Neuronal cell-based high-throughput screen for enhancers of mitochondrial function reveals luteolin as a modulator of mitochondria-endoplasmic reticulum coupling

Abstract

Background: Mitochondrial dysfunction is a common feature of aging, neurodegeneration, and metabolic diseases. Hence, mitotherapeutics may be valuable disease modifiers for a large number of conditions. In this study, we have set up a large-scale screening platform for mitochondrial-based modulators with promising therapeutic potential.

Results: Using differentiated human neuroblastoma cells, we screened 1200 FDA-approved compounds and identified 61 molecules that significantly increased cellular ATP without any cytotoxic effect. Following dose response curve-dependent selection, we identified the flavonoid luteolin as a primary hit. Further validation in neuronal models indicated that luteolin increased mitochondrial respiration in primary neurons, despite not affecting mitochondrial mass, structure, or mitochondria-derived reactive oxygen species. However, we found that luteolin increased contacts between mitochondria and endoplasmic reticulum (ER), contributing to increased mitochondrial calcium (Ca²⁺) and Ca²⁺-dependent pyruvate dehydrogenase activity. This signaling pathway likely contributed to the observed effect of luteolin on enhanced mitochondrial complexes I and II activities. Importantly, we observed that increased mitochondrial functions were dependent on the activity of ER Ca²⁺-releasing channels inositol 1,4,5-trisphosphate receptors (IP₃Rs) both in neurons and in isolated synaptosomes. Additionally, luteolin treatment improved mitochondrial and locomotory activities in primary neurons and Caenorhabditis elegans expressing an expanded polyglutamine tract of the huntingtin protein.

Conclusion: We provide a new screening platform for drug discovery validated in vitro and ex vivo. In addition, we describe a novel mechanism through which luteolin modulates mitochondrial activity in neuronal models with potential therapeutic validity for treatment of a variety of human diseases.

Keywords: High-throughput screen; Luteolin; Mitochondria; Mitochondria-ER contacts; Mitochondrial calcium.

Fig. 1

Differentiated cells grown in pyruvate show increased mitochondrial function and higher susceptibility to mitochondrial complex III inhibition. a Schematic representation of SH-SY5Y differentiation protocol. SH-SY5Y cells were cultured for five days in high glucose media (25 mM) supplemented with 10% FBS and 10 μM RA, followed by 3 days culture in media devoid of FBS but supplemented with 25 ng/mL BDNF and either with 25 mM or 10 mM glucose or 10 mM pyruvate. b Neurite length was quantified using ImageJ in SH-SY5Y cells differentiated for 8 days (n = 28 ND, n = 32 25 G, n = 30 10 G, n = 34 10 P from 4 independent experiments). c–e SH-SY5Y cells were fixed and imaged by TEM. Number of mitochondria profile per cell and mitochondrial profile area (in μm²) are represented (n = 20 for all conditions from 4 independent experiments). Scale bar = 1 μm. f, g ΔΨ_m was measured using TMRM fluorescent probe in non-quenching conditions (5 nM) in a confocal microscope (f) and total fluorescence was quantified (g) by ImageJ (4 independent experiments). Scale bar: 15 μm. h, i Total levels of ATP were measured by luminescence in differentiated SH-SY5Y or mice primary cortical neurons (as indicated) in the presence or absence of mitochondrial complex III inhibitor Ant A (3.6 μM, 30 min) (4 independent experiments run in triplicates). j–m OCR was quantified using Seahorse flux analyzer and basal, maximal, and oligomycin-sensitive respiration were calculated after the sequential injection of oligomycin (1 μM), FCCP (1 μM), and Ant A (0.5 μM), respectively (5 independent experiments run in quadruplicates). ND = non-differentiated; 25 G = 25 mM glucose; 10 G = 10 mM glucose; 10 P = 10 mM pyruvate. Statistical significance: ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, ^****p < 0.0001 using non-parametric Kruskal-Wallis test or two-way ANOVA followed by Sidak’s multiple comparison test (in H and I)

Fig. 2

HTS reveals luteolin as a single hit to increase mitochondrial bioenergetic function. a Pipeline of the HTS design and validation for compounds targeting mitochondria. b 10 P-differentiated SH-SY5Y cells were incubated with 1200 Prestwick library-selected compounds (10 μM, 24 h), which were automatically transferred to 384-well plates using an acoustic dispenser. Total ATP levels were measured, and hits were selected when ATP z-scores ≥ 2 and cytotoxicity z-scores ≤ 2 (green dots). Luteolin is marked in red. c Heatmap representing the ATP z-scores of selected hits from B in a 3-CRC (24 h incubation). Arrow heads represent compounds with highest z-scores (> 6) or dose-dependent increase in ATP z-scores that were selected to be tested in a 9-CRC in d. d Quantification of ATP levels and cytotoxicity in a 9-CRC (24 h incubation) of the selected hits from c. Green background indicates range of concentrations with observed increases in ATP levels, while red background indicates increased cytotoxicity

Fig. 3

Luteolin boosts mitochondrial respiration and ATP production without affecting mitochondrial mass or structure. a–d Cortical neurons were treated with luteolin (Lut) for 16 h at the indicated concentrations. OCR was quantified using the Seahorse analyzer. Basal respiration, maximal respiration, and oligomycin-sensitive respiration coupled to ATP synthesis were calculated after the sequential injection of oligomycin (1 μM), FCCP (1 μM), and Ant A (0.5 μM) plus Rot (0.5 μM) (n = 6–7 run in triplicate; each point in the graph represents one independent experiment). e Cortical neurons were treated with luteolin (2.5 μM, 16 h), and total ATP levels were quantified by luminescence (n = 6 run in quadruplicate; each point in the graph represents one independent experiment). f, g Neurons were treated with 2.5 μM luteolin for 16 h. Mitochondrial H₂O₂ levels were quantified using MitoPY1 (10 μM, 25 min) in an epifluorescence cell observer microscope. After 10 min of basal reading, mitochondria were challenged with Ant A (2 μM). Representative images show MitoPY1 fluorescence before and after (insets) Ant A addition (scale bar = 25 μm; 10 μm for insert). Two-way ANOVA revealed an effect of Ant A F(1,137) = 65.85, p < 0.0001 (n = 37 −Lut −AntA, n = 38 +Lut −AntA, n = 33 –Lut +AntA, n = 33 +Lut +AntA, from 3 independent experiments). h Representative images of TEM in primary cortical neurons. Insets show an amplified image of mitochondrial cristae that is highlighted in green (scale bar = 500 nm; 150 nm for insert). i–k Mitochondria profile area (in nm²), aspect ratio, and cristae number per mitochondrial perimeter were quantified by TEM (in i and j, n = 23 Ctr, n = 22 Lut 2.5 μM from 3 independent experiments; in k, n = 19 from 3 independent experiments). Each n represents one cell. Statistical significance: ^*p < 0.05, ^**p < 0.01 using non-parametric Kruskal-Wallis or Mann-Whitney test (in b–e), ^****p < 0.0001 using 2-way ANOVA followed by Tukey’s multiple comparisons test (in g); ns = non-significant

Fig. 4

Mitochondria-ER contacts are increased by luteolin regulating mitochondrial respiration in neuronal models. a Representative TEM images from primary neurons evidence mitochondria (in yellow) in close contact with ER (in purple) (scale bar = 80 nm) (upper panels). Representative confocal images of differentiated 10 P SH-SY5Y cells transfected with SPLICS_S (in green), and f-actin labeled with phalloidin (gray) (middle and lower panels) (scale bar = 15 μm). b, c Number of MERCS per mitochondria profile and MERCS length were quantified by TEM (n = 31 Ctr, n = 32 Lut 2.5 μM from 4 independent experiments). d Differentiated 10 P SH-SY5Y cells were transfected with SPLICS_S and treated with luteolin for 16 h, and the number of green dots quantified. Each dot indicates a contact between mitochondria and ER (n = 15 Ctr, n = 18 Lut 2.5 μM from 4 independent experiments). e, f Schematic representation of proteins in MERCS (in e). MERCS proteins were quantified by western blotting using specific antibodies. Proteins were extracted from neurons treated for 16 h with DMSO or luteolin (in f) (n = 4–5; each point in the graph represents one independent experiment). OMM: outer mitochondrial membrane; IMM: inner mitochondrial membrane. g–i Cortical neurons (in g, h) or differentiated 10 P SH-SY5Y cells (in i) were treated with DMSO or luteolin (2.5 μM) for 24 h and incubated, when indicated, with XeC (3 μM) for 30 min. OCR was measured using the Seahorse flux analyzer. Spider chart lines represent the fold increase in OCR considering basal respiration of the control cells equal to 1 (n = 6–8 run in quadruplicate; each point in the graph represents one independent experiment). SRC: spare respiratory capacity. Statistical significance: ^*p < 0.05, ^****p < 0.0001 using non-parametric Mann-Whitney test (in b, d); ^*p < 0.05, ^**p < 0.01, ^***p < 0.001 using non-parametric Kruskal-Wallis (in h, i)

Fig. 5

Luteolin increases mitochondrial Ca²⁺ levels, Ca²⁺ transfer from ER, and bioenergetics in synaptosomes in an IP₃R-dependent manner. a Representative traces of mitochondrial (mt-Ca²⁺, upper panel) and cytosolic (cyt-Ca²⁺, lower panel) Ca²⁺ levels in cortical neurons. The arrow indicates the addition of a mix of IP₃-generating agonists (100 μM ATP, 300 μM carbachol, and 100 μM glutamate) in a Ca²⁺-free, EGTA-containing solution. b–d Basal mt-Ca²⁺ levels, mean peak area of mt-Ca²⁺ and cyt-Ca²⁺ were quantified in cortical neurons treated with DMSO or luteolin (2.5 μM, 16 h) (in b, n = 121 Ctr, n = 147 Lut 2.5 μM; in c, n = 33 Ctr, n = 28 Lut 2.5 μM; in d, n = 40 Ctr, n = 43 Lut 2.5 μM from 3 independent cultures). e, f Schematic representation of PDH activation by calcium (in e). In the presence of calcium, PDH phosphatases (PDPs) are activated, favoring PDH dephosphorylation and activation. Phosphorylated and total levels of PDH E1α were quantified in neurons treated with DMSO or luteolin (2.5 μM, 16 h) by western blotting using specific antibodies (n = 5–6; each point in the graph represents one independent experiment). g NADH levels were quantified in mitochondrial-enriched extracts obtained from cortical neurons treated with DMSO or luteolin (2.5 μM, 16 h) (n = 6 run in triplicates; each point in the graph represents one independent experiment). h, i Electron flow was measured in cortical mitochondria isolated from WT mice using a Seahorse flux analyzer after a short 30 min incubation with luteolin or DMSO. Mitochondrial complex inhibitors or substrates, 2 μM Rot, 10 mM succinate (Succ), 4 μM Ant A, and 1 mM ascorbate (Asc)/ 100 mM TMPD, were sequentially injected to calculate mitochondrial complex I–IV activities, respectively (n = 4; each point in the graph represents data from one animal’s brain). j, k Pure synaptosome preparations (represented in j) were incubated for 30 min with luteolin (2.5 μM) and/or XeC (5 μM), and total ATP levels were quantified by luminescence. Ant A (3.2 μM, 30 min) was used as a positive control, which decreased ATP levels by 56% (n = 5; each point in the graph represents data from one animal’s brain). Scale bar = 1 μm (upper image); 200 nm (inset). Statistical significance: ^*p < 0.05, ^**p < 0.01, ^***p < 0.001 by Mann-Whitney test when comparing two independent samples (in b, c, f, i). ^*p < 0.05, ^**p < 0.01, ^****p < 0.0001 by non-parametric Kruskal-Wallis test (in k)

Fig. 6

Luteolin is neuroprotective in models of HD. a, b YAC128 cortical neurons were treated with luteolin at indicated concentrations for 16 h. OCR was quantified using a Seahorse analyzer and basal respiration, maximal respiration, ATP production (oligomycin-sensitive respiration), and spare respiratory capacity (SRC) were calculated after the sequential injection of oligomycin (1 μM), FCCP (1 μM), and Ant A plus rotenone (0.5 μM) (n = 5 run in duplicate or triplicates; each point in the graph represents one independent culture). c Adult C. elegans carrying rmIs110 transgene were exposed to DMSO (as a control) or luteolin (25 μM) for 96 h and trashing assay was performed (n = 41 from 3 biological replicates). Statistical significance: ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, ^****p < 0.0001 by non-parametric Kruskal-Wallis test (in b); unpaired t test ^**p = 0.0037 (in c)