Intact mitochondrial Ca²⁺ uniport is essential for agonist-induced activation of endothelial nitric oxide synthase (eNOS)

Abstract

Mitochondrial Ca²⁺ uptake regulates diverse endothelial cell functions and has also been related to nitric oxide (NO^•) production. However, it is not entirely clear if the organelles support or counteract NO^• biosynthesis by taking up Ca²⁺. The objective of this study was to verify whether or not mitochondrial Ca²⁺ uptake influences Ca²⁺-triggered NO^• generation by endothelial NO^• synthase (eNOS) in an immortalized endothelial cell line (EA.hy926), respective primary human umbilical vein endothelial cells (HUVECs) and eNOS-RFP (red fluorescent protein) expressing human embryonic kidney (HEK293) cells. We used novel genetically encoded fluorescent NO^• probes, the geNOps, and Ca²⁺ sensors to monitor single cell NO^• and Ca²⁺ dynamics upon cell treatment with ATP, an inositol 1,4,5-trisphosphate (IP₃)-generating agonist. Mitochondrial Ca²⁺ uptake was specifically manipulated by siRNA-mediated knock-down of recently identified key components of the mitochondrial Ca²⁺ uniporter machinery. In endothelial cells and the eNOS-RFP expressing HEK293 cells we show that reduced mitochondrial Ca²⁺ uptake upon the knock-down of the mitochondrial calcium uniporter (MCU) protein and the essential MCU regulator (EMRE) yield considerable attenuation of the Ca²⁺-triggered NO^• increase independently of global cytosolic Ca²⁺ signals. The knock-down of mitochondrial calcium uptake 1 (MICU1), a gatekeeper of the MCU, increased both mitochondrial Ca²⁺ sequestration and Ca²⁺-induced NO^• signals. The positive correlation between mitochondrial Ca²⁺ elevation and NO^• production was independent of eNOS phosphorylation at serine¹¹⁷⁷. Our findings emphasize that manipulating mitochondrial Ca²⁺ uptake may represent a novel strategy to control eNOS-mediated NO^• production.

1. Introduction

The transfer of Ca²⁺ into mitochondria influences cell signaling and functions by various means. The organelles themselves immediately respond to Ca²⁺ elevations as they house Ca²⁺-sensitive enzymes and transporters [1]. Ca²⁺ within mitochondria is known to elevate their metabolic activity, but can also trigger cell death [2,3]. In addition, mitochondrial Ca²⁺ uptake and release shapes local and global cellular Ca²⁺ signals, which control a variety of cell signaling events [4]. Particularly, in vascular endothelial cells mitochondrial Ca²⁺ signals have been debated to directly or indirectly control key functions such as the synthesis and release of vasoactive compounds [5,6]. However, the exact role of mitochondrial Ca²⁺ uptake for endothelial nitric oxide (NO^•) production, which is accomplished predominately by the Ca²⁺ calmodulin-dependent endothelial nitric oxide synthase (eNOS) [7], is unknown. While mitochondria-derived NO^• has been detected in isolated mitochondria and permeabilized endothelial cells [5,8], the impact of mitochondrial Ca²⁺ uptake on Ca²⁺-evoked NO^• production in intact cells has not been investigated so far.

In recent years the key components of mitochondrial Ca²⁺ channels have been identified [9,10]. These studies unveiled that the mitochondrial Ca²⁺ uniporter basically consists of the pore forming mitochondrial calcium uniporter (MCU) [9], the essential MCU regulator (EMRE) [11], and the mitochondrial calcium uptake 1 (MICU1) [10]. Interestingly, MICU1 shields the MCU/EMRE-dependent mitochondrial Ca²⁺ channel and thereby impedes mitochondrial Ca²⁺ uptake under basal conditions [11]. A very recent study unveiled that MICU1 methylation by protein arginine methyltransferase-1 (PRMT1) hampers MCU activation by Ca²⁺, indicating that mitochondrial Ca²⁺ uptake is tightly controlled by posttranslational modifications [12]. However, the knock-down of these key components of the mitochondrial Ca²⁺ uniporter allows to efficiently manipulate mitochondrial Ca²⁺ uptake [13]. Based on these findings the role of mitochondrial Ca²⁺ uptake for endothelial NO^• production in intact cells has not been reinvestigated so far. Here we used novel genetically encoded fluorescent NO^• probes, the geNOps, to monitor NO^• signals on the level of individual cells. Recently, geNOps were used to specifically visualize the biotransformation of nitroglycerine to NO^• by aldehyde dehydrogenase-2 (ALDH2) in single vascular smooth muscle cells [14]. Herein we combined the geNOp technology with Ca²⁺ sensors and siRNA-mediated knock-down of mitochondrial Ca²⁺ channel proteins [15]. This approach allowed us to demonstrate that Ca²⁺-triggered NO^• production by eNOS is positively controlled by mitochondrial Ca²⁺ uniport independently of global cytosolic Ca²⁺ signals. Hence, this study unveiled a clear link between mitochondrial Ca²⁺ uptake and eNOS-mediated NO^• generation which might have several implications in health and diseases.

2. Materials and methods

2.1. Chemicals

Dulbecco's modified Eagle's medium (DMEM), antimycin, oligomycin, and L-N^G-nitro arginine were purchased from Sigma-Aldrich (Vienna, Austria). Fura-2-acetoxymethyl ester (fura-2/am), tetramethylrhodamine methyl ester perchlorate (TMRM) and cell culture supplements were obtained from Invitrogen (San Diego, USA). TransFast™ transfection reagent was obtained from Promega (Mannheim, Germany). Antibodies against total and phosphorylated eNOS (pS1177) were from BD Transduction Laboratories™, alpha-tubulin was from Cell Signaling Technology^®. Adenosine-5´-triphosphate (ATP) and all other chemicals were purchased from Roth (Karlsruhe, Germany) unless otherwise indicated.

2.2. Cell culture, transfection and adenoviral transduction

The human umbilical vein endothelial cell line EA.hy92 (passage 35–45) [16] was cultured in DMEM medium containing 1% HAT (5 mM hypoxanthine, 20 μM aminopterin, 0.8 mM thymidine), 10% fetal bovine serum, 100 U/ml penicillin, and 100 μg/ml streptomycin in a humidified incubator (37 °C, 5% CO₂). Human embryonic kidney cells HEK293 were grown in a culture medium without HAT. Prior to transfection or adenoviral transduction cells were plated on 30 mm glass cover slips. At 50–60% confluence, cells were co-transfected with 100 μM of the respective siRNA(s) and 1.5 μg of a plasmid coding either for the NO^•-sensitive probe G-geNOp, its mutated NO^•-insensitive negative control (G-geNOp^mut) (NGFI Next Generation Fluorescence Imaging GmbH, Graz, Austria), the mitochondrial cameleon 4mtD3cpv or the endothelial nitric oxide synthase C-terminally fused to a red fluorescent protein (eNOS-RFP) as described previously [13]. All siRNAs were purchased from Microsynth (Balgach, Switzerland) and their (5´–3´) sequences were: GCAGCUCAAGAAGCACUUCAA (hMICU1-si1), GCAAUGGCGAACUGAGCAAUA (hMICU1-si2), AGAAGUCUGUGAUGAUAAA (hMICU1-si3), GCCAGAGACAGACAAUACU (hMCU-si1), GGAAAGGGAGCUUAUUGAA (hMCU-si2), GAACUUUGCUGCUCUACUU (hEMRE-si) and a scrambled negative control siRNA UUCUCCGAACGUGUCACGU (Scrambled siRNA). Human umbilical vein endothelial cells (HUVECs) (CC-2519; Lonza, Basel, Switzerland) were grown at 37 °C and 5% CO₂ on tissue culture dishes coated with 1% Gelatin (#49391; Sigma) in endothelial basal medium (EBM) culture medium (CC-3121; Lonza) supplemented with endothelial growth medium (EGM) SingleQuots and growth factors (CC-4143; Lonza). HUVECs were cultured maximally 14 days before experiments. For imaging experiments HUVECs were seeded on 1% Gelatin coated 30 mm glass cover slips. At a confluence of 50–60% siRNA transfections and infection of BacMam 4mtD3cpv virus (Life Technologies, Vienna, Austria) encoding the mitochondrial cameleon was performed as recently described [12]. The G-geNOp expression in endothelial cells was achieved by adenoviral transduction at a multiplicity of infection (MOI) of 10 for HUVECs or MOI 1000 for EA.hy926 cells. All experiments were performed 42–54 h after transfection or infection, respectively.

2.3. Buffer solutions for imaging experiments

For imaging of cytosolic Ca²⁺ or mitochondrial membrane potential, cells were loaded at room temperature with either 3.3 μM fura2/am or 100 nM TMRM, respectively supplemented in a storage buffer composed of 138 mM NaCl, 2 mM CaCl₂, 5 mM KCl, 1 mM MgCl₂, 1 mM HEPES, 2.6 mM NaHCO₃, 0.44 mM KH₂PO₄, 0.34 mM Na₂HPO₄, 10 mM d-glucose, 0.1% vitamins, 0.2% essential amino acids, and 1% penicillin/streptomycin, pH 7.4 for 40 min. Prior to NO^• measurements G-geNOp-expressing EA.hy926 cells were incubated at room temperature in a ferrous fumarate/ascorbic acid buffer for 20 min and maintained in storage buffer for at least 1 h. For dual recordings of NO^• and cytosolic Ca²⁺, Fura2/am loading was done after ferrous fumarate incubation as recently described [15]. All imaging experiments were performed by perfusing cells in a HEPES-buffered solution (HBS) containing 138 mM NaCl, 5 mM KCl, 2 mM CaCl₂, 1 MgCl₂, 10 mM D-glucose, 10 mM HEPES, pH 7.4. For Ca²⁺-free conditions HBS containing 1 mM EGTA instead of CaCl₂ was used.

2.4. Fluorescence microscopy

For fluorescent recordings an advanced wide-field fluorescent microscope (Till Photonics, Graefling, Germany) equipped with a motorized sample stage, a polychrome V (Till Photonics), a 40x objective (alpha Plan Fluar 40x, Zeiss, Göttingen, Germany) and a charge-coupled device camera (AVT Stingray F145B, Allied Vision Technologies, Stadtroda, Germany) was used. The FRET-based mitochondrial Ca²⁺ sensor 4mtD3cpV was excited at 430 nm and emission was collected using the dichrotome dual emission filter set (dichroic 535dcxr, CFP emitter 482/18 nm and YFP emitter 535/3 nm). TMRM was visualized at an excitation of 550 nm and emission of 575 nm. For monitoring NO^• dynamics G-geNOp expressing cells were exposed to 480 nm and emitted at 515 nm. For simultaneous measurements of cytosolic Ca²⁺and NO^•, fura-2/am loaded and G-geNOp expressing cells were alternately excited at 340, 380 and 480 nm, respectively and emission was recorded at 515 nm (515dcxr). Data acquisition and control was carried out by the Live Acquisition 2.0.0.12 software (Till Photonics). Confocal imaging of adenoviral infected G-geNOp expressing EA.hy926 or transient transfected eNOS-RFP expressing HEK293 cells was performed as described previously by using a 40× objective (Plan-Neofluar 40×/1.3 oil, Zeiss) or a 100× objective (Plan-Fluar 100×/1.45 oil, Zeiss), respectively [17].

2.5. Immunoblotting

Transfected EA.hy926 cells (siControl, siMCU/EMRE or siMICU1) were incubated with vehicle (water) or 100 μM ATP at 37 °C for 3 or 5 min. Cells were then harvested and homogenized by sonication (3×5 s) in ice-cold RIPA lysis buffer (Sigma, Vienna, Austria) containing 2 mM EDTA, protease and phosphatase inhibitors (CompleteTM, PhosSTOPTM, Roche, Vienna, Austria). Protein concentration was determined with the PierceTM BCA Protein Assay Kit using bovine serum albumin as standard (Fisher Scientific Austria GmbH, Vienna, Austria). Denatured samples (30 μg) were separated by SDS-PAGE on 10% gels and transferred electrophoretically to nitrocellulose membranes. After blocking with 5% non- fat dry milk in Tris-buffered saline containing 0.1% (v/v) TWEEN-20 for 1 h, membranes were incubated overnight at 4 °C with a primary antibody against eNOS (1:2000; BD Transduction Laboratories), Ser¹¹⁷⁷ phospho-eNOS (1:1000; BD Transduction Laboratories) or β-actin (1:200,000; Sigma). Thereafter, membranes were washed 3 times and incubated for 1 h with a horseradish peroxidase-conjugated anti-mouse IgG secondary antibody (1:5000). Immunoreactive bands were visualized by chemiluminescence using ECL detection reagent (Biozym, Germany) and quantified densitometrically using the Fusion SL system (Peqlab, Erlangen, Germany).

2.6. Statistical analyses

The acquired data were analyzed by the GraphPad Prism software version 5.04 (GraphPad Software, San Diego, CA, USA). Data are presented as mean ± standard error of mean (SEM) of independent experiments (N) throughout the whole manuscript. For comparisons between two groups, two-tailed Student t-test was used for evaluation of statistical significance and a p value between 0.01 and 0.05 (pStudent´s t-test) was considered significant and indicated with “*”, p between 0.001 and 0.01 as very significant with “**” and p < 0.001 as highly significant with “***”. For comparisons across multiple groups, one-way ANOVA with Barlett's test for equal variances and Tukey's Multiple Comparision test were used for evaluating statistical significance expressed as described above. Data shown are either average or representative curves of at least three independent experiments, including analyses from imaging and western blot experiments.

3. Results

3.1. Real-time monitoring of NO^• production in single endothelial cells

To monitor single cell NO^• dynamics in EA.hy926 cells we used the novel geNOp technology [15]. The geNOps are protein-based fluorescent NO^• reporters that immediately and specifically respond to physiological cellular NO^• elevations by fluorescence quenching in a reversible manner (Fig. 1A). Adenoviral transduction resulted in very high numbers of geNOp-positive EA.hy926 cells (Fig. 1B). To elucidate NO^• production in response to Ca²⁺ release from the endoplasmic reticulum (ER), endothelial cells expressing G-geNOp (green-geNOp) were first stimulated with the inositol 1,4,5-trisphosphate (IP₃)-generating agonist ATP in the absence of extracellular Ca²⁺. Under these conditions ATP evoked a transient NO^• signal in endothelial cells (Fig. 1C-D). Cytosolic NO^• levels recovered rapidly upon the subsequent addition of Ca²⁺ to ATP stimulated cells and remained elevated in the presence of the IP₃-generating agonist (Fig. 1C-D). These findings demonstrate the importance of Ca²⁺ entry for sustained eNOS activation in EA.hy926 cells. However, within a few minutes elevated NO^• signals gradually decreased back to basal levels upon the washout of ATP (Fig. 1C-D).

Fig. 1. Live-cell imaging of Ca²⁺-triggered NO^• formation in endothelial cells expressing G-geNOp.

(A) Schematic illustration of G-geNOp consisting of the enhanced green fluorescent protein (EGFP) fused to the bacteria-derived NO^• binding domain GAF (grey). Binding of NO^• in the vicinity of EGFP in G-geNOp induces fluorescence quenching. (B) Representative image showing EA.hy926 cells expressing G-geNOp. Notably, the NO^• probe localizes to the nucleus and cytosol. (C) Average curve ( ± SEM) showing NO^• signal over time obtained from single EA.hy926 cells obtained from independent experiments (N =40) upon treatment with 100 μM ATP in the absence of Ca²⁺(1 mM EGTA) and upon Ca²⁺(2 mM) addition. (D) Representative simultaneous recordings of NO^• (solid grey line) and Ca²⁺(dashed black line) signals over time in a single fura-2/am loaded EA.hy926 cell expressing G-geNOp. As indicated the cell was treated with 100 μM ATP in the absence of Ca²⁺(1 mM EGTA) and upon Ca²⁺ addition (2 mM) for 14 min.

The geNOp signals were abolished in the presence of L-N^G-nitro arginine, a potent eNOS inhibitor [18], which did not affect respective cytosolic Ca²⁺ elevations (Supplementary data – Fig. 1). To further confirm the specificity of the G-geNOp signals, endothelial cells expressing the NO^• insensitive mutated version of G-geNOp, referred to as G-geNOp^mut, were investigated. Cell treatment with the IP₃-generating agonist ATP did not evoke significant fluorescence signals in cells expressing G-geNOp^mut (Supplementary data – Fig. 2), demonstrating that geNOps do not report intracellular pH changes under these experimental conditions [15]. As the NO^• probe did not respond to NO^• related species such as nitrite, nitrate [14], H₂S (Supplementary data – Fig. 3) and other NO^• related compounds [15], these experiments confirm that geNOp-expressing EA.hy926 cells represent a suitable model to specifically study Ca²⁺-triggered eNOS-dependent generation of NO^• on the level of individual cells. The simultaneous measurement of NO^• and Ca²⁺ in the same single cell unveiled that cellular NO^• dynamics strictly follow respective Ca²⁺ signals (Fig. 1D), indicating that NO^• production in EA.hy926 cells is primarily controlled by the free cytosolic Ca²⁺ concentration ([Ca²⁺]_cyto).

3.2. Genetic ablation of mitochondrial Ca²⁺ uptake reduces eNOS-mediated NO^• formation

Next we set out to investigate how mitochondrial Ca²⁺ uptake affects NO^• production in endothelial cells. For this purpose EA.hy926 cells were treated with siRNA against both MCU and EMRE, two recently identified key components of the mitochondrial Ca²⁺ uniporter pore [9,19]. In line with other reports [13,20–22], the siRNA-mediated knock-down of these proteins substantially reduced mitochondrial Ca²⁺ signals in response to ER Ca²⁺ release and Ca²⁺ entry, while the basal mitochondrial Ca²⁺ levels were only slightly reduced (Fig. 2A). Under the same experimental conditions global cytosolic Ca²⁺ signals were only diminished at the plateau phase upon Ca²⁺ addition (Fig. 2B), confirming that mitochondrial Ca²⁺ uptake facilitates store operated Ca²⁺ entry (SOCE) in endothelial cells [20,23,24]. However, cells ablated from MCU and EMRE showed considerably reduced NO^• production upon Ca²⁺ mobilization in the absence and presence of Ca²⁺ entry (Fig. 2C). As expected both control cells and cells treated with siRNA against MCU and EMRE did not show significant changes of the fluorescence signal of the NO^•-insensitive G-geNOp^mut in response to ATP (Fig. 2C, right panel). Compared to respective control cells, the Ca²⁺-triggered NO^• signals were significantly reduced in MCU- and EMRE silenced EA.hy926 cells that were stimulated with either 1 μM, 5 μM or 100 μM ATP (Fig. 2D), indicating that mitochondrial Ca²⁺ uptake is relevant for eNOS activation over a broad range of physiological and supra-physiological levels of Ca²⁺ mobilization.

Fig. 2. Knock-down of EMRE and MCU reduces Ca²⁺-triggered NO^• production in EA.hy926 cells.

(A) Average curves of [Ca²⁺]_mito (left panel) in EA.hy926 cells treated with either scrambled siRNA (black curve and white bars, N=3) or siRNAs against EMRE and MCU (siEMRE/siMCU, red curve and red bars, N=3) upon stimulation with 100 μM ATP in the absence of extracellular Ca²⁺(1 mM EGTA) followed by addition of Ca²⁺(2 mM). Bar graphs showing average basal Ca²⁺ level (basal, p=0.0581) and Ca²⁺ transients in response to 100 μM ATP-stimulated ER Ca²⁺ release (ATP + EGTA, **p=0.0029) and store-operated Ca²⁺ entry (ATP + Ca²⁺, *p=0.0169). (B) Curves representing average of [Ca²⁺]_cyto in control (Scrambled siRNA, black curve, N=6) versus EMRE/MCU ablated (siEMRE/siMCU, red curve, N=7) EA.hy926 cells (left panel) stimulated with 100 μM ATP in the absence (1 mM EGTA) and addition of Ca²⁺(2 mM) as indicated. Respective statistical analyses of basal levels (basal, p=0.397), maximum cytosolic Ca²⁺ release (ATP + EGTA, p=0.6373) and maximum cytosolic Ca²⁺ entry (ATP + Ca²⁺, p=0.6988). (C) Representative curve of [NO^•]_cyto in response to ATP stimulation. Statistical analyses of average [NO^•]_cyto are shown in bar graphs (right panels). Maximal signals (amplitude) of scrambled siRNA- or siMCU/siEMRE-treated cells are defined as 100%. Cells were first stimulated with 100 μM ATP in the absence of extracellular Ca²⁺ in control cells (Scrambled siRNA, white column, ATP + EGTA, N =17) and cells reduced of EMRE and MCU (siEMRE/siMCU, red column, ATP + EGTA, N =17) and upon Ca²⁺( 2 mM) addition (right middle column pair). **p=0.0068 (ATP+EGTA) and **p=0.0042 (ATP+Ca²⁺) vs Scrambled siRNA. Right columns represent maximal fluorescence changes of G-geNOp^mut in response to 100 μM ATP in EGTA (1 mM) in control cells (white column, N =11) and cells treated with siRNA against EMRE and MCU (red column, N =10). (D) Average curves of Ca²⁺-dependent NO formation upon stimulation with 1 μM (upper left panel) 5 μM (upper middle panel) or 100 μM ATP (upper right panel) in the absence of extracellular Ca²⁺ in EA.hy926 cells treated with either negative control siRNA (Scrambled siRNA, black curves) or siRNA against EMRE and MCU (siEMRE/siMCU, red curves). Statistical analysis of maximum NO^• formation in cells stimulated with either 1 μM ATP (lower left panel) representing Scrambled siRNA (white bar, N=3) and siEMRE/siMCU (red bar, N=3); **p=0.0045, with 5 μM ATP (lower middle panel) representing Scrambled siRNA (white bar, N=3) and siEMRE/siMCU (red bar, N=3); **p=0.0087, or with 100 μM ATP (lower right panel) representing Scrambled siRNA (white bar, N=3) and siEMRE/siMCU (red bar, N=3); **p=0.0012.

In order to investigate if silencing of MCU and EMRE also impacts on Ca²⁺-triggered NO^• production in non-immortalized endothelial cells, primary human umbilical vein endothelial cells (HUVEC) were tested. HUVECs pooled from different donors showed more transient cytosolic Ca²⁺ signals in response to ATP compared to EA.hy926 cells, while the basal cytosolic Ca²⁺ levels were identical in both cell types (Fig. 3A). Respective mitochondrial Ca²⁺ elevations tended to be stronger in the primary endothelial cells (Fig. 3B). In average the ATP-evoked NO^• signals in HUVECs expressing G-geNOp (Fig. 3C) were less pronounced and occurred with rather slow kinetics compared to the faster and stronger NO^• elevations in EA.hy926 cells (Fig. 3C). Nevertheless, the knock down of MCU and EMRE, which reduced mitochondrial Ca²⁺ signals in HUVECs (Fig. 3D), also resulted in diminished ATP-evoked NO^• formation in the primary endothelial cells (Fig. 3E). The siRNA-mediated knock-down of MCU and EMRE significantly reduced both the amplitude and slope of ATP-triggered NO^• signals in HUVECs (Fig. 3E).

Fig. 3. Knock-down of EMRE/MCU reduces Ca²⁺-evoked NO^• formation in HUVECs.

(A) Cytosolic Ca²⁺ signals in EA.hy926 cells (black curve and white bars) and primary HUVECs (red curve and red bars) in response to 100 μM ATP. [Ca²⁺]_cyto was imaged using fura-2/am loaded cells. Middle bars represent basal fura-2 ratio (F₃₄₀/F₃₈₀) values of EA.hy926 cells (white bars, N =6, p=0.9708) and HUVECs (red bars, N =6). Right bars show maximal fura-2 ratio amplitudes in EA.hy926 (white bars, N =6) and HUVEC cells (red bars, N =6). **p=0.005 vs. EA.hy926 cells. (B) Mitochondrial Ca²⁺ signals in EA.hy926 cells (black curve and white bars) and primary HUVECs (red curve and red bars) in response to 100 μM ATP. [Ca²⁺]_mito was imaged using cells expressing 4mtD3cpV. Middle bars represent basal 4mtD3cpV ratio (F₅₃₅/F₄₈₀) values of EA.hy926 cells (white bars, N =3, p=0.4753) and HUVECs (red bars, N =6). Right bars show maximal 4mtD3cpV ratio amplitudes in EA.hy926 (white bars, N =3) and HUVEC cells (red bars, N =6). (C) Cytosolic NO^• signals in EA.hy926 cells (black curve and white bars) and primary HUVECs (red curve and red bars) in response to 100 μM ATP. [NO^•]_cyto was imaged using cells expressing G-geNOp. Middle bars represent maximal fluorescence changes of G-geNOp (1-F/F₀) expressed in EA.hy926 cells (white bars, N =10) and HUVECs (red bars, N =8); *p=0.0366 vs EA.hy926. Right bars show maximal slopes of G-geNOp fluorescence changes in response to 100 mM ATP in EA.hy926 (white bars, N =10) and HUVEC cells (red bars, N =8). *p=0.015 vs. EA.hy926 cells. (D) Mitochondrial Ca²⁺ signals in HUVECs under control conditions (Scrambled siRNA, black curve, white bars, N =6) and in cells treated with siRNA against EMRE and MCU (siEMRE/siMCU, red curve, red columns, N =6). Middle bars represent basal 4mtD3cpV ratio values (p=0.1502). Cells were stimulated with 100 μM ATP in the absence of extracellular Ca²⁺. Right bars show maximal 4mtD3cpV ratio changes upon cell stimulation.**p=0.0032 vs Scrambled siRNA. (E) Cytosolic NO^• signals ([NO^•]_cyto) in control HUVECs (Scrambled siRNA, black curve and white bars, N =8) and cells treated with siRNA against EMRE and MCU (siEMRE/siMCU, red curve and bars, N =8) in response to 100 μM ATP in EGTA (1 mM). Middle columns represent mean of maximal ATP-evoked G-geNOp responses; *p=0.0193. Right columns show respective maximal slopes of G-geNOp signals; *p=0.0206.

3.3. Mitochondrial depolarization reduces Ca²⁺-triggered NO^• formation in EA.hy926 cells

The impact of impaired mitochondrial Ca²⁺ uptake was studied by depolarization of the mitochondrial membrane potential with a combination of oligomycin and antimycin (Fig. 4A and B). In the presence of these compounds the Ca²⁺-triggered NO^• signal was significantly reduced in response to Ca²⁺ release and entry (Fig. 4C). The fluorescence signal of the NO^•-insensitive G-geNOp^mut remained unaffected by oligomycin and antimycin independently of the presence of Ca²⁺(Fig. 4C, right panel). Under the same experimental conditions cytosolic Ca²⁺ signals were, however, only partially reduced during Ca²⁺ addition (Fig. 4D). These findings again indicate that mitochondrial Ca²⁺ signals positively regulate eNOS activity.

Fig. 4. Cell treatment with oligomycin/antimycin reduces Ca²⁺-triggered NO^• formation.

(A) Average [Ca²⁺]_mito signals ± SEM of EA.hy926 cells expressing 4mtD3cpv under control conditions (black curve N =8) and in the presence of 1 μM oligomycin and 5 μM antimycin (red curve, N =12) in response to 100 μM ATP first in the absence of extracellular Ca²⁺(1 mM EGTA) and upon addition of Ca²⁺(2 mM).**p=0.0096 (in EGTA) and *p=0.0159 (in Ca²⁺) vs. Control. (B) TMRM signals of EA.hy926 cells over time to estimate changes in Ψ_mito (mitochondrial membrane potential) under control conditions (black curve N =6) and upon addition of 1 μM oligomycin and 5 μM antimycin (red curve, N =6). *p=0.0286. (C) NO^• signals of EA.hy926 cells over time expressing G-geNOp in the absence (control, black curve, white bars, N =11) or presence of 1 μM oligomycin and 5 μM antimycin (red curve and bars, N =9). Maximal signals under control conditions in response to 100 μM ATP were defined as 100% (middle right bars). **p=0.0074 (in EGTA) and *p=0.0200 (in Ca²⁺) vs. Control. Right bars represent maximal fluorescence changes of the NO^•-insensitive G-geNOp^mut in control cells (white bars N =6) and in the presence of the mitochondria toxins (red bars, N =6). p=0.8016 (in EGTA) and p=0.1508 (in Ca²⁺). (D) Fura-2 ratio signals of EA.hy926 cells over time under control conditions (black curve, N =6) and in the presence of 1 μM oligomycin and 5 μM antimycin (red curve, N =6). *p=0.0286 (in EGTA) and *p=0.0283 (in Ca²⁺) vs. Control.

3.4. Genetic augmentation of mitochondrial Ca²⁺ uptake increases eNOS-mediated NO^• formation

We next investigated whether Ca²⁺-evoked NO^• production is gradable by increasing mitochondrial Ca²⁺ uptake. For this purpose the expression level of MICU1, a negative regulator of mitochondrial Ca²⁺ uptake [10], was silenced by siRNA in EA.hy926 cells. As expected, cells treated with siRNA against MICU1 showed increased basal mitochondrial Ca²⁺ levels and ATP-evoked Ca²⁺ uptake in the absence and presence of extracellular Ca²⁺(Fig. 5A). Respective global cytosolic Ca²⁺ elevations remained almost unaffected by the knock-down of MICU1 (Fig. 5B). However, the Ca²⁺-triggered NO^• production was significantly higher in the MICU1-depleted EA.hy926 cells upon both Ca²⁺ release and entry (Fig. 5C), emphasizing that increasing mitochondrial Ca²⁺ uptake augments NO^• production. The fluorescence signal of the NO^•-insensitive G-geNOp^mut remained unaffected by cell treatment with the IP₃-generating agonist independently of MICU1 expression (Fig. 5C, right panel). In primary HUVECs siRNA-mediated silencing of MICU1 also increased mitochondrial Ca²⁺ signals in response to ATP (Fig. 5D). However, the ATP-induced NO^• formation (Fig. 5D) was only slightly increased in HUVECs that had been treated with siRNA against MICU1 compared to respective control cells (Fig. 5E).

Fig. 5. Silencing of MICU1 increases Ca²⁺-triggered NO^• production in endothelial cells.

(A) Average curves of [Ca²⁺]_mito (left panel) in EA.hy926 cells treated with either scrambled siRNA (black curve and white bars, N=3) or siRNAs against MICU1 (siMICU1, red curve and red bars, N=3) upon stimulation with 100 μM ATP in the absence of extracellular Ca²⁺(1 mM EGTA) followed by addition of Ca²⁺(2 mM). Bar graphs showing average basal Ca²⁺ level (basal, *p=0.0356) and Ca²⁺ transients in response to 100 μM ATP in 1 mM EGTA (ATP + EGTA, **p=0.0098) and upon Ca²⁺ addition (ATP + Ca²⁺, *p=0.0151). (B) Curves representing average of [Ca²⁺]_cyto in control (Scrambled siRNA, black curve, N=6) versus MICU1 ablated (siMICU1, red curve, N=6) EA.hy926 cells (left panel) stimulated with 100 μM ATP in the absence (1 mM EGTA) and addition of Ca²⁺(2 mM) as indicated. Respective statistical analyses of basal levels (basal, p=0.9621), maximum cytosolic Ca²⁺ release (ATP + EGTA, p=0.9911) and maximum cytosolic Ca²⁺ entry (ATP + Ca²⁺, p=0.4977). (C) Representative curve of [NO^•]_cyto in response to ATP stimulation. Statistical analyses of average [NO^•]_cyto are shown in bar graphs (right panels). Maximal signals (amplitude) of scrambled siRNA- or siMICU1-treated cells are defined as 100%. Cells were first stimulated with 100 μM ATP in the absence of extracellular Ca²⁺ in control cells (Scrambled siRNA, white column, ATP + EGTA, N =17) and cells reduced of MICU1 (siMICU1, red column, ATP + EGTA, N =16) and upon Ca²⁺(2 mM) addition (right middle column pair). **p=0.0083 (ATP + EGTA) and **p=0.0079 (ATP+Ca²⁺) vs Scrambled siRNA. Right columns represent maximal fluorescence changes of G-geNOp^mut in response to 100 μM ATP in EGTA (1 mM) in control cells (white column, N =11) and cells treated with siRNA against MICU1 (siMICU1, red column, N =8); p=0.4424 vs Scrambled siRNA. (D) Mitochondrial Ca²⁺ signals in HUVECs under control conditions (Scrambled siRNA, black curve, white bars, N =6) and in cells treated with siRNA against MICU1 (siMICU1, red curve, red columns, N =6). Middle bars represent basal 4mtD3cpV ratio values (p=0.1502). Cells were stimulated with 100 μM ATP in the absence of extracellular Ca²⁺. Right bars show maximal 4mtD3cpV ratio changes upon cell stimulation.*p=0.0211 vs Scrambled siRNA. (E) Cytosolic NO^• signals ([NO^•]_cyto) in control HUVECs (Scrambled siRNA, black curve and white bars, N =8) and cells treated with siRNA against MICU1 (siMICU1, red curve and bars, N =10) in response to 100 μM ATP in EGTA (1 mM). Middle columns represent mean of maximal ATP-evoked G-geNOp responses; p=0.2532. Right columns show respective maximal slopes of G-geNOp signals; p=0.4567.

3.5. The eNOS expression levels as well as basal and ATP-stimulated eNOS phosphorylation remain unaffected by silencing MCU, EMRE or MICU1

Phosphorylation of several sites within the eNOS protein significantly modulates its activity [25]. Particularly, eNOS phosphorylation at serine¹¹⁷⁷ is known to enhance its activity [25,26]. Accordingly, we performed western blot analysis using EA.hy926 cells to investigate whether or not the knock-down of MCU/EMRE and MICU1 affects eNOS phosphorylation at serine¹¹⁷⁷. As shown in Fig. 6 the knock-down of neither MCU/EMRE nor MICU1 affected eNOS phosphorylation under basal (unstimulated) conditions. Cell treatment with ATP raised eNOS phosphorylation levels within 3 and 5 min (Fig. 6), while the ATP-induced increase of phosphorylated eNOS (peNOS) was not altered by MCU/EMRE and MICU1 silencing (Fig. 6). In line with these findings the expression of eNOS in EA.hy926 cells was not affected by knock-down of MCU/EMRE or MICU1 (total eNOS, Fig. 6A). These findings again confirm that mitochondrial Ca²⁺ uptake supports NO^• production in endothelial cells without affecting the global cytosolic Ca²⁺ concentration, eNOS-expression levels and phosphorylation at serine¹¹⁷⁷.

Fig. 6. The expression level and phosphorylation status of eNOS remain unaffected by transient down-regulation of MCU, EMRE and MICU1.

(A) Representative Western blot of total homogenates of transfected EA.hy926 cells (30 μg of protein) showing phosphorylated eNOS (peNOS; 140 kDa), total eNOS (140 kDa) and β-actin (43 kDa) under basal conditions (unstimulated) or stimulated with 100 μM ATP for 3 or 5 min. Purified human eNOS (30 and 50 ng) was used as standard. Bands of control cells (Scrambled siRNA) and cells treated with siRNA against both EMRE and MCU (siEMRE/siMCU) or against MICU1 (siMICU1) are shown. (B) Bars show the ratio of phosphorylated to total eNOS protein in transfected EA.hy926 cells (Scrambled siRNA, siEMRE/siMCU and siMICU1) under basal conditions (unstimulated), after 3 (left panel) and 5 (right panel) minutes of incubation with 100 μM ATP. N =3 for all conditions.

3.6. Mitochondrial Ca²⁺ uptake significantly contributes to Ca²⁺-triggered NO^• formation in HEK293 cells transiently expressing eNOS-RFP

There is general agreement that eNOS is mainly localized at the plasma membrane [27]. However, in a previous study it has been suggested that mitochondrial Ca²⁺ uptake stimulates NO^• formation by a mitochondrial NOS isoform in bovine endothelial cells [5]. Therefore, we transiently expressed wild-type eNOS-RFP in HEK293 cells, which do not express any NOS isoforms. Confocal microscopy of eNOS-RFP expressing HEK293 cells confirmed that the NO^•-generating enzyme is located predominately at the plasma membrane and intracellular vesicular structures but not within mitochondria of cells (Fig. 7A). HEK293 cells expressing eNOS-RFP showed a considerable increase of cellular NO^• levels in response to the IP₃-generating agonist ATP (Fig. 7B). Control cells which were not transfected with eNOS-RFP were unable to produce NO^• upon the addition of the agonist (Fig. 7B). However, the response of the NO^• probe to the NO^•-donor sodium nitroprusside (SNP) was the same in both control (HEK) and eNOS-RFP expressing HEK293 cells (HEK + eNOS-RFP) (Fig. 7B). Knock-down of MCU/EMRE in HEK293 cells significantly reduced mitochondrial Ca²⁺ signals in response to ATP (Fig. 7C), while the respective cytosolic Ca²⁺ elevation remained unaffected by silencing these MCU components (Fig. 7D). As expected, knock-down of MICU1 resulted in increased mitochondrial Ca²⁺- levels and uptake without affecting respective cytosolic Ca²⁺ signals (Fig. 7C and D). In line with our results using primary endothelial cells ATP-evoked NO^•-signals were significantly reduced in cells silenced of MCU and EMRE (Fig. 7E), while MICU1 silencing only marginally increased respective NO^• signals in HEK293 cells expressing eNOS-RFP.

Fig. 7. Mitochondrial Ca²⁺ uptake controls NO^• formation in HEK293 cells expressing eNOS-RFP.

(A) Representative image showing HEK293 cells expressing eNOS-RFP. (B) Representative curves of cytosolic NO^• signals ([NO^•]_cyto) over time in response to 100 μM ATP and 1 mM SNP in control HEK293 cells (black curve and bars, N =5) and HEK293 cells co-expressing eNOS-RFP (red curve and bars, N =3). Left Bars show maximal changes of G-geNOp fluorescence in response to ATP; ***p=0.0005 vs HEK293. Right bars show maximal G-geNOps signals in response to SNP, p=0.9386. (C) Mitochondrial Ca²⁺ signals in HEK293 cells under control conditions (Scrambled siRNA, grey curve and white columns, N =8), in cells treated with siRNA against EMRE and MCU (siEMRE/siMCU, red curve and red columns, N =8) or against MICU1 (siMICU1, black curve and columns, N =7). Cells were stimulated with 100 μM ATP in EGTA (1 mM). Left columns represent basal 4mtD3cpV ratio values; p=0.4088 (siEMRE/siMCU) vs Scrambled siRNA; ***p=0.0004 (siMICU1) vs Scrambled siRNA. Right columns show mean values of maximal 4mtD3cpV ratio signals; *p=0.0145 (siEMRE/siMCU) vs Scrambled siRNA; *p=0.0152 (siMICU1) vs Scrambled siRNA. (D) Representative cytosolic Ca²⁺ signals of fura-2/am loaded HEK293 cells treated with scrambled siRNA (Scrambled siRNA, grey curve and white columns, N =3) siRNA against EMRE and MCU (siEMRE/siMCU, red curve and red columns, N =3) or MICU1 (siMICU1, black curve and columns, N =4). Cells were stimulated with 100 μM ATP in EGTA (1 mM). Columns represent mean values of maximal fura-2 ratio signals; p=0.6842 (siEMRE/siMCU) vs Scrambled siRNA; p=0.4253 (siMICU1) vs Scrambled siRNA. (E) Representative cytosolic NO^• signals eNOS-RFP expressing HEK293 cells treated with scrambled siRNA (Scrambled siRNA, grey curve and white columns, N =9) siRNA against EMRE and MCU (siEMRE/siMCU, red curve and red columns, N =11) or MICU1 (siMICU1, black curve and columns, N =8). Cells were stimulated with 100 μM ATP in EGTA (1 mM). Columns represent mean values of maximal G-geNOp signals; ***p=0.0005 (siEMRE/siMCU) vs Scrambled siRNA; p=0.0829 (siMICU1) vs Scrambled siRNA.

4. Discussion

Taking advantage of the novel genetically encoded fluorescent NO^• probes [15] and the knowledge of recently identified key components of mitochondrial Ca²⁺ channels, we could specifically investigate the impact of mitochondrial Ca²⁺ uptake on eNOS-mediated NO^• formation on the level of individual cells. Our data show a positive association of mitochondrial Ca²⁺ signals and NO^• production upon cell treatment with a physiological Ca²⁺-mobilizing agonist and, hence, emphasize that mitochondrial Ca²⁺ uptake promotes the Ca²⁺-triggered biosynthesis of this vasoactive molecule by eNOS.

Our simultaneous recordings of Ca²⁺ and NO^• signals confirm that the Ca²⁺/calmodulin-dependent activation mechanism of the eNOS [7] is under the control of the global cytosolic Ca²⁺ concentration [28,29]. However, genetic manipulations of the mitochondrial Ca²⁺ uptake capacity markedly attenuated the ATP-evoked NO^• formation without affecting global cytosolic Ca²⁺ signals. These observations strongly indicate that mitochondrial Ca²⁺ uptake contributes to the regulation of eNOS activity. Several data point to a localization of eNOS or other NOS isoforms within mitochondria or at their surface [5,30–33]. It has been suggested that mitochondrial Ca²⁺ uptake stimulates NO^• formation within mitochondria of calf pulmonary artery endothelial cells by a mitochondria-specific NOS isoform [5]. The positive association between mitochondrial Ca²⁺ uptake and NO^• generation, found in the present study, is in line with the concept of Ca²⁺-dependent mitochondria targeted NOS isoforms [5,34,35]. However, because of the lack of a clear proof, the existence of mitochondria-specific NOS isoforms in endothelial cells remains an issue of debate [36]. While we did not investigate whether or not a mitochondria located NOS is present in EA.hy926 cells and HUVECs, our data with HEK293 cells expressing eNOS-RFP indicate that mitochondrial Ca²⁺ uptake contributes to eNOS activity even if the enzyme is outside the organelle. It is, hence, feasible, that mitochondria in the vicinity of extra-mitochondrial NOS might positively control the cellular NO^• production by shaping the patterns of local Ca²⁺ signals [24] and/or contributing to the availability of essential NOS cofactors [37]. Further experiments are necessary to clarify by which molecular mechanisms mitochondrial Ca²⁺ uptake supports NO^• production by eNOS. In this study we could, however, exclude that eNOS phosphorylation at serine¹¹⁷⁷, which is known to enhance eNOS activity [25,26], is under the control of mitochondrial Ca²⁺ uptake. Cell treatment with ATP significantly increased eNOS phosphorylation at serine¹¹⁷⁷. This finding is in line with other reports that demonstrated interrelations between the Ca²⁺/calmodulin- and phosphorylation-dependent activation of eNOS [25,26]. Different cellular kinases such as the AMPK [38], PKC delta [39] or the PI3K/Akt pathway [40,41] might be responsible for the ATP-dependent phosphorylation of eNOS at serine¹¹⁷⁷ in EA.hy926 cells.

Our results strongly suggest that targeting the mitochondrial Ca²⁺ homeostasis represents a promising approach to therapeutically intervene in endothelium-governed NO^•-dependent vasoreactivity. However, the availability of pharmacological tools that specifically target the MCU machinery is very limited. We used a combination of oligomycin, an inhibitor of the ATP synthase, and antimycin, an inhibitor of complex III, in order to abolish mitochondrial Ca²⁺ uptake indirectly. This kind of cell treatment strongly affects the metabolic activity of the organelle and can lead to a severe cellular ATP depletion [42], which in turn affects the local and global cellular Ca²⁺ homeostasis [24]. While our data show that cell treatment with the mitochondria depolarizing compounds reduces eNOS-mediated NO^• production in response to the IP₃-generating agonist in EA.hy926 cells, the molecular mechanisms responsible for this effect on eNOS activity might be different from those in cells with reduced levels of MCU components. Due to the lack of suitable agents that specifically target the mitochondrial Ca²⁺ homeostasis, the impact of an acute alteration of mitochondrial Ca²⁺ fluxes on eNOS activity is difficult to investigate. It is possible that the siRNA-mediated knock-down of components of the MCU machinery introduces secondary effects on the cellular bioenergetics that may interfere with eNOS function. Such effects cannot be ruled out, although other studies have demonstrated that silencing or even the knock-out of MCU components have surprisingly little impact on the cellular metabolic homeostasis in situ [12] and in vivo [43,44]. Eventually, our data suggest that manipulating mitochondrial Ca²⁺ uptake may represent an attractive strategy for controlling the eNOS/NO^•-dependent vascular reactivity. Nevertheless, the molecular mechanism linking mitochondrial Ca²⁺ uptake to eNOS function and/or NO^• bioavailability remains to be clarified in more detail.

Supplementary Material

Appendix A. Supporting information

Supplementary data associated with this article can be found in the online version at doi:10.1016/j.freeradbiomed.2016.11.049.

Nonstandard Abbreviations and Acronyms

Ψ_mito

mitochondrial membrane potential

EMRE

essential mitochondrial calcium uniporter regulator

eNOS

endothelial nitric oxide synthase

geNOps

genetically encoded fluorescent nitric oxide probes

EGFP

enhanced green fluorescent protein

ER

endoplasmic reticulum

IP₃

inositol 1,4,5-trisphosphate

MCU

mitochondrial calcium uniporter

MICU1

mitochondrial calcium uptake 1

NO^•

nitric oxide

P-eNOS

phosphorylated endothelial nitric oxide synthase

RFP

red fluorescent protein

SOCE

store operated Ca²⁺ entry

TMRM

tetramethylrhodamine methyl ester perchlorate