Electron Leak From the Mitochondrial Electron Transport Chain Complex I at Site I_Q Is Crucial for Oxygen Sensing in Rabbit and Human Ductus Arteriosus

Abstract

Background

As partial pressure of oxygen (pO₂) rises with the first breath, the ductus arteriosus (DA) constricts, diverting blood flow to the pulmonary circulation. The DA's O₂ sensor resides within smooth muscle cells. The DA smooth muscle cells’ mitochondrial electron transport chain (ETC) produces reactive oxygen species (ROS) in proportion to oxygen tension, causing vasoconstriction by regulating redox‐sensitive ion channels and enzymes. To identify which ETC complex contributes most to DA O₂ sensing and determine whether ROS mediate O₂ sensing independent of metabolism, we used electron leak suppressors, S1QEL (suppressor of site I_Q electron leak) and S3QEL (suppressor of site III_Qo electron leak), which decrease ROS production by inhibiting electron leak from quinone sites I_Q and III_Qo, respectively.

Methods and Results

The effects of S1QEL, S3QEL, and ETC inhibitors (rotenone and antimycin A) on DA tone, mitochondrial metabolism, O₂‐induced changes in intracellular calcium, and ROS were studied in rabbit DA rings, and human and rabbit DA smooth muscle cells. S1QEL's effects on DA patency were assessed in rabbit kits, using micro computed tomography. In DA rings, S1QEL, but not S3QEL, reversed O₂‐induced constriction (P=0.0034) without reducing phenylephrine‐induced constriction. S1QEL did not inhibit mitochondrial metabolism or ETC‐I activity. In human DA smooth muscle cells, S1QEL and rotenone inhibited O₂‐induced increases in intracellular calcium (P=0.02 and 0.001, respectively), a surrogate for DA constriction. S1QEL inhibited O₂‐induced ROS generation (P=0.02). In vivo, S1QEL prevented O₂‐induced DA closure (P<0.0001).

Conclusions

S1QEL, but not S3QEL, inhibited O₂‐induced rises in ROS and DA constriction ex vivo and in vivo. DA O₂ sensing relies on pO₂‐dependent changes in electron leak at site I_Q in ETC‐I, independent of metabolism. S1QEL offers a therapeutic means to maintain DA patency.

Nonstandard Abbreviations and Acronyms

DA

ductus arteriosus

DASMC

ductus arteriosus smooth muscle cell

ECAR

extracellular acidification rate

OCR

oxygen consumption rate

S1QEL

suppressor of site I_Q electron leak

S3QEL

suppressor of site III_Qo electron leak

Research Perspective.

What Is New?

• Electron leak and partial pressure of oxygen‐dependent changes in reactive oxygen species production by complex I, but not complex III, in ductus arteriosus smooth muscle cells are critical to O₂ sensing in term rabbits and humans.

• The complex I suppressor of site I_Q electron leak (S1QEL) mimics hypoxia and maintains key mitochondrial metabolic parameters while inhibiting O₂‐induced ductus arteriosus constriction in vitro in ductus arteriosus smooth muscle cells and in vivo in fetal rabbits.

What Question Should Be Addressed Next?

• S1QEL is a potential novel therapeutic agent for maintaining ductus arteriosus patency in vivo, although further studies are required to assess the safety and efficacy of S1QEL or similar drugs in humans as an alternative to prostaglandin E for maintenance of ductus arteriosus patency in infants with congenital heart disease.

The ductus arteriosus (DA) is a vital fetal vessel that shunts placentally oxygenated blood from the pulmonary artery to the aorta, bypassing the unventilated fetal lung (Figure 1A). In utero, fetal partial pressure of oxygen (pO₂) is low (≈40 mm Hg), maintaining the DA in a state of hypoxic vasodilation and eliciting hypoxic pulmonary vasoconstriction. ¹ In utero, the patent DA diverts blood away from the unventilated pulmonary circulation to the aorta, and only ≈10% of the cardiac output passes through the lungs, the majority transiting the DA into the descending aorta. ² At birth, the circulatory system must quickly adapt to obtaining O₂ and eliminating carbon dioxide through respiration, rather than relying on the fetoplacental circulation. With the first breath, the DA and pulmonary artery experiences a similar rise in pO₂; however, O₂ constricts the DA, ³ rapidly eliciting functional DA closure, ⁴ while causing the pulmonary artery to relax, increasing lung perfusion (Figure 1B).

Figure 1. Contrasting the normal changes in cardiopulmonary circulation with closure of the ductus arteriosus (DA) at the transition from fetal to neonatal life with the presence of a patent DA.

A, In the fetus, hypoxia maintains DA patency and promotes hypoxic pulmonary vasoconstriction, causing blood exiting the right ventricle to transit the DA and enter the descending aorta (Ao). B, With the first breath, a rise in the partial pressure of oxygen (pO₂) constricts the DA, rapidly eliciting functional closure while simultaneously causing the pulmonary artery (PA) to relax, increasing lung perfusion. C, In the presence of a patent DA, the DA can permit either left to right shunting of oxygenated blood to the PAs, causing heart failure (illustrated), or right‐to‐left shunting of deoxygenated blood to the aorta, causing cyanosis (not shown).

Functional closure of the ductus typically occurs within the first 48 hours of life in full‐term infants ⁵ and is mediated by DA constriction, reflecting contraction of DA smooth muscle cells (DASMCs). This is followed within days by anatomical closure, which creates a fibrous ligamentum arteriosum. ⁶ Patent DA (PDA) (Figure 1C) remains an important clinical problem and accounts for ≈5% to 10% of all congenital heart disease. ² Although the incidence of PDA is low for full‐term infants (57 per 100 000), 55% of infants who weigh <1000 g have PDA, ⁷ , ⁸ and ≈70% of affected preterm infants require either medical or surgical treatment to close their PDA. ⁹ , ¹⁰

DA constriction to O₂ is modulated by endothelial‐derived vasodilators, such as prostaglandin E, ¹¹ and vasoconstrictors, such as endothelin ¹² ; however, human DAs constrict to O₂ in the absence of endothelium ¹³ and despite blockade of the endothelin pathway. ¹⁴

The core of the DA's O₂‐sensor mechanism resides in DASMCs. ¹³ , ¹⁴ When pO₂ rises, this activates dynamin‐related protein 1 leading to mitochondrial fission, ¹⁵ which activates complex I and increases mitochondrial reactive oxygen species (ROS) production. ¹⁵ Mitochondrial‐derived ROS, notably H₂O₂, inhibit redox‐sensitive voltage‐gated potassium (Kv) channels ¹⁶ , ¹⁷ , ¹⁸ and activate redox‐sensitive enzymes, which promote vasoconstriction and DA closure, such as rho kinase ¹⁶ , ¹⁸ and the epidermal growth factor receptor/p38 mitogen‐activated protein kinase/c‐Jun N‐amino‐terminal kinase (EGFR/p38/JNK) pathway. ¹⁹ Kv channel inhibition depolarizes the DASMCs' membrane potential and activates voltage‐gated calcium channels, thereby increasing cytosolic calcium, cytosolic calcium concentration [Ca²⁺]_i, and triggering DA constriction. The effects of hypoxia on DASMC Kv currents (activation and hyperpolarization) and, [Ca²⁺]_i (lowering) relax DA vascular tone. Both the ionic and hemodynamic effects of hypoxia are mimicked by classical electron transport chain (ETC) inhibitors, like rotenone and antimycin A, ¹⁶ and by reducing agents, ²⁰ indicating a role for the mitochondrial ETC in DA O₂ sensing. However, a precise understanding of the source of ROS within the ETC is lacking. Prior studies that relied on ETC inhibitors like rotenone and antimycin A are confounded by the possibility that these inhibitors may inhibit mitochondrial metabolism.

In this study, we used complex‐specific electron leak suppressors to (1) identify which ETC complexes mediate DA O₂ sensing, and (2) determine whether changes in ROS, rather than metabolic inhibition, determine changes in DA tone. S1QEL (suppressor of site I_Q electron leak) and S3QEL (suppressor of site III_Qo electron leak) are small molecules that prevent superoxide production at specific sites within ETC complexes I and III, respectively. ²¹ , ²² S1QEL and S3QEL selectively block electron leak by inhibiting formation of superoxide anions from a single Q site in each megacomplex, without blocking forward electron flow, disturbing mitochondrial membrane potential (∆Ψm), or inhibiting oxidative phosphorylation. S1QEL binds the complex I quinone binding site (I_Q), whereas S3QEL binds the quinone binding site in complex III (III_Qo). ²² Thus, unlike rotenone and antimycin A, electron leak suppressors have the potential to not only identify which ETC complex serves as a sensor but also to reveal the relative importance of changes in ETC‐derived ROS production versus changes in mitochondrial metabolism in DA O₂ sensing. This study of DA O₂ sensing strives to improve our mechanistic understanding of O₂‐induced DA closure and offers a novel, preclinical, proof of concept therapeutic strategy for congenital heart diseases that require maintenance of DA patency.

Methods

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Experimental Animals

All animal experiments conducted at Queen's University were done in accordance with the Canadian Council on Animal Care regulations and were approved by the Queen's University Animal Care Committee (protocol number 2017–1715). All researchers were blinded to sex for in vivo and in vitro studies, with kits randomly assigned to treatment conditions before postmortem sex determination. No animals were excluded based on sex or response, with the only exclusions based on technical issues such as tearing of the DA ring before mounting in the physiologic bath. For in vitro studies, there were 5 male and 3 female rabbit DASMC lines used, and 1 female and 4 male human DASMC lines used (as shown in Figure S1–S2.). Both male and female DAs were used to assess responses in the physiologic ring bath; however, DAs were randomized before sex determination resulting in some treatments having unequal female‐to‐male distributions: DMSO (3:3), S1QEL (7:2), rotenone (3:2), antimycin A (3:3), and S3QEL (1:3).

Human DA Tissues

DA samples from humans were harvested during congenital heart surgery at either the University of Chicago (institutional review board number A3523‐01) or the University of Nebraska (institutional review board number 100‐11‐EP), after ethics approval from each institution. Based on our ethics approval, which required fully anonymization, we were not able to acquire detailed demographic information on these infants. Ethics approval from Queen's University Health Sciences and Affiliated Teaching Hospitals Research Ethics Board (Tools for Research at Queen's, number 6007784) was obtained before the use of human DASMC (hDASMC) lines for ongoing research.

Tension Measurements in Term Rabbit DA Rings

Pregnant New Zealand White rabbits at 29 days gestation (term=31 days, n=11 pregnant rabbits; n=29 kits as donors of individual DA rings) were anesthetized with 10 mg/kg ketamine and 0.2 mg/kg medetomidine and maintained under anesthesia with 2% to 4% isoflurane. Rabbit kits were delivered via cesarean section, and a midline sternotomy was performed immediately, before the onset of respiration. The heart and lungs were excised en bloc and placed in hypoxic Krebs solution (≈40 mm Hg pO₂, pH 7.35–7.45). The DA was dissected free from adventitia under a Scribolux ×2.8 magnifier (Eschenbach, Danbury, CT) and severed at its connections with the pulmonary trunk and the descending thoracic aorta. Within ≈10 minutes of harvesting the DAs, vessels were mounted on a force transducer, using titanium wires, at an experimentally determined, optimal resting tension of ≈800 mg, in hypoxic Krebs solution. Meclofenamate (10 μmol/L, M4531; Sigma‐Aldrich, Oakville, ON, Canada), an inhibitor of prostaglandin synthase, and L‐Nitro arginine methyl ester (100 μmol/L, N5751; Sigma‐Aldrich), an inhibitor of nitric oxide synthase, were present in all ring bath experiments. After equilibration for 30 minutes, DA rings were exposed to hypoxic or normoxic pO₂ levels (35 and 155 mm Hg, respectively). At peak O₂‐induced constrictions, tissues were treated with either ETC leak suppressors, S1QEL (SML1948; Sigma‐Aldrich) or S3QEL (SML1554; Sigma‐Aldrich) (1–100 μmol/L in DMSO), ETC complex inhibitors, rotenone (10 μmol/L, R8875; Sigma‐Aldrich) or antimycin A (10 μmol/L, A8674; Sigma‐Aldrich), or a vehicle control (DMSO, D8418; Sigma‐Aldrich). The doses of each drug were determined in preliminary experiments to establish doses that had effects on DA tone greater than the respective vehicle, or were based on previous research. ¹⁶ At the end of each experiment, 80 mmol/L KCl (P3911; Sigma‐Aldrich) was administered during hypoxia, to establish the maximal, O₂‐independent constrictor capacity of the ring. To determine the specificity of the effects of electron leak suppressors for O₂‐induced constriction, S1QEL or S3QEL was added to the bath before challenge with phenylephrine (P6126; Sigma‐Aldrich) or KCl.

Isolation of DASMCs

To establish rabbit and hDASMC lines, DA tissues were minced using a razor blade. Minced tissue was placed in 50 mL Falcon tubes containing 2 mL of warm digestion buffer (Roswell Park Memorial Institute media with 2% FBS, 13 WU/mL Liberase, and 100 mg/mL DNase). Falcon tubes were vortexed for 20 seconds before incubation for 15 minutes at 37 °C in an orbital shaker. Digested tissues were then filtered through a wet 70‐μm cell strainer, and then the filtered cell suspension was washed with 5 mL PBS. Cells were centrifuged at 1200 rpm for 7 minutes at 21 °C. After centrifugation, the cell pellet was suspended in 2 mL hypoxic M231 smooth muscle cell (SMC) growth media (supplemented with 5% smooth muscle growth supplement, 10% FBS, 1% L‐glutamine, 1% penicillin/streptomycin, and 10 μg/mL ciprofloxacin HCl). All cell culture reagents were obtained from Gibco (Carlsbad, CA). Cells were plated and cultured at 37 °C in a hypoxic incubator (2.5% O₂ [40 mm Hg pO₂], 5% CO₂, balance N₂).

Establishing Purified DASMC Populations

To obtain a colony of purified DASMCs, negative fluorescence‐activated cell sorting experiments were performed on cells collected from DA tissues to exclude fibroblasts (CD90⁺) and endothelial cells (CD31⁺). Cultured cells were trypsinized (0.25% Trypsin–EDTA, ThermoFisher Scientific, Mississauga, ON, Canada), washed with PBS, and resuspended in 1 mL of hypoxic M231 media with 4% FBS. After counting, the cell suspensions were aliquoted into 4 Eppendorf tubes, 2 for single staining using phycoerythrin anti‐CD90/Thy1 (MRC OX‐7, ab33694; Abcam, Toronto, ON, Canada; 5 μL per 10⁶ cells), and APC (adenomatous polyposis coli) anti‐CD31 (clone 390; catalog number 10243; BioLegend, San Diego, CA; 1.25 μL per 10⁶ cells), 1 for double staining using both antibodies, and 1 to serve as an unstained control. Then, all Eppendorf tubes were vortexed and incubated for 30 minutes at 37 °C in a hypoxic incubator. Cells were then centrifuged to wash away excess antibodies, and the pellet was resuspended in 1 mL PBS. Cells were sorted using the SH800S cell sorter (Sony Biotechnology, San Jose, CA) to collect only CD90⁻ CD31⁻ cells. Unstained and single‐stained controls were used to optimize sorting and compensation parameters. Cells were first gated based on forward scatter and side scatter to exclude debris. Cells were then negatively sorted using antibodies against CD90 and CD31, to exclude fibroblasts and endothelial cells, respectively. The resulting pure SMC population was collected in hypoxic M231 supplemented media and cultured at 37 °C in a hypoxic incubator. Finally, cells were confirmed to be SMC positive by staining using Alexa‐Fluor 488 anti‐α‐smooth muscle actin (clone 1A4, catalog number 53–9760‐82; ThermoFisher Scientific).

Measuring Cytosolic Ca²⁺ Concentration in hDASMCs

hDASMCs were plated at an experimentally determined, optimal density for fluorescence imaging in 35‐mm glass bottom dishes (number 1.5 uncoated $\mathrm{\gamma }$‐irradiatred, P35G‐1.5‐14‐C; MatTek, Ashland, MA), and all imaging was performed in an O₂‐sensing buffer (modified Krebs solution: 115 mmol/L NaCl, 5.4 mmol/L KCl, 1 mmol/L MgCl₂, 2 mmol/L CaCl₂, 25 mmol/L NaHCO₃, 10 mmol/L HEPES, 10 mmol/L D‐glucose). The next day, cells were loaded with 10 μmol/L of Cal‐520 AM dye (ab171868; Abcam) in the presence of 0.02% Pluronic F‐127 (P6866; ThermoFisher Scientific). Cal‐520 AM was incubated with the cells for 30 minutes at 37 °C in a hypoxic incubator, with either DMSO, 50 μmol/L S1QEL, 10 μmol/L rotenone, or 50 μmol/L MitoTEMPO. After loading, cells were imaged using an OkoLab stage‐top microscope incubator (OkoLab Bold Line, Pozzuoli, Italy), and imaging was performed using a Leica TCS SP8 X confocal microscope (excitation 491 nm, emission 498–590 nm, 1.3 frames/s; Leica Microsystems, Wetzlar, Germany). Cells were allowed to equilibrate in hypoxia (2.5% O₂, 5% CO₂, balance N₂) for 10 minutes, before imaging for 15 minutes in hypoxia, followed by 15 minutes of normoxia (18% O₂, 5% CO₂, balance N₂). Images were obtained using LAS‐X software (Leica), and representative images made using Fiji (ImageJ). For each treatment group, regions of interest (ROI) were drawn around cells to track changes in Cal‐520 fluorescence over time. To compare changes in Cal‐520 fluorescence induced by increased O₂ content, ROI intensity over the last 10 images of normoxia were compared with the ROI intensity during the initial 10 hypoxia images.

Mitochondrial Metabolism Micropolarimetry Measurements

O₂ consumption rate (OCR) and extracellular acidification rate (ECAR), a measure of glycolysis, were measured simultaneously using an XFe24 extracellular flux analyzer (Agilent, Santa Clara, CA), as previously described. ²³ The day before performing the experiment, rabbit DASMCs (rbDASMCs) were plated at an experimentally determined, optimal density in a 24‐well XFe24 cell culture microplate (Seahorse Bioscience), and an extracellular FluxPak (Seahorse Bioscience) was hydrated by loading 1 mL of XF calibrant solution (Seahorse Bioscience) into each well in the utility plate and incubating at 37 °C overnight in a CO₂‐free incubator. The next day, measurements were taken after equilibration in XF assay media (Seahorse Bioscience) supplemented with 4.5 g/L glucose, with pH adjusted to 7.4 at 37 °C. To measure changes in OCR and ECAR, S1QEL, S3QEL, rotenone or vehicle control (DMSO) was added to the injection port A of the FluxPak. Oligomycin (O4876; Sigma‐Aldrich), FCCP, (C2920; Sigma‐Aldrich), and rotenone in combination with antimycin A were added to ports B, C, and D of the micropolarimeter, respectively. After FluxPak equilibration, the hydrating cartridge was replaced with the cell culture microplate. For each phase of the experiment (baseline, intervention, oligomycin, FCCP, and rotenone/antimycin A), 3 measurement cycles were performed, each consisting of a 3‐minute mixing period, a 2‐minute incubation period, and then a 3‐minute measurement period. For all experiments, cells were challenged with 1 μmol/L oligomycin (to inhibit ATP synthase), 1 μmol/L FCCP (an uncoupling agent that maximizes OCR), and 1 μmol/L of rotenone/antimycin A (to inhibit mitochondrial respiration). All OCR values were normalized to total protein content in each well, based on protein quantification using the Pierce BCA Protein Assay Kit (23227; ThermoFisher Scientific).

Complex I Activity Measurements

Complex I activity was quantified using the Complex I Enzyme Activity Dipstick Assay (ab109720;Abcam). rbDASMCs were collected 3 hours after incubation with 50 μmol/L S1QEL or vehicle (DMSO). After incubation with extraction buffer and centrifugation, protein concentration of supernatants was quantified using the Pierce BCA Protein Assay, and 10 μg of protein/condition was used on each dipstick. Dipsticks were imaged using the ChemiDoc MP Imaging System (Bio‐Rad, Mississauga, ON, Canada) and analyzed using ImageJ.

Measuring ROS Production in hDASMCs

hDASMCs were plated at an experimentally determined, optimal density for fluorescence imaging in a μ‐Slide 8 Well Glass Bottom plate (part number 80827; Ibidi, Fitchburg, WI). The next day, cells were loaded with 10 μmol/L of MitoROS 580 dye using the Mitochondrial Superoxide Detection Kit (ab219943; Abcam). MitoROS was incubated with the cells for 30 minutes at 37 °C in a hypoxic incubator, in the presence of either DMSO, 50 μmol/L S1QEL, 100 μmol/L S3QEL, or 10 μmol/L rotenone. After loading, cell chambers were placed in an OkoLab stage‐top microscope incubator, and imaging was performed using a Leica TCS SP8 X confocal microscope (excitation 540 nm, emission 570–720 nm, 2.5 frames/second, Leica Microsystems). Cells were allowed to equilibrate in hypoxia for 10 minutes, before imaging for 10 minutes in hypoxia followed by 10 minutes of normoxia. Images were obtained using LAS‐X software (Leica) and representative images made using Fiji (ImageJ). For each treatment group, ROI were drawn around cells to track changes in MitoROS fluorescence over time. To compare changes in MitoROS fluorescence induced by increased O₂ content, ROI intensity over the last 4 images of normoxic incubation were compared with the initial 4 images taken during hypoxic incubation.

MitoROS Plate Assay

hDASMCs were plated at an experimentally determined, optimal density in a black‐bottomed 96‐well plate in hypoxia the night before the experiment. The mitochondrial superoxide detection kit (ab219943; Abcam) with the MitoROS 580 dye was used. Cells were treated with DMSO, 50 μmol/L S1QEL, or 50 μmol/L MitoTEMPO combined with MitoROS working solution (prepared as per manufacturer's guide), 80 μL working solution and 20 μL effectors for a total volume of 100 μL per well. Plates were incubated with dye and effectors for 1 hour in hypoxia or normoxia at 37°C in the dark. Fluorescence was measured using SpectraMax M3 (Molecular Devices, San Jose, CA), with excitation 540 and emission 590. Fluorescence intensity was normalized to wells with dye only (no cells).

Micro‐Computed Tomography Imaging of DA in Rabbit Kits

Pregnant New Zealand White rabbits at 29 days gestation (n=7) were anesthetized with 2 mg/kg alfaxalone, and anesthesia was maintained with 2% to 4% isoflurane, and kits (n=15) were delivered via cesarean section. Rabbit kits were imaged by micro‐computed tomography (CT) under 3 conditions: hypoxia (group A), normoxia (group B), and normoxia + S1QEL (group C). Upon removal from the amniotic sac, kits were either immediately anesthetized with sodium pentobarbital (group A), before their first breath, or placed in a 100% O₂ chamber for 20 minutes where they were allowed to breathe spontaneously, before being anesthetized with sodium pentobarbital (groups B and C). Rabbit kits were placed in the lateral decubitus position, and a left thoracotomy was performed to prevent bleeding via the umbilical vasculature. A right femoral arteriotomy was also performed to permit the exit of blood and perfusion solution. Three luer lock syringes were filled with heparinized PBS (10 units, 2 mL), 4% PFA (8 mL), and the contrast agent iohexol 300 mgI/mL (Omnipaque with 3% gelatin, 8 mL). Syringes were capped and kept at 37 °C before use. A 25‐gauge butterfly catheter was inserted in the beating left ventricle (≈1.5 mm). The catheter was then connected to the 3 solutions described above. Contrast infusion was performed manually at ≈2 mL/min, sequentially as follows: heparinized PBS, 4% PFA, and finally Omnipaque with 3% gelatin. For kits in group C, 250 μL S1QEL (≈100 μmol/L) was injected ≈2 minutes before heparinized PBS infusion. After perfusion, rabbit kits were positioned on ice with their heads elevated at a 45° angle to the table to prevent leak of the contrast agent. Micro‐CT images were acquired using a VECTor ⁴ CT preclinical scanner equipped with a cone‐beam x‐ray CT system (MILabs B.V., Utrecht, the Netherlands). Percent change in DA diameter was calculated by comparing the diameter of the DA at its midpoint to its diameter at its connection with the aorta. The x‐ray source rotates around a fixed bed allowing the rabbit kit to be kept in the horizontal position in the scanner. Imaging was performed at an acceleration voltage of 50 kVp and an x‐ray tube current of 430 μm in an ultra‐focused mode that took 3600 projections over a 360° scan with an exposure time of 40 ms. The tomographic CT images were reconstructed using MILabs' reconstruction software to generate a 3‐dimensional CT image. Reconstructed slice data and maximum intensity projection images were processed using PMOD 3.9 software (PMOD Technologies, Zurich, Switzerland).

Statistical Analysis

Values are expressed as the mean±SEM. For normally distributed data, either a Student t test or 1‐way ANOVA was used to assess differences between groups. For nonparametric data, the Kruskal‐Wallis test was performed. P values <0.05 were considered statistically significant.

Results

Tension Measurements in Rabbit DA Rings

Normoxia caused an average 1100±62 mg increase in DA tension. Exposure to S1QEL (10 μmol/L, 50 μmol/L, and 100 μmol/L) caused concentration‐dependent decreases in DA tension (Figure 2A). In contrast, equimolar caused no significant change in DA tension during normoxia (Figure 2B). S1QEL (50 μmol/L) caused a significant reduction in DA tension (−25.3±17.6%, P<0.01, n=8). Likewise, 10 μmol/L of the complex III inhibitor antimycin A reduced DA tension (−20.1±16.2%, P<0.05, n=7). In contrast, S3QEL (100 μmol/L) and vehicle (DMSO) had minimal effect on DA tension, whereas the complex I inhibitor rotenone (10 μmol/L) demonstrated a statistically nonsignificant DA relaxation (Figure 2C). S1QEL (50 μmol/L) did not inhibit phenylephrine (10 μmol/L)‐induced constriction (Figure 2D). The phenylephrine induced constriction was of similar magnitude as normoxia‐induced constriction (Figure 2E). Thus, electron leak from complex I site I_Q, but not complex III site III_Qo, is implicated in DA O₂ sensing. Moreover, S1QEL's vasodilatory effects are specific for O₂‐induced DA constriction.

Figure 2. S1QEL (suppressor of site I_Q electron leak) reverses ductus arteriosus (DA) constriction ex vivo.

A and B, DA tension was measured using a force transducer, and constriction was induced by switching from hypoxia (≈35 mm Hg partial presure of oxygen [pO₂]) to normoxia (≈155 mm Hg pO₂). Dose responses of constricted DA to 10 μmol/L, 50 μmol/L, and 100 μmol/L S1QEL (suppressor of site I_Q electron leak) (A) and S3QEL (suppressor of site III_Q electron leak) (B) are shown. C, Percent reduction in DA tension was quantified 5 minutes after the addition of vehicle (DMSO, n=6), S1QEL (50 μmol/L, n=8), S3QEL (100 μmol/L, n=4), rotenone (10 μmol/L, n=5), or antimycin A (10 μmol/L, n=7). D, DAs in hypoxia were exposed to phenylephrine (PE; 10 μmol/L) to induce constriction (white bars). After relaxing fully, DAs were exposed to vehicle (DMSO, n=3) or S1QEL (50 μmol/L, n=5) before a second PE challenge (gray bars). E, DAs in hypoxia were either exposed to PE (10 μmol/L, n=8) or normoxia (n=11). Analysis was performed using 1‐way ANOVA followed by post hoc Dunn test (C). *P<0.05, **P<0.01.

Isolation and Purification of DASMCs

After digestion of DA tissues, colonies of rbDASMCs and hDASMCs were grown and purified using negative fluorescence‐activated cell sorting, to exclude CD90⁺ and CD31⁺ cells (Figure S1A–D). Isolated rbDASMC and hDASMC colonies were confirmed to be >99% smooth muscle cells based on uniformly positive staining for $\propto$‐smooth muscle actin (Figure S1–S2.C). The limited demographic data available for patients whose DAs were used to obtain hDASMC colonies are shown in Figure S1–S2.D.

[Ca²⁺]_i Measurements

hDASMC cell lines from 5 human infants were loaded with Cal‐520 AM, and their normoxia‐induced increase in fluorescence intensity was quantified in the presence of vehicle, S1QEL, rotenone, or the ROS scavenger, MitoTEMPO. Representative images taken during these experiments are shown in Figure 3A through 3D. The normoxic increase in [Ca²⁺]_i, seen in each DASMC line, was significantly reduced by S1QEL (50 μmol/L, P<0.05), rotenone (10 μmol/L, P<0.01), and MitoTEMPO (50 μmol/L, P<0.01) (Figure 3E and 3F). S1QEL reduced the normoxic increase in [Ca²⁺]_i, a proxy for DA constriction in vitro, without altering the response to KCl (Figure 3G). This is also consistent with S1QEL's vasodilatory effects being specific for O₂‐induced DA constriction.

Figure 3. S1QEL (suppressor of site IQ electron leak) blunts the O₂‐induced rise in cytosolic calcium concentration, [Ca^{2 +}]_i, in DASMCs.

[Ca²⁺]_i was measured in human ductus arteriosus smooth muscle cells (hDASMCs) using Cal‐520 AM. Representative images showing changes in fluorescence intensity immediately upon beginning imaging in hypoxia (2.5% O₂) or after 15 minutes of exposure to normoxia (18% O₂) are shown in (A) through (D). E, Across 5 different hDASMC lines, each from an individual newborn, normoxia‐induced changes in Cal‐520 fluorescence relative to their hypoxic baseline were measured in cells exposed to vehicle (DMSO, yellow), S1QEL (suppressor of site I_Q electron leak; 50 μmol/L, blue), rotenone (10 μmol/L, red), or MitoTEMPO (50 μmol/L, orange). Change in Cal‐520 fluorescence was quantified in each cell line, with individual data points representing individual cells (regions of interest). F, Data obtained from all hDASMC lines were compiled together to analyze for intergroup differences using a 1‐way ANOVA followed by post hoc Dunnett test (n=5 per group). G, Compared with vehicle, the presence of S1QEL did not alter Cal‐520 fluorescence changes in response to KCl (80 mmol/L, n=5 per group) *P<0.05, **P<0.01.

Mitochondrial Metabolism Measurements

Figure 4A outlines the modified Seahorse XF Mito Stress Test protocol used to evaluate mitochondrial metabolism. Representative results from OCR and ECAR measurements from 1 rbDASMC cell line are shown in Figure 4B and 4C after treatment with S1QEL and rotenone. A representative OCR and ECAR tracing from 1 rbDASMC cell line after treatment with S3QEL is also shown in Figure S2. S1QEL did not inhibit complex I activity, as measured using a dipstick assay (Figure 4D). In contrast to vehicle (DMSO), S1QEL and rotenone both significantly decreased OCR (Figure 4E). However, unlike rotenone, S1QEL preserved key parameters of mitochondrial metabolism and caused no inhibition of spare respiratory capacity, maximal respiration, or ATP‐linked OCR (Figure 4F through 4H). Moreover, S1QEL did not inhibit mitochondrial coupling efficiency, a surrogate measure for proton leakage, and preserved normoxic OCR/ECAR ratios, unlike rotenone, which had significant inhibitory effects on both parameters (Figure 4I and 4J). Thus, unlike rotenone, S1QEL, has minimal inhibitory effects on mitochondrial respiration or ETC complex I function in rbDASMCs.

Figure 4. S1QEL (suppressor of site IQ electron leak), but not rotenone, preserves the mitochondrial respiration, O₂ consumption rate (OCR)/extracellular acidification rate (ECAR) ratio.

A, Various respiratory parameters are derived using the OCR tracings from the SeaHorse XF Mito Stress Test, with sample OCR and ECAR tracing measurements shown in (B) and (C), respectively. D, Mitochondrial complex I activity was measured in cells exposed to control (DMSO) or S1QEL (50 μmol/L) via a dipstick assay (n=6 per group). E through J, Mitochondrial respiratory parameters were compiled across 8 different rabbit ductus arteriosus smooth muscle cell lines for cells exposed to vehicle (DMSO, n=8), S1QEL (1 or 10 μmol/L, n=6 and 8, respectively), S3QEL (suppressor of site III_Q electron leak; 10 μmol/L, n=4), or rotenone (1 μmol/L, n=6).Data were normalized to total protein within each well. Analysis was performed using 1‐way ANOVA followed by post hoc Dunn test. FCCP indicates carbonyl cyanide p‐trifluoro methoxyphenylhydrazone. *P<0.05, **P<0.01, ***P<0.001.

ROS Production During Normoxia in DASMCs

Five hDASMC cell lines were loaded with MitoROS 580, which measures mitochondrial superoxide production, and the effects of switching from hypoxia to normoxia were quantified. Increased superoxide production was observed after switching from hypoxia to normoxia (Figure 5A through 5D). O₂‐induced increases in MitoROS in hDASMCs were inhibited by S1QEL (50 μmol/L, P<0.05) (Figure 5E and 5F) but not by rotenone or S3QEL. Thus, S1QEL, which inhibits O₂‐induced DA constriction, also attenuates O₂‐induced ROS production in hDASMCs. Using the MitoROS plate assay, normoxia significantly increased superoxide production in the presence of vehicle (DMSO, P<0.001), but not S1QEL (50 μmol/L) or MitoTEMPO (50 μmol/L) (Figure 5G). Compared with vehicle, superoxide production during hypoxia was also found to be significantly decreased in the MitoTEMPO‐treated cells (P<0.0001), and superoxide production during normoxia was found to be significantly decreased in the S1QEL‐treated cells.

Figure 5. S1QEL (suppressor of site IQ electron leak) blunts the O₂‐induced rise in mitochondrial reactive oxygen species (ROS) production in ductus arteriosus smooth muscle cells.

Superoxide was measured in human ductus arteriosus smooth muscle cells (hDASMCs) using MitoROS 580. A through D, Representative images showing changes in fluorescence intensity immediately upon beginning imaging in hypoxia (2.5% O₂) or after 10 minutes of exposure to normoxia (18% O₂). E, Across 5 different hDASMC lines, normoxia‐induced changes in MitoROS fluorescence relative to their hypoxic baseline were measured in cells exposed to vehicle (DMSO, yellow), S1QEL (suppressor of site I_Q electron leak; 50 μmol/L, blue), S3QEL (suppressor of site III_Q electron leak; 100 μmol/L, magenta), or rotenone (10 μmol/L, red), with individual data points representing individual cells (regions of interest). F, Data obtained from all hDASMC lines were averaged to analyze for intergroup differences using 1‐way ANOVA followed by post hoc Dunnett test (n=5 per group). G, Using a plate assay, MitoROS 580 fluorescence was measured in cells exposed to hypoxia (blue) or normoxia (red), in the presence of vehicle (DMSO), S1QEL (50 μmol/L), or MitoTEMPO (50 μmol/L). Analysis was performed using 2‐way ANOVA followed by post hoc Tukey test. *P<0.05, ***P<0.001, ****P<0.0001. AU indicates arbitrary units.

Micro‐CT Imaging of DA in Rabbit Kits

Representative maximum intensity projection images of rabbit kits perfused with Omnipaque (with 3% gelatin) immediately after cesarean delivery are shown under the following 3 conditions: (1) hypoxia (no breathing), (2) breathing 100% O₂ for 20 minutes in the presence of vehicle (DMSO, normoxia), and (3) breathing 100% O₂ for 20 minutes in the presence of S1QEL (100 μmol/L; Figure 6A). Rabbit kit thoracic vasculature exposed by midline sternotomy shows the DA is ≈0.5 mm in diameter (Figure 6B). Kits imaged after delivery but before the first breath had ductal patency, whereas kits beathing O₂ showed complete functional DA closure (n=5; Figure 6C). Kits exposed to normoxia and S1QEL (100 μmol/L) retained DA patency in all cases (n=5). Consistent with this, S1QEL prevented DA constriction, expressed as percent change in DA diameter (Figure 6D). Thus, suppression of site I_Q electron leak via S1QEL prevents O₂‐induced DA constriction in vivo.

Figure 6. S1QEL (suppressor of site IQ electron leak) reverses ductus arteriosus (DA) constriction in vivo.

A, Maximum intensity projection images were taken after perfusion with the contrast agent Omnipaque (with 3% gelatin) and subsequent micro‐computed tomography imaging. On the left, the ductus arteriosus (DA) is patent, connecting to the pulmonary artery (PA) and descending thoracic aorta (Ao) in these rabbit kits, who were hypoxic because they were perfused immediately following cesarean delivery prior to thier first breath. In the middle panel are rabbit kits exposed to 20 minutes of 100% O₂ before perfusion. Note the DA is closed, and only a residual beak remains, at the site of connection to the aorta. Finally, in normoxic rabbit kits exposed to normoxia plus S1QEL (suppressor of site I_Q electron leak; 100 μmol/L) before perfusion, the patent DA connects the PA and Ao, as shown on the right. To show the tiny size of the DA in rabbit kits (≈0.5 mm), we show a photograph of the rabbit kit vasculature (B) after performing a midline sternotomy. Percent DA patency (C) and change in DA diameter (D) were quantified for rabbit kits exposed to hypoxia, normoxia with vehicle (DMSO), and normoxia with S1QEL (100 μmol/L, n=5 per group). For (C) and (D), analysis was performed using 1‐way ANOVA followed by post hoc Tukey test. ****P<0.0001.

Discussion

We used novel electron leak suppressors S1QEL and S3QEL to pharmacologically define the site and mechanism of O₂ sensing in the human and rabbit DA. The 3 main findings of this study are as follows: (1) electron leak and ROS production by complex I, but not complex III, are critical for DASMCs contraction to O2, (2) S1QEL, but not S3QEL, mimics hypoxia and inhibits O₂‐induced DA constriction, and (3) S1QEL maintains rabbit DA patency in vivo. Because S1QEL does not inhibit mitochondrial respiratory reserve, ATP‐linked OCR, or complex I activity, we conclude that it is electron leak and ROS production from complex I, not metabolic consequences of ETC inhibition, that triggers O₂‐induced DA constriction. Moreover, we show that electron leak from complex III does not play a major role in DA constriction. The molecular specificity of S1QEL allows us to identify ETC complex I's I_Q site as the primary source of pO₂‐sensitive electron leak and ROS production in DASMCs. The observation that S1QEL maintains DA patency in vivo could potentially have translational relevance to patients with congenital heart diseases, like hypoplastic left heart syndrome, who must maintain DA patency while awaiting corrective surgery, such as the Norwood procedure.

S1QEL was first identified by Brand et al by using high‐throughput screening to select for H₂O₂ suppressing compounds resulting from electron leak at I_Q, but not sites III_Qo or II_F, without also impairing bioenergetic function. ²¹ In their study, S1QEL was found to protect astrocytes against endogenous oxidative damage, protect against ER stress in cardiomyocytes, as well as decrease ischemia–reperfusion injury in the mouse heart. Importantly, these effects were found to be a consequence of changes in superoxide‐H₂O₂ production, because oxidative phosphorylation remained unaltered when measured using the Seahorse XF24 assay. ²¹ Although the exact mechanism of S1QEL is unknown, it is believed that it may indirectly modulate the quinone‐redox reaction by binding to subunit ND1 of complex I to induce structural changes. ²⁴

We demonstrated robust O₂ constriction of isolated term rabbit DA over a physiologically relevant range of pO₂ (35–155 mm Hg) and showed that inhibition of electron leak at site I_Q in ETC complex I, but not site III_Qo in ETC complex III, reverses O₂‐induced DA constriction (Figure 2A through 2C). Importantly, this dilatory effect of S1QEL is not due to a nonspecific suppression of vasoconstriction, because phenylephrine constriction in DA rings is unaltered by a dose of S1QEL that inhibits O₂ constriction (Figure 2D). Likewise, in DASMCs, the ability of KCl to increase [Ca²⁺]_i is also unaltered by S1QEL (Figure 3G).

S1QEL and S3QEL have previously been shown to inhibit formation of superoxide anions from a single Q site in each megacomplex (I_Q and III_Qo, respectively) without blocking forward electron flow ²⁵ or disturbing oxidative phosphorylation. ²¹ , ²⁴ , ²⁶ Important to our conclusions, we confirmed in the ductus that S1QEL and S3QEL selectively block electron leak from ETC complexes I and III, with little confounding effects of metabolic inhibition, although high‐dose S1QEL does reduce basal OCR (Figure 4). These new observations add precision as to the location of the O₂ sensor within the DASMCs’ mitochondrial ETC, because S1QEL is specific for electron leak from the I_Q site in this megacomplex, the preponderant site of superoxide production in complex I. ²⁷ Classical ETC inhibitors are relevant comparators because they too have ETC complex specificity and modulate DA and pulmonary artery tone, ¹⁶ mimicking hypoxia. We reaffirm these findings in the current study showing that the ETC complex inhibitors, particularly antimycin A, dilate the O₂‐constricted DA ex vivo (Figure 2C), ¹⁶ but we acknowledge that the mechanism by which they alter vascular tone is potentially confounded because they not only impair electron flux but also inhibit metabolism (Figure 4).

Our research also explores the effects of S1QEL on key steps in the O₂‐sensing pathway. In these studies, we carefully ensured all DASMCs were maintained in hypoxia until the purposeful exposure to O₂, and superoxide production was quantified using the MitoROS 580 probe. MitoROS 580 is a hydroethidine fluorogenic probe, and although its exact structure is proprietary, it likely accumulates in the mitochondrial matrix due to the covalent attachment of a cationic compound, such as triphenylphosphonium. Although hydroethidine probes can react with other forms of ROS, absorption and excitation wavelengths can be controlled for to select for superoxide production. ²⁸ Normoxia rapidly (in ≈2 minutes) increased superoxide production in DASMC, and this rise in ROS was inhibited by S1QEL but not by S3QEL (Figure 5E through 5F). This is consistent with the extensive literature showing that site I_Q is a major source of ROS production in various tissue types. ²¹ , ²⁹ , ³⁰ These findings indicate that site I_Q, but not site III_Qo, is a significant source of superoxide production required for cell signaling during O₂ sensing in DASMCs. In addition, S1QEL prevents the O₂‐induced rise in [Ca²⁺]_i (Figure 3), a cellular surrogate for O₂‐induced DA constriction. It is noteworthy that rotenone and MitoTEMPO, a mitochondria‐targeted ROS scavenger, also inhibited O₂‐induced increases in [Ca²⁺]_i. MitoTEMPO has previously been shown to rescue H₂O₂‐induced cell death and reduce mitochondrial ROS production, establishing its therapeutic capacity to reduce H₂O₂ cytotoxicity. ³¹ The antioxidant nature of MitoTEMPO comes from its piperidine nitroxide moiety, which has been shown to metabolize superoxide ion radicals into H₂O₂ and O₂, and also act as a catalase to catalyze the conversion of H₂O₂ into water and O₂. ³² , ³³ , ³⁴ , ³⁵ Our findings suggest that similar to S1QEL, MitoTEMPO reduces the O₂‐induced increase in superoxide formation (Figure 5G), and as a consequence of reduced H₂O₂ production, prevents the O₂‐induced increase in cytosolic Ca²⁺ (Figure 3F). Although H₂O₂ production was not directly measured, it is known to be the modulator of DA tone, because reduction of H₂O₂ levels via the enzyme catalase causes hyperpolarization of DASMC membrane potential, whereas more rapid metabolism of superoxide via the actions of superoxide dismutase has no effect on Kv channels. ¹⁶ , ²⁰ In aggregate, the ROS and calcium data confirm S1QEL's ability to inhibit the upstream ROS‐signaling portion of the proposed O₂‐sensing pathway and support the role of ROS derived from ETC complex I in the signal transduction pathway.

Our in vitro [Ca²⁺]_i measurements highlight important similarities between the complex I inhibitor rotenone and electron leak suppressor S1QEL. Both compounds decrease the normoxia‐induced rises in [Ca²⁺]_i production; however, micropolarimetry shows that S1QEL and rotenone have quite different effects on mitochondrial respiratory parameters. OCR measurements in DASMC show that neither electron leak suppressor inhibit mitochondrial metabolism, as judged by the OCR/ECAR ratio, spare respiratory capacity, and ETC I activity (Figure 4). In contrast, rotenone completely and immediately abolishes mitochondrial‐linked OCR, and this inhibition is sustained (Figure 4B). These studies in rbDASMCs show that S1QEL, but not rotenone, maintains key parameters linked to mitochondrial respiration, including spare respiratory capacity and ATP‐linked OCR (Figure 4F through 4H). Taken together with our findings showing greater reductions in DA tension after treatment with S1QEL versus rotenone, these data suggest electron leak and ROS production at ETC complex I, rather than mitochondrial metabolism, are the driving force behind O₂ signaling.

Our research agrees with the redox theory of O₂ sensing, which postulates that ROS production by the ETC is proportional to rates of mitochondrial respiration. ³⁶ When rates of respiration are low, during hypoxia due to reduced O₂ supply, rates of ETC electron flux are also low, causing a proportionate reduction in electron leak and ROS production. Conversely, as rates of respiration increase with increased O₂ supply during normoxia, increased electron flux increases electron leak and subsequent ROS production by the ETC. Our findings support the interpretation that site I_Q constitutes a key site of electron leak in that when it is inhibited by S1QEL, there is attenuated production of vasoactive ROS and inhibition of O₂‐induced DA constriction (Figure 7). Interestingly, the complex I inhibitor rotenone also binds site I_Q, further highlighting the importance of this portion of ETC complex I in fetal O₂ sensing. Although S3QEL did not relax the normoxic DA, antimycin A, which also binds complex III, did relax the DA, consistent with our prior reports ¹⁶ (Figure 2C). Interestingly, S3QEL binds to site III_Qo, or the quinone oxidation site of complex III, whereas antimycin A binds to site Q_i, the quinone reductase site. Although III_Qo is known to be a source of ROS production in diverse cell signaling events and pathologies, ²² our data suggest that ROS production by this site is not implicated in O₂‐sensing in the DA. Site Q_i of complex III does not appear to be a major source of electron leak and ROS production in DASMC, suggesting that the vasodilatory effects of antimycin A may reflect a general decrease in complex III function and metabolic inhibition.

Figure 7. Proposed mechanism of ductus arteriosus smooth muscle cell (DASMC) relaxation by S1QEL (suppressor of site IQ electron leak) and MitoTEMPO.

The left panel reflects the hypoxic DASMCs. In hypoxia, low reactive oxygen species (ROS) production by the mitochondria maintains K⁺ channels (blue) open, and the outward K⁺ current maintains a polarized membrane potential that reduces the open‐state probability of voltage‐gated Ca²⁺ channels (pink). The resulting low intracellular calcium levels maintain the DASMCs in a relaxed state. In the middle panel, in response to increased partial pressure of oxygen (pO₂) with the first breath, ROS produced by leak of electrons from site I_Q of the electron transport chain (ETC) complex I increases. These ROS, notably hydrogen peroxide, diffuse to the cytosol, where they oxidize and inhibit K⁺ channels, causing DASMC depolarization and Ca²⁺ channel activation, leading to vasoconstriction. In the right panel, both the actions of S1QEL (suppressor of site I_Q electron leak, pink arrows) and MitoTEMPO (orange arrows) are shown. S1QEL acts on site I_Q of the ETC, decreasing ROS production and thereby reversing ROS‐induced ion channel activity changes. MitoTEMPO acts within the mitochondrial matrix, converting superoxide radicals into hydrogen peroxide and then water, thereby lowering overall ROS levels and reversing ROS‐induced ion channel activity changes. Kv indicates voltage‐gated potassium channel; CaL, large conductance voltage‐gated calcium channel; SH‐SH, reduced sulfhydryl groups; S‐S, oxidized sulfhydryl groups; NADH, reduced nicotinamide adenine dinucelotide; NAD, oxidized nicotinamide adenine dinucelotide; and SOD2, superoxide dismutase 2.

The significance of site I_Q in O₂‐sensing has recently been demonstrated in the mechanism of hypoxic pulmonary vasoconstriction in adult pulmonary artery smooth muscle cells (PASMCs). In adult PASMCs (rat and human) hypoxia decreases ROS production by mitochondria, leading to inhibition of redox‐sensitive Kv channels, as reviewed by Weir et al. ³⁶ Closure of these channels results in PASMCs depolarization, which triggers an influx of Ca²⁺ through voltage sensitive Ca²⁺ channels, triggering vasoconstriction. Dunham‐Snary et al found that Ndufs2, a subunit of complex I within the I_Q site, is required for O₂‐sensing in adult PASMCs and is critical to the mechanism of hypoxic pulmonary vasoconstriction. ²³ Treatment with siNdufs2 prevented hypoxic increases in [Ca²⁺]_i in PASMC, and significantly impaired hypoxic pulmonary vasoconstriction in vivo in rats. Fernandez‐Aguera et al likewise found that Ndufs2 was crucial for O₂ sensing in the type 1 cells of the carotid body, implicating a conserved role for the I_Q site in O₂ sensing. ³⁷ Ndufs2 is located at the distal portion of the iron–sulfur chain in complex I, and forms part of the binding pocket of the terminal iron–sulfur cluster, which transfers electrons to ubiquinone at site I_Q. Thus, site I_Q may represent a conserved site for O₂‐sensitive ROS production in the tissues within the mammalian homeostatic O₂‐sensing system. These findings suggest potential areas for future research, including examining the role of Ndufs2 in O₂ sensing within the DASMCs and assessing the impact of S1QEL on O₂‐sensing within the adult and fetal pulmonary vasculature.

Finally, the observation that in term rabbits S1QEL maintains DA patency in vivo has potential translational implications. Medical management of the DA after birth remains a clinically relevant problem. In most cases, physicians are tasked with closing the patent DA, often in preterm infants. Ibuprofen closes the PDA in only ≈50% of infants who weigh <1000 g ³⁸ and, even combined with acetaminophen, ibuprofen fails to close PDAs in 45% of cases. ³⁹ Complications of ibuprofen are not rare, including intestinal perforation (5%), necrotizing enterocolitis (3%), acute renal failure (11%), and thrombocytopenia (2%). ⁴⁰ In other infants, such as those awaiting cardiac surgical correction of hypoplastic left heart syndrome, maintenance of circulation depends on maintaining DA patency using prostaglandin E infusion; however, this therapy may be accompanied by several short‐ and long‐term adverse effects, including apnea, fever, and hypotension. ⁴¹ Thus, understanding the mechanism by which O₂ constricts the DA is crucial. Unlike rotenone, which is toxic (ie is used to create preclinical models of Parkinson disease), ⁴² S1QEL has been shown to be nontoxic across a variety of cell types and has been used to improve survival rates in mouse models of cardiac arrest. ²¹ , ⁴³ The observation that S1QEL inhibits O₂‐induced constriction of the DA in rabbit term kits (Figure 6) has potential translational relevance for newborns with congenital heart disease, such as those with a hypoplastic left heart, pulmonary atresia, or aortic stenosis, who depend on a PDA for maintaining their systemic or pulmonary circulation. ⁴⁴ Drugs like S1QEL might, if proven safe in humans, prove a useful alternative to prostaglandin E as a means of maintaining DA patency.

Conclusions

We conclude that electron leak and ROS production from the I_Q site within ETC complex I is critical to DA O₂‐sensing in term rabbits and humans. S1QEL is a potential novel therapeutic agent for maintaining DA patency in vivo. These observations highlight the importance of site I_Q‐generated ROS, consistent with observations in other specialized O₂‐sensing tissues.

Sources of Funding

This work was funded by the Canadian Institute of Health Research (CIHR) grant (S.L.A.), National Institutes of Health R01‐HL071115 (S.L.A.), 1RC1HL099462 (S.L.A.), R01‐HL133675 (W.S.), a Tier I Canada Research Chair (S.L.A.), and the William J. Henderson Foundation (S.L.A.).

Disclosures

None.

Supporting information

Figures S1–S2