High-Throughput Microscopy Analysis of Mitochondrial Membrane Potential in 2D and 3D Models

Abstract

Recent proteomic, metabolomic, and transcriptomic studies have highlighted a connection between changes in mitochondria physiology and cellular pathophysiological mechanisms. Secondary assays to assess the function of these organelles appear fundamental to validate these -omics findings. Although mitochondrial membrane potential is widely recognized as an indicator of mitochondrial activity, high-content imaging-based approaches coupled to multiparametric to measure it have not been established yet. In this paper, we describe a methodology for the unbiased high-throughput quantification of mitochondrial membrane potential in vitro, which is suitable for 2D to 3D models. We successfully used our method to analyze mitochondrial membrane potential in monolayers of human fibroblasts, neural stem cells, spheroids, and isolated muscle fibers. Moreover, by combining automated image analysis and machine learning, we were able to discriminate melanoma cells from macrophages in co-culture and to analyze the subpopulations separately. Our data demonstrated that our method is a widely applicable strategy for large-scale profiling of mitochondrial activity.

Keywords: NSCs; TMRM; co-culture; machine learning; mitochondrial membrane potential; single muscle fibers; spheroids.

Figure 1

High-content measurement of ΔΨm in cell cultures. (A) Representative images of primary human fibroblasts (HDFa, IMR-90) stained with TMRM and Hoechst (scale bar 50 µm). (B) Mitochondrial membrane potential (ΔΨm) kinetics of human fibroblasts. To study the changes in ΔΨm, the cells were treated with oligomycin and FCCP. Data represent the mean ± SEM of 5–6 biological replicates.

Figure 2

High-content imaging of mitochondrial membrane potential in 3D cell models. (A) Representative images of TMRM, Hoechst, and brightfield of LUHMES cells; scale bar: 200 μm. TMRM fluorescence intensity analysis of different classes: (B) single cells, (C) grouped, (D) aggregates, and (E) spheroids. The values reported in the graph have been normalized on basal levels. Data represent the mean ± SEM of 3 biological replicates.

Figure 3

High-content microscopy of ΔΨm kinetics in isolated muscle fibers. (A) Fluorescent microscopy images of TMRM of muscle fibers isolated from extensor digitorum longus (EDL; top) and tibialis anterior (TA; bottom) muscles. Intact, unblemished myofibers appear as translucent cylinders with striated patterns (as shown in the brightfield images). White scale bars: 200 μm. (B) Mitochondrial membrane potential (TMRM fluorescence intensity) was quantified and analyzed in the selected individual myofibers. Values are normalized to basal intensity levels. Data represent the mean ± SEM of 3 biological replicates.

Figure 4

High-content imaging of ΔΨm in cell co-cultures and machine learning quantification. (A) Representative images of melanoma A375 cells (cultured either alone or co-cultured) and macrophages treated with TMRM. White scale bars: 200 μm. (B,C) the respective TMRM analysis in basal conditions or exposed to oligomycin and FCCP. Data represent the mean ± SEM from 4 biological replicates. (D) Cell cycle phase distribution of melanoma A375 cells cultured either alone or in co-culture with macrophages. Dashed lines indicate the division of the three cell cycle phases: G0/G1, S, and G2/M. Representative graph reports the percentage of the frequency distribution of nuclei intensity in each cell cycle phase. Data represent the mean ± SEM from 4 biological replicates.