Adp-ribosyl cyclase and cyclic ADP-ribose hydrolase act as a redox sensor. a primary role for cyclic ADP-ribose in hypoxic pulmonary vasoconstriction.

Hypoxic pulmonary vasoconstriction is unique to pulmonary arteries and serves to match lung perfusion to ventilation. However, in disease states this process can promote hypoxic pulmonary hypertension. Hypoxic pulmonary vasoconstriction is associated with increased NADH levels in pulmonary artery smooth muscle and with intracellular Ca(2+) release from ryanodine-sensitive stores. Because cyclic ADP-ribose (cADPR) regulates ryanodine receptors and is synthesized from beta-NAD(+), we investigated the regulation by beta-NADH of cADPR synthesis and metabolism and the role of cADPR in hypoxic pulmonary vasoconstriction. Significantly higher rates of cADPR synthesis occurred in smooth muscle homogenates of pulmonary arteries, compared with homogenates of systemic arteries. When the beta-NAD(+):beta-NADH ratio was reduced, the net amount of cADPR accumulated increased. This was due, at least in part, to the inhibition of cADPR hydrolase by beta-NADH. Furthermore, hypoxia induced a 10-fold increase in cADPR levels in pulmonary artery smooth muscle, and a membrane-permeant cADPR antagonist, 8-bromo-cADPR, abolished hypoxic pulmonary vasoconstriction in pulmonary artery rings. We propose that the cellular redox state may be coupled via an increase in beta-NADH levels to enhanced cADPR synthesis, activation of ryanodine receptors, and sarcoplasmic reticulum Ca(2+) release. This redox-sensing pathway may offer new therapeutic targets for hypoxic pulmonary hypertension.

Since it was first described over 50 years ago, hypoxic pulmonary vasoconstriction (HPV) 1 has been recognized as the critical and distinguishing characteristic of the blood vessels of the lung (1). Thus, in marked contrast to systemic arteries, which dilate in response to hypoxia, pulmonary arteries constrict. Physiologically, HPV contributes to the matching of lung perfusion and ventilation. However, when alveolar hypoxia is global, as it is in disease states such as cystic fibrosis, emphysema, and mountain sickness, it results in pulmonary hypertension, which can ultimately lead to right heart failure. Unfortunately, the precise mechanisms that underpin HPV remain to be identified, and current therapies for hypoxic pulmonary hypertension are poor.
Certain key characteristics of HPV have been described. In isolated pulmonary arteries, HPV is biphasic. An initial transient constriction (phase 1) is followed by a slowly developing, sustained phase of constriction (phase 2). It is widely thought that the first phase of constriction is initiated by a reduction in membrane K ϩ conductance in pulmonary artery smooth muscle cells (2)(3)(4), membrane depolarization, and Ca 2ϩ influx through voltage-gated Ca 2ϩ channels (5)(6)(7)(8). Phase 2 of the constriction is tonic and may depend on the release of a vasoconstrictor from the endothelium, which sensitizes the contractile apparatus to Ca 2ϩ (9,10).
Our recent findings (11) do not support the above hypothesis. They suggest that hypoxia may, by activating a mechanism intrinsic to pulmonary artery smooth muscle cells, induce intracellular Ca 2ϩ release from ryanodine-sensitive stores in the absence of transmembrane Ca 2ϩ influx. This proposal is also supported by the findings of others (12,13). Also, we have established that the hypoxia-induced SR Ca 2ϩ release initiates and maintains acute HPV (11). Details of the signal transduction pathway remain to be clarified. One possibility is that the ␤-NAD ϩ metabolite cyclic ADP-ribose (cADPR; Refs. 14 -16), a messenger that regulates SR Ca 2ϩ release via ryanodine receptors (RyRs) in a variety of cell types (17)(18)(19), plays a role in this process.
We have investigated the role of the ␤-NAD ϩ :␤-NADH ratio in regulating cADPR synthesis in pulmonary artery smooth muscle because of the fact that (a) the cellular redox couple ␤-NAD ϩ is the recognized substrate for cADPR synthesis; (b) the redox state of O 2 -sensing cells is uniquely sensitive to changes in the level of O 2 (20); (c) hypoxia has been shown to reduce the NAD(P): NAD(P)H ratio in all O 2 -sensing cells studied to date, including carotid body type I cells (20,21), airway neuroepithelial cells (22), and pulmonary artery smooth muscle cells (3,23); and (d) the hypoxia-induced fall in the NAD(P): NAD(P)H ratio in carotid body type I cells occurs with a time course similar to the hypoxia-induced increase in intracellular Ca 2ϩ (21).
We report that the enzyme activities for cADPR synthesis and metabolism are particularly high in pulmonary artery smooth muscle, as opposed to systemic artery smooth muscle. * This work was supported by the Wellcome Trust and the BBSRC. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  We show that hypoxia, by increasing ␤-NADH levels, inhibits cADPR hydrolase, which in turn promotes an increase in cADPR levels in pulmonary artery smooth muscle. Furthermore, we show that a membrane-permeant cADPR antagonist, 8-bromo-cADPR, abolishes the second, sustained phase of HPV. The role of ADP-ribosyl cyclase and cADPR hydrolase as a redox sensor is discussed.
Homogenates-The endothelium was removed by rubbing the luminal surface with a cotton bud, and 2-mm artery strips were cut and placed in 1 ml of Ca 2ϩ -free sucrose-HEPES buffer (250 mM and 20 mM, respectively, pH 7.2). The preparation was homogenized by 2ϫ 5-s bursts using an ultra-turrex homogenizer; 2ϫ 5 strokes using a glass Dounce homogenizer (pestle D). Homogenates were centrifuged for 12 min at 2,000 ϫ g and 4°C to remove debris and nuclei. The supernatant was stored at Ϫ70°C. Production of cADPR from 2.5 mM ␤-NAD ϩ and metabolism of 5-10 M cADPR, respectively, was assessed in 10 -50-l samples of smooth muscle homogenate at 37°C. Test samples (5 l) were assayed for cADPR using a Ca 2ϩ release bioassay (see below).
Protein Concentration-Estimated using the BCA protein assay (Sigma).
Electrophysiology-Membrane potential recordings in isolated pulmonary artery smooth muscle cells were made at 22°C, under current clamp (I ϭ 0) and using the whole cell configuration of the patch clamp technique , as described previously (31). Data were acquired and analyzed using an Axopatch 200B amplifier and pCLAMP 6.0 data acquisition and analysis software (Axon Instruments). The pipette solution was 140 mM KCl, and 10 mM HEPES, pH 7.4. The bath solution was 130 mM NaCl, 5 mM KCl, 2 mM glucose, 10 mM HEPES, 1.7 mM CaCl 2 , and 2 mM MgCl 2 , pH 7.45.
Small Vessel Myography-Vessels were always obtained from the same anatomical position within the lung, i.e. third order branches of the pulmonary arterial tree (inner diameter, 300 -400 m). Vessels (2-3 mm in length) were mounted on to the jaws of an automated myograph (AM10, Cambustion Biological, Cambridge, UK) using 50-m tungsten wire. Initial tension was set to be equivalent to typical pulmonary arterial pressure. The detailed technique, protocol, and theory have been described previously (32). Vessels were bathed in PSS B (constituents as for PSS A, but with 24 mM NaHCO 3 and no HEPES) at 37 Ϯ 1°C and in a bath volume of 4 ml. The solution was bubbled with 75% N 2 , 20% O 2 , 5% CO 2 to maintain a pH of 7.4.
All arteries were first subjected to four exposures of high K ϩ (75 mM) to test their responsiveness and stability. Resting tension was taken to be zero. The bath chambers containing PSS B were individually sealed and bubbled with either normoxic (154 -160 Torr; 75% N 2 , 20% O 2 , 5% CO 2 ) or hypoxic (16)(17)(18)(19)(20)(21); 93% N 2 , 2% O 2 , 5% CO 2 ) gas, supplied via a gas-mixing flowmeter (Cameron Instruments Ltd., Port Aransas, TX). The desired PaO 2 was confirmed using an Oxel O 2 electrode and meter (World Precision Instruments). All drugs were applied to the bath directly. All solutions were warmed to 37°C before adding them to the bath.
When required, the endothelium of the arteries was removed by rubbing the inner surface with braided silk surgical thread. Removal of the endothelium was assessed by the ability of 100 M acetylcholine to relax constrictions induced by 1 M prostaglandin F 2 ␣.
Extraction and Measurement of Endogenous cADPR-Second and third order branches of the pulmonary artery were placed in chambers containing PSS B that were individually sealed and bubbled with either normoxic (154 -160 Torr; 75% N 2 , 20% O 2 , 5% CO 2 ) or hypoxic (16 -21 Torr; 93% N 2 , 2% O 2 , 5% CO 2 ) gas as described above (see small vessel myography). Vessels were quickly removed from the experimental chambers and snap frozen in liquid nitrogen. Acid extraction of nucleotides/cADPR was carried out using a variation of a method described previously (33). Briefly, the frozen tissue was powdered, added to icecold 3 M perchloric acid (1:1 w/v) and sonicated for 20 s. After sonication the samples were left in an ice-salt bath (Ϫ5°C) for 30 min to allow for extraction of nucleotides. Precipitated protein was then removed by centrifugation (15,000 ϫ g for 10 min). The supernatant was neutralized by the addition of 2 M KHCO 3 . The potassium perchlorate precipitate was removed by centrifugation at 15,000 ϫ g for 10 min. To remove contaminating nucleotides that weakly interfere with [ 32 P]cADPR binding, the neutralized acid extracts were treated with NADase (0.25 unit/ml), nucleotide pyrophosphatase (1.75 units/ml), alkaline phosphatase (50 units/ml), and apyrase (5 units/ml) for 4 h, as previously described (15). The concentration of cADPR in acid extracts was assessed by comparing inhibitory effects on the [ 32 P]cADPR binding assay, with a standard curve constructed using authentic cADPR. As a control each sample was heat treated at 85°C for 45 min, to hydrolyze cADPR to ADPR. The inhibitory effect on [ 32 P]cADPR binding of all samples was abolished by heat treatment. The recovery of cADPR, monitored by recovery of [ 32 P]cADPR added prior to acid extraction, was 69.9 Ϯ 5.2% (n ϭ 3). Correction was introduced for recovery of cADPR.
Statistical Significance-Statistical significance was assessed by analysis of variance and assumed if p Ͻ 0.05. Data were expressed as the means Ϯ S.E. for n animals tested unless stated. In biochemical assays each sample was assayed in triplicate. Fig. 1A illustrates the time course of cADPR synthesis from 2.5 mM ␤-NAD ϩ in smooth muscle homogenates from third order branches of the pulmonary arterial tree, as determined by the sea urchin homogenate bioassay. The synthesis was complete after 45 min, when 18 Ϯ 1.7 nmol of cADPR were synthesized per mg of protein (n ϭ 5 animals). Homogenates showed a similar timedependent hydrolysis of 5 M cADPR, assessed by bioassay. This was complete at 60 min, when 10 Ϯ 0.6 nmol of cADPR had been metabolized per mg of protein (n ϭ 4 animals; Fig.  1B). The inset in Fig. 1A shows that 250 M NGD ϩ , an alternative substrate for ADP-ribosyl cyclase, yielded a fluorecent cyclic product (cyclic GDP-ribose) when NGD ϩ was added to smooth muscle homogenates of third order branches of the pulmonary artery. When NGD ϩ was excluded, no change in fluorescence was observed with time (trace 1). Upon addition of NGD ϩ a significant increase in the fluorescence was observed with time (trace 2). The increase in fluorescence was clearly dependent on cyclic GDP-ribose formation and therefore ADPribosyl cyclase, because it was inhibited by the ADP-ribosyl cyclase antagonist nicotinamide (10 mM; trace 3; see also Refs. 25 and 26).

Synthesis and Degradation of cADPR Occurs in Pulmonary Artery Smooth Muscle Homogenates-
cADPR Synthesis Occurs at an Intracellular Site-Because both ecto-enzymes and intracellular enzymes with ADP-ribosyl cyclase activities have been reported (37), it was important to show whether or not cADPR could be synthesized intracellularly, because this would probably be more relevant to an intracellular signaling role. To achieve this we studied the enzyme activities in the cellular fractions obtained from differential centrifugation of smooth muscle homogenates from third order branches of the pulmonary artery. Fig. 2A shows that the synthesis of cADPR was detected in each of the membrane fractions. However, the rate of synthesis was highest in the P3 (microsomal; SR and mitochondria) fraction at 9.5 Ϯ 1.1 nmol/mg of protein/h (n ϭ 3 animals), compared with 2.5 Ϯ 0.7 nmol/mg of protein/h (n ϭ 3 animals) in the P2 fraction (plasma membrane), 0.25 Ϯ 0.01 nmol/mg of protein/h (n ϭ 3 animals) in the P1 fraction (nuclear and whole cell debris), and 0.01 Ϯ 0.1 nmol/mg of protein/h (n ϭ 3 animals) in the S3 fraction (cytosolic). Fig. 2B shows a similar distribution for cADPR metabolism. The rate of metabolism was highest in the P3 (microsomal; SR and mitochondria) fraction at 2.6 Ϯ 0.3 nmol/mg of protein/h (n ϭ 3 animals), compared with 0.92 Ϯ 0.02 nmol/mg of protein/h (n ϭ 3 animals) in the P2 fraction (plasma membrane). In contrast, the hydrolase activity in the P1 fraction (nuclear and whole cell debris; n ϭ 3 animals) and in the S3 fraction (cytosolic; n ϭ 3 animals), respectively, was undetectable. Fig. 2C shows the distribution of a 5Ј nucleotidase reaction used to estimate the relative level of plasma membrane associated protein in all four fractions. The highest level of 5Ј nucleotidase activity was 129 Ϯ 19 units/liter for the P2 (plasma membrane) fraction (n ϭ 3 animals), compared with 86 Ϯ 5.5 U/liter (n ϭ 3 animals) for the P3 (microsomal) fraction and 56 Ϯ 2.2 U/liter (n ϭ 3 animals) for the P1 (nuclear and whole cell debris) fraction. Clearly, the P2 fraction had a much lower level of ADP-ribosyl cyclase and cADPR hydrolase activity than the P3 pellet. The ADP-ribosyl cyclase and cADPR hydrolase activity did not, therefore, follow the plasma membrane contamination.  2D shows cADPR synthesis using whole and saponin (50 g/ml) permeabilized pulmonary artery smooth muscle cells isolated from the large extrapulmonary artery. A significantly greater level of synthesis was detected in permeabilized smooth muscle cells. Thus, after a 120-min incubation with 2.5 mM ␤-NAD ϩ the production of cADPR measured 0.51 Ϯ 0.07 nmol/mg protein (n ϭ 3 animals) in whole cells and 1.5 Ϯ 0.4 nmol/mg protein (n ϭ 3 animals) in paired and time-matched permeabilized cells (Fig. 2D). A clear separation in the level of cADPR synthesis in whole versus permeabilized pulmonary artery smooth muscle cells, respectively, was not, however, observed at the 60-min time point. This apparent nonlinearity may be a product of (a) the sensitivity of the Ca 2ϩ release bioassay to cADPR and (b) the time required for sufficient quantities of the newly synthesized cADPR to accumulate in the samples tested. Note that the saponin concentration used was found to have no effect on the Ca 2ϩ release bio-assay (see methods). Fig. 2D (inset) shows a picture of an isolated and saponin permeabilized pulmonary artery smooth muscle cell in suspension. Fig. 3A compares the level of cADPR synthesis by homogenates of pulmonary and systemic arterial smooth muscle following a 1-h incubation with 2.5 mM ␤-NAD ϩ . The quantity of cADPR produced was high in smooth muscle homogenates from all sections of the pulmonary arterial tree, whereas no significant production could be detected in homogenates of either the aorta or mesenteric artery. Moreover, the level of cADPR synthesis was inversely related to pulmonary artery diameter, measuring 1.0 Ϯ 0.4 nmol/mg of protein/h (n ϭ 4 animals) in homogenates of the large conduit or extrapulmonary artery, 9.1 Ϯ 2.8 nmol/mg protein/h (n ϭ 5 animals) in the second order branches of the pulmonary artery (intrapulmonary artery) and 18 Ϯ 1.7 nmol/mg protein/h (n ϭ 6 animals) in the third order branches. Fig. 3B shows that the rate of hydrolysis of 5 M cADPR in the same series of artery homogenates followed a similar pattern, measuring 11 Ϯ 1.5 nmol/mg protein/h (n ϭ 4 animals) in smooth muscle homogenates of the third order branches, 5.7 Ϯ 2.3 nmol/mg protein/h (n ϭ 5 animals) in second order branches, and 2.3 Ϯ 2.2 nmol/mg protein/h (n ϭ 4 animals) in homogenates of the extrapulmonary artery. In contrast to cADPR synthesis, a small amount of cADPR hydrolysis was detected in homogenates of both the aorta (0.6 Ϯ 0.3 nmol/mg protein/h; n ϭ 3 animals) and the near-resistance sized mesenteric arteries (200 -1000 m external diameter; 0.4 Ϯ 0.2 nmol/mg protein/h, n ϭ 3 animals). These findings suggest that ADP-ribosyl cyclase and cADPR may play an important role in the regulation of pulmonary artery smooth muscle cell function.

ADP-ribosyl Cyclase and Hydrolase Activity Is Differentially Distributed in Pulmonary and Systemic Artery Smooth Muscle-
Modulation by ␤-NADH of cADPR Synthesis and Metabolism- Fig. 4A compares the rate of synthesis of cADPR from a maximally effective concentration (2.5 mM) of ␤-NAD ϩ to the rate of synthesis of cADPR from 25 mM ␤-NADH (n ϭ 4 animals). From these data ␤-NADH can be seen to be a poor substrate for ADP-ribosyl cyclase in pulmonary artery smooth muscle. However, addition of ␤-NADH in combination with ␤-NAD ϩ could produce up to a 3-fold increase in cADPR synthesis when compared with the time-matched synthesis obtained in the presence of ␤-NAD ϩ alone. Fig. 4B shows the effect of ␤-NADH (1-10 mM) on the synthesis of cADPR from a maximally effective concentration of ␤-NAD ϩ (2.5 mM) in smooth muscle homogenates from third order branches of the pulmonary artery. ␤-NADH increased the rate of synthesis of cADPR from ␤-NAD ϩ in a concentration-dependent manner. The rate of synthesis of cADPR increased from 13.52 Ϯ 1.5 nmol/mg of protein/h (n ϭ 4 animals) in the absence of ␤-NADH to a maximum of ϳ30 nmol/mg of protein/h (n ϭ 4 animals) in the presence of 4 -10 mM ␤-NADH. The fact that ␤-NADH increased cADPR synthesis from a maximally effective concentration of ␤-NAD ϩ raised the possibility that ␤-NADH may mediate this effect via the inhibition of cADPR hydrolase. Fig.  4C, shows that 4 mM ␤-NADH inhibited the hydrolysis of cADPR (10 M) in pulmonary artery smooth muscle homogenates. In smooth muscle homogenates of third order branches of the pulmonary artery, 3.63 Ϯ 0.49 nmol of cADPR were metabolized/mg protein/h (n ϭ 3 animals) in the absence of ␤-NADH. This fell to 0.61 Ϯ 0.37 nmol/mg protein/h (n ϭ 3 animals) in the presence of 2 mM ␤-NADH and was undetectable in the presence of 4 mM ␤-NADH. Taken together these findings suggest that net amount of cADPR accumulated from ␤-NAD ϩ may be increased by ␤-NADH in a nonadditive and synergistic manner. This may be due, at least in part, to the inhibition by ␤-NADH of cADPR hydrolase. Note that the small amount of cADPR synthesis observed with 25 mM ␤-NADH may be due, in part, to the 0.5% ␤-NAD ϩ contamination.
Hypoxia Increases cADPR Content in Pulmonary Artery Smooth Muscle-We used a sea urchin egg homogenate [ 32 P]cADPR receptor binding assay to compare the cADPR content of pulmonary artery smooth muscle under normoxic (151-160 Torr) and hypoxic (16 -21 Torr) conditions. Fig. 5   FIG. 2. The enzymes for the synthesis of cADPR are located at an intracellular site. A, the rate of cADPR synthesis (nmol/mg protein/h) in the P1 (nuclear and whole cell debris), P2 (plasma membrane), P3 (microsomal), and S3 (cytosolic) fractions of smooth muscle homogenates of third order branches of the pulmonary artery; each fraction was separated by differential centrifugation. B, the rate of cADPR metabolism (nmol/mg protein/h) in each fraction. C, the relative plasma membrane contamination as indicated by the 5Ј nucleotidase (5ЈND) activity (units of activity/liter) in P1, P2, P3, and S3 fractions. D, production of cADPR by whole (open bars) and saponin permeabilized (black bars) pulmonary artery smooth muscle cells acutely isolated from the extrapulmonary artery. The inset shows a typical image of an isolated pulmonary artery smooth muscle cell in sucrose-HEPES buffer. All incubations were carried out at 37°C and in triplicate. Bars represent the means Ϯ S.E. for Ն 3 animals. *, significance of p Ͻ 0.05.

FIG. 3. Enzyme activities for cADPR synthesis and metabolism in rabbit pulmonary and systemic arteries.
A, the rate of cADPR production from 2.5 mM ␤-NAD ϩ in homogenates of smooth muscle from third order branches of the pulmonary artery (RP), second order branches of the pulmonary artery (IP), and the main or extrapulmonary artery (EP). These are compared with the rate of cADPR production in the thoracic aorta, third order branches of the mesenteric artery (RM) and the main conduit mesenteric artery (CM). B, rate of degradation of 5 M cADPR in smooth muscle homogenates. All incubations were carried out at 37°C and in triplicate. The bars represent the means Ϯ S.E. for Ն3 animals.
shows that the level of cADPR in second order branches of the pulmonary artery increased ϳ2-fold in the presence of hypoxia from 1.5 Ϯ 0.4 pmol/mg protein to 3.6 Ϯ 0.5 pmol/mg protein (n ϭ 3 animals). Significantly, the increase in cADPR induced by hypoxia was greater still in third order branches of the pulmonary artery, in which we measured a 10-fold increase in cADPR from 1.5 Ϯ 0.6 pmol/mg protein to 19.7 Ϯ 1.1 pmol/mg protein (n ϭ 3 animals).
cADPR Releases Ca 2ϩ from Ryanodine-sensitive SR Stores in Isolated Pulmonary Artery Smooth Muscle Cells-We examined the sensitivity of pulmonary artery smooth muscle Ca 2ϩ stores to cADPR. This was achieved by applying cADPR intracellularly from a patch pipette in the whole cell configuration and in current clamp mode (I ϭ 0). Changes in intracellular Ca 2ϩ were monitored by recording changes in the membrane potential mediated via the activation of large conductance Ca 2ϩ -activated potassium channels (BK Ca ) in smooth muscle cells isolated from second and third order branches of the pulmonary artery. In the absence of cADPR the membrane potential was stable at approximately Ϫ45 Ϯ 5 mV (n ϭ 3 cells), in agreement with the findings of others (38). In marked contrast, Fig. 6A shows the effect of intracellular dialysis of 30 M cADPR from the patch pipette; dialysis began immediately after breaking the membrane patch under the pipette tip, i.e. at the start of the record shown. Clearly, cADPR induced a pronounced hyperpolarization of the membrane potential. The hyperpolarization reached a maximum after ϳ1-2 min, at which point pronounced oscillations in the membrane potential were observed. When the oscillations in membrane potential commenced the smooth muscle cells began to contract, and it proved impossible to hold the smooth muscle cells in a tight whole cell configuration for any length of time. On average (mean Ϯ S.E.; n ϭ 5 cells), the membrane potential hyperpolarized from Ϫ48.6 Ϯ 5.3 mV (recorded immediately after entering the whole cell configuration) to Ϫ73 Ϯ 2.4 mV. The hyperpolarization, oscillations in membrane potential, and cell contraction were not observed after the ryanodine-sensitive SR stores had been depleted by prior application of 10 M ryanodine and 10 mM caffeine (Fig. 6B). Under these conditions the membrane potential remained stable at Ϫ55 Ϯ 3 mV (n ϭ 4 cells) for the duration of the experiment (Յ20 min). The activation of BK Ca channels was confirmed by the fact that the hyperpolarization was not observed in the presence of TEA (10 mM; not shown) or iberiotoxin (100 nM; not shown). These data suggest that cADPR can induce Ca 2ϩ release from ryanodinesensitive SR stores in pulmonary artery smooth muscle cells. (25), on HPV in isolated rabbit pulmonary artery rings. Fig. 7A shows that hypoxia induced a characteristic biphasic constriction in an intact pulmonary artery ring. Fig. 7B shows the response obtained in the absence of the pulmonary artery endothelium. The initial fast transient constriction seen in Fig. 7A remains, but the second slowly developing, tonic phase of constriction is lost. Instead the constriction falls to a plateau level above base line that is maintained for the duration of exposure to hypoxia. Previous studies (11) have shown that the initial fast transient constriction and the maintained plateau rely on Ca 2ϩ release from ryanodine-sensitive SR stores. Panels C and D of Fig. 7 show the effect on the hypoxic constriction of preincubating (10 min) intact (C) and de-endothelialized (D) arteries with 300 M 8-bromo-cADPR. Panels E and F of Fig. 7 show the mean (1-min intervals) Ϯ S.E. (5-min intervals) for the constriction obtained in the presence and absence of 300 M 8-bromo-cADPR, both with (E) and without (F) the endothelium. The peak of the fast transient constriction remained unaffected in the presence of 300 M 8-bromo-cADPR (n ϭ 6 arteries). In marked contrast, the second phase of the hypoxia-induced constriction was abolished both in the presence and absence of the endothelium (n ϭ 6 arteries). These data suggest that prolonged, cADPR-dependent SR Ca 2ϩ release is required to maintain hypoxia-induced constriction of pulmonary artery smooth muscle. However, the full development of phase 2 of HPV also requires the release of an endothelium-derived vasoconstrictor (10). Because 8-bromo-cADPR abolished phase 2 of HPV when the endothelium was present, we cannot rule out that this was due, in part, to its inhibition of vasoconstrictor release from the endothelium or the blockade of the action of the vasoconstrictor in the smooth muscle. We know that the endothelium-derived vasoconstrictor induces constriction by sensitizing the smooth muscle myofilaments to Ca 2ϩ and that it does not appear to mobilize Ca 2ϩ in its own right (10). Thus, we would expect arteries that have been preconstricted by Ca 2ϩ derived from another source (e.g. the extracellular fluid) to constrict further on release of the endothelium-derived vasoconstrictor during hypoxia, even when maintained SR Ca 2ϩ release has been blocked by 8-bromo-cADPR. Because K ϩ induces pulmonary artery constriction by depolarizing the smooth muscle cell membrane and activating voltage-gated Ca 2ϩ channels, we were able to preconstrict pulmonary arteries with K ϩ even in the presence of 8-bromo-cADPR. We used 20 mM K ϩ to provide a plateau constriction similar in magnitude (2.4 Ϯ 0.2 mN/mm, n ϭ 6) to the endothelium-independent constriction that was maintained by hypoxia and blocked by 8-bromo-cADPR (2.8 Ϯ 1.1 mN/mm, n ϭ 6; see also Fig. 7 D). Fig. 8A shows a control response to hypoxia (16 -21 Torr) as before. Fig. 8B shows the response of the same vessel to hypoxia after the vessel had been preconstricted with 20 mM K ϩ . Fig. 8C shows the effect of 300 M 8-bromo-cADPR on the constriction to hypoxia, after the vessel had been preconstricted with 20 mM K ϩ . Under these conditions and when reported as a percentage of the constriction to 75 mM K ϩ , hypoxia induced a typical biphasic constriction similar in magnitude to control both in the presence and absence of 8-bromo-cADPR. The peak of phase 1 measured 62 Ϯ 11% in the absence of 8-bromo-cADPR and in the absence of K ϩ (20 mM)-induced preconstriction; 55 Ϯ 10% in the absence of 8-bromo-cADPR and in the presence of K ϩ (20 mM)-induced preconstriction; and 53 Ϯ 7% in the presence of 300 M 8-bromo-cADPR and K ϩ (20 mM)-induced preconstriction (n ϭ 5 arteries). The peak of phase 2 of HPV after 40 min of exposure to hypoxia measured 32 Ϯ 8% in the absence of 8-bromo-cADPR and in the absence of K ϩ (20 mM)-induced preconstriction; 36 Ϯ 8% in the absence of 8-bromo-cADPR and in the presence of K ϩ (20 mM)-induced preconstriction; and 25 Ϯ 9% in the presence of both 300 M 8-bromo-cADPR and K ϩ (20 mM)-induced preconstriction (n ϭ 5 arteries). This finding suggests that although 8-bromo-cADPR inhibits SR Ca 2ϩ release to hypoxia, it does not affect hypoxiainduced release of the endothelium-derived vasoconstrictor(s), subsequent myofilament sensitization, or the constriction to Ca 2ϩ per se.

8-Bromo-cADPR Inhibits Hypoxic Pulmonary Vasoconstriction in Pulmonary Artery Rings in Vitro-We next investigated the effect of a membrane-permeant antagonist of cADPR, 8-bromo-cADPR
Extracellular application of 300 M cADPR, which is membrane-impermeant, had no effect on either resting tone (after 60 min) or acute HPV (20 min of preincubation) in isolated pulmonary artery rings (n ϭ 4; not shown). This supports the view that the cADPR antagonist 8-bromo-cADPR is acting intracellularly. DISCUSSION We have investigated the possible role of ADP-ribosyl cyclase, cADPR hydrolase, and cADPR as a redox sensor in pulmonary artery smooth muscle. Our findings suggest that increased cADPR synthesis may mediate, in part, the hypoxiainduced increase in SR Ca 2ϩ release in pulmonary artery smooth muscle and hence contribute to HPV.
We measured a high level of cADPR production in smooth muscle homogenates from pulmonary arteries, whereas little synthesis could be detected in smooth muscle homogenates from aortic or mesenteric arteries. A similar trend was observed with respect to the metabolism of cADPR, although a small amount of metabolism was observed in homogenates of both aortic and mesenteric artery smooth muscle. Furthermore we have shown that synthesis and metabolism of cADPR occurs intracellularly. These findings point to an important role for ADP-ribosyl cyclase, cADPR hydrolase and cADPR in the regulation of pulmonary artery function. Support for this proposal comes from the fact that the level of cADPR synthesis in pulmonary artery homogenates was inversely related to artery diameter, being 18-fold higher in homogenates from the small third order branches than it was in the homogenates of the main extrapulmonary artery. Because the magnitude of the hypoxic constriction is also inversely related to artery diameter (39,40), the ability of hypoxia to constrict pulmonary arteries increases with the increasing levels of smooth muscle ADPribosyl cyclase and cADPR hydrolase activity. An intriguing possibility, therefore, is that hypoxia may increase the rate of synthesis of cADPR in pulmonary artery smooth muscle.
In a separate series of experiments, we found that ␤-NADH induced a concentration-dependent increase in the rate of cADPR synthesis from a fixed and maximally effective concentration of its substrate ␤-NAD ϩ . Because ␤-NADH was found to be a poor substrate for cADPR synthesis, this effect was clearly nonadditive, i.e. synergistic. We also found ␤-NADH to inhibit, in a concentration-dependent manner, cADPR metabolism in pulmonary artery smooth muscle homogenates. Thus, ␤-NADH may promote an increase in cADPR accumulation from ␤-NAD ϩ , at least in part, by blocking the metabolism of cADPR by a cADPR hydrolase. The effect of ␤-NADH on cADPR synthesis may be of importance to the regulation of O 2 -sensing cells because the redox state of such cells is uniquely sensitive to changes in O 2 (3,20,21,22), and hypoxia has been shown to increase ␤-NADH levels in all O 2 -sensing cells studied to date (3,20,21,22,23). Previous investigations of pulmonary artery smooth muscle suggest that total tissue ␤-NAD ϩ levels may be in the mM range during normoxia and hypoxia and that during hypoxia a small fall in ␤-NAD ϩ levels yields a large increase in ␤-NADH levels. Estimates suggest that ␤-NADH concentration in the smooth muscle may increase at least 5-fold during hypoxia from ϳ0.03 mM (normoxia; Ref. 23) to 0.15 mM (hypoxia; Ref. 23). However, it should be noted that any estimates of total tissue ␤-NAD ϩ and ␤-NADH levels are likely underestimates of the true levels that may be experienced in intact cells. The FIG. 7. Hypoxic pulmonary vasoconstriction is inhibited by 8-bromo-cADPR in the presence and absence of the pulmonary artery endothelium. A, control constriction to 75 mM K ϩ . The pulmonary artery ring was then exposed to hypoxia (16 -21 Torr) for 30 min. After the artery had recovered from the exposure to hypoxia, the vessel was then exposed once more to 75 mM K ϩ . B, first shows a control response of a de-endothelialized pulmonary artery ring to 75 mM K ϩ and then the effect of exposing the vessel to hypoxia (16)(17)(18)(19)(20)(21). C, control response to K ϩ (75 mM) and then the effect of preincubation (10 min estimates given here assume a total tissue H 2 O content of 0.75 liter/Kg and assume that the extracellular fluid would account for 20% of the total tissue H 2 O. They do not take into account known subcellular compartmentalization, which may result in significant concentration of ␤-NAD ϩ and ␤-NADH levels and, indeed, cADPR. In fact ␤-NAD ϩ /␤-NADH and the enzymes for cADPR synthesis and metabolism may be localized in cellular compartments from the plasma membrane to the mitochondria (41). In addition hypoxia may reduce the redox state of the cytoplasmic compartment, by modulating glycolysis (42) and/or the mitochondrial redox state (20). It is, however, clear that these data give equivalent ␤-NAD ϩ :␤-NADH ratios of ϳ17 and 5 (23), respectively, and the latter is close to the range (5-0.7) of ␤-NAD ϩ :␤-NADH ratios, over which our data indicate that we may see an increase in cADPR synthesis from ␤-NAD ϩ (Fig. 4B).
Given the above findings and the fact that the hypoxiainduced increase in NAD(P)H autofluorescence in carotid body type I cells occurs over the same time scale as does the hypoxiainduced increase in intracellular Ca 2ϩ concentration (21), it seemed likely that hypoxia may promote SR Ca 2ϩ release by reducing the NAD(P):NAD(P)H ratio and by subsequently increasing cADPR synthesis. This proposal gained support from the finding that tissue cADPR levels increased when pulmo-nary arteries were exposed to hypoxia under physiological conditions. Furthermore, a 5-fold greater increase in cADPR content was observed in third order branches than in second order branches of the pulmonary artery. Thus, like the enzyme activities (see above), the magnitude of the hypoxia-induced increase in cADPR content was inversely related to artery diameter, as is the sensitivity of the arteries to hypoxia (39,40).
In functional studies we first showed that cADPR induced Ca 2ϩ release from ryanodine-sensitive SR stores in isolated pulmonary artery smooth muscle cells, as does hypoxia (11,12,13). We therefore investigated the role of cADPR in HPV, using a membrane-permeant cADPR antagonist, 8-bromo-cADPR. OurfindingsstronglysuggestthathypoxiaactivatesbothcADPRdependent and cADPR-independent Ca 2ϩ release from ryanodine-sensitive SR stores. The first transient phase of constriction to hypoxia remained unaffected in the presence of 8-bromo-cADPR. Thus, a cADPR-independent O 2 -sensing mechanism must initiate the ryanodine-sensitive SR Ca 2ϩ release associated with this phase of HPV (11). Recent investigations have suggested that phase 1 of HPV may result from an initial fall in ATP levels and inhibition of the SR Ca 2ϩ ATPase, leading to an increase in the net efflux of Ca 2ϩ from the SR (Ref. 43 and Fig. 9). This may in turn promote calcium influx because of the subsequent activation of the store refilling current (43). These findings (43) and our own (11), do not support the view (2-8) that phase 1 of HPV in isolated vessels is mediated by Ca 2ϩ influx through voltage-gated Ca 2ϩ channels. In marked contrast, the subsequent sustained phase of acute HPV in de-endothelialized pulmonary artery rings (Fig. 7B), which is also mediated by ryanodine-sensitive SR Ca 2ϩ release (11), was blocked by 8-bromo-cADPR. This finding suggests that a sustained increase in SR Ca 2ϩ release and hence constriction may depend on an increase in cADPR synthesis during HPV (Fig. 9). Perhaps the most important observation, however, was that the second tonic phase of HPV in intact isolated pulmonary artery rings was abolished by 8-bromo-cADPR and could be reconstituted if the arteries were preconstricted with K ϩ (Figs. 7 and 8). These findings suggest that an increase in cADPR levels and subsequent SR Ca 2ϩ release in FIG. 8. Hypoxic pulmonary vasoconstriction is not inhibited by 8-bromo-cADPR in pulmonary arteries preconstricted with potassium. A, control constriction to 75 mM K ϩ . The pulmonary artery ring was then exposed to hypoxia (16 -21 Torr) for 30 min. After the artery had recovered from the exposure to hypoxia, the vessel was then exposed once more to 75 mM K ϩ . B, first shows a control response to 75 mM K ϩ . The vessel was then preconstricted with 20 mM K ϩ and subsequently exposed to hypoxia (16)(17)(18)(19)(20)(21). C, shows a control response to K ϩ (75 mM). The vessel was then preincubated (10 min) with 300 M 8-bromo-cADPR, preconstricted with 20 mM K ϩ and subsequently exposed to hypoxia (16)(17)(18)(19)(20)(21). The results were obtained from the same artery at 37°C.
FIG. 9. Schematic representation of the proposed redox sensing pathway. We propose that hypoxia may increase ␤-NADH levels and thereby increase the net amount of cADPR accumulated from ␤-NAD ϩ , at least in part, by inhibiting cADPR metabolism by the cADPR hydrolase (CH). The increase in cADPR synthesis in combination with the inhibition of the SR Ca 2ϩ ATPase by hypoxia may then promote SR Ca 2ϩ release from ryanodine-sensitive SR stores and constriction via Ca 2ϩ -dependent activation myosin light chain kinase (MLCK). The release by hypoxia of a vasoconstrictor from the pulmonary artery endothelium may then promote further constriction by activating a Rho-associated kinase (ROCK), leading to the inhibition of smooth muscle myosin phosphatase (SMMP) and an increase in myofilament Ca 2ϩ sensitivity. AC, ADP-ribosyl cyclase; ADPR, ADP-ribose. the smooth muscle is essential for the maintenance of acute HPV. Furthermore they show that 8-bromo-cADPR did not block (a) constriction to Ca 2ϩ per se, (b) the release of the endothelium-derived vasoconstrictor in response to hypoxia, or (c) the increase in myofilament Ca 2ϩ sensitivitiy promoted by the released vasoconstrictor. Thus, the release of the endothelium-derived vasoconstrictor during the second tonic phase of HPV (9, 10, 40) may be insufficient to promote pulmonary artery constriction in the absence of maintained cADPR-dependent SR Ca 2ϩ release, despite the fact that the vasoconstrictor may activate a Rho associated kinase and increase smooth muscle myofilament Ca 2ϩ sensitivity (Ref. 44 and Fig. 9).
Given the above, it may be of some significance that the level of cADPR synthesis is at least 2 orders of magnitude higher in pulmonary artery smooth muscle than it is in systemic artery smooth muscle. Thus, in pulmonary artery smooth muscle hypoxia may induce an increase in cADPR levels at least 2 orders of magnitude greater than that observed in systemic artery smooth muscle. This may provide a platform for the evident pulmonary selective effects of hypoxia; systemic arteries dilate in response to hypoxia. This is a significant point, because all three ryanodine receptor subtypes (RyR1, RyR2, and RyR3) may be present in both systemic and pulmonary vascular smooth muscle (45), and all three RyR subtypes can be expressed in a cADPR-sensitive form (46 -48).
In summary, our findings suggest that (a) the enzyme activities for the metabolism of cADPR are significantly higher in pulmonary artery smooth muscle than they are in systemic artery smooth muscle; (b) the expression of these enzyme activities was highest in the smaller pulmonary arteries, which exhibit the highest sensitivity to hypoxia; (c) an increase in ␤-NADH concentration increased the net amount of cADPR synthesized from ␤-NAD ϩ , and this was due, at least in part, to the inhibition by ␤-NADH of cADPR metabolism; (d) hypoxia increased cADPR levels in pulmonary arteries; (e) cADPR induced Ca 2ϩ release from ryanodine-sensitive SR stores in pulmonary artery smooth muscle cells; and (f) 8-bromo-cADPR, a cADPR antagonist, abolished the sustained phase of acute HPV in intact arteries. We propose that ADP-ribosyl cyclase, cADPR hydrolase, and cADPR act as a redox sensor that couples changes in the cellular redox potential to SR Ca 2ϩ release ( Fig.  9) and that cADPR is the primary mediator of acute HPV. This pathway may offer an important new therapeutic target for the treatment of hypoxic pulmonary hypertension. It is also possible that this pathway may play a role in (a) ischemia/ischemic reperfusion injury and (b) Ca 2ϩ signaling in pancreatic ␤ cells, in which cADPR may regulate stimulus secretion coupling and in which glucose induces concomitant oscillations in NAD(P)H autofluorescence and in cytoplasmic Ca 2ϩ (34,49,50).