Redox and metal profiles in human coronary endothelial and smooth muscle cells under hyperoxia, physiological normoxia and hypoxia: Effects of NRF2 signaling on intracellular zinc

Zinc is an important component of cellular antioxidant defenses and dysregulation of zinc homeostasis is a risk factor for coronary heart disease and ischemia/reperfusion injury. Intracellular homeostasis of metals, such as zinc, iron and calcium are interrelated with cellular responses to oxidative stress. Most cells experience significantly lower oxygen levels in vivo (2–10 kPa O2) compared to standard in vitro cell culture (18kPa O2). We report the first evidence that total intracellular zinc content decreases significantly in human coronary artery endothelial cells (HCAEC), but not in human coronary artery smooth muscle cells (HCASMC), after lowering of O2 levels from hyperoxia (18 kPa O2) to physiological normoxia (5 kPa O2) and hypoxia (1 kPa O2). This was paralleled by O2-dependent differences in redox phenotype based on measurements of glutathione, ATP and NRF2-targeted protein expression in HCAEC and HCASMC. NRF2-induced NQO1 expression was attenuated in both HCAEC and HCASMC under 5 kPa O2 compared to 18 kPa O2. Expression of the zinc efflux transporter ZnT1 increased in HCAEC under 5 kPa O2, whilst expression of the zinc-binding protein metallothionine (MT) decreased as O2 levels were lowered from 18 to 1 kPa O2. Negligible changes in ZnT1 and MT expression were observed in HCASMC. Silencing NRF2 transcription reduced total intracellular zinc under 18 kPa O2 in HCAEC with negligible changes in HCASMC, whilst NRF2 activation or overexpression increased zinc content in HCAEC, but not HCASMC, under 5 kPa O2. This study has identified cell type specific changes in the redox phenotype and metal profile in human coronary artery cells under physiological O2 levels. Our findings provide novel insights into the effect of NRF2 signaling on Zn content and may inform targeted therapies for cardiovascular diseases.


A B S T R A C T
Zinc is an important component of cellular antioxidant defenses and dysregulation of zinc homeostasis is a risk factor for coronary heart disease and ischemia/reperfusion injury. Intracellular homeostasis of metals, such as zinc, iron and calcium are interrelated with cellular responses to oxidative stress. Most cells experience significantly lower oxygen levels in vivo (2-10 kPa O 2 ) compared to standard in vitro cell culture (18kPa O 2 ). We report the first evidence that total intracellular zinc content decreases significantly in human coronary artery endothelial cells (HCAEC), but not in human coronary artery smooth muscle cells (HCASMC), after lowering of O 2 levels from hyperoxia (18 kPa O 2 ) to physiological normoxia (5 kPa O 2 ) and hypoxia (1 kPa O 2 ). This was paralleled by O 2 -dependent differences in redox phenotype based on measurements of glutathione, ATP and NRF2-targeted protein expression in HCAEC and HCASMC. NRF2-induced NQO1 expression was attenuated in both HCAEC and HCASMC under 5 kPa O 2 compared to 18 kPa O 2 . Expression of the zinc efflux transporter ZnT1 increased in HCAEC under 5 kPa O 2 , whilst expression of the zinc-binding protein metallothionine (MT) decreased as O 2 levels were lowered from 18 to 1 kPa O 2 . Negligible changes in ZnT1 and MT expression were observed in HCASMC. Silencing NRF2 transcription reduced total intracellular zinc under 18 kPa O 2 in HCAEC with negligible changes in HCASMC, whilst NRF2 activation or overexpression increased zinc content in HCAEC, but not HCASMC, under 5 kPa O 2 . This study has identified cell type specific changes in the redox phenotype and metal profile in human coronary artery cells under physiological O 2 levels. Our findings provide novel insights into the effect of NRF2 signaling on Zn content and may inform targeted therapies for cardiovascular diseases.

Introduction
Coronary artery and heart disease are leading causes of mortality worldwide, yet only recent studies have focused on the association between these conditions and trace metal micronutrients such as zinc (Zn) [1][2][3][4][5][6]. Dysregulation of Zn 2+ homeostasis is associated with ischemia/reperfusion injury [4,5,[7][8][9]. Serum zinc levels are low in patients undergoing cardiac surgery and, although an association between zinc deficiency and postoperative outcomes remains to be established [10], elevated zinc levels and zinc supplementation are associated with reduced post MI scarring [11] and cardiomyopathy [12].
Although zinc is a redox-inert metal, physiological concentrations of zinc have antioxidant, anti-inflammatory and anti-proliferative properties [13][14][15][16][17][18], whilst zinc deficiency or overload generates oxidative stress [19]. Intracellular zinc concentrations change dynamically to regulate both rapid cellular events and slow transcriptional responses [3,[20][21][22]. In human cells, the concentration of free Zn 2 ions r estimated at a few hundred picomolar but the total zinc concentration approximates one hundred micromolar [3]. Free Zn 2+ can increase transiently above low nanomolar concentrations, yet its role as a signaling ion is tightly regulated by cytosolic buffering and the activity of transporters that remove Zn 2+ ions from the cytosol [20,22,23]. ZnT (solute-linked carrier 30) transport proteins lower zinc concentrations through cellular efflux or uptake into cellular compartments, whilst ZIP (solute-linked carrier 39) proteins increase zinc levels through influx into cells or export from organelles [24,25]. In addition, metallothioneins (MT), small proteins containing thiolate clusters, bind Zn 2+ ions with a range of affinities [19,26]. Thus, tight control of cellular Zn 2+ levels via multiple transporters maintains cellular function, including thiol/disulfide redox homeostasis because the majority of intracellular zinc is bound to redox-sensitive thiols of the amino acid cysteine [3,23]. Increases in free Zn 2+ alter gene transcription through their modification of critical proteins controlling cellular signaling.
Endothelial and smooth muscle take up Zn 2+ [27], and zinc protects the endothelium against oxidative stress [28,29]. Sustained increases in intracellular Zn 2+ in human endothelial cells leads to apoptotic cell death as a consequence of impaired glutathione homeostasis and mitochondrial ATP synthesis [30], further highlighting the intricate cross-talk between zinc, reactive oxygen species and endogenous antioxidant defenses regulated by nuclear factor (erythroid-derived 2)-like 2 (NRF2) [31][32][33][34][35]. In contrast, zinc deficiency induces an inflammatory endothelial cell phenotype, associated with increased monocyte adhesion which can be reversed by zinc supplementation [36]. Moreover, zinc supplementation has been shown to restore autophagic flux in heart failure patients [12].
The importance of physiologically relevant O 2 levels in cell culture in vitro has been reviewed [37][38][39][40][41][42]. Compared to cells cultured in standard incubators under atmospheric oxygen (~18 kPa O 2 ), most cells experience significantly lower O 2 levels in vivo [37][38][39][40], with the blood-dissolved O 2 gradient in coronary arteries ~5 kPa. We previously reported that long-term (5 days) adaptation of endothelial cells to physiological O 2 (5 kPa) increases the bioavailability of nitric oxide [43,44] and activity of sarco/endoplasmic reticulum Ca 2+ ATPase (SERCA), protecting cells against ionomycin-induced Ca 2+ overload [45]. Furthermore, we have shown that NRF2-regulated redox signaling is attenuated in umbilical vein and brain microvascular endothelial cells under physiological normoxia [46,47]. The present study correlates redox and metal profiles in human coronary artery endothelial cells (HCAEC) and smooth muscle cells (HCASMC) under hyperoxia (18 kPa O 2 ), physiological normoxia (5 kPa O 2 ) and hypoxia (1 kPa O 2 ), focusing on the effects of NRF2 on intracellular zinc. To our knowledge, our findings provide the first insight into cell type specific regulation of intracellular metals in coronary artery endothelial and smooth muscle cells adapted to different pericellular O 2 levels.

Intracellular O 2 levels in HCAEC and HCASMC measured using MitoXpress®-INTRA
Intracellular O 2 levels in live cells were monitored using MitoX-press®-INTRA (Agilent Technologies, USA), which is a cell-penetrating phosphorescent platinum-porphyrin based nanoparticle probe [46,47]. HCAEC and HCASMC were seeded into black clear bottomed 96-well plates and loaded with the nanoparticle probe (10 μg/ml) for 16 h in complete medium. Phosphorescence signals at 655 ± 55 nm after excitation at 355 ± 55 nm were measured after 30 μs (D1) and 70 μs (D2) with a 30 μs window and converted to probe lifetime (T) [T=(D2-D1)/ln (l w1 /l w2 )], where T represents emission lifetime and l w1 and l w2 represent signals measured at window 1 and window 2. Averaged lifetime values measured at 7 ambient O 2 tensions (18, 15, 10, 7.5, 5, 2.5, 1 kPa) were plotted against the known O 2 tension and subjected to an exponential fit analysis. Lifetime values were then converted to O 2 (kPa) by a curve based on the parameters of the fit. Lifetime values were then interpolated from this curve to give the dissolved intracellular O 2 level in HCAEC and HCASMC cytosol (see Fig. 2C and D). Dissolved O 2 in culture medium was also measured in parallel by diluting MitoX-press®-INTRA (2.5 μg/ml) in complete medium.

Oxygen consumption rate in HCAEC and HCASMC
Cells were adapted for 5 days to 18 or 5 kPa O 2 in an O 2 -controlled workstation and seeded into Thermo Nunc flat bottom 96-well microplates in standard culture medium for 48 h. After cells reached ~80% confluence under 18 or 5 kPa O 2 , the medium was changed and a Resipher oxygen sensing lid (Lucid Scientific Inc., USA) was attached [48]. Basal oxygen consumption rate (OCR) was measured over 10 h in cells under 18 or 5 kPa O 2 . Data were analyzed using the Resipher web application (Lucid Scientific) and OCR values expressed as fmol/mm 2 /s/μg protein (see Supplementary Fig. 1).

Metallomic profiling in cells adapted to 18, 5 or 1 kPa O 2 using ICP-MS and LA-ICP-MS
HCAEC and HCASMC lysates were collected in purified trace metal free water with a resistivity ≥18.2 MΩcm obtained from a Milli-Q system (Merck Millipore, USA). Lysis solutions were subjected to 3 freezethaw cycles and sonication before heating with nitric acid 65% Supra-pur® (Merck Millipore, USA) to 95 • C for 2 h. Subsequently, the lysis solution was cooled and diluted with purified ddH 2 O to a final concentration of 0.5% nitric acid (0.1 M). Quantification of total zinc (Zn), copper (Cu), manganese (Mn) and magnesium (Mg) concentrations in HCAEC and HCASMC lysates was conducted using a PerkinElmer NexION 350D Inductively Coupled Plasma Quadrupole Mass Spectrometer (ICP-QMS) under Dynamic Reaction Cell (DRC) and Kinetic Energy Discrimination (KED) modes [49]. Cell lysate samples were introduced to the ICP-QMS via a Cetac ASX-520 autosampler (Teledyne, Cetac, USA) coupled to a SeaSpray glass nebulizer fitted to a quartz cyclonic spray chamber. Measurements in counts per second were converted to a concentration by applying a regression model from calibration standards (multi-element standard solution VI, Sigma-Aldrich). Concentrations were then normalised to the amount of cells in the lysis solution based on the protein concentration.
Laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) was used to map Zn distribution in HCAEC and HCASMC cultured on 8-well glass removable chamber slides (Ibidi, Germany). For the LA-ICP-MS experiments, an Analyte Excite 193 nm ArF*excimerbased laser ablation system (Teledyne Photon Machines, USA) equipped with a HelEx II two-volume ablation cell was coupled to an iCAP TQ ICP-MS (Thermo Fisher Scientific, USA) via a Aerosol Rapid Introduction System (ARIS). Elemental images of cell monolayers were acquired under fixed dosage mode (10 laser shots per pixel) with a laser energy density of 0.8 j cm − 2 and a beam waist diameter of 2 μm (effective image resolution). ICP-MS measurement acquisition parameters were optimised based on cell monolayer washout times, which provided a dwell time of 36.8 ms for Zn under KED mode. Total cell content of metals was calculated using ImageJ software (National Institute of Health, USA).

Measurement of intracellular ATP and glutathione levels
HCAEC and HCASMC were adapted to 18, 5 or 1 kPa O 2 for 5 d and ATP and glutathione (GSH) extracted using 6.5% trichloroacetic acid (Sigma-Aldrich, UK). To measure ATP, cell extracts were incubated with firefly lantern extract (Sigma, UK) containing both luciferase and luciferin, while GSH levels were determined using a fluorometric assay [46,47,50]. Luminescence and fluorescence were measured in a plate reader (CLARIOstar, BMG Labtech, Germany) and expressed as nmol/mg protein.

siRNA Nrf2 silencing
HCAEC and HCASMC were adapted to 18 or 5 kPa O 2 for 5 d and then transfected with 15 pmol/well and 8.84 pmol/well respectively with either scrambled siRNA or Nrf2 siRNA (Santa Cruz, USA) for 24 h, using FuGENE (Promega, UK) or Dharmafect 1 transfection reagent (Thermo Scientific, UK) according to the manufacturer's instructions [47].

Statistical analysis
Data denote the mean ± S.E.M. of 3-6 different HCAEC or HCASMC cultures and were analyzed using Graphpad Prism 9. Significance was assessed using either an unpaired Student's t-test or one-or two-way ANOVA followed by a Bonferroni Post Hoc test where appropriate, with *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001 considered significant.

Intracellular O 2 levels in HCAEC and HCASMC during culture under physiological normoxia
We previously reported that NRF2 regulated antioxidant defense enzymes are upregulated during culture of human umbilical vein (HUVEC) and murine brain microvascular (bEnd.3) endothelial cells under 18 compared to 5 kPa O 2 [46,47]. As cellular responses to changes in pericellular O 2 levels are cell type specific (reviewed in Ref. [37]), we conducted real-time measurements of intracellular O 2 in HCAEC and HCASMC monolayers using the phosphorescent nanoparticle probe MitoXpress®-INTRA. Intracellular oxygen content was measured during stepwise reductions in ambient O 2 (18-1 kPa) in an O 2 -controlled, time-resolved fluorescent plate reader and calculated from phosphorescent lifetime measurements [46,47]. Intracellular O 2 levels of 4.31 ± 0.29 kPa and 3.98 ± 0.11 kPa were measured respectively in HCAEC and HCASMC cultured under 5 kPa O 2 ( Fig. 1A and H), recapitulating physiological O 2 levels measured in human coronary vasculature in vivo [37].

Effects of O 2 on HIF-1α stabilization and intracellular ATP and GSH in HCAEC and HCASMC
To determine whether 5 d culture under 5 or 1 kPa O 2 induces a hypoxic response in HCAEC and HCASMC, HIF-1α stabilization was examined by immunoblotting. As shown in Fig. 1D and K, increased HIF-1α expression was only detected in cells adapted to 1 kPa O 2 , confirming the absence of an hypoxic phenotype in cells cultured under 5 kPa O 2 . We next examined whether changes in pericellular O 2 levels affect intracellular ATP and total glutathione (GSH) levels. In HCAEC, ATP levels were only decreased under 1 kPa O 2 (Fig. 1B), whereas ATP levels in HCASMC were affected negligibly (Fig. 1I). Intracellular GSH levels were significantly lower in HCAEC adapted to 5 and 1 kPa compared to 18 kPa O 2 (Fig. 1C), consistent with our findings in bEnd.3 brain endothelial cells and airway epithelial cells [47,51]. Notably, basal GSH levels in HCASMC cultured under 18 kPa O 2 (13.76 ± 0.58 nmol/mg protein) were significantly lower than levels in HCAEC (50.36 ± 4.11 nmol/mg protein, Fig. 1C) and reduced further during culture under 1 kPa O 2 (Fig. 1J).

Oxygen dependent changes in antioxidant enzyme expression in HCAEC and HCASMC
Our previous studies indicated that cells cultured under physiological O 2 levels are under lower oxidative stress and therefore express decreased levels of antioxidant enzymes [46,47]. Many of these enzymes such as CuZnSOD (SOD1) and MnSOD (SOD2) have metal constituents that are key for their function with cellular levels potentially regulated by oxygen. We therefore examined expression of catalase, responsible for the decomposition of H 2 O 2 , and SOD1 and SOD2, responsible for dismutation of superoxide. Catalase expression in HCAEC was decreased under both 5 and 1 kPa O 2 compared to 18 kPa O 2 (Fig. 1E), whilst CuZnSOD expression was affected negligibly (Fig. 1F). MnSOD expression was significantly lower in HCAEC under 1 kPa O 2 compared to 5 kPa O 2 (Fig. 1G). By comparison, CuZnSOD and MnSOD expression decreased in HCASMC under 1 kPa O 2 compared to 18 kPa O 2 ( Fig. 1M  and N). Taken together these data indicate a reduction in antioxidant enzyme expression in cells adapted to lower redox stress.

Adaptation to defined O 2 levels alters intracellular metal content in a cell type specific manner
There is evidence that metal homeostasis, storage and channel expression are all affected by the oxygen environment and/or oxidative stress [52][53][54][55][56]. In particular, Zn has been reported to be protective against heart disease and ischemia-reperfusion. To determine whether metal homeostasis is influenced by the O 2 environment, HCAEC and HCASMC monolayers were adapted to 18, 5 and 1 kPa O 2 for 5 d and lysates analyzed by ICP-MS. In HCAEC, total Zn content decreased significantly between 18 and 1 kPa O 2 (18 kPa = 0.345 ± 0.056 ng/μg protein, 5 kPa = 0.267 ± 0.032 ng/μg protein, 1 kPa = 0.177 ± 0.020 ng/μg protein) ( Fig. 2A). However, no significant differences in Zn levels were detected in HCASMC under different O 2 levels (18 kPa = 0.345 ± 0.090 ng/μg protein, 5 kPa = 0.298 ± 0.020 ng/μg protein, 1 kPa = 0.441 ± 0.058 ng/μg protein) (Fig. 2I). Interestingly, total Cu levels decreased significantly between 18 and 1 kPa O 2 in HCAEC (Fig. 2B) but increased in HCASMC (Fig. 2J). There were no significant differences in total Mn content across the O 2 conditions tested in either cell type (Fig,  2C and K). Under 5 kPa O 2 , HCAEC showed a significant increase in total Mg compared to both 18 and 1 kPa O 2 (Fig. 2D), whilst HCASMC showed no difference (Fig. 2L). ICP-MS analysis of cell free culture media used for HCAEC and HCASMC revealed Zn concentrations of 2.0 μM and 2.8 μM, respectively (Table 1). Since the extracellular content of zinc in the culture media remained constant throughout experiments, our results indicate that changes in intracellular Zn levels are due to transport and not a passive reflection of the extracellular concentration.
We then employed LA-ICP-MS to map the spatial distribution of Zn in HCAEC (Fig. 2E) and HCASMC (Fig. 2M). LA-ICP-MS analysis showed that 66 Zn per cell trended to decrease in HCAEC as pericellular O 2 levels were lowered (Fig. 2F), supporting our ICP-MS data ( Fig. 2A). HCAEC demonstrated a concentrated distribution of zinc at the nucleus (Fig. 2  E). Whilst HCASMC also demonstrated a high nuclear localizaion, there was a stronger cytoplasmic signal compared to HCAEC at all O 2 levels (Fig. 2M), with analysis of 66 Zn per cell indicating a trend for a decrease in Zn content at 1 kPa O 2 (Fig. 2N). The implications of these findings is that pericellular O 2 levels may regulate the intracellular content of metals in a cell type specific manner, even in vascular cells isolated from the human coronary artery.

Zinc transporter ZnT1 and zinc-binding protein metallothionein (MT) expression correlate with intracellular zinc
Due to the importance of zinc in cardiovascular disease, we further interrogated the effects of O 2 on Zn homeostasis by investigating the expression of some of the key proteins involved in zinc storage and transport. ZnT1 is the main zinc channel on the cell membrane responsible for zinc efflux [57]. Notably, ZnT1 protein expression was significantly lower in HCAEC adapted to 18 kPa O 2 compared to 5 and 1 kPa O 2 (Fig. 2G). Unlike HCAEC, ZnT1 expression in HCASMC was affected negligibly by changes in pericellular O 2 levels (Fig. 2O). Metallothioneins are a family of cysteine-rich metal-binding proteins that bind a large proportion of the intracellular zinc pool [58,59]. Metallothionein (MT1/2) expression decreased as O 2 levels decreased and was significantly lower in HCAEC adapted to 1 kPa O 2 compared to 18 kPa O 2 (Fig. 2H). This suggests that whilst the intracellular zinc-binding pool (MT) is increased at 18 kPa O 2 , cellular efflux capacity may be reduced, resulting in an increased total intracellular zinc. However, this does not hold true for HCASMC, where changes in O 2 had no effect on total Zn content (Fig. 2N), although MT expression decreased significantly in cells adapted to 1 compared to 5 kPa O 2 (Fig. 2P).
Consistent with our previous findings in other cell types [46,47], NQO1 expression was reduced in HCAEC and HCASMC transfected with   (Fig. 3B and F). In HCASMC, unlike HCAEC, NRF2-siRNA resulted in a modest reduction in NQO1 expression, although another NRF2 targeted HO-1 was significantly downregulated (data not shown). When we investigated whether changes Zn content correlated with changes in ZnT1 and metallothionein (MT) expression in HCAEC ( Fig. 3C and D) and HCASMC ( Fig. 3G and H), NRF2-siRNA had no effect on ZnT1 or MT expression.

Effects of NRF2 overexpression on total intracellular Zn and NQO1 expression
Overexpression of NRF2 was achieved through transfection with a plasmid vector which constitutively expresses NRF2. Quantification of total Zn content in HCAEC by ICP-MS showed no differences in cells transfected with control vector (3.1C) or NRF2 expressing vector (hNRF2). However, overexpression of NRF2 at 5 kPa O 2 tended to increase total Zn to levels in HCAEC under 18 kPa O 2 (Fig. 4A). The NRF2 target NQO1 showed robust induction under 18 kPa O 2 , which was attenuated at 5 kPa O 2 (Fig. 4B), consistent with our findings in HUVEC [46] and brain microvascular endothelial cells [47]. In HCAEC adapted to 5 kPa O 2 , overexpression of NRF2 increased ZnT1 expression (Fig. 4C) but had a negligible effect on MT expression (Fig. 4D). Total intracellular Zn in HCASMC was affected negligibly by NRF2 overexpression (Fig. 4E). In HCASMC, NQO1 was upregulated by NRF2 overexpression at 18 kPa O 2 and attenuated at 5 kPa O 2 (Fig. 4F), whilst ZnT1 and MT expression were affected negligibly ( Fig. 4G and H).  Supplementary Fig. 3B).

Effects of sulforaphane on NRF2 activation and total Zn levels
We previously reported that NRF2 activated HO-1 expression is attenuated in human endothelial cells under 5 kPa O 2 due to an increase in expression of the repressor protein Bach1 [46]. When we probed Bach1 expression in HCAEC and HCASMC adapted to 18 or 5 kPa O 2 , basal and SFN induced Bach1 levels were only increased significantly in HCASMC under 5 kPa O 2 (see Supplementary Fig. 4B).

Discussion
In view of the importance of zinc in influencing redox dysregulation in vascular [29], cardiac [9,38,57,60] and other cell types [33,61,62], our study in human coronary endothelial and smooth muscle cells cultured long-term under defined pericellular O 2 levels has identified cell type specific differences in redox and metal profiles. In a recent expert recommendation, we emphasized the importance of recapitulating O 2 levels in cell culture that reflect O 2 levels in vivo. Adaptation of coronary endothelial and smooth muscle cells to 5 kPa O 2 in our O 2 -controlled workstations enabled us to measure intracellular O 2 levels that recapitulate measurements in vivo. Importantly, the absence of HIF-1α stabilization in both cell types under 5 kPa O 2 confirms that cells were not exposed to hypoxia, consistant with our previous studies in other endothelial cell types [43,44,46,47].
Although intracellular ATP levels in HCAEC were affected negligibly when lowering O 2 from 18 to 5 kPa, GSH content was significantly lower under 5 kPa O 2 , reflecting lower oxidative stress [47]. In HCASMC, lowering pericellular O 2 from 18 to 5 kPa had negligible effects on GSH, antioxidant enzymes or NRF2 activated NQO1 expression. In contrast, GSH, catalase and NRF2 induced NQO1 expression were decreased in HCAEC under 5 kPa O 2 , highlighting differences in their redox phenotype and sensitivity to changes in pericellular O 2 . Moreover, the phenotype of HCAEC and HCASMC under 1 kPa O 2 was characterized by lower intracellular GSH and antioxidant enzymes (e.g. catalase, SOD1 and SOD2), providing insights into the effects of long-term hypoxia on human coronary vascular cells. Despite enhanced expression of Bach1 in HCASMC under 5 kPa O 2 , these cells retained the ability to induce HO-1 expression most likey through non-NRF2 related transcription. In the context of cardiac function, inhibition of Bach1 may provide a novel target to protect the heart against pressure overload [63].
Endothelial cells synthesize ATP primarily via glycolysis with a low rate of O 2 consumption [64]. Their relatively low mitochondrial content, low energy demand and high glycolytic activity enables endothelial cells to transfer more oxygen to perivascular cells (reviewed in Ref. [65]). Although the O 2 demand of coronary endothelial cells is relatively low, glucose deprivation and hypoxia semstize these cells to changes in pericellular O 2 levels [66]. In vascular smooth muscle cells, the rate of O 2 consumption and lactate production are nearly equal on a molar basis under resting conditions, with ~30% of the ATP supply from aerobic glycolysis and at least 90% of the flux via glycolysis converted to lactate [67]. Differences in metabolism may in part account for the differential sensitivity of HCAEC and HCASMC to changes in pericellular O 2 . Although basal oxygen consumption rate (OCR) was similar in HCAEC and HCASMC (see Supplementary Fig. 1), Sekine et al. reported marked differences in OCR between cardiomyocytes and other coronary cell types [68]. The high OCR in human iPSC cardiomyocytes was attributed to their large quantity of mitochondria and energy demand for spontaneous contraction [68]. In this context, a recent transcriptomic analysis of different human cancer cell lines cultured for 14 days under 18 versus 5 kPa O 2 established that changes in O 2 levels affect the expression of mitochondrial DNA-encoded genes in a highly cell type specific manner [69].
To our knowledge, the current study is the first to investigate the effects of controlled pericellular O 2 levels on metal profiles in coronary artery endothelial and smooth muscle cells during long-term culture in vitro. Notably, total intracellular Cu content shows an opposite trend in HCAEC and HCASMC as O 2 levels decreased from 18 to 1 kPa. The significant increase in Cu content in HCASMC under 1 kPa O 2 may be associated with accelerated atherogenesis, as Heinecke et al. have shown that Cu supplementation in culture medium modifies low density lipoprotein (LDL) in human aortic smooth muscle cells, resulting in cholesteryl ester accumulation in macrophages [70]. Morevoer, HIF-1α mediated upregulation of mRNA/protein expression of Cu transposter 1 (CRT1) leads to increased Cu uptake in human pulmonary artery smooth muscle cells exposed to hypoxia, with elevated intracellular Cu enhancing proliferation and migration during hypoxia [71].
In recent years, an increasing number of studies have focused on the effects of hypoxia/reoxygenation or ischemia/reperfusion on intracellular Zn [4,72]. Notably, the majority of studies in vitro have compared the effects of hypoxia versus atmospheric O 2 , which is well known to Fig. 4. Effects of NRF2 overexpression on total Zn levels and NQO1, ZnT1 and metallothionein expression in HCAEC and HCASMC adapted to 18 kPa or 5 kPa O 2 HCAEC and HCASMC were adapted for 5 d to 18 or 5 kPa O 2 and then transfected with control vector pcDNA3.1 (3.1C) or NRF2 overexpression vector pcDNA3.1-hNrf2 (hNRF2) and cell lysates harvested after 24 h. A and E, Total Zn levels in HCAEC and HCASMC were measured by ICP-MS. B-D and F-H, Cell lysates were immunoblotted for NQO1, ZnT1 and MT expression relative to β-actin and analyzed by densitometry. Data denote mean ± S.E.M., n = 4-6 independent cell cultures (color coded), two-way ANOVA followed by a Bonferroni Post Hoc test analysis, *P < 0.05, **P < 0.01, ***P < 0.001. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) enhance oxidative stress and reactive oxygen species generation in cultured cells [37]. Thus, we felt it important to measure total Zn content in coronary artery endothelial and smooth muscle cells adapted long-term to physiological normoxia (5 kPa O 2 ) before examining the relationship between Zn and NRF2 regulated redox signaling. We have obtained the first comparative measurements of total Zn content in HCAEC and HCASMC under 18, 5 or 1 kPa O 2 and established that O 2 levels affect Zn content in HCAEC but not in HCASMC. This cell type specific sensitivity to pericellular O 2 mediated changes in Zn content may reflect differences in the expression/activity of Zn transporters and/or binding proteins [73], but also the differing redox sensitivity of HCAEC and HCASMC to oxygen tension. In agreement, studies have shown that zinc modulation alters cardiac proteostasis [12].
To further invesitigate effects of NRF2 signaling on Zn content, HCAEC and HCASMC were transfected with NRF2-siRNA or NRF2 overexpression vector, or treated with the NRF2 inducer sulforaphane. Together our findings suggest that NRF2 activation/overexpression tends to increases total Zn content in HCAEC with negligible changes detected in HCASMC, consistent with the observed changes in ZnT1 and MT protein expression. In a previous study in HCAEC, downregulation of ZnT1 expression under depleted/low and high Zn conditions was attributed to a reduced turnover of new ZnT proteins [74], whilst in HUVEC low or oscillatory shear stress led to an upregulation ZnT1 and MT and lower intracellular free Zn [75]. As endothelial nitric oxide (NO) is decreased under oscillatory shear stress [76], Conway et al. suggested that reduced NO generation in atheroprone regions together with increased ZnT1 and MT expression could account for decreased intracellular free zinc [75].
The physiological concentration range of Zn is quite narrow and is strictly regulated by uptake, storage and secretion, mainly mediated by ZnT1-10 and ZIP1-14 and MT [21,77,78]. Notably, Tran et al. have shown that Zn transporters are located differentially among endothelial and smooth muscle cells in human subcutaneous microvessels [73]. ZIP10 and 14 are approximately equally expressed in both cell types, while ZIP1, 2, 8 and 12 are relatively more abundant in endothelial cells and ZIP14 is more abundant in smooth muscle cells [73]. Activation of NRF2 in HepG2 cells significantly increases mRNA levels of ZnT1, 3 and 6 and decreases ZnT10 and ZIP3 mRNA levels [31].
Transcriptional activation of metal regulatory transcription factor 1 (MTF-1) and NRF2 by oxidants is linked via a pool of available zinc controlled by MTF-1 [19]. Oxidation of Keap1 cysteine residues releases zinc, leading to nuclear accumulation of NRF2 and gene transcription, but it remains to be determined whether zinc release from Keap1 leads to transcriptional activation of MTF-1 [79]. Although in endothelial cells zinc activates NRF2 regulated glutamate-cysteine ligase and glutathione synthesis independent of MTF-1 [29], MTF-1 does play a role in detoxifying heavy metals and protection of cells against hypoxia and oxidative stress.
NRF2 and zinc influence the dynamic balance of cellular redox homeostasis, as evidenced in the present study in coronary vascular cells and in other cell types [31][32][33]. We have identified differences in the redox phenotype and metal profile in human coronary endothelial and smooth muscle cells adapted long-term to hyperoxia (standard cell culture), physiological normoxia ('physioxia') or hypoxia. Our findings further emphasize the importance assessing intracellular redox signaling and changes in total and labile zinc levels in cells under pericellular O 2 levels encountered in vivo. To further recapitulate the physiological environment of coronary endothelial and smooth muscle cells, we encourage researchers not only to mimic O 2 levels in vivo but to also investigate the effects of laminar shear stress and/or mechanical stress, noting that gene expression of metallothionein isoforms and ZnT1 are upregulated by low and oscillatory shear stress [75].

Declaration of competing interest
The authors declare that they have no conflicts of interest that could have influenced the study.

Declaration of interest statement
Authors declare no conflicts of interest in this study.

Data availability
No data was used for the research described in the article.