Ndufs2, A Core Subunit of Mitochondrial Complex I, Is Essential for Acute Oxygen-Sensing and Hypoxic Pulmonary Vasoconstriction

Abstract

Rationale:

Hypoxic pulmonary vasoconstriction (HPV) optimizes systemic oxygen delivery by matching ventilation to perfusion. HPV is intrinsic to pulmonary artery smooth muscle cells (PASMC). Hypoxia dilates systemic arteries, including renal arteries. Hypoxia is sensed by changes in mitochondrial-derived reactive oxygen species, notably H₂O₂ ([H₂O₂]_mito). Decreases in [H₂O₂]_mito elevate pulmonary vascular tone by increasing intracellular calcium ([Ca²⁺]_i) through redox regulation of ion channels. Although HPV is mimicked by the Complex I inhibitor, rotenone, the molecular identity of the O₂-sensor is unknown.

Objective:

To determine the role of NADH dehydrogenase [ubiquinone] iron-sulfur protein 2 (Ndufs2), Complex I’s rotenone binding site, in pulmonary vascular oxygen-sensing.

Methods & Results:

Mitochondria-conditioned media (MCM) from pulmonary and renal mitochondria isolated from normoxic and chronically hypoxic rats, were infused into an isolated lung bioassay. MCM from normoxic lungs contained more H₂O₂ than MCM from chronic hypoxic lungs or kidneys and uniquely attenuated HPV via a catalase-dependent mechanism. In PASMC acute hypoxia decreased H₂O₂ within 112±7s, followed, within 205±34s, by increased intracellular calcium concentration, [Ca²⁺]_i. Hypoxia had no effects on [Ca²⁺]_i in renal artery SMC (RASMC). Hypoxia decreases both cytosolic and mitochondrial H₂O₂ in PASMC while increasing cytosolic H₂O₂ in RASMC. Ndufs2 expression was greater in PASMC versus RASMC. Lung Ndufs2 cysteine residues became reduced during acute hypoxia and both hypoxia and reducing agents caused functional inhibition of Complex I. In PASMC, siNdufs2 decreased normoxic H₂O₂, prevented hypoxic increases in [Ca²⁺]_i, and mimicked aspects of chronic hypoxia, including decreasing Complex I activity, elevating the NADH/NAD⁺ ratio and decreasing expression of the O₂-sensitive ion channel, Kv1.5. Knocking down another Fe-S center within Complex I (Ndufs1) or other mitochondrial subunits proposed as putative oxygen sensors (Complex III’s Rieske Fe-S center and COX4i2 in Complex IV) had no effect on hypoxic increases in [Ca²⁺]_i. In vivo, siNdufs2 significantly decreased hypoxia- and rotenone-induced constriction while enhancing phenylephrine-induced constriction.

Conclusions:

Ndufs2 is essential for oxygen-sensing and HPV.

INTRODUCTION

Hypoxic pulmonary vasoconstriction (HPV) is a rapid, reversible, homeostatic process intrinsic to the pulmonary vasculature¹. In response to physiologic airway hypoxia, small pulmonary arteries constrict, diverting blood away from poorly oxygenated alveoli, thereby optimizing ventilation-perfusion matching and increasing systemic oxygen delivery¹. In contrast, systemic arteries, such as the renal arteries, dilate during hypoxia, which also serves to increase systemic oxygen delivery^{2, 3}. Though modulated by the endothelium, HPV is intrinsic to pulmonary artery smooth muscle cells (PASMC) in small intrapulmonary arteries. HPV is exploited in single-lung anesthesia⁴, and is a crucial homeostatic mechanism that optimizes systemic PO₂ during pneumonia⁵ and atelectasis⁴. However, excessive HPV can be problematic, and contributes to high altitude pulmonary edema (HAPE)⁶.

The mechanism underlying the opposing pulmonary versus systemic vascular hypoxic responses has been the subject of intense research^7-16, but the molecular identity of the pulmonary vascular oxygen-sensor remains unknown. Prior research shows that PASMC from resistance level PAs are enriched both in the oxygen sensor (mitochondria that can vary production of reactive oxygen species, ROS, in direct proportion to PO₂), and the vasoconstrictor effector (oxygen-sensitive potassium channels, including the voltage-gated channel, Kv1.5^{2, 3}).

Physiologic levels of ROS regulate PASMC tone through reduction-oxidation (redox) mechanisms. Cysteine-rich K⁺ channels in PASMCs close when reduced by hypoxia or reducing agents, and conversely open when oxidized by ROS, such as hydrogen peroxide (H₂O₂)¹⁵. Redox sensitive Kv channels, such as Kv1.5 and Kv2.1, contribute to PASMC membrane potential¹⁷. Redox-mediated inhibition of these Kv channels results in (i) PASMC depolarization, (ii) increased opening of large-conductance, voltage-sensitive Ca²⁺ channels, (iii) increased [Ca²⁺]_i, and subsequently, (iv) vasoconstriction¹⁸.

Although ROS can derive from enzymes, including xanthine oxidase and NAPDH oxidases, the mitochondrial electron transport chain (ETC) is a primary source of ROS and peroxides in most eukaryotic cells^{19, 20}. Mitochondria utilize metabolically-derived electron donors (NADH and FADH₂) as a source of the electrons that cascade down a potential gradient to molecular oxygen, the terminal electron receptor in the ETC. Electron transport is accompanied by proton translocation across the inner mitochondrial membrane, which generates an electrochemical gradient that drives Complex V, ATP synthase. As a byproduct of electron transport, superoxide radical (O₂^{• -}) is generated at Complexes I and III^21-23. This small amount of uncoupled electron transport produces superoxide anion, which would be toxic were it not rapidly converted to H₂O₂ by superoxide dismutase 2 (SOD2). Production of ROS by the ETC reflects both oxygen supply and metabolism, suggesting it as a plausible oxygen sensor.

Further implicating the mitochondrion in oxygen-sensing and suggesting that hypoxia creates a reduced redox milieu are data demonstrating that certain proximal ETC inhibitors (rotenone and antimycin A) as well as reducing agents (e.g. dithiothreitol) mimic the cellular response to acute hypoxia in PASMC^{3, 11}. We studied Complex I as a putative site of O₂-sensing because rotenone, an inhibitor of ETC Complex I, and hypoxia have many similarities, both in vitro and in vivo. Both hypoxia and rotenone cause pulmonary vasoconstriction while dilating systemic arteries, and both reduce ROS production in small PAs³. Pulmonary vasoconstriction to both acute hypoxia and rotenone are selectively suppressed in animals exposed to chronic hypoxia whereas constriction to other agents is increased¹⁴. Rotenone also recapitulates the effects of hypoxia in other O₂-sensing tissues, activating type 1 cells of the carotid body²⁴ and stimulating catecholamine secretion from adrenal medullary cells²⁵. While the role of the mitochondria as O₂-sensing organelles is well accepted, there remains controversy as to the effect of hypoxia on ROS production^{26, 27} and the molecular identity of the ETC protein(s) that functions as the O₂-sensor is unknown.

Mitochondrial Complex I, responsible for the oxidation of NADH, is suspected to be an integral component of the O₂-sensing pathway. Upon inhibition of Complex I, a decrease in ROS levels and accumulation of NADH can inhibit Kv channels^{15, 28}. The mitochondrial subunit NADH dehydrogenase [ubiquinone] iron-sulfur protein 2 (Ndufs2) is the binding site for both quinones (for subsequent transfer of electrons to Complex II)^{29, 30} and the Complex I inhibitor, rotenone³¹. Complete loss of Ndufs2 leads to Mitochondrial Complex I deficiency³², an umbrella diagnosis encompassing mutations of Complex I-encoding genes, that results in myopathy and neurological impairment, partially due to ATP depletion³³. These facts implicate this iron-sulfur subunit as one of the master regulators of Complex I activity. Additionally, Lopez-Barneo et al. recently demonstrated that Ndufs2 knockdown eliminated O₂-sensing in the murine carotid body. Mice lacking Ndufs2 did not hyperventilate in response to hypoxia, but maintained their ventilatory response to hypercapnia³⁴.

Here, we assess the role of Ndufs2 as an O₂-sensor in HPV both in vitro using cultured human and rodent resistance PA-derived PASMC, as well as in in vivo experiments using rodents. We report that Ndufs2 is required for HPV and is itself redox-sensitive, becoming chemically reduced during acute hypoxia. Silencing Ndufs2, using small interfering RNA (siRNA), eliminates acute hypoxia-induced calcium influx and reduces Complex I activity, leading to an accumulation of reduced NADH in isolated PASMC. siNdufs2 eliminates HPV in rats and mice in vivo and ex vivo. Lung mitochondria produce a relaxing factor, which is inhibited by hypoxia, and which we identify as H₂O₂. Mitochondria-conditioned media (MCM) from lung mitochondria is rich in H₂O₂ and inhibits HPV whereas MCM from renal mitochondria contains little H₂O₂ and does not alter HPV. Using compartment-specific redox probes we show that acute hypoxia reduces H₂O₂ levels within 2 minutes, both in the mitochondria and cytosol. This is followed immediately by a rise in intracellular calcium [Ca²⁺]_i, a surrogate for HPV. Finally, molecular inhibition of Ndufs2 in normoxia not only eliminates HPV but also decreases expression of Kv1.5 channels and creates a reduced redox milieu, recapitulating the effects of chronic hypoxia. Correspondingly, chronic hypoxia suppresses Ndufs2 expression in PASMC. Knocking down another Fe-S center within Complex I (Ndufs1) or other ETC components that have been proposed as oxygen sensors (e.g. the Rieske Fe-S center in Complex III⁹ and COX4i2 in Complex IV³⁵) had no effect on the cell-based HPV surrogate. We conclude that Ndufs2 is a critical component of the mitochondrial oxygen sensor that initiates HPV.

METHODS

An expanded Methods section is available in the Online Supplement. The authors declare that all supporting data are available within the article (and its Online Supplement).

RESULTS

Lung mitochondria produce hydrogen peroxide - a normoxic pulmonary vasodilator

Filtered, conditioned media incubated with purified, normoxic lung mitochondria (nLM) significantly decreased HPV when injected into the isolated perfused lung (56 ± 8% reduction in HPV), compared to injection of vehicle (9 ± 0% reduction of HPV). Media incubated with mitochondria derived from chronically hypoxic lungs (hLM) caused less inhibition of HPV (27 ± 1% reduction of HPV). Media from normoxic kidney mitochondria (KM) did not alter HPV (1.5 ± 0.0% reduction of HPV, Figure 1A-1B). nLM produce significantly more H₂O₂ than both hLM and KM, as measured by the Amplex Red assay (Figure 1C).

Figure 1: HPV is specific to the lung and is mediated by H₂O₂:

(A) Representative pulmonary artery pressure (PAP) traces of isolated perfused rat lung (IPL) treated at peak HPV with mitochondria-conditioned media (MCM; indicated by shaded area; veh – vehicle, Hx – hypoxia, Nx – normoxia, KM – kidney mitochondria, nLM – normoxic lung mitochondria, hLM – hypoxic lung mitochondria, // indicates separate perfusion experiments). (B) % reduction of HPV in response to MCM (shaded areas of A); data expressed as mean ± SEM of n ≥ 3 interventions per group; *p < 0.05, ****p < 0.0001 compared to control. (C) Quantitation of H₂O₂ in MCM from normoxic lung (nLM), kidney (KM) and chronic hypoxic lung (hLM); data expressed as mean ± SEM of n ≥ 6. *p < 0.05, ****p < 0.0001 (D) Representative PAP trace of IPL treated at peak HPV with vehicle (Veh), tBOOH (100 μM), nLM + Cat (120kU/mg), or nLM + Rote (50μM); treatments indicated by shaded area. (E) % reduction of HPV in response to redox mediators (shaded areas of C); data expressed as mean ± SEM of n ≥ 3 interventions per group; *p < 0.05, ****p < 0.0001 compared to control; tBOOH – tert-buytl hydroperoxide, nLM – normoxic lung mitochondria, Cat – catalase, Rote – rotenone.

Addition of catalase (nLM + Cat, 120 kU/mg) eliminated the hemodynamic effects of nLM (10 ± 2 % reduction of HPV, vs. 56% reduction with nLM alone). Conversely, exogenous H₂O₂, (tert-butyl-hydroperoxide, tBOOH, 100 μM), decreased HPV 75 ± 3%, mimicking the effect of nLM MCM. Addition of rotenone to normoxic lung mitochondria largely eliminated the reduction in HPV caused by the MCM (Figure 1D-1E). Rotenone alone given at peak HPV did not enhance vasoconstriction (Online Figure I). Both exogenous tBOOH and catalase supplementation of nLM augment HPV in a linear fashion (Online Figure I).

Acute hypoxia induces a reduced redox environment with less mitochondrial and cytosolic H₂O₂ production in PASMC

Transfection of rat PASMC with either the mitochondria- or cytosolic-targeted, H₂O₂-specific probes HyPer-dMito (Figure 2A) or HyPer-dCyto (Figure 2B) permits dynamic, real-time monitoring of intramitochondrial or cytosolic H₂O₂ levels, respectively³⁶. These probes are specific to hydrogen peroxide with a sensitivity limit < 5μM (in E. coli) and respond with immediate changes in fluorescence³⁶. The specificity, dynamic range and hypoxic responsiveness of these probes were verified in PASMC (Figure 2C-2D, Online Figure II) and RASMC (Online Figure II); we were able to detect changes in HyPer fluorescence with 100nM H₂O₂. Imaging under flow conditions permitted sequential alternation of normoxic and hypoxic buffers, and addition of exogenous H₂O₂ and catalase (Figure 2E-2G). Acute hypoxia (pO₂ ~20 mmHg) elicited a decrease in mitochondrial H₂O₂ levels within 2 minutes (Figure 2E-2F, Online Figure III). The effects of hypoxia on H₂O₂ were mimicked by administration of rotenone during normoxia (Figures 2G-2H). Hypoxia elicited no significant changes in mitochondrial H₂O₂ levels in RASMC (although cytosolic H₂O₂ increased in response to acute hypoxia, Online Figure II). Micropolarimetry analysis of rat PASMC and RASMC indicate that PASMC have significantly higher basal and ATP-linked oxidative metabolism (Online Figure II).

Figure 2: PASMC H₂O₂ levels decrease during acute hypoxia under static and flow conditions:

(A) Representative fluorescence microscopy of dual-excitation-single-emission H₂O₂ pHyPer-dMito (“HyPer-mito”) vector-transfected primary rat PASMC during normoxia (top) and acute hypoxia (bottom) under flow conditions. 500nm : 420nm ratio (HyPer-mito fluorescence) increases as mitochondrial H₂O₂ increases. (B) Representative fluorescence microscopy of dual-excitation-single-emission H₂O₂ pHyPer-dCyto (“HyPer-cyto”) vector-transfected primary rat PASMC during normoxia (top) and acute hypoxia (bottom) under flow conditions. 500nm : 420nm ratio (HyPer-cyto fluorescence) increases as cytosolic H₂O₂ increases. (C) Quantification of hypoxia-induced decrease in mitochondrial H₂O₂ and subsequent normoxic recovery in rat PASMC under static conditions, relative to normoxia. Data expressed as mean ± SEM of n = 5. **p < 0.01. (D) Quantification of hypoxia-induced decrease in cytosolic H₂O₂ in rat PASMC under static conditions. Data expressed as mean ± SEM of n = 5. *p < 0.05. (E) Representative ratiometric trace of HyPer-mito rat PASMC, reported as % of normoxic baseline. (F) Quantification of hypoxia-induced decrease in H₂O₂ in HyPer rat PASMC. (G) Representative ratiometric trace of HyPer rat PASMC treated with redox mediators, reported as % of normoxic baseline. (H) Quantification of rotenone-induced decreases in H₂O₂ in HyPer rat PASMC.

Data expressed as mean ± SEM of n = 4-5 fields of view from ≥ 3 primary cell lines, relative to normoxia; *p < 0.05.

Nx – normoxia, Hx – hypoxia, tBOOH – tert-butyl hydroperoxide, Cat – catalase, Rote – rotenone, 150 mmHg – normoxic PO₂, 20 mmHg – hypoxic PO₂.

Acute hypoxia increases intracellular calcium [Ca²⁺]_i in PASMC

[Ca²⁺]_i was measured using the ratiometric indicator Fura2-AM. An increase in [Ca²⁺]_i upon exposure to hypoxia was considered a surrogate for HPV (Figure 3A, Online Figure III). Acute hypoxia (~20 mmHg pO₂, ~35 mmHg pCO₂, pH 7.4, Online Figure III) elicited a reversible increase in [Ca²⁺]_i within 3-4 minutes in PASMC (Figure 3B-3C, Online Figure III). 80 mM KCl was used as control because it stimulates the maximal increases in [Ca²⁺]_i possible with a depolarizing stimulus. The dynamic range of calcium detectable by Fura 2 was determined using ionomycin to permeabilize the PASMC membrane, exposing the indicator to extracellular calcium concentrations (and stimulate calcium release from intracellular stores). EGTA was then administered to chelate all calcium and achieve a minimum [Ca²⁺]_i concentration (>35 μM and 0 μM, respectively). Addition of rotenone to the flow chamber increased [Ca²⁺]_i, mimicking hypoxia (Figure 3D-3E). Exogenous oxidants, such as tBOOH reduced [Ca²⁺]_i, whereas reducing agents, such as dithiothreitol, and catalase increased [Ca²⁺]_i levels, consistent with the interpretation that normoxia is a state of increased oxidation and low [Ca²⁺]_i whilst hypoxia is a state of reduction which increase [Ca²⁺]_i (Figure 3F-3G, Online Figure III).

Figure 3: PASMC intracellular calcium levels as a surrogate for HPV confirm decreased H₂O₂ levels during acute hypoxia under flow conditions:

(A) Representative widefield fluorescence microscopy of dual-excitation-single-emission Fura-2 AM in primary rat PASMC during normoxia (top) and hypoxia (bottom) under flow conditions. Cytosolic 340nm : 380nm ratio (Fura2 intensity) increases as intracellular calcium [Ca²⁺]_i increases. (B) Representative ratiometric trace of Fura-2 AM-loaded rat PASMC, expressed as % of normoxic signal. (C) Intracellular calcium ([Ca²⁺_i]) expressed as Fura-2 AM intensity in rat PASMC during normoxia, acute hypoxia and 80mM KCl treatment. (D) Representative ratiometric trace of Fura-2 AM-loaded rat PASMC treated with rotenone, expressed as % of normoxic signal (E) Intracellular calcium ([Ca²⁺_i]) expressed as Fura-2 AM intensity in rat PASMC during 50nM rotenone treatment, mimicking acute hypoxia. (F) Representative ratiometric trace of Fura-2 AM-loaded rat PASMC treated with redox mediators, expressed as % of normoxic signal. (G) Quantification of intracellular calcium ([Ca²⁺_i]) in Fura-2 AM-loaded rat PASMC during 1μM tBOOH and 40kU/mg catalase treatment, mimicking acute hypoxia.

Data expressed as mean ± SEM (% normoxia) of n = 5-10 fields of view from ≥ 3 primary cell lines. *p < 0.05, **p < 0.01, ****p < 0.0001.

Nx – normoxia, Hx – hypoxia, KCl – potassium chloride, Iono – ionomycin, EGTA - ethylene glycol-bis(β-aminoethyl ether)-N,N,N’,N’-tetraacetic acid, Rote – rotenone, tBOOH – tert-butyl hydroperoxide, Cat – catalase, 150 mmHg – normoxic PO₂, 20 mmHg – hypoxic PO₂).

Mitochondrial Complex I subunit Ndufs2 is necessary for oxygen-sensing in isolated PASMC

Immunocapture and subsequent electrophoresis of mitochondrial Complex I was performed to resolve the expression level of components of this megacomplex. This revealed differences in Complex I composition between PASMC and RASMC (Figure 4A). PASMC were enriched in a ~50 kDa gel band (relative quantification in Figure 4B), which was identified as Ndufs2, one of seven iron-sulfur (Fe-S) clusters in Complex I. Ndufs2 was visualized using super-resolution confocal microscopy, confirming its association with mitochondrial Complex I in normal rat PASMC (Figure 4C). Ndufs2 expression was significantly greater in PASMC versus RASMC Complex I (confirmed by RT-PCR and western blot, Figure 4D-4E). Knockdown of Ndufs2 in rat PASMC (verified by RT-PCR, western blot and immunofluorescence, Figure 4F-4H) eliminated the hypoxia-induced rise in [Ca²⁺]_i, but had no effect on KCl-induced elevation of [Ca²⁺]_i (Figure 4I-4J, Online Figure III). Cells were not tolerant of simultaneous transfection with HyPer-dMito and Ndufs2 siRNA, so Amplex Red was used to determine the effect of siNdufs2 on H₂O₂. This assay confirmed that siNdufs2 decreased normoxic levels of H₂O₂ (Figure 4K). Silencing Ndufs2 also reduced Complex I activity without altering Complex I quantity or distal ETC (Complex IV) activity (Figure 4L-4N, Online Figure IV). siNdufs2 also suppressed mitochondrial oxidative metabolism (measured via micropolarimetry), leading to a normoxic accumulation of reduced NADH (Figure 4O-4R). The elevated concentrations of reduced NADH and an elevated NADH/NAD⁺ ratio suggests that siNdufs2 effectively blocks electron flux from NADH to Complex I, resulting in a retrograde reduction of the cellular milieu, that is reinforced by reduced H₂O₂ production.

Figure 4: Mitochondrial Complex I subunit Ndufs2 is essential for acute pulmonary oxygen sensing in rat PASMC:

(A) Complex I immunocapture showing differential expression of protein subunits between rat pulmonary (left) and renal (right) artery smooth muscle cells; 49kDa subunit Ndufs2 marked. (B) Relative intensity of ~50kDa gel band (Ndufs2) in Complex I isolated from rat PASMC and RASMC. Gel bands below graph are ~50kDa gel band seen in Figure 4A; mean ± SEM of n = 3 per group, normalized to lane total protein signal. (C) Visualization of Ndufs2 in normal rat PASMC via confocal microscopy (i), and stimulated emission-depletion (STED) confocal nanoscopy (ii-iv, v - schematic of nanoscopy in panels iii-iv); blue – nucleus, red – mitochondria, cyan – ETC Complex I immunocapture, green – Ndufs2. (D) Quantification of Ndufs2 mRNA levels in rat PASMC and rat RASMC, normalized to control siRNA; mean ± SEM of n = 5 per group. (E) Representative immunoblot and densitometric quantification demonstrating reduced Ndufs2 protein levels (normalized to TOMM20) in rat RASMC compared to rat PASMC. (F) Ndufs2 mRNA levels in rat PASMC treated with negative control siRNA or Ndufs2 siRNA for 48 hours, normalized to control siRNA; mean ± SEM of n = 5 per group. (G) Representative immunoblot and densitometric quantification of Ndufs2 protein levels (normalized to TOMM20) in control and Ndufs2 siRNA-treated rat PASMC. (H) Confocal microscopy verification of siRNA knockdown of Ndufs2 in control rat PASMC (siNC) and Ndufs2 siRNA-treated rat PASMC (siNdufs2); blue – nucleus, red – ETC Complex I immunocapture, green – Ndufs2. (I) Representative ratiometric trace of Fura-2 AM-loaded rat PASMC treated with Ndufs2 siRNA under flow. (Nx – normoxia, Hx – hypoxia, KCl – potassium chloride, Iono – ionomycin, EGTA - ethylene glycol-bis(β-aminoethyl ether)-N,N,N’,N’-tetraacetic acid). (J) Intracellular calcium ([Ca²⁺_i]) expressed as Fura-2 AM intensity in siNdufs2-treated rat PASMC during normoxia, and subsequent acute hypoxic and 80mM KCl challenges; mean ± SEM of n = 10 fields of view from ≥ 3 primary cell lines. (K) H₂O₂ present in tissue culture media (24 hours) of control and siNdufs2 rat PASMC during normoxia via the Amplex Red assay; mean ± SEM of n = 6 wells of ≥ 3 primary cell lines. (L) Mitochondrial Complex I activity of control and siNdufs2 rat PASMC during normoxia via dipstick assay; mean ± SEM of n = 6 wells of ≥ 3 primary cell lines. (M) Mitochondrial Complex I quantity in control and siNdufs2 rat PASMC during normoxia via dipstick assay; mean ± SEM of n = 6 wells of ≥ 3 primary cell lines. (N) Mitochondrial Complex IV activity control and siNdufs2 rat PASMC during normoxia via dipstick assay; mean ± SEM of n = 6 wells of ≥ 3 primary cell lines. (O) Oxygen consumption rate (OCR) measured in control and siNdufs2 rat PASMC; mean ± SEM of 3-5 technical replicates of ≥ 3 primary cell lines. (Oligo – oligomycin, FCCP – carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone, AA – antimycin A, Rote – rotenone). (P) Quantification of OCR in control and siNdufs2 rat PASMC; mean ± SEM of 3-5 technical replicates of ≥ 3 primary cell lines. (ATP – ATP-linked, Max – maximal, Res Cap – reserve capacity). (Q) NADH concentration (reduced) in control and siNdufs2 rat PASMC; mean ± 5 measurements of ≥ 3 primary cell lines. (R) NADH/NAD⁺ ratio in control and siNdufs2 rat PASMC; mean ± SEM of 5 measurements of ≥ 3 primary cell lines.

*p < 0.05, **p < 0.005, ***p < 0.001.

To assess whether other ETC subunits are also critical to pulmonary oxygen-sensing, we performed targeted, in vitro knockdown of three ETC proteins, including two that have been proposed as putative oxygen sensors (Online Figure V). Compared to control siRNA, we found no significant impairment of acute hypoxic increases in [Ca²⁺]_i after effective knockdown of another Complex I Fe-S center, Ndufs1, the Rieske Fe-S protein in Complex III, Uqcrfs1, or COX4i2, a subunit of Complex IV. These results demonstrate both that siNdufs2 is not leading to a nonspecific effect on Complex I, and that pulmonary vascular oxygen-sensing requires Ndufs2 (i.e. other putative sensors are not crucial to HPV). These experiments further strengthen and support our conclusion that Ndufs2 is a uniquely critical part of the pulmonary vascular oxygen sensor.

Silencing Ndufs2 eliminates HPV in vivo

We next explored the translational nature of our in vitro findings, performing targeted in vivo knockdown of Ndufs2 in rats and mice (Figure 5, Online Figure VI, Movies MI-MII). Silencing Ndufs2 (Figures 5A-5B), achieved by airway nebulization of 1 nmole siRNA, eliminated HPV in mice and rats when tested 48 hours post-siRNA treatment. siNdufs2 decreased the rise in mean pulmonary artery pressure (mPAP) caused by acute hypoxia (10% FiO₂, PaO₂ ~40mmHg, PaCO2 ~30mmHg, pH 7.4, Online Figure VI) in control rats (Figures 5C-5E) and reduced expression of Kv1.5 in rat lungs (Figure 5F). siNdufs2 also significantly attenuated HPV in the mouse, as measured by intravital confocal microscopy and confirmed with confocal microscopy of excised lung segments (Figure 5G, Movies MI-MII). Collectively these data indicate that Ndufs2 is required for oxygen-sensing in the pulmonary vasculature of two different rodent models.

Figure 5: Silencing Ndufs2 attenuates HPV in vivo:

(A) Ndufs2 mRNA levels in rat lung homogenate from rats nebulized with control and Ndufs2 siRNA, normalized to control siRNA; mean ± SEM of n = 4 per group. (B) Confocal microscopy verification of siRNA knockdown of Ndufs2 in rat lungs; red – smooth muscle actin, green – Ndufs2. Quantitation of SMA-specific Ndufs2 area; mean ± SEM of 5 vessels per group. (C) Representative mean pulmonary artery pressure (mPAP) trace measured via pulmonary artery catheterization in rats nebulized with control siRNA. (D) Representative mean pulmonary artery pressure (mPAP) trace measured via pulmonary artery catheterization in rats nebulized with Ndufs2 siRNA. (E) Mean pulmonary artery pressure (mPAP) in rats nebulized with control (black line) or Ndufs2 (grey) siRNA during normoxia, hypoxia and normoxic recovery; mean ± SEM of n = 4 per group. (F) Kv1.5 mRNA levels in rat lung homogenate from rats nebulized with control and Ndufs2 siRNA, normalized to control siRNA; mean ± SEM of n = 4 per group. (G) % reduction in vessel diameter during hypoxia in mice nebulized with control and Ndufs2 siRNA, measured via intravital microscopy, expressed as mean ± SEM of n = 13-16 blood vessels per group. *p < 0.05, ***p < 0.0005.

siNdufs2 elicits downstream changes to the effector mechanism of HPV, and partially mimics chronic hypoxia

To determine if chronic hypoxia and siNdufs2 elicit similar cellular changes, we compared rat and human PASMC exposed to both conditions (Figure 6). Chronic hypoxia (10% O₂ for 72+ hours) reduced Ndufs2 expression (Figure 6A-6B, 6J), Complex I activity (Figure 6C-6D), H₂O₂ production (Figure 6E-6F), and led to an accumulation of NADH (Figure 6G-6H) in both normal rat and normal human PASMC. Changes in the redox milieu during chronic hypoxia inhibited the expression of both Ndufs2 and the voltage-gated potassium (Kv) channel Kv1.5 in rat PASMC (Figure 6I-6J). In whole rat lung, chronic hypoxia also suppressed Ndufs2 and caused a modest decrease in Kv1.5 expression (Figure 6K). Collectively, these data demonstrate that silencing Ndufs2 elicits a cellular phenotype similar to chronic hypoxia.

Figure 6: siNdufs2 mimics chronic hypoxia in vitro:

(A) Ndufs2 mRNA levels in rat PASMC (control, chronic hypoxia – 10% O₂ for 72+ hours, siNdufs2), relative to control. (B) Ndufs2 mRNA levels in normal human PASMC (control, chronic hypoxia – 10% O₂ for 72+ hours, siNdufs2), relative to control. (C) Mitochondrial Complex I activity in rat PASMC (control, chronic hypoxia – 10% O₂ for 72+ hours, siNdufs2) via dipstick assay. (D) Mitochondrial Complex I activity in normal human PASMC (control, chronic hypoxia – 10% O₂ for 72+ hours, siNdufs2) via dipstick assay. (E) H₂O₂ present in tissue culture media (24 hours) of rat PASMC (control, chronic hypoxia – 10% O₂ for 72+ hours, siNdufs2) via Amplex Red assay. (F) H₂O₂ present in tissue culture media (24 hours) of normal human PASMC (control, chronic hypoxia – 10% O₂ for 72+ hours, siNdufs2) via Amplex Red assay. (G) NADH/NAD⁺ ratio in rat PASMC (control, chronic hypoxia – 10% O₂ for 72+ hours, siNdufs2). (H) NADH/NAD⁺ ratio in normal human PASMC (control, siNdufs2, normoxic control, and chronic hypoxia (10% O₂ for 72+ hours) normal human PASMC. (I) Kv1.5 mRNA levels in rat PASMC (control, chronic hypoxia – 10% O₂ for 72+ hours, siNdufs2), relative to control. (J) Immunoblot and densitometric quantitation of Ndufs2 and Kv1.5 levels in chronic hypoxic (C-Hx, left) and normoxic (Nx, right) rat PASMC, normalized to GAPDH expression. . (K) Immunoblot and densitometric quantitation of Ndufs2 and Kv1.5 levels in chronic hypoxic (C-Hx, left) and normoxic (Nx, right) rat lung homogenate, normalized to GAPDH expression.

Data expressed as mean ± SEM of n ≥ 3 per group. *p < 0.05, **p < 0.005, ***p < 0.0005, ****p < 0.0001.

To address whether Ndufs2 knockdown in vivo would mimic chronic hypoxia, we administered four doses of siRNA to rats over 10-12 days, before performing PA catheterization and subsequent immunohistochemical analysis (Figure 7, Online Figure VII). Sequential doses of Ndufs2 siRNA successfully reduced Ndufs2 expression (Figure 7A-7B). Serial siNdufs2 treatment elicited a subtle (but statistically insignificant) increase in baseline mPAP (Online Figure VII), eliminated acute hypoxia-induced vasoconstriction and eliminated rotenone-induced vasoconstriction while enhancing the vasoconstrictive response to phenylephrine (an α₁-adrenergic receptor agonist), (Figure 7C-7E). Serial siNdufs2 did not cause vascular remodeling consistent with chronic hypoxia nor did it elicit pulmonary hypertension (Online Figure VII). However, these experiments did confirm our previous in vivo studies using a single dose of siRNA findings (siNdufs2 prevents acute hypoxia-induced rise in mPAP). The concordant responses to hypoxia and rotenone are consistent with our in vitro data and confirm similarities between rotenone and acute hypoxia. These findings are also consistent with previous reports that chronic hypoxia induces a parallel loss of rotenone-induced constriction and enhanced phenylephrine-induced constriction¹⁴. Taken together, these results provide additional evidence for Ndufs2 as a critical component of the oxygen sensor.

Figure 7: Serial siNdufs2 mimics chronic hypoxia in vivo.

(A) Ndufs2 mRNA levels in rat lung homogenate from rats nebulized four times with control and Ndufs2 siRNA, normalized to control siRNA. (B) Confocal microscopy verification of siRNA knockdown of Ndufs2 in rat lungs; red – smooth muscle actin, green – Ndufs2. Quantitation of SMA-specific Ndufs2 area; mean ± SEM of 5 vessels per group. (C) Representative mean pulmonary artery pressure (mPAP) trace measured via pulmonary artery catheterization in rats serially nebulized with control siRNA. (D) Representative mean pulmonary artery pressure (mPAP) trace measured via pulmonary artery catheterization in rats serially nebulized with Ndufs2 siRNA. (E) Mean pulmonary artery pressure (mPAP) in rats serially nebulized with control (black) or Ndufs2 (grey) siRNA during normoxia, hypoxia, normoxic recovery, 50 μg/mL phenylephrine, and 50 μM rotenone.

Data expressed as mean ± SEM of n = 4-5 per group. *p < 0.05, **p < 0.005; PE – phenylephrine, Rote – rotenone.

Acute hypoxia increases Ndufs2 thiol reduction and reducing agents inhibit Complex I activity.

To determine how hypoxia effects Ndufs2’s redox status, we used biotin-phenylarsinic acid (PAA) columns to isolate and quantify chemically reduced Ndufs2 from mouse lung exposed to normoxia versus 10% O₂ for 30 minutes. Acute hypoxia significantly increased the amount of reduced Ndufs2 in mouse lung, without affecting total Ndufs2 levels (Figure 8A-8B). To further assess the redox sensitivity of Ndufs2, we performed Complex I activity measurements in control and siNdufs2 rat PASMC exposed to normoxia, acute hypoxia, diamide (an oxidizing agent) and dithiothreitol (DTT, a potent reducing agent). Acute hypoxia and DTT both reduce Complex I activity in control PASMC (Figure 8C, left). In siNdufs2 PASMC, where Complex I activity is already reduced in normoxia (Figure 4L), acute hypoxia no longer has an effect on activity, but DTT still modestly reduces activity of the complex (Figure 8C, right). Thus, chemical reduction of Ndufs2 inhibits its function; these data reaffirm that acute hypoxia is a state of reduction.

Figure 8: Ndufs2 is a redox-sensitive molecular oxygen sensor in hypoxic pulmonary vasoconstriction:

(A) Biotin-PAA immunocapture blot of reduced Ndufs2 protein (upper) and standard immunoblot of total Ndufs2 input (lower) of whole lung homogenate from mice exposed to normoxia (left) or 10% O₂ for 30 minutes (right). (B) Quantitation of immunoblots from 8A, with biotin-PAA immunocapture signal only (i), input signal only (ii) and biotin-PAA immunocapture normalized to total input signal (iii); mean ± SEM of n = 5 animals per group. **p < 0.005. (C) Mitochondrial Complex I activity of control (left) and siNdufs2 (right) rat PASMC during normoxia, acute hypoxia, 10mM DTT and 1μM diamide treatment via dipstick assay; mean ± SEM of n = 3 primary cell lines. *p < 0.05, **p < 0.005 (D) Schematic showing the pulmonary vasculature and PASMC mitochondria during normoxia (left) and hypoxia (right). a) Under normoxic conditions, mitochondrial H₂O₂ production (from Complexes I and III due to uncoupled electron transport ³⁹ and elevated NAD/NADH ratio result in an oxidative environment and leads oxidation of sulfhydryl groups (S-S) on Kv channels, thereby increasing their open state probability, while the Ca_L channel remains closed. b) Hypoxia inhibits the production of uncoupled electrons at Ndufs2, leading to lower levels of superoxide anion and thus lower concentrations of the dismutation product, hydrogen peroxide. This, coupled with accumulated NADH result in depolarization of the cell, closing Kv channels and subsequent opening of Ca_L channels and triggering of vasoconstriction. c) Intact Ndufs2 is required for optimal Complex I function, maintenance of circulating normoxic H₂O₂ levels and sensing of changes in O₂. d) Inhibition of Complex I, whether caused by hypoxic, pharmacological or molecular inhibition of Ndufs2 results in a more reduced SMC redox state, inhibiting Kv channel expression and a loss of activation of the Ca_L channel and vasoconstriction.

PASMC – pulmonary artery smooth muscle cell, H₂O₂ – hydrogen peroxide, NAD(H) – nicotinamide adenine dinucleotide, Kv – oxygen-sensitive, voltage-gated potassium channel, Ca_L – L-type calcium channel

DISCUSSION

This study demonstrates that HPV is initiated by a decrease in production of mitochondria-derived H₂O₂, which originates primarily from ETC Complex I. The oxygen-sensitive production of H₂O₂ is unique to PASMC, is not found in RASMC, and is mediated by the ETC Complex I subunit Ndufs2. Ndufs2 is also the binding site for rotenone. Molecular downregulation of Ndufs2 mimics the effects of chronic hypoxia on PASMC redox chemistry, metabolism and the expression of a key oxygen-sensitive potassium channel, Kv1.5. Super resolution confocal microscopy allowed us to image Ndufs2 within ETC Complex I for the first time. Using various microscopic and protein chemistry techniques we identified greater expression of Ndufs2 in PASMC versus RASMC, likely contributing to the tissue specificity of HPV. We carefully confirmed that knockdown of Ndufs2 by siRNA was effective in vivo and in vitro and showed that loss of Ndufs2 selectively eliminated PASMC responses to hypoxia without inhibiting responses to other vasoconstrictor agents. Knockdown of other components of ETC Complex I, III and IV did not alter the response of PASMC to acute hypoxia, consistent with the unique importance of Ndufs2. Finally, we show that acute oxygen-sensing in vivo requires Ndufs2. Interestingly, Ndufs2 and Complex I activity are themselves redox-sensitive, being inhibited by hypoxia and reducing agents.

Prior studies have identified the mitochondria as the pulmonary vascular oxygen sensor^{8, 28} and suggested ETC Complex I as a key locus for sensing hypoxia²⁸. To determine the molecular identity of the oxygen sensor and its signaling mediator, we created a bioassay using mitochondria isolated from lungs versus kidneys to condition media, on the assumption that the redox mediator was a diffusible substance that must be able to egress the mitochondria and thus should be sufficiently stable to be detected in bioassay (Figure 1). This mitochondria-conditioned media (MCM) was administered to isolated perfused lungs at the peak of their hypoxic pressor response. The MCM from lungs (but not kidneys) contained substantial H₂O₂ (Figure 1C), which was shown to be the bioactive factor that inhibited HPV. Notably the vasodilator substance in nLM MCM was inhibited by catalase. These data indicate that the pulmonary vasculature is maintained in a normoxic state of vasodilation by a prooxidant, normoxic vasodilator stimulus (H₂O₂) and that HPV reflects the withdrawal of H₂O₂, findings consistent with the Redox Theory of HPV³⁷. The most abundant ROS produced by the ETC is superoxide radical (O₂^{• -}), which largely derives from ETC Complexes I and III^{11, 38, 39}, O₂^{• -} is unstable and rapidly converted to H₂O₂, which is more stable and has a larger diffusion radius, allowing it to serve as a redox second messenger.

Proximal ETC inhibitors, rotenone (Complex I) and antimycin A (Complex III) mimic hypoxia in the pulmonary vasculature (causing constriction which is not additive to HPV), whereas distal ETC inhibitors, such as the Complex IV inhibitor cyanide, cause pulmonary vasoconstriction but increase ROS levels and have no effect on acute HPV^{3, 10}. We replicated these results in the isolated perfused rat lung (Figure 1, Online Figure I). Moreover, rotenone mimicked hypoxia in isolated PASMC, decreasing H₂O₂ production and increasing [Ca²⁺]_i. Acute hypoxia elicited a rapid and reversible decrease in mitochondrial (and subsequently, cytosolic) H₂O₂ in PASMC, while exogenous H₂O₂ (tert-butyl hydroperoxide, tBOOH) and dithiothreitol mimicked the effects of normoxia and hypoxia, respectively (Figure 2-3, Online Figure II-III). The cellular specificity of the hypoxic constrictor response, a hallmark of HPV, was confirmed by the demonstration that in RASMC, hypoxia neither raised [Ca²⁺]_i nor altered mitochondrial H₂O₂ levels (Online Figure II).

Several groups report that HPV is stimulated by a decrease in ROS and is a condition of reduced redox state^{40, 41}, while others propose the mechanism of HPV involves an increase in mitochondrial ROS^{9, 35, 42, 43}. The basis for the discrepancy (ROS up versus down with hypoxia) remains unclear, although Schumacker et al. have noted that hypoxia decreases ROS in the mitochondrial matrix⁹. Further, in all theories, whether ROS are increased or decreased by hypoxia, ROS are identified as diffusible, mitochondrial-derived signaling mediators. Therefore, they must reach the cytosol and cell membrane to modulate the activity of the mediators of vascular tone, rho kinase and ion channels, respectively. Thus, while there are redox compartments within the cell and even within the mitochondria, the redox mediator responsible for normoxic pulmonary vasodilatation must cross boundaries. Using dynamic, compartment-targeted probes that are sensitive to rapid changes in mitochondrial or cytosolic H₂O₂, we demonstrate that the net redox state of the PASMC cytosol and mitochondria are harmonized-both being reduced by acute, physiologic, hypoxia (Figure 2C-2D). The measurement technique we used is superior in sensitivity and specificity to many ethidium-based ROS probes³⁶ and directly shows rapid hypoxic suppression of H₂O₂ production.

We used hypoxia-induced increases in [Ca²⁺]_i, as a surrogate for HPV in PASMC (Figure 3-4, Online Figure V). PASMC responded to hypoxia with a rapid and reversible decline in H₂O₂ concentrations (within a minute of hypoxia) that preceded the increase in [Ca²⁺]_i, which began 205 ± 34 seconds after onset of hypoxia (Figure 3, Online Figure III). The PASMC’s hypoxic response was mimicked by rotenone, increasing [Ca²⁺]_i. Conversely, the oxidant tBOOH decreased hypoxia-induced [Ca²⁺]_i, and this suppression of Ca²⁺ influx could be rescued by addition of catalase, further supporting the conclusion that mitochondria-derived H₂O₂ modulates vasoconstriction during acute hypoxia.

We have previously reported that PASMC and RASMC mitochondria are qualitatively distinct³. Renal mitochondria generate less H₂O₂ and have reduced expression of the H₂O₂-generating enzyme SOD2 than PA mitochondria³. Additionally, renal mitochondria do not modulate production of H₂O₂ in response to acute hypoxia, as was observed in the lung³. We speculate that this difference in mitochondrial function may be epigenetic, related to the environmental conditions experienced by the mitochondria in PASMC of small PAs, which are exposed to high alveolar oxygen concentrations, versus renal artery SMC, which experience the lower arterial PO₂. Alternatively, the origin of the hypoxic ROS observed in RASMC may derive either from a different portion of the mitochondrial ETC or from a nonmitochondrial oxidase, such as NOX. To determine the degree of variation between these cell types, we compared the substructure of Complex I in PASMC and RASMC (Figure 4). Multiple protein subunits of Complex I are differentially expressed between PASMC and RASMC. PASMC exhibit increased abundance of Fe-S center proteins Ndufs1 (79 kDa), Ndufs2 (MW: 49 kDa) and Ndufv1 (51 kDa), as well as structural subunits NDUFA9 and MT-ND1 (MW: 36-40 kDa). Subunits Ndufs1 and Ndufv1 reside in the N module of Complex I and are responsible for the binding and oxidation of NADH. NDUFA9 is also known to harbor an NADH/NADPH site, whereas MT-ND1 is situated at the edge of the P module, and is responsible for anchoring the Complex within the inner mitochondrial membrane and pumping protons^44-46. We focused on Ndufs2 because of its role in electron transfer to ubiquinone and in rotenone binding^{29, 31}. Rotenone inhibits the ETC directly by blocking the quinone-binding site, inhibiting electron transfer to ubiquinone^{31, 47}. Additionally, we were guided by the work of Lopez-Barneo et al. who identified a key role for Ndufs2 in oxygen sensing in the carotid body. Mice lacking Ndufs2 selectively lost responsiveness to hypoxia and did not exhibit a hypoxic hyperventilatory response. Moreover, glomus cells in the carotid body that lacked Ndufs2 were unable to modulate the Kv channels responsible for membrane depolarization and subsequent neurotransmitter release, and exhibited a more depolarized resting membrane potential compared to control, indicative of an “hypoxic phenotype”³⁴.

Acute hypoxia and rotenone decrease mitochondrial H₂O₂ production in isolated PASMC, consistent with Complex I serving as a critical component of the oxygen sensor (Figures 2-3). We used siRNA knockdown of Ndufs2 to determine whether it is critical to pulmonary vascular oxygen-sensing. Ndufs2 knockdown in isolated rat PASMC eliminated the hypoxia-induced rise in [Ca²⁺]_i, but had no effect on KCl-induced increases in [Ca²⁺]_i (Figure 4I-4J, Online Figure III), indicating that siNdufs2 impairs the sensor and metabolic functions of mitochondria, leaving downstream, oxygen-insensitive, ionic mechanisms of vasoconstriction intact. Silencing Ndufs2 also reduced normoxic H₂O₂ production. The parallel effects of rotenone and authentic hypoxia on H₂O₂ production, cytosolic calcium, and vascular tone support the contention that the loss of HPV caused by siNdufs2 reflects interruption of the PASMC’s vascular oxygen-sensor mechanism, schematized in Figure 8D. Inhibition of ETC Complex I, whether caused by hypoxia, pharmacological or molecular inhibition of Ndufs2 results in a concordant reduced SMC redox state and elicits pulmonary vasoconstriction.

Additional mitochondrial ETC subunits have been proposed as acute oxygen sensors by other groups, who report increased ROS production during hypoxia at both Complex III^{9, 42, 43} and Complex IV³⁵, as well as proposing separate mechanisms for acute and chronic hypoxia. Inhibition of ETC Complex IV has been shown to increase normoxic ROS production in perfused rat lungs¹⁰. However, reports of HPV attenuation during Complex IV subunit knockdown³⁵ conflict with previous reports demonstrating that HPV is unaltered by Complex IV inhibition (with cyanide), whereas HPV is inhibited by Complex I inhibition (with rotenone^{3, 10}). Current literature also supports the conclusion that acute and chronic hypoxic oxygen-sensing are a spectrum of the same mechanism^{14, 48-50}, a conclusion supported by the data herein demonstrating downregulation of both Ndufs2 and Kv1.5 during chronic hypoxia (Figure 6). We show that selective, effective knockdown of putative oxygen sensors in Complexes III and IV does not inhibit hypoxia-induced increases in intracellular Ca²⁺ (Online Figure V). It is unlikely that Complex I inhibition is indirectly inhibiting HPV by depriving distal complexes of electrons required to generate a hypoxic ROS signal, since direct knockdown of these distal complexes does not affect acute oxygen-sensing. This interpretation is further supported by the maintenance of hypoxia-induced increases in [Ca²⁺]_i influx after knockdown of a different subunit of Complex I (Online Figure V) and a lack of change in Complex IV activity in siNdufs2 rat PASMC (Figure 4N).

To translate our cellular and organ-level observations, we assessed a transient knockdown of the putative O₂ sensor by silencing Ndufs2 in vivo and ex vivo (Figure 5, Movies MI-MII). Nebulization of siNdufs2 resulted in an ~35% knockdown of subunit expression in whole rat lung homogenate; this partial knockdown was sufficient to eliminate HPV. Similarly, siNdufs2 attenuated HPV in healthy mice, as measured via intravital microscopy and confocal imaging of excised mouse lungs. These in vivo results across multiple models provide robust evidence that Ndufs2 is both a master regulator of Complex I activity, and an integral component of the O₂-sensing pathway in the pulmonary vasculature.

We have demonstrated that rotenone mimics acute hypoxia in vitro, ex vivo, and in vivo, confirming previously published observations from our group and others^{10, 14, 15, 51, 52}. While acute inhibition of Ndufs2 by rotenone mimics acute hypoxia, chronic downregulation of Ndufs2 by siRNA mimics chronic hypoxia (Figures 6-7, Online Figure VI). It is well-documented that chronically hypoxic PA cells and arteries have a blunted constrictor response to both acute hypoxia and rotenone, whereas the constrictor response to other agonists, such as phenylephrine and KCl is preserved or increased¹⁴. Additionally, in chronic hypoxia, the mitochondria no longer make sufficient H₂O₂ to dynamically modulate HPV^{14, 17}. This is consistent with the selective impairment of HPV which occurs following exposure of rats to chronic hypoxia, a phenomenon first described by McMurtry et al⁵³. Impaired acute HPV is also seen in patients with chronic lung disease and in residents of high altitude that have undergone adaptation⁵³. We found that chronic hypoxia and siNdufs2 had similar effects with respect to Complex I activity, H₂O₂ levels, NADH concentrations and Kv1.5 expression in both rat and human PASMC (Figure 6). Additionally, we show both ex vivo and in vivo that attenuated HPV in chronic hypoxia is due, in part, to reduced expression of both the key oxygen sensor (Ndufs2) and reduced production of H₂O₂, as well as downregulation of an important downstream effector of HPV, the Kv1.5 channel (Figures 1, 6-7). These results parallel our previous observations, wherein silencing mitochondrial superoxide dismutase 2 (SOD2) in rat PASMC reduced H₂O₂ levels, resulting in normoxic stabilization of HIF-1α and reduced expression of effector Kv channels⁵⁴. Thus, loss of Ndufs2 expression contributes to impaired oxygen-sensing in chronic hypoxia.

Oxygen-sensing is also impaired in numerous pathologies, including pulmonary hypertension^{48, 55}. While serial administration of Ndufs2 siRNA mimicked many aspects of chronic hypoxia, it did not induce hypoxic pulmonary hypertension. However, chronic hypoxia is a more multifaceted stimulus than Ndufs2 knockdown, resulting in polycythemia, endothelial dysfunction, and autonomic activation. Thus, it is not expected that a brief and partial suppression of the oxygen sensor would engender full blown chronic hypoxic pulmonary hypertension. In summary, we have identified Ndufs2 as a redox-sensitive pulmonary vascular oxygen sensor, and hydrogen peroxide (H₂O₂) as the diffusible mediator involved in HPV (Figure 8).

Limitations.

We recognize that Ndufs2-deficient animals are available and may provide greater specificity and magnitude of gene silencing compared to siRNA. However, currently available Ndufs2-knockout nice are embryonic or pre-weaning lethal. The knockout mouse used by Lopez-Barneo has a carotid body specific Ndufs2 knockout³⁴. Mice with a tamoxifen-induced global Ndufs2 knockout die of pleiotropic organ failure before 4 months of age³⁴. Therefore, none of the available knockout mice are useful for studying HPV. Further research is required to create a healthy, lung-specific Ndufs2 knockout mouse. We also recognize that Ndufs2 simultaneously regulates cellular redox status while being itself redox-regulated. Thus, there are likely other key players contributing to the redox modulation of Ndufs2. Further investigation will be required to determine the specific mechanism driving changes in Ndufs2 thiol redox status during acute hypoxia. We speculate that the observed reductive shift in Complex I’s electron donors may lead to a reduction of Ndufs2 and contribute to functional inhibition of this subunit. Indeed, our original redox theory in the 1980s highlighted the interplay between cytosolic redox balance and mitochondrial ROS production, with cytosolic redox becoming reduced and ROS decreasing with physiologic levels of hypoxia^{10, 37}.

There are limited tools to accurately quantify ROS generation and these measurements are especially difficult in isolated organelles. However, using the Amplex Red assay kit, with appropriate positive and negative controls, our data clearly indicate that H₂O₂ is being produced by these isolated lung mitochondria and the organelles remain intact after the incubation, as evidenced by our included Complex I activity data. Moreover, the ability of mitochondria to produce diffusible H₂O₂ is not in question in the literature. Mitochondria were recognized by Boveris and Chance in 1972-1973³⁹ as an active site of H₂O₂ production. Boveris summarized this, writing: “Mitochondria isolated from rat heart, liver, kidney and brain (respiratory control 4.0-6.5) release NO and H₂O₂ at rates that depend on the mitochondrial metabolic state⁵⁶. Other groups, working in other organs using other measurement assays also find mitochondria generate diffusible H₂O₂ (such as Gill et al in T-cells⁵⁷ and Lee et al in neurons⁵⁸).

NOVELTY & SIGNIFICANCE.

What Is Known?

• Acute hypoxia induces a pulmonary vasoconstriction.

• Mitochondria in pulmonary artery smooth muscle cells (PASMC) of small pulmonary arteries (PAs) are recognized as the pulmonary the oxygen-sensor.

• The sensor resides within the electron transport chain (ETC) and acts by altering production of diffusible signaling molecules [reactive oxygen species (ROS)] in proportion to FiO2.

• The specific ETC protein(s) responsible for oxygen-sensing remain unclear.

What New Information Does This Article Contribute?

• PASMC mitochondria uniquely produce hydrogen peroxide (H₂O₂), a normoxic vasodilator, in proportion to PO₂.

• Expression of the mitochondrial Complex I subunit, Ndufs2, is upregulated in PASMC relative to renal SMC.

• Ndufs2 is necessary for oxygen-sensing in isolated PASMC and Ndufs2 knockdown reduces the production of H₂O₂ and inhibits hypoxic pulmonary vasoconstriction (HPV) in rats and mice in vivo.

• siNdufs2 mimics chronic hypoxia, eliciting a pseudo-hypoxic phenotype in PASMC, with suppressed oxidative metabolism, a reduced redox state (low levels of H₂O₂ and an increased NADH/NAD⁺ ratio), decreased expression of Kv1.5 channels and suppression of acute HPV.

The identity of the specific protein(s) responsible for oxygen-sensing in the lung have remained elusive for decades. While there is agreement in the literature that mitochondria are a key player in HPV, the specific protein subunits that serve as oxygen sensor(s) have been subject to significant debate. In this study, we have demonstrated for the first time, that Ndufs2, a protein subunit of ETC Complex I is required for oxygen-sensing in the pulmonary vasculature. Silencing Ndufs2 in vitro inhibits hypoxia-induced ROS signaling and calcium uptake, as well as attenuating hypoxic PA vasoconstriction in vivo. These findings identify Ndufs2 as a redox-sensitive master regulator of pulmonary oxygen-sensing, mitochondrial activity and metabolism and mirror recent reports identifying Ndufs2 as the oxygen-sensor in the carotid body, another oxygen-sensitive tissue. . Modulating Ndufs2 expression may provide new avenues for treatment of diseases characterized by impaired oxygen-sensing, including pulmonary hypertension and high-altitude pulmonary edema.

Nonstandard Abbreviations and Acronyms:

ΔΨ

membrane potential

ΔΔC_T

double delta cycle threshold

AII

angiotensin 2

C-Hx

chronic hypoxia (within figures)

Ca²⁺_(i)

(intracellular) calcium

CAC

citric acid cycle

Ca_L

L-type calcium channel

Cat

catalase

CB

carotid body

CH

chronic hypoxia (or chronic hypoxic), within text

Cx

complex (used to abbreviate “electron transport chain complex” in Figure 8)

DIC

differential interference contrast (microscopy)

DTT

dithiothreitol

EGTA

ethylene glycol-bis(β-aminoethyl ether)-N,N,N’,N’-tetraacetic acid

ETC

electron transport chain

FADH₂

flavin adenine dinucleotide

H₂O_2(mito)

(mitochondrial) hydrogen peroxide

HAPE

high altitude pulmonary edema

HIF-1α

hypoxia-inducible factor one-alpha

hLM

hypoxic lung mitochondria

HPV

hypoxic pulmonary vasoconstriction

Hx

hypoxia

Iono

ionomycin

KCl

potassium chloride

KM

kidney mitochondria

Kv1.5/2.1

voltage gated potassium channel 1.5/2.1

MCM

mitochondria conditioned media

NAD⁺

nicotinamide adenine dinucleotide (oxidized)

NADH

nicotinamide adenine dinucleotide (reduced)

NADPH

nicotinamide adenine dinucleotide phosphate (reduced)

Ndufs2

NADH dehydrogenase [ubiquinone] iron-sulfur protein 2

nLM

normoxic lung mitochondria

Nx

normoxia

O₂-sensing

oxygen-sensing

OCR

oxygen consumption rate

PA(SMC)

pulmonary artery (smooth muscle cell)

RA(SMC)

renal artery (smooth muscle cell)

roGFP

reduction-oxidation sensitive green fluorescent protein

Rote

rotenone

RPL32

ribosomal protein L32

siNdufs2

cells/tissue treated with Ndufs2 siRNA

SMA

smooth muscle actin (α)

SOD2

superoxide dismutase 2

tBOOH

tert-butyl hydroperoxide

TOMM20

translocase of outer mitochondrial membrane 20

Ubiq

ubiquinone