Neither Kir6 nor SUR genes contain mitochondrial target sequences, and Kir6/SUR proteins are not found in mitochondrial proteome databases or prediction engines.^{4, 5} Furthermore, immune-based methods to detect K_IR/SUR subunits in mitochondria are plagued by issues of antibody specificity⁶ and mitochondrial purity/contamination. Several of the key pharmacologic reagents used to study mK_ATP channels (e.g. the agonist diazoxide (DZX) and antagonist 5-hydroxydecanoate (5-HD)) are also known to exhibit off-target effects.^{7, 8} Targeted gene deletion in mice to identify the mK_ATP channel involved in IPC has proven futile, due to the confounding cardiovascular effects of knocking out K_IR6 and SUR genes (Kcnj8, Kcnj11, Abcc8, and Abcc9) on surface K_ATP channel function. In general, K_IR and SUR knockouts exhibit profound defects in glucose/insulin handling,^9–12 which impacts the response to IPC.¹³

A recent study using custom-made antibodies and SUR knockout mice identified shortform splice variants of SUR2 in mitochondria.¹⁴ Furthermore, recent pharmacological evidence suggests that complex II of the respiratory chain (succinate dehydrogenase) may be a regulatory component of the mK_ATP channel.^{15, 16} However, both these findings leave the identity of the K⁺ channel-forming subunit of mK_ATP unknown. In this regard, mK_ATP is similar to other mitochondrial ion channels which exist at a phenomenological level but have not been molecularly identified (e.g., the mitochondrial Ca²⁺ uniporter).

A major obstacle in studying the mK_ATP channel has been the availability of a reliable assay. Most studies to date have utilized an isolated mitochondrial rapid swelling assay, in which K⁺ uptake into mitochondria is followed by osmotically-obliged water, leading to mild swelling that is assayed as light scattering in a spectrophotometer.^{17, 18} This assay has been criticized as irreproducible by some laboratories,¹⁹ with the precise timing of mitochondrial isolation appearing to be a critical factor.²⁰

A novel fluorescence-based Tl⁺ flux assay for mK_ATP activity was also developed. The ionic radii of Tl⁺ (0.154 nm) and K⁺ (0.144 nm) are similar,²⁷ and thus Tl⁺ is widely used as an analog to study membrane K⁺ transport.^{21, 27–30} The assay made use of the fluorescent indicator BTC-AM, which is better known as a ratiometric Ca²⁺ sensor, but is also sensitive to Tl⁺ with a distinct spectral response preventing signal overlap between these sensitivities. Mitochondria were loaded with BTC-AM during the isolation procedure and stored on ice until use. In the assay, 0.3 mg BTC-AM loaded mitochondria were added to a rapidly stirred cuvet containing 2 ml of chloride-free Tl⁺ assay buffer at 37°C. Tested compounds were present from the beginning of the assay, and baseline fluorescence was recorded for 10 s. prior to addition of TlSO₄ (2 mmol/L final) via a syringe port. Fluorescence was monitored in a Varian Cary Eclipse spectrofluorometer (λ_ex=488nm, λ_em525nm) and normalized to baseline. Full details including the concentrations and preparation methods for all reagents used in the assay, are in the online supplement.

Isolated rat heart perfusions (Langendorff) were performed as previously described.¹⁶ Following 20 min. equilibration, hearts were divided into 7 groups: (i) IR alone, comprising 20 min. vehicle (water or DMSO) infusion, 30 s. wash-out, 25 min. global ischemia, 120 min. reperfusion; (ii) FLX + IR, comprising 20 min. FLX infusion (5 μmol/L), 30 s. wash-out, then IR; (iii) IPC + IR, comprising 3 × 5 min. ischemia interspersed with 5 min. reperfusion, then IR; (iv) FLX + IPC + IR, comprising 5 min. FLX infusion (5 μmol/L), plus FLX infused throughout the 3 reperfusion phases of IPC (i.e. 20 min. total FLX delivery), 30 s. wash-out, then IR; (v) Zimelidine + IPC + IR. As above, replacing FLX with zimelidine (5 μmol/L). (vi) FCCP + IR, comprising 20 min. FCCP infusion (30 nmol/L ^{31, 32}), then IR; vii) FCCP + FLX + IR, comprising 20 min. infusion of both FCCP (30 nmol/L) and FLX (5 μmol/L), then IR. Following reperfusion hearts were stained with tetrazolium chloride (TTC), imaged, and infarct size measured as previously described.¹⁶

Figure 1A shows the addition of Tl⁺ to BTC-AM loaded mitochondria resulted in increased fluorescence due to rapid Tl⁺ influx and the establishment of a new steady-state. The fluorescence increase was largely inhibited by ATP, consistent with Tl⁺ transport by a K_ATP channel. Furthermore, the effect of ATP could be overridden by the mK_ATP channel opener AA5, and this effect was in-turn blocked by the mK_ATP antagonist 5-HD. These data are quantified in Figure 1B, which also shows the effects of mK_ATP reagents DZX (agonist) and glyburide (antagonist). The ionophores valinomycin and nonactin, both of which transport Tl⁺,³³ resulted in maximal Tl⁺ flux into mitochondria, and the mitochondrial uncoupler FCCP inhibited Tl⁺ uptake indicating a requirement for membrane potential. It was hypothesized that the steady state is likely due to balancing of Tl⁺ influx by its efflux through the K⁺/H⁺ exchanger (KHE). However, attempts to modulate KHE activity with the inhibitors DCCD and quinine were inconclusive (data not shown). Validation of the Tl⁺ assay for mK_ATP channel activity was also performed by a direct comparison with results from mK_ATP osmotic swelling assays run in parallel under a variety of open/closed conditions. Figure 1C shows that the 2 assays correlated well (r² = 0.855).

A general property of K_IR channels is their tendency to “run-down” over time, a phenomenon attributed to loss of the phospholipid PIP₂ from a binding site on the channel.³⁴ The mK_ATP channel (which is constitutively open in isolated mitochondria) also loses activity following mitochondrial isolation, which may underlie the reported poor reproducibility of mK_ATP channel activity measurements.¹⁹ Upon investigating the relationship between these phenomena, it was found that incubation of mitochondria on ice for 5 hrs. resulted in complete loss of mK_ATP channel activity, and that PIP₂ addition restored channel activity (Figure 2). Furthermore, the full pharmacologic profile of mK_ATP channel activity (i.e. inhibition by ATP, activation by AA5, and re-inhibition by 5HD) was recovered in PIP₂-treated aged mitochondria. The same concentrations of the PIP₂ breakdown products inositol triphosphate (IP₃), 1,2-dioctanoyl glycerol (DOG), or 1,2-dipalmitoyl glycerol (DPG) did not affect mK_ATP activity. The polyvalent cation neomycin, which is known to inhibit K_IR channel activity by sequestering PIP₂,³⁵ was able to reverse the mK_ATP channel-restorative effects of PIP₂. Identical results were obtained with the osmotic swelling mK_ATP channel assay (Figure S1). Overall these data suggest that the mK_ATP channel contains a PIP₂ sensitive subunit, possibly a K_IR channel. Consistent with this, both the Tl⁺ and swelling assays revealed that mK_ATP sensitivity to the nucleotides UDP and ATP (Figure S2) was closer to that of the K_IR6.2 than the K_IR6.1 isoform.^36–39

Several classes of K_IR channel are known to be inhibited by fluoxetine (Prozac™), an antidepressant of the selective serotonin reuptake inhibitor (SSRI) class.^{24, 25} As shown in Figure S3, K_IR6 channels (components of K_ATP channels) are an order of magnitude more sensitive to FLX than K_IR4 channels, a well-known FLX target.⁴⁰ Thus, we investigated the possibility that FLX might block mK_ATP channels. As shown in Figure 3, FLX blocked mK_ATP channel activity with an IC₅₀ of 2.3 μmol/L, while a related SSRI zimelidine (ZM) did not. Identical results were obtained with the osmotic swelling mK_ATP assay (Figure S3). Furthermore, FLX blocked AA5- or DZX-mediated opening of mK_ATP (Figure S3).

Work on the mK_ATP channel to date has relied on a variety of assays, many of which measure downstream effects of mitochondrial K⁺ uptake such as changes in respiration,⁴² matrix alkalinization,⁴² flavoprotein fluorescence,⁴³ and swelling induced light scatter.⁴² Such methods are limited by the ability of other mitochondrial phenomena (e.g. electron transport chain activity, volume changes, membrane potential) to interfere with the measured parameters. Direct measurement of mitochondrial K⁺ fluxes using the potassium-binding fluorescent indicator (PBFI) is difficult because its K_d for K⁺ of ~8 mmol/L⁴² would result in saturation at typical intramitochondrial K⁺ levels (~150 mmol/L).⁴⁴ Thus herein we chose to exploit another property of K⁺ channels, their ability to transport Tl⁺ as a surrogate for K⁺.^{21, 27} While Tl⁺ acetate has previously been used in swelling-based studies on mK_ATP⁴⁵, this study is the first application of a Tl⁺ sensitive probe, BTC-AM,^28–30 to study mK_ATP.

The kinetics of the Tl⁺ based mK_ATP channel assay are superior to those of the swelling based assay.¹⁶ Following Tl⁺ addition maximal fluorescence is attained within 2–4 s., compared to a time-lag of 20–30 s. for maximal signal intensity in the osmotic swelling assay. Unfortunately the high flux rate of Tl⁺ through K⁺ channels (~2x K⁺ flux ²¹), coupled with the relatively slow mixing time in the fluorescence cuvet, does not permit precise channel kinetics to be determined in this apparatus. Current best estimates for mK_ATP channel conductivity range from 10 to 300 pS.^{15, 46, 47}

Another barrier to investigating the mK_ATP channel has been the rapid loss of channel activity over time in isolated mitochondrial preparations.²⁰ Previous work showed that the purified mK_ATP channel runs-down in an electrophysiology setting and can be re-activated by very high concentrations of UDP.⁴⁸ However, the cause of channel activity loss in intact mitochondria was unknown, and could easily be due to proteolytic degradation. The finding herein that time-dependent mK_ATP channel inactivation in intact mitochondria can be reversed by PIP₂ indicates this is a genuine run-down phenomenon, which is a common property of K_IR channels.⁴⁹

K_ATP channels were the first channels identified to depend on phosphoinositides such as PIP₂,^{22, 23} and this is the first study to identify a mitochondrial ion channel that responds to PIP₂. Such regulation of mK_ATP channel activity by PIP₂ may have implications for the function of this channel in IPC. PIP₂ has been found in mitochondrial membranes, ⁵⁰ but its endogenous source in mitochondria is unknown. Notably, the run-down of a planar lipid bilayer reconstituted mK_ATP was reversed by ATP/Mg²⁺,⁵¹ suggesting the mitochondrial high energy phosphate pool may be important in maintaining membrane lipid phosphorylation status. However, this phenomenon required very specific experimental conditions (i.e. addition and removal of ATP/Mg²⁺ to and from different sides of the membrane in a particular order), and attempts to reproduce this in isolated mitochondria were unsuccessful (data not shown). The role of lipid kinases (e.g., PI₃K), phospholipases (e.g., PLC), and other components of the IP₃/DAG signaling pathway in regulating mK_ATP is also unknown, but the involvement of such signaling components in IPC^{52, 53} suggests a potential novel pharmacological target (i.e., mitochondrial PIP₂ turnover) to modulate preconditioning.

The Tl⁺ assay was also used to probe the response of mK_ATP to nucleotides (Figure S2). The mK_ATP EC₅₀ for UDP (~20 μmol/L) was closer to that of K_IR6.2 (~200 μmol/L) than K_IR6.1 (~4 mmol/L), and the mK_ATP IC₅₀ for ATP (~4.5 μmol/L) was also closer to that of K_IR6.2 (~15 μmol/L) than K_IR 6.1 (~350 μmol/L). While these data agree with previous studies on mK_ATP,⁴⁸ a variety of labeling, electrophysiological and genetic studies across multiple species and tissues have suggested the presence of either K_IR6.1, K_IR6.2, both, or neither in mitochondria. ^{6, 10, 54–62} An overall consensus is that mK_ATP likely contains a K_IR channel, but the definitive assignment of a particular K_IR isoform is not yet possible.

The discovery that mK_ATP activity is blocked by FLX is also consistent with the consensus that mK_ATP contains a K_IR. Fluoxetine has previously been shown to inhibit K_IR channels, while related SSRIs (e.g. zimelidine) had no effect.^{24, 25} Our data (Figure S3) suggest that K_IR6 channels may be the most sensitive to FLX of all K_IR isoforms,^{24, 25, 40, 63} and in agreement with this the mK_ATP exhibits a strikingly low FLX IC₅₀ of 2.4 μmol/L (Figure 3). Physiological concentrations of FLX are in the range of 1 – 20 μmol/L.⁶³ The fact that FLX is a lipophilic cation (LogP 4.8),⁶⁴ coupled with the highly membranous nature of mitochondria, may serve to concentrate FLX in the organelle. In a mitochondria-rich tissue such as myocardium, the mitochondrion may be a primary target for FLX.

The discovery that FLX can block IPC-mediated cardioprotection is both consistent with its effect on mK_ATP activity, and consistent with a critical role for mK_ATP in IPC signaling.^{1, 16, 65} The lack of effect of another SSRI, zimelidine, on either IPC or mK_ATP activity suggests that this effect is not mediated via the SSRI mode of action. The observation that FLX also blocks mK_ATP channel opening by the highly specific agonist AA5 also suggests a direct mK_ATP effect. Furthermore, the lack of effect of FLX on FCCP-mediated cardioprotection, which is completely independent of mK_ATP channels,^{31, 32} suggests that the protection-blocking effect of FLX is specific to mK_ATP channel-mediated protection, and does not extend to all modes of protection. The current lack of a molecular identity for the mK_ATP does not permit decisive knock-out experiments to verify whether the effects of FLX observed in the intact heart are mediated by mK_ATP.

In the US, antidepressants are the most commonly prescribed class of medication,⁶⁶ with FLX alone prescribed >23 million times in 2008.⁶⁷ Although SSRIs are known to negatively impact the outcome of cardiac surgery,⁴¹ they are widely prescribed to patients with acute coronary syndrome.⁶⁸ Notably, while IPC elicits solid protection in animal models of IR injury, its application in humans is limited by confounding effects such as age,⁶⁹ gender,⁷⁰ diabetes,¹³ and other medications.⁷¹ To this list of medications FLX must now be added, with the implication that successful cardioprotection in humans may require FLX withdrawal. Furthermore, the mood enhancer lithium is also known to both increase cardiac PIP₂ lelvels⁷² and to induce cardioprotection,⁷³ suggesting that some of the protective effects of lithium previously attributed to GSK-3β inhibition⁷³ may be mediated via the mK_ATP channel.

In summary, we have developed herein a novel assay for the mK_ATP channel, and used this assay to reveal novel sensitivities of the channel to phosphoinositides and antidepressants. It is anticipated that this assay may find widespread use in the mK_ATP field, leading ultimately to the identification of this important channel.

1