Rho kinase is an effector underlying Ca2+-desensitizing hypoxic relaxation in porcine coronary artery

Robert L. Wardle, Min Gu, Yukisato Ishida, Richard J. Paul

Abstract

Acute hypoxia dilates most systemic arteries leading to increased tissue perfusion. We have previously shown that at high-stimulus conditions, porcine coronary artery was relaxed by hypoxia without a change in intracellular [Ca2+] (27). This Ca2+-desensitizing hypoxic relaxation (CDHR) was validated in permeabilized porcine coronary artery smooth muscle (PCASM) in which hypoxia decreased force and myosin regulatory light chain phosphorylation (p-MRLC) despite fixed [Ca2+] (10). Rho kinase-dependent phosphorylation of myosin phosphatase-targeting subunit 1 (p-MYPT1) is associated with decreased MRLC phosphatase activity and increased Ca2+ sensitivity of both p-MRLC and force. We recently reported that p-MYPT1 dephosphorylation was a key effector in CDHR (33). In the current study, we tested the hypothesis that Rho kinase and not p-MYPT1 phosphatase is the regulated enzyme involved in CDHR. We used α-toxin to permeabilize deendothelialized PCASM. CDHR was attenuated in contractions attributable to myosin light chain kinase (MLCK, in the presence of the Rho kinase inhibitor Y-27632). In contrast, hypoxia relaxed contractions attributable to Rho kinase phosphorylation of MYPT1 and MRLC or MRLC alone (in the presence of the MLCK inhibitor ML7). Using an in situ assay, we showed that Rho kinase activity, measured as thiophosphorylation of MYPT1 and MRLC, was nearly abolished by hypoxia. The in vitro activity of the catalytically active fragment of Rho kinase was not affected by hypoxia. Our evidence strongly implicates that hypoxia directly inhibits Rho kinase-dependent phosphorylation of MYPT1. This underlies the decreases in both p-MYPT1 and p-MRLC and thereby leads to the Ca2+-desensitizing hypoxic relaxation.

  • smooth muscle
  • oxygen sensing
  • RhoA
  • contractility

systemic vessels, including porcine coronary arteries, dilate to acute hypoxia (27), leading to increased tissue perfusion (35). Smooth muscle relaxation underlies this hypoxic dilation. Smooth muscle is activated by phosphorylation of the LC20 myosin light chain (MRLC). This is achieved by the Ca2+-dependent myosin light chain kinase (MLCK) and modulated by myosin light chain phosphatase (MLCP). The latter can be phosphorylated by a cascade leading to the activation of Rho kinase and phosphorylation of the MLCP subunit myosin phosphatase-targeting subunit 1 (MYPT1) (29). Most proposed mechanisms for hypoxic vasorelaxation ultimately depend on a reduction of myoplasmic intracellular Ca2+ concentration ([Ca2+]i) (16). However, there are reports of hypoxic relaxation in smooth muscle without a decrease in [Ca2+]i (19, 30) and for vascular smooth muscle in particular (1, 12, 27). Since [Ca2+]i remains elevated and force decreases, by definition this involves a decrease in Ca2+ sensitivity of force.

We have previously demonstrated that hypoxia can relax intact porcine coronary artery without a change in myoplasmic [Ca2+]i (27). Hypoxia also relaxed permeabilized coronary artery, in which [Ca2+]i is fixed by EGTA buffers. We termed this Ca2+-desensitizing hypoxic relaxation (CDHR) (10, 33). CDHR may complement mechanisms of hypoxic relaxation that result from a reduction of [Ca2+]i (16). We have established an important role for phosphorylation of MYPT1 (p-MYPT1) in CDHR (33). The nature of the molecular effector(s) and the mechanism(s) leading to CDHR and specifically whether Rho kinase itself may act as an O2 sensor are the focus of this investigation.

MATERIALS AND METHODS

Coronary artery preparation and isometric force measurement.

Young adult porcine hearts were obtained immediately after death from a local slaughterhouse and placed in a cold (4°C) physiological salt solution (PSS). The left anterior descending coronary artery was dissected and cleaned of fat and connective tissue. Each artery was cut into two to twenty adjacent rings (1 mm wide, 6–10 mm circumference). The rings were mechanically deendothelialized by gentle rubbing between the thumb and index finger, and the blotted wet weight of each ring was determined. One ring was used for isometric force measurements, and groups of four adjacent rings were mounted under tension on stainless steel wires for rapid freezing and protein phosphorylation measurements (see Protein phosphorylation). One ring was mounted on wire hooks with one hook fixed and the other connected to a Harvard Apparatus Capacitance force transducer and placed into a tissue bath (1.5 ml) at 37°C containing MOPS-PSS (pH 7.4) of the following composition (in mM): 140 NaCl, 4.7 KCl, 1.2 NaH2PO4, 20 MOPS, 0.02 EDTA, 1.2 MgSO4, 2.5 CaCl2, and 11.2 glucose. The bathing solution was aerated with hydrated air or N2 (after it passed through a water trap that also removed soluble impurities in the gases). Aeration was achieved by directing a stream of gas over the surface of the bath; a plastic shield was positioned on top of the bath to reduce evaporation and aid more rapid equilibration of the solution with the aerating gas.

The rings were equilibrated for 45–60 min. Baseline tension during this period was adjusted to 8 mN, which sets the tissue to the optimal length range for maximum isometric force generation. Following equilibration, rings were stimulated with 80 mM KCl for 10 min for at least two contraction/relaxation cycles until reproducible forces were generated. Test contractions were induced by 30 mM KCl, and force was allowed to plateau. After an attainment of a steady-state force (at ∼15 min poststimulation), the aeration of the bathing solution was switched from air to 100% N2 for 10–20 min. We defined these conditions as “hypoxia.” Hypoxic relaxation of permeabilized porcine coronary artery smooth muscle (PCASM) occurred at 1% to 2% O2, previously measured polarographically (6, 27). Isometric force was recorded by AcqKnowledge software, a digital data acquisition system (Biopac). The force was normalized to cross-sectional area [F/A] = (change in force × circumference)/(2 × wet weight). The decrease in force in response to hypoxia was characterized in terms of the maximum hypoxic relaxation (at 10–20 min), expressed as a percentage of the isometric force immediately preceding aeration with N2. Similar normalization for maximum force and hypoxic relaxation was used for the permeabilized preparations.

Tissue permeabilization.

After hypoxic relaxation for an intact coronary artery ring was validated, the bath temperature was lowered to ∼23°C and the ring was permeabilized. The ring was first calcium depleted in a solution containing (in mM) 5 EGTA, 20 imidazole, 50 KCl, and 150 sucrose, pH 7.4 for 10 min at room temperature. The ring was then incubated with 50 μg/ml α-toxin in the Ca2+-depleting solution for 20–60 min. The ring was rinsed once and equilibrated in a relaxing solution.

Solutions.

Relaxing solution for skinned fibers consisted of (in mM) 10 MgCl2, 7.5 Na2ATP, 4 EGTA, 20 imidazole (pH 6.7), and 10 phosphocreatine. This nominally Ca2+-free solution had a free Ca2+ <10 nM, calculated using a multiple ionic equilibrium program (9). Ca2+-contracting solution for skinned fibers was similar to relaxing solution but also contained (in mM) 2.0 CaCl2 with calculated 1.94 free Mg2+ and 7.2 MgATP (ionic strength of 110) and a free Ca2+ of 1 μM. Rigor solution contained (in mM) 2 MgCl2, 4 EGTA, 20 imidazole (pH 6.7), and 50 KCl with an ionic strength of 110. The solution for thiophosphorylating MYPT1 (MYPT1γS) was similar to the relaxing solution but included 1 mM Na2ATPγS and 300 μM ML7 to inhibit MRLC kinase.

Protocol for thiophosphorylating MYPT1.

Permeabilized artery rings were subjected to a control contraction/relaxation cycle, with the relaxing solution containing 300 μM ML7 to inhibit MRLC kinase. They were then incubated for at least 10 min in MYPT1γS solution (+/−1 μM Y-27632) and finally transferred back to the relaxing solution.

Protein isolation.

At defined times and under specified conditions, a group of four adjacent artery rings were rapidly frozen in 5% trichloroacetic acid, 10 mM 2-mercaptoethanol, and 95% acetone cooled on dry ice and incubated for 20 min. The frozen tissues were then washed twice in dry ice-cooled 5 mM 2-mercaptoethanol-acetone, each for 20 min, followed by a third wash in the same solution overnight. The ring segments were frozen in liquid nitrogen and pulverized vigorously in a dental amalgamator. The resulting powder was dissolved in ice-cold homogenization buffer [0.5 M NaCl, 10 mM MgCl2, 50 mM Tris·HCl (pH 7.5), 0.01% SDS, 1% Triton X-100, and 0.5% deoxycholic acid] containing protease and phosphatase inhibitors (Sigma P8340, P2850, and P5726). Homogenates were centrifuged at 12,000 g for 30 min. The supernatant was saved, and total protein concentration was determined from this sample using the Bradford assay. Protein samples (20 μg) were then subjected to duplicated SDS 4–20% polyacrylamide gel electrophoresis for 2 h. Proteins were transferred from gels to wet nitrocellulose membranes overnight using the transfer apparatus (Bio-Rad, Hercules, CA).

Protein phosphorylation.

The top of the membrane from one of the duplicate gels was used for pT696-MYPT1 measurement. The top of the membrane from the other duplicate gel was used for pT853-MYPT1 measurement. The bottom of at least one of the duplicate gels was used for pS19-MRLC measurement. Membranes were blocked with 1% blocking solution for 1 h, incubated with rabbit anti-pT696-MYPT1 (No. 07–251, Upstate) or rabbit anti-pT853-MYPT1 (No. 36–003, Upstate) or rabbit anti-pS19-MRLC (No. 3671, Cell Signaling Technology) at room temperature for 1.5 h, followed by 1-h incubation in anti-rabbit secondary antibody. After three washes with Tris-buffered saline (TBS) containing Tween 20 (TBST) and one wash in TBS, the membranes were treated with enhanced chemiluminescence reagents (No. RPN 2209, Amersham Biosciences) and exposed to X-ray film.

After phosphospecific immunolabeling was detected, antibodies were stripped from the membranes by incubating them overnight at room temperature in stripping solution (No. 21059, Pierce). After stripped membranes were rinsed in TBST, pan antibodies were used to label total MYPT1 and total MRLC. Membranes were blocked again with 1% blocking solution for 1 h, incubated with mouse anti-MYPT1 (No. 612164, BD Bioscience Pharmingen) or rabbit anti-MRLC (No. 3672, Cell Signaling Technology) at room temperature for 1.5 h, followed by 1-h incubation in anti-mouse or anti-rabbit secondary antibody. After three washes with TBST and one wash in TBS, the membranes were treated with enhanced chemiluminescence reagents (No. RPN 2209, Amersham Biosciences) and exposed to X-ray film. Films were developed and scanned on a desktop scanner.

In situ Rho kinase activity assay.

Paired PCASM rings (from n = 4 arteries, each from a different heart) were exposed to rigor solution for 15 min to deplete ATP. All rings were then frozen at the end of an additional 15 min of treatment, under normoxia or hypoxia (aerating with N2), and with 1 mM GTPγS in relaxing solution (estimated free Ca2+, <10 nM) containing 7.5 mM ATP (control) or relaxing solution without ATP and in the absence (−) or presence (+) of 1 mM ATPγS as the only adenine nucleotide substrate for the kinase-catalyzed transfer of the gamma thiophosphate group to MYPT1 and/or MRLC. Protein extracts were separated on SDS 4–20% acrylamide gels, transblotted to nitrocellulose, and immunolabeled (see Fig. 2 legend for further details).

Fig. 1.

The effects of hypoxia on Ca2+-induced contractions with and without Y-27632 inhibition of Rho kinase or with and without ML7 inhibition of myosin regulatory light chain (MRLC) kinase in α-toxin-permeabilized porcine coronary artery smooth muscle (PCASM). Isometric myograms illustrate general protocols in which activated intact (A) and then permeabilized (A–C) PCASM rings relax to hypoxia (N2). A: the second Ca2+-induced contraction, this time in the presence of Y-27632 (Rho kinase inhibited), exhibited a much attenuated relaxation to hypoxia. This inhibition of hypoxic relaxation is similar to results previously observed in β-escin-permeabilized arteries (10). B–C: Ca2+-desensitizing hypoxic relaxation (CDHR) of PCASM rings with or without myosin phosphatase-targeting subunit 1 (MYPT1) thiophosphorylated (MYPT1γS). B: in the continuous presence of ML7 (MRLC kinase inhibited), 1 μM Y-27632 completely blocked Ca2+-induced contractions. The subsequent Ca2+-induced contraction in the absence of Y-27632 (Rho kinase no longer inhibited) exhibited CDHR similar to that elicited before MYPT1γS. C: in the continuous presence of ML7, the Ca2+-induced contraction in the absence of Y-27632 (maximized dependence on Rho kinase) exhibited CDHR similar to that elicited during a GTPγS-induced (activated Rho and Rho kinase) contraction. Both B and C are typical of experiments on 3 arteries. The unstimulated isometric force baseline was set to 8 mN.

Fig. 2.

Hypoxia inhibits in situ GTPγS-induced (Rho kinase) thiophosphorylation of MYPT1 and MRLC in α-toxin permeabilized PCASM. Paired PCASM rings (from n = 4 arteries, each from a different heart) were exposed to rigor solution for 15 min to deplete ATP. All rings were then frozen at the end of an additional 15 min of treatment, under normoxia or hypoxia (aerating with N2, see materials and methods), and with 1 mM GTPγS in relaxing solution (estimated free Ca2+, <10 nM) containing 7.5 mM ATP (control) or relaxing solution without ATP and in the absence (−) or presence (+) of 1 mM ATPγS as the only adenine nucleotide substrate for the kinase-catalyzed transfer of the gamma thiophosphate group to MYPT1 and/or MRLC. Protein extracts were separated on SDS 4–20% acrylamide gels, transblotted to nitrocellulose, and immunolabeled. Western blots of samples from 2 representative arteries show that normoxic ATP (+) control samples and ATPγS (+) samples both exhibited MYPT1 phosphorylated or thiophosphorylated at both T696 and T853 (pT696- and pT853-MYPT1; A) and MRLC phosphorylated or thiophosphorylated at S19 (pS19-MRLC; B). In contrast, ATPγS (+) samples frozen under hypoxia did not exhibit appreciable thiophosphorylation of either MYPT1 or MRLC. Controls in the absence of ATPγS or ATP were not phosphorylated, and the pan-antibodies (which specifically recognize both phosphorylated/thiophosphorylated and unphosphorylated protein) detected comparable levels of antigen under normoxia and hypoxia.

In vitro Rho kinase activity assay.

A commercial assay (Cat. Nos. CY-1160 and CY-E1160–1; CycLex in Japan and distributed by MBL International, Woburn, MA) was used to determine the relation between the amount of constitutively active Rho kinase catalytic subunit and the phosphorylated T696 (pT696) of a MYPT1 COOH-terminal peptide under normoxia and hypoxia. The entire assay was performed at room temperature and otherwise according to assay kit instructions. We used 0.0025, 0.005, 0.01, and 0.02 units of activity of Rho kinase catalytic subunit per 100 μl reaction solution in each well of the microtiter plate. A dual-wavelength spectrophotometer was used to measure absorbance that was proportional to previous Rho kinase-induced pT696 (reported as absorbance at 450 nm/absorbance at 550 nm minus background). In a control experiment, under the same conditions used for each assay, we used an electrode (YSI) to measure O2 levels in 9 ml of the reaction solution inside the incubation chamber. After the constant flow of air containing ∼21% O2 into the chamber (normoxia) was switched to the same constant flow of 100% N2, the O2 dropped to <2% within 2 min, to <1% within 5 min, and to <0.5% within 9 min, below which it remained for the 60-min incubation period. It is likely that O2 levels changed even more rapidly in the much smaller 100 μl assay reaction solutions.

Statistical analysis.

Data for the kinase assay were analyzed using two-way, repeated-measures ANOVA and two-way ANOVA on mean values. Statistical significance was accepted for P < 0.05. Values are expressed as means ± SE; n represents the number of hearts with one coronary artery ring used for force or four rings used per treatment for protein phosphorylation measurements.

RESULTS

The first experimental series was designed to investigate the effect of hypoxia on contractions that maximized the dependence on MRLC kinase by inhibition of Rho kinase. Figure 1A shows an isometric myogram of a near-maximal KCl contracture in the endothelium-denuded, sarcolemma-intact porcine coronary artery. Aerating the artery ring bathing solution with N2 instead of 21% O2 induced a rapid and reversible relaxation as previously reported (6, 10, 27). After the sarcolemma was permeabilized with α-toxin and the artery ring was activated with a submaximal concentration of Ca2+ (1 μM, fixed with an EGTA buffer), hypoxia induced a CDHR. Previously, we concluded that CDHR under these conditions was attributable to hypoxia-induced reductions in both Rho kinase-phosphorylated MYPT1 and MRLC phosphorylation (33). After pretreatment with relaxing solution (calculated free Ca2+, <10 nM) containing ATPγS, the MRLC kinase inhibitor ML7 (300 μM), and the Rho kinase inhibitor Y-27632 (1 μM), neither MYPT1 nor MRLC was thiophosphorylated (33). After removal of ATPγS and ML7, the artery ring did not contract when returned to ATP-containing relaxation solution (Fig. 1A), verifying functionally that MRLC was not thiophosphorylated. In addition, it shows that proteins thiophosphorylated by kinases not inhibited by ML7 and Y-27632 do not lead to contraction. With the effects of Rho kinase minimized, the effects of hypoxia were greatly attenuated (Fig. 1A), which agrees with our previous studies with β-escin-permeabilized coronary artery [see Fig. 5 (10)]. As a control for tissue viability, at the end of the experiment, arteries were exposed to 6.6 μM Ca2+, which elicited a near maximal contraction.

Conversely, the next two experimental series were designed to investigate the effect of hypoxia on contractions in which the effects of Rho kinase were maximized by inhibition of MRLC kinase. The permeabilized artery rings illustrated in Fig. 1B were pretreated with ATPγS in the continuous presence of ML7 to thiophosphorylate MYPT1 (MYPT1γS) and thereby to reduce MRLC phosphatase activity (33). In the continued presence of ML7 to inhibit MRLC kinase, ATPγS was removed and 1 μM Y-27632 was added to inhibit Rho kinase. A subsequent addition of Ca2+ (1 μM) in this case did not elicit a contraction. This step again shows that kinases not inhibited by ML7 and Y-27632 do not generate a Ca2+ contracture. After Y-27632 was removed, but in the continued presence of the MRLC kinase inhibitor ML7, an addition of Ca2+ now elicited a contraction (Fig. 1B). With MLCP activity reduced by MYPT1γS and with MLCK inhibited by ML7, this contraction is likely due to direct phosphorylation of MRLC by a Ca2+-activated Rho kinase, as previously argued (3, 14). Again, hypoxia induces a rapid Ca2+-desensitizing hypoxic relaxation. This contracture was further relaxed by Y-27632, supporting our contention that it was largely mediated by Rho kinase phosphorylation of MRLC.

The third experimental series was also designed to investigate the effect of hypoxia on contractions that were due to Rho kinase. In this series, in contrast to Fig. 1B, the Rho kinase-dependent contractions and CDHR were elicited without thiophosphorylating MYPT1 and thus not eliminating Rho kinase modulation of MLCP at this site. After CDHR of a Ca2+ contraction was demonstrated, the artery was relaxed to baseline force with relaxing solution (Fig. 1C, left). The artery rings were then activated with GTPγS (1 mM), which activates Rho, and, in the presence of a low level of Ca2+ (0.2 μM), leads to phosphorylated MRLC and contraction. After CDHR of this contraction was demonstrated, the artery was relaxed to baseline force with relaxing solution (Fig. 1C, middle). Finally, the artery rings were pretreated with relaxing solution containing ATPγS, ML7 (300 μM), and Y-27632 (1 μM); again, as in previous series, a return to ATP-containing relaxing solution in the continued presence of ML7 to inhibit MRLC kinase did not elicit an increase in force. The addition of Ca2+ (1 μM) elicited a Rho kinase-dominated contraction, which was comparable in magnitude with that induced by GTPγS and also similarly relaxed by hypoxia (Fig. 1C, right).

To summarize thus far, contractions mediated by Rho kinase are relaxed by hypoxia and this CDHR is nearly abolished for contractions mediated by MLCK. The next experimental series was designed to provide biochemical evidence that further distinguishes between whether hypoxia induces the inhibition of Rho kinase or to the activation of p-MYPT1 phosphatase. To this end, we used Western blot analysis to measure Rho kinase activity in situ in the permeabilized arteries. We used 1 mM GTPγS to activate endogenous Rho and Rho kinase and 1 mM ATPγS as substrate for thiophosphorylation of MYPT1 and MRLC in relaxing solution containing 4 mM EGTA. The thiophosphoproteins are resistant to phosphatases, and the Ca2+-free (<10 nM) relaxing solution was used to minimize MRLC kinase activity. Figure 2 shows typical Western blots from this protocol that was applied to paired coronary artery rings from four different porcine hearts. All rings were frozen at the end of a 15-min treatment with GTPγS in relaxing solution with or without ATP or ATPγS. Control samples, frozen in the presence of air and with ATP as the gamma phosphate donor, showed that MYPT1 phosphorylated at both T696 and T853 (pT696- and pT853-MYPT1; Fig. 2A, left) and that MRLC phosphorylated at S19 (pS19-MRLC; Fig. 2B, left). As a negative control, no phosphorylation of either MYPT1 or MRLC was seen for samples frozen in the absence of both ATP and ATPγS. Samples that were frozen in the presence of air and ATPγS as the gamma thiophosphate donor were thiophosphorylated at both sites of MYPT1 and at S19 of MRLC (Fig. 2, A and B, right). In contrast, under hypoxic conditions, thiophosphorylated MYPT1 and MRLC were barely detectible, if at all. The lack of thiophosphorylated proteins was not attributable to a reduction in total protein, since the pan-antibodies (which specifically recognize both phosphorylated/thiophosphorylated and unphosphorylated protein) detected comparable levels of antigen under normoxia and hypoxia.

We have previously shown under these conditions that neither MYPT1 phosphatase nor MRLC phosphatase is effective against these thiophosphorylated substrates (33). Thus the experiment in Fig. 2 does not rule out that MYPT1 phosphatase may be activated by hypoxia since it is not effective against thiophosphorylated proteins. Importantly, however, it shows that, in situ, Rho kinase activity is inhibited under hypoxic conditions.

We further investigated whether the activity of recombinant Rho kinase catalytic subunit was inhibited by hypoxia in vitro. We utilized a commercial assay to determine the relation between the amount of constitutively active Rho kinase catalytic subunit and the phosphorylated T696 of a MYPT1 COOH-terminal peptide under normoxia and hypoxia. As shown in Fig. 3, this relation was linear at low concentrations and saturable around 2 units of the catalytic subunit per 10 ml. Particularly, as can be seen in the more stringent three separate paired assays, hypoxia did not reduce Rho kinase activity (P > 0.50) nor did the mean values of phosphorylation of T696-MYPT1 fragments differ (P > 0.29) at any concentration of the constitutively active Rho kinase catalytic subunit.

Fig. 3.

The effects of hypoxia on the in vitro Rho kinase catalytic subunit phosphorylation of T696-MYPT1 fragments. A commercial assay was used to determine the relation between the amount of constitutively active Rho kinase catalytic subunit and the phosphorylated T696 of a MYPT1 COOH-terminal peptide under normoxia and hypoxia. We conducted 1 independent trial under normoxia only (•) and 3 (normoxia and hypoxia)-paired independent trials (normoxia, black symbols; hypoxia, white paired symbols). Large white symbols and black with bars indicate the mean values for hypoxia and air, respectively. A450/A550, absorbance at 450 nm/absorbance at 550 nm. The measurement of T696 phosphorylated under hypoxia was not consistently reduced relative to the paired measurement of T696 phosphorylated under normoxia, and there were no statistically significant differences (P > 0.29) between means at any level of added Rho kinase.

DISCUSSION

In previous studies we showed that hypoxia could relax both intact (27) and permeabilized porcine coronary arteries (10). In permeabilized coronary, aerating with N2 relaxed contractions elicited by Ca2+ (fixed at 1 μM by EGTA) as well as by GTPγS (1 mM)-induced activation of the small G protein RhoA. In more recent studies (33), we showed that hypoxia in addition to relaxation decreased p-MYPT1 and phosphorylated MRLC (p-MRLC). Importantly, this CDHR was abolished after treatment with ATPγS to thiophosphorylate MYPT1 and MRLC (10, 33). Thiophosphorylation of MYPT1 but not MRLC (pretreatment with ATPγS plus ML7, preserving subsequent reversible Ca2+-dependent phosphorylation of MRLC) blocked hypoxic relaxation of a Ca2+-activated contraction, demonstrating that dephosphorylation of p-MYPT1 was upstream of p-MRLC during CDHR, which ruled out MLCK as a hypoxic effector. Thus MYPT1 is an important hypoxic effector under these conditions. In our current study, hypoxic relaxations in the presence of Rho kinase inhibition (Y-27632) were severely attenuated (Fig. 1A). Conversely, hypoxia elicited significant relaxations when the contractions were primarily attributed to Rho kinase (Fig. 1, B and C). Finally and critically, hypoxia nearly abolished in situ thiophosphorylation of MYPT1 and MRLC (Fig. 2). Since the phosphatase does not dethiophosphorylate MYPT1 (33), we cannot rule out MYPT1 phosphatase(s) as another potentially hypoxic effector. But importantly, these data imply that Rho kinase activity is directly inhibited by hypoxia.

There are a number of paths by which Rho kinase activity could be inhibited by hypoxia. First, hypoxia could affect the activation of RhoA. Maximum activation in permeabilized muscle can occur with as little as 10 μM GTPγS (10). In our study, contractions induced by high GTPγS (1 mM) were relaxed by hypoxia (Fig. 1C). Thus, simply by mass action, dissociation of GTPγS by hypoxia and deactivation of RhoA are unlikely to be involved. Second, hypoxia could affect activated RhoA binding to and/or activation of Rho kinase. The amount of RhoA activated by 1 mM GTPγS would also likely overcome any potential effects of hypoxia on its binding to and activation of Rho kinase. However, any mechanism whereby hypoxia might attenuate the translocation of Rho kinase to a subcellular compartment that favors interaction, for example, the Ca2+/calmodulin kinase II-dependent translocation of RhoA kinase to caveolae (21, 24, 31), cannot be excluded. Third, hypoxia could affect the interactions between Rho kinase and RhoA-independent activation mechanisms, such as those mechanisms operating via arachidonic acid (8) or the sphingosylphosphorylcholine and Src family tyrosine kinases pathway (18, 28). And finally, hypoxia could affect activated Rho kinase binding to and/or activating its substrate(s). Interestingly, the Rho kinase thiophosphorylation of both MYPT1 and MRLC was inhibited by hypoxia (Fig. 2), suggesting that the inhibition was not specific to a substrate.

Most of these pathways include upstream kinases in addition to Rho kinase and MRLC kinase. Downstream, there are also kinases reported to phosphorylate MYPT1 or MRLC, for example, ZIP kinase or integrin-linked kinase (11, 34). We showed that hypoxic relaxation is preserved when permeabilized artery rings are pretreated with ATPγS in the presence of Y-27632 and ML7, respectively, to protect MYPT1 and MRLC from thiophosphorylation (33). During the pretreatment with ATPγS, we expect that kinases not inhibited by ML7 and Y-27632 would have thiophosphorylated their substrate proteins, which, if hypoxic effectors, should have blocked the hypoxic relaxation. We had previously concluded (33), from studies with thiophosphorylated MYPT1, that MRLC kinase is not likely to be a hypoxic effector (33).

At the concentrations used, Y-27632 and ML7 are reasonably selective inhibitors for Rho kinase (7) and MRLC kinase (22), and to our knowledge no other kinases are more sensitive than Rho kinase is to Y-27632 or MRLC kinase is to ML7. Therefore, the overall weight of our evidence strongly suggests that Rho kinase itself is the hypoxic effector.

Hypoxia relaxes not only clearly Rho-dependent, GTPγS-activated contractures (Fig. 1C) but also those stimulated by Ca2+ alone (Fig. 1). Our hypothesis involving Rho kinase in hypoxic vasorelaxation raises the question of whether it is activated in Ca2+ contractures. There is a growing literature suggesting a role for Ca2+ in the activation of Rho kinase. We previously showed that 1 μM Ca2+ induced increases in both pT696-MYPT1 and pT853-MYPT1, in addition to increases in pS19-MRLC and force (33). This is similar to studies with high KCl depolarization in intact vascular and airway smooth muscles (13, 15, 17, 21, 23, 24, 31), also implying Ca2+-dependent activation of Rho kinase. Rho kinase also modulates myogenic contractions of the phasic ureter smooth muscle in the absence of agonist by altering [Ca2+]i as well as by altering the Ca2+ sensitivity of MRLC phosphorylation and force (26).

As a first step toward understanding whether Rho kinase itself may act as an O2 sensor that triggers Ca2+-desensitizing hypoxic relaxation, we showed that the truncated, constitutively active Rho kinase fragment was not inhibited by N2 in any consistent manner (Fig. 3). This suggests that neither Rho kinase binding to MYPT1 nor Rho kinase catalytic activity to phosphorylate T696 of MYPT1 was affected. Although purely speculative, this suggests that the Rho-kinase holoenzyme is required for hypoxic inhibition, which may involve the autoinhibitory region (2, 5, 8).

Our experiments show that Rho kinase activity is inhibited by hypoxia, which underlies, at least in part, Ca2+-desensitizing hypoxic vasorelaxation. Our experiments were performed by aerating with N2, leading to low Po2. Exactly how large a role Ca2+ desensitization plays under physiological conditions remains to be determined. An anoxic core in coronary artery in vivo can be calculated to begin to develop at an Po2 of ∼10% (20) in the physiological range for tissue. Understanding the molecular mechanisms that lead to CDHR may be of major significance beyond coronary artery since Rho kinase is known to be involved in many important cellular functions (4, 25, 29, 32).

GRANTS

This study was supported in part by National Heart, Lung, and Blood Institute Grants HL-66044 (to R. J. Paul) and 5T32-HL-07571 (to R. J. Paul and M. Gu) and a grant from the American Heart Association (to R. J. Paul and R. L. Wardle).

Acknowledgments

We acknowledge Daniel Bullard for technical assistance and Dr. J. Clark for use of the oxygen electrode.

Footnotes

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

REFERENCES

View Abstract