Mitochondrial Remodeling in Endothelial Cells under Cyclic Stretch is Independent of Drp1 Activation

Mitochondria in endothelial cells remodel morphologically when supraphysiological cyclic stretch is exerted on the cells. During remodeling, mitochondria become shorter, but how they do so remains elusive. Drp1 is a regulator of mitochondrial morphologies. It shortens mitochondria by shifting the balance from mitochondrial fusion to fission. In this study, we hypothesized that Drp1 activation is involved in mitochondrial remodeling under supraphysiological cyclic stretch. To verify the involvement of Drp1, its activation was first quantified with Western blotting, but Drp1 was not significantly activated in endothelial cells under supraphysiological cyclic stretch. Next, Drp1 activation was inhibited with Mdivi-1, but this did not inhibit mitochondrial remodeling. Intracellular Ca increase activates Drp1 through calcineurin. First, we inhibited the intracellular Ca increase with Gd and thapsigargin, but this did not inhibit mitochondrial remodeling. Next, we inhibited calcineurin with cyclosporin A, but this also did not inhibit mitochondrial remodeling. These results indicate that mitochondrial remodeling under supraphysiological cyclic stretch is independent of Drp1 activation. In endothelial cells under supraphysiological cyclic stretch, reactive oxygen species (ROS) are generated. Mitochondrial morphologies are remodeled by ROS generation. When ROS was eliminated with N-acetyl-L-cysteine, mitochondrial remodeling was inhibited. Furthermore, when the polymerization of the actin cytoskeleton was inhibited with cytochalasin D, mitochondrial remodeling was also inhibited. These results suggest that ROS and actin cytoskeleton are rather involved in mitochondrial remodeling. In conclusion, the present results suggest that mitochondrial remodeling in endothelial cells under supraphysiological cyclic stretch is induced by ROS in association with actin cytoskeleton rather than through Drp1 activation.


Stretch chamber
Stretch chambers were prepared as described previously [Shinmura, Tsukamoto, Hamada et al. (2015)] with modifications. Base and curing agents of PDMS (Sylgard 184, Dow Corning Toray) were mixed at a ratio of 10:1. After being poured into a handmade mold, PDMS was solidified at 80°C for 1 h. The molded stretch chambers were treated with oxygen plasma (PDC-32G, Harrick Plasma) and then incubated with (3-aminopropyl) trimethoxysilane solution (Sigma, 1% in ddH2O) for 5 min. The stretch chambers were washed with ddH2O 3 times and heated at 80°C for 1 h. The stretch chambers were then activated with glutaraldehyde solution (Wako, 0.5% in PBS) for 10 min and washed with PBS 3 times. Finally, the stretch chambers were incubated with collagen solution (Koken, 10 µg/ml in PBS) at room temperature for 20 min and washed with PBS 3 times. On the stretch chambers, HAECs were cultured until reaching 70-100% confluence before experiments.

Cyclic stretch and fluorescence imaging
HAECs were stained with MitoTracker Orange (Molecular Probes, 1 µM in serum-free Medium 200) for 30 min or with MitoTracker Green (Molecular Probes, 0.2 µM in serum-free Medium 200) for 15 min at 37°C and washed with fresh serum-free Medium 200 3 times. After HAECs were stained, a stretch chamber was connected to a stepper motor Sigma Koki). Before loading the cyclic stretch, the stretch chamber was stabilized in a microscope incubator (Tokai Hit) under the conditions of 5% CO2 at 37°C for 30 min. The HAECs were subjected to cyclic stretch with an elongation of 5% or 20% and a frequency of 1 Hz. The cyclic stretch was paused for 2 min every 10 min in order to obtain fluorescence images. Fluorescence images of mitochondria were obtained in 6 imaging fields with a XY-stage-adapted confocal microscope (FV-1000, Olympus) through an objective lens (60×, N.A 1.1, Olympus) as described previously [Shinmura, Tsukamoto, Hamada et al. (2015)]. The wavelengths of excitation light and emission light for MitoTracker Orange were 516 nm and 576 nm, respectively. Those for MitoTracker Green were 490 nm and 543 nm, respectively.

Image analysis
The mitochondrial lengths were measured using fluorescence images as described previously [Shinmura, Tsukamoto, Hamada et al. (2015)]. In brief, fluorescence images were trimmed and their brightness was adjusted with ImageJ (NIH). The image size was standardized with Irfan View (freeware). Mitochondrial morphology was evaluated with FibrilTool [Boudaoud, Burian, Borowska-Wykręt et al. (2014)]. The mitochondrial morphologies were evaluated using the "anisotropy" analytical function of FibrilTool. The mitochondrial lengths were estimated using anisotropy [Shinmura, Tsukamoto, Hamada et al. (2015)]. For example, when the anisotropy values were low, the mitochondria were estimated to be short, and vice versa.

Western blotting for Drp1 activation
HAECs were subjected to cyclic stretch for 12 min and then were lysed with RIPA buffer. Protein concentrations were determined by a BCA protein assay kit (Pierce). Equal amounts of proteins were subjected to SDS-polyacrylamide gel electrophoresis. Proteins were transferred to PVDF membranes, which were blocked with 5% skim milk in TBS-T. The membranes were incubated with primary antibodies followed by HRP-conjugated secondary antibodies. Signals were obtained by Clarity/Clarity Max ECL Western blot substrate kits (Bio-Rad). The primary antibodies used in this study were anti-total Drp1 (BD, 611112), anti-phospo-Drp1 (Ser616) (Cell Signaling Technology, #4494), and antiactin (Millipore, MAB1501).

Inhibitors
Drp1 was inhibited by loading Mdivi-1 (Sigma, 20 µM) on HAECs for 15 min before the experiments. Ca 2+ influx through the Ca 2+ channel was inhibited by loading Gd 3+ (Sigma, 10 µM) for 5 min before the experiments. Ca 2+ release from the intracellular Ca 2+ store was inhibited by loading thapsigargin (Sigma, 1 µM) for 30 min before the experiments. Calcineurin was inhibited by loading cyclosporin A (Sigma, 10 µM) for 15 min before the experiments. Actin polymerization was inhibited by loading cytochalasin D (Sigma, 1 µM) for 30 min before the experiments. All the inhibitors were diluted in Medium 200 and loaded on HAECs under the conditions of 5% CO2 at 37°C. When cells were loaded with Gd 3+ or thapsigargin, they were used in experiments without being washed. When loaded with other inhibitors, cells were washed 3 times before the experiments.

Statistical analysis
All data are expressed as mean ± SEM. Data were analyzed using one-way analysis of variance (ANOVA), followed by Dunnett`s test for post-hoc analysis. Statistical significance is shown by the symbol ''**'' when p<0.01 and ''*'' when p<0.05. Otherwise, the symbol ''NS'' is shown for nonsignificant cases as needed.

Mitochondria remodeled in HAECs under supraphysiological cyclic stretch
Mitochondria in endothelial cells remodel their morphologies under supraphysiological 20% cyclic stretch [Shinmura, Tsukamoto, Hamada et al. (2015)]. In a previous study [Shinmura, Tsukamoto, Hamada et al. (2015)], bovine aortic endothelial cells (BAECs) were used as a model of endothelial cells. There was a scarcity of antibodies required for the detection of Drp1 activation in bovine cells. In BAECs, it was thus difficult to detect Drp1 activation with its antibodies. On the other hand, Drp1 antibodies for human cells were readily available. Drp1 activation could be easily detected in human cells. In this study, the model of endothelial cells was changed from BAECs to HAECs. Under physiological 5% cyclic stretch, mitochondria failed to remodel their morphologies in HAECs ( Fig. 1(A)). Under supraphysiological 20% cyclic stretch, mitochondria remodeled their morphologies ( Fig. 1(B)). The remodeling in HAECs was remarkable within the first 12 min of the stretch (Figs. 1(C), 1(D)). This remodeling under supraphysiological 20% cyclic stretch is analogous to that in BAECs [Shinmura, Tsukamoto, Hamada et al. (2015)]. These results validated that HAECs are available as a model of endothelial cells in which mitochondria remodel their morphologies under cyclic stretch with supraphysiological 20% elongation.

Figure 3:
Mitochondrial remodeling does not involve intracellular Ca 2+ increase or calcineurin. Inhibition of the intracellular Ca 2+ increase by Gd 3+ (10 µM) and thapsigargin (Tg, 1 µM), as well as that by calcineurin with cyclosporin A (CsA, 10 µM), all failed to inhibit mitochondrial remodeling under supraphysiological 20% cyclic stretch

Discussion
In this study, Drp1 was not involved in mitochondrial remodeling in endothelial cells under supraphysiological cyclic stretch. This implies that mitochondrial remodeling is not realized through mitochondrial fission, which is typically induced by Drp1 phosphorylation at Ser616. However, mitochondrial fission independent of Drp1 can occur, e.g., as observed in cells infected by bacteria [Stavru, Palmer, Wang et al. (2013)]. Actin cytoskeleton is involved in mitochondrial fission induced by bacterial infection. This involvement of actin cytoskeleton in mitochondrial remodeling coincides with its involvement in mitochondrial remodeling in endothelial cells under supraphysiological cyclic stretch (Fig. 4). Thus, it is also possible that mitochondrial remodeling under supraphysiological cyclic stretch is realized through mitochondrial fission. Mitochondrial remodeling is not limited to fission and fusion [Peng, Lin, Chen et al. (2011)]. Other mitochondrial dynamics, including globules, twisted tubules, and loop fragmentation, can be involved in mitochondrial remodeling under cyclic stretch. Those dynamics can apparently shorten mitochondrial. Fluorescence images were obtained with single slices under a confocal microscope in this study. To understand mitochondrial distributions in the Z-axis direction, i.e., in a direction parallel to fluorescence excitation light, 3D stack images are required. To understand whether those mitochondrial dynamics including lobules, twisted tubules, and loop fragmentation are involved in mitochondrial remodeling in endothelial cells under supraphysiological cyclic stretch, detailed observation and morphological analyses are required. Mitochondrial remodeling in endothelial cells under supraphysiological cyclic stretch was dependent on ROS (Fig. 4). This result follows previous findings that ROS generation is enhanced in endothelial cells under cyclic stretch [Ali, Mungai and Schumacker (2006); Ali, Pearlstein, Mathieu et al. (2004); Muliyil and Narasimha (2014); Sung, Yee, Eskin et al. (2007)]. Opa1 regulates mitochondrial morphology by enhancing mitochondrial fusion. When ROS is generated, Opa1 is cleaved and mitochondria are fragmented [Garcia, Innis-Whitehouse, Lopez et al. (2018)]. Thus, it is also possible that ROS generated by cyclic stretch released Opa1 from mitochondria and in turn enhanced mitochondrial fragmentation. Mitochondria anchor to the cytoskeleton via cross-linkers. Mitochondrial morphology, mobility, and positioning are physically supported by cytoskeletons [Jufri, Mohamedali, Avolio et al. (2015); Tang, Luo, Chen et al. (2014)]. Microtubules support mitochondrial long-range movements via microtubule motors. Actin cytoskeletons support those shortrange movements and dynamics [Boldogh and Pon (2006); Hoppins (2014); Jayashankar and Rafelski (2014)]. When the cells are subject to high-magnitude stretch of 20% or above, actin is reoriented perpendicular to the stretch direction [Kaunas, Nguyen, Usami et al. (2005); Shinmura, Tsukamoto, Hamada et al. (2015)]. Under the reorientation, actin cytoskeletons remodel. Due to this remodeling, mitochondria could remodel as well. Although we show that cyclic stretch with large elongation remodel mitochondrial morphologies independently of Drp1, it is not clarified whether this remodeling involves in pathologies including hypertension. Increased ROS generation might trigger endothelial cell dysfunction, possibly contributing to the development of hypertension [Tang, Luo, Chen et al. (2014)]. Although the precise mechanisms by which ROS triggers endothelial cell dysfunction remain elusive, mitochondrial remodeling induced by cyclic stretch can be involved. Further studies are required to understand whether and how mitochondrial remodeling is involved in the development of hypertension.