence on the mitochondrial membrane potential.U-2OS cells were treated with the compounds for 24 h prior to the addition of tetramethylrhodamine, methyl ester (TMRE) to determine mitochondrial membrane potential.FCCP was used as a positive control (n =3; mean values ± SD).

Figure S8 (
Figure S8 (related to Figure 5): Influence of the compounds on HIF1 signaling and iron chelation.(A) HIF1-α-dependent reporter gene assay.HEK293 cells transfected with pGL4.22-PGK1-HRE::dLUC and Renilla luciferase-expressing plasmids were treated with the compounds or DMSO as a control for 24 h prior to detection of firefly and Renilla luciferase activities.ML228 and CoCl2 were used as controls for HIF1 induction.Mean values ± SD (n =3).(B) Detection of HIF1-α protein levels in U-2OS cells after treatment with the compounds for 24 h.Cells were treated with the compounds prior to detection of HIF1-α and -actin as a reference protein using immunoblotting.Representative blot is shown (n =3).Lanes in the immunoblot were rearranged to fit the figure.See Figure S9 for the full blots.(C) Using Ferrozine-Fe(II) complex formation to determine iron chelation by the compounds.Mean values ± SD (n =3).

Figure S10 (
Figure S10 (related to Figure 7): Median cluster subprofiles of the 13 defined clusters.The median cluster subprofiles for all clusters besides MitoStress were previously reported 2 .Blue color: decreased feature, red color: increased feature.

Figure
Figure S12 (related to Figure 7): Cluster biosimilarities for the profiles of compounds studied for mitotoxicity.(A) Compounds from Seal et al. 3 (B) Compounds from Trapotsi et al. 4 Percent values are given.Cpd: compound; Conc: concentration; PYR: pyrimidine.

Figure S13 (related to Figure 7 )
Figure S13 (related to Figure 7): Profile analysis for salubrinal.(A) Chemical structures of salubrinal and Sal003.(B, C) Profile similarity for salubrinal and Sal003, respectively.The top line of the heatmap profile is set as a reference profile (100 % biological similarity) to which the following profiles are compared.Blue color, decreased feature; red color, increased feature.(D) Cluster biosimilarity heatmap for salubrinal.Percent values are given.(E) Profile cross-similarity for salubrinal and Sal003.Values are biosimilarity in %.Cpd: compound; BioSim: biosimilarity; Ind: induction; Conc: concentration.L/CH: Lysosmotropism/cholesterol homeostasis; PYR: pyrimidine.

Figure S14 (
Figure S14 (related to Figure 7): Profile analysis for ponatinib.(A) Chemical structure of ponatinib.(B) Cluster biosimilarity heatmap for ponatinib.Percent values are given.(C) Profile similarity for ponatinib compared to SB525334.The top line of the heatmap profile is set as a reference profile (100 % biological similarity) to which the following profiles are compared.Blue color, decreased feature; red color, increased feature.Only the MitoTracker-related features of the MitoStress cluster subprofile were used for the comparison.Cpd: compound; BioSim: biosimilarity; Ind: induction; Conc: concentration.L/CH: Lysosmotropism/cholesterol homeostasis; PYR: pyrimidine.

Table S5 (related to Figure 5 and 6):
Proteome profiling data (see separate XLS file).

Table S7 (related to Figure 5 and 6): Overlap of downregulated proteins in Quiros et al. 10 and this study.
The table lists proteins, which were reported by Quiros et al. to be downregulated upon the treatment with mitochondrial stressor, that were also downregulated by ciclopirox (CPX), GSK-J4, SB525334 or compound 2 (cpd 2).Data for SB525334 is shown for log fold change (FC) of 0.3.and 0.2 as only little changes were detected at log FC of 0.2.

Table S8 (related to Figure 5 and 6): Upregulated genes and proteins in Quiros et al. 10 and this study.
The table lists genes and proteins, which were reported by Quiros et al. to be downregulated upon the treatment with mitochondrial stressor, that were also upregulated by ciclopirox (CPX), GSK-J4, SB525334 or compound 2 (cpd 2).Data for SB525334 is shown for log fold change (FC) of 0.3.and 0.2 as only little changes were detected at log FC of 0.2.