Hydrogen sulfide inhibits Kir2 and Kir3 channels by decreasing sensitivity to the phospholipid phosphatidylinositol 4,5-bisphosphate (PIP2)

Inwardly rectifying potassium (Kir) channels establish and regulate the resting membrane potential of excitable cells in the heart, brain, and other peripheral tissues. Phosphatidylinositol 4,5-bisphosphate (PIP2) is a key direct activator of ion channels, including Kir channels. The gasotransmitter carbon monoxide has been shown to regulate Kir channel activity by altering channel–PIP2 interactions. Here, we tested in two cellular models the effects and mechanism of action of another gasotransmitter, hydrogen sulfide (H2S), thought to play a key role in cellular responses under ischemic conditions. Direct administration of sodium hydrogen sulfide as an exogenous H2S source and expression of cystathionine γ-lyase, a key enzyme that produces endogenous H2S in specific brain tissues, resulted in comparable current inhibition of several Kir2 and Kir3 channels. This effect resulted from changes in channel-gating kinetics rather than in conductance or cell-surface localization. The extent of H2S regulation depended on the strength of the channel–PIP2 interactions. H2S regulation was attenuated when channel–PIP2 interactions were strengthened and was increased when channel–PIP2 interactions were weakened by depleting PIP2 levels. These H2S effects required specific cytoplasmic cysteine residues in Kir3.2 channels. Mutation of these residues abolished H2S inhibition, and reintroduction of specific cysteine residues back into the background of the cytoplasmic cysteine-lacking mutant rescued H2S inhibition. Molecular dynamics simulation experiments provided mechanistic insights into how potential sulfhydration of specific cysteine residues could lead to changes in channel–PIP2 interactions and channel gating.

Inwardly rectifying potassium (Kir) channels establish and regulate the resting membrane potential of excitable cells in the heart, brain, and other peripheral tissues. Phosphatidylinositol 4,5-bisphosphate (PIP 2 ) is a key direct activator of ion channels, including Kir channels. The gasotransmitter carbon monoxide has been shown to regulate Kir channel activity by altering channel-PIP 2 interactions. Here, we tested in two cellular models the effects and mechanism of action of another gasotransmitter, hydrogen sulfide (H 2 S), thought to play a key role in cellular responses under ischemic conditions. Direct administration of sodium hydrogen sulfide as an exogenous H 2 S source and expression of cystathionine ␥-lyase, a key enzyme that produces endogenous H 2 S in specific brain tissues, resulted in comparable current inhibition of several Kir2 and Kir3 channels. This effect resulted from changes in channel-gating kinetics rather than in conductance or cell-surface localization. The extent of H 2 S regulation depended on the strength of the channel-PIP 2 interactions. H 2 S regulation was attenuated when channel-PIP 2 interactions were strengthened and was increased when channel-PIP 2 interactions were weakened by depleting PIP 2 levels. These H 2 S effects required specific cytoplasmic cysteine residues in Kir3.2 channels. Mutation of these residues abolished H 2 S inhibition, and reintroduction of specific cysteine residues back into the background of the cytoplasmic cysteinelacking mutant rescued H 2 S inhibition. Molecular dynamics simulation experiments provided mechanistic insights into how potential sulfhydration of specific cysteine residues could lead to changes in channel-PIP 2 interactions and channel gating.
The gaseous mediator, hydrogen sulfide (H 2 S), is increasingly garnering the reputation of a major physiological messen-ger molecule with robust effects in ischemia-related insults to the heart, brain, and other peripheral tissues (1). H 2 S is generated from L-cysteine by three distinct enzymes: cystathionine ␤-synthase, cystathionine ␥-lyase (CSE), 2 and 3-mercaptopyruvate sulfurtransferase/cysteine aminotransferase (2), and signals via sulfhydration (also known as persulfidation), a posttranslational modification of reactive cysteine residues analogous to N-nitrosylation by nitric oxide (3). In these studies, sulfhydration was detected by techniques such as a Biotinswitch assay (4), a Tag-switch assay (5), and mass spectrometry (6). In experiments in vivo and in vitro, the endogenous production of H 2 S was modulated by several chemicals and/or H 2 S was applied exogenously to tissue/cells, using sulfide salts, mainly sodium hydrosulfide (NaHS). Collectively, H 2 S has been shown to play a role in diverse physiological processes, such as cellular necrosis, apoptosis, oxidative stress, or inflammation (7). One theme emerging from these studies is that H 2 S increases the excitability of neuronal compartments (8 -12), whereas it depresses excitability of cardiac tissues (13,14), correlating to neurotoxic (12,15) and cardioprotective (16 -20) effects, respectively. Therefore, exactly how H 2 S is affecting differentially the ion channels expressed in cells that rapidly control membrane excitability has become a central question. In the cardiovascular system for example, H 2 S stimulates the activity of the ATP-sensitive Kir6 (K ATP ) channel through the direct sulfhydration of an N-terminal cysteine (Cys-43) residue, hyperpolarizing and dilating blood vessels (13).
The K ATP channel belongs to the Kir family of K ϩ channels that establish and regulate the resting membrane potential of cardiomyocytes, smooth muscle cells, pancreatic cells, neurons, and various other excitable cells (21). A wealth of electro-physiological, structural, and biochemical studies have revealed that a biological phospholipid, phosphatidylinositol 4,5-bisphosphate (PIP 2 ), interacts with Kir channels as a key direct activator (22). The cysteine residue (Cys-43) implicated in the effect of H 2 S on vascular K ATP happens to be conserved throughout the Kir subfamily and is positioned next to a positively-charged residue critical for coordinating PIP 2 binding in Kir3.2 channels. Therefore, sulfhydration of this nearby cysteine may be comparable with other post-translational modifications, such as phosphorylation of specific channel residues that directly or allosterically alter channel interactions with PIP 2 (23). The effect of H 2 S on phosphoinositide control of channel activity is thus a critical molecular theme to be clarified. More specifically, we explored the putative effect of H 2 S on cytoplasmic cysteine residues of Kir channels (both conserved and distinct).
Here, we demonstrate that the effect of H 2 S on Kir channels (Kir2 and Kir3), heterologously expressed in Xenopus laevis oocytes or CHO-K1 cells, is inhibitory, as opposed to being stimulatory on K ATP channels. We used two approaches to study the effects of H 2 S on Kir channels by exogenous application of NaHS to Kir2 and Kir3 channels or by co-expressing CSE, the principal enzyme producing H 2 S in peripheral tissues. We report that channel inhibition does not involve changes in cell-surface localization or single-channel conductance.
Instead, open-time kinetics decrease while closed-time kinetics increase. We show that H 2 S-mediated inhibition of Kir channels is dependent on the strength of channel-PIP 2 interactions by employing the voltage-clamp technique on whole oocytes as well as patch-clamp in excised membrane patches from oocytes. Using site-directed mutagenesis, we determined which cysteine residues are involved in the inhibitory effect of H 2 S on this channel. Finally, using molecular dynamics simulation experiments, we provide mechanistic insight as to how potential sulfhydration of these specific cysteine residues could lead to changes in channel-PIP 2 interactions and channel gating.

Hydrogen sulfide regulation of Kir channels
treatment up to 400 min did not elicit additional inhibitory effects (Fig. 1C).
Therefore, the effect of NaHS was observed and compared when oocyte groups were treated for 100 min in successive experiments. To explore whether the NaHS effect was direct on the Kir3.2 protein, we attempted unsuccessfully to employ the Tag-switch assay (for description see under "Experimental procedures"). To test for possible changes in cell-surface localization induced by NaHS, an HA-tagged Kir3.2 was used in a chemiluminescence assay adopted to provide high sensitivity at the single oocyte level (see "Experimental procedures") (25,26).
NaHS did not alter significantly the cell-surface localization compared with untreated controls (Fig. 1D).

NaHS inhibits Kir3.2 channel activity in CHO-K1 cells by altering the channel-gating kinetics
Although NaHS inhibition of Kir3.2* channels in Xenopus oocytes took over 100 min, the same effect was observed in mammalian cells within 300 s (Fig. 2, A-C). Thus, whole-cell patch-clamp recording of Kir3.2* channels expressed in CHO-K1 cells showed that a 300-s treatment with 100 M NaHS decreased currents by ϳ40%, from Ϫ52 Ϯ 11 to Ϫ31 Ϯ

Figure 2. Exogenous H 2 S inhibits Kir3.2 channels in mammalian cells by decreasing the open probability.
A, whole-cell currents recorded from CHO-K1 cells expressing Kir3.2* are decreased by 100 M NaHS. Recording protocols are described under "Experimental procedures." B, 300-s treatment with 100 M NaHS reduces Kir3.2* channel currents in CHO-K1 cells by ϳ40% (n ϭ 8 cells) and Kir3.2-mediated currents by ϳ60% (n ϭ 10) and assessed for statistical significance using an unpaired Student's t test; **, p Ͻ 0.01. C, NaHS reduces Kir3.2* channel activity in a time-dependent manner. Currents were elicited by test pulses to Ϫ120 mV from a holding level of Ϫ80 mV every 10 s. Treatment with 100 M decreased the current by ϳ40% within 300 s (n ϭ 8), an effect that was stable when NaHS was washed out. D, left, example images of CFP-Kir3.2 channels at the surface of CHO-K1 cells, captured by total internal reflection fluorescence microscopy. The mean surface-density of CFP-pixels was assessed every 100 s following exposure to 100 M NaHS. Right, mean surface-density of CFP-pixels decreases in a non-significant (N.S.) manner over 600 s with or without 100 M NaHS (unpaired Student's t test used to determine statistical significance between groups). E, single Kir3.2* channels were studied in cell-attached mode before and after a 300-s treatment with 100 M NaHS. The effective voltage was Ϫ120 mV. F, histograms show that the distribution of single-Kir3.2* channel amplitudes is not changed from control (upper) by a 300-s treatment with 100 M NaHS (lower). The histograms are fit with a Gaussian function (black line) to determine the mean (Ϯ) single channel amplitudes of 1.2 Ϯ 0.26 pA (ϳ30 pS) before and 1.18 Ϯ 0.25 pA (ϳ30 pS) after treatment with 100 M NaHS, based on Ͼ80,000 events per condition. G, dwell time histograms for single Kir3.2* channels show that the distribution of open times (left) is decreased by a 300-s treatment with 100 M NaHS, while the closed time is increased. Data were calculated from Ͼ80,000 events per condition. N.S., non-significant.
8.5 pA/pF (n ϭ 8, p Ͻ 0.01) at a test potential of Ϫ120 mV. NaHS also inhibited currents mediated by wildtype Kir3.2 channels by ϳ60% from Ϫ11 Ϯ 6 to Ϫ4.5 Ϯ 3 pA/pF (n ϭ 10, p Ͻ 0.01; Fig. 2B). Consistent with the notion that H 2 S is a cell-permeable signaling molecule, treatment with NaHS inhibited Kir3.2* channel currents in a time-dependent manner, with the maximal effect observed at ϳ300 s (Fig. 2C). The inhibition remained stable, even when NaHS was washed out by the control bath solution. Next, we sought to determine whether H 2 S decreased Kir3.2 currents by reducing the number of channels at the cell surface, the conductance of each channel, or the probability that a single channel would open. To assess the effect of NaHS on the number of Kir3.2 channels at the cell membrane, we expressed CFP-tagged Kir3.2 in CHO-K1 cells and monitored cyan fluorescence by total internal reflection fluorescence (TIRF) microscopy. Although the number of CFP pixels decreased by ϳ15% during a 600-s treatment with 100 M NaHS, the change could be attributed to photobleaching of the fluorophore, as it also occurred in cells that were only exposed to control solutions (Fig. 2D). Next, the conductance and open probability of Kir3.2* channels were assessed in CHO-K1 cell-attached patches. A 300-s treatment with 100 M NaHS did not change the conductance of single channels (ϳ30 pS) studied at an effective test potential of Ϫ120 mV (Fig. 2, E and F, and Table 1). However, treatment with NaHS did decrease the open probability of Kir3.2* channels by ϳ40% from 0.101 to 0.062 (Table 1). Studying the time that single channels dwell in the open or closed state showed that the decrease in open probability resulted from a diminution in the time that Kir3.2* channels spent open and an increase in the duration between channel openings ( Fig. 2G and Table 1).

NaHS inhibition of Kir3.2* channels is dependent on channel-PIP 2 interactions
A common feature of Kir channels is that they all require the membrane phospholipid PIP 2 to maintain their activity (29). The strength of Kir-PIP 2 interactions determines the sensitivity of Kir channels to several regulatory factors such as pH, protein kinase C, and G-proteins (30,31). Kir2.1 channels, with the highest apparent affinity to PIP 2 among the channels tested, were unaffected by NaHS treatment, suggesting that the extent of the NaHS effect may depend on the strength of channel-PIP 2 interactions. To test this hypothesis, we first altered cytoplasmic PIP 2 levels and monitored NaHS-mediated inhibition of Kir3.2* channels. First, NaHS-mediated inhibition of Kir3.2* channels diminished when the intracellular pool of PIP 2 was increased through the co-expression of PIP5K, a kinase that phosphorylates phosphatidylinositol 4-phosphate, into PI(4,5)P 2 (Fig. 4A). Conversely, depleting intracellular levels of PIP 2 through co-administration of wortmannin (Wort), a known blocker of PI4Ks at micromolar concentrations, permitted greater inhibition of Kir3.2* currents by NaHS (Fig. 4A). Furthermore, the NaHS-mediated inhibition was enhanced by Ci-VSP (voltage-activated PIP5 phosphatase)-dependent inhibition of Kir3.2* channels (Fig. 4B), another approach used to deplete the intracellular pool of PIP 2 . Consistent with these results, bolstering Kir3.2*-PIP 2 interactions by co-expressing the G-protein subunit, G␤␥, also attenuated NaHS inhibition of the Kir3.2* current (Fig. 4C). Next, we tested what would happen to the extent of NaHS-mediated inhibition when Kir3.2 channels themselves were altered intrinsically using known point mutations that increase channel-PIP 2 interactions. For example, the E152D point mutation in Kir3.2* was shown previously to strengthen channel-PIP 2 interactions (30), and thus it was not surprising to see that NaHS had a greater inhibitory effect on wildtype Kir3.2 homomeric channels with a weaker apparent PIP 2 affinity than Kir3.2* channels (Fig. 4D), results consistent with those obtained in the CHO-K1 cells (see Fig.  2B). Moreover, the inhibitory effect of NaHS was virtually abolished when the point mutation (I234L) in the background of Kir3.2* channels further strengthened the apparent channel-PIP 2 affinity (Fig. 4D).

NaHS inhibits Kir2.3 in macropatch recordings at micromolar concentrations
NaHS incubation of whole oocytes expressing Kir channels led to significant inhibition of various Kir2.x and Kir3.x channel activities, although it stimulated Kir6.2 activity. We proceeded to apply NaHS to the cytoplasmic side of excised inside-out patches expressing Kir channels to examine directly H 2 S effects, isolating the channel and other membrane-delimited elements from the rest of the cellular cytoplasm. We used Kir2.3, whose relative phosphoinositide-interaction profile has been characterized as moderate in previous experiments, thus yielding reasonable currents in excised patches. Kir2.3 was inhibited by NaHS in TEVC recordings (Fig. 3B). First, we tested the effect of applying NaHS at the micromolar range of

Hydrogen sulfide regulation of Kir channels
concentrations and chose to use a single (sub-maximal) concentration of soluble PIP 2 , diC8 (25 M), to compare the relative NaHS effects. Not surprisingly, Kir2.3 channel activity was inhibited (Fig. 5, A and B) by NaHS in a dose-dependent manner (IC 50 82.95 M), and channel activity was maximally inhibited by ϳ64%. Interestingly, NaHS also elicited dose-dependent inhibition of Kir2.3 when the channels were maximally activated with long chain PIP 2 (Fig. 5C), but it showed a lesser extent of inhibition (ϳ37% maximal inhibition) than when Kir2.3 channels were activated by 25 M diC8 PIP 2 (see Fig. 5, A and B), consistent with the decrease of NaHS effects when channel-PIP 2 interactions are strengthened, as seen in the whole oocyte experiments (Fig. 4, A, C, and D). Conversely, we performed dose-response measurements of PIP 2 before and after inhibition by NaHS (100 M, close to IC 50 in Fig. 5, A and B) to examine how NaHS treatment might affect channel-PIP 2 interactions. We determined that NaHS treatment to the patch (Fig. 5, D and E) right-shifted the PIP 2 dose-response curve, decreasing the apparent affinity of PIP 2 demonstrated by the increase of EC 50 from 6.20 to 9.60 M. Interestingly, direct NaHS treatment dramatically reduced the efficacy of PIP 2 , indicated by an ϳ29% reduction in maximal activity following NaHS treatment.

NaHS inhibition of Kir3.2* channels is dependent on specific cytoplasmic cysteine residues
H 2 S is a gasotransmitter that has been shown to signal via sulfhydration, a post-translational modification of reactive cysteine residues and conversion of the Cys -SH group to an -SSH group. It was previously shown that NaHS is able to sulfhydrate a specific cysteine residue, of vascular K ATP channels (Kir6.1), leading to eventual hyperpolarization and vasodilation of superior mesenteric arteries, and its substitution with serine (C43S) abolished the NaHS effects (13). This specific cysteine residue is conserved throughout Kir subfamily members and is positioned adjacent to a positively charged cytoplasmic residue shown in the Kir3.2 crystal structure to interact electrostatically with one of the negatively charged phosphate groups of PIP 2 , critical for channel gating (32). In the cytoplasmic region of the Kir3.2 channel, there are three putative cysteine residues Cys-190, Cys-221, and Cys-321 that may be sulfhydrated in addition to Cys-65, the conserved cysteine residue corresponding to Cys-43 in Kir6.1 channels (Fig. 6A). In addition to the four cytoplasmic cysteine residues, Kir3.2 channels also have two extracellular cysteine residues, Cys-134 and Cys-166, located in the extracellular loop region and involved in intersubunit disulfide Treatment with NaHS led to ϳ42% inhibition of Kir3.4* channels (0.58 Ϯ 0.20, n ϭ 13) and ϳ67% inhibition of Kir3.4/Kir3.1 channels (0.33 Ϯ 0.14, n ϭ 10). B, treatment with NaHS significantly inhibited Kir2.3 channels (0.64 Ϯ 0.27, n ϭ 17) but did not significantly alter Kir2.1 channel activity (1.01 Ϯ 0.24, n ϭ 13). C, currents from both groups of oocytes expressing Kir6.2/SUR2A or Kir6.2Delta36, in which the 36 C-terminal amino acids were removed from Kir6.2 to permit cell-surface expression in the absence of sulfonylurea receptor (SUR2A) subunits, increased ϳ6-fold (normalized current values of 6.39 Ϯ 1.82, n ϭ 12, and 6.48 Ϯ 3.98, n ϭ 9, respectively) upon incubation with 2 mM NaHS, consistent with hyperpolarizing findings in Mustafa et al. (13), in which Kir6.1 was expressed in HEK293 cells. Unpaired Student's t test was used to assess the significant differences between NaHS-untreated (control) and NaHS-treated groups. * indicates comparison to the untreated group. Single, double, and triple symbols indicate p Ͻ 0.05, p Ͻ 0.01, and p Ͻ 0.001, respectively. N.S., non-significant.

CSE co-expression mimics exogenous NaHS effects on Kir3.2* channels
CSE is an enzyme central to the production of H 2 S in cardiac tissue and specific regions of the brain (hippocampus and striatal neurons) in physiological and pathophysiological conditions (2,3), and therefore it is most likely to be co-expressed in neurons with Kir3.2 channels and Kir2.x channels (as well as Kir3.4 channels in cardiac cells). Among the three enzymes that produce H 2 S, CSE seems to be the most inducible and is mod- , enhanced H 2 S inhibition from ϳ40% to ϳ80% inhibition. B, decreasing PIP 2 though activation of co-expressed Ci-VSP, a voltage-activated PIP5 phosphatase, also reduces intracellular PIP 2 , and H 2 S is able to enhance channel inhibition (remaining currents from 0.43 Ϯ 0.09, n ϭ 8 relative current value to 0.31 Ϯ 0.08, n ϭ 7 post-NaHS treatment). C, co-expression with G␤␥, another allosteric enhancer of Kir3.2-PIP 2 interactions, also reduced the NaHS inhibitory effect on Kir3.2* channels (increased NaHS treated current values from 0.68 Ϯ 0.17 n ϭ 10 to 1.08 Ϯ 0.20, n ϭ 8). D, H 2 S effect on Kir3.2 weakens when channel-PIP 2 interactions are strengthened through point mutations. Whereas Kir3.2 (WT) channel is greatly inhibited by NaHS treatment (0.36 Ϯ 0.06), pore point mutation (E152D) in the background of Kir3.2 strengthens channel-PIP 2 interaction and diminishes Kir3.2 inhibition by H 2 S (0.62 Ϯ 0.36, n ϭ 9). Additional point mutation (I234L), in the background of the Kir3.2* (E152D) mutation, further strengthens channel-PIP 2 , completely abrogating the H 2 S inhibition (1.04 Ϯ 0.16, n ϭ 8). Unpaired Student's t test was used to assess the significant differences between NaHS-untreated (control) and NaHS-treated groups, and one-way repeated-measures ANOVA and the Holm-Sidak post-test (p Ͻ 0.05) were applied to compare the NaHS-treated groups in D. Double and triple symbols p Ͻ 0.01, and p Ͻ 0.001 respectively. † indicates comparison with NaHS treated Kir3.2 group, and † † indicates comparison with NaHS-treated Kir3.2* group. Single, double, and triple symbols indicate p Ͻ 0.05, p Ͻ 0.01, and p Ͻ 0.001, respectively. N.S., non-significant.

Hydrogen sulfide regulation of Kir channels
ulated in diverse conditions such as inflammation mediated by tumor necrosis factor and lipopolysaccharides, metabolites such as glucocorticoids and glucose, as well as dietary restriction and endoplasmic reticulum stress thereby making it an attractive therapeutic target (2, 3). Co-expression of CSE with Kir3.2* in Xenopus oocytes recapitulated the inhibition of Kir3.2* produced by exogenously applied H 2 S (NaHS), in which increasing the cRNA amount co-injected (ranging from 0.5 to 8.0 ng) resulted in a dose-dependent inhibition of Kir3.2* currents (Fig. 8A). Acute treatment with 5 mM propargylglycine (PAG) for 100 min, an inhibitor of CSE activity, reversed CSE inhibition of Kir3.2*, whereas it did not affect significantly Kir3.2* channels expressed in the absence of CSE (Fig.  8B). The Cys-less Kir3.2* channel, refractory to exogenous NaHS treatment, was also unaffected by CSE co-expression (Fig. 8C).

Molecular dynamics experiments show sulfhydration of specific cysteine residues alters channel gating and interactions with PIP 2
To gain a better mechanistic understanding about how sulfhydration of a specific cysteine residue (the functional effects demonstrated experimentally in Fig. 6B) may affect channel-PIP 2 interactions as well as channel gating, and because we could not biochemically demonstrate using the Tag-switch assay direct sulfhydration upon NaHS exposure, we turned to molecular dynamics (MD) simulations with Kir3.2 co-crystallized with PIP 2 (PDB code 3SYQ) to model the sulfhydrationinduced Kir channel inhibition (32,34). More specifically, we induced sulfhydration of three cysteine residues, Cys-65, Cys-321 (positive effects in Fig. 6B), or Cys-221 (no effect in Fig. 6B) and performed a 25-ns MD simulation run to observe whether dynamic changes occurred to (a) the G-loop gate, integral for Kir3.2 channel gating, and (b) the channel-PIP 2 interactions. First, we surveyed the minimal distances between one of the two phosphates on the PIP 2 inositol ring and the lysine residue, Lys-64, previously characterized as critical for Kir3.2-PIP 2 channel interactions (32), and we observed an increase in the distances when Cys-65 (red) or Cys-321 (gold) residues were sulfhydrated but not when Cys-221 (green) was sulfhydrated (Fig. 9A). Snapshots were taken at the times during the simulations when the distance between Lys-64 and PIP 2 was the largest. Dashed lines in the model snapshots taken from the molecular dynamics experiment detail how the Lys-64 residue moved away from the phosphate on the PIP 2 headgroup when Cys-65 or Cys-321 was sulfhydrated (Fig. 9B). Consistent with these

Hydrogen sulfide regulation of Kir channels
observations, scoring the average channel-PIP 2 channel interactions (the limit was specified as being within 0.7 nm) revealed that sulfhydration of Cys-65 or Cys-321 (and not Cys-221) residues decreased interactions between the PIP 2 and the channel ( Table 2). We then sought to determine whether sulfhydration affects channel gating, and we monitored the minimal distance of the G-loop gate between the ␣ carbons of Thr-317 on opposite Kir3.2 subunits. We observed a substantial decrease when the Cys-65 or Cys-321 (but not the Cys-221) residue was sulfhydrated, suggestive of constriction of the G-loop gate closure, whereas a minimal change was observed when Cys-221 was sulfhydrated (Fig. 9C). The change in G-loop gate conformation is highlighted by model snapshots (Fig. 9D) of the Thr-317 ␣ carbon moving from no sulfhydration (gray) to Cys-65 (red), Cys-221 (green), or Cys-321 (yellow). These snapshots reflect the same time points during the simulations that gave the largest distances between Lys-64 and PIP 2 (as was shown in Fig.  9A). In stark contrast, we observed an increase in channel-PIP 2 interactions when a homology model of Kir6.2, based on the Kir3.2 structure (PDB code 3SYQ), was subjected to sulfhydration at Cys-42 (homologous to Cys-65 in Kir3.2) (Fig. 10 and Table 2). The sulfhydration of the Cys-42 also led to increases in distance between the analogous ␣ carbons (Thr-294) in an MD simulation, suggesting that the G-loop gate opened (Fig. 10A), highlighted by the shift of Thr-294 ␣ carbons (red) away from the non-sulfhydrated Kir6.2 (gray) (Fig. 10B).

Discussion
The "dual-functional effect" exerted by hydrogen sulfide is exemplified by neuronal versus cardiovascular cells that are deprived of oxygen. H 2 S mediates ischemic damage pursuant to stroke in neuronal tissue (15), whereas it serves as a powerful cardioprotective agent upon comparable conditions. One categorical correlation that emerged in the studies highlighting this "dual-physiological effect" by the gaseous mediator is that while H 2 S depolarizes cerebral components (8 -12), it hyperpolarizes cardiomyocytes (13,14). In other words, H 2 S increases the excitability of cerebral tissue, worsening seizure-like symptoms and rendering neurons more susceptible to ischemic insult (12,15). In contrast, H 2 S depresses the excitability of the heart, in which H 2 S is cardioprotective in ischemia and reperfusion (I/R) injury models when administered either through a pre-I/R (16,18,19) or post-I/R (20) treatment regime. The link between the effect of H 2 S on tissue excitability and the clinical phenotype is especially evident when the blocking of a single ion channel, K ATP , by glibenclamide abrogated the cardioprotective effects H 2 S had on the I/R injury models (19,20). Thus, subsequent efforts have been made to identify key ion channels that contribute to the depolarizing effect of H 2 S on neuronal cells. To date, the identity of H 2 S-sensitive ion channels in specific neuronal tissues has remained elusive (8 -12). Relying on ion channel involved in intersubunit disulfide interactions, were not mutated. Highlights show location of Cys-65, and how sulfhydration (space-filling models) may directly interfere with channel-PIP 2 interactions, whereas Cys-321, when sulfhydrated, may give rise to allosteric modifications at the G-loop directly as well as indirectly affect channel-PIP 2 interactions. B, Kir3.2* Cys-less mutant activity is entirely insensitive to H 2 S. NaHS incubation is unable to inhibit Kir3.2*Cys-less mutant (0.985 Ϯ 0.28, n ϭ 23). Preliminary data reveal that the strength of Cys-less mutant-PIP 2 interaction is not significantly different from that of Kir3.2*. Re-introducing single cysteine residues to positions 65 and 321 partially restores H 2 S inhibition by ϳ18% (0.82 Ϯ 0.22, n ϭ 58) and ϳ25% (0.75 Ϯ 0.29, n ϭ 57), respectively, roughly additive to ϳ38% inhibition of Kir3.2* (0.62 Ϯ 0.21, n ϭ 75) with all four cytoplasmic cysteine residues. Re-introducing both cysteine residues to positions 65 and 321 simultaneously led to ϳ36% inhibition of Kir3.2* (0.64 Ϯ 0.15, n ϭ 34), comparable with H 2 S inhibition of Kir3.2* (ϳ38%). Re-introducing single cysteine residues to positions 190 and 221 do not significantly restore NaHS inhibition (0.91 Ϯ 0.21, n ϭ 11, and 0.94 Ϯ 0.50, n ϭ 11, respectively). Unpaired Student's t test was used to assess the significant differences between NaHS-untreated (control) and NaHS-treated groups. Nonsignificant (N.S.) and *** indicate comparisons with untreated groups, p Ͻ 0.001.

Hydrogen sulfide regulation of Kir channels
blockers with limited specificity has made it difficult to reliably correlate the effect of H 2 S on observed depolarizations in neurons to specific ion channels.
Here, we show for the first time that NaHS inhibits Kir2.3 and Kir3.x channels expressed heterologously in Xenopus oocytes and CHO-K1 cells (Figs. 1 and 2), although it activates the closely related cardiac and pancreatic subunit of the K ATP channel subtype, Kir6.2 (confirming previous experiments to show the effect H 2 S has on other Kir6.x subtypes) (Fig. 3C). The significant inhibition of Kir2.x and Kir3.x channels is surprising given their homology to Kir6.x and is consistent with the depolarizations observed in NaHS-treated native cerebral components. As with other known modulators of Kir channel activity, we find that the effect of H 2 S on Kir channels proceeds by altering the strength of channel-PIP 2 interactions. Bolstering channel-PIP 2 interactions (Fig. 4, A, C and D) reduces the effect of H 2 S on Kir channels, whereas Kir channels became more susceptible to H 2 S when channel-PIP 2 interactions were weakened (Fig. 4, A and B). These results are all consistent with our excised macropatch findings showing that exposure to NaHS decreases PIP 2 potency and efficacy in activating Kir channels. We also show here that the inhibitory effect of H 2 S on the Kir3.2* channel is mediated by two of four cytoplasmic cysteine residues, Cys-65 and Cys-321 (but not Cys-190 and Cys-221) in each of four channel subunits that form tetrameric Kir3.2 channels. The substitution of all four cytoplasmic cysteines abolished the inhibitory effect of H 2 S (Fig. 6B and Fig. S1), without altering the strength of channel-PIP 2 interactions (Fig. 7). The reintroduction of cysteines into positions 65 or 321 in the background of the Kir3.2* Cys-less mutant partially restored H 2 Smediated inhibition, and reintroduction of both cysteines fully restored the extent of H 2 S-mediated inhibition ( Fig. 6B and Fig.  S1). Our computational MD simulation experiments that modeled the structural changes upon sulfhydration of the Kir3.

Hydrogen sulfide regulation of Kir channels
and Table 2) as well as changes in the G-loop gate favoring the closed conformation of the channel (Fig. 9, C and D) providing mechanistic insight into the inhibitory effect of H 2 S.
Previous experiments have applied micromolar (5-200 M) concentrations of NaHS, within the physiological range of reported hydrogen sulfide in mammals (15), and our experiments in the mammalian cell line CHO-K1 (Fig. 2) were consistent with effects seen with micromolar concentrations. Yet, in our two-electrode voltage-clamp (whole-cell) studies in Xenopus oocytes, 10-fold higher concentrations of NaHS (2,000 M) were needed to elicit a significant response. We attribute this need for higher concentrations and extended treatment (for 100 min) to the expansive size of the Xenopus oocyte, as well as the stockpiling of ␤-globin in these cells that could serve as a major buffer for H 2 S (35). It is noteworthy that this stringent NaHS administration led to channel-specific inhibitory effects, as H 2 S was unable to affect Kir2.1 channel activity as well as the Kir3.2* (I234L) mutant, two channels displaying higher apparent affinities to PIP 2 , whereas it activated (rather than inhibited) Kir6.2 channels. Still, to ensure we are not observing toxic effects of NaHS on Kir channels, we demonstrated that the inhibitory effect of NaHS on Kir3.2* channels could be mimicked by facilitating endogenous production of hydrogen sulfide through the co-expression in Xenopus oocytes of CSE, a key producer of endogenous H 2 S in native cardiovascular and specific neuronal tissues. Perhaps the most convincing evidence that the whole-cell oocyte studies did not report an artifact was the requirement of micromolar range of NaHS concentrations when applied directly to the cytoplasmic side of Xenopus patches expressing Kir 2.3 channels (i.e. in inside-out macropatches, separating the membrane-delimited elements from the rest of the cellular cytoplasm). As expected, Kir2.3 channel activity was greatly inhibited (Fig. 5, A and B) by much lower concentrations of NaHS in a dose-dependent manner (IC 50 82.95 M), and channel activity was maximally inhibited by ϳ64%, comparable with the inhibition measured in whole oocytes incubated with high levels of NaHS (Fig. 3B). The oocyte inside-out macropatch results are also consistent with the CHO-K1 cell data in terms of the micromolar NaHS effective concentrations but are also consistent with other aspects of the results obtained in the two-electrode voltage-clamp experiments. Interestingly, maximally activating Kir3.2* channels in excised membrane macropatches with long chain PIP 2 (2 M) decreased NaHS-mediated inhibition of the current (ϳ37% maximal inhibition) compared to when Kir2.3 channels are activated by a submaximal concentration of water-soluble short-chain PIP 2 (25 M diC8 PIP 2 , Fig. 5, A and B). This finding is consistent with the diminished NaHS inhibition observed when channel-PIP 2 interactions were strengthened in the whole-oocyte experiments (Fig. 4). Also, PIP 2 efficacy and potency were weakened by NaHS treatment (Fig. 5E), as indicated by the ϳ29% reduction in maximal channel activity and a rightward shift of the diC8 PIP 2 dose-response curve, respectively, following NaHS treatment. All these electrophysiologi- Figure 8. Expression of CSE, enzymatic producer of H 2 S, mimics NaHS effects. A, co-injection of CSE, integral to the production of H 2 S in specific mammalian tissues at physiological concentrations, recapitulates inhibition of Kir3.2* produced by exogenously applied NaHS (2 mM) and assessed for statistical significance using one-way repeated-measures ANOVA and the Holm-Sidak post-test (p Ͻ 0.05). Increasing CSE mRNA co-injected into oocyte (with Kir3.2* channel) increases significantly Kir3.2* inhibition in a dose-dependent manner: 0.81 Ϯ 0.12 (n ϭ 11) at 0.5 ng injection, 0.72 Ϯ 0.14 (n ϭ 10) at 4.0 ng injection, 0.59 Ϯ 0.18 (n ϭ 11) at 8.0 ng injection. B, treatment with PAG, an inhibitor of CSE activity, reversed CSE inhibition of Kir3.2* (from 0.67 Ϯ 0.28, n ϭ 10, to 0.94 Ϯ 0.40, n ϭ 13). C, Kir3.2* Cys-less activity is refractory to CSE co-injection. Mutating all four cytoplasmic residues mentioned in Fig. 6B removes the inhibitory effect of co-expressing CSE (0.94 Ϯ 0.26, n ϭ 6) compared with that of Kir3.2* WT (0.48 Ϯ 0.27, n ϭ 10), strongly suggesting that the reduced Kir3.2* activity is not due to taxing the mRNA machinery in oocytes with a non-specific RNA load. In addition, CSE inhibition requires cytoplasmic cysteine residues as was the case with exogenously applied NaHS. Unpaired Student's t test was used to assess the significant differences between NaHS-untreated (control) and NaHS-treated groups (B) as well as between control groups and CSE co-injected oocytes. * indicates comparison with the only channel injected. † indicates comparison with co-injection with 0.5 ng of CSE. Double, and triple symbols indicate p Ͻ 0.01, and p Ͻ 0.001, respectively. N.S., non-significant.

Hydrogen sulfide regulation of Kir channels
cal findings are consistent with the MD computational simulation results, in which conformational changes are induced by sulfhydration of cysteine residues, leading to weakened channel-PIP 2 interactions and the closing of the G-loop gate ( Fig. 9). Interestingly, our computational predictions are also in agreement with the activation of Kir6.2 by H 2 S (Fig. 3C), showing that sulfhydration of Cys-42 (works in a way opposite to modification of cysteines in the Kir3.2 model) in a homology model of Kir6.2 (based on the Kir3.2 structure) leads to strengthening of channel-PIP 2 interactions and opening of the G-loop gate (Fig. 10).
We demonstrate the effect of H 2 S on Kir2.x and Kir3.x is inhibitory, in the opposite direction of the previously characterized Kir6.x channel activation despite their homology. The inhibition of Kir2.x and Kir3.x channels by H 2 S could be a critical factor in the NaHS-mediated depolarization of neuronal cells. The activation of Kir channels (Kir6.) is expected to limit the excitability of cardiomyocytes or colonic smooth muscle cells (36). Although our findings provide a potential explanation for H 2 S-mediated depolarization of neuronal cells, they do not rule out a role for effects of H 2 S on other ion channels or membrane receptors. Future studies evaluating the collective effect of H 2 S on specific native brain tissues will need to at least

Hydrogen sulfide regulation of Kir channels
address the differential expression of Kir channels (37). One explanation as to why H 2 S raises the excitability of specific tissues may be because its depolarizing influence via inhibition of Kir2.x and Kir3.x channels is simply greater than its opposing influence on Kir6.x channels (whereas the hyperpolarizing influence of H 2 S dominates in specific tissue cell types, like native cardiomyocytes and smooth muscle cells).
Here, we show that the inhibitory effect of H 2 S is PIP 2 -dependent. Thus, increasing the strength of channel-PIP 2 interactions limits H 2 S-mediated inhibition of Kir2.x and Kir3.x channels while weakening channel-PIP 2 interactions augments H 2 S inhibition. Our work identifies new physiological targets of H 2 S and offers mechanistic insights into the inhibition of Kir channels by H 2 S.

Experimental procedures
Materials diC8 forms of PI(4,5)P 2 were purchased from Echelon (Salt Lake, UT). They were dissolved in water to make 6.0 mM stock solutions, which were divided into aliquots and stored at Ϫ80°C. Further dilutions were made in bath solutions on the day of the experiment. LC arachidonyl-stearyl PI(4,5)P 2 was purchased from Avanti (Alabaster, AL). Aqueous stock and working solutions were prepared and sonicated as described (38,39). NaHS was purchased from Strem (Newburyport, MA) and dissolved in water to make 100 mM stock solutions, which were divided into aliquots and stored at Ϫ20°C. Further dilutions were made in incubation solution on the day of the experiment. DL-PAG was purchased from Sigma, was stored at 100 mM stocks, and was used at 5 mM working concentrations. Wortmannin was purchased from Tocris (Bristol UK) and dissolved in DMSO from Sigma to a stock concentration of 20 mM, stored at Ϫ80°C, and was diluted to a 5 M working concentration in experiments.

Molecular biology
A Cys-less mutant (C65V, C190S, C221S, and C321V) of Kir3.2C (33) was provided by the Slesinger Laboratory (Icahn School of Medicine at Mount Sinai, New York). Because Kir3.2C (425 aa) is a longer splice variant with an additional 11 aa at the C terminus, we generated a Cys-less Kir3.2 (414 aa) isoform by PCR. The Kir3.2(E152D) (referred to as Kir3.2*) mutation that boosts homomeric currents (Yi et al. (24)) was introduced, and the resulting Kir3.2* (Cys-less) channel was subcloned in the oocyte and mammalian expression vector pXoom, so that it may be compared with the Kir3.2* subtype. Desired point and double point mutations were introduced by the commercial QuikChange (Agilent Technologies) method. All mutations were confirmed by DNA sequencing (Genewiz). CSE constructs in pcDNA3 were cut with KpnI and EcoRI and subcloned into the pGEMHE vector.

X. laevis oocyte expression
All pGEMHE constructs were linearized using the NheI restriction enzyme, whereas pXoom constructs were linearized using the XhoI restriction enzyme and in vitro-transcribed using the mMessage mMachine kit (Ambion) kit. Complementary RNA (cRNA) concentrations were quantified by optical density. Xenopus oocytes were surgically extracted, dissociated, and defolliculated by collagenase treatment and microinjected with 50 nl of a water solution containing the desired cRNAs. The use of X. laevis frogs for this study was approved by the IACUC at Virginia Commonwealth University. All constructs used in this study were injected to achieve between 5 and 20 ng per oocyte, depending on the channel and co-injection of CSE, Ci-VSP, or PIP5K cRNA. Oocytes were incubated for 2-4 days at 18°C.

Kir3.2 surface localization assay in Xenopus oocytes
Oocytes expressing Kir3.2, in which the HA epitope was inserted in the extracellular loop between Ile-126 and Glu-127 (26), were incubated for 100 min in the presence and absence of 2 mM NaHS, in dishes sealed with Parafilm. At the end of this incubation, both groups of oocytes were transferred to OR2(ϩ) medium (5 mM KCl, 5 mM NaOH, 1 mM MgCl 2 , 1.8 mM CaCl 2 , 82.5 mM NaCl, 5 mM HEPES, adjusted to pH 7.4, 200 mosM)  (a and b). B, top view (right) showed how the Thr-294 residue moves away (gate opening) from no modification (gray) to when Cys-42 was sulfhydrated (red).

Hydrogen sulfide regulation of Kir channels
containing 1% bovine serum albumin (BSA) and 0.5 g/ml rat anti-HA high affinity antibody 3F10 (Roche Applied Science, catalog no. 11 867 423 001) and incubated for 2 h at room temperature under gentle shaking. Oocytes were then extensively washed (OR2ϩ/BSA solution), incubated with an HRP-coupled goat anti-rat secondary antibody (KPL, catalog no. 474 -1612, 2 g/ml in OR2ϩ/BSA) for 1 h at room temperature, washed as above, and transferred to OR2(ϩ) without BSA. Individual oocytes were placed in 96-well plates (one oocyte/well) and incubated at room temperature for 2 min with 120 l of SuperSignal Elisa Pico chemiluminescent substrate (Thermo Fisher Scientific, catalog no. 37070). Chemiluminescence intensity for each oocyte was quantified in a POLARstar OMEGA plate reader (BMG*LABTECH) and expressed as a percentage of the mean chemiluminescence intensity obtained for the non-NaHS-exposed oocytes. Not injected oocytes, subjected to the same treatment as above, served as a negative control. Surface labeling for un-injected oocytes was less than 0.5%.

Electrophysiology
Whole-oocyte currents were measured by conventional TEVC with a GeneClamp 500 amplifier (Axon Instruments) or TEC-03X (NPI) amplifiers. Agarose cushion microelectrodes were filled with 1.5% (w/v) agarose in 3 M KCl and were used with resistances between 0.1 and 1.0 megohms. Oocytes were held at 0 mV, and currents were assessed by 800-ms ramps from Ϫ80 to ϩ80 mV. Currents at Ϫ80 mV were recorded. Bariumsensitive basal currents were defined as the difference between the steady-state currents while perfusing a high-potassium (HK) solution or a barium solution. The HK solution contained the following: 96 mM KCl, 1 mM NaCl, 1 mM MgCl 2 , 5 mM KOH/HEPES, pH 7.4. The barium solution consisted of HK ϩ 3 mM BaCl 2 . Five to 20 oocytes from the same batch were recorded for each group, and the experiments were repeated in at least two batches. A specialized voltage protocol was designed to activate Ci-VSP submaximally when they were coinjected in several experiments, whereby Ci-VSP was activated at ϩ10 mV, and current levels were monitored by brief pulses to Ϫ80 mV. Recovery was analyzed by holding at Ϫ80 mV following a 1-min ϩ80-mV depolarization protocol that fully inhibited channel current. Data acquisition and analysis were carried out using pClamp9 (Molecular Devices) and OriginPro (Microcal) software.

NaHS treatment in TEVC experiments
Channels were expressed in X. laevis oocytes. After ϳ72 h of expression, the Vehicle group of injected oocytes was transferred to a Petri dish containing 2 mM NaCl (and recorded subsequently) added to HK solution, whereas the ϩNaHS group was transferred to a Petri dish containing 2 mM NaHS added to HK, sealed with Para film, and left to incubate for 100 min prior to recording. Barium-sensitive basal currents from both groups, 1) Vehicle and 2) ϩNaHS in the presence of HK solution, were assessed at Ϫ80 mV using two-electrode voltage clamp. Barium-sensitive currents were normalized to average basal current of vehicle. All experiments were tested in at least two batches.

Wortmannin treatment and recording in TEVC experiments
Channels were expressed in X. laevis oocytes. After ϳ72 h of expression, the Vehicle group of injected oocytes was transferred to a Petri dish containing 2 ml of HK and 5 l of DMSO. The ϩWort group of injected oocytes was transferred to a Petri dish containing 2 ml of HK and 5 l of 10 mM stock solution of Wort (in DMSO). Oocytes (5-10) from Vehicle and ϩWort oocytes were then assayed as described above under "Electrophysiology" (TEVC), 120 min following incubation. Oocytes from ϩWort dish were then transferred to solution containing same solution as the Vehicle group (ϩ 5 l of DMSO in HK). Subsequently, NaHS was added to both groups for 100 min at 2 mM working concentration (as described) and assayed through TEVC recordings.

Inside-out macropatch
The vitelline membrane was manually peeled using forceps to gain access to the oocyte plasma membrane. Pipettes were pulled using a Flaming-Brown micropipette puller and then fire-polished to give a final resistance of 0.8 to 1.2 megohms. Currents were acquired at 10 kHz and filtered at 2 kHz using pClamp software and the Axopatch 200A amplifier (Axon Instruments). Inward currents were monitored using a ramp protocol from Ϫ100 to ϩ100 mV, with the holding potential at 0 mV. Following giga-seal formation, the pipette was pulled away from the membrane to achieve an inside-out patch and perfused directly using a multibarrel gravity-driven perfusion apparatus. Pipette solution contained the following (in mM): 96 KCl, 10 HEPES, 1.8 CaCl 2 , 1 MgCl 2 , 1 NaCl, pH 7.4. Bath solution contained the following (in mM): 76 KCl, 20 NaCl, 5 EGTA, 10 HEPES, pH 7.4.

NaHS treatment in macropatch experiments
NaHS dose-response curve (diC8 PIP 2 )-Once the inside-out patch (technique described above) expressing Kir2.3 channels was isolated and once the channel activity ran down, the patch was subjected to 25 M diC8 PIP 2 until the current reached a steady state, after which the patch was subjected to NaHS treatment in pipette solution for 200 s and reactivated by diC8 PIP 2 . Subsequently, the patch was subjected to various concentrations of NaHS for a 200-s duration, and currents produced by subsequent 25 M diC8 PIP 2 pulses were recorded and compared.
NaHS dose-response curve (LC-PIP 2 )-We used the same technique as we did to generate the NaHS dose-response curve pulsed by 25 M diC8 PIP 2 , but now we applied ϳ10 M solubilized long-chain PIP 2 until currents reached steady state and then we subjected them to varying concentrations of NaHS following 200-s intervals.
diC8 PIP 2 dose-response curve-Inside-out patches expressing Kir2.3 channels were allowed to run down and were pulsed by varying diC8 PIP 2 concentrations, and then the same patch was exposed to 100 M NaHS for 200 s and pulsed by the same range of diC8 PIP 2 concentrations.

Hydrogen sulfide regulation of Kir channels
analysis, demonstrated to be mycoplasma-free by Hoechst DNA stain and agar by culture, and maintained in F12K medium supplemented with 10% FBS (ATCC). Plasmids were transfected into cells with Lipofectamine 2000 according to the manufacturer's instructions (Thermo Fisher Scientific). CHO cells for patch-clamp studies were co-transfected with eGFP for identification and studied 24 -36 h post-transfection at room temperature.

Patch-clamp recording
Whole-cell currents were recorded with an Axopatch 200B amplifier and pCLAMP software (Molecular Devices, Sunnyvale, CA) at filter and sampling rates of 5 and 25 kHz, respectively. CHO-K1 cells were bathed in a solution containing (in mM): 130 NaCl, 4 KCl, 1.2 MgCl 2 , 2 CaCl 2 , and 10 HEPES; pH was adjusted to 7.4 with NaOH. Recording electrodes were fabricated from borosilicate glass (Warner, Hamden, CT) and firepolished to have resistances of 3-4 megohms when filled with a solution containing (in mM): 130 KCl, 1 MgCl 2 , 5 EGTA, 5 K 2 ATP, and 10 HEPES; pH was adjusted to 7.4 with KOH. The whole-cell capacitance of CHO cells was 7-12 pF; series resistance was typically Ͻ5 megohms, and the voltage-error of Ͻ3 mV was not adjusted. Cells for study were selected based on eGFP fluorescence. Currents were evoked from CHO cells with 500-ms test pulses from Ϫ120 to Ϫ20 mV from a holding potential of Ϫ80 mV, with 10-mV increments. To assess the time dependence of NaHS block, currents were assessed by 500-ms test pulses to Ϫ120 mV, from a holding potential of Ϫ80 mV every 10 s.
Single Kir3.2* channels were studied in cell-attached patches from cells perfused with the bath solution described above. Recording electrodes were fabricated to a resistance of 8 -9 megohms and were also filled with bath solution. Channels were studied at ϩ40 mV, to give an effective membrane potential of ϳϪ120 mV and a driving force of ϳϪ40 mV. All the experiments were performed at room temperature, and all patch-clamp data were analyzed in pClamp (Molecular Devices) and OriginLab software (Northampton, MA).

TIRF microscopy
Single protein particle complexes at the surface of live CHO cells were identified and studied by TIRF microscopy as described previously (40). Briefly, the critical angle for TIRF was adjusted using a CellTIRF illuminator (Olympus, Waltham, MA) and a high numerical-aperture apochromat objective (ϫ150, 1.45 NA; Olympus) mounted on an automated fluorescence microscope (IX81; Olympus), controlled by Metamorph software (Molecular Devices). CFP was excited by a 445-nm laser line, and the emission was collected through a 470 -500 nm bandpass filter using a back-illuminated EM-CCD (Hamamatsu, Japan). To study the effect of NaHS on the surface-density of CFP-tagged Kir3.2 subunits, images were captured with a 400-ms exposure every 100 s. The surface density of single fluorescent pixels was determined from three to five separate 30 by 30 pixel regions of interest for five cells per group using the Analyze plugin in ImageJ.

TAG-switch assay
Presence of sulfhydration was attempted with a recently designed modified Tag-switch assay (5). First, Kir3.2* was expressed and purified from Pichia pastoris as described previously (30). We also purchased from Sigma recombinant glyceraldehyde-3-phosphate dehydrogenase (GAPDH) purified from Escherichia coli. Then, ϳ0.6 g/l of each of the two proteins was exposed to 1 mM NaHS for 16 min at room temperature to trigger sulfhydration of cysteines. Next, 10 mM methylsulfonyl benzothiazole (MSBT-A) exposure for 15 min at room temperature was used to form R-S-BT and R-S-S-BT adducts at the R-SH and R-S-SH moieties of the cysteines, respectively. Excess MSBT-A was removed by Micro Bio-Spin P-6 Gel columns. A combination nucleophile and reporter molecule, CN-biotin, was employed at 2 mM for 1 h at room temperature in a reaction that is highly preferential to R-S-S-BT over all other moieties, resulting in a biotin tag detectable on sulfhydrated cysteines only. Excess CN-biotin was removed by Micro Bio-Spin P-6 gel columns. We aimed to detect biotinylated purified proteins by Western blotting using anti-biotin antibodies. Unfortunately, regardless of H 2 S treatment, both polyclonal (Abcam catalog no. ab53496) and monoclonal (Sigma catalog no. B7653) antibiotin antibodies detected GAPDH (Sigma catalog no. SRE0024), a well-known H 2 S target that was used as a positive control. We next turned to BSA (Sigma catalog no. A7030) that has also been used as another positive control for H 2 S treatment. BSA could not be detected by the monoclonal anti-biotin antibody (even though it was detected by the polyclonal antibody), thus showing specificity for biotin over untreated BSA. Unfortunately, H 2 S treatment of BSA also failed to yield detection of sulfhydrated BSA by the specific monoclonal anti-biotin antibody. Purified Kir3.2 was exposed to 1 mM NaHS, but sulfhydration of Kir3.2 anticipating a band corresponding to ϳ47 kDa could not be detected with either anti-biotin antibody. From these negative experiments, we concluded that the Tagswitch assay in our hands lacked the needed sensitivity to detect sulfhydration of protein targets.

Preparation of MSBT
Periodic acid (1.3 g, 5.80 mmol) was dissolved in CH 3 CN (15 ml) by vigorous stirring at room temperature for 1 h. Then CrO 3 (27.5 mg, 0.27 mmol, 10 mol %) was added to the solution. The mixture was stirred at room temperature for 5 min to give a clear orange solution. The H 5 IO 6 /CrO 3 solution was then added dropwise over a period of 45 min to a solution of 2-(methylthio)benzo[D]thiazole (0.5 g, 2.76 mmol) in ethyl acetate (30 ml) at 0°C. After the addition was completed, the reaction mixture was stirred at room temperature for 12 h. The reaction was quenched by addition of saturated Na 2 SO 3 solution (2 ml) and filtered, and the solids were washed with ethyl acetate (60 ml). The filtrate was washed, respectively, with saturated aqueous Na 2 SO 3 solution and brine and then dried over anhydrous Na 2 SO 4 . The solvent was evaporated under reduced pressure. The crude reaction mixture was purified by flash chromatography (hexane/ethyl acetate) to obtain MSBT (0.470 g, 82%) as a white solid. 1 [3,4-d]imidazol-4-yl) pentanamido)ethyl 2-cyanoacetate (CN-biotin). This compound was synthesized by following the procedure reported in the literature (5), and the spectral data were matched with the reported compound.

Molecular dynamics experiments
The Kir3.2* control channel or the sulfhydrated Kir3.2* (added -SH to Cys-65, Cys-221-negative control, or Cys-321) channel with four PIP 2 molecules were subjected to MD simulations, in which GROMACS version 4.5 was used to conduct the simulation, applying the GROMOS96 53a6 force field. The topology files and charges for the PIP 2 atoms were calculated using the PRODRG web server, as described in previous work. Meanwhile, the topology file of the sulfhydrated Cys residue was generated based on existing force field parameters. The channel-PIP 2 structures were immersed in an explicit POPC bilayer using the VMD membrane package and solvated with SPC water molecules with 150 mM KCl. To mimic the activated state, we applied a constant depolarizing electric field of Ϫ0.128 V⅐nm Ϫ1 . Energy minimization was performed, followed by a 300-pS position-restrained (1000 kJ/mol/nm 2 ) MD run. Subsequently, the four systems were subjected to a 25-ns MD simulation. For analysis, the SIMULAID program was used to analyze/cluster structures and to calculate interaction networks, including hydrogen bonds, salt bridges, and hydrophobic contacts.

Statistical analysis
Data are reported as mean Ϯ S.D.; n denotes the number of oocytes. Unpaired (two-sample) t tests were performed in most statistical analyses of TEVC experiments, and normalized currents from the oocyte group exposed to NaHS were compared with that of the vehicle (untreated) group with an overall significance level of p Ͻ 0.05 using OriginPro 2016 (Microcal). Currents were normalized to the average current of vehicle (untreated) group. To determine whether there are significant changes upon exposing oocyte groups to NaHS at increasing time periods, as well as increasing RNA injection amounts of CSE, one-way repeated-measures ANOVA were performed, and pairwise comparisons were made by the Holm-Sidak method with an overall significance level of p Ͻ 0.05 in Origin-Pro 2016 (Microcal). Inhibition (ϮS.D.) of currents by NaHS in inside-out macropatches was normalized to current elicited by PIP 2 (prior to NaHS treatment), and non-linear fitting of current inhibition was done in OriginPro 2016 using the Growth/ Sigmoidal category, DoseRep function, and Levenberg Marquardt iteration function to the following Equation 1, where y is the normalized current; x is the concentration of NaHS or diC8 applied to the patch; A1 is minimum value; A2 is the maximum value; and p is the parameter for the variable hill slope.