Studies on Hypoxic Pulmonary Vasoconstriction Detect a Novel Role for the Mitochondrial Complex I Subunit Ndufs2 in Controlling Peroxide Generation for Oxygen-Sensing

The molecular mechanisms for the oxygen sensor that appears to be present in pulmonary arterial smooth muscle cells (PASMC) mediating hypoxic pulmonary vasoconstriction (HPV) has been the focus of extensive research and remains controversial (1,2). In this issue of Circulation Research, Dunham-Snary et al (3) describe a novel involvement of the mitochondrial electron transport chain (ETC) rotenone binding site Complex I subunit, NADH dehydrogenase [ubiquinone] iron-sulfur protein 2 (Ndufs2), in controlling an oxygen dependent vasodilator source of peroxide generation that appears to contribute to acute hypoxic pulmonary vasoconstriction (HPV).

In this study (3), silencing Ndufs2 with in vivo lung targeted siRNA treatments attenuated acute HPV and rotenone-induced vasoconstriction, and enhanced vasoconstriction to phenylephrine. Remarkably, siRNA depletion in cultured PASMC of Ndufs2 (but not siRNA depletion of other key Fe-S subunits in complex I (Ndufs1) or other potential oxygen sensors in Complex III (Rieske Fe-S), or Complex IV (heme containing cytochrome oxidase subunit 4i2)) attenuated hypoxia-elicited increases in PASMC intracellular calcium levels. Exposure of PASMC to hypoxia was associated with evidence for the detection of decreases in cytosolic and mitochondrial peroxide by thiol oxidation of the HyPer protein detectors targeting both mitochondrial and cytosolic regions. In contrast, renal artery smooth muscle cells showed hypoxia-elicited decreases in intracellular calcium and increases only in cytosolic oxidation of the HyPer protein detector. Perfusion of lungs with conditioned media from mitochondria isolated from lungs (but not from kidneys which show lower levels of Ndufs2) rapidly attenuated the acute HPV response in a manner dependent on the mitochondrial release of hydrogen peroxide. Interestingly, the siRNA depletion of Nfuds2 appeared to decrease PASMC peroxide release and mitochondrial respiration, along with elevation of NADH, without inhibiting NADH-dependent electron transfer (to nitroblue tetrazolium) by Complex I or depleting multiple other ETC subunits. These results together with the absence of effects of the other mitochondrial ETC subunits, suggests that the HPV responses studied are potentially most dependent on hypoxia decreasing hydrogen peroxide, and perhaps minimally dependent on mitochondrial NADH redox, ETC activity or energy metabolism to support changes in intracellular calcium or force generation during HPV. Thus, Ndufs2 influences HPV and PASMC increases in intracellular calcium responses to hypoxia in a potentially unique manner consistent with it being a key hypoxia inhibited source of vasodilator levels of hydrogen peroxide under the conditions studied. The study of Dunham-Snary et al (3) also documents that this property of lung-derived mitochondria is not seen in mitochondria derived from kidneys, supporting specialization of the HPV mechanism for controlling the matching of lung ventilation to perfusion. Interestingly, chronic hypoxia associated with pulmonary hypertension development showed effects similar to the silencing of Ndufs2.

This work evolved from early studies by Archer et al. (4) documenting that hypoxia, and the mitochondrial electron transport chain inhibitor rotenone promoted pulmonary vasoconstriction associated with decreasing detection of reactive oxygen species (ROS) and a closure potassium channels. Similarities in properties of oxygen sensing mechanisms between HPV with the carotid body, together with recent evidence (5) for the mitochondrial Complex I subunit Ndufs2 having a critical role in the carotid body sensing of hypoxia contributed to development of novel evidence in the current study for Ndufs2 regulating hypoxia-elicited decreases in H₂O₂ as an oxygen sensing mechanism in HPV. One key difference in the carotid body study is that hypoxia appears to be increasing ROS in a Ndufs2-dependent manner. The observations of rotenone and antimycin A decreasing ROS in the rat pulmonary vasculature by Archer (2) and by our own group in bovine pulmonary arteries (6) was initially difficult to rationalize based on what was known at the time about actions of these mitochondrial ETC inhibitors. This is because rotenone was thought to increase ROS production by Complex I and decrease their production by Complex III of the ETC, whereas antimycin was thought to increase ROS production by these sites. Moreover, the studies of Schumaker’s group primarily in pulmonary artery-derived smooth muscle cells showed evidence for hypoxia increasing mitochondrial-derived ROS originating from complex III based on the predicted actions of these and other mitochondrial ETC inhibitors at the time (2). This work evolved into evidence for the Rieske Fe-S protein (which removes an electron from ubiquinol (QH₂) at the Qo site in Complex III) forming a transient free radical ubisemiquinone (Q^.-) which potentially reacts with oxygen to form superoxide (2). While the literature contains minimal evidence for a specific role of Ndufs2 in controlling mitochondrial ROS, Bland and colleagues defined in skeletal muscle mitochondria a novel site of superoxide production inhibited by rotenone in the region of Complex I (termed I_Q), associated with the site of binding and electron transfer to ubiquinone (Q) (7). This site appeared to participate in superoxide generation from reverse electron transport from Complex II to Complex I promoted by succinate dehydrogenase that is inhibited by rotenone under conditions of “high protomotive force” or a large pH gradient across the inner mitochondrial membrane that produces mitochondrial hyperpolarization (See Figure 1). While the mechanism of superoxide generation by I_Q is not known, rotenone binds to the Q binding site in the Ndufs2 subunit of Complex I. High levels of QH₂, the two electron reduced form of Q promotes superoxide generation by I_Q into the mitochondrial matrix (7,8), and this would result in the release of hydrogen peroxide due to the high activity of mitochondrial matrix superoxide dismutase 2 (SOD2). In the current study, Archer’s group has continued to document differences in the expression of key ETC components and properties in lung versus kidney mitochondria that may help explain how increased Ndufs2 may function in pulmonary arterial smooth muscle ROS generation. Their studies provide evidence for elevated levels of expression of multiple subunits of Complex I in addition to Ndufs2 (3) and mitochondrial matrix SOD2 expression (9), which may also contribute to the detected markedly elevated levels of mitochondrial peroxide release seen in PASMC mitochondria.

Fig. 1.

Model showing the site of Ndufs2 in Complex I of the mitochondrial electron transport chain illustrating the pathways for electron transfer and redox forms of ubiquinone (Q) discussed in this article that potentially participate in its hypothesized oxygen sensing role through controlling the generation and release of a source of vasodilator levels of mitochondrial hydrogen peroxide (H₂O₂) which are removed by hypoxia to produce hypoxic pulmonary vasoconstriction. Based on studies in skeletal muscle mitochondria, metabolic-redox bioenergetic conditions promoting high levels of ubiquinol (QH₂) binding the Ndufs2 subunit or I_Q region of Complex I, and a high H⁺ gradient associated with mitochondrial hyperpolarization may be key factors enabling this site to reduce oxygen to superoxide and promote levels of peroxide that function in oxygen sensing mechanisms such as the removal of a peroxide-mediated vasodilation in hypoxic pulmonary vasoconstriction. The star symbols in the figure show approximate locations of sites of superoxide generation in the mitochondrial electron transport chain discussed in this article.

There are many observations on roles for different sources and actions of superoxide and hydrogen peroxide in vascular oxygen sensing mechanisms (1,2,10). Studies on skeletal muscle mitochondria have detected strong evidence for changes in the generation of superoxide or hydrogen peroxide from multiple mitochondrial sites under different bioenergetic conditions seen in physiology and/or pathophysiology (8). Thus, differences in vascular oxygen sensing mechanisms reported to be associated with detected changes in ROS or related metabolic redox systems may originate from environmental adaptations which can shift aspects of the mechanisms studied. Factors such as diet, exercise, hypoxia, elevated vascular pressure, exposure to mediators influencing vascular function, cell culture conditions, etc. may shift the properties of vascular oxygen sensing mechanisms. Many additional sources of ROS and metabolic-redox processes contribute to the regulation of pulmonary vascular function and the progression of pathophysiological changes that could potentially influence oxygen sensing mechanisms (10). The study of Dunham-Snary et al (3) also provides evidence for a new example where chronic hypoxia promotes a loss of the HPV response associated with decreased PASMC Ndufs2 and mitochondrial peroxide release.

Evidence reported in this study supporting the novel role of Ndufs2 in the oxygen dependence of mitochondrial hydrogen peroxide generation has implications that may markedly influence aspects of future work in this controversial field. For example, it is not known if reverse electron transport from QH₂ generation by Complex II to the Ndufs2 site in Complex I is a key factor controlling when oxygen is being sensed at this site through superoxide and/or peroxide generation. Based on Ndufs2 controlling a hypoxia-mediated increase in ROS in the carotid body (5), compared to the decrease seen in the Archer study (3), there may be fundamental differences in the function of metabolic-redox systems and/or components of the ETC between these two cell types that remain to be defined, and defining these differences may help in resolving controversy in HPV oxygen sensing mechanisms field. Ndufs2 may also have additional roles in oxygen sensing by other vascular or non-vascular systems which remain to be identified. Lung perfusion studies showing a potent reversal of HPV by peroxide released from lung mitochondria (3) suggest that increased mitochondrial hydrogen peroxide release may be a fundamental property of multiple cell types in the lung. This could enable extracellular peroxide derived from these other cells to influence HPV and other aspects of pulmonary vascular function. Ndufs2 is shown to undergo thiol redox changes in PASMC with hypoxia that could be a factor in how it functions, or is subject to oxidant or metabolic conditions seen in pathophysiology such as iron-derived highly reactive oxidant species or peroxynitrite. The development of an understanding of how Ndufs2 functions and is altered could also provide insight into metabolic-redox modulating therapeutic approaches (10) which could help restore beneficial oxygen sensing or peroxide regulated processes.