ER-mitochondria contacts mediate lipid radical transfer via RMDN3/PTPIP51 phosphorylation to reduce mitochondrial oxidative stress

Abstract

The proximal domains of mitochondria and the endoplasmic reticulum (ER) are linked by tethering factors on each membrane, allowing the efficient transport of substances, including lipids and calcium, between them. However, little is known about the regulation and function of mitochondria-ER contacts (MERCs) dynamics under mitochondrial damage. In this study, we apply NanoBiT technology to develop the MERBiT system, which enables the measurement of reversible MERCs formation in living cells. Analysis using this system suggests that induction of mitochondrial ROS increases MERCs formation via RMDN3 (also known as PTPIP51)-VAPB tethering driven by RMDN3 phosphorylation. Disruption of this tethering caused lipid radical accumulation in mitochondria, leading to cell death. The lipid radical transfer activity of the TPR domain in RMDN3, as revealed by an in vitro liposome assay, suggests that RMDN3 transfers lipid radicals from mitochondria to the ER. Our findings suggest a potential role for MERCs in cell survival strategy by facilitating the removal of mitochondrial lipid radicals under mitochondrial damage.

Fig. 1. Design of a reversible MERCs quantification assay system – MERBiT.

a Schematic of the construct and strategy for detection of MERCs. b Schematic of the MERBiT system. c Representative images of V5-TOMM20-SmBiT and LgBiT−3×HA-Sec61β localization in HeLa cells stably expressing V5-TOMM20-SmBiT and LgBiT−3×HA-Sec61β (MERBiT cells). Cells were stained with V5, HA, HSP60 and calnexin antibodies. HSP60 is used as a mitochondrial marker and calnexin is used as an ER marker. d, Representative immunoblots for each component of MERBiT cells. The lysates of MERBiT cells were analyzed by immunoblotting for V5 (V5-TOMM20-SmBiT), HA (LgBiT-3×HA-Sec61β), TOMM20, HSP60, calnexin, and α-tubulin. Black and white arrowheads indicate tagged and endogenous TOMM20, respectively. e Luminescence of MERBiT cells. Quantification of the luminescence of HeLa cells, MERBiT cells, and stably expressing V5-TOMM20-SmBiT HeLa cells. Data are mean ± s.e.m. (n = 9). f Quantification of MERCs reduction during recovery from starvation in MERBiT cells. Cells were starved in HBSS for 1 h and then recovered in 10% FBS DMEM for the indicated times before luminescence was measured. Data are mean ± s.e.m. (n = 3, triplicate). g, h Effects of knockdown of different MERCs tethering factors on MERBiT luminescence in MERBiT cells. Cells were transfected with the indicated siRNAs and then luminescence was measured or WB was performed with the indicated antibodies to confirm protein expression levels. Data are mean ± s.e.m. (n = 3, triplicate). i MERCs linker increases luminescence. MERBiT cells were transfected with MERCs linker (pCAG-AKAP1(1-30 aa)-mTagBFP-V5-SACM1L (521-587 aa)) and luminescence was detected. Data are mean ± s.e.m. (n = 3, triplicate). Statistical significance was analyzed by one-way analysis of variance (ANOVA) (e, f, g) or Student’s t-test, Two-tailed (i). P values are indicated as; **p < 0.01; ****p < 0.0001.

Fig. 2. Mitochondrial ROS increases MERCs formation.

a Mitochondrial ROS produced by rotenone and antimycin A treatment increases luminescence in MERBiT cells. The cells were treated with rotenone (50 nM) and antimycin A (50 nM) with or without Mito-TEMPO for 1 h and then luminescence was measured. b Antimycin A and rotenone stimulation increases MERCs in EM imaging. Representative electron micrographs of HeLa cells treated with or without antimycin A (50 nM) or rotenone (50 nM) for 1 h. Mitochondria and ER were labeled with green and magenta, respectively. ER-mitochondria contact sites are indicated by dashed white lines ( < 25 nm distance between membranes). c Quantification of MERCs length normalized to mitochondrial circumference. Data are mean ± s.e.m. n = 30 cells for control, antimycin A treatment, and rotenone treatment in HeLa cells. d, e oxNAC and NACS2 treatment reduces MERCs formation by removing excessive ROS in MERBiT cells. The cells were treated with rotenone (50 nM) and antimycin A (50 nM) with or without oxNAC (30 µM) and NACS2 (30 µM) for 1 h and then luminescence was measured. oxNAC and NACS2 were treated for 6 h before addition of rotenone and antimycin A. f Mitochondrial ROS production by CARS2 and SLC25A39 knockdown increases MERCs formation in MERBiT cells. The cells were transfected with indicated siRNAs for 3 days and treated with or without Mito-TEMPO for 1 h before luminescence measurement. g, h Mitochondrial, but not cytosolic, H₂O₂ generation increases MERCs formation in MERBiT cells. MERBiT cells were transfected with the cytoDAAO and mitoDAAO expressing constructs for 1 day and treated with or without D-Met (1 mM) for 1 h before luminescence measurement. i, j The MERBiT cells with higher mitochondrial ROS production form more MERCs than cells with lower mitochondrial ROS production. The cells were stained with MitoSOX 10 µM for 30 min and the MitoSOX high or low MERBiT cells were sorted by a cell sorter. Then the reseeded cells were measured for luminescence. k NAC treatment decreases MERCs formation in MERBiT cells and mitochondrial ROS high MERBiT cells. The cells or mitochondrial ROS high cells were treated with NAC for 2 days and then luminescence was measured. Data are mean ± s.e.m. (n = 3, triplicate), and statistical significance was analyzed by one-way analysis of variance (ANOVA) (a, c, e, f, h, j, k). P values are indicated as **p < 0.01; ***p < 0.001; ****p < 0.0001; n.s., not significant.

Fig. 3. MERCs induced by mitochondrial ROS are tethered by RMDN3 and VAPB.

a RMDN3 and VAPB are critical tethering factors for MERCs formation induced by antimycin A and rotenone stimulation. MERBiT cells were transfected with the indicated siRNAs for 3 days and treated with or without rotenone (50 nM) and antimycin A (50 nM) for 1 h before luminescence measurements. Data are mean ± s.e.m. (n = 3, triplicate). b Interaction between RMDN3 and VAPB increase in rotenone and antimycin A treatment. HeLa cells were transfected with the indicated vectors and treated with rotenone (50 nM) or antimycin A (50 nM) for 1 h. Cell lysates were subjected to IP assay (left). Ratio of RMDN3-VAPB interaction (right). Data are mean ± s.e.m. (n = 3). c Schematic model of the RMDN3 domain. d The expression levels of RNAi-resistant RMDN3 vectors. The HeLa cells were transfected with RMDN3 siRNA for 2 days and then transfected with the indicated vectors such as RMDN3 RNAi-resistant vectors for 1 day. e FFAT but not TPR domain is important for mitochondrial ROS-induced MERCs formation. The MERBiT cells were transfected with the indicated siRNAs for 2 days. Then transfected with empty vectors or indicated RNAi-resistant vectors for 1 day. Before measuring luminescence, cells were treated with or without rotenone (50 nM) and antimycin A (50 nM) for 1 h. Data are mean ± s.e.m. (n = 3, triplicate). f Threonine 160 mutant of RMDN3 decrease phosphorylation by antimycin A stimulation. HeLa cells were transfected with the indicated vectors and then treated with or without antimycin A (50 nM). Cell lysates were subjected to IP assay and then beads were incubated with or without lambda phosphatase (λPP). Pull-down lysates were subjected to Phos-tag-PAGE or SDS-PAGE. g Phosphorylation of RMDN3 T160 was important for interaction with VAPB by antimycin A treatment. HeLa cells were transfected with the indicated vectors and treated with antimycin A (50 nM) for 1 h. Cell lysates were subjected to IP assay and IB assay (left). Ratio of RMDN3-VAPB interaction (right). Data are mean ± s.e.m. (n  =  3). h Phosphorylation of RMDN3 T160 is important for MERCs formation induced by antimycin A stimulation. The MERBiT cells were transfected with the indicated siRNAs for 2 days and then transfected with vectors for 1 day before treatment with or without antimycin A (50 nM) for 1 h, and then the luminescence was measured. Data are mean ± s.e.m. (n = 3, triplicate). Statistical significance was analyzed by one-way analysis of variance (ANOVA) (a, b, e, g, h). P values are indicated as *p < 0.05; **p < 0.01; ****p < 0.0001; n.s., not significant.

Fig. 4. Disruption of mitochondrial ROS-derived MERCSs attenuates cell viability.

a Rotenone or antimycin A treatment of RMDN3 and VAPB knockdown cells reduced cell viability. The HeLa cells were transfected with the indicated siRNAs for 5 days with or without rotenone (50 nM) and antimycin A (50 nM) for 2 days before cell viability was measured. Cell viability was determined and expressed as a fold change of si-NT. Viable cells were detected by cell viability assay using Cell Counting Kit-8. b The HeLa cells were transfected with the indicated siRNAs for 5 days with or without rotenone (50 nM) and antimycin A (50 nM) for 2 days before measuring cell viability. oxNAC (50 µM), NACS2 (50 µM) and mito-TEMPO (100 nM) were treated for 3 days before measuring cell viability. Cell viability was determined and expressed as a fold change of si-NT. Viable cells were detected by cell viability assay using Cell Counting Kit-8. c Lack of TPR domain does not rescue cell viability of RMDN3 knockdown with rotenone or antimycin A treatment. HeLa cells were transfected with RMDN3 siRNA for 5 days and with the indicated vectors for 3 days before measuring cell viability. Rotenone (50 nM) and antimycin A (50 nM) were treated for 2 days before cell viability was measured. Cell viability was determined and expressed as a fold change of si-NT. Viable cells were detected by cell viability assay using Cell Counting Kit-8. FLAG-resi-RMDN3 WT, FLAG-resi-RMDN3ΔFFAT, and FLAG-resi-RMDN3ΔTPR are RMDN3 RNAi-resistant vectors. d HeLa cells were transfected with RMDN3 siRNA for 5 days with or without rotenone (50 nM) and antimycin A (50 nM) for 2 days before measuring cell viability. The indicated inhibitors were treated for 2 days before measuring cell viability. Cell viability was determined and expressed as a fold change of si-NT. Viable cells were detected by cell viability assay using Cell Counting Kit-8. z-VAD-FMK (20 µM), necrostatin-1 (20 µM), ferrostatin-1 (Fer-1) (5 µM), and deferoxamine (DFO) (100 µM). Data are mean ± s.e.m. (n = 3, triplicate), and statistical significance was analyzed by one-way analysis of variance (ANOVA) (a-d). P values are indicated as ****p < 0.0001; n.s., not significant.

Fig. 5. RMDN3 removes lipid radicals from mitochondria via MERCs.

a Suppression of RMDN3 and antimycin A treatment leads to accumulation of lipid radicals on mitochondria. Representative LipiRADICAL Green staining images in HeLa cells transfected with RMDN3 siRNA for 3.5 days before staining with LipiRADICAL Green. Cells were treated with antimycin A (50 nM) for 3 h before staining with LipiRADICAL Green and also labeled with MitoTracker as mitochondrial marker. b Quantification of colocalization between LipiRADICAL Green signal and MitoTracker in relation to (a). c Quantification of colocalization between LipiRADICAL Green signal and ER-RFP as an ER marker. Cells were treated with antimycin A (50 nM) for 3 h before staining with LipiRADICAL Green. d Representative images of LipiRADICAL Green staining images in HeLa cells transfected with RMDN3 siRNA for 3.5 days and with the indicated vectors for 2 days before staining with LipiRADICAL Green. Cells were treated with Antimycin A (50 nM) for 1 day before staining with LipiRADICAL Green and MitoTracker. FLAG-resi-RMDN3WT-IRES-EBFP and FLAG-resi-RMDN3ΔTPR-mCherry are RMDN3 RNAi resistant vectors. Blue line indicates cells expressing FLAG-resi-RMDN3WT-IRES-EBFP. Red line indicates cells expressing FLAG-resi-RMDN3ΔTPR-IRES-mCherry. e Quantification of LipiRADICAL Green staining patterns from (d) following patterns in Supplementary Fig. 7a. EBFP (FLAG-resi-RMDN3WT-IRES-EBFP) and mCherry (FLAG-resi-RMDN3ΔTPR-IRES-mCherry) negative cell were counted as non-expressing cells (n = 106 cells). Resi-RMDN3WT cells were counted for EBFP-positive cell staining patterns (n = 92 cells). Resi-RMDN3ΔTPR cells were counted for mCherry-positive cell staining patterns (n = 74 cells). f Representative images of LipiRADICAL Green staining images in HeLa cells transfected with RMDN3-siRNA for 3.5 days and with the indicated vectors for 2 days before staining with LipiRADICAL Green. Cells were treated with antimycin A (50 nM) for 1 day before staining with LipiRADICAL Green and MitoTracker. FLAG-resi-RMDN3WT-IRES-EBFP and FLAG-resi-RMDN3ΔFFAT-mCherry are RMDN3 RNAi-resistant vectors. Blue line indicates cells expressing FLAG-resi-RMDN3WT-IRES-EBFP. Red line indicates cells expressing FLAG-resi-RMDN3ΔFFAT-IRES-mCherry. g Quantification of LipiRADICAL Green staining patterns from (f) following patterns in (Supplementary Fig. 7a). EBFP (FLAG-resi-RMDN3WT-IRES-EBFP) and mCherry (FLAG-resi-RMDN3ΔTPR-IRES-mCherry) negative cell were counted as non-expressing cells (n = 87 cells). Resi-RMDN3WT cells were counted as EBFP-positive cell staining patterns (n = 104 cells). Resi-RMDN3ΔFFAT cells were counted as mCherry-positive cell staining patterns (n = 106 cells).

Fig. 6. RMDN3 TPR domain binds and transfers lipid radicals.

a Schematic of the liposome assay to monitor lipid radical transfer activity. b RMDN3 shows higher lipid transfer activity in the presence of lipid radicals. The y‐axes in the graphs show the percentage of acceptor liposome fluorescence, which was calculated using the following formula: 100×Facceptor /(Facceptor+Fdonor). c Liposome assay to measure lipid radicals transfer by RMDN3 TPR. Donor liposome (containing biotinyl-cap lipids and unsaturated lipids), which bind to streptavidin magnetic beads, were treated with or without AAPH (500 µM), a radical inducer. After washed out AAPH, acceptor liposomes and RMDN3 TPR were mixed with donor liposome-bead lysates. LipiRADICAL Green was then added to the mixture. After removing donor liposome-bead by binding to magnetic racks, the supernatant (including acceptor liposomes and RMDN3 TPR mixture) LipiRADICAL Green fluorescence signals were measured. d The RMDN3 TPR binds and transfers lipid radicals. The donor liposome (containing biotinyl-cap lipids and unsaturated lipids), which bind streptavidin magnetic beads, lysates were treated with or without AAPH (500 µM), a radical inducer. After washed out AAPH, acceptor liposomes and RMDN3 TPR or RMDN3 TPR alone were mixed with donor liposome-bead lysates. Then, LipiRADICAL Green was added to the mixture. After removing donor liposome-bead by binding to magnetic racks, the supernatant (including acceptor liposomes and RMDN3 TPR or RMDN3 TPR alone) LipiRADICAL Green fluorescence signals were measured. e Schematic illustration: induction of mitochondrial ROS promotes MERC formation, and RMDN3 transfers lipid radicals from mitochondria to the ER at MERCs to attenuate mitochondrial oxidative stress. Data are mean ± s.e.m. (n = 3, triplicate), and statistical significance was analyzed by one-way analysis of variance (ANOVA) (b–d). P values are indicated as ****p < 0.0001.

Fig. 7. RMDN3 attenuates lipid peroxide accumulation in brown adipocytes under cold-mimetic stimuli.

a, b Suppression of RMDN3 does not affect the induction of thermogenic genes. Cells were transfected with the indicated siRNAs and harvested at 0 or 4 days after differentiation. The cell lysates were analyzed by immunoblotting with the indicated antibodies (a). mRNA levels of differentiation markers were measured by qRT-PCR. Data were normalized to s18 mRNA and expressed relative to si-NT on day 4 (b). c Representative images of lipid droplets (LDs) in cells treated with the indicated siRNAs. Cells were fixed on day 4. The LDs and mitochondria were labeled with LipidTOX and anti-TOMM20 antibodies. The LDs and mitochondria were quantified for total LD area in (d), average LD size in (e), total mitochondrial area in (f), and the ratio of total LD area to total mitochondrial area in (g) from the ROI of (c). 2−3 cells from three independent experiments for the control and si-RMDN3#1 cells, respectively. Data are mean ± s.e.m. (n = 3) h Lipid peroxide production increases with NE stimulation. MitoPeDPP (10 µM) was stained 30 min after stimulation with or without NE (1 µM) for 1 h and MitoPeDPP signals were detected in brown adipocytes (day 6). i, j RMDN3 and VAPB binding is increased by mitochondrial ROS generated under NE stimulation. Cell lysates were subjected to IP assay with anti-RMDN3 antibody and IB assay with the indicated antibodies (j). Ratio of RMDN3-VAPB interaction, plotted data for NE (1 µM) with or without Mito-TEMPO (10 µM) treatment versus control. Data are mean ± s.e.m. (n means three independent experiments). k Phosphorylation of RMDN3 by mitochondrial ROS and binding with VAPB is necessary for the suppression of lipid peroxide production. Cells were transfected with the indicated siRNAs and expressed human RMDN3 or human RMDN3 T160A before measuring MitoPeDPP fluorescence. NE was treated for 1 h. Data are mean ± s.e.m. (n means three independent experiments). Statistical significance was analyzed by one-way analysis of variance (ANOVA) (b, j, k) or Student’s t-test, Two-tailed (d–h). P values are indicated as *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; n.s., not significant.