Mitochondrial connexin40 regulates mitochondrial calcium uptake in coronary endothelial cells

Abstract

Connexins (Cxs) are a group of integral membrane proteins that can form gap junctions between adjacent cells. Recently, it was reported that Cx43 is expressed not only in the plasma membrane but also in the inner mitochondrial membrane and that it regulates mitochondrial functions. Cx40 is predominantly expressed in vascular endothelial cells (ECs) and plays an important role in the electrical propagation between ECs and endothelial/smooth muscle cells. However, it is unknown whether Cx40 is expressed in the mitochondria and what the role of mitochondrial Cx40 is in endothelial functions. We observed in coronary ECs that Cx40 protein was expressed in the mitochondria, as determined by Western blot and immunofluorescence studies. We found that mouse coronary ECs (MCECs) isolated from Cx40 knockout (Cx40 KO) mice exhibited significantly lower resting mitochondrial calcium concentration ([Ca²⁺]_mito) than MCECs from wild-type (WT) mice. After increase in cytosolic Ca²⁺ concentration ([Ca²⁺]_cyto) with cyclopiazonic acid, calcium uptake into the mitochondria was significantly attenuated in MCECs from Cx40 KO mice compared with WT MCECs. There was no difference in resting [Ca²⁺]_cyto and store-operated calcium entry in MCECs from WT and Cx40 KO mice. We also detected a significant decrease in the concentration of mitochondrial reactive oxygen species (ROS) in Cx40 KO MCECs. Cx40 overexpression in ECs significantly increased resting [Ca²⁺]_mito level and calcium uptake by mitochondria in response to increased [Ca²⁺]_cyto and augmented mitochondrial ROS production. These data suggest that mitochondrial Cx40 contributes to the regulation of mitochondrial calcium homeostasis.

gap junctions are cell membrane channels, and gap junction intercellular communication (GJIC) is a pathway for the intercellular electrical propagation and transportation of small molecules (e.g., ATP) (50). One gap junction is composed of two hemichannels (i.e., connexon), and each hemichannel is made of six connexins (Cxs) (17). It has been reported that some hemichannels can also pass small molecules between the cytoplasm and extracellular space (18, 45). To date, 21 Cx genes have been identified in the human genome and 20 in the mouse. Cx37, Cx40, Cx43, and Cx45 and Cx32 (in some cases) are expressed in the vascular wall, and Cx40 is highly expressed in endothelial cells (ECs) (7, 9, 15, 19, 26, 27, 44, 56, 63). It should be noted that levels of these Cxs might be slightly different in ECs from different organs. Cx40 knockdown or inhibition of Cx40 attenuates vascular relaxation (13) and inhibits EC proliferation and migration (3, 21). Cx40 knockout (Cx40 KO) mice develop hypertension due to enhanced renin secretion (31). Although endothelium-specific Cx40 KO mice are normotensive (10), previous studies demonstrated that the loss of functional Cx40 in ECs is associated with accelerated atherosclerosis and exercise-induced hypertension (10, 41). We have also demonstrated that decreased Cx40 expression in mouse coronary endothelial cells (MCECs) is responsible for impaired endothelium-derived hyperpolarization-dependent vascular relaxation in coronary arteries and decreased capillary density in the left ventricle in diabetic mice (36). These data indicate the critical role of Cx40 in the plasma membrane in endothelial functions.

In addition to the role of Cxs in the regulation of GJIC, there is increasing evidence that some Cxs are expressed and function in the mitochondrial membranes. Mitochondrial Cx43 has been reported in ECs (33, 40, 62) and cardiac myocytes (4–6, 23, 30, 39, 48, 58). Inhibition of Cx43 attenuates K⁺ uptake (6, 39), Ca²⁺ uptake (58), and ADP-stimulated complex I respiration (5) in the mitochondria of cardiac myocytes. Stress stimulation (e.g., ischemia-reperfusion) increases the translocation of Cx43 to the mitochondria, and, moreover, Cx43 reduces the damage induced by stress (23, 30, 33, 48, 60). These data suggest that mitochondrial Cx43 exerts a cardioprotective effect upon stress stimulation. Fowler et al. (20) demonstrated that Cx32 is expressed in the inner mitochondrial membrane of mouse hepatocytes; however, its function is not clear.

In the present study, we demonstrate for the first time that Cx40 is expressed in the mitochondria of coronary ECs and functions as a regulator of mitochondrial Ca²⁺ homeostasis.

MATERIALS AND METHODS

Reagents.

Medium 199 (M199), streptomycin-penicillin, trypsin-EDTA, EC growth supplement, FBS, anti-Cx40 (for immunofluorescent study), anti-GAPDH, MitoTracker Green, MitoSOX Red, fura-2 AM, rhod-2 AM, Dynabeads sheep anti-rat IgG, and paraformaldehyde were obtained from Thermo Fisher Scientific. Collagenase II and dispase II were purchased from Worthington Biochemical. The Qproteome Mitochondria Isolation Kit was from Qiagen. Anti-Cx40 [for Western blot (WB)], anti- ATP synthase subunit α (ATP5A), and anti-Actin were from Santa Cruz Biotechnology. Rat anti-mouse CD31 was obtained from BD Biosciences. All other chemicals were from Sigma-Aldrich.

Animals and cell lines.

This study was conducted in accordance with the guidelines established by the Institutional Animal Care and Use Committee at the University of Arizona. All animal use was in compliance with all current US government regulations concerning the care and use of laboratory animals. The University of Arizona has been certified (Animal Welfare Assurance no. A3248-01), and the approved protocol number for this study is 14-520. All surgery was performed under anesthesia with a mixture of ketamine (100 mg/kg ip) and xylazine (5 mg/kg ip), and all efforts were made to minimize suffering. Male C57BL/6 mice were purchased from The Jackson Laboratory. Cx40 KO and Cx37 KO mice were kindly provided by Dr. Janis Burt (Univ. of Arizona) and bred in the animal facility of the University of Arizona. KO mice were used for the experiments starting at 5 wk of age (age range: 5–24 wk), and age-matched control mice were used for each time experiment. Human coronary endothelial cells (HCECs) were obtained from Lonza Group and cultured in EC medium composed of M199 supplemented with 10% FBS, 100 U/ml penicillin, 100 μg/ml streptomycin, 20 μg/ml EC growth supplement, and 16 U/ml heparin.

Isolation of mouse coronary endothelial cells.

MCECs were isolated as described previously (34–36). Briefly, dissected mouse heart was minced and incubated with M199 containing 1 mg/ml collagenase II and 0.6 U/ml dispase II for 1 h at 37°C. The digested material was collected and incubated with magnetic beads that were prepared as follows: Dynabeads sheep anti-rat IgG was incubated with rat anti-mouse CD31 monoclonal antibody (1 μg/ml) at 4°C overnight. The cell suspension was incubated with beads for 1 h at 4°C, and then MCECs were captured and isolated by the Dynal magnet.

Mitochondria isolation.

Mitochondria were isolated from HCECs with the Qproteome Mitochondria Isolation Kit according to the manufacturer’s instructions.

Western blot.

HCECs, MCECs, and isolated mitochondria from HCECs were lysed in lysate buffer (in mM: 25 HEPES, 150 NaCl, 10 MgCl₂, and 1 EDTA, with 1% IGEPAL, 1% protease inhibitor, and 1% phosphate inhibitor, pH 7.5). Proteins were separated with a 10% SDS-polyacrylamide gel and transferred to nitrocellulose membranes. Blots were incubated with the following primary antibodies: anti-Cx40 (Santa Cruz, sc-20466; 1:2,000), anti-ATP5A (Santa Cruz, sc-136178; 1:2,000), anti-Actin (Santa Cruz, sc-1616; 1:4,000), anti-Cx37 (Thermo Fisher Scientific, 42-4400; 1:1,000), anti-Cx43 (Thermo Fisher Scientific, 71-0700; 1:2,000), anti-Cx45 (Thermo Fisher Scientific, 41-5800; 1:2,000), or anti-GAPDH (Thermo Fisher Scientific, MA5-15738; 1:4,000), followed by horseradish peroxidase-linked secondary antibody incubation. The immunoblots were identified with the SuperSignal West Pico Chemiluminescent Substrate (Thermo Fisher Scientific, 34087). Band intensity was calculated with ImageJ 1.48v (National Institutes of Health), and the intensity data from focus proteins were normalized to GAPDH or Actin (for whole cell and cytosolic samples) or ATP5A (for mitochondrial samples) and are expressed in arbitrary units.

Detection of Cx40 localization in mitochondria by superresolution microscopy.

MCECs on coverslips were fixed with 100% methanol for 10 min at −20°C. Cells were blocked with 5% iron-fortified calf serum (IFCS) in PBS-Tween for 1 h followed by incubation with primary antibody against Cx40 (Thermo Fisher Scientific, 36-4900; 1:200) and ATP5A (Santa Cruz, sc-136178; 1:200) overnight. Anti-rabbit Alexa Fluor 488 and anti-mouse Alexa Fluor 594 antibodies (Thermo Fisher Scientific, A-21441 and A-21201; 1:4,000) and Hoechst 33342 (Thermo Fisher Scientific, H1399; 1 μg/ml) in 1% IFCS were used in secondary probing. Images were taken with a ×63 oil lens (numerical aperture: 1.40) and a Zeiss Elyra S.1 Structured Illumination Microscope and processed with Structured Illumination using ZEN 2 software (Carl Zeiss Microscopy).

Cytosolic Ca²⁺ concentration measurement with fura-2 AM.

Cytosolic Ca²⁺ concentration ([Ca²⁺]_cyto) in MCECs was measured with a modification of previously described methods (16). Cells on coverslips were loaded with fura-2 AM (4 µmol/l) for 1 h in the dark at room temperature. The dye was washed for 30 min with Ca²⁺-physiological salt solution (PSS) (in mM: 140 NaCl, 4.7 KCl, 1.8 CaCl₂, 1.2 MgCl₂, 10 HEPES, 10 glucose). The fura-2 fluorescence (340- and 380-nm excitation; 510-nm emission) from the cells and background fluorescence were imaged with an Eclipse Ti-E inverted fluorescence microscope (Nikon, Tokyo, Japan). Data are described as the ratio F₃₄₀/F₃₈₀. The peak [Ca²⁺]_cyto increase (Δpeak) and the area under the curve (AUC) of the peak were calculated (see Fig. 3A). Cells were isolated from at least three mice.

Fig. 3.

Effects of Cx40 deletion on [Ca²⁺]_cyto in MCECs. A: typical record of the change in [Ca²⁺]_cyto in MCECs. 1st peak describes the rise in [Ca²⁺]_cyto by CPA treatment in the absence of extracellular Ca²⁺ (indirect indicator of [Ca²⁺]_ER), and 2nd peak indicates store-operated Ca²⁺ entry (SOCE). Data are described as a ratio (F₃₄₀/F₃₈₀). B–F: summarized data of resting level of [Ca²⁺]_cyto (B), Δ1st peak (C), AUC of 1st peak (D), Δ2nd peak (E), and AUC of 2nd peak (F) in MCECs of WT mice (n = 74 cells) and Cx40 KO mice (n = 96 cells). Data are means ± SE. *P < 0.05 vs. WT.

Mitochondrial calcium concentration measurement with rhod-2 AM.

Mitochondrial calcium concentration ([Ca²⁺]_mito) in MCECs was measured with a method described in our previous publication (52). Cells on coverslips were loaded with rhod-2 AM (2 µmol) for 20 min in the dark at 37°C and washed with Ca²⁺-PSS for 70 min at 32°C. Rhod-2 fluorescence (530-nm excitation; 580-nm emission) from the cells and background fluorescence were imaged with a Nikon Eclipse Ti-E inverted fluorescence microscope with a ×20 objective lens. The background intensity was subtracted from the cell intensity. The peak [Ca²⁺]_mito increase and the AUC of the peak were calculated (see Fig. 4A). Cells were isolated from at least three mice.

Fig. 4.

Deletion of Cx40 decreases the resting level of [Ca²⁺]_mito, Δ2nd peak, and AUC of 2nd peak in MCECs. A: typical record of the change in [Ca²⁺]_mito in MCECs. B–F: summarized data of resting level of [Ca²⁺]_mito (B), Δ1st peak (C), AUC of 1st peak (D), Δ2nd peak (E), and AUC of 2nd peak (F) in MCECs of WT mice (n = 50 cells) or Cx40 KO mice (n = 100 cells). Data are means ± SE. *P < 0.05 vs. WT.

[Ca²⁺]_mito measurement with mitocameleon.

We first generated Tie2-mitocameleon-adenovirus (Tie2-mtcam-Adv). The CMV-mtcam plasmid was kindly provided by Dr. Roger Tsien from the University of California, San Diego (46). Mtcam encodes duplicated mitochondria-targeting sequences and the cameleon sequence [enhanced cyan fluorescent protein (ECFP)-calmodulin-enhanced yellow fluorescent protein (EYFP) that generates a fluorescence resonance energy transfer (FRET) signal when Ca²⁺ binds to calmodulin]. Cx40-Adv was generated as described in our previous report (36). HCECs were transduced with Tie2-mtcam-Adv (100 pfu/cell) with Cx40-Adv or control-Adv (100 pfu/cell). Two days after transduction, the experiments for [Ca²⁺]_mito were conducted. FRET (excitation 430 nm, emission 535 nm) and CFP (excitation 430 nm, emission 470 nm) fluorescence were imaged with a Nikon Eclipse Ti-E inverted fluorescence microscope with a ×20 objective lens. Data are described as the ratio FRET/CFP. The peak [Ca²⁺]_mito increase and the AUC of the peak were calculated for individual cells. We used rhod-2 in MCECs instead of Tie2-mtcam-Adv because a high titer of Tie2-mtcam-Adv (1,000 pfu/cell) was required to achieve sufficient signal in MCECs and this titer was cytotoxic.

Measurement of mitochondrial reactive oxygen species.

Mitochondrial reactive oxygen species ([ROS]_mito) was measured as described previously (11, 37). Briefly, MCECs were stained with 5 µM MitoSOX Red (a mitochondrial ROS indicator; excitation 510 nm, emission 580 nm) and 100 nM MitoTracker Green (excitation 490 nm, emission 516 nm) for 30 min at 37°C. Images were captured with a Nikon Eclipse Ti-E inverted fluorescence microscope with a ×60 objective lens (numerical aperture: 1.4). The fluorescence intensities were calculated with ImagePro-PLUS 7.0 software (Media Cybernetics). MitoTracker Green signal was used to define the mitochondrial area, and the intensity of MitoSOX Red was measured in the mitochondria. The background intensity was subtracted from the cell intensity.

Statistical analysis.

Statistical analysis was performed with SigmaPlot 12.5 (Systat Software). Data are expressed as means ± SE. Student’s t-test for unpaired samples was used to identify significant differences. Differences were considered to be statistically significant when P < 0.05.

RESULTS

Evidence of Cx40 existence in mitochondria of coronary ECs.

HCECs were used and mitochondria and cytosol were subfractionated from whole cell lysate. We used mitochondrial ATP5A as a mitochondria-specific marker and GAPDH as a marker for the cytosolic fraction for the WB study. The strong immunoreactivity of ATP5A and a lack of GAPDH signal in the mitochondrial fraction demonstrate the high purity of our mitochondrial preparation. Strong Cx40 signal was detected in the mitochondrial fraction isolated from HCECs (Fig. 1A). In addition, superresolution microscopy images showed the colocalization of Cx40 with Tom20 (a mitochondrial marker) in ECs isolated from mice (MCECs) (Fig. 1B).

Fig. 1.

Mitochondrial Cx40 expression in coronary ECs. A: Western blots showing Cx40, GAPDH, and ATP5A expression in whole cell lysate, mitochondrial fraction, and cytosolic fraction of HCECs. ATP5A was used as a specific marker for the mitochondria and GAPDH as a marker for cytosolic proteins. B: photomicrographs show typical image of Cx40 localization in MCECs. MCECs were stained with Cx40 (green) and Tom20 (a marker of mitochondria; red). Left: white arrows indicate the lining of Cx40 in the plasma membrane. Right: yellow arrows indicate the Cx40 in (or on) the mitochondria.

Cx protein levels in MCECs from Cx40 KO mice.

Previous studies found that the loss of Cx40 alters the expression of other Cxs in ECs (29, 55). To confirm the expression profile of Cxs in Cx40 KO mice, MCECs were isolated from Cx40 KO and wild-type (WT) mice. As shown in Fig. 2, A and B, Cx40 protein was not detectable in MCECs from Cx40 KO mice. We found that there is no significant change in Cx37, Cx43, and Cx45 levels (Fig. 2. C–F) in MCECs from these mice. Systemic Cx37 deletion did not affect Cx40 level in MCECs. Immunofluorescent images demonstrate Cx40 expression in the plasma membrane, mitochondria, and cytosol in WT MCECs and diminished signals of Cx40 in Cx40 KO MCECs (Fig. 2G).

Fig. 2.

Protein levels of Cx40, Cx37, Cx43, and Cx45 in MCECs. A and D: typical WB images of Cx40, Cx37, Cx43, and Cx45 protein levels in the whole lysate of MCECs isolated from WT, Cx40 KO, and Cx37 KO mice. Actin was used as a loading control for samples of whole lysate from MCECs. B: Cx40 protein level normalized to actin: WT (n = 5) and Cx40 KO (n = 5). Data are means ± SE. C: Cx37 protein levels in MCECs of WT and Cx40 KO mice (n = 4 in each group). Data are means ± SE. E: Cx43 protein levels in MCECs of WT and Cx40 KO mice (n = 5 in each group). Data are means ± SE. F: Cx45 protein levels in MCECs of WT and Cx40 KO mice (n = 6 in each group). Data are means ± SE. G: photomicrographs show typical image of Cx40 expression in WT MCECs (left) and Cx40 KO MCECs (right). MCECs were stained with Cx40 (green) and Tom20 (a marker of mitochondria; red).

Effects of Cx40 deletion on cytosolic Ca²⁺ concentration.

Figure 3A illustrates the time course of [Ca²⁺]_cyto measured with fura-2 AM. After the resting level of [Ca²⁺]_cyto was obtained in Ca²⁺-PSS, perfusate was changed to Ca²⁺-free-PSS for 5 min. Then, we added 10 µM cyclopiazonic acid [CPA, a sarco(endo)plasmic reticulum Ca²⁺-ATPase blocker] to Ca²⁺ free-PSS. This “1st peak” of [Ca²⁺]_cyto is representative of the Ca²⁺ leak from the endoplasmic reticulum (ER), and it is an indirect indicator of Ca²⁺ concentration in the ER ([Ca²⁺]_ER). Ten minutes after the application of CPA, the perfusate was changed to Ca²⁺-PSS supplemented with CPA to initiate the “2nd peak” of [Ca²⁺]_cyto. This 2nd peak results from store-operated calcium entry (SOCE). As shown in Fig. 3B, the resting level of [Ca²⁺]_cyto was not different between MCECs from WT and Cx40 KO mice. Interestingly, we observed that Δ1st peak was slightly, but significantly, increased in MCECs from Cx40 KO mice compared with WT MCECs (Fig. 3C). However, Δ2nd peak and AUC of 1st peak and 2nd peak did not show any difference between WT MCECs and Cx40 KO MCECs (Fig. 3, D–F).

[Ca²⁺]_mito was decreased in Cx40 KO MCECs.

Next, we used rhod-2 AM to measure [Ca²⁺]_mito in MCECs. Figure 4 demonstrates that the resting level of [Ca²⁺]_mito was significantly decreased in Cx40 KO MCECs compared with MCECs from WT mice. Although there was no difference in Δ1st peak and AUC of 1st peak, Δ2nd peak and AUC of 2nd peak were significantly decreased in Cx40 KO MCECs compared with the control.

Cx40 deletion decreased the level of [ROS]_mito.

It has been reported that change of [Ca²⁺]_mito influences [ROS]_mito (1, 8, 22, 52). Therefore, we examined whether [ROS]_mito would be altered in MCECs isolated from Cx40 KO mice. MitoSOX was used to measure [ROS]_mito, and MitoTracker Green was used to label mitochondria. As shown in Fig. 5, [ROS]_mito was significantly decreased in MCECs from Cx40 KO mice compared with the control.

Fig. 5.

MCECs isolated from Cx40 KO mice exhibit decreased mitochondrial ROS ([ROS]_mito) formation. A: representative images showing [ROS]_mito in MCECs isolated from WT (left) and Cx40 KO mice (right). Bars, 10 μm. B: summarized data of [ROS]_mito. I_c, intensity of the cell; I_b, intensity of background. n = 66 cells (WT) and 106 cells (Cx40 KO). Data are means ± SE. *P < 0.05 vs. WT.

Effect of Cx40 overexpression on [Ca²⁺]_mito.

To further confirm the effect of Cx40 on [Ca²⁺]_mito, we overexpressed Cx40 in HCECs with an adenovirus carrying the Cx40 gene (Cx40-Adv). As shown in Fig. 6, A–C, Cx40-Adv transduction significantly increased the level of Cx40 protein in mitochondrial, cytosolic, and whole cell samples, respectively. [Ca²⁺]_mito was measured with mtcam in this study. mtcam consists of YFP and CFP that are linked by calmodulin (46) and includes a mitochondria-targeting sequence. When [Ca²⁺]_mito increases, the ratiometric FRET signal to CFP signal also increases. In contrast to the results from Cx40 KO MCECs, overexpression of Cx40 significantly increased the resting level of [Ca²⁺]_mito, Δ1st peak, Δ2nd peak, and AUC of 2nd peak (Fig. 7, A–E). We also repeated the experiments using MCECs (Fig. 7, F–J). After Cx40-Adv transduction, [Ca²⁺]_mito in MCECs was measured with rhod-2 AM. In line with the results in HCECs, Cx40 overexpression in MCECs significantly increased all parameters of [Ca²⁺]_mito.

Fig. 6.

Cx40 adenovirus transduction in HCECs increases Cx40 levels in mitochondrial fraction. A: mitochondrial fraction. B: cytosolic fraction. C: whole cell lysate. Graphs show the summarized data of Cx40 protein level normalized to respective loading control. n = 5 (A), 5 (B), and 8 (C). Cont Adv, control adenovirus; Cx40 Adv, Cx40-adenovirus. Data are means ± SE. *P < 0.05 vs. Cont Adv.

Fig. 7.

Overexpression of Cx40 increases resting level of [Ca²⁺]_mito, Δ1st peak, Δ2nd peak, and AUC of 2nd peak in HCECs and MCECs. A–E: summarized data of resting [Ca²⁺]_mito level (A), Δ1st peak (B), AUC of 1st peak (C), Δ2nd peak (D), and AUC of 2nd peak (E) in HCECs transduced with Cont Adv (n = 75) or Cx40 Adv (n = 76). Data are means ± SE. *P < 0.05 vs. Cont Adv. F–J: summarized data of resting [Ca²⁺]_mito level (F), Δ1st peak (G), AUC of 1st peak (H), Δ2nd peak (I), and AUC of 2nd peak (J) in MCECs transduced with Cont Adv (n = 68) or Cx40 Adv (n = 69). Data are means ± SE. *P < 0.05 vs. Cont Adv.

Cx40 overexpression increased the level of [ROS]_mito.

Cx40 deletion decreased [Ca²⁺]_mito and [ROS]_mito (Fig. 4 and Fig. 5). Because Cx40 overexpression significantly increased [Ca²⁺]_mito (Fig. 7), we examined whether [ROS]_mito was altered by Cx40 overexpression. Cx40-Adv was used to transduce ECs, and [ROS]_mito was measured with MitoSOX. As shown in Fig. 8, Cx40 overexpression significantly increased the level of [ROS]_mito in HCECs and MCECs. These data suggest that the increase in Cx40 in mitochondrial might be harmful to the cells.

Fig. 8.

Cx40 overexpression augments mitochondrial ROS ([ROS]_mito) formation in HCECs and MCECs. [ROS]_mito was measured 48 h after Cx40 gene transduction. A: summarized data of [ROS]_mito in HCECs. n = 63 (Cont Adv) and 52 (Cx40 Adv). Data are means ± SE. *P < 0.05 vs. Cont Adv. B: summarized data of [ROS]_mito in MCECs. n = 81 (Cont Adv) and n = 93 (Cx40 Adv). Data are means ± SE. *P < 0.05 vs. Cont Adv.

DISCUSSION

The role of Cxs located at the plasma membrane in the regulation of hemichannel activity has been well studied for many decades, whereas the functional role of mitochondrial Cxs in cellular functions has been less well explored. The first publication that evidenced the existence of a mitochondrial Cx was in 2002 by Li et al. (33). They identified the expression of Cx43 in the mitochondrial fraction with WB and measured the signal of Cx43 in the mitochondria with immunofluorescence in human umbilical vein endothelial cells. Later, two groups reported that retinal ECs also expressed mitochondrial Cx43 and demonstrated that high-glucose treatment decreased mitochondrial Cx43 expression, which resulted in altered mitochondrial function (40) and mitochondrial morphology (62). Most studies on mitochondrial Cx43 have been conducted in cardiac myocytes, perhaps because Cx43 is strongly expressed in these cells, and the localization and function of mitochondrial Cx43 are therefore better understood in cardiac myocytes (4–6, 23, 30, 39, 48, 58).

In vascular ECs, Cx40 is strongly expressed and plays an important role in the regulation of GJIC (7, 9, 15, 19, 24, 26, 27, 56, 63). There are, however, no reports investigating mitochondrial Cx40 in ECs. Therefore, we examined in this study whether Cx40 is expressed in the mitochondria from ECs. First, we confirmed that the mitochondrial isolation yielded highly purified mitochondria, as determined by WB (Fig. 1A). With this method, we observed a strong Cx40 signal in the mitochondrial fraction. It should be noted that the same amount of protein (8 μg/lane) was loaded for whole cell, mitochondrial, and cytosolic samples. From WB it appears that Cx40 is most abundant in the mitochondrial fraction; however, this may not be the case. The pool of mitochondrial proteins is much smaller than the pool of cytosolic and whole cell proteins, and Cx40 constitutes a higher percentage of the protein found in mitochondria than in the cytosol or whole cell. By loading an equal amount of protein for each sample type, a larger relative abundance of Cx40 is therefore expected in mitochondria, as we observed.

As shown in Fig. 1B, left, the majority of Cx40 is located at the plasma membrane, where it forms hemichannels and gap junctions (indicated with white arrows). However, at higher magnification (Fig. 1B, right) Cx40 is shown to colocalize with a mitochondrial marker, suggesting that Cx40 also exists on (or in) the mitochondria. In this study we do not identify where Cx40 is in the mitochondria (e.g., inner or outer mitochondrial membrane) because the low concentration of mitochondria that was obtained from coronary ECs precluded the further separation of inner and outer membrane fractions. It has been shown that Cx43 is expressed in the inner mitochondrial membrane in cardiac myocytes (48), and it will be of interest to determine the mitochondrial membrane localization of Cx40 in future studies.

To examine the functional role of mitochondrial Cx40 in ECs, we used systemic Cx40 KO mice to isolate MCECs. In Fig. 2, we confirmed the expression of Cx40, Cx37, Cx43, and Cx45 in WT MCECs and the absence of detectable Cx40 in MCECs of Cx40 KO mice. Cx45 expression levels were not altered in Cx40 KO MCECs, while Cx40 deletion markedly decreased Cx37 levels and increased Cx43 levels. This finding is in line with previous reports (2, 54), and there is as yet no definitive explanation as to why Cx40 KO decreases Cx37 expression. It may be possible that Cx40 somehow acts as a regulator of gene expression for other Cxs, but further experiments are needed to evaluate this hypothesis. We also tested Cx32 expression level in MCECs and found that Cx32 was not detectable in MCECs (data not shown).

The available information about the functional role of mitochondrial Cxs comes only from the studies on Cx43 in cardiac myocytes. Miro-Casas et al. (39) demonstrated that Cx43 hemichannels on the mitochondrial membrane gate K⁺ and that inhibition of Cx43 decreases K⁺ uptake into mitochondria. Boengler et al. (6) used Cx43 mimetic peptide to inhibit Cx43 and showed an inhibition of K⁺ influx in the mitochondria. Mitochondrial K⁺ channels are important regulators of inner mitochondrial membrane integrity by balancing potassium concentration in the intermembrane space and matrix and maintaining the mitochondrial membrane potential (12, 14, 28, 49, 53, 59, 61). It is known that the opening of mitochondrial K⁺ channels leads to the depolarization of the mitochondrial membrane, reduces mitochondrial Ca²⁺ overload, and results in cardioprotection (38, 42, 43, 57). There is increasing evidence that mitochondrial Cx43 reduces mitochondrial damage induced by pathophysiological stress (23, 30, 33, 48, 60).

Srisakuldee et al. (58) demonstrated that phosphorylation of Cx43 attenuates Ca²⁺-induced mitochondrial swelling; however, there is no evidence showing that mitochondrial Cx43 directly regulates [Ca²⁺]_mito. In most cells, including ECs, mitochondrial Ca²⁺ overload leads to mitochondria-induced cell apoptosis (25, 32). Therefore, inhibition of mitochondrial Ca²⁺ overload is a beneficial approach to decrease cell apoptosis under pathophysiological conditions (52). Our data show that the loss of Cx40 decreased the resting [Ca²⁺]_mito and attenuated Ca²⁺ influx into the mitochondria after SOCE without changing the levels in [Ca²⁺]_cyto, suggesting that mitochondrial Cx40 directly regulates mitochondrial Ca²⁺ uptake. In addition, overexpression of Cx40 increased the resting level of [Ca²⁺]_mito and the peak of Ca²⁺ influx in the mitochondria. These data suggest that pathophysiological increase of mitochondrial Cx40 leads to augmented Ca²⁺ uptake into the mitochondria and might result in enhanced mitochondria-induced cell apoptosis. In contrast to the result from mitochondrial Cx40, the loss of Cx40 in the plasma membrane has maladaptive effects on endothelial functions (10, 31, 35, 36, 41). We have demonstrated that restoring Cx40 levels in diabetic MCECs had a beneficial effect on diabetes-induced endothelial dysfunction (35, 36). The result of the present study was therefore contrary to our expectations, and we will examine the role of mitochondrial Cx40 in endothelial functions under pathophysiological conditions (e.g., diabetes) in future experiments.

In the present study, Δ1st peak of [Ca²⁺]_cyto was slightly enhanced in MCECs from Cx40 KO mice compared with WT MCECs. The increase of [Ca²⁺]_cyto indicates that 1) Ca²⁺ leak from the ER is enhanced and/or 2) [Ca²⁺]_ER is increased. Since AUC of 1st peak was not significantly changed, the cause of the increase of Δ1st peak is likely the upregulation of Ca²⁺ leak from the ER. Ca²⁺ leak from the ER in ECs is mediated by the inositol 1,4,5-trisphosphate receptor on the ER. One possible explanation for the increase in Δ1st peak in Cx40 KO MCECs is that Cx40 deletion influences inositol 1,4,5-trisphosphate expression as shown in other proteins (2, 54).

It has been reported that Cx40 inhibition did not affect stimulation-induced [Ca²⁺]_cyto increase in uterine artery ECs (64). The cause of these contradictory results might be the difference in the tissue of origin. The increase of [Ca²⁺]_mito induced by SOCE was prominently altered by the deletion of Cx40, but no change in Δ1st peak of [Ca²⁺]_mito, suggesting that Cx40-mediated Ca²⁺ uptake into the mitochondria might not be mediated through the direct Ca²⁺ transport from the ER.

Mitochondria are constantly generating ROS as a by-product of the activity of the electron transport chain. However, excess ROS production is harmful to cellular functions and potentially leads to apoptosis. We and other investigators have shown that mitochondrial ROS formation is also controlled by [Ca²⁺]_mito (1, 47, 51, 52, 65). We found that Cx40 deletion decreased mitochondrial ROS formation while Cx40 overexpression increased [ROS]_mito, implying that mitochondrial Cx40 might regulate the level of [ROS]_mito via altering [Ca²⁺]_mito.

In this study, we do not identify whether mitochondrial Cx40 forms gap junctions or acts as hemichannels in the mitochondrial membrane. Miro-Casas et al. (39) implied that mitochondrial Cx43 in cardiac myocytes might potentially form hemichannels; however, it is methodologically difficult to confirm the form of connexin channels (i.e., hemichannel or gap junctions) in mitochondria. It has been reported that hemichannels in the plasma membrane can also transfer small molecules like gap junctions (18, 45).

In conclusion, coronary ECs express Cx40 not only in the plasma membrane but also in the mitochondria. Mitochondrial Cx40 functions as a Ca²⁺ gate, regulating Ca²⁺ uptake into the mitochondria and mitochondrial ROS formation. The findings in this study provide important and useful information that can be used to further study mitochondrial Cx40 in ECs under pathophysiological conditions.

GRANTS

This work was supported by National Heart, Lung, and Blood Institute Grant HL-115578 to A. Makino.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

R.G., R.S., and A.M. performed experiments; R.G., R.S., and A.M. analyzed data; R.G. and A.M. interpreted results of experiments; R.G. and A.M. prepared figures; R.G. drafted manuscript; R.G., R.S., B.T.S., and A.M. approved final version of manuscript; B.T.S. and A.M. edited and revised manuscript.