(–)-Epicatechin Alters Reactive Oxygen and Nitrogen Species Production Independent of Mitochondrial Respiration in Human Vascular Endothelial Cells

Introduction Vascular endothelial dysfunction is characterised by lowered nitric oxide (NO) bioavailability, which may be explained by increased production of reactive oxygen species (ROS), mitochondrial dysfunction, and altered cell signalling. (-)-Epicatechin (EPI) has proven effective in the context of vascular endothelial dysfunction, but the underlying mechanisms associated with EPI's effects remain unclear. Objective(s). Our aim was to investigate whether EPI impacts reactive oxygen and nitrogen species (RONS) production and mitochondrial function of human vascular endothelial cells (HUVECs). We hypothesised that EPI would attenuate ROS production, increase NO bioavailability, and enhance indices of mitochondrial function. Methods HUVECs were treated with EPI (0-20 μM) for up to 48 h. Mitochondrial and cellular ROS were measured in the absence and presence of antimycin A (AA), an inhibitor of the mitochondrial electron transport protein complex III, favouring ROS production. Genes associated with mitochondrial remodelling and the antioxidant response were quantified by RT-qPCR. Mitochondrial bioenergetics were assessed by respirometry and signalling responses determined by western blotting. Results Mitochondrial superoxide production without AA was increased 32% and decreased 53% after 5 and 10 μM EPI treatment vs. CTRL (P < 0.001). With AA, only 10 μM EPI increased mitochondrial superoxide production vs. CTRL (25%, P < 0.001). NO bioavailability was increased by 45% with 10 μM EPI vs. CTRL (P = 0.010). However, EPI did not impact mitochondrial respiration. NRF2 mRNA expression was increased 1.5- and 1.6-fold with 5 and 10 μM EPI over 48 h vs. CTRL (P = 0.015 and P = 0.001, respectively). Finally, EPI transiently enhanced ERK1/2 phosphorylation (2.9 and 3.2-fold over 15 min and 1 h vs. 0 h, respectively; P = 0.035 and P = 0.011). Conclusion(s). EPI dose-dependently alters RONS production of HUVECs but does not impact mitochondrial respiration. The induction of NRF2 mRNA expression with EPI might relate to enhanced ERK1/2 signalling, rather than RONS production. In humans, EPI may improve vascular endothelial dysfunction via alteration of RONS and activation of cell signalling.


Introduction
Globally, cardiovascular disease (CVD) is the leading cause of morbidity and mortality [1]. One major risk factor for CVD is vascular endothelial dysfunction, which is typified by impaired vasodilation and diminished blood flow [2,3].
Vascular endothelial health is also impacted by mitochondrial function. Indeed, ageing is associated with reduced mitochondrial content in endothelial cells of conduit arteries, feed arteries, and capillaries [4,[17][18][19], which could be due to blunted transcriptional responses [4,19]. This potential reduction in mitochondrial biogenesis with ageing may result from diminished NO bioavailability [20,21] and/or lowered AMP-activated protein kinase (AMPK) signalling [22], both of which are known to activate peroxisome proliferator-activated receptor gamma coactivator 1alpha (PGC-1α). A link between mitochondria and vascular endothelial dysfunction has also been made from observations that human skeletal muscle feed arteries exhibit impaired respiratory capacity and lower coupling efficiency in middle-(55 years) and older-age (70 years) compared to young adults [18,23]. Furthermore, mitochondrial-targeted antioxidants like mitoquinone (MitoQ) restore vascular endothelial dysfunction in aged mice and patients with peripheral artery disease, which likely results from reductions in levels of mitochondrial superoxide [23,24]. Together, evidence points towards mitochondria as promising targets for interventions aimed at combatting vascular endothelial dysfunction.
(-)-Epicatechin (EPI) belongs to a subclass of flavonoids known as the flavanols. Not only is EPI highly bioavailable-reaching up to 10 μM in circulation in humans [25]-but EPI is also associated with several health benefits [26,27], including improved vascular endothelial function [28][29][30] and increased NO bioavailability [30]. Additional mechanisms thought to underly the therapeutic effects of EPI include increased eNOS phosphorylation [31][32][33], enhanced content or activity of enzymatic antioxidant proteins [34,35], and augmented mitochondrial biogenesis [32,[36][37][38]. Whilst the potential of EPI to enhance markers of mitochondrial biogenesis and antioxidant capacity seems promising, it is unclear whether this effect translates to enhanced respiratory function or altered ROS production of vascular endothelial cells. In fact, one recent study reported no impact of EPI on mitochondrial respiration [39], but the authors demonstrated that EPI may lower mitochondrial ROS production of HUVECs in the presence of high glucose. The challenge remains to resolve whether EPI modulates ROS production and mitochondrial function of vascular endothelial cells.
To this end, our aim was to investigate whether EPI modulates reactive oxygen and nitrogen species (RONS) production and mitochondrial function of human vascular endothelial cells. We hypothesised that EPI would attenuate ROS production, augment NO bioavailability, and enhance indices of mitochondrial function.

Materials and Methods
2.1. Cell Culture and Treatment. Human umbilical endothelial vein endothelial cells (HUVECs; Thermo Fisher Scientific, Waltham, MA, USA) at passages 3-7 were used in this study. HUVECs were not passaged more than 7 times because of changes in cell phenotype that can occur with multiple population doublings that ultimately lead to senescence [40][41][42]. Following the plating of cells onto pregelatinised well-plates (0.2% gelatin) in complete endothelial cell growth medium (EGM; Cell Applications Inc., San Diego, CA, USA),~80% confluent HUVECs were washed twice with Dulbecco's phosphate-buffered saline (D-PBS) and switched to prewarmed (37°C) EGM in the absence (vehicle [H 2 O], "CTRL") or presence of EPI (0.5-20 μM) over 24 h and 48 h. Human umbilical endothelial vein endothelial cells (HUVECs; Thermo Fisher Scientific, Waltham, MA, USA) at passages 3-7 were used in this study. HUVECs were not passaged more than 8 times because of changes in cell phenotype that can occur with multiple population doublings that ultimately lead to senescence [40][41][42]. Following the plating of cells onto pregelatinised well-plates (0.2% gelatin) in complete endothelial cell growth medium (EGM; Cell Applications Inc., San Diego, CA, USA),~80% confluent HUVECs were washed twice with Dulbecco's phosphatebuffered saline (D-PBS) and switched to prewarmed (37°C) EGM in the absence (vehicle [H 2 O], "CTRL") or presence of EPI (0-20 μM) over 24 h and 48 h.

Cell
Viability. The fluorescent CyQUANT® Proliferation Assay kit was used to determine cell viability. HUVECs were grown to 60-70% confluency in EGM in gelatinised 96-well plates. Cells were subsequently dosed for 24 h in EGM +/-EPI at 0-20 μM. After 24 h, wells were aspirated, washed twice with D-PBS, and then frozen immediately at -80°C. On the day of the assay, plates were thawed at room temperature, and CyQUANT® GR dye/cell-lysis buffer was added to each well according to manufacturer instructions. Plates were gently mixed on an orbital shaker (80 rpm) for 5 minutes protected from light. Sample fluorescence was measured using a CLARIOStar plate reader (BMG Labtech, Ortenberg, Germany) with 485/520 Ex/Em.

Mitochondrial ROS Production.
Mitochondrial superoxide was detected in HUVECs using MitoSOX (Thermo Fisher Scientific, Waltham, USA), a hydroethidine probe which is targeted to mitochondria by a conjugated triphenyl-phosphonium moiety [43]. HUVECs were seeded at 3 × 10 4 cells/mL in gelatinised 12-well microplates and at~80% confluence dosed +/-EPI (0-10 μM) for 24 h. Next, cells were washed in Krebs-Ringer buffer (KRH; 135 mM NaCl, 3.6 mM KCl, 10 mM HEPES (pH 7.4), 0.5 mM MgCl 2 , 1.5 mM CaCl 2 , 0.5 mM NaH 2 PO 4 , 2 mM glutamine, and 5 mM D(+)-Glucose) prior to incubation at 37°C for 30 minutes, with or without 15 μM antimycin A (AA) to stimulate mitochondrial superoxide production. Next, AAcontaining KRH was removed and MitoSOX was loaded into cells in fresh pre-warmed KRH to a final concentration of 2.5 μM. Plates were immediately transferred to a plate reader (ClarioStar, BMG Labtech), and fluorescence was monitored continuously at 30 sec intervals over 30 min at excitation/ emission of 510/580 nm. Rates of mitochondrial superoxide production were determined from the slope of the resultant progress curve over the 30-minute recording. Finally, plates  Table 1. After preparation, reaction tubes were transferred to a Rotor-Gene Q PCR thermal cycler for product amplification using a one-step protocol (QuantiFast SYBR® Green RT-PCR Kit, Qiagen, UK). The amplification protocol was as follows: reverse transcription (10 minutes at 50°C), transcriptase inactivation, and initial denaturation (95°C for 5 min) followed by 40× amplification cycles consisting of 95°C for 10 s (denaturation) and 60°C for 30 s (annealing and extension); followed by melt curve detection. Critical threshold (C T ) values were derived from setting a threshold of 0.09 for all genes. To quantify gene expression, C T values were used to quantify relative gene expression using the comparative Delta Delta C T (2 -ΔΔCT ) equation [44], whereby the expression of the gene of interest was determined relative to the internal reference gene (RPL13a) in the treated sample, compared with the untreated zero-hour control. The cells were incubated in this medium for 45 minutes at 37°C in a non-CO 2 incubator and then transferred to a Seahorse XFe24 extracellular flux analyser (Agilent, Santa Clara, CA, USA) maintained at 37°C. After an initial 15-minute calibration, oxygen consumption rate (OCR) was measured by a 3-4 loop cycle consisting of a 1 min mix, 2 min wait, and 3 min measure to record cellular respiration. After measuring basal respiration, 2 mM oligomycin was added to selectively inhibit the mitochondrial ATP synthase. Subsequently, 3 μM BAM15 was added to uncouple OCR to determine maximal respiration, and finally, a mixture of 2 μM rotenone and 2 μM AA was added to inhibit complex I and III of the electron transfer system, respectively, to determine nonmitochondrial respiration. Rates of oxygen consumption were corrected for nonmitochondrial respiration and expressed relative to the cell number of the appropriate well, determined by the CyQUANT® assay. The raw values of extracellular acidification rate (ECAR) and OCR were divided into component rates to calculate the relative contribution of glycolytic (ATP glyc ) and oxidative ATP-producing reactions (ATP ox ) to total ATP production, as previously described [45]. Three independent experiments were performed that contained at least two technical replicates.
2.9. Statistical Analysis. One-way ANOVAs were employed to detect effects of EPI treatment. Two-way ANOVAs were employed to detect main effects (e.g., dose, time, or antimycin A) and potential significant interactions between two main independent factors. Multiple comparisons between experimental conditions were adjusted for multiple tests, using Dunnett's or Sidak's where appropriate. All data are presented as mean ± SEM and significance accepted when P < 0:05.

EPI Does Not Cause Vascular Endothelial Cell Toxicity.
After 24 h EPI treatment (0.5-20 μM dose responses), there was a significant main effect of dose on cell proliferation (P = 0:018; Figure 1). However, multiple comparisons revealed no significant difference between doses of EPI versus CTRL. Given that EPI did not cause cell toxicity, and prior knowledge of physiologic EPI concentrations in vivo (up to 10 μM), subsequent experiments were performed with doses of 5 and 10 μM.

EPI Dose-Dependently Impacts the Expression of Genes Associated with Energy Metabolism in Vascular Endothelial
Cells. Next, experiments were performed to resolve whether EPI alters the expression of genes linked with mitochondrial function and the antioxidant response. First, the expression of genes associated with mitochondrial function and remodelling was quantified. In the presence of EPI, there was a significant main effect of dose (P = 0:018) and time (P = 0:002) on dynamin-related protein 1 (DRP1) expression (Figure 3(a)), but no significant dose × time interaction. At 48 h, 10 μM EPI increased DRP1 expression 2.2-fold compared to CTRL (P = 0:010). There was a significant main    Figure 3). Genes associated with the antioxidant response were also quantified. There was no effect of dose or time on catalase or eNOS expression. There was a significant main effect of dose and time on NADPH oxidase 4 (NOX4) expression in EPI-treated cells, respectively (P = 0:015 and P = 0:006). A significant main effect of dose was found on nuclear factor-erythroid factor 2-related factor 2 (NRF2) expression in the presence of EPI (P < 0:001; Figure 3(c)). NRF2 mRNA abundance was increased 1.5-fold and 1.6-fold with 5 and 10 μM EPI over 48 h when compared to CTRL (P = 0:015 and P = 0:001, respectively). There was a significant effect of time on superoxide dismutase 2 (SOD2) expression in EPI-treated cells only (P = 0:024). Multiple comparisons revealed that SOD2 expression was increased 2.1-fold in the presence of 10 μM EPI versus CTRL conditions (P = 0:040; Figure 3(d)).

EPI Does Not Alter Mitochondrial Bioenergetics of Vascular Endothelial Cells.
Having described that EPI influences RONS production and alters the expression of genes linked with mitochondrial function, we tested whether EPI impacted vascular endothelial cell bioenergetics. There was no significant main effect of EPI on rates of basal respiration, maximal respiration, ADP phosphorylation, proton leak, spare respiratory capacity (%), or coupling efficiency, regardless of dose (Figures 4(a), 4(b), and 4(d)). There was no effect of EPI on the relative contribution of ATP glyc or ATP ox to total ATP production (Figure 4(c)).    assessed cell signalling responses. There was a significant main effect of treatment and time on AMPKα phosphorylation, and a significant interaction was observed (P < 0:001; Figure 5 Whilst ERK1/2 is not involved in the AMPK/eNOS pathway, ERK1/2 signalling may mediate the effects of EPI upon vascular endothelial cell adaptation. There was no main effect of treatment (P = 0:141), but a significant main effect of time on ERK1/2 phosphorylation (P = 0:039; Figure 5(b)). There was also a significant treatment × time interaction (P = 0:003). Under CTRL conditions, despite a 2.7-fold increase in ERK1/2 phosphorylation at 3 h, ERK1/ 2 phosphorylation did not reach significance vs. 0 h until 24 hours posttreatment (0 h: 0:38 ± 0:10 vs. 24 h: 1:15 ± 0:28 AU; P = 0:022). By contrast, EPI significantly increased  To help establish whether the increase in NO brought about by EPI was associated with eNOS signalling, phosphorylation of eNOS at Ser1177 was assessed. There was no main effect of treatment or time on eNOS phosphorylation ( Figure 5(c)), and no treatment × time interaction (P = 0:100). At 3 h, eNOS phosphorylation was~60% higher under CTRL versus EPI conditions (CTRL: 0:38 ± 0:10 vs. EPI: 1:15 ± 0:28 AU, P = 0:038).

Discussion
Resolving EPI's mode of action using vascular endothelial cells as a model will help to establish its potential efficacy in mitigating vascular endothelial dysfunction. We tested the hypothesis that EPI would attenuate ROS production, augment NO bioavailability, and enhance mitochondrial function of human vascular endothelial cells in culture. We demonstrated that physiologic EPI concentrations, dosedependently, modulated RONS emission but did not directly impact mitochondrial respiration. Moreover, the influence of EPI on RONS emission was associated with the induction of increased NRF2 mRNA, which appears downstream of rapid and transient activation of ERK1/2 signalling. Taken together, our findings expand our knowledge of EPI's mechanisms of action in vitro and support further research on EPI as a potential instigator of cell signalling and NRF2 activation in vivo.

EPI Dose-Dependently Modulates RONS Production.
We have demonstrated that EPI dose-dependently altered mitochondrial ROS production in the absence of additional cell stress. Increased rates of mitochondrial ROS production, reported in the presence of 5 μM EPI, support previous observations that 10 days EPI supplementation increased superoxide production in mitochondria isolated from mouse heart tissue [46]. However, EPI has been demonstrated to increase the abundance and/or activity of key antioxidant proteins like SOD2 and catalase [47,48], and even to blunt hydrogen peroxide production from isolated brain and heart mitochondria [49]. In a similar way, we found that higher (10 μM) EPI concentrations attenuated the rate of mitochondrial superoxide production, which may have been facilitated by increased SOD2 mRNA expression. Clearly and importantly, the divergent effects of EPI on mitochondrial ROS production suggest EPI's biological effects are highly dose-dependent. Future studies should investigate whether EPI alters the vascular endothelial cell redox state and its potency relative to known mitochondrial antioxidants. Another noteworthy observation of this study was that EPI did not rescue AA-induced increases in mitochondrial ROS. Likewise, one recent study reported that 1 μM EPI did not lower the production of mitochondrial superoxide in HUVECs after acute AA treatment [39], suggesting limited efficacy for EPI in overcoming conditions associated with elevated mitochondrial ROS production in the vascular endothelium of humans.
The potent stimulation of NO production by EPI is well documented in studies using cell, human, and rodent models [30,32,33,50,51], although not all studies have reported such effects [52]. Here, we demonstrated that EPI increased NO levels of vascular endothelial cells, suggesting that EPI may be a promising strategy to combat vascular endothelial dysfunction, although further in vivo studies are required. Previous studies have attributed elevated NO bioavailability in the presence of EPI to increased phosphorylation of eNOS at Ser177 [31,32,[53][54][55]. Given that we and others demonstrated unaltered eNOS phosphorylation in the presence of EPI [39], it is plausible that arginase inhibition is the potential mechanism by which EPI increases NO production in HUVECs [56]. However, this may not be the case in arterial endothelial cells, where EPI has repeatedly been shown to enhance eNOS Ser1177 phosphorylation [31,32,55].

EPI Augments the Expression of Genes Linked with the Antioxidant Response and Mitochondrial Dynamics.
To better understand the mechanisms underlying EPI's effects, we measured the transcription of genes associated with energy metabolism. Interestingly, EPI increased the expression of genes DRP1 and MFN2, respectively, involved in mitochondrial fission and fusion. Despite these effects, EPI did not impact mitochondrial respiration, at least over 24 hours. Although mitochondrial dynamics can influence respiratory function [57,58], the lack of functional change in mitochondrial respiration could reflect a discord between cellular mRNA responses and changes in protein abundance and/ or function. Regardless of dose, EPI significantly enhanced NRF2 mRNA expression. This observation implies that the induction of NRF2 mRNA by EPI is independent of RONS production and might be explained by alternate factors that control NRF2 activity, like phosphorylation status [59]. Although posttranslational modifications of NRF2 were not assessed in this study, flavonoids have been shown to promote NRF2 phosphorylation and nuclear translocation [60,61], which would likely increase the transcription of genes related to the antioxidant response, including NRF2 itself.

EPI Does Not Impact Mitochondrial Respiration.
One important finding was that EPI does not impact mitochondrial respiration of HUVECs in culture. Although mitochondria have been proposed as potential molecular targets of EPI [62], previous investigations into EPI's effects on mitochondrial respiration have produced equivocal results. Some studies have demonstrated increased state 3 respiration in rat beta cells following EPI (0.1-2.5 μM) supplementation [63,64], whilst others have reported inhibited or similar state 3 respiration rates with EPI, depending on the substrates provided [65]. One recent study using HUVECs as a model reported that 0.1 and 1 μM EPI treatment over 2 hours had a negligible impact on mitochondrial respiration [39]. Taken together, our data suggest that the 8 Oxidative Medicine and Cellular Longevity therapeutic potential of EPI is not related to changes in mitochondrial respiration in vascular endothelial cells, pointing to alternate potential mechanisms of action like cell signalling activation [66].

EPI Rapidly and Transiently Activates ERK1/2
Signalling. The serine/threonine protein kinase AMPK is an important regulator of mitochondrial adaptation [67]. However, in our studies, AMPK phosphorylation was suppressed in the presence of EPI for up to 1 h and was without further impact for up to 24 h, compared with controls. Supporting these findings, 2 h EPI treatment (1 μM) did not affect the phosphorylation of AMPK in HUVECs [39]. Although EPI is capable of augmenting AMPK activity in liver and muscle tissue and several cell types [68][69][70], it seems that AMPK activation is not responsible for the therapeutic actions of EPI in HUVECs in culture. Together with the negligible impact of EPI on mitochondrial respiration, the data suggest that EPI does not directly alter vascular endothelial cell energy metabolism, in vitro. Importantly and of novelty, we reported that ERK1/2 phosphorylation at Thr202/Tyr204 was rapidly and transiently increased by EPI in HUVECs ( Figure 5). This finding supports recent observations of increased ERK1/2 signalling after 0.1 μM EPI treatment in bovine coronary artery endothelial cells that may be associated with phosphorylation of CaMKII [51]. How EPI promotes ERK1/2 phosphorylation in vascular endothelial cells remains to be described, but cur-rent evidence suggests EPI activates ERK1/2 by binding to the G-protein coupled estrogen receptor (GPER) on the cell membrane [51,71]. Given the induction of NRF2 mRNA expression found in this study, it would be useful to determine if acute activation of ERK1/2 signalling by EPI is a prerequisite for the induction of NRF2 activity.

4.5.
Limitations. We did not use metabolites of EPI that appear in circulation after ingestion of EPI-containing foods or supplements in vivo [72]. Therefore, the present findings may have limited translational potential. In our study, we harnessed HUVECs to model the vascular endothelial cell, and given that these cells are venous in nature, their physiology may not well reflect the arterial vasculature or microcirculation where EPI potentially exerts its beneficial effects. Further, not all assays were performed in the presence of additional cell stress. Thus, caution should be taken when translating our findings to human populations with disease. Finally, posttranslational modifications of NRF2, which are critical for regulating NRF2's activity, were not measured in this study.

Conclusion
We demonstrate that physiologic EPI concentrations do not impact mitochondrial respiration but do modulate reactive oxygen and nitrogen species production and the signalling and transcriptional activities of vascular endothelial cells  Figure 6: Schematic of the potential mechanisms by which EPI exerts its biological effects in vascular endothelial cells. Solid arrows/lines signify EPI's mode of action demonstrated in this study. Dashed arrows represent the potential activity of EPI not examined in this study but reported previously. 9 Oxidative Medicine and Cellular Longevity in vitro. EPI's dose-dependent alteration of reactive oxygen and nitrogen species production occurred in parallel with enhanced and transient ERK1/2 signalling and the induction of NRF2 mRNA ( Figure 6). The fact that EPI enhanced NRF2 mRNA expression regardless of dose implies that alterations in reactive oxygen and nitrogen species production alone were not solely responsible. Further research will help clarify the precise way in which EPI promotes ERK1/2 signalling and NRF2 activity and its relevance to vascular endothelial health in vivo.

Data Availability
The datasets generated and analysed during this study are available from the corresponding author upon request.

Conflicts of Interest
Daniel G. Sadler, Jonathan Barlow, Helen Jones, Dick H. J. Thijssen, and Claire E. Stewart had no conflict of interest associated with this manuscript. Richard Draijer is employed by Unilever.
Supplementary data containing primer sequences, western blot images, and additional qPCR figures. Table S1: primer sequences for homo sapiens with product length. All primers were used under the same cycling conditions. Figure S1: stain-free blot images of individual western blot experiments. Repeat one (a), two (b), and three (c). Figure S2: western blot analysis of HUVEC lysates. (a) pThr172-AMPKα and total AMPKα, (b) pThr202/Tyr204-p44/42 MAPK and total p44/42 MAPK, (c) pSer1177-eNOS and total eNOS. Figure S3: gene expression responses following acute EPI treatment. HUVECs were treated with 0, 5, and 10 μM EPI over 48 h and lysed for analysis of gene expression. (a) Parkin, (b) PGC-1α, (c) Sirt1, (d) Tfam, (e) Catalase, (f) eNOS, and (g) NOX4. Data are means ± SEM from 3 independent experiments. Statistical significance was determined by a two-way ANOVA, with dose and time as factors. Multiple comparisons were performed by Dunnett's test to determine differences in gene expression between conditions. β Main effect of time (P < 0:05); * P < 0:05. (Supplementary Materials)