Fe²⁺-Mediated Activation of BK_Ca Channels by Rapid Photolysis of CORM-S1 Releasing CO and Fe²⁺

Abstract

Heme catabolism by heme oxygenase (HO) with a decrease in intracellular heme concentration and a concomitant local release of CO and Fe²⁺ has potential to regulate BK_Ca channels. Here we show that the iron-based photolabile CO-releasing molecule CORM-S1 [dicarbonyl-bis(cysteamine)iron(II)] co-releases CO and Fe²⁺, making it a suitable light-triggered source of these downstream products of HO activity. To investigate the impact of CO, iron and cysteamine on BK_Ca channel activation, human Slo1 (hSlo1) was expressed in HEK293T cells and studied with electrophysiological methods. Whereas hSlo1 channels are activated by CO and even more strongly by Fe²⁺, Fe³⁺ and cysteamine possess only marginal activating potency. Investigation of hSlo1 mutants revealed that Fe²⁺ modulates the channels mainly through the Mg²⁺-dependent activation mechanism. Flash photolysis of CORM-S1 suits for rapid and precise delivery of Fe²⁺ and CO in biological settings.

INTRODUCTION

Heme degradation by heme oxygenase (HO), including the inducible enzyme HO-1 and the constitutive form HO-2, results in the formation of biliverdin, ferrous iron (Fe²⁺) and carbon monoxide (CO).¹ Biliverdin is known for its antioxidant and antiviral properties,^{2, 3} and is further broken down to bilirubin via the activity of biliverdin reductase. Fe²⁺ is typically considered a co-factor for protein function within either iron/sulfur clusters or heme. When liberated during heme catabolism, Fe²⁺ has oxidative potential but was so far considered an inconsequential by-product, most likely because of scavenging by ferritin and iron transport mechanisms.⁴ CO, although toxic at high concentration, meanwhile is accepted as an important gaseous signaling molecule.⁵

The vasodilatory effects of CO led to the hypothesis that CO may be a biological mediator of cellular functions.⁶ This hypothesis is supported in various animal models of human diseases.⁷ CO contributes to important physiological functions including the regulation of vascular smooth muscle tone and blood pressure, suppression of inflammation, and protection against ischemia, septic shock and hypoxia.⁸ The induction of HO-1 under stress conditions strongly hinted that endogenously produced CO had a beneficial or therapeutic effect.^{9, 10} These findings have established CO as an important gaseous messenger molecule akin to NO and H₂S.¹¹

To fully realize the promise of CO as a therapeutic agent, a key challenge is to find novel routes of CO delivery specifically to diseased tissues in need of treatment. Several CO vehicles, so-called CO-releasing molecules (CORMs), have therefore been developed.^{7, 12, 13} To enable a safe and site-specific administration of CO, CORMs releasing CO only upon tailored triggers should be beneficial. CORM-S1 [dicarbonyl-bis(cysteamine)iron(II)] (SFigure 1A) holds this promise as this compound liberates CO under illumination with visible light (SFigure 2).¹⁴ Furthermore, because CORM-S1 is based on ferrous iron, it is expected that illumination results in liberation of CO, Fe²⁺, and cysteamine, the latter of which being an endogenously produced breakdown product of coenzyme A.¹⁵ Therefore, to some degree CORM-S1 may constitute a light-triggered mimic of HO activity.

In physiological experiments, CORM-S1 activates CO-sensitive ion channels in a light-dependent manner including large-conductance Ca²⁺- and voltage-dependent K⁺ channels (BK_Ca, MaxiK, K_Ca, K_Ca1.1).¹⁴ BK_Ca channels open in response to an increase in intracellular Ca²⁺ concentration to μM levels and/or membrane depolarization, typically providing a negative feedback influence on cellular excitability.¹⁶ The channels are found in almost every tissue and their physiological roles are well documented in many phenomena, including regulation of smooth muscle tone, determination of action potential duration and neurotransmitter release.¹⁶ Dysfunction of BK_Ca channels is associated with multiple human diseases.^17–21 While activation of BK_Ca channels promotes vascular relaxation, channel inhibition leads to vascular constriction.²² Consistent with this framework, CO released from CORMs was shown to dilate blood vessels by opening of BK_Ca channels in smooth muscle cells.⁶

BK_Ca channels are composed of four pore-forming α subunits encoded by the Slo1 gene (KCNMA1 in humans). Each subunit contains 7 transmembrane segments (S0-S6) and a large C-terminal cytoplasmic region,¹⁶ which harbors two homologous structural domains, RCK1 and RCK2.²³ Four sets of RCK1/RCK2 dimers form the “gating ring” of BK_Ca channels.²⁴ Recently determined cryo-EM structural models based on Aplysia Slo1 and human Slo1 (hSlo1) provide insight into the molecular arrangements of the functional domains.^{25, 26} Each subunit possesses 3 structurally and functionally distinct divalent cation sensors: RCK1 sensor in the gating ring, RCK2 sensor in the gating ring (“Ca²⁺ bowl“), and the lower-affinity so-called Mg²⁺ sensor encompassing both the transmembrane domain and the gating ring. The cytosolic gating-ring domain harbors multiple regulatory sites ²⁷ including those for the primary activating ions Ca²⁺ and Mg²⁺.^{28, 29} In addition, H⁺,¹² Zn²⁺,³⁰ and CO ^{31, 32} can augment channel activation. Some transition metal ions have also been reported to interfere with BK_Ca channels.³³ Contradictory reports on the impact of Fe²⁺ on BK_Ca channels^{34, 35} ask for a closer examination using improved experimental protocols.

Modulation of BK_Ca channels by small molecular compounds may be particularly important in the carotid body where the channels are inhibited during hypoxia, thus promoting cell depolarization, Ca²⁺ influx, and transmitter release.^{36, 37} A key link between hypoxia and BK_Ca activity appears to be HO-2, which is postulated to be closely associated with the channel and serves as an oxygen sensor of native and recombinant hSlo1 BK_Ca complexes.³⁸ However, the molecular mechanism by which HO-2 activity affects BK_Ca channel function is unclear. Besides the alteration of the pool of available heme, the downstream products CO and Fe²⁺ may have signaling function when delivered close to the target. While CO has a large diffusional radius, Fe²⁺ is subject to an intricate buffer system and moreover may be oxidized to Fe³⁺, both aspects asking for molecular tools for deliberate release of Fe²⁺ at specific sites of action.

Here we studied how hSlo1 BK_Ca channels are activated by heme breakdown products by using CORM-S1 and examined if this CO-releasing compound might serve as a chemical and physiological tool to concurrently deliver CO and Fe²⁺. In particular, we address to what extent BK_Ca channels are subject to regulation by Fe²⁺ ions.

RESULTS AND DISCUSSION

CORM-S1 is a Light-Triggered Releaser of CO and Fe²⁺

Dicarbonyl-bis(cysteamine)iron(II) (CORM-S1, SFigure 1A) was introduced as a photosensitive CO-releasing molecule ¹⁴ and, hence, it might be an easy-to-use tool for studying the impact of CO on physiological systems. Moreover, CORM-S1 is expected to liberate Fe²⁺ upon illumination. In this respect, this low-molecular weight compound should mimic the activity of HO to produce CO and Fe²⁺ with the advantage of light serving as a direct trigger. We therefore compared the stimulation of light-exposed CORM-S1 on excised membrane patches with hSlo1 BK_Ca channels using either fluoride solutions (KF) or EGTA-buffered solution; the latter is expected to more efficiently chelate both Ca²⁺ and Fe²⁺ ions in solution. KF, in contrast, selectively eliminates Ca²⁺ by precipitating it as CaF₂. While illumination of CORM-S1 potently increased the hSlo1 BK_Ca currents at 100 mV in KF solutions (Figure 1A, B), it only resulted in a small activating effect with a slow onset in EGTA solutions (Figure 1B). The residual increase in current was similar in magnitude and kinetics to what was previously observed for the application of CO gas.³¹ To infer about the redox state of iron released from CORM-S1 upon illumination, we investigated the liberation of Fe²⁺ ions using a photometric assay of 1 mM 1,10-phenanthroline, which upon binding of Fe²⁺ forms ferroin. This reaction was monitored as a change in absorbance at 510 nm (Figure 1C). Control recordings showed that CORM-S1 did not noticeably release Fe²⁺ in the absence of light during an experiment of 3 min (Figure 1C). For quantitative analysis of Fe²⁺ release from CORM-S1, a ferroin calibration curve with FeSO₄ was acquired (Figure 1D). Illumination of CORM-S1 samples resulted in comparable ferroin absorption values as FeSO₄ samples at the same concentration (Figure 1D), suggesting a full release of the central iron of CORM-S1 as Fe²⁺.

Figure 1.

Activation of hSlo1 BK_Ca channels by light-exposed CORM-S1. (A) Current recordings (bottom) in the inside-out patch-clamp configuration of HEK293T cells expressing hSlo1 BK_Ca channels for the indicated pulse protocol (top) before (Control) and after 1-min light exposure of 50 μM CORM-S1 in KF intracellular solutions. (B) Time course of relative current at 100 mV in the presence of 50 μM CORM-S1 for the indicated light stimulation (bar) in KF solutions (open circles) and in EGTA (10 mM)-buffered solutions (filled circles). Straight lines connect data points for clarity. The dashed horizontal line denotes 1. (C) Analysis of iron released from CORM-S1 in a cuvette measured by the absorbance at 510 nm of ferroin. 50 μM CORM-S1 was dissolved in KF solution containing 1 mM 1,10-phenanthroline. Samples with (blue) or without (gray) illumination as indicated. Error bars are smaller than the symbol size. (D) Fe²⁺/1,10-phenanthroline calibration curve (triangles) in KF solution with a superimposed linear fit (red line). The maximal OD_{510 nm} after illumination of CORM-S1 (from C) is depicted as blue circle (mean ± standard error, SEM, n = 3).

For an application of CORM-S1 in a physiological setting, it is mandatory to know the kinetics of CO/Fe²⁺ release and ascertain stability of the compound in the physiological buffers used. We therefore assayed the CO release using a myoglobin absorption assay ¹⁴ under various conditions. When illuminated with white light, CORM-S1 rapidly releases two CO molecules, and termination of illumination immediately stops the release (SFigure 1C). The release kinetics was described with a single-exponential function with a time constant of 50 ± 1 s (n = 4) when the measurements were performed in PBS with CORMs dissolved from dry samples immediately before use. When stored as frozen stock solutions in DMSO for up to 48 days, CORM-S1 retained its activity (SFigure 1E). Freshly prepared, CORM-S1 released CO to a maximal extent regardless of whether it was dissolved in PBS or in EGTA- or KF-solutions; the latter buffers were used in subsequent physiological assays (SFigure 1E). In comparison, the commonly used CO releaser CORM-2 (tricarbonyldichlororuthenium(II) dimer) is considerably less stable in storage (SFigure 1B, D, F). Thus, CORM-S1 is a rapid light-triggered releaser of CO and Fe²⁺ with long-term stability in DMSO, making it a valuable and easy-to-use tool in physiological research. While CO release from CORM-2 is strongly accelerated by the presence of dithionite,³⁹ the breakdown of CORM-S1 entirely relies on illumination, thus justifying the use of the myoglobin release assay (SFigure 1).

Activation of BK_Ca Channels by CO and Fe²⁺

BK_Ca channels are likely down-stream targets of HO; one of its reaction products, CO, can activate these channels.^{31, 32} Because Fe²⁺ is also released when HO breaks down heme, we tested the effect of Fe²⁺ on excised inside-out membrane patches from HEK293T cells expressing hSlo1. In KF solution, i.e. with a very low concentration of free Ca²⁺, membrane depolarization to 100 mV opens a small fraction of channels (P_open ~ 2%). Application of 100 μM Fe²⁺ (from an FeSO₄ stock solution at low pH, 0.1 N HCl) increased channel activity by about 25-fold (Figure 2A). Application of an equivalent amount of HCl did not affect the channels (data not shown). The time course of the current at 100 mV, normalized to control conditions, is shown in Figure 2A, B; a rapid and robust increase in current is followed by a slower (time constant about 120 s) decline of current size. The latter phenomenon might be attributed to progressive oxidation of Fe²⁺ to Fe³⁺ and subsequent precipitation of iron(III)-hydroxide (FeO(OH)), which is insoluble in water, since all experiments were performed under ambient conditions. Fe²⁺ stability assays in the same KF buffer used for patch-clamp experiments showed an about 5-fold slower decrease in the Fe²⁺ concentration (SFigure 3A-C), presumably because greater availability of oxygen in an open cell-culture dish compared to a test tube; for stability of Fe²⁺ under anoxic conditions, see SFigure 3D.

Figure 2.

Effects of CORM-S1 breakdown products on hSlo1 BK_Ca channels. (A) Current recordings in the inside-out patch-clamp configuration of HEK293T cells expressing hSlo1 BK_Ca channels for the indicated pulse protocol before (Control) and after application of 100 μM Fe²⁺ solution to the intracellular side. (B) Time course of the current at 100 mV relative to the control value (I / I_control) with application of 100 μM Fe²⁺ solutions at time zero. (C-F) Time course of the relative current at 100 mV upon application of 100 μM Fe²⁺ in the presence of 500 μM reduced GSH (C), 100 μM Fe³⁺ (from FeCl₃) (D), 400 μM cysteamine (E), and 200 μM CO gas dissolved in buffer (F). Straight lines connect data points in B-F for clarity. Internal solutions in A-D were fluoride-buffered, and in E-F EGTA-buffered. Data points in B-F are means ± standard error (SEM) with n in parentheses.

Oxidation of Fe²⁺ to Fe³⁺ is slowed by reducing agents. We therefore performed experiments with 100 μM Fe²⁺ in the presence of 500 μM reduced glutathione (GSH) and found a fractional BK_Ca current increase of about 12 (compared to a factor of 25 without GSH) (Figure 2C). GSH alone also augmented the currents by a factor of 2. Further, the current increase was much more sustained (time constant about 230 s; Figure 2C). Application of 100 μM Fe³⁺ (from a FeCl₃ stock solution) only marginally increased BK_Ca currents (Figure 2D). Thus, hSlo1 BK_Ca channels are preferentially activated by Fe²⁺ but not by Fe³⁺.

Besides CO and Fe²⁺, breakdown of CORM-S1 also produces cysteamine, which was ineffective in activating hSlo1 BK_Ca channels (Figure 2E). Application of CO gas dissolved in the recording buffer (200 μM) only slowly increased the current by a factor of 2 (Figure 2F), as previously reported.³¹ The fractional increase in current by CO gas was similar in magnitude as the hSlo1 BK_Ca activation observed with illuminated CORM-S1 in the presence of EGTA (Figure 1B). Taken together, CORM-S1 exerts its activating effect on BK_Ca channels primarily via Fe²⁺ ions, while CO plays a secondary role, and cysteamine has almost no impact on channel function.

Fe²⁺ Activates BK_Ca Channels via Ca²⁺- and Mg²⁺-Binding Sites

Because of the limited redox stability of Fe²⁺ under ambient conditions (Figure 2), the activating effect of Fe²⁺ on BK_Ca channels was rapidly quantified by measuring currents in response to voltage ramps. Described with a Boltzmann function and a linear ion-permeation characteristic (Equation 1, Figure 3A), a shift in the half-maximal activation voltage (ΔV_0.5) was extracted as our primary data-description parameter. ΔV_0.5 was measured at pH 7.4 for various Fe²⁺ concentrations yielding a half-maximal effective concentration, EC₅₀, of 25 μM and a maximum shift of about –45 ± 5 mV (n = 6; Figure 3B). This EC₅₀ value is an upper limit because the real concentration of free Fe²⁺ must be lower: some Fe²⁺ is expected to be oxidized and to precipitate as FeO(OH). Because of the stability of Fe²⁺ inside the CORM-S1 complex, using CORM-S1 as a light-triggered source of Fe²⁺ will be a preferential tool for studying Fe²⁺ effects on biological systems, particularly at physiological pH which strongly favors oxidation and precipitation of Fe²⁺ as FeO(OH).

Figure 3.

pH-dependent hSlo1 BK_Ca activation by Fe²⁺. (A) bottom Current recordings of hSlo1 BK channels at pH 7.4 upon stimulation with a voltage ramp (−60 to 240 mV in 150 ms) before (Control) and after application of 100 μM Fe²⁺ to the intracellular side with superimposed fits (red; Eq. 1), used to estimate the shift in half-maximal activation voltage, ΔV_0.5. top Fractional current change with Fe²⁺ application. (B) ΔV_0.5 as a function of [Fe²⁺] with superimposed Hill fit (Eq. 2). Fit parameters were: EC₅₀ = 25 μM, n_H = 1.5, ΔV_{0.5, max} = −50 mV. (C, D) Data as for A and B at pH 6.8. Fit parameters in D were n_H = 0.7, ΔV_{0.5, max} = −29 mV, EC₅₀ = 5 μM.

Overall, the effect of Fe²⁺ on BK_Ca channels is reminiscent of the effects exerted by H⁺ and divalent cations, such as Ca²⁺, Mg²⁺ and Zn²⁺.^{27, 30, 40, 41} We therefore tested whether Fe²⁺-mediated BK_Ca activation is sensitive to the intracellular pH. As shown in Figure 3C and D, hSlo1 BK_Ca activation by Fe²⁺ was weaker at pH 6.8 (as compared to pH 7.4) with a maximum shift in V_0.5 of –27 ± 2 mV (n = 6). However, the half-maximal activation concentration was 5 μM, indicating a higher affinity. At pH 6.0, 100 μM Fe²⁺ only marginally activated hSlo1 BK_Ca channels with a ΔV_0.5 of –5 ± 2 mV (n = 5). At pH 8.0 the maximal Fe²⁺-induced shift was not much greater than that at pH 7.4, ΔV_0.5 = –48 ± 7 mV (n = 5).

To infer the mechanism of Fe²⁺-mediated channel activation, we compared ΔV_0.5 obtained with 100 μM Fe²⁺ for hSlo1 BK_Ca wild-type channels with those found for various mutants with modified cation sensors. The double mutation H365A:H394A in the RCK1 domain impairs activation of BK_Ca channels by intracellular H⁺ and CORM-2; ^{31, 32} H365A eliminates the activating effect of Zn²⁺.³⁰ While the double mutation H365A:H394A diminished the channel activation by CORM-2 to a factor of 2 originating from the released CO,³¹ ΔV_0.5 obtained with Fe²⁺ was only marginally affected (Figure 4B).

Figure 4.

Structural requirements for BK_Ca channel activation by Fe²⁺. (A) Topological model of a hSlo1 BK_Ca channel α subunit with the transmembrane region and the cytosolic RCK1 and RCK2 regulatory domains. Residues important for channel activation by H⁺ (green: H365, H394), Mg²⁺ (magenta: D99, N172, E374, E399), and Ca²⁺ (blue/red: M513, E535 in RCK1; black: D894 and D895 in RCK2) are indicated. (B) Shifts in half-maximal activation voltage induced by 100 μM Fe²⁺ for hSlo1 wild type (top) and the indicated channel mutants. ΔCB refers to the deletion of D894 and D895, and HA:HA to the double mutation H365A:H394A. n values are indicated in parentheses. * p < 0.05 and ** p < 0.01 for a test versus the wild-type channel after post-hoc Bonferroni correction.

We furthermore utilized the mutants M513I and E535A known to selectively impair the RCK1 Ca²⁺-sensor site:⁴² ΔV_0.5 values of these mutants for Fe²⁺ were only half as large as for the wild type (Figure 4B). In the two-residue deletion mutant ∆D894:D895, the RCK2 “Ca Bowl” Ca²⁺ sensor is not functional (ΔCB; ⁴³); ΔV_0.5 induced in this mutant by Fe²⁺ was only 50% of that for the wild type (Figure 4B). Even with the mutations designed to disrupt both the RCK1 and RCK2 Ca²⁺ sensors ([dummy]∆CB:M513I and ∆CB:E535A), about 50% of ΔV_0.5 found in the wild type remained. In contrast, the mutations D99R, N172R, E374R, and E399A, each of which disrupt the activation effect of mM levels of intracellular Mg²⁺,^{27, 28} rendered the channels almost insensitive to Fe²⁺ (Figure 4B). The ΔV_0.5 by Fe²⁺ was essentially absent in ∆CB:E399A with the impaired Mg²⁺ and RCK2 Ca²⁺ sensors. The mutagenesis results show that the effects of Fe²⁺ on V_0.5 are complex, affecting all three divalent cation sensors in a non-additive manner. However, because those mutations affecting the low-affinity so-called Mg²⁺ sensor markedly diminish Fe²⁺-induced ΔV_0.5, the Mg²⁺ site must play a major role.

Activation of BK_Ca channels by Mg²⁺ requires voltage-sensor activation.⁴¹ We tested whether BK_Ca channels can be activated by Fe²⁺ when the voltage sensors are preferentially in the resting state by recording single-channel events at –100 mV. In fact, voltage-sensor activation does not seem to be required for the activation of BK_Ca channels by Fe²⁺ (SFigure 4), indicating that the action of Fe²⁺ similar to that of Ca²⁺.⁴⁰ However, while activation of BK_Ca channels by Ca²⁺ is augmented by inclusion of human Slo β1 subunits,⁴⁴ Fe²⁺-mediated current increase of BK_Ca channels containing human Slo β1 subunits was not enhanced compared to homomeric hSlo1 BK_Ca channels (SFigure 5A, B), and ΔV_0.5 for channel activation was even smaller (SFigure 5C). Taken together, these results show that Fe²⁺ ions bind to and enhance activity of BK_Ca channels in a complex manner with some similarity and differences to Mg²⁺ and Ca²⁺ ions.

To test whether physiologically relevant concentrations of Mg²⁺ may attenuate BK_Ca activation by Fe²⁺, F^– was replaced with Cl^– because F^– also precipitates Mg²⁺. To minimize the concentration of contaminating Ca²⁺, which is not eliminated by precipitation as CaF₂ under these conditions, would become a confounding factor, a solution containing 2 mM MgCl₂ was produced from ultrapure salts and water. Under such conditions, which resemble a physiological situation under which BK_Ca channels are exposed to about μM levels of free Ca²⁺ and ~2 mM Mg²⁺, 100 μM Fe²⁺, either directly added from an FeSO₄ stock solution or photoreleased from 100 μM CORM-S1, induced a shift in BK_Ca V_0.5 by –52.6 mV and of –59.6 mV (n = 6), respectively (Figure 5). Such shifts in V_0.5 are indistinguishable from those obtained in fluoride-based solutions (e.g., Figure 4C). Therefore, photoreleased Fe²⁺ from CORM-S1 retains the potential of activating BK_Ca channels even in a physiological environment with potentially competing Mg²⁺ ions because Fe²⁺ has a much higher affinity to the target sites than Mg²⁺.

Figure 5.

Fe²⁺ activates hSlo1 BK_Ca channels in the presence of competing Ca²⁺ and Mg²⁺. (A) bottom Representative inside-out patch current recordings in response to stimulation with a voltage ramp (−60 to 190 mV in 125 ms) in a near-physiological internal solutions with 2 mM Mg²⁺ and about 2 μM Ca²⁺ at pH 7.4 before (black) and after application of 100 μM FeSO₄ to the intracellular side (brown). top Fractional current change with Fe²⁺ application. (B) Similar data as in (A) with application of 100 μM CORM-S1 and 5-s illumination with blue light (450−490 nm bandpass, 100-W HBO lamp, 40× objective) (blue). Ramp currents were fit as in Figure 3 to determine mean ΔV_0.5 values: −52.6 ± 0.2 mV for 100 μM Fe²⁺; −59.7 ± 0.3 mV for Fe²⁺ released from 100 μM CORM-S1. In fluoride-buffered solutions 100 μM Fe²⁺ yielded ΔV_0.5 of −45.5 ± 0.5 mV (n = 6 for all).

CORM-S1, a Tool for Rapid Release of Fe²⁺ and CO

Due to the low solubility and stability of Fe²⁺ under ambient conditions, its physiological reactions are difficult to study. To overcome this problem, CORM-S1 may serve as a useful light-triggered Fe²⁺ releaser. Using a flash of light of about 500 μs in duration ⁴⁵ passed through a GFP excitation filter (450–490 nm), we measured the change in transmission through the bath solution viewed with a 20× objective. With 100 μM CORM-S1 and 2 mM 1,10-phenanthroline in the bath of KF solution, a single light flash resulted in a rapid decrease in transmitted light, which saturated after about 5 ms (Figure 6A, top, green traces). In similar experiments with hSlo1 BK_Ca channel-containing inside-out patches in the focus of the objective and 100 μM CORM-S1 (no phenanthroline) in the bath, BK_Ca channels were first preactivated with voltage steps to 120 mV to reach a P_open of a few %. A single light flash then caused a rapid increase in current, well characterized by a single exponential function with time constant of about 60 ms (Figure 6A, bottom). Since Fe²⁺ release from CORM-S1 was much faster, this time course therefore must reflect binding of Fe²⁺ to the channel and/or subsequent channel gating. The current-voltage relationship of BK_Ca channels determined with voltage-ramp stimulations revealed that a single light flash in 100 μM CORM-S1 solutions resulted in a ΔV_0.5 of –52 mV (Figure 6B, C), i.e. slightly more than that obtained when 100 μM Fe²⁺ was applied directly. This indicates that in the focus volume of the microscope objective a single light flash quantitatively photolyzes all CORM-S1. Figure 6C also illustrates that Fe²⁺ results in current decline at high voltages, presumably because of voltage-dependent open-channel block.

Figure 6.

Rapid release of Fe²⁺ by flash photolysis of CORM-S1. (A) Representative experiments to illustrate flash photolysis of CORM-S1 (100 μM) and its effect on BK_Ca currents in inside-out patches. The green trace (top) shows transmission (20× objective) through a 2-mM 1,10-phenanthroline solution. A light flash of about 500 μs duration, passed through a GFP filter set, was applied as indicated by the dashed vertical line. As Fe²⁺ is released, absorbance increases within a few ms. (bottom) BK_Ca currents recorded from inside-out patches with step depolarization to 120 mV from −60 mV (solution without 1,10-phenanthroline) with the pipette tip positioned in the focus of the microscope objective for the same application of flash light as above. The time course of additional channel activation was fit with a single-exponential function with time constant of 67 ms, superimposed in red. (right) Expanded time range as also indicated by the gray bar in the left panel. (B) Ramp currents before and after a flash with fractional current increase at the top. (C) Mean normalized conductance (n = 3, standard error, SEM, in shading) shows that a single flash results in a ΔV_0.5 of about −52 mV, i.e. about as much as was obtained with direct application of 100 μM Fe²⁺.

CORM-S1 can also be photolyzed using a cold light source or conventional shuttered light from a mercury lamp passed through a GFP filter set (SFigure 6). The results are qualitatively similar to those obtained with short-duration light flashes (Figure 6). However, at low light intensities or only short exposure, the kinetics of Fe²⁺ release may become rate-limiting when using BK_Ca channel activation as readout. As exemplified for BK_Ca channels, light-triggered decomposition of CORM-S1 may also be used as a sole source of CO if the released Fe²⁺ ions are readily chelated (e.g. by EGTA) (Figures 1, 2).

Despite the sensitivity of CORM-S1 to visible (blue/cyan, 450–500 nm) light, it remains stable for several days when frozen in DMSO, and even in physiological buffer under ambient temperature the compound retains its efficaciousness for several minutes (SFigure 3E). This stability of CORM-S1 is in contrast with the widely used CORM-2, which spontaneously releases its CO in DMSO, even in frozen stock solutions (SFigure 1).

The biological impact of free Fe²⁺ is difficult to investigate under ambient conditions, but Fe²⁺ is much more stable while integrated in CORM-S1. Utilizing a principle applied for the triggered release of Ca²⁺ from, e.g., DM-nitrophen,⁴⁶ the results here show that photorelease of Fe²⁺ from CORM-S1 is a preferred method for transiently supplying Fe²⁺ in biological settings.

Mechanism of BK_Ca Channel Activation by Fe²⁺

Our results clearly show that Fe²⁺, but not Fe³⁺, potently activates BK_Ca channels. For Fe²⁺ released by flash photolysis from 100 μM CORM-S1 in < 5 ms, the resulting channel activation time constant of about 60 ms (Figure 6) reflects binding of Fe²⁺ to the channel protein and/or the subsequent channel gating leading to pore opening. Channel activation is mainly brought about by an Fe²⁺-induced shift in half-maximal activation voltage by about –50 mV. As for BK_Ca channel activation by Zn²⁺,³⁰ voltage-sensor activation was not required and ΔV_0.5 was diminished by lowering pH (Figure 3B, D). Yet, the molecular determinants for the Fe²⁺-dependent channel activation clearly differ from those for protons and Zn²⁺: the histidine mutants H365A and H365A:H394A completely eliminate activation by Zn²⁺ and protons ^{32, 47} but not by Fe²⁺. The observed lower EC₅₀ for Fe²⁺ for the activation shift at low pH may therefore involve other sites or result from allosteric protein modulation. Similarly, the affinities of the Ca²⁺ sensors are modulated by voltage and the state of the channel gate.^{27, 42, 48}

Opening of BK_Ca channels is mainly enhanced by two physiological divalent cations, Ca²⁺ and Mg²⁺, working through different structural and functional mechanisms. Ca²⁺ with the ionic radius of ~1 Å binds to the divalent cation sensors located in the RCK1 and RCK2 domains with estimated affinities in the low μM range.^{40, 48} The Ca²⁺-dependent activation of the channel is independent of voltage-sensor activation.⁴¹ In contrast, Mg²⁺, with the smaller ionic radius of ~0.72 Å, is coordinated in the interdomain space between the transmembrane and the cytoplasmic gating ring domains,²⁶ and the activating effect of Mg²⁺ at mM levels requires voltage-sensor activation.⁴¹ Our mutagenesis results targeting the aforementioned three divalent cation sensors suggest that the activating effect of Fe²⁺ encompasses all three divalent cation sensor sites in a non-additive manner. However, because those mutations targeting the Mg²⁺-coordinating residues essentially eliminate the Fe²⁺ action on ΔV_0.5 (Figure 4), we postulate that Fe²⁺ primarily but not exclusively works through the so-called interdomain Mg²⁺ sensor site, yet without the strict requirement of voltage-sensor activation. The ionic radius and the dehydration enthalpy of Mg²⁺, 0.72 Å and 1926 kJ/mol, are similar to those of Fe²⁺, 0.77 Å and 1950 kJ/mol. It is perhaps not surprising that Mg²⁺ and Fe²⁺ may activate the channels in part through a common mechanism. Considering that the mM levels of Mg²⁺ are required to engage the interdomain Mg²⁺ sensor but that less than 100 μM of Fe²⁺ is sufficient, for this site, Fe²⁺ is clearly a higher-affinity ligand, by up to 100-fold. The ΔV_0.5 by 100 μM Fe²⁺ (∆V_0.5 = –50 mV) is, however, smaller than that by 10 mM Mg²⁺ (~–100 mV).⁴¹ The observed selective activation by Fe²⁺ over Fe³⁺ is probably explained by the smaller ionic radius and the greater hydration enthalpy of Fe³⁺, 0.65 Å and 4429 kJ/mol, rendering coordination of Fe³⁺ by the channel much less likely.

Our mutagenesis results collectively suggest that Fe²⁺ has potential to influence the three main Ca²⁺ and Mg²⁺ sensors. We suggest that the interdomain Mg²⁺ sensor may be the primary target of Fe²⁺. We cannot, however, exclude potential secondary effects of Fe²⁺, for example redox changes affecting channel gating because BK_Ca channels are dependent on the redox conditions in a complex manner.¹²

BK_Ca Channels are Targets of Local Fe²⁺ Concentration Increases

BK_Ca channels are multimodal targets of heme and heme degradation products. Intracellular heme has inhibitory effects on the channel at positive voltages,⁴⁹ while it promotes channel opening under hyperpolarized conditions.⁵⁰ Here we have demonstrated that Fe²⁺, which is co-released along with CO when heme oxygenase cleaves the protoporphyrin ring of heme, activates BK_Ca channels with greater efficacy than CO. Considering that HO-2 and Slo1 may be in close proximity of each other,⁵¹ Fe²⁺ generated by the action of HO-2 has potential to reach the BK_Ca channels where the activating efficacy is greater than that of CO. Local Fe²⁺, escaping scavenging by ferritin or other Fe²⁺-binding proteins, may reach the μM range. Taken the previous results and those presented here together, heme catabolism by HO-2 has potential to effect a trilateral impact on BK_Ca channels: (i) local removal of the channel modulator heme, (ii) liberation of the potential channel activator CO, whose action may depend on the presence of heme,⁵² and (iii) release of Fe²⁺ to activate the channels in part via the Mg²⁺-binding sites. Further, heme itself or to an even greater extent the liberated local Fe²⁺ may catalyze Fenton reactions such that redox processes may have an additional impact on the channel function.

Another scenario for physiologically relevant BK_Ca activation by Fe²⁺ is iron release from lysosomes. BK_Ca channels have been shown to alleviate lysosomal storage disease by providing a positive feedback loop: lysosomal Ca²⁺, locally released via TRPML1 channels, directly activated co-localized BK_Ca channels.⁵³ The resulting K⁺ influx into the lysosomes counteracts the depolarization resulting from Ca²⁺ efflux, thereby maintaining an electrochemical gradient for Ca²⁺. A similar mechanism involving Fe²⁺ instead of Ca²⁺ is likely because TRPML1 channels have been shown to release Fe²⁺ from the very reducing lysosomes.⁵⁴

The exact contributions of all above-mentioned components in the heme degradation pathway on the channel function are difficult to predict a priori because their effects strongly depend on the overall cellular redox status and the iron buffer capacity of the cytosol. However, ferrous iron needs to be considered a local messenger molecule, presumably in all cells that harbor heme oxygenases and/or specific Fe²⁺ transport systems, and target molecules, besides BK_Ca channels, of local changes in Fe²⁺ concentrations remain to be investigated.

METHODS

Spectrophotometric Measurements

Sample illumination was performed using a cold light source (20 W halogen lamp, Osram GZX4) with 2 W output power at the end of the light guide, which was placed 20 mm above a 1-ml cuvette.

The CO release from CORM-S1 was measured in a spectrophotometric assay as described previously ⁵⁵ by monitoring absorbance changes on conversion of deoxymyoglobin (Mb) to the myoglobin-CO complex (MbCO). Myoglobin (10 mM) in phosphate-buffered saline (PBS, in mM: 137 NaCl, 2.7 KCl, 10 Na₂HPO₄, 2 KH₂PO₄, pH 7.4) was converted to Mb by addition of 0.1% sodium dithionite (Na₂S₂O₄), present during all spectrophotometric measurements. The time course of MbCO formation was monitored by measuring the absorbance at either 422 nm or 540 nm every 10 s. Absorbance changes were converted into CO release based on the difference in absorbance at 540 nm of 100 μM Mb or at 422 nm of 10 μM Mb (in PBS) of CO-free solutions and MbCO (in CO-saturated PBS), corresponding to 100% CO release from 50 μM or 5 μM CORM-S1, respectively.

For spectrophotometric detection of Fe²⁺, 1 mM 1,10-phenanthroline was dissolved in KF solution, consisting of (in mM): 100 KF, 40 KCl, 10 HEPES (pH 7.4). 1,10-phenanthroline forms stable complexes with free Fe²⁺ (ferroin), which was detected by measuring the absorbance at 510 nm.⁵⁶ For Fe²⁺ release experiments, 50 μM CORM-S1 was dissolved in KF buffer containing 1 mM 1,10-phenanthroline. Time course of Fe²⁺ release from CORM-S1 was monitored by measuring the absorbance at 510 nm at intervals of 15 s with 12 s of illumination in between. For Fe²⁺ calibration, FeSO₄ was dissolved to the respective concentration in KF buffer with 1 mM 1,10-phenanthroline.

Flash Photolysis of CORM-S1

Rapid photolysis of CORM-S1 was achieved with a discharge xenon arc lamp (UV flash lamp, T.I.L.L. Photonics). Light was passed through an epifluorescence GFP filter set (BP 470/40 HE; FT 495 HE) and a 20– objective (NA 0.75, fluar). A single flash had a peak intensity of about 2.2 kW/cm² and a half-width duration of 500 μs.⁴⁵ Bandpass-filtered transmission light (BP 520/15) was measured with a photodiode (FDU photodiode with view finder, T.I.L.L. Photonics). The kinetics of Fe²⁺ release from CORM-S1 was estimated by measuring the transmission through the bath solution (about 2 mm path length) with additional 2 mM 1,10-phenanthroline in the bath because Fe²⁺ binds to phenanthroline and increases absorbance.

Chemicals

CORM-S1 (CAS 127589–49-5) was prepared as described previously ⁵⁵ and stored under nitrogen. All other chemicals were from Sigma.

Channel Constructs and Mutagenesis

Pore-forming channel subunits of human Slo1 (hSlo1 α, KCNMA1, acc. no. U11058), were subcloned in a pCI-neo-vector (Promega). Mutations were introduced by PCR-based methods. All constructs were verified by DNA sequencing.

Cell Culture

HEK293T cells (CAMR) were cultured in Dulbecco’s modified Eagle’s medium containing 45% Ham’s F12 medium (PAA) and supplemented with 10% fetal bovine serum in a 5% CO₂ incubator at 37 °C.

Cells were trypsinized, diluted with culture medium, and seeded on 12-mm glass coverslips one day prior to transfection. Electrophysiological experiments were performed 1–3 days after transfection. Cells were transfected with the respective plasmids using the Rotifect (Roth) transfection reagent. Co-transfection of a plasmid coding for CD8 was used for identification of transfected cells by means of anti-CD8-coated microbeads (Dynabeads, Deutsche Dynal GmbH).

Electrophysiological Measurements

Inside-out patch-clamp experiments were performed at 20–23 °C using an EPC-10 patch-clamp amplifier controlled via PatchMaster software (both from HEKA Elektronik). Patch pipettes fabricated from borosilicate filament glass had resistances from 1 to 3 MΩ when used with the solutions described below. Current recordings were low-pass filtered at 10 kHz and sampled at 50–100 kHz. Linear leak was subtracted using a p/4 protocol. The standard internal “EGTA” solution contained (in mM): 140 KCl, 10 EGTA, 10 HEPES (pH 7.4 with KOH). Alternatively, KF solution was used (see above). External solution (in mM): 140 KCl, 2 CaCl₂, 2 MgCl₂, 10 HEPES (pH 7.4 with KOH). For the application of chemical compounds, appropriate amounts of freshly prepared stock solutions (100 mM FeSO₄ or FeCl₃ in 0.1 N HCl and 50 mM CORM-S1 in DMSO) were added directly to the internal solution. Illumination of CORM-S1 was performed similarly to the myoglobin assay. Voltage dependence, measured with ramp protocols, was quantified by fitting the currents (I(V)) assuming a linear single-channel current and a Boltzmann function for the open probability:

$$I(V)=\frac{G_{\max }\left(V-E_{rev}\right)}{1+\exp \left\{-q_{app}\left(V-V_{0.5}\right)/kT\right\}}$$ (1)

G_max is the maximal conductance, E_rev the reversal potential, k the Boltzmann constant, T the absolute temperature, V_0.5 the half-maximal activation voltage, and q_app the apparent gating charge.

Shifts in half-maximal activation voltage (ΔV_0.5) as a function of Fe²⁺ concentration was described with a Hill equation:

$$\mathrm{\Delta }{V}_{0.5}=\mathrm{\Delta }{V}_{0.5,\max }\frac{{\left(c/E{C}_{50}\right)}^{n_H}}{1+{\left(c/E{C}_{50}\right)}^{n_H}}$$ (2)

ΔV_0.5,max is the maximally obtained shift, EC₅₀ the half-maximal effective concentration, c the concentration, and n_H the Hill coefficient.

For experiments probing competition between Fe²⁺ and Ca²⁺/Mg²⁺, a solution containing (in mM) 140 KCl, 2 MgCl₂ and 10 HEPES (pH 7.4 with KOH) was prepared from ultrapure compounds and water. Based on the voltage dependence of BK_Ca channel activation, the free Ca²⁺ concentration in these solutions was estimated to be about 2 μM.

Data Analysis

Data were analyzed with FitMaster (HEKA Elektronik) and IgorPro (WaveMetrics). Averaged data are presented as means ± standard error (SEM) (n, number of independent measurements). Averaged data were compared with a two-sided Student’s t test assuming unequal variances, followed by a Bonferroni post-hoc correction, when appropriate.

Abbreviations

CO

carbon monoxide

CORMs

carbon monoxide releasing molecules

ΔCB

deletion of D894 and D895 in hSlo1

EGTA

ethyleneglycol-bis(β-aminoethyl)-N,N,N’,N’-tetraacetic acid

GFP

green fluorescent protein

GSH

glutathione

H₂S

hydrogen sulfide

HEPES

4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid

HO

heme oxygenase

hSlo1

human Slo1

NO

nitric oxide

RCK(1/2)

regulators of conductance for K⁺