Hypoxic regulation of cardiac Ca²⁺ channel: possible role of haem oxygenase

Abstract

Acute and chronic hypoxias are common cardiac diseases that lead often to arrhythmia and impaired contractility. At the cellular level it is unclear whether the suppression of cardiac Ca²⁺ channels (Ca_V1.2) results directly from oxygen deprivation on the channel protein or is mediated by intermediary proteins affecting the channel. To address this question we measured the early effects of hypoxia (5–60 s, ) on Ca²⁺ current (I_Ca) and tested the involvement of protein kinase A (PKA) phosphorylation, Ca²⁺/calmodulin-mediated signalling and the haem oxygenase (HO) pathway in the hypoxic regulation of Ca_V1.2 in rat and cat ventricular myocytes and HEK-293 cells. Hypoxic suppression of I_Ca and Ca²⁺ transients was significant within 5 s and intensified in the following 50 s, and was reversible. Phosphorylation by cAMP or the phosphatase inhibitor okadaic acid desensitized I_Ca to hypoxia, while PKA inhibition by H-89 restored the sensitivity of I_Ca to hypoxia. This phosphorylation effect was specific to Ca²⁺, but not Ba²⁺ or Na⁺, permeating through the channel. CaMKII inhibitory peptide and Bay K8644 reversed the phosphorylation-induced desensitization to hypoxia. Mutation of CAM/CaMKII-binding motifs of the α_1c subunit of Ca_V1.2 fully desensitized the Ca²⁺ channel to hypoxia. Rapid application of HO inhibitors (zinc protoporphyrin (ZnPP) and tin protoporphyrin (SnPP)) suppressed the channel in a manner similar to acute hypoxia such that: (1) I_Ca and I_Ba were suppressed within 5 s of ZnPP application; (2) PKA activation and CaMKII inhibitors desensitized I_Ca, but not I_Ba, to ZnPP; and (3) hypoxia failed to further suppress I_Ca and I_Ba in ZnPP-treated myocytes. We propose that the binding of HO to the CaM/CaMKII-specific motifs on Ca²⁺ channel may mediate the rapid response of the channel to hypoxia.

Key points

• Lack of oxygen to the heart, hypoxia, triggers a cascade of events that within minutes leads to irreversible tissue damage.

• One of the initial steps at the onset of hypoxia is suppression of the cellular calcium channels that control the strength of the heartbeat.

• The role of oxygen in energy production is well known, but it is not known how lack of oxygen within seconds may suppress the calcium channels.

• We tested the involvement of different regulatory pathways that are known to modulate the calcium channel in response to various stimuli, and succeeded to link it to a haem oxygenase protein that may act as the primary oxygen sensor.

• Our findings provide a better understanding of the initial events that occur during episodes of acute hypoxia induced by coronary thrombosis and bring into focus a new set of therapeutic considerations.

Introduction

O₂ deprivation has significant detrimental effects on the myocardium that include ischaemic infarction, defective Ca²⁺ homeostasis and arrhythmia, leading eventually to cardiomyopathy and heart failure. It is generally thought that prolonged ischaemia leads to influx of Ca²⁺ on the Na⁺/Ca²⁺ exchanger, increasing the Ca²⁺ load of the sarcoplasmic reticulum (SR), which in turn is mobilized during reperfusion injury causing cell death (Piper et al. 2004). Conversely, the L-type Ca²⁺ current (I_Ca) is suppressed by cardiac hypoxia (Fearon et al. 1997) possibly reducing I_Ca-gated Ca²⁺ release, force development and phosphorylation by CaM-kinase II (Vittone et al. 2002). Thus, it may be critical to determine whether the suppression of Ca²⁺ channel by hypoxia serves as a regulatory mechanism that protects against Ca²⁺ overload during ischemia–reperfusion insult.

Previous studies of hypoxic modulation of ionic channels, on longer time scales and in different tissues, have yielded no consensus on the underlying mechanisms of oxygen sensing, the intermediate signalling pathways involved, and their convergence on the channel proteins. Rapid suppression of the L-type Ca²⁺ channel has been observed in smooth muscle (Franco-Obregon et al. 1995) and single glomus cells (Montoro et al. 1996) with either Ca²⁺ (I_Ca) or Ba²⁺ (I_Ba) as charge carrier, in the HEK-293 cell lines, where cytoplasmic residues 1786–1821 of the recombinant human cardiac α_1C subunit were found to be involved (Fearon et al. 1997, 2000), and in rat ventricular myocytes (Movafagh & Morad, 2010). Some studies have tested the possibility that the hypoxic regulation of the Ca²⁺ channel may involve oxidation of intra- and extracellular thiol residues by reactive oxygen species (ROS) (Fearon et al. 1999; Hool, 2000; Hool & Corry, 2007). However, this idea has encountered some skepticism because (1) ROS production during hypoxia is either moderate or even decreased (Hool et al. 2005; Korge et al. 2008), (2) accumulation of mitochondrial superoxides is only moderate in the first 40–50 min of hypoxia, long after suppression of I_Ca (Hool, 2000), and (3) the evidence in support of involvement of extracellular thiol groups comes primarily from application of exogenous thiol-oxidizing or -reducing agents.

Here we have examined the possibility that the rapid hypoxic suppression of I_Ca might involve a more direct O₂ sensing mechanism mediated e.g. by haem oxygenase (HO) or its reaction products (CO, Fe²⁺ and biliverdin). Mechanisms of this type have been associated with Ca²⁺-activated K⁺ channels in carotid bodies (Williams et al. 2004; Ortega-Saenz et al. 2007), anoxic activation of ganglionic nociceptive neurons (D’Agostino et al. 2009), and CO-mediated redox modulation of L-type Ca²⁺ channels (Scragg et al. 2008). The intriguing finding that calmodulin (CaM) binds to both HO-2, and the Ca²⁺ sensitive domain of the carboxyl tail of the α_1C subunit of the Ca_V1.2 (Boehning et al. 2004), where Ca²⁺-dependent inactivation of the channel takes place (Zuhlke & Reuter, 1998; Erickson et al. 2001), provided further impetus to probe this possibility.

To test whether HO might serve as a rapid O₂ sensor for the Ca²⁺ channel, we measured the response of the Ca²⁺ channel current only in the first 5–60 s of withdrawal of O₂ () using experimental conditions with varying degrees of Ca²⁺-dependent inactivation and altered signalling by CaM and HO. We found that the suppression of the L-type Ca²⁺ channel in response to acute O₂ loss has a rapid onset (>5 s) both in cardiomyocytes and in HEK-293 cells expressing the recombinant channel. The hypoxic response depended critically on Ca²⁺ influx through the channel, and on its suspected phosphorylation state, but not on release of Ca²⁺ from the SR. Using different isoforms of the recombinant Ca²⁺ channel, an 80 amino acid domain within the Ca²⁺/CaM/CaMKII binding motif of the C-terminus was identified as mediating the hypoxic response. Interestingly, the suppressive effects of withdrawal of O₂ and HO-oxygenase inhibitors on the Ca²⁺ channel had similar profiles with respect to magnitude, kinetics, phosphorylation and ionic specificity, and were non-additive. Our data are consistent with the idea that HO may confer O₂ sensitivity on the cardiac Ca²⁺ channel.

Methods

Isolation of rat and cat cardiomyocytes

Protocols for experiments with rats (AR no. 2791) and cats (ACORP no. 454) were approved and supervised by the Institutional Animal Care and Use Committees (IACUC) of Georgetown University, the Medical University of South Carolina (A 3428–01), University of South Carolina (A 3049-01) and the Department of Veterans Affairs according to national and international guidelines.

Freshly dissociated rat cardiomyocytes were prepared by enzymatic digestion as previously described (Knollmann et al. 2000) using one male Wistar rat (200–300 g) per experimental day. Briefly, each rat was anaesthetized with sodium pentobarbital (150 mg kg⁻¹ i.p.) and the heart was rapid excised from the chest cavity and subjected to retrograde Langendorff perfusion using Tyrode solutions with different concentrations of Ca²⁺ and digestive enzymes (1.4 mg ml⁻¹ collagenase and 0.16 mg ml⁻¹ protease). The dispersed cells were equilibrated in Tyrode solution with 0.2 mm CaCl₂ and were allowed 30 min to settle and adhere to coverslips before they were used in electrophysiological experiments within the next 4–8 h.

The preparation of cat ventricular myocytes was similar (Mann et al. 1991), but was performed under sterile conditions yielding cells that were kept in culture medium for up to 4 days before use. As with rats, the method of killing was exsanguinations following excision of the heart under deep anaesthesia (50 mg kg⁻¹ ketamine hydrochloride and 5 mg kg⁻¹ acepromazine maleate i.m.).

Expression of recombinant human Ca²⁺ channels in HEK-293 cells

was accomplished with the standard lipofectamine method (Lipofectamine 2000, Invitrogen) using cDNAs of all subunits (α_1C77-YFP, β₁, α₂δ) with yellow fluorescent protein (YFP) for positive fluorescence identification (Chroma Technology, Bellows Falls, VT, USA; 470 nm excitation/525 nm emission). HEK-293 cells were cultured to 90% confluence in six-well plates using Dulbecco's modified Eagle's medium (DMEM) with 10% fetal bovine serum (FBS). The lipofectamine reagent was mixed with 1–6 μg of each cDNA, incubated for 20 min and dispensed into each well while swirling the culture plates. Cells were then incubated for 24 h allowing for gene expression to take place. The next day, cells were split onto coverslips at 10% confluence and allowed to incubate for another 48 h before being screened based on green fluorescent protein (GFP) fluorescence (60–70%) and used in voltage-clamp experiments similar to those conducted with cardiomyocytes. Mutant channels were expressed in HEK cells by replacing the α_1c77 subunit with the α_1c86 subunit. All cDNAs were generously provided by Dr Nikolai Soldatov (National Institute on Aging, NIH, Baltimore, MD, USA).

Electrophysiological experiments at different O₂ pressures

Rat and cat cardiomyocytes and HEK-293 cells expressing recombinant Ca²⁺ channels were voltage-clamped in the whole-cell configuration on the stage of an inverted Nikon microscope using a Dagan amplifier and pCLAMP software (Axon Instruments, Union City, CA, USA). A rapid perfusion system was used to change between solutions equilibrated with different gasses or with different ionic composition. Epi-fluorescence detection was used for ratiometric measurements of intracellular Ca²⁺ in cardiomyocytes (Fura-2) and evaluation of protein expression in HEK-293 cells (YFP).

To measure Ca²⁺ current (I_Ca), patch pipettes with 2–5 MΩ resistance were filled with an internal solution containing (in mm): 120 Cs⁺, 120 Asp⁻, 20 TEACl, 2 EGTA, 20 Hepes and 5 MgATP, at pH 7.2 using CsOH. A series resistance compensation of ∼80% was applied and monitored throughout recordings. The liquid junction potential was corrected before seal formation. In many experiments, 0.2 mm cyclic adenosine monophosphate (cAMP) was added to the patch pipette solutions to prevent rundown of I_Ca and fully activate the SR Ca²⁺-ATPase via PKA-mediated phosphorylation. The extracellular Tyrode solution contained (in mm): 140 NaCl, 4.7 mm KCl or CsCl, 2–5 CaCl₂, 1 MgCl₂, 10 glucose and 10 Hepes (pH 7.4). For measurement of Ba²⁺ current (I_Ba), CaCl₂ was replaced with 2–10 mm of BaCl₂. To allow permeation of Na⁺ through the Ca²⁺ channels (I_Na), 2 mm EGTA was used to chelate traces of Ca²⁺ and Mg²⁺. To inhibit K⁺ currents, membrane currents were recorded with an extracellular solution where K⁺ was replaced by Cs⁺. A holding potential of −50 mV was chosen to inactivate Na⁺ channels.

The current through the Ca²⁺ channel was generally measured using 25–50 ms depolarizing pulses from −50 to +10 mV applied every 5 or 10 s. Pulses of 200 ms duration were used for simultaneous measurements of I_Ca and intracellular Ca²⁺ (Fig. 2). Current–voltage (I–V) relations were obtained with depolarizing pulses ranging from −50 to +80 mV in increments of 10 mV. To generate I–V relations that were free of secondary effects of prolonged hypoxia and rundown, we limited exposure to hypoxic solutions to 2 min. The measured currents were filtered at 1 or 10 KHz, digitized at 10 or 100 kHz, and plotted and analysed in terms of magnitude and time constants of decay using Origin 6.0 (OriginLab Corp., Northampton, MA, USA) and pCLAMP 9.0 software.

Figure 2. The suppression of I_Ca by hypoxia is accompanied by a similar suppression of the I_Ca-triggered intracellular Ca²⁺ transients.

A, simultaneous measurements of I_Ca and Ca²⁺ transients measured with 0.4 mm K₄Fura-2 in the patch pipette in non-cAMP dialysed rat cardiomyocytes before (O₂) and after 40 s of hypoxia (N₂). B, time course of suppression I_Ca and Ca²⁺ transients during exposure to hypoxic solution.

To study the effects of anoxia on ionic currents, the voltage-clamped single cells were superfused with external solutions equilibrated with atmospheric O₂ (normoxic) or 100% N₂ (hypoxic). Solution exchange took place within 50 ms using an electronically controlled perfusion system equipped with five barrels loaded with control and hypoxic and Tyrode solutions containing varying electrolyte concentrations and pharmacological agents (Cleemann & Morad, 1991). The O₂ pressure was measured with a needle probe which registered >5 mmHg for the hypoxic solutions both in the bubbled reservoirs and near the port for solution outflow into the main chamber. Due to size of the O₂ electrode (1.2 mm diameter) it measured the O₂ pressure near the much smaller outflow tube (280 μm diameter) with a delay of 1–2 s. However, considering that the voltage clamped cell (∼100 μm in length) was located near the outflow port (100–150 μm) and entirely within its rapidly flowing stream (∼7 cm s⁻¹, Belmonte & Morad, 2008), it is plausible that changes in O₂ pressure at the surface of the cell occur on the same time scale (<50 ms) as measured when the Ca²⁺ channels were subjected to ionic interventions. In another set of experiments we verified rapid changes in local by using cells that were plated onto cover slips with an oxygen-sensitive fluorescent coating (Supplemental Fig. 1).

Simultaneous measurement of Ca²⁺ transients and I_Ca were performed with a dialysing solution containing 0.5 mm of the cell impermeant fluorescent ratiometric Ca²⁺ indicator K₄Fura-2 (Invitrogen) together with 0.21 mm Ca²⁺, but no EGTA, yielding an intracellular concentration of 100 nm of free Ca²⁺. As previously described (Cleemann & Morad, 1991) a mirror vibrating at 1150 Hz was used to alternate between excitation wavelengths centred at 340 and 405 nm. The plotted Ca²⁺ signal (Fig. 2) is the ratio (F₃₄₀/F₄₀₅) of the corresponding fluorescence intensities measured at wavelengths above 515 nm.

When dialysing compounds of relatively high molecular weight into voltage-clamped cells, we used an equilibration period of 5–10 min between break-in and the first recordings of membrane current. Such compounds include K₄Fura-2, the PKA inhibitor PH-89, CaM-binding domain peptide_290−309, and the HO inhibitors zinc protoporphyrin (ZnPP) or tin protoporphyrin (SnPP). All experiments were done at room temperature (23–25°C), except those with cat cardiomyocytes that were conducted at 30–32°C.

Reagents

The CAMKII inhibitor CK59 was dissolved in dimethyl sulphoxide (DMSO) and was applied extracellularly at a concentration of 10 μm (Sigma-Aldrich, St Louise, MO, USA). CaM-binding domain peptide_290−309 (Calbiochem Inc.) and the PKA inhibitor H-89 were internally dialysed using concentrations of 0.1 mm and 10 μm, respectively in the dialysing solution. The Ca²⁺ channels were subjected to phosphorylating conditions using 0.2 mm cAMP or 60 nm okadaic acid (OA, Sigma-Aldrich). (S)-(–)-Bay K8644 (1,4-dihydro-2,6-dimethyl-3-nitro-4-(2-trifluoromethylphenyl)-pyridine-5-carboxylic acid methyl ester) was from Sigma-Aldrich. Zinc protoporphyrin IX (ZnPP) and tin protoporphyrin IX (SnPP, Frontier Scientific Inc., Logan, UT, USA) were dissolved in DMSO and applied to external solutions in a maximal concentration of 0.005% of DMSO or dialysed into the cell for 10 min before recording of membrane current. K₄Fura-2 was purchased from Invitrogen. Protease enzyme, collagenase A, was obtained from Roche Applied Sciences (Indianapolis, IN, USA).

Statistics

Student's t test was used to compare means, which are presented plus/minus their standard errors (SEM), with n indicating the number of experiments. Significance levels are indicated with an increasing number of asterisks (*P < 0.05, **P < 0.01, ***P < 0.001) as opposed to being not significant (NS, P > 0.05).

Results

Current through the cardiac Ca²⁺ channel is rapidly suppressed by acute hypoxia

Direct effects of acute O₂ withdrawal on the Ca²⁺ channel were studied in different preparations using conditions that allowed different ions (Ca²⁺, Ba²⁺ or Na⁺) to permeate through the channel in the absence and presence of PKA activation. To achieve rapid changes in extracellular , the stream of solution surrounding the voltage-clamped cell was rapidly switched from one saturated with 100% O₂ to another with 100% N₂ () and peak I_Ca was measured during repeated depolarizations from −50 to +10 mV. In freshly dissociated rat ventricular cells at room temperature, the hypoxic solution produced ∼10% suppression of I_Ca within the first 5 s followed by a gradually increasing suppression that levelled off at 23 ± 3% (n = 15) in 45–60 s (Fig. 1A open circles, summarized in the bar graph, Fig. 1C). Current–voltage relations measured before (filled circles) and 1–2 min after (open circles) the onset of hypoxia showed that the suppression of I_Ca was equally pronounced at all potentials with no detectable change in either its activation range (Fig. 1B) or the rate of its inactivation (Inset panels). Similar suppressions of I_Ca (28 ± 3%, n = 9) were also measured in HEK-293 cells expressing all the recombinant subunits of the Ca²⁺ channel (filled circles in Fig. 1A and inset tracings), and, once again, with no significant change in the inactivation kinetics (inset panel). Short-term cultured feline cardiomyocytes (examined at 30–32°C) were similarly suppressed by O₂ withdrawals (24 ± 6%, n = 3).

Figure 1. Suppression of current through the L-type Ca²⁺ channel during brief periods of hypoxia.

A, time course of hypoxia-induced changes in the peak inward current in rat cardiomyocytes where the permeating ion was Ca²⁺ (I_Ca), Ba²⁺ (I_Ba) or Na⁺ (I_Na) and in HEK cells where the recombinant channel conducted Ca²⁺ (I_Ca). B, current–voltage relations for I_Ca in rat ventricular cardiomyocytes measured under normoxic control conditions (O₂) and after 1–2 min exposure to hypoxia (N₂). C, average suppression of peak inward current measured with different permeating ions (Ca²⁺ vs. Ba²⁺ vs. Na⁺) and preparations (rat vs. cat cardiomyocytes vs. HEK-cell expression system) after 45–60 s superfusion with solutions equilibrated with N₂, He or O₂. The inset panels show that currents activated by depolarization from −60 to +10 mV displayed different rates of inactivation, but similar degrees of suppression during hypoxia (I_Ca vs. I_Ba in rat cardiomyocytes vs. I_Ca in HEK cells). All illustrated experiments were conducted in the absence of cAMP or other interventions that might evoke cAMP-dependent phosphorylation. The experiments with cat cardiomyocytes were performed at 30–32°C, the rest at room temperature, 23–25°C.

To minimize the effects of Ca²⁺ channel run-down, only the effects of the first 45–60 s of withdrawal of O₂ were quantified and statistically analysed. On this time scale, I_Ca in rat cardiomyocytes showed a small degree of run-down (4 ± 3%, n = 4, Fig. 1C) in fully oxygenated solutions. This was much smaller than the effect of hypoxia, but did preclude full recovery after re-oxygenation (not shown). To exclude possible direct effects of N₂ on I_Ca, He was substituted for N₂ resulting in similar suppressive effects (24 ± 4%, n = 9).

Substitution of Ba²⁺ for Ca²⁺, as the charge carrier through the calcium channel, in cardiomyocytes resulted in suppressive effects of hypoxia that were somewhat larger (31 ± 5%, n = 11) and faster (Fig. 1A, Δ; see also Supplemental Fig. 1) compared to those with Ca²⁺ as charge carrier. The characteristic slower inactivation kinetics of I_Ba, which is known not to trigger SR Ca²⁺ release (Nabauer et al. 1989), remained also insensitive to withdrawal of O₂ (inset panel). Readmission of O₂ resulted in rapid and nearly complete recovery of I_Ba (Fig. 1A, open triangles) although some run-down appeared to be present (dotted line). Interestingly, the suppressive effect of hypoxia on the channel was significantly enhanced (52 ± 2%, n = 3) in the first minute of O₂ withdrawal when Na⁺ was the permeating cation through the Ca²⁺ channel (in ‘zero’ Ca²⁺ solution with 2 mm EGTA, Fig. 1A, filled triangles).

The similarity in the suppressive effects of hypoxia on the Na⁺, Ba²⁺ and Ca²⁺ currents through the channel suggests that the suppressive effect of hypoxia must be independent of the Ca²⁺-triggering property of I_Ca on the SR. To test the role of the SR Ca²⁺ stores more directly, Fura-2 dialysed myocytes were used to measure I_Ca-triggered Ca²⁺-transients. Figure 2 shows that the hypoxia-induced suppression of I_Ca was accompanied by parallel decrease in Ca²⁺ transients, without any changes in either baseline [Ca²⁺]_i or the release kinetics, suggesting that the hypoxic suppression of I_Ca is likely to result in a weakening of the heartbeat, thus reducing the requirements for metabolic energy.

The results thus far suggest that hypoxia in the absence of PKA stimulation uniformly and rapidly suppresses the Ca²⁺ channels both in freshly isolated rat and cat cardiomyocytes and in HEK-293 cells expressing the human Ca_V1.2, irrespective of the O₂ substituents (N₂ or He) or permeating cations (Ca²⁺, Ba²⁺ or Na⁺).

PKA phosphorylation desensitizes the Ca²⁺-transporting channel to hypoxia

Ca²⁺ channels are phosphorylated by the protein kinase A (PKA)/cAMP cascade causing altered gating kinetics and enhanced current. In rat ventricular myocytes dialysed with 0.2 mm cAMP, peak I_Ca at +10 mV increased from 2–6 pA pF⁻¹ to 8–15 pA pF⁻¹. Figure 3A shows that phosphorylation induced by cAMP significantly reduced the suppressive effects of hypoxia on I_Ca from 23 ± 3% to 6.8 ± 2% (Fig. 3A and C; P = 0.005). Similarly, inclusion of okadaic acid (an inhibitor of protein phosphatases PP1 and PP2a) in the patch pipette also reduced the response of I_Ca to hypoxia to 6.7 ± 2% (Fig. 3C; OA, P < 0.01).

Figure 3. Phosphorylation of the Ca²⁺ channel makes it insensitive to hypoxia when the permeating ion is Ca²⁺ (I_Ca), but not when it is Ba²⁺ (I_Ba) or Na⁺ (I_Na).

A, time course of I_Ca and I_Ba during 40 s hypoxia (N₂) in rat ventricular myocytes dialysed with 0.2 mm cAMP. B, the I–V relation is insensitive to hypoxia even when SR Ca²⁺ stores are depleted by pre-incubation with 1 μm thapsigargin (Thap). The inset panel shows sample currents recorded with depolarization to +10 mV. C, bar graph showing the reduction in the currents through the Ca²⁺ channel during 45–60 s of hypoxia in cells dialysed with 0.2 mm cAMP or incubated with 1 μm of the phosphatase inhibitor okadaic acid (OA). Results obtained with the PKA inhibitor H-89 and thapsigargin (Thap) are included. Results obtained under non-phosphorylating conditions are reproduced for comparison from Fig. 1C as hatched bars. All illustrated results were obtained using rat ventricular cells and hypoxic solutions equilibrated with 100% N₂.

To probe whether the insensitivity of the Ca²⁺ channel to hypoxia was related to PKA-dependent phosphorylation rather than to other cAMP-mediated pathways, a subset of cells that were dialysed with cAMP were then exposed to10 μm H-89, a PKA inhibitor. Figure 3C shows that cells dialysed with cAMP, but exposed to H-89, recovered their sensitivity to hypoxia (26 ± 5% suppression, n = 4, P = 0.002, vs. 6.7%). Thus, there was no significant difference between cells dialysed with cAMP-free and those dialysed with cAMP and exposed to H-89 (Fig. 3C). It appears therefore that PKA-mediated phosphorylation protects the Ca²⁺ channels against acute hypoxia.

Phosphorylation by PKA is known to affect the gating of the Ca²⁺ channel not only directly, but also indirectly through enhanced Ca²⁺ release caused by phosphorylation of the ryanodine receptors and phospholamban of the SR (Reuter & Scholz, 1977; Yamaoka & Kameyama, 2003). To examine whether the observed PKA effects were related, in part, to modulation of SR Ca²⁺ release, we incubated cAMP dialysed cells with 1 μm thapsigargin, which is known to deplete the SR Ca²⁺ content by inhibiting the SR Ca²⁺-ATPase (Treiman et al. 1998). Figure 3B shows that, even though thapsigargin greatly slowed the inactivation kinetics of I_Ca (changing τ from 5.4 ± 0.5 ms to 15.3 ± 0.8 ms; cf. insets in Figs 1 and 3), it did not restore the O₂ sensitivity of phosphorylated Ca²⁺ transporting channels (n = 7). These data imply that the phosphorylation-induced protection against hypoxia is independent of release of Ca²⁺ from the SR.

Since Ba²⁺ transporting Ca²⁺ channels fail to trigger Ca²⁺ release (Nabauer et al. 1989), but are suppressed by acute hypoxia (Fig. 1C), we examined whether phosphorylation could also protect the Ba²⁺ transporting channels against acute hypoxia. Surprisingly, phosphorylation failed to desensitize the Ba²⁺-transporting channels to hypoxia. Statistical analysis revealed significant differences in the response of phosphorylated channels to O₂ withdrawal when the permeating ion was Ba²⁺ or Na⁺ instead of Ca²⁺ (Fig. 3C) suggesting that phosphorylation-induced protection against hypoxia required entry of Ca²⁺ through the channel.

In ventricular cells dialysed with okadaic acid, to achieve channel phosphorylation without the use of cAMP, the desensitizing effects of phosphorylation continued to be specific to permeation of Ca²⁺ (Fig. 3A). Thus the data obtained with different charge carriers (Ca²⁺ vs. Ba²⁺ vs. Na⁺) and different means of channel phosphorylation (cAMP vs. inhibition of phosphatases vs. PKA-inhibitors), suggest that the degree of hypoxic suppression of the Ca²⁺ channel is somewhat dependent on the permeating cationic species (Na⁺ > Ba²⁺≍ Ca²⁺), while the PKA-induced desensitization against hypoxia required strictly the permeation of Ca²⁺, but, surprisingly, Ca²⁺ released from the SR did not substitute for the permeating Ca²⁺.

Role of calmodulin (CaM) and CaM kinase II (CAMKII) in the O₂ sensing of the Ca²⁺ channel

cAMP is thought to increase the endogenous activity of CaMKII, by a sequence of intermediary steps that include PKA-mediated phosphorylation of inhibitor 1 and protein phosphatase PP1 leading to Ca²⁺- and CaM-dependent auto-phosphorylation at T₂₈₆ (Fig. 4A) (Bradshaw et al. 2003; Huke & Bers, 2007). The activity of CAMKII seems to depend on higher local rather than global cytosolic concentrations of Ca²⁺ during excitation–contraction coupling (Song et al. 2008). This mechanism may be, in part, responsible for the PKA-induced desensitization of the Ca²⁺-, but not Ba²⁺- or Na⁺-transporting channel to acute hypoxia. To test this possibility we first used inhibitory peptide_290−309 (PEP), which comprises the CaM-binding domain of CaMKII and by competitive binding of available CaM may block the activation of CaMKII (Payne et al. 1988) and other CaM-dependent processes tested in subsequent experiments. Following 8 min of dialysis with an internal solution containing inhibitory peptide and 0.2 mm cAMP, the voltage-clamped cardiomyocytes were exposed to hypoxic solution. Figure 4 shows that the presence of the peptide in the cytosol (PEP) restored the sensitivity of the cAMP-stimulated channels to the suppressive effects of hypoxia. The suppression of I_Ca was gradual (panel B, filled circles) and reached 26 ± 5% after 60–90 s of hypoxia (panel C). For comparison the dashed curve in Fig. 4B show the time course of the much smaller effect of hypoxia on I_Ca in cAMP-dialysed myocytes in the absence of the inhibitory peptide (6.8 ± 2%, cf. Fig. 3A).

Figure 4. Inhibition of Ca²⁺-dependent inactivation of I_Ca by calmodulin domain peptide_290−309 and Bay K is associated with enhanced suppression of I_Ca by anoxia.

A, putative CaM-binding (IQ and hydrophobic 1–5–10 motifs) and phosphorylation sites (grey highlight) in the partial AA sequences of CaMKII, peptide_290−309 (PEP), wild-type and mutant α_1c subunits of the Ca²⁺ channel (α_1c77 and α_1c86) and haem oxygenase (HO-2) (Colbran, 1993; Rhoads & Friedberg, 1997; Soldatov et al. 1997; Boehning et al. 2004). B, time course of suppression of I_Ca upon removal of extracellular O₂ in cAMP-dialysed rat cardiomyocytes that in addition were dialysed with 100 μm PEP, treated with 1 μm Bay K8644 (BayK) or both. For comparison purposes, I_Ca in cardiac control cells dialysed with 0.2 mm of cAMP is reproduced from Fig. 3A (dashed line). C, comparison of the anoxic suppressions of I_Ca in cardiomyocytes measured after 60–90 s of hypoxia (during recording of I–V relations). A, amino acid sequences of α_1c77 and α_1c86 subunits.

We also examined whether pharmacological inhibition of both Ca²⁺-dependent inactivation (using Ca²⁺ channel agonist (S)-(–)Bay K8644; Adachi-Akahane et al. 1999) and Ca²⁺-dependent facilitation (using CaMKII inhibiting peptide_290−309) would further enhance the suppressive effects of hypoxia on I_Ca. Figure 4B shows, unexpectedly, that Ca²⁺ channels that had been rendered insensitive to hypoxia by cAMP, once again became sensitive to hypoxia when the myocytes were exposed to the I_Ca enhancing effects of Bay K (open circles). Experimentally this was then followed by measurements of the voltage dependence of current, which confirmed that the insensitivity of I_Ca to hypoxia in cAMP-dialysed myocytes was reversed by Bay K causing 29 ± 8% suppression at all voltages tested, somewhat similar to the effect of the CaM inhibitory peptide_290−309. The suppression in the presence of Bay K or the inhibitory peptide was comparable (separately causing 29 ± 8%vs. 26 ± 5% inhibition), roughly summing to about 49 ± 6% when Bay K and inhibitory peptide were used in combination. Our data therefore suggest that pharmacological inhibition of Ca²⁺-dependent inactivation and facilitation can reverse the PKA-induced protection of the Ca²⁺ channel against hypoxia, even when Ca²⁺ is the charge carrier, and implicates a critical role for calmodulin in the O₂ sensing pathway.

In probing the role of CaM-dependent signalling in O₂ sensing we next targeted the CaM-binding motifs that is located on the carboxyl-terminal tail of the Ca_V1.2 channel (Fig. 4A) (Soldatov et al. 1998). In this set of experiments the native α_1c77 channel was replaced by its isoform α_1c86 channel with alternative splicing at exons 40–41, involving residues 1572–1651 that include both pre-IQ and IQ domains of the CaM binding motif (Soldatov et al. 1997) (Fig. 4A). This mutant channel lacks Ca²⁺-dependent inactivation, but shows similar fast voltage-dependent inactivation kinetics for both I_Ca and I_Ba (Soldatov et al. 1997; Erickson et al. 2003). Figure 5 confirms the rapid voltage-dependent inactivation of Ba²⁺ currents in α_1c86 (Fig. 5B) and shows, in addition, that the mutant channel is completely insensitive to acute hypoxia (Fig. 5C). While peak Ba²⁺ current in α_1c77 channels was markedly suppressed by hypoxia by an average of 52% (Fig. 5D; n = 8, P < 0.001 at 10 mV), both I_Ba, and I_Ca in the mutant channel were equally resistant to hypoxic suppression (Fig. 5D, n = 7). There were also no significant changes in the I–V relations of α_1c86 channel in response to hypoxia (n = 4, P = 0.39, data not shown). Together these findings support the idea that the CaM association site of the Ca²⁺ channel C-terminus is critically important in hypoxic regulation of I_Ca.

Figure 5. The cytoplasmic CaM/CaMKII binding sites of the α_1C subunit of the Ca²⁺ channel is required for its hypoxia-induced suppression.

The measured Ca²⁺ and Ba²⁺ currents (I_Ca and I_Ba) show that Ca²⁺-dependent inactivation (A) and anoxic suppression (C) are present in HEK-293 cell expressing the α_1c77 of the L-type Ca²⁺ channel, but not in cells expressing the mutant α_1c86 subunit lacking CaM/CAMKII binding sites (B and D; cf. Fig. 4B). The currents in panels A and B were normalized to emphasize differences and similarities in kinetics. Extracellular Ba²⁺ at 20 mm was used to enhance the currents shown in panels C and D. All measurements were done under non-phosphorylating conditions.

Role of haem oxygenase in the hypoxic regulation of the cardiac Ca²⁺ channel

Since haem oxygenase binds oxygen and has been implicated in the O₂-dependent regulation of Ca²⁺-activated K⁺ channels (Williams et al. 2004; Ortega-Saenz et al. 2007), and has been reported to form a Ca²⁺-sensitive complex with CaM (Fig. 4A; Boehning et al. 2004), we considered the possibility that this protein might be involved in the O₂-dependent regulation of the cardiac Ca²⁺ channel by, for instance, interacting with a CaM/CaMKII binding site near the inner mouth of the channel. Figure 6A shows that rapid application of 10–100 nm of the HO inhibitor ZnPP reversibly suppressed I_Ca without changing its kinetics. To exclude the possibility that contaminating free Zn²⁺ in ZnPP solutions may cause the Ca²⁺ channel suppression, 10 nm ZnCl₂ (an upper estimate of possible free Zn²⁺ released from 100 nm ZnPP in solution) was added before the application of the 100 nm ZnPP-containing solutions. Figure 6B shows that 10 nm of Zn²⁺ hardly affected I_Ca, while ZnPP again blocked I_Ca rapidly and reversibly. Application of ZnPP for 1–2 min blocked I_Ba equally at all voltages tested (Fig. 6C) with ∼64.3% suppression at 0 mV (n = 3). Similarly, another HO inhibitor, SnPP, caused 48 ± 20% suppression of I_Ba at 100 nm concentrations (n = 4, data not shown). In cells dialysed for 9 min with ZnPP through the patch pipette, I_Ba was consistently smaller than in control experiments with cells that were equilibrated for the same duration with control internal solutions. In this case, peak I_Ca was suppressed by 69 ± 10% at 0 mV and by similar amounts at other potentials (Fig. 6D, n = 4). These findings are consistent with the idea that HO modulates L-type Ca²⁺ channels at an intracellular binding site.

Figure 6. Suppression the Ca²⁺ channel by inhibition of haem oxygenase.

A, brief exposures to the haem oxygenase inhibitor ZnPP in concentrations of 10 and 100 nm cause rapid, reversible, dose-dependent suppression of I_Ca. B, effect of 10 nm ZnCl₂ or 100 nm ZnPP on I_Ca. Insets in panels A and B show traces of I_Ca measured at the indicated times (a, b and c). C and D, normalized I–V relations of I_Ba measured in control cells (Control, open circles) and in cells exposed to 100 nm ZnPP (C,filled triangles) that was applied either in the external solution (C) or via the patch pipette by 9 min of dialysis (D). All experiments were performed on rat ventricular cardiomyocytes under normoxic and non-phosphorylating conditions.

Does PKA phosphorylation desensitize Ca²⁺- but not Ba²⁺-transporting channels to HO-inhibitors?

We compared the effects of HO inhibitors to those of hypoxia in rat ventricular myocytes with and without dialysis of either 0.2 mm cAMP or 60 nm okadaic acid (OA). Under these conditions the suppression of I_Ca produced by 50 s exposures to 100 nm ZnPP was reduced by the phosphorylating interventions from 21.7 ± 4.3% (n = 5) in control cells to 6.2 ± 2.9% (n = 6) in cells dialysed with cAMP and to 7.4 ± 7.5% (n = 4) in cells with OA (Fig. 7A and E). In comparison I_Ba continued to be suppressed by ZnPP when the channel was phosphorylated (Fig. 7B and E). The suppression of I_Ba by ZnPP was somewhat enhanced in OA-dialysed cells (69.1 ± 13.2%, n = 4) compared to control 43.7 ± 4.7%, n = 4). Figure 7C shows that the CAMKII inhibitor CK59 (10 μm), which strongly reduced I_Ca in cells dialysed with 0.2 mm cAMP, also restored the suppressive effect of ZnPP (double arrow). These effects of phosphorylation on the ZnPP-induced suppression of I_Ca and I_Ba are summarized Fig. 7E in a format that facilitates comparison with the hypoxia-induced effects (cf. Fig. 3C). The similarity in the responses suggest that HO inhibitors and the effects of hypoxia on L-type Ca²⁺ channels may be mediated by the same channel entity. Consistent with this idea, we found that acute hypoxia failed to further suppress the Ca²⁺ channel in the presence of 1 μm ZnPP (Fig. 7D). Similarly there was no additive effect when ZnPP was applied after the current through the channel had been suppressed by hypoxia (Supplemental Fig. 2).

Figure 7. Inhibition of the Ca²⁺ channel in rat cardiomyocytes by haem oxygenase inhibitors with respect to phosphorylation, permeating ion and hypoxia.

A and B, effect of 100 nm ZnPP on I_Ca (A) and I_Ba (B) measured in the absence of phosphorylating interventions (no cAMP, filled circles) and with dialysis of either 0.2 mm cAMP (cAMP, open circles) or 60 nm okadaic acid (OA, open triangles). C, pretreatment with an inhibitor of CaMKII (10 μm CK59) restored the suppressive effect (double arrow) of 100 nm ZnPP on I_Ca in cardiomyocytes dialysed with 0.2 mm cAMP. Notice that in this, as in all our experiments, the suppression of I_Ca was quantified based on a normalization that yielded 100% immediately before the exposure to hypoxia or ZnPP. D, hypoxia (N₂) had no additional suppressive effects after I_Ba had been suppressed by 1 μm ZnPP. E, summary of the quantitative effects of 100 nm ZnPP on I_Ca and I_Ba, as measured in panels A–C. This bar graph has the same layout as used for hypoxic effects in Fig. 3C, but shows data obtained with a different protein kinase inhibitor (CK59 vs. H-89).

Discussion

Several studies have attempted to identify the mechanisms by which L-type Ca²⁺ channels are regulated by hypoxia including redox-modification of thiol containing residues and involvement of residues 1786–1821 of the C-terminus of the channel (Fearon et al. 1997; Hool, 2000; Yatani & Kamp, 2000). Here we propose that the CaM/CaMKII binding domain of the L-type Ca²⁺ channel is a critical site in rapid O₂ sensing. Furthermore, our experiments suggest that HO may mediate the rapid sensing of O₂ loss. Consistent with this idea, we found that the pharmacology of the fast suppressive effect of hypoxia on cardiac L-type Ca²⁺ channels to be similar to that of HO inhibitors (cf. Figs 3 and 7). Our finding of rapid suppression of I_Ca within the first 5 s of O₂ withdrawal (Figs 1A and 2B), long before accumulation of metabolic products and O₂ radicals, suggests a direct role for the Ca²⁺-channel protein as a likely first step in sensing of O₂. In native channels and without PKA stimulation this suppressive effect of hypoxia was independent of permeating cations (Na⁺, Ba²⁺, Ca²⁺), but became specific to permeation of Ba²⁺ and Na⁺ in PKA-phosphorylated myocytes (Fig. 1C vs. 3C). Interestingly, the suppressive effect of withdrawal of O₂ on the Ca²⁺ transporting channel was almost completely abolished by phosphorylating interventions, but surprisingly the requirement for permeating Ca²⁺ could not be substituted by release of Ca²⁺ from the SR (Fig. 3B). The dependence on permeating Ca²⁺ held under different conditions in native and recombinant Ca²⁺ channels, except when myocytes were exposed to the Ca²⁺ agonist drug Bay K8644 (Fig. 4B). This finding, though appearing peripheral to the major thrust of the proposed O₂ sensing pathway, maybe relevant in the light of finding that Bay K8644 suppresses the Ca²⁺-dependent inactivation (CDI) property of the channel (Adachi-Akahane et al. 1999) (perhaps by interfering with the binding of Ca²⁺ to CaM). The specificity for Ca²⁺ also implicates a critical role for CaM in the Ca²⁺ channel O₂ sensing. Consistent with this idea, the CaM-inhibitory peptide reversed the protective effects of phosphorylation in cAMP-dialysed myocytes, making the channel once again responsive to withdrawal of O₂ (Fig. 4B). Further, possible involvement of CaMKII phosphorylation became apparent based on experiments showing that (1) an isoform of α_1C channel, deleted from CaM/CaMKII binding motifs, was insensitive to O₂ withdrawal (Fig. 5), and (2) inhibition of CaMKII phosphorylation by CK59 in cAMP dialysed myocytes restored the channel sensitivity to O₂ (Fig. 3C). Even though these findings implicated the Ca²⁺-regulatory domains of Ca²⁺ channel in the hypoxic response, we could not identify a candidate AA sequence that would bind O₂. The possible involvement of haem oxygenase became likely when we considered that this molecule had structural and functional similarities with CaMKII as to allow it to interact with the carboxyl-tail of the Ca²⁺ channel (Colbran, 1993; Boehning et al. 2004; Xu et al. 2010). Supporting this idea were the findings that (1) HO inhibitors (ZnPP and SnPP) also rapidly suppressed I_Ca (Fig. 6), (2) the blocking effect of HO inhibitors was abolished in cAMP-dialysed myocytes only if Ca²⁺, but not Ba²⁺ was the permeating cation (Fig. 7A, B and D), and (3) the suppressive effects of hypoxia and HO-2 inhibitors were not additive, i.e. either intervention blocked the suppressive effect of the other (Fig. 7C). A possible pathway for O₂ sensing pathway, based on these findings, is illustrated in Fig. 8. Though HO-2 inhibitors have the same pharmacology and phosphorylation fingerprints as hypoxia, making HO-2 a likely candidate for O₂ binding, it remains to be determined how this interaction leads to suppression of the Ca²⁺-channel current. Future experiments, possibly through mutagenesis, or genetic engineering, may provide more insight on the detailed molecular signalling of O₂ sensing.

Figure 8. Proposed model for modulation of L-type Ca²⁺ channel by oxygen.

The carboxy-tail of the α_1C subunit includes LA and IQ motifs for calmodulin- (CaM) mediated interactions and a binding site for protein phosphatase 2a (PP2a). CaM is shown as also interacting with CaMKII, haem oxygenase-2 (HO-2) and inhibitory peptide_290−309 (PEP). The primary inputs to the system are thought to be: cAMP leading to PKA-mediated phosphorylation (left), permeation of Ca²⁺ through the channel to a restricted space (middle), and molecular oxygen (O₂) that binds to haem/HO. PKA may act directly or via CaMKII and may have multiple phoshporylation targets (including HO-2). Similarly okadaic acid (OA) may promote phosphorylation inhibiting different phosphatases. ROS generation by mitochondria or membrane bound NOX proteins, CO-mediated signalling and I_Ca-triggered release of Ca²⁺ from the SR are thought to be of little or no importance for the rapid suppression of the Ca²⁺ channel at the onset of hypoxia.

L-type Ca²⁺ channels as primary oxygen sensor during acute cardiac hypoxia

In our studies we mostly compared baseline values of I_Ca in experimental cells to their values during the first minute of hypoxia. This approach raises the question as to what extent rundown of I_Ca may have contributed to its suppression during hypoxia. Based on control experiments without hypoxia (Fig. 1C), readmission of oxygen (Fig. 1A), and washout of HO inhibitors (Figs 6A and B and 7D) we estimate a rundown of ∼5% during the first 50–60 s of repetitive I_Ca measurements. Corrections of this magnitude are warranted when considering the hypoxic suppressions of I_Ca that were consistently 20–30% under non-phosphorylating conditions (Figs 1 and 2). The observed rundown may have masked the effectiveness of cAMP-induced desensitization of I_Ca (Figs 3 and 7), but it is not of sufficient magnitude to invalidate the experiments testing protein kinase inhibitors (H-89, C-59, Figs 3C and 7D), peptide_290−309 (Fig. 4), or the α_1c86 isoform of the channel (Fig. 5) and it is of little importance when considering the larger and faster suppressions of I_Ba and I_Na.

Our experimental approach was designed to simulate the response of cardiac I_Ca to acute hypoxia under different degrees of β-adrenergic stimulation and PKA-mediated phosphorylation (Herzig et al. 1993). The macromolecular signalling complex that provides effective regulation of I_Ca appears to be located at the C-terminus of α_1C subunit and includes the adenylate cyclase/PKA signalling complex and protein phosphatases PP2a and PP2b (Xu et al. 2010). We propose that the haem oxygenase may be integrated into this system perhaps via the regulatory pathways that start with (1) cAMP, (2) permeation of Ca²⁺ through the channel, and (3) molecular oxygen (O₂) all converging on calmodulin (Fig. 8).

Roles of CaM and CaMKII in the hypoxic regulation of the cardiac Ca²⁺channel

It is a novel idea that CaM and CaMKII play a critical role in the hypoxia-induced suppression of the Ca²⁺ channel (Movafagh & Morad, 2010). Our observations implicating these signalling molecules are that (1) O₂-dependent suppression of I_Ca is absent in HEK cells expressing mutant α_1c86 channel lacking the CaM/CaMKII binding domain (Fig. 5); (2) in cardiac myocytes the suppressive effect of hypoxia, absent in cAMP-phosphorylated cells, could be restored by dialysing peptide_290−309 (Fig. 4); (3) given the reported interactions between HO-2 and CaM/CaMKII (Boehning et al. 2004), there is striking similarity in the suppressive effects of hypoxia (Fig. 3C) and HO inhibitors on I_Ca (ZnPP, Fig. 7E); and (4) CaMKII inhibitor, CK59, restored ZnPP-induced suppression of I_Ca in cardiac cells dialysed with cAMP (Fig. 7C). Considering the previously proposed signalling pathway for cAMP, Ca²⁺ channel, and CaM/CaMKII (as simplified in the schematic diagram of Fig. 8) and the findings reported here, we feel it is reasonable to include an O₂-signalling pathway that interacts with CaM/CaMKII signalling as suggested in Fig. 8.

PKA phosphorylation and desensitization of O₂ sensing

The finding that PKA phosphorylation significantly desensitizes the channel to acute hypoxia (Fig. 3) was somewhat unexpected. Hool et al. (2000, 2005, 2007) reported that hypoxia enhances β-adrenergic sensitivity of I_Ca via specific activation of a PKC isoform and suggested that O₂ loss enhances PKA phosphorylation. Although we do not find enhancement of I_Ca of the phosphorylated channel on withdrawal of O₂, the channel did become less sensitive to O₂ withdrawal (Fig. 3A) and the O₂ sensitivity of the channel could be restored by the PKA inhibitor H-89 in cAMP dialysed cells (Fig. 3C). It is unlikely that these findings are related to activation of PKC, and they do not address whether crosstalk between the PKA and PKC phosphorylation pathways (Yang et al. 2009) is of importance for the hypoxic modulation of I_Ca or if different phosphorylation patterns might explain the hypoxic enhancement of I_Ca found by Hool et al. Possibly the use of isoproterenol in combination with strong intracellular Ca²⁺ buffering (10 mm EGTA/1 mm Ca²) may have blunted CaM/CaMKII signalling in the experiments reported by these authors. Furthermore, in our experiments, the desensitization induced by PKA phosphorylation of the channel was only present when cAMP levels were elevated by either internal dialysis of cAMP or by phosphatase inhibitors, and not by basal cAMP levels (Fig. 1A).

The role of Ca²⁺-dependent signalling in the hypoxic response

The endogenous activity of CaMKII depends on its auto-phosphorylation at T₂₈₆, which is in turn modulated by the phosphatase PP1. Stimulation of β-adrenergic receptors suppresses PP1 activity via PKA-dependent phosphorylation of the PP1 inhibitor I-1 at T₃₅ (Bradshaw et al. 2003; Huke & Bers, 2007). Thus, PKA-induced auto-phosphorylation of CaMKII may be the pathway by which PKA controls the channel's sensitivity to hypoxia. Our data clearly show that the desensitization of the phosphorylated Ca²⁺ channel to hypoxia requires permeation of Ca²⁺, but not Ba²⁺ or Na⁺, through the channel, consistent with CaMKII activity requiring high local rather than global cytosolic Ca²⁺ concentrations (Zuhlke et al. 1999). The specific requirement for Ca²⁺ influx, but not SR Ca²⁺ release, in desensitizing the phosphorylated channel to O₂ withdrawal was unexpected considering the report that SR Ca²⁺ release accelerates the decay kinetics of I_Ca (Adachi-Akahane et al. 1996), but at the same time emphasizes the specificity of interactions of Ca²⁺ in the nano-domains of the C-terminus as evident by re-sensitization of the phosphorylated channel to hypoxia by dialysis of CaMKII inhibitory peptide_290−309 suggesting that CaMKII phosphorylation desensitizes the channel to O₂ loss. Since the primary site of CaMKII-induced Ca²⁺-dependent facilitation is within the C-terminus (residues 1654–1659) and mutation of the IQ region abolishes such effects (Hudmon et al. 2005), we posit that the inhibitory peptide_290−309 may prevent the binding of CaMKII to the IQ region (Payne et al. 1988) thus allowing the recovery of O₂ sensitivity of the channel. Mutation of residues 1572–1651 (α_1c86 channel isoform), which includes the entire CaM/CaMKII-binding motifs of the C-terminus (Fig. 4), completely desensitizes the channel to O₂ withdrawal. A different region of the C-terminus has previously been implicated in the hypoxia-induced suppression of the cardiac Ca²⁺ channel (AA 1786–1821; Fearon et al. 2000) and is now known to bind a protein phosphatase (AA 1795–1818, PP2a; Xu et al. 2010) that antagonizes PKA-mediated phosphorylation at S₁₉₂₈ (Fig. 8, left; Davare et al. 2000). This is consistent with the notion that modulation of I_Ca by O₂ and HO inhibitors is governed by the balance between kinases and phosphatases (Figs 3 and 7).

Curiously, hypoxia does not change the kinetics of the measured currents (Figs 1 and 3) despite the fact that the removal of the CaMKII/CaM binding (where IQ motif is placed) impairs responses elicited by hypoxia on the Cav1.2 currents. This may be related to the possible role of this site in transmitting O₂ sensing to the pore region of the channel, but not serving as the O₂ sensor itself.

Possible role of haem oxygenase as an O₂ sensor of the Ca²⁺ channel

HO-2 has been suggested to serve as the O₂ sensor for potassium channels in the carotid bodies (Williams et al. 2004; Lahiri et al. 2006). Significant levels of the constitutively active isoform of HO-2 are expressed in cardiomyocytes, and are enhanced by β-adrenergic agonists (Ding et al. 2011). It is further known that HO-2 complexes with CaM (Boehning et al. 2004), which led us to examine whether it may serve as an O₂ sensor feeding into the CaM/CaMKII signalling pathway. In support of this idea we found that HO-inhibitors (1) rapidly suppressed I_Ca with kinetics similar to those of hypoxic suppression, and (2) blocked completely the suppressive effects of hypoxia on I_Ca (Figs 6 and 7), consistent with a unitary locus of action for hypoxia and HO inhibitors. We posit that in PKA-activated myocytes Ca²⁺ influx through the channel phosphorylates CaMKII allowing it to associate with CaM and possibly preventing HO/CaM binding, and the resultant suppression of I_Ca. In this scheme, dephosphorylated CaMKII would facilitate CaM association with HO and facilitate modulation of I_Ca by rapid changes in O₂ concentrations. This may help to explain why the deletion of CaM/CaMKII binding motifs in the α_1c86 mutant channel renders it insensitive to O₂ removal. These considerations implicate CaMKII as a modulatory partner of the O₂ sensing site, rather than the oxygen sensor itself. Indeed, if activated CaMKII were needed for hypoxia-induced effects on Cav1.2, the use of other cations to replace Ca²⁺ would not have resulted in an increased O₂ sensing at all. In this scheme the sensing of O₂ by HO is thought to be transmitted rapidly and directly to the Ca²⁺ channel via the CaM/CamKII system. In an alternative role, HO is thought to suppress the Ca²⁺ channel by generating CO thereby stimulating mitochondrial ROS production and redox modulation of the carboxy-tail of α_1C (Scragg et al. 2008).

Do reactive oxygen species (ROS) mediate the rapid O₂ sensing of the channel?

Since other groups have proposed that the modulation of I_Ca by O₂ depends on a redox mechanism (Fearon et al. 1999; Hool, 2000; Hool & Corry, 2007) we examined this possibility by treating the cells with intra- or extracellular reduced glutathione (0.5–10 mm) to examine the role of ROS in the early onset of I_Ca suppression, but found no significant alterations in time course or magnitude of the O₂ withdrawal response (Movafagh & Morad, unpublished data). This led us to conclude that reduced thiol residues were unlikely to be responsible for the fast suppression of I_Ca within the first 5 s of O₂ withdrawal (Figs 1, 2 and 4B) and consequently we pursued experiments testing the role of haem oxygenase in rapid O₂ sensing. In sharp contrast to rapid suppression of I_Ca, ROS generation by mitochondria proceeds too slowly (Hool et al. 2005; Korge et al. 2008) to be compatible with the kinetics of I_Ca suppression. Alternatively it may be argued that I_Ca could be modulated by ROS generated by an extra-mitochondrial O₂-dependent pathway, e.g. involving a membrane-bound NADPH oxidase, like NOX-2, that generates superoxide that in turn triggers release of Ca²⁺ from the SR within a fraction of a second (Prosser et al. 2011). Inconsistent with such a possibility, we found no involvement of the SR Ca²⁺ release in the rapid hypoxic response (Fig. 3B). Further, NOX-2, though known to bind haem, is not modulated by haem analogues like ZnPP. Yet another redox mechanism to be considered involves CO, which is reported, to modulate the BK (Yi et al. 2010) and Ca²⁺ channels (Scragg et al. 2008). We did not examine the effects of CO, in part because it is a likely modulator of any haem-binding protein and therefore non-specific with respect to the identity the oxygen sensor of the Ca²⁺ channel. Thus, our data and the existing literature point to haem oxygenase as the likely oxygen sensor of the Ca²⁺ channel, but they do not exclude the possibility that other O₂-sensitive proteins and processes are also involved in the cumulative effects of hypoxia.

Pathophysiological implications

Our data clearly show that Ca²⁺ channels of the heart respond to the decrease in O₂ tension well within 50 s, suggesting a rapid detection mechanism for hypoxia. In general this early mechanism is embedded within the sequence of ischaemic events that occur over periods of minutes to hours. Under these conditions, early and transient suppression of I_Ca may work as a protective mechanism against subsequent Ca²⁺-mediated toxicity. In this regard inhibition of Ca²⁺ channels by nifedipine may reduce the degree of phospholamban phosphorylation by CaMKII within the first 5 min of ischaemia (Vittone et al. 2002). In our simultaneous measurements of I_Ca and Ca²⁺ transients in non-phosphorylated cells, we found an ∼10% decrease in Ca_i-transients after 40 s of hypoxia compared to an ∼20% decrease in I_Ca. The small hypoxia-induced decrease in released Ca²⁺ is consistent with the modest rapid suppression of I_Ca and its role in regulating Ca²⁺-induced Ca²⁺ release. Considering the high degree of cooperativity in activation of contraction by binding of Ca²⁺ to troponin and in cross bridge recycling, it is plausible that associated changes in energy demands may be considerably larger (Fabiato & Fabiato, 1975; Han et al. 2009; ter Keurs, 2012).

The pharmacological approach of using (S)-(–)-Bay K8644 in combination with the CaMKII inhibitor peptide_290−309, where we found significant reduction of I_Ca even in phosphorylated channels, provides an intriguing conceptual possibility for using a similar clinical strategy, but such an approach is far from compelling given the arrhythmogenicity associated with Ca²⁺ channel agonists. Nevertheless, a mixed DHP agonist/antagonists therapeutic approach in combination with a CaMKII inhibitor could be a potential therapeutic avenue to consider.

Glossary

HO

haem oxygenase

PEP

inhibitory peptide_290−309

PKA

protein kinase A

ROS

reactive oxygen species

SnPP

tin protoporphyrin

YFP

yellow fluorescent protein

ZnPP

zinc protoporphyrin

Author contributions

A.O.R and S.M. did almost all of the experimental work. All authors took part in design, analysis and preparation of the manuscript and have approved the manuscript. The work was done both at Georgetown University and the Medical University of South Carolina.

Disclosures

None.

Supplementary material

Supplemental Figure 1

Supplemental Figure 2