Hypoxic pulmonary vasoconstriction in the absence of pretone: essential role for intracellular Ca²⁺ release

Abstract

Hypoxic pulmonary vasoconstriction (HPV) maintains blood oxygenation during acute hypoxia but contributes to pulmonary hypertension during chronic hypoxia. The mechanisms of HPV remain controversial, in part because HPV is usually studied in the presence of agonist-induced preconstriction (‘pretone’). This potentiates HPV but may obscure and distort its underlying mechanisms. We therefore carried out an extensive assessment of proposed mechanisms contributing to HPV in isolated intrapulmonary arteries (IPAs) in the absence of pretone by using a conventional small vessel myograph. Hypoxia elicited a biphasic constriction consisting of a small transient (phase 1) superimposed upon a sustained (phase 2) component. Neither phase was affected by the L-type Ca²⁺ channel antagonists diltiazem (10 and 30 μm) or nifedipine (3 μm). Application of the store-operated Ca²⁺ entry (SOCE) blockers BTP2 (10 μm) or SKF96365 (50 μm) attenuated phase 2 but not phase 1, whereas a lengthy (30 min) incubation in Ca²⁺-free physiological saline solution similarly reduced phase 2 but abolished phase 1. No further effect of inhibition of HPV was observed if the sarco/endoplasmic reticulum Ca²⁺-ATPase inhibitor cyclopiazonic acid (30 μm) was also applied during the 30 min incubation in Ca²⁺-free physiological saline solution. Pretreatment with 10 μm ryanodine and 15 mm caffeine abolished both phases, whereas treatment with 100 μm ryanodine attenuated both phases. The two-pore channel blocker NED-19 (1 μm) and the nicotinic acid adenine dinucleotide phosphate (NAADP) antagonist BZ194 (200 μm) had no effect on either phase of HPV. The lysosomal Ca²⁺-depleting agent concanamycin (1 μm) enhanced HPV if applied during hypoxia, but had no effect on HPV during a subsequent hypoxic challenge. The cyclic ADP ribose antagonist 8-bromo-cyclic ADP ribose (30 μm) had no effect on either phase of HPV. Neither the Ca²⁺-sensing receptor (CaSR) blocker NPS2390 (0.1 and 10 μm) nor FK506 (10 μm), a drug which displaces FKBP12.6 from ryanodine receptor 2 (RyR2), had any effect on HPV. HPV was virtually abolished by the rho kinase blocker Y-27632 (1 μm) and attenuated by the protein kinase C inhibitor Gö6983 (3 μm). Hypoxia for 45 min caused a significant increase in the ratio of oxidised to reduced glutathione (GSSG/GSH). HPV was unaffected by the NADPH oxidase inhibitor VAS2870 (10 μm), whereas phase 2 was inhibited but phase 1 was unaffected by the antioxidants ebselen (100 μm) and TEMPOL (3 mm). We conclude that both phases of HPV in this model are mainly dependent on [Ca²⁺]_i release from the sarcoplasmic reticulum. Neither phase of HPV requires voltage-gated Ca²⁺ entry, but SOCE contributes to phase 2. We can detect no requirement for cyclic ADP ribose, NAADP-dependent lysosomal Ca²⁺ release, activation of the CaSR, or displacement of FKBP12.6 from RyR2 for either phase of HPV. Sustained HPV is associated with an oxidising shift in the GSSG/GSH redox potential and is inhibited by the antioxidants ebselen and TEMPOL, consistent with the concept that it requires an oxidising shift in the cell redox state or the generation of reactive oxygen species.

Key points

• Hypoxic pulmonary vasoconstriction (HPV) is a mechanism by which pulmonary arteries maintain blood oxygenation during alveolar hypoxia.

• HPV is generally studied using a vasoconstricting co-stimulus that amplifies the HPV but may also distort its properties; we therefore characterised HPV in isolated rat intrapulmonary arteries during 40 min hypoxic challenges in the absence of any such stimulus.

• Immediate (phase 1) and sustained (phase 2) components of HPV were unaffected by blocking voltage-gated Ca²⁺ channels but were abolished by depletion of sarcoplasmic reticulum Ca²⁺. Phase 2 was attenuated by blockade of store-operated Ca²⁺ entry (SOCE), although it largely persisted in Ca²⁺-free physiological saline solution.

• HPV was associated with an increase in the intrapulmonary artery ratio of oxidised to reduced glutathione and was inhibited by antioxidants.

• HPV resulted primarily from intracellular Ca²⁺ release, with SOCE making a contribution, particularly to phase 2. Sustained HPV involves oxidation of the pulmonary artery redox state.

Introduction

Pulmonary arteries constrict to hypoxia. This phenomenon, termed hypoxic pulmonary vasoconstriction (HPV), acts to divert the flow of deoxygenated blood away from hypoxic regions of the lung, thus matching ventilation to perfusion. However, in the face of global hypoxia, occurring for example in chronic obstructive pulmonary disease and during sleep apnoea, HPV contributes to an increase in pulmonary vascular resistance which can lead to right heart failure (Ward & McMurtry, 2009). Insights into HPV have largely emerged from studies employing isolated IPAs (Leach et al. 1994; Jabr et al. 1997; Robertson et al. 2000a,b, 2008), buffer-perfused lungs (McMurtry et al. 1976; Weigand et al. 2005; Weissmann et al. 2006a,b), or cultured pulmonary artery smooth muscle cells (PASMCs; Salvaterra & Goldman, 1993; Cornfield et al. 1994; Ng et al. 2005; Wang et al. 2005, 2012a). As a result, a number of mechanisms have been proposed to explain how hypoxia causes contraction in IPAs, although with little agreement as to which of these is of primary importance. Potential effector mechanisms include voltage-gated Ca²⁺ influx (Weir & Archer, 1995), Ca²⁺ release from the sarcoplasmic reticulum (SR; Jabr et al. 1997), store-operated Ca²⁺ entry (Ng et al. 2005; Weigand et al. 2005) and rho kinase (ROK)-mediated Ca²⁺ sensitisation (Robertson et al. 2000a). SR Ca²⁺ release has been suggested to be mediated by cyclic ADP ribose (cADPR; Dipp & Evans, 2001; Wilson et al. 2001), possibly acting in concert with lysosomal Ca²⁺ release leading to Ca²⁺-induced Ca²⁺ release (Evans, 2010). Ca²⁺ release has alternatively been proposed to be due to activation of the CaSR (Zhang et al. 2012) or the displacement of FKB12.6 from the ryanodine receptor (Liao et al. 2011). Similarly, the mechanisms by which PASMCs sense a fall in the O₂ concentration remain controversial, with evidence having been put forward that hypoxia is detected as a fall in cellular reactive oxygen species (ROS) concentration due to the lack of O₂ (Weir & Archer, 1995), a rise in [ROS] generated by the mitochondria (Waypa et al. 2001) and NADPH oxidase (Weissmann et al. 2006b; Rathore et al. 2008), or the activation of AMP-activated protein kinase (AMPK) by a hypoxia-induced increase in the [AMP]:[ATP] ratio (Evans et al. 2005; Evans, 2006). Additional controversy concerns whether HPV can persist in the absence of extracellular Ca²⁺ (Hoshino et al. 1988; Leach et al. 1994).

Crucially, when studying HPV in arteries and perfused lungs (and sometimes cells), most investigators have applied hypoxia in the ongoing presence of a low concentration of a vasoconstrictor such as angiotensin II or PGF_2α (see review by Sylvester et al. 2012), the rationale for this being that HPV is typically very small or unstable in preparations in the absence of such ‘pretone’. A notable exception to this is a series of reports from Evans’ laboratory that described HPV in non-prestimulated small pulmonary arteries from rabbits or rats (Dipp et al. 2001; Dipp & Evans, 2001; Wilson et al. 2001). The results of these studies, which proposed a role for novel mechanisms for oxygen sensing and [Ca²⁺] regulation in HPV, have not yet been independently verified.

As well as being useful methodologically, the employment of exogenously imposed pretone for examining HPV in in vitro preparations is probably physiologically relevant, since it is very likely that some form of HPV-enhancing pretone exists in vivo. It is known, for example, that exercise results in a pronounced fall in pulmonary vascular resistance in humans which is only partially accounted for by the recruitment of areas of the pulmonary vascular bed which are not perfused at rest (Merkus et al. 2008), this implying that some pulmonary vascular tone must exist under basal conditions. It has, for example, been proposed that endothelin may provide such pretone in vivo (Sato et al. 2000). Moreover, the addition of blood to the physiological saline solution perfusing isolated lungs has been shown to strongly potentiate HPV (McMurtry, 1984).

However, the nature of pretone in vivo remains unknown, and this has unfortunate implications for the use of pretone to study HPV in vitro since its properties may be strongly affected by the nature and concentration of the particular pretone agent that is employed. This is probably because there is an overlap between the constricting mechanisms activated by pretone agents and by hypoxia (Ng et al. 2005; Aaronson et al. 2006). Therefore, a consequence of the use of pretone is that the ‘pure’ or ‘intrinsic’ response of pulmonary arteries to hypoxia, i.e. that which would occur in the absence of potential modulatory factors, remains relatively unexplored.

It is also noteworthy that mechanistic studies of HPV in PASMCs, isolated IPAs, perfused lungs or in vivo have generally used relatively brief (≤10 min) exposures to hypoxia. HPV is typically seen to reach a peak and then wane during such studies (e.g. Weissmann et al. 1995, 2004; Weigand et al. 2005). Conversely, investigations using longer periods of hypoxia in perfused lungs (Emery et al. 2003; Weissmann et al. 2006a,b) and isolated arteries (e.g. Leach et al. 1994; Dipp et al. 2001) show that this transient phase of HPV is followed by a sustained second phase, and there is evidence that HPV is also biphasic in humans in vivo (e.g. Talbot et al. 2005). Because this sustained contraction has seldom been studied, the mechanisms causing sustained HPV remain particularly obscure.

The few investigations in which sustained (i.e. phase 2) HPV in isolated IPAs has been elicited in the absence of preconstriction have indicated that intracellular Ca²⁺ release is the primary factor underpinning the contraction (Hoshino et al. 1988; Dipp et al. 2001). However, these results remain unconfirmed and are difficult to reconcile with observations made in perfused lungs (usually in the presence of angiotensin II) and cultured cells, which indicate a pivotal role for Ca²⁺ influx (McMurtry et al. 1976; Archer et al. 1995; Ng et al. 2005; Wang et al. 2005, 2012a; Weigand et al. 2005).

Since such discrepancies might arise from distortion of the response by pretone in perfused lungs or by phenotypical changes occurring as a result of cell culture (Ng et al. 2008, see also discussion in Ward & McMurtry, 2009), we have undertaken extensive studies of HPV in isolated rat IPAs in order to determine its properties in the absence of pretone, using hypoxic challenges of 40 min to allow evaluation of the mechanisms of sustained HPV. Although the use of mouse IPAs for such studies would have had obvious advantages, we used the rat because it is difficult to elicit HPV even with pretone in mouse IPAs (Sylvester et al. 2012). Our results indicate that sustained HPV is crucially dependent on intracellular Ca²⁺ release from the SR, but that a number of mechanisms which have been proposed to cause this release, including cADPR (Wilson et al. 2001), lysosomal Ca²⁺ release (Evans, 2010), and activation of the CaSR (Zhang et al. 2012) are not involved.

Methods

Ethical approval

This study conforms with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 85–23, revised 1996) and is in accordance with UK Home Office regulations (Animals (Scientific Procedures) Act, 1986). Male Wistar rats (200–300 g; 6–8 weeks old) were killed by lethal injection (i.p.) of sodium thiopental.

IPA mounting and measurement of tension development

The heart and lungs were excised and placed in cold physiological salt solution (PSS), which contained (mm): 118 NaCl, 24 NaHCO₃, 1 MgSO₄, 0.435 NaH₂PO₄, 5.56 glucose, 1.8 CaCl₂, and 4 KCl. Rings of intrapulmonary artery (IPA; inner diameter 0.5–1.0 mm) were dissected free of adventitia and parenchyma under a dissection microscope, mounted on a conventional small vessel wire myograph and stretched to give a basal tension of ∼3 mN (equivalent to an internal pressure of ∼7 mmHg). They were then equilibrated with three brief exposures to PSS containing 80 mm KCl (80KPSS; isotonic replacement of NaCl by KCl). Hypoxia was induced by switching from 95% air/5% CO₂ to 0% O₂/5% CO₂/balance N₂, the latter of which we have previously shown gives a of 15–20 mmHg during hypoxia in our myograph chambers (Robertson et al. 2000b). All experiments were conducted at 37°C.

Measurements of reduced and oxidised glutathione

The measurement of GSH and GSSG was made with the method of Griffith (1980) as described in detail (Gonzalez et al. 2004). Spontaneous reduction of 5,5′-dithiobis-(2-nitrobenzoic acid) (DTNB) by GSH generates 5-thio-2-nitrobenzoic acid (TNB; peak absorbance at 412 nm) and GSSG. The GSSG formed is back-reduced enzymatically to GSH by glutathione reductase coupled to NADPH oxidation. This enzymatic step makes the assay highly specific, and cycling renders it very sensitive (10⁻¹⁰ mol assay⁻¹). Reactant concentrations (DTNB, NADPH and glutathione reductase) were selected to obtain a linear rate of colour formation during 2–3 min. The slope of the linear colour formation (absorbance min⁻¹) is directly proportional to the glutathione in the sample. Assaying standards with different amounts of GSH and plotting absorbance change as a function of time yield several lines with different slopes. The plot of slopes of these lines as a function of the GSH used is also linear and defines the standard curve. This standard curve allows calculation by interpolation of the concentrations of total glutathione (GSt = GSH + GSSG) in the experimental samples. To specifically determine GSSG, the GSH present in the supernatants is destroyed with 2-vinylpyridine and then GSSG is assayed as above, except for the use of higher glutathione reductase amounts due to the lower concentration of GSSG in the tissues. Standards and tissue samples were assayed in duplicate. In tissue samples, the concentration of GSH is obtained as the difference GSt − GSSG. The standard curves to measure GSt and GSSG are usually prepared with GSH because the enzymatic cycling makes it irrelevant whether one starts with GSH or GSSG. It should be noted that each molecule of GSSG yields two molecules of GSH and therefore, when we determined GSt in the tissue we were actually measuring GSH + 0.5(GSSG), and when we were measuring GSSG we were in fact measuring half its molar concentration. Appropriate corrections were made when plotting the data. Glutathione redox potential (E_GSH) was calculated as a Nernst potential (Schafer & Buettner, 2001).

Estimation of changes in NAD(P)H

Changes in NAD(P)H levels were assessed as described in Leach et al. (2001). Briefly, the autofluorescence of IPAs mounted on a myograph was recorded while the preparation was alternately excited with light at 340 and 380 nm while florescence at 500 nm was recorded. This approach depends on the fact that NAD(P)H fluorescence at ∼500 nm is greater when excited at 340 nm than at 380 nm, such that the 340/380 emission ratio is proportional to changes in NAD(P)H. Movement of the preparation does not affect the ratio as it equally affects emission at both wavelengths.

Solutions, drugs and chemicals

Diltiazem, nifedipine, acetylcholine, cyanide, metformin, TEMPOL, dihydrolipoic acid (DHLA), (Sigma-Aldrich, UK), Y-27632 (Calbiochem, UK) and prostaglandin F_2α (PGF_2α; Biomol, UK) and ryanodine (Tocris, UK) were made as stock solutions in distilled water. Compound C, Gö6983, BZ194, concanamycin (Calbiochem, UK), rotenone, antimycin A, 8-bromo-cyclic ADP ribose, VAS2870, FK506 (Sigma, UK), NPS2390 (Tocris), BTP2 (Cayman) and NED-19 (Enzo) were made up as a stock solutions in dimethylsulphoxide (DMSO). Drug concentrations in DMSO stock solutions were such that the highest DMSO concentration in the bath was 0.3%, which itself had no effect on responses; most often the DMSO concentration was 0.1%. EGTA (Sigma-Aldrich, UK) was diluted from a 500 mm stock solution (pH 7.4) in distilled water.

Experimental protocol, data representation and statistical analysis

IPAs were exposed to hypoxia for three 40 min periods, each separated by ∼40 min. Except where otherwise mentioned, the HPV evoked during the 2nd hypoxic challenge served as the control, and experimental interventions (e.g. application of inhibitors or zero Ca²⁺ solution) were made 30 min prior to the 3rd hypoxic challenge. However, in some cases where we wanted to examine the effect of a drug on an ongoing response, drugs were applied starting during the 2nd hypoxic challenge (at 20 min) and remained present during the 3rd hypoxic challenge. For the illustration of results, increases in tension above baseline throughout the 2nd and 3rd hypoxic challenges were expressed as a percentage of the increase in force above baseline recorded at 40 min of hypoxia during the 2nd (control) response (see Fig. 1B). Data were analysed with GraphPad Prism.

Figure 1. Basic properties of HPV in isolated non-preconstricted rat PAs.

A, contractile response of an isolated unstimulated rat PA to three successive hypoxic challenges of 40 min duration. B, average amplitude of HPV over 40 min during 3 successive exposures to hypoxia under control conditions. The magnitude of each HPV was normalised to the amplitude of the 2nd HPV recorded at 40 min in each PA (just before reoxygenation), and the resulting normalised data from 29 IPAs, each from a different animal, are shown (mean ± SEM). Grey circles, 1st HPV; black circles, 2nd HPV; black triangles, 3rd HPV. C, two successive HPVs in the absence of pretone followed by an HPV in the presence of PGF_2α (10 μm; representative of 10 experiments). D, two successive HPVs in the absence of pretone followed by an HPV in the presence of angiotensin 2 (10 nm, representative of 4 experiments). In this and subsequent figures, ‘W’ (wash) indicates replacement of the solution with fresh PSS.

The amplitude of phase 1 HPV was measured as the initial peak in tension development occurring within the first 3 min after the onset of HPV, minus the baseline. In some arteries phase 1 was less prominent and manifested as a shoulder in the response where the initial rapid rise in tension gave way to a slower rise, in which case the level of tension at the point where the slope changed was taken to represent the amplitude of phase 1. The amplitude of phase 2 was taken to be the level of tension above baseline measured after 40 min of hypoxia, immediately before reoxygenation.

In order to analyse the effects of interventions on either phase, we first conducted 29 control experiments, during which hypoxia was imposed 3 times sequentially in PSS (see Figs 1A and B, and 13A). These revealed that the amplitude of HPV was very similar throughout the 2nd and 3rd hypoxic challenges, although there was a small (non-significant) trend for the response to diminish between the 2nd and 3rd responses. The changes in the amplitudes of the two HPV phases between the 2nd and 3rd HPVs were then recorded in experiments in which the 2nd hypoxic challenge was carried out under control conditions and the 3rd hypoxic challenge occurred during an intervention (e.g. in the presence of an inhibitor). The change in amplitude of phases 1 and 2 between the 2nd and 3rd HPVs in the intervention experiments was then compared to the changes recorded in the control experiments using Student's unpaired t test in order to determine whether the intervention had significantly changed either phase, with the threshold for statistical significance being set at P < 0.05. This approach was used to control for the slight intrinsic rundown of the response under control conditions.

Figure 13. Summary of the changes in amplitude of phase 1 and phase 2 between the 2nd (control) and 3rd (intervention) HPVs under the various conditions described in the text.

For purposes of comparison, the leftmost bars in each panel depict the slight (non-significant) fall in the amplitudes of both phases under control conditions. The changes in amplitude of both phases of HPV after the interventions described in the text and Figs 2–11 are shown in the rest of the figure. Asterisks indicate where an intervention applied to the 3rd HPV caused the amplitude of a phase to change from that observed during the 2nd HPV to a significantly (P < 0.05) greater extent than was observed when both HPVs were evoked under control conditions. Related interventions are grouped by panel to facilitate visualisation of the results.

Most figures illustrate the mean (±SEM) amplitude of HPV measured at nine points after hypoxia was imposed. The 1st point represents the amplitude of phase 1 measured as described above, and the next eight points represent the amplitude of HPV measured at 5 min intervals, starting 5 min after hypoxia was initiated. In most cases, the sets of points from the 2nd (control) and 3rd (intervention) HPVs have been superimposed. These results are normalised as described above. Asterisks over the 1st and final points during the hypoxic challenge indicate a significant effect of any intervention (i.e. the change in amplitude of a phase between the 2nd and 3rd HPVs was significantly different to that observed in the 29 control experiments; see above).

Results

Basic properties of HPV in isolated non-preconstricted rat IPA

An example of the contractile response of a rat isolated IPA to three successive 40 min hypoxic challenges is shown in Fig. 1A, whilst Fig. 1B depicts the mean response obtained from 29 such experiments. As described in Methods, the force development during each hypoxic challenge has been normalised to the force development observed at the 40 min point of the 2nd (control) HPV. This normalisation procedure has also been used to produce the averaged results (mean ± SEM) illustrated in subsequent figures, with asterisks over the 1st and final points during the hypoxic challenge indicating a significant effect of any intervention.

Imposition of hypoxia led to a rapidly developing constriction (phase 1) which peaked within 5 min. Tension then fell slightly, after which it tended to increase gradually and become sustained (phase 2). Force development then returned rapidly to baseline upon the restoration of normoxia. There was a tendency for the shape of the response to change between the 1st and 2nd hypoxic challenges, such that the biphasic profile became less prominent, and in some arteries phase 1 appeared as a ‘shoulder’ on the response rather than as a distinct transient contraction during the 2nd exposure to hypoxia. The shape and amplitude of the response then became more stable, so that the 2nd and 3rd HPVs under control conditions did not differ significantly over 40 min. We therefore generally used the 2nd HPVs in each artery as the control and initiated an intervention designed to examine the mechanism (e.g. addition of channel or kinase inhibitor) 30 min before the 3rd HPV, although alternative approaches were used where appropriate.

Figure 1C and D illustrates that HPV observed in the absence of pretone was much smaller than that obtained if pretone was applied with either PGF_2α (Fig. 1C) or angiotensin II (Fig. 1D), which have often been used as a source of pretone. Both stimuli particularly enhanced the initial transient response, which occurred over the duration of the hypoxic challenges generally used to record HPV in perfused lungs and responses in cells (e.g. 5–10 min). The effects on HPV of these two pretone-inducing drugs were, however, not identical, highlighting the importance of evaluating the properties of the response in the absence of any pre-existing stimulus.

The role of Ca²⁺ influx mediated by VGCCs and SOCE

Treatment of IPAs with nifedipine (3 μm) or diltiazem (10 and 30 μm), two structurally dissimilar L-type voltage-gated Ca²⁺ channel (VGCC) antagonists, had no significant effect on either phase of HPV (Fig. 2A, B and C). Addition of these drugs during an ongoing hypoxic challenge, which might more easily reveal a small inhibition, also had no effect (Fig. 2D). In contrast to this lack of effect of Ca²⁺ channel antagonists on HPV in the absence of pretone, when HPV was evoked in the presence of PGF_2α-induced pretone, 3 μm nifedipine almost abolished the much more pronounced phase 1 which occurred under these conditions, although again it had no effect on phase 2 (Fig. 2E).

Figure 2. Effect of VGCC antagonists on HPV with and without pretone.

HPV in the absence of pretone was unaffected by 30 min pretreatment with 3 μm nifedipine (n= 16, n.s.; A), 10 μm diltiazem (n= 9, n.s.; B), or 30 μm diltiazem (n= 8; C). D, addition of diltiazem (10 and 30 μm, upper panel) or nifedipine (1 or 10 μm, lower panel) during phase 2 had no effect on tension (n= 1 for each). E, effect of 3 μm nifedipine on HPV in the presence of a concentration of PGF_2α sufficient to evoke a contraction of 10–15% of that caused by 80 mm K⁺ solution (n= 9). For all panels *indicates P < 0.05 vs. control.

As shown in Fig. 3A, the SOCE blocker BTP2 (10 μm) reduced the amplitude of phase 2, but had no significant effect on phase 1 HPV. Similar results were obtained with the non-selective SOCE blocker SKF96365 (Fig. 3B), used at a concentration (50 μm) previously shown to fully inhibit HPV and the response to angiotensin II in saline-perfused rat lung (Weigand et al. 2005). In order to eliminate Ca²⁺ influx through all pathways, we also recorded the response to hypoxia after incubation of arteries for 30 min in ‘Ca²⁺-free’ PSS (containing 0.2 mm EGTA but no added Ca²⁺). Phase 1 was virtually abolished by this procedure, whereas phase 2 was attenuated to the same extent as it was by BTP2 (Fig. 3C). The contraction to (Ca²⁺-free) 80KPSS was virtually abolished in Ca²⁺-free PSS (not shown) indicating that the extracellular space had been thoroughly depleted of Ca²⁺.

Figure 3. Effects of SOCE antagonists and Ca²⁺-free PSS on HPV.

A, comparison of HPV under control conditions (squares) and following incubation in 10 μm BTP2 (triangles; n= 13). B, comparison of HPV under control conditions (squares) and following incubation in 50 μm SKF96365 (triangles; n= 7). C, HPV under control conditions (squares) and following incubation of PAs for 30 min in Ca²⁺-free PSS containing 200 μm EGTA (triangles; (n= 15). D, HPV under control conditions (squares) and following incubation of PAs in Ca²⁺-free PSS containing 200 μm EGTA and 30 μm CPA (triangles; n= 11). For all panels * indicates P < 0.05 vs. control.

Robertson et al. (2000b) previously showed that the sarco/endoplasmic reticulum Ca²⁺-ATPase (SERCA) inhibitor cyclopiazonic acid (CPA) selectively inhibits phase 1 in the presence of pretone. Dipp et al. (2001) reported a similar effect in the absence of pretone, leading to the proposal that CPA depletes Ca²⁺ selectively from an element of the SR located close to the plasma membrane (Boittin et al. 2003; Clark et al. 2010). In the presence of Ca²⁺, we found that CPA (30 μm) caused a contraction (amounting to 17.3 ± 1.9% of the contraction to 80KPSS, n= 11) which decayed gradually and incompletely, making it difficult to examine its effect on HPV under these conditions. This contraction was completely suppressed by Ca²⁺-free PSS (and also by 10 μm BTP2), indicating that it was due to SOCE (not shown). As it might be predicted that Ca²⁺-free PSS would deplete the CPA-sensitive Ca²⁺ store because of its superficial location, we reasoned that co-application of CPA and Ca²⁺-free solution for 30 min would not cause any further reduction in HPV over and above that obtained in Ca²⁺-free solution alone. As shown in Fig. 3D, this prediction was borne out.

HPV requires intracellular Ca²⁺ release from the SR but is not dependent on lysosomes

As CPA may not release Ca²⁺ from the entirety of the SR, we also used the alternative approaches of suppressing intracellular Ca²⁺ release by blocking the ryanodine receptor (RyR) with a high concentration of ryanodine and depleting the SR with a combination of ryanodine and caffeine (Jabr et al. 1997). Pretreatment of IPAs with ryanodine at a concentration (100 μm) which should block RyR-mediated Ca²⁺ release (Laporte et al. 2004), appeared to reduce both phases of HPV, although only the effect on phase 2 was statistically significant (Fig. 4A). A brief application of 15 mm caffeine and 10 μm ryanodine to deplete the SR of Ca²⁺, followed by continuing incubation of arteries in 10 μm ryanodine in order to prevent store refilling (Jabr et al. 1997), caused an almost complete inhibition of HPV (P < 0.05 compared to control; Fig. 4B).

Figure 4. HPV requires intracellular Ca²⁺ release from the SR via RyRs.

A, HPV before (squares) and after (triangles) incubation of PAs with 100 μm ryanodine (n= 9). B, HPV before (squares) and after (triangles) incubation with a combination of 15 mm caffeine and 10 μm ryanodine (n= 6). For all panels * indicates P < 0.05 vs. control.

There is evidence that Ca²⁺ release from lysosomes activated by nicotinic acid adenine dinucleotide phosphate (NAADP) via activation of two-pore channels (Calcraft et al. 2009) can activate Ca²⁺ release from the SR by triggering Ca²⁺-induced Ca²⁺ release by RyRs in PASMCs (Boittin et al. 2002; Kinnear et al. 2004, 2008). Lysosomes contain a vacuolar H⁺-ATPase, thereby generating a proton gradient which is proposed to allow Ca²⁺ to be concentrated in the organelle via an as yet undefined Ca²⁺/H⁺ exchange process. Aley et al. (2010) have recently provided evidence for a role of lysosomes in the spontaneous contractile activity of rat uterus by applying the vacuolar H⁺-ATPase inhibitor bafilomycin, and the two-pore channel inhibitor NED-19, both of which inhibited spontaneous contractile activity.

Application of 1 μm NED-19 during an ongoing HPV had no discernible effect, and the subsequent HPV in the continued presence of the drug was also not affected (Fig. 5A; P > 0.05, n= 5). Application of 1 or 10 μm NED-19 between the 2nd and 3rd hypoxic challenges according to our usual protocol also had no significant effect (Fig. 5B and C). Using the same protocol, application of another NAADP antagonist, BZ194 (200 μm; Dammermann et al. 2009), similarly had no effect on HPV (Fig. 5D).

Figure 5. Lack of effect of NED-19 and BZ194 on HPV.

A, application of NED-19 (1 μm) midway through a hypoxic challenge (left panel: squares → triangles) had no effect on HPV. PAs were then re-oxygenated and re-challenged with hypoxia ∼45 min later with NED-19 present throughout, giving rise to a third HPV (triangles) of similar amplitude (n= 6). B, NED-19 (10 μm) was applied after the control HPV (squares), and another HPV was evoked subsequently in the continued presence of drug (triangles; n= 7). C, BZ194 (200 μm) was applied after the control HPV (squares), and a third HPV was subsequently recorded in the continued presence of drug (triangles; n= 9).

Lysosomal Ca²⁺ can be released and depleted by the drug concanamycin, which, like bafilomycin, blocks the vacuolar H⁺-ATPase (Huss & Wieczorek, 2009). Notably, although concanamycin caused little or no contraction when applied under baseline conditions at a concentration of 1 μm, it further increased contraction when applied in the presence of a low concentration of PGF_2α (Fig. 6A); this is consistent with the concept that it releases lysosomal Ca²⁺ (Morgan et al. 2011). NED-19 (1 μm) partially reversed the concanamycin contraction, inhibiting it by 65.8 ± 9.5% (n= 6; P < 0.05). As shown in Fig. 6B and C, application of 1 μm concanamycin during a hypoxic challenge also caused a further sustained rise in tension. However, the amplitude of HPV during a subsequent period of hypoxia in the continued presence of concanamycin was similar to that observed before concanamycin had been applied.

Figure 6. Concanamycin causes contraction but does not affect the amplitude of HPV.

A, example trace showing the effect of applying concanamycin (1 μm) and then NED-19 (1 μm) in the presence of 5 μm PGF_2α. B, example trace showing the effect of adding concanamycin (1 μm) during HPV, and also the subsequent HPV recorded in the continuing presence of concanamycin. C, averaged data from a group of experiments using the protocol shown in B; the HPV during which concanamycin was added is shown on the left (squares → triangles) and the points on the right (triangles) represent the subsequent HPV evoked in the continuing presence of concanamycin (n= 6). All points have been normalised to the amplitude of the HPV recorded just before addition of concanamycin (20 min into the hypoxic challenge). Concanamycin caused a significant increase in tension when added during an ongoing HPV (left), but not during a subsequent HPV evoked in its continuing presence (right).

Effects of pharmacological manipulation of cyclic ADP ribose, NADH and AMP-activated protein kinase on HPV

SR Ca²⁺ release during phase 2 has also been suggested to require cyclic ADP ribose, the concentration of which was proposed to increase as a result of a hypoxia-induced increase in [NADH] (Wilson et al. 2001). However, pretreatment of IPAs with 8-Br-cyclic ADPR (30 μm), an antagonist of cADPR, did not inhibit HPV (Fig. 7A). Application of 30 and 60 μm 8-Br-cyclic ADPR during phase 2 also had no effect on tension in one additional experiment (not shown).

Figure 7. The cyclic ADP ribose antagonist 8-Br-cyclic ADPR does not inhibit HPV, and HPV can be elicited independently of an increase in NAD(P)H autofluorescence.

A, comparison of average control HPVs (squares) with subsequent HPVs recorded following 30 min pretreatment with 8-Br-cyclic ADP ribose (triangles; n= 11). B, HPV before (squares) and after (triangles) the successive application of succinate (5 mm) and rotenone (1 μm; n= 8). C, the effect of hypoxia alone on NAD(P)H autofluorescence compared with the effect of hypoxia following application of succinate (5 mm) and then rotenone (1 mm). One of two experiments of this type is shown; similar results were obtained in both. For panels A and B* indicates P < 0.05 vs. control.

In order to determine whether HPV without pretone required an increase in [NADH], we applied 1 μm rotenone (which prevents the oxidation of NADH by complex 1 and should therefore raise its concentration) after first adding 5 mm succinate, which bypasses rotenone-induced block of the electron transport chain by feeding electrons into complex 2 and thereby rescues HPV from block by rotenone (Leach et al. 2001). Application of rotenone in the presence of succinate caused a small transient contraction (not shown). Phase 2 HPV was preserved during a subsequent hypoxic challenge under these conditions (Fig. 7B) although phase 1 was significantly attenuated. Figure 7C shows that NAD(P)H was strongly increased by rotenone in the presence of succinate, and was then little affected when hypoxia was applied. Thus, an increase in [NAD(P)H] similar to that caused by hypoxia causes little tension development, and HPV can then occur without a further rise in [NAD(P)H].

Pretreatment of IPAs with compound C (40 μm), a inhibitor of AMPK, which might be responsible for stimulating cADPR production (Evans, 2006), partially inhibited HPV (e.g. by 48 ± 10% at 40 min, n= 13; Fig. 8A). However, this effect was not selective: as shown in Fig. 8B compound C similarly inhibited the contraction to 5 μm PGF_2α (by 56 ± 9%; n= 4).

Figure 8. The AMP kinase inhibitor compound C blocks and the AMPK activator metformin enhances HPV and also PGF_2α-induced contractions.

A, comparison of average control HPVs (squares) with subsequent HPVs recorded following 30 min pretreatment with 40 μm compound C (triangles; n= 13). B, effect of compound C (40 μm) on contractions evoked by 5 μm PGF_2α (n= 4). C, HPV recorded before (squares) and following (triangles) pretreatment with, and in the continuing presence of, metformin (3 mm; n= 9). D, example trace showing HPV and contractions induced by 1 and 10 μg ml⁻¹ antimycin A in a PA. Similar results were obtained in 3 other experiments of this type. In panels A–C* indicates P < 0.05 vs. control.

As additional approaches to investigating the role of AMP kinase, we examined whether IPAs contracted when exposed to the AMP kinase activator metformin (Sung & Choi, 2012) or three blockers of the electron transport chain (cyanide, rotenone, antimycin) which would be expected to mimic hypoxia in increasing the [AMP]/[ATP] ratio and thereby stimulate AMP kinase. At concentrations between 0.1 and 3.0 mm metformin caused no contraction in nine IPAs, although a small contraction (∼0.05 mN) was observed with 3.0 μm metformin in one artery. Interestingly, 3 mm metformin caused a significant increase in the amplitude of HPV (Fig. 8C). However, this effect was not specific to HPV, as the contraction evoked by 30 μm PGF_2α was increased to a similar extent (26.1 ± 8.3%, n= 12, P < 0.05) in the presence of 3 mm metformin. Neither cyanide (10 μm) nor rotenone (1 μm) caused contraction (n= 5 and 7, respectively, not shown). As shown in Fig. 8D, antimycin caused a small contraction at concentrations of both 1 and 10 μg ml⁻¹, but this response was much smaller than HPV (see Sylvester et al. 2012).

Alternative mechanisms of Ca²⁺ release

Zhang et al. (2012) have recently presented evidence that hypoxia activates Ca²⁺ release via a stimulation of the PASMC Ca²⁺-sensing receptor (CaSR), and that this occurs due to a rise in extracellular [H₂O₂]. We investigated this using the CaSR antagonist NPS2390 (Kwak et al. 2005; Smajilovic et al. 2006). As shown in Fig. 9A, application of a high concentration (10 μm) of NPS2930 had no effect on HPV. However, because the free [Ca²⁺] in plasma in vivo is 1.1–1.2 mm (Moore, 1970), a concentration considerably lower than that in PSS, it is possible that the CaSR may have been saturated and thus insensitive to further stimulation in our normal PSS. We therefore investigated the effect of NPS2390 on HPV using PSS containing 1.2 mm Ca²⁺. In these experiments we first applied 0.1 μm NPS2390 before the 3rd HPV, and then increased the NPS2390 concentration to 10 μm and applied a 4th hypoxic challenge. As shown in Fig. 9B, HPV was unaffected by either concentration of NPS2390. Moreover, neither phase of HPV was inhibited in the presence of the H₂O₂ scavenger catalase (200 units ml⁻¹; not shown, n= 7).

Figure 9. Neither the CaSR antagonist NPS2390 nor the FKB12.6 binding protein modulator FK506 have any effect on HPV.

A, comparison of average control HPVs (squares) with subsequent HPVs recorded following pretreatment with, and in the continuing presence of 10 μm NPS2390 (inverted triangles) in 5 arteries incubated in normal PSS (n.s.). B, comparison of average control HPVs (squares) with HPVs recorded following pretreatment with, and in the continuing presence of 0.1 μm (3rd HPV, normal triangles, dashed line) and then 10 μm NPS2390 (4th HPV, inverted triangles) in 9 arteries incubated in PSS containing 1.2 mm[Ca²⁺]. C, effect on tension of the cumulative addition of increasing concentrations of FK506 (1.0–50 μm) during normoxia (n= 6–12). D, comparison of average control HPVs (squares) with subsequent HPVs recorded following pretreatment with 10 μm FK506 (triangles; n= 6). E, comparison of average control HPVs (squares) with subsequent HPVs recorded following pretreatment with 10 μm cyclosporine (triangles; n= 7).

Displacement of FKBP12.6 from ryanodine receptor 2 (RyR2) has also been proposed to be important in mediating SR Ca²⁺ release during HPV, a concept supported by the observation that the drug FK506, which also causes this displacement, enhanced the hypoxia-induced [Ca²⁺]_i rise in mouse PASMCs when applied at a concentration of 10 μm (Liao et al. 2011). FK506 itself caused a small concentration-dependent contraction (Fig. 9C), consistent with the idea that displacement of FKBP12.6 from RyR2 can cause release of Ca²⁺ from the SR. However, pre-incubation of IPAs in 10 μm had no effect on HPV (Fig. 9D). Since FK506 also blocks calcineurin, we tested cyclosporine which inhibits calcineurin but does not interact with FKBP12.6 (Liao et al. 2011), and it also had no effect on HPV (Fig. 9E, n= 7).

Effect of protein kinase inhibitors on HPV

Ca²⁺ sensitisation permits contraction independent of changes in [Ca²⁺]_i via the inhibition of myosin light chain phosphatase and can be mediated either by ROK phosphorylating myosin phosphatase target subunit 1 (MYPT-1) or protein kinase C (PKC) acting through CPI-17 (Somlyo & Somlyo, 2003). We have previously presented evidence that ROK-mediated Ca²⁺ sensitisation contributes to the sustained phase of HPV in isolated preconstricted IPAs and perfused lungs from rat (Robertson et al. 2000a). Conversely PKC inhibition with Ro 31-8220 reduced the amplitude of phase 1, whilst having no effect on phase 2 in preconstricted rat IPAs (Robertson et al. 1995).

Pretreatment of IPAs with Y-27632 (1 μm) virtually abolished both phases of HPV (Fig. 10A). Lower concentrations of Y-27632 also caused some block of HPV, and notably HPV was significantly more sensitive to inhibition by Y-27632 than was the contraction to the thromboxane (TP) receptor agonist U46619 (Fig. 10B). The broad spectrum PKC inhibitor Gö6983 (3 μm) also significantly attenuated HPV (Fig. 10C), but had no effect on tension development elicited by either 5 μm PGF_2α or 30 mm KPSS (Fig. 10D). It is noteworthy that similar experiments with the broad spectrum PKC inhibitor Ro 31-8220 and the conventional PKC blocker Gö6976 (both at 3 μm) similarly attenuated HPV. However, both compounds also reduced the amplitude of the contractions caused by 30 mm KPSS and PGF_2α, indicating their lack of selectivity (not shown).

Figure 10. Block of HPV by rho kinase inhibitor Y-27632 and broad spectrum PKC antagonist Gö6983.

A, comparison of control HPVs (squares) with subsequent HPVs recorded following pretreatment with, and in the continuing presence of, 1 μm Y-27632 (triangles; n= 10). B, concentration dependency of the block of HPV and the contraction to 100 nm U46619 by Y-27632 (n= 9–11 for HPV and 4 for U46619). *P < 0.05, block of contraction; #P < 0.05, difference between the block of the two types of contraction. C, comparison of average control HPVs (squares) with subsequent HPVs recorded following pretreatment with, and in the continuing presence of, 3 μm Gö6983 (triangles; n= 15, *P < 0.05). D, effect on phase 2 HPV (from panel C) compared with the lack of effect of Gö6983 on responses to 30 mm K⁺ PSS and 5 μm PGF_2α (n= 6 for both stimuli). All three responses recorded in Gö6983 are normalised to the control responses (represented by the dashed line) previously recorded in the same IPA.

Effects of antioxidants on HPV

As shown in Fig. 11A, the antioxidant TEMPOL (3 mm) almost abolished phase 2, but did not significantly affect the amplitude of phase 1. Another antioxidant, ebselen, caused a similar pattern of effects at a concentration of 100 μm; it significantly attenuated phase 2 but had no effect on phase 1 (Fig. 11B). Both antioxidants also significantly diminished the response to 30 μm PGF_2α (Fig. 11D). In contrast to the effects of TEMPOL and ebselen, the NADPH oxidase inhibitor VAS2870 (10 μm; ten Freyhaus et al. 2006) had no effect on HPV (Fig. 11C), although it attenuated the PGF_2α contraction to the same extent as did ebselen (Fig. 11D).

Figure 11. Effect of antioxidants and NADPH oxidase blockade on HPV.

A, comparison of average control HPVs (squares) with subsequent HPVs recorded following pretreatment with, and in the continuing presence of, 3.0 mm TEMPOL (triangles; n= 5). B, comparison of average control HPVs (squares) with subsequent HPVs (triangles) recorded following pretreatment with, and in the continuing presence of, 100 μm ebselen (n= 9). C, comparison of average control HPVs (squares) with subsequent HPVs (triangles) recorded following pre-treatment with, and in the continuing presence of, 30 μm VAS2870. D, effect of TEMPOL, ebselen and VAS2870 (3.0 mm, 100 μm and 10 μm, respectively) on the contraction caused by 30 μm PGF_2α (n= 5 for TEMPOL, 5 for DHLA, and 8 for VAS2780). For all panels * indicates P < 0.05 vs. control.

Effect of sustained hypoxia on IPA redox state

The effects of TEMPOL and ebselen implied that phase 2 might require a rise in cellular ROS or an oxidising shift in the cell redox state. We therefore examined whether 45 min of hypoxia affected the redox potential of the GSSG/GSH redox couple, as this is thought to reflect the cytoplasmic redox state (Schafer & Buettner, 2001). As shown in Fig. 12A, hypoxia caused an increase in the GSSG/GSH ratio when IPAs were gassed with either 0 or 1% O₂ (note that HPV amplitude was slightly smaller when IPAs were gassed with 1% compared to 0% O₂; not shown). In contrast, hypoxia had no significant effect on the GSSG/GSH redox potential in rat aorta (Fig. 12B).

Figure 12. Effect of 45 min hypoxia on the GSH/GSSG redox potential in IPA (A) and aorta (B).

GSH and GSSG concentrations were measured as described in Methods following a 45 min incubation under normoxic or hypoxic conditions (gassing with 1 or 0% O₂) and the redox potential of this couple was calculated using the equation shown above the graphs. Number of replicates shown for each condition; * indicates a significant effect of hypoxia; P < 0.05 vs. control).

Figure 13 presents a summary of the effects of the various interventions described above on the two phases of HPV.

Discussion

The relative importance of Ca²⁺ influx vs. the release of Ca²⁺ from the SR, together with the mechanism(s) by which release is triggered and the pathway predominantly responsible for causing Ca²⁺ influx, currently constitute key areas of disagreement with regard to the mechanisms by which hypoxia causes HPV. The purpose of this study was to attempt to resolve some of these controversies by doing a systematic study of various proposed HPV mechanisms using an in vitro model of HPV which is uncontaminated by any exogenously applied stimuli. Using rat IPAs made hypoxic in the absence of pretone, we find that both phases of HPV are strongly dependent on Ca²⁺ released from the SR, with SOCE making some contribution to phase 2. Neither phase, however, is affected by antagonists of cyclic ADP ribose (Dipp et al. 2001), NAADP-dependent lysosomal Ca²⁺ release (Evans, 2010), the CaSR (Zhang et al. 2012), or by displacement of FKBP12.6 from the RyR (Zheng et al. 2004; Liao et al. 2011). We also observe that hypoxia causes an increase in the GSSG/GSH ratio after 45 min of hypoxia and that phase 2 HPV is blocked by the antioxidants ebselen and TEMPOL. These results are consistent with the proposal, first presented by Jabr et al. (1997), that the fundamental effector mechanism in HPV is a RyR-mediated release of SR Ca²⁺. Although we did not assess the cell redox state during phase 1, our measurements of GSH and GSSG indicate that the sustained release-dependent contraction is accompanied by an oxidising shift in the cell redox state, and the results therefore are in accordance with the mitochondrial ROS hypothesis of Schumacker and colleagues (Waypa et al. 2001). We speculate that sustained release may be due to oxidation-induced modifications of the RyR, which contains a number of highly reactive cysteines (Hidalgo & Donoso, 2008).

Voltage-gated Ca²⁺ entry during phase 1 is pretone dependent

Hypoxia causes an increase in [Ca²⁺]_i in both cultured PASMCs (Salvaterra & Goldman, 1993; Vadula et al. 1993; Cornfield et al. 1994; Ng et al. 2005; Wang et al. 2005) and intact preconstricted IPAs (Robertson et al. 2000b). The question as to the relative contribution of various potential pathways (e.g. VGCCs, SOCE, Ca²⁺ release) to this rise in [Ca²⁺]_i remains, however, unresolved.

We found that neither phase of HPV without pretone was affected by either diltiazem or nifedipine, indicating that L-type VGCCs play no role in the response to hypoxia in this model. This observation runs counter to a considerable body of evidence (see reviews by Moudgil et al. 2005; Sylvester et al. 2012) supporting the concept that HPV is due primarily to Ca²⁺ influx through these channels (McMurtry et al. 1976). This could be caused by membrane depolarisation (Harder et al. 1985) resulting from the closure of K⁺ channels (Post et al. 1992; Yuan et al. 1994; Archer et al. 2004; Olschewski et al. 2006) or the opening of non-selective cation channels (Ng et al. 2005; Wang et al. 2005; Weigand et al. 2005; Weissmann et al. 2006a).

Importantly, however, whereas neither phase of HPV was affected by VGCC inhibition in the absence of pretone, phase 1 in the presence of PGF_2α-induced pretone was strongly inhibited by nifedipine (Fig. 2E), although phase 2 was still unaffected. This is noteworthy because influential studies of the role of VGCCs in perfused lung have utilised brief (∼5 min) hypoxic exposures and/or pretone to magnify HPV (e.g. McMurtry et al. 1976; Weigand et al. 2005).This may have led to an indirect attenuation by the VGCC blocker of HPV due to an effect on pretone, a possibility supported by the observation of Rodman et al. (1989) that the profound block of HPV by nifedipine in isolated phenylephrine-preconstricted IPAs during very brief exposures to hypoxia was ablated when the inhibitory effect of nifedipine on pretone was compensated for by raising the vasoconstrictor concentration to restore the pretone level. An implication of these observations is that since most in vitro studies of the effects of VGCCs on HPV have used short periods of hypoxia and uncompensated pretone, it is possible that the role of these channels in HPV, particularly when it is sustained beyond a few minutes, has been greatly overestimated by these observations.

A similar possibility also exists with regard to in vivo measurements of HPV. Sato et al. 2000 found that the endothelin receptor antagonists BQ123 and PD145065 strongly suppressed HPV in rats in vivo and that this effect could be reversed by infusion of angiotensin II, which presumably restored pretone. Remarkably, although BQ123 and PD145065 suppressed HPV in this study, they did not decrease resting (normoxic) pulmonary vascular resistance, implying that the level of endothelin-induced stimulation required to enable the development of HPV was so small that its effect on pulmonary vascular tone could not be measured. Similarly, Liu et al. (2001) reported that the abolition of HPV by BQ123 in small IPAs from pig could be reversed by a concentration of endothelin (10⁻¹⁰ m) that caused little or no contraction (Liu & Sylvester, 1999). Therefore, HPV in vivo appears to be powerfully potentiated by a level of endogenous stimulation which may not cause detectable increases in IPA tone. Such ‘endogenous pretone’ could be due to any of a variety of vasoconstrictors, and may be associated with membrane depolarisation and therefore subject to block by VGCC antagonists (Sato et al. 2000; Liu et al. 2001; Shimoda et al. 2001).

In light of our observations and these considerations, we propose that the ∼50% attenuation of HPV caused by VGCC blockers in vivo (Naeije et al. 1982; Bishop & Cheney, 1983; Burghuber, 1987) reflects the suppression of a component of endogenous prestimulus which is sensitive to these drugs rather than an effect on the intrinsic constriction of IPAs to hypoxia. The level of prestimulus required to facilitate HPV may itself be so small that it is difficult to record, but even so it may serve to strongly amplify the response.

Ca²⁺ release and SOCE in phases 1 and 2: a role for two Ca²⁺ stores?

The two phases of HPV demonstrated different properties. Phase 1 was not significantly blocked by BTP2 and SKF96365, but was abolished in Ca²⁺-free solution. It was also not significantly inhibited by 100 μm ryanodine, unlike phase 2 (Fig. 4). We have previously shown that blocking the RyRs with either ryanodine or dantrolene selectively antagonises phase 2 in the presence of pretone, which by exaggerating the size of phase 1 (e.g. Fig. 1) makes it easier to visualise (Becker & Aaronson, 2006; Becker et al. 2006).

Conversely, phase 2 was inhibited, and similarly, by BTP2 and SKF96365 (35 and 44%, respectively), with both drugs being used at concentrations which should maximally block SOCE (He et al. 2005; Weigand et al. 2005; Harper & Poole, 2011). Treatment with Ca²⁺-free PSS led to a similar inhibition of phase 2 (32%), and no further reduction of phase 2 occurred when arteries were incubated in Ca²⁺-free solution containing CPA. Phase 2 was also strongly suppressed by 100 μm ryanodine and by the combination of 10 μm ryanodine and 15 mm caffeine, an observation which is in accordance with previous reports that sustained HPV in isolated IPAs is inhibited by caffeine/ryanodine or genetic knockdown of the RyR (Jabr et al. 1997; Dipp et al. 2001; Zheng et al. 2005; Li et al. 2009).

Based on these observations and previous work by others (e.g. Jabr et al. 1997; Clark et al. 2010), we propose a model in which HPV in the absence of pretone is mediated by the release of two functionally and spatially discrete compartments of the SR (Fig. 14).

Figure 14. Proposed model for the role of Ca²⁺ release by two compartments of the SR in HPV.

According to this scheme, hypoxia causes the release of Ca²⁺ from two spatially and functionally discrete compartments of the SR. One of these (SR_A) is located superficially within PASMCs and is depleted by CPA or incubation in Ca²⁺-free PSS. The release of this store causes phase 1, and a somewhat delayed activation of SOCE which contributes to phase 2. The other compartment (SR_B) is located more deeply within PASMCs. Hypoxia releases Ca²⁺ from SR_B via the RyR, leading to a sustained contraction which accounts for about 2/3rds of the amplitude of phase 2. The resulting changes in [Ca²⁺]_i are small and sensitisation mediated by rho kinase strongly amplifies their contractile effects. The model is based on observations in the present study, and also owes much to previous results and proposals (Jabr et al. 1997; Evans & Dipp, 2002; Clark et al. 2010).

One compartment of the SR (SR_A) in this scheme is selectively depleted by Ca²⁺ removal and is insensitive to ryanodine (although it may be at least partially sensitive to the combination of ryanodine and caffeine). Ca²⁺ released from this store by hypoxia is directly responsible for phase 1 and also initiates SOCE, which increases progressively to become significant during phase 2. A study by Clark et al. (2010) demonstrated that the SR in freshly isolated rat PASMCs is spatially segregated into two compartments which can be defined by the presence of differential distributions of RyR and SERCA isoforms. One of these is located close to the plasmalemma, leading Evans and colleagues to propose that it equates to a CPA-sensitive Ca²⁺ store which, when released by β-agonists, leads to the activation of BK_Ca channels and relaxation (Boittin et al. 2003; Clark et al. 2010). CPA has also been shown to abolish phase 1 HPV but has less effect on phase 2 both in the presence (Robertson et al. 2000b) and absence (Dipp et al. 2001) of pretone.

The subplasmalemmal distribution of this compartment implies that it would be well positioned to activate SOCE, and our observation that CPA causes a contraction that is inhibited by BTP2 and also by Ca²⁺ removal also strongly supports this possibility. Although it might therefore be predicted that SOCE should be most prominent during phase 1, we speculate that SOCE might develop slowly if there is a delay of several minutes before the store is sufficiently depleted of Ca²⁺ for SOCE to make a significant contribution to HPV.

The second compartment of the SR in this model (SR_B) is essentially insensitive to Ca²⁺ removal and CPA, but its release is blocked by 100 μm ryanodine. It is not well coupled to SOCE, and the Ca²⁺ it releases directly mediates ∼60% of phase 2. This compartment may correspond to an element of the SR identified by Clark et al. which they showed is located more centrally within PASMCs and is released by endothelin-1 but not CPA. It also has some similarities to the ryanodine-sensitive and CPA-insensitive component of the SR proposed by Jabr et al. (1997) to mediate sustained HPV in canine PAs, although it differs in that they found that this store was quickly depleted by Ca²⁺ removal and nisoldipine treatment.

The deeper location of this store would explain why its release during hypoxia does not activate SOCE and could also account for the observation (which we also made) that depletion of this store with ryanodine and caffeine is not associated with a contraction (Jabr et al. 1997; Dipp et al. 2001). It would also be consistent with its remarkable lack of sensitivity to the removal of extracellular Ca²⁺, a property which is reminiscent of observations of agonist-releasable Ca²⁺ stores in vascular smooth muscle that are able to mediate sustained and repeatable contractions almost indefinitely in Ca²⁺-free PSS (Casteels et al. 1981; Aaronson & Jones, 1985).

Our finding that Ca²⁺ release rather than influx plays such a major role in HPV is surprising, yet agrees with previous reports that sustained HPV in isolated arteries recorded in the absence of pretone is largely unaffected by the removal of extracellular Ca²⁺ (Hoshino et al. 1988; Evans & Dipp, 2002, but see Yuan et al. 1990). On the other hand, this observation is inconsistent with evidence showing that SOCE and VGCCs make the predominant contribution to hypoxia-induced rises in [Ca²⁺]_i and HPV in isolated/cultured PASMCs (Wang et al. 2005; Ng et al. 2005, 2012; Lu et al. 2009) and perfused lungs (Weigand et al. 2005), respectively. The reasons for this discrepancy are unclear, although there are a number of factors which could potentially explain it. These include phenotypic changes in Ca²⁺ stores which occur during cell culture that might potentiate SOCE (Ng et al. 2008), and also the possibility that hypoxia might activate SOCE (and, as a result, voltage-gated Ca²⁺ entry) in the presence of pretone more strongly than in its absence because it is potentiating the effect of the co-stimulus which has already initiated this process (Wang et al. 2005). Nevertheless, the reconciliation of these apparently contradictory observations will clearly require further study.

Mechanisms of SR Ca²⁺ release

Evidence that SR Ca²⁺ release mediated by RyRs is pivotal to HPV (Dipp et al. 2001; Zheng et al. 2005; Li et al. 2009) has led to many to investigate the link between hypoxia and RyR opening, and we examined a number of the proposals emerging from these studies. Evans and colleagues (Dipp & Evans, 2001; Wilson et al. 2001) proposed that Ca²⁺ release during HPV is the result of the opening of the RyRs by cyclic ADP ribose (cADPR). The concentration of cADPR was suggested to increase due to a rise in cellular NADH resulting from hypoxia-induced inhibition of the electron transport chain. They observed that phase 2 in non-preconstricted IPAs was abolished following pretreatment with 3 μm of the cADPR inhibitor 8-Br-cADPR, and that direct application of this inhibitor during phase 2 rapidly reversed the contraction, causing relaxations of ∼30 and ∼60% at concentrations of 10 and 30 μm, respectively. However, we were unable to confirm this observation, as 8-Br-cADPR (30 μm) did not suppress HPV (Fig. 7A).

In addition, we sought to determine whether a rise in [NADH] similar to that caused by hypoxia would cause a contraction. We therefore pre-treated IPAs with rotenone to block complex 1 and raise [NADH], and also applied succinate at the same time to maintain the flow of electrons through the distal electron transport chain, which enables HPV to occur (Leach et al. 2001). As shown in Fig. 7B, application of 1 μm rotenone and 5 mm succinate caused a rise in [NAD(P)H] which was similar to that elicited by hypoxia. However, application of rotenone + succinate was not associated with a contraction. Moreover, hypoxia then elicited a normal phase 2 HPV, even though there was no further rise in [NAD(P)H]. Thus, our results do not support the concept that a rise in cADPR associated with an increase in [NAD(P)H] causes phase 2 HPV.

Evans et al. (2005) proposed that inhibition of mitochondrial oxidative phosphorylation in response to hypoxia produces a change in the [AMP]:[ATP] ratio, thus activating AMPK, which goes on to trigger HPV. In support of this hypothesis, the AMPK inhibitor compound C at a concentration of 40 μm was shown to block phase 2 in isolated PGF_2α-preconstricted IPAs (Robertson et al. 2008) and also the hypoxia-induced rise in [Ca²⁺]_i in human cultured PASMCs (Tang et al. 2010). Compound C also markedly inhibited HPV in the absence of pretone (Fig. 8A). However, this effect was not specific to HPV as compound C reduced the PGF_2α contraction to the same extent (Fig. 8B). As an alternative approach, we examined whether application of the AMPK activator metformin (Sung & Choi, 2012) would cause contraction on its own, or affect HPV. Although metformin did not itself cause contraction, it did slightly enhance HPV (Fig. 8C). However, as with the effect of compound C, this does not appear to have anything to do with O₂ sensing since the PGF_2α contraction was similarly affected.

In order to address the putative role of AMPK using a third approach, we took advantage of the fact that inhibition of the electron transport chain should also raise the [AMP]/[ATP] ratio, and should therefore evoke contraction if AMPK activation has this effect. However, the three inhibitors we tested either caused no contraction (cyanide, rotenone) or caused a contraction which was much smaller than HPV (antimycin; Fig. 8D). Similar observations have previously been made in other models of HPV (see review by Sylvester et al. 2012). Taken together, these results again do not support a role for AMP kinase in HPV, and also are not consistent with the concept that a fall in mitochondrial ROS or an increase in NADH can trigger a contraction in pulmonary arteries, as proposed by Weir & Archer (1995).

Boittin et al. (2002) demonstrated in rat PASMCs that intracellular application of the Ca²⁺-mobilising second messenger and relative of cADPR, NAADP (0.01–10 μm) caused contraction via release of Ca²⁺ from a thapsigargin-insensitive store, which was amplified by Ca²⁺-induced Ca²⁺ release via RyRs but not InsP₃ receptors. The thapsigargin-insensitive Ca²⁺ store was subsequently discovered to be lysosomal (Kinnear et al. 2004). Moreover, a subpopulation of lysosomes was seen to colocalise with a RyR3-rich component of the SR in the perinuclear region of PASMCs, leading to the hypothesis that these structures formed an NAADP-sensitive trigger zone where lysosomal Ca²⁺ release can act on the RyRs to initiate SR-mediated Ca²⁺ waves (Kinnear et al. 2004, 2008). This process has also been proposed to be crucial for HPV (Evans, 2010), although to date evidence for this has been lacking. We examined the possible role of lysosomes in HPV using concanamycin, which depletes lysosomal Ca²⁺ by inhibiting the vacuolar proton pump (Huss & Wieczorek, 2009). Additional experiments examined the effects of NED-19, which blocks the two-pore channels shown to mediate lysosomal Ca²⁺ release (Aley et al. 2010) and BZ194, an analogue of NAADP which was reported to act as an antagonist with an IC₅₀ of 90 μm (Dammermann et al. 2009). Concanamycin caused an enhancement of contraction when applied in the presence of either PGF_2α or hypoxia, indicating that lysosomal Ca²⁺ release can either directly or indirectly (via CICR) raise [Ca²⁺]_i in PASMCs. However, concanamycin had no effect on the amplitude of the subsequent HPV, indicating that depletion of lysosomal Ca²⁺ prior to hypoxia had no effect. Similarly, neither NED-19 nor BZ194 attenuated either phase of HPV.

Additionally, diltiazem at a concentration of 30 μm has been shown to almost completely block NAADP-induced lysosomal Ca²⁺ release in sea urchin egg homogenates (Genazzani et al. 1997). Since diltiazem has a pK_a of 7.7 (Smirnov & Aaronson, 1998), it will be partly in its membrane-permeable uncharged form at a pH of 7.4 and therefore its intracellular and extracellular concentrations would equalise under our experimental conditions (diltiazem is thought to access and block L-type VGCCs from within cells; Smirnov & Aaronson, 1998). However, we saw no block of HPV by 30 μm diltiazem. This provides additional evidence that NAADP-induced lysosomal Ca²⁺ release is not required for HPV.

Although the release of Ca²⁺ from lysosomes by vacuolar H⁺-ATPase inhibitors such as concanamycin can be explained as being the result of the reversal of lysosomal Ca²⁺/H⁺ exchange, this remains an unproven assumption and can be questioned because the exchangers required for this are not expressed in the endolysosomal system (Morgan et al. 2011). Notably, this mechanism would not explain the novel observation, which we did not pursue further, that the two-pore channel blocker NED-19 reversed the concanamycin-induced contraction (Fig. 6B).

Zhang et al. (2012) recently proposed that activation of the CaSR caused by a hypoxia-induced rise in H₂O₂ causes HPV by triggering the opening of the RyRs through a cyclic AMP-mediated process. However, we observed no effect of the CaSR blocker NPS2390 on either phase of HPV. Similarly, incubation of IPAs in catalase, which should scavenge extracellular H₂O₂ and prevent activation of the CaSR according to Zhang et al. (2012), had no effect on HPV.

Protein kinases and HPV

Although the abolition of HPV by Y-27632 and other rho kinase antagonists supports a role for Ca²⁺ sensitisation in HPV, Wang et al. (2007) proposed that these drugs were instead acting by preventing Ca²⁺ release from the SR. In order to examine this possibility, we studied the effects on HPV of concentrations of Y-27632 (≤1 μm) which were shown in this paper not to reduce the [Ca²⁺]_i rise caused by hypoxia. As shown in Fig. 10C, HPV was attenuated in the presence of low concentrations of Y-27632, with a ∼40% reduction and virtually complete block at 0.2 and 1 μm, respectively (Fig. 10B). These data, together with our previous finding that both acute and sustained hypoxia are associated with a Y-27632-sensitive increase in MYPT-1 phosphorylation (Knock et al. 2008), suggest that the effects of this drug on HPV are the result of its effect on Ca²⁺ sensitisation rather than on Ca²⁺ handling.

HPV during short (5–10 min) periods of hypoxia has been shown to be abolished by PKC inhibition in perfused rabbit lung (Weissmann et al. 1999) and isolated rat IPAs studied in the absence of pretone (Cogolludo et al. 2009), with evidence presented that this effect in the latter preparation is due to stimulation of PKCζ by ceramide released as a result of the activation of neutral sphingomyelinase. However, a role for PKC in sustained HPV is more controversial. Phase 2 in isolated preconstricted IPAs has been found to be either unaffected (Robertson et al. 1995) or inhibited (Tsai et al. 2004) by pharmacological PKC inhibitors. However, the effects of PKC inhibition on sustained HPV have not been examined in the absence of preconstriction until now, and the finding that Gö6983 caused a partial inhibition of HPV, whilst having no effect on PGF_2α or depolarisation-induced contraction, adds to the evidence that both phases of HPV involve PKC activation. Gö6976, which has previously been shown to block sustained HPV in dog lung (Barman, 2001), also attenuated HPV, but this effect was not selective since Gö6976 had a similar effect on contractions caused by 80KPSS and PGF_2α.

Role of cell redox state in HPV

The role of cell redox state and reactive oxygen species in triggering HPV is controversial. Although antioxidants have generally been observed to attenuate HPV (Sylvester et al. 2012), their effect on the two phases of HPV in isolated arteries has previously not been examined. We found that phase 2 but not phase 1 was inhibited by the antioxidants TEMPOL and ebselen, indicating that phase 1 does not require an increase in ROS or cell oxidation, whereas phase 2 might. Evidence has been presented that phase 1 HPV in perfused mouse and rabbit lung is partly dependent on the activity of NADPH oxidase (Weissmann et al. 2006b), but VAS2870, at a concentration which blocks all isoforms of NADPH oxidase inhibitor (ten Freyhaus et al. 2006; Altenhöfer et al. 2012; Wingler et al. 2012), had no effect on either phase of HPV in isolated IPAs (Fig. 11C). It has been shown that block of NADPH oxidase diminishes TP receptor-mediated contraction in PASMCs by ∼50% (Cogolludo et al. 2006). Accordingly, VAS2870 reduced the PGF_2α contraction, which in these arteries is mediated by TP receptors (Snetkov et al. 2006), to a similar extent, as did ebselen and TEMPOL.

These results implied that these drugs were acting as would be predicted if their effects were due to their antioxidant properties, and also that phase 2 should be associated with an increase in cell ROS or an oxidising shift of the PASMC redox potential. Evidence from studies which have used chemical indicators to record ROS levels during HPV has presented a confusing picture, as hypoxia has been seen to cause both rises and falls in cellular [ROS] (see review by Sylvester et al. 2012). More recent experiments which have used RhoGFP, a genetically encoded redox-sensitive fluorescent protein indicator designed to avoid the well-known pitfalls of small-molecule ROS indicators (e.g. Wardman, 2007; Kalyanaraman et al. 2012) have shown in cultured PASMCs and lung slices that hypoxia causes the oxidation of RhoGFP targeted to the cytoplasm and mitochondrial intramembrane space (Waypa et al. 2010; Desireddi et al. 2010), whereas RhoGFP targeted to the mitochondrial matrix becomes more reduced during hypoxia. In order to examine the effect of hypoxia on the redox state in freshly isolated IPAs, we employed the alternative method of measuring the GSSG/GSH ratio during sustained hypoxia. This ratio increases as a result of increases in cellular [ROS], since GSH is converted to GSSG as it is used by glutathione peroxidase to reduce H₂O₂ to H₂O and to reverse the oxidative modification of cell thiols (Janssen-Heininger et al. 2008). The redox potential of the GSSG/GSH redox couple, which can be calculated from this ratio, is thought to closely reflect the cytoplasmic redox state (Schafer & Buettner, 2001). We found that this was shifted by +13 and +17 mV in tissues gassed with 1 and 0% O₂, respectively, indicating that cellular oxidation had occurred. A similar effect was not observed in rat aorta, suggesting that hypoxia has different effects on the redox state in pulmonary compared to systemic arteries. This observation provides novel and compelling evidence that hypoxia causes an oxidative shift in the PASMC redox potential in non-cultured PAs. Together with the observations emerging from the use of RhoGFP, and the finding that hypoxia reduced GSH levels in cultured PASMCs (Waypa et al. 2001), these results powerfully support the concept that sustained HPV is associated with an increase in cytoplasmic ROS. Moreover, there are a number of potential mechanisms by which ROS could induce Ca²⁺ release and stimulate SOCE (Hidalgo & Dinoso, 2008; Hawkins et al. 2010; Mungai et al. 2011).

In summary, these observations, which constitute the most extensive characterisation of the effects of hypoxia on isolated non-preconstricted IPAs to date, demonstrate that the release of intracellular Ca²⁺ plays a pivotal role in HPV, both as a direct source of Ca²⁺ during both phases, and as a trigger for SOCE during phase 2. Our results are consistent with the concept that phase 2 but not phase 1 requires an increase in intracellular [ROS] or an oxidising shift in the cytoplasmic redox potential. This concept is also supported by our previous observation that 30 μm H₂O₂ also causes a sustained ryanodine-sensitive rise in [Ca²⁺] in these arteries which, like much of phase 2, persists in the absence of extracellular Ca²⁺ (Pourmahram et al. 2008). We speculate that the remarkable maintenance of phase 2 in the absence of extracellular Ca²⁺ may be possible because the stores involved in releasing Ca²⁺ are deep within the cell, which could minimise the exposure of the plasmalemmal Ca²⁺-ATPase and Na⁺/Ca²⁺ exchanger to a high subplasmalemmal [Ca²⁺] upon their release, and also because the activity of these extrusion mechanisms may be attenuated by hypoxia, as demonstrated for PASMC Na⁺/Ca²⁺ exchange (Wang et al. 2000).

Limitations of the study

We sought in this study to define the mechanisms by which hypoxia causes contraction in PAs, as HPV has long been held to be a response intrinsic to these arteries. This assumption has recently been challenged, however, by Wang et al. (2012b), who have proposed that HPV, at least in the mouse, is largely a function of a hypoxia-induced depolarisation of the endothelium in the pulmonary capillaries which surround the alveoli. This is then conducted upstream to the PA and initiates HPV by causing endothelial release of vasoconstricting epoxyeicosatrienoic acids. This proposal is provocative, but remains to be confirmed. In any case, it is clear that HPV in vivo is powerfully modulated by a host of factors, including endogenous pretone and nitric oxide (Sylvester et al. 2012). Therefore, since our results were obtained only from isolated PAs in the absence of pretone, it remains to be determined to what extent the mechanisms we have described apply to HPV in the intact animal.

Glossary

AMPK

AMP-activated protein kinase

cADPR

cyclic ADP ribose

CaSR

Ca²⁺-sensing receptor

CPA

cyclopiazonic acid

DHLA

dihydrolipoic acid

GSH

reduced glutathione

GSSG

oxidised glutathione

HPV

hypoxic pulmonary vasoconstriction

NAADP

nicotinic acid adenine dinucleotide phosphate

IPA

intrapulmonary artery

PA

pulmonary artery

PASMC

pulmonary artery smooth muscle cell

PKC

protein kinase C

PSS

physiological saline solution

ROK

rho kinase

ROS

reactive oxygen species

RyR

ryanodine receptor

SERCA

sarco/endoplasmic reticulum Ca²⁺-ATPase

SR

sarcoplasmic reticulum

SOCE

store-operated Ca²⁺ entry

TP receptor

thromboxane receptor

VGCCs

voltage-gated Ca²⁺ channels

80KPSS

PSS containing 80 mm KCl

Additional information

Competing interests

None declared.

Author contributions

M.J.C., J.P.-L. and P.I.A. contributed to the conception and design of the experiments, collection, analysis and interpretation of data, drafting the article or revising it critically for important intellectual content. S.B. contributed to the conception and design of experiments. J.P.T.W. contributed to the conception and design of the experiments, drafting the article or revising it critically for important intellectual content. All authors approved the final submitted version. All experiments were carried out in the Division of Asthma, Allergy and Lung Biology, King's College London.

Funding

This work was supported by a Wellcome Trust Programme grant (087776) to J.P.T.W. and P.I.A. and a British Heart Foundation PhD studentship (FS/05/117/19967) which supported M.J.C.