Identification of a C-terminus domain critical for the sensitivity of Kir2.1 to cholesterol

Abstract

A variety of ion channels are regulated by cholesterol, a major lipid component of the plasma membrane whose excess is associated with multiple pathological conditions. However, the mechanism underlying cholesterol sensitivity of ion channels is unknown. We have recently shown that an increase in membrane cholesterol suppresses inwardly rectifying K⁺ (Kir2) channels that are responsible for maintaining membrane potential in a variety of cell types. Here we show that cholesterol sensitivity of Kir2 channels depends on a specific region of the C terminus of the cytosolic domain of the channel, the CD loop. Within this loop, the L222I mutation in Kir2.1 abrogates the sensitivity of the channel to cholesterol whereas a reverse mutation in the corresponding position in Kir2.3, I214L, has the opposite effect, increasing cholesterol sensitivity. Furthermore, the L222I mutation has a dominant negative effect on cholesterol sensitivity of Kir2.1 WT. Mutations of 2 additional residues in the CD loop in Kir2.1, N216D and K219Q, partially affect the sensitivity of the channel to cholesterol. Yet, whereas these mutations have been shown to affect activation of the channel by the membrane phospholipid phosphatidylinositol 4,5-bisphosphate [PI(4,5)P₂], other mutations outside the CD loop that have been previously shown to affect activation of the channel by PI(4,5)P₂ had no effect on cholesterol sensitivity. Mutations of the lipid-facing residues of the outer transmembrane helix also had no effect. These findings provide insights into the structural determinants of the sensitivity of Kir2 channels to cholesterol, and introduce the critical role of the cytosolic domain in cholesterol dependent channel regulation.

A growing number of studies have demonstrated that the level of membrane cholesterol is a major regulator of ion channel function (1, 2). The most common effect is cholesterol-induced suppression of channel activity, demonstrated for inwardly rectifying K⁺ (Kir) channels (3, 4), Ca²⁺-sensitive K⁺ channels (5), N-type Ca²⁺ channels (6), and volume-regulated anion channels (7). In contrast, epithelial Na⁺ channels and TrpC channels were shown to be inhibited by cholesterol depletion (8, 9). In general, 3 basic mechanisms have been proposed: (i) specific interaction of a channel protein with cholesterol as a boundary lipid (10, 11), (ii) hydrophobic mismatch between the channel and the lipid core of the membrane (12, 13), and (iii) indirect effects through the interactions of ion channels with regulatory lipids or proteins within the environment of cholesterol-rich membrane domains (1, 14, 15). The structural basis of the sensitivity of ion channels to cholesterol, however, is unknown. In this study, we present insights into the structural determinants of cholesterol sensitivity of Kir channels.

Our studies focus on Kir2 channels, a subfamily of inward rectifiers, that are responsible for maintaining membrane potential and K⁺ homeostasis in a variety of cell types, including heart, vascular smooth muscle, and endothelial cells (16). Down-regulation of Kir is associated with heart failure (17), prolongation of the QT interval and arrhythmia (18), and impairment of vasodilation of cerebral arteries (19). Our earlier studies have shown that Kir2 channels are strongly suppressed by the elevation of membrane cholesterol (3, 4) and by diet-induced hypercholesterolemia in endothelial cells and bone marrow-derived progenitor cells (20, 21), suggesting that cholesterol-induced suppression of Kir may significantly impair vascular function.

The first insights into the mechanism responsible for cholesterol-induced Kir suppression came from comparing the effects of cholesterol and its optical isomer, epicholesterol (3). Surprisingly, whereas cholesterol suppresses Kir channels, epicholesterol has the opposite effect, suggesting that Kir channels are regulated by specific lipid-protein interactions rather than by changes in the physical properties of membrane lipid bilayer. We hypothesized, therefore, that cholesterol may alter Kir2 activity by interfering with Kir-phosphatidylinositol 4,5-bisphosphate (PIP₂) interactions, which are well known to be critical for Kir function (22–25). Here, we identify specific Kir2.1 residues that confer cholesterol sensitivity to the channels. Our data show that cholesterol sensitivity of Kir2.1 critically depends on a subset of PIP₂-sensitive residues located within the CD loop in the C terminus cytosolic domain but is unaffected by PIP₂-sensitive residues outside of this loop or by the lipid-facing residues of the outer helix transmembrane (TM) domain. This study identifies a cluster of specific amino acids that play a key role in cholesterol sensitivity of an ion channel.

Results

Differential Effects of PIP₂-Sensitive Residues on Cholesterol Sensitivity of Kir2.1.

Earlier studies have shown that Kir2.1 channels are more sensitive to PIP₂ than Kir2.3 and that the difference in PIP₂ sensitivity depends on having a leucine (Kir2.1) at position 222 (Fig. 1A) versus an isoleucine at a corresponding position (Kir2.3) (25, 26). Kir2.1 is also more sensitive to cholesterol than Kir2.3 (4). Therefore, we tested whether the position 222 leucine-to-isoleucine substitution that decreases the sensitivity of Kir2.1 to PIP₂ also affects the sensitivity of the channels to cholesterol. The effect of Kir2.1 L222I substitution on cholesterol sensitivity of the channels was most striking: whereas Kir2.1 WT currents, as expected, were significantly enhanced by cholesterol depletion, Kir2.1-L222I currents were not affected at all (Fig. 1 B-D). This substitution also abrogated sensitivity of the channels to cholesterol enrichment [supporting information (SI) Fig. S1]. We also tested whether other PIP₂-sensitive residues that reside in the same cluster (Fig. 1A) also alter the sensitivity of Kir2.1 to cholesterol. Specifically, the charged residues K219 and R228 were substituted with a non-charged residue glutamine, the same substitutions that were shown earlier to weaken the interaction of the channels with PIP₂. Here we show that, whereas—as expected from the previous studies (23, 25, 26)—Kir2.1-K219Q and Kir2.1-R228Q generated significantly smaller currents than Kir2.1-WT, sensitivity of the channels to cholesterol was only slightly affected by the K219Q substitution and not affected at all by the R228Q mutation. In contrast, cholesterol sensitivity of Kir2.1 was significantly decreased by substituting asparagine 216 with aspartic acid (N216D). Thus, cholesterol sensitivity of Kir2.1 depends on leucine 222, asparagine 216, and lysine 219 but to progressively smaller degrees. Furthermore, we also show here that co-expression of Kir2.1-L222I with the WT at a 1:1 ratio results in the loss of cholesterol sensitivity of the channels, demonstrating that L222I mutation has a dominant-negative effect on cholesterol sensitivity of Kir2.1 WT (Fig. 1 E and F).

Fig. 1.

Differential effects of PIP₂-sensitive residues on Kir2.1 sensitivity to cholesterol. (A) Sequence of Kir WT with marked PIP₂-sensitive mutations analyzed for sensitivity to cholesterol and the homology model showing 2 opposite-facing subunits of the channel with the positions of these residues. (B) Typical current traces of Kir2.1-WT, Kir2.1-R228Q, Kir2.1-K219Q, and Kir2.1-N216D in control cells (gray) and in cells depleted of cholesterol (black). Three superimposed traces are shown for each condition. Note that, in contrast to WT, L222I currents in cholesterol-depleted cells are slightly lower than those in control cells. (C) Mean peak current densities for control and cholesterol-depleted cell populations (n = 17–43 cells per condition). (D) Ratios of mean peak current densities in cholesterol-depleted and control cells. (E) Typical traces for Kir2.1 WT, L222I, and WT co-expressed with L222I at 1:1 ratio under control and cholesterol-depleted cell populations. (F) Mean peak current densities for the same cell populations (n = 23–50 cells)

PIP₂-Sensitive Residues of a Juxtamembrane Cluster Have No Effect on Cholesterol Sensitivity of Kir2.1.

Another group of residues that affects Kir2-PIP₂ interactions has been identified in the C terminus proximal to the inner leaflet of the membrane including lysines 182, 185, and 187 (23, 25, 26). However, we show here that substitutions of these residues with glutamine (K182Q, K185Q, and K187Q), which have been previously shown to weaken Kir-PIP₂ interactions, have no effect on cholesterol sensitivity of the channels (Fig. 2). All 3 mutations resulted in a significant decrease of Kir current density, as expected. Finally, we also mutated a more distal PIP₂-sensitive residue, arginine 312, but R312Q mutant did not generate any detectable Kir current in CHO cells (not shown).

Fig. 2.

PIP₂-sensitive residues of C terminus juxtamembrane cluster have no effect on cholesterol sensitivity of Kir2.1. (A) Sequence of Kir WT with marked PIP₂-sensitive mutations at the juxtamembrane analyzed for cholesterol sensitivity and the homology model showing the positions of these residues. (B) Typical current traces of Kir2.1-WT, Kir2.1-K182Q, Kir2.1-K185Q, and Kir2.1-K187Q in control (gray) and cholesterol-depleted (black) cells. (C) Mean peak current densities for control and cholesterol-depleted cells (n = 15–40 cells). (D) Ratios of mean peak current densities in cholesterol-depleted and in control cells.

Lipid-Facing Residues of TM1 Helix Have No Effect on Cholesterol Sensitivity of Kir2.1.

Earlier studies have identified the residues that are predicted to constitute the protein-lipid interface between Kir2.1 TM domains and the plasma membrane (27). To address the role of the outer TM helix (TM1) in cholesterol sensitivity of Kir2.1, we substituted all of the predicted lipid-facing residues of this helix either with leucine or with alanine, as described earlier (Fig. 3). Consistent with the earlier studies, both TM1 mutants formed functional channels, although the currents were significantly smaller than the WT. However, neither of the 2 mutants lost their sensitivity to cholesterol, and the ratio between the currents in cholesterol depleted and in control cells remained the same as in control.

Fig. 3.

Lipid-facing residues of TM1 domain have no effect on cholesterol sensitivity of Kir2.1. (A) Sequence of Kir WT with marked lipid-facing mutations analyzed for cholesterol sensitivity and homology model showing the positions of these residues. (B) Typical current traces of Kir2.1-WT, Kir2.1-Leu, and Kir2.1-Ala in control (gray) and cholesterol-depleted (black) cells. (C) Mean peak current densities for control and cholesterol-depleted cells (n = 12–25 cells). (D) Ratios of mean peak current densities in cholesterol-depleted and in control cells.

Cholesterol-Sensitive Kir2.1 Residues Do Not Affect Surface Expression of the Channels.

As mutations that alter cholesterol sensitivity of Kir2.1 also affect Kir current densities, we tested whether these mutations alter channel expression or their ability to traffic to the plasma membrane. The channels were tagged with an extracellular HA tag that allows selective identification of the channels that are inserted into the plasma membrane, as described earlier (4, 28). The sensitivities of Kir2.1-HA and Kir2.1 without the tag to cholesterol were similar (not shown). By using 2 independent methods, fluorescent microscopy and flow cytometry, we established that none of the mutations described earlier has any effect on the surface expression of the channels (Figs. S2 and S3).

Loss of Cholesterol Sensitivity Does Not Prevent Lipid Raft Targeting of Kir2.1.

Multiple studies suggested that sensitivity of ion channels to cholesterol depends on their partitioning to cholesterol-rich membrane domains (i.e., lipid rafts) (1, 14). We have shown earlier that Kir2.1 channels partition to both raft and non-raft membrane fractions (29). Lipid rafts were also implicated in the regulation of Kir3 channels (30). In this study, we addressed the question whether the loss of the sensitivity of Kir2.1 to cholesterol can be a result of mis-targeting of the channels and failure to be incorporated into the raft domains. However, our observations show that this is not the case. Whereas the distribution of cholesterol-insensitive L222I mutant between the fractions is slightly different from that of the WT channels (Fig. 4A), the overall distribution between low-density raft and high-density non-raft membrane fractions is not affected by the L222I substitution (Fig. 4B). Partitioning of R228Q mutant to different membrane fractions was also similar to that of Kir2.1-WT. Furthermore, the impact of cholesterol depletion on the raft/non-raft distribution of L222I was similar to that of the WT channels (Fig. 4 C and D).

Fig. 4.

Partitioning of Kir 2.1-WT, Kir2.1-L222I, and Kir2.1-R228Q to low-density membrane fractions. (A) Typical immunoblots of Kir2.1-WT, L222I, and R228Q probed with mouse anti-HA antibody. The vertical lanes represent samples prepared from the sucrose gradient fractions (1–11, increasing densities, 1 mL each). (B) Densitometric analysis of Kir2.1-WT, L222I, and R228Q in low-density (1–6 fractions, black bars) and high-density (7–11 membrane fractions, gray bars) membrane fractions normalized to total band intensity for each (n = 3–7). (C) Typical immunoblots of Kir2.1-WT and L222I under control and cholesterol-depleted conditions. (D) The ratios of raft/non raft distributions for the 2 experimental conditions for Kir2.1-WT and L222I (n = 3–7).

Enhancement of Cholesterol Sensitivity of Kir2.3 by a Reverse Mutation.

As described earlier, Kir2.3 channels that have isoleucine in position 214 corresponding to position 222 in Kir2.1 are less sensitive to PIP₂ (26) and to cholesterol than Kir2.1 (4). Therefore, we hypothesized that, as the leucine-to-isoleucine substitution abrogates cholesterol sensitivity of Kir2.1, a reverse mutation in Kir2.3 ought to have the opposite effect. Here we show that this was indeed the case (Fig. 5). Consistent with our earlier observations, Kir2.3 WT was significantly less sensitive to changes in membrane cholesterol than Kir2.1: depleting membrane cholesterol enhanced Kir2.3 current 1.55 ± 0.17 fold and enriching the cells with cholesterol decreased the current 1.4 ± 0.10 fold (Kir2.1 was enhanced by 2.8 ± 0.25 fold and suppressed by 1.9 ± 0.31 fold, respectively, under the same cholesterol conditions). However, Kir2.3-I214L substitution resulted in an increase of cholesterol sensitivity of Kir2.3 channels to 1.97 ± 0.16 fold increase for cholesterol depletion and to 1.98 ± 0.13 fold decrease for cholesterol enrichment. In contrast, substituting Kir2.3 arginine 220 with glutamine (R220Q), a substitution that corresponds to cholesterol-sensitive R228Q mutation in Kir2.1, also had no impact on cholesterol sensitivity of Kir2.3 channels. These observations verify and underscore the importance and the specificity of the corresponding leucine residues for cholesterol sensitivity of Kir2 channels.

Fig. 5.

Enhancement of cholesterol sensitivity of Kir 2.3 by I214L substitution. (A) Typical current traces of Kir2.3-WT, Kir2.3-I214L, and Kir2.3-R220Q channels expressed in CHO cells under different cholesterol conditions. (B) Mean peak current densities for control, cholesterol-depleted, and cholesterol-enriched cells expressing Kir2.3 WT or mutants defined above (n = 25–45 cells). (C) Ratios of mean peak current densities in cholesterol-depleted and control cells.

Sequestering PIP₂ Does Not Affect the Sensitivity of Kir2.1/2.3 to Cholesterol.

There are several approaches to sequester PIP₂ and decrease its availability. One approach is to expose cells to neomycin (31), which, as expected, resulted in the rundown of Kir current (Fig. S4). An increase in membrane cholesterol had no effect on the current rundown, but cholesterol depletion resulted in a rundown delay. Substitution of cholesterol with its chiral analogue epicholesterol resulted in an even longer delay. However, despite the longer delay induced by epicholesterol, the effect of epicholesterol on Kir current density was lower than that of cholesterol depletion, suggesting that the 2 effects are distinct. Another approach to sequester PIP₂ is to transfect the cells with the pleckstrin homology domain of phospholipase δ1 (PH-PLCδ1) which has high affinity to PIP₂ (31). In this series of experiments, Kir 2.1 and Kir 2.3 channels were co-transfected with PH-PLCδ1 tagged with GFP. The rationale of this approach was that, if cholesterol sensitivity of Kir2 channels depends on Kir-PIP₂ interactions, then sequestering PIP₂ should decrease cholesterol sensitivity of the channels. As expected, co-expression of the channels with PH-PLCδ1 significantly decreased current density under control conditions for both Kir2.1 and Kir2.3 channels (Fig. 6A), confirming that expression of PH-PLCδ1 decreases PIP₂ availability to the channels. However, sequestering of PIP₂ had no effect on the sensitivities of Kir2.1 and Kir2.3 channels to cholesterol: in both control cells that expressed GFP only and in cells that expressed PH-PLCδ1-GFP, cholesterol depletion significantly increased the density of Kir2.1 and 2.3 currents (Fig. 6B). The ratio between the currents in cholesterol-depleted and control cells was the same in cells expressing PH-PLCδ1-GFP and in cells expressing GFP only (Fig. 6C).

Fig. 6.

Sequestering PIP₂ with PH-PLC δ1 domain has no effect on cholesterol sensitivity of Kir2.1/Kir2.3. (A) Mean peak current densities of Kir2.1 and Kir2.3 WT channels when expressed alone (with GFP) or when co-expressed with PH-PLC-δ1. (B) Mean peak current densities of Kir 2.1 and of Kir2.3 co-expressed with either GFP or PH-PLC-δ1-GFP under control or cholesterol-depleted conditions (n = 12–21). (C) Ratios of mean peak current densities in control and cholesterol-depleted cells.

Discussion

The first fundamental question in elucidating the mechanism responsible for the sensitivity of ion channels to cholesterol is whether it is conferred by their TM or cytoplasmic domains. In general, as cholesterol resides in the plasma membrane, it is logical to expect that it is the TM domains that are critical for the sensitivity of the channels to cholesterol. Indeed, 2 current models of cholesterol-ion channel interactions, a lipid belt model (10, 11) and a hydrophobic mismatch model (12, 13), both focus on the interactions of cholesterol, directly or indirectly, with the TM domains of the channels. In the lipid belt model, it is proposed that channels interact directly with the boundary lipids surrounding the TM domains (10, 11), and in the hydrophobic mismatch model, it is proposed that the hydrophobic mismatch between the TM domains and the lipid core of the membrane determines the energy required to change the conformation state of the channels from closed to open (12, 13). However, little is known about specific residues that are responsible for ion channel-cholesterol interactions. Our study provides structural insights into cholesterol sensitivity of Kir channels by identifying a specific region, the CD loop in the C terminus cytoplasmic domain, that plays a critical role in the sensitivity of Kir2.1 channels to cholesterol.

Our first hypothesis was that cholesterol may regulate Kir2.1 channels indirectly by interfering with Kir-PIP₂ interactions. This hypothesis was based on a partial overlap between PIP₂-sensitive residues and the residues in the CD loop of Kir2.1 channels that affect cholesterol sensitivity of the channels, as well as on earlier studies suggesting that sensitivity of Kir channels to diverse modulators correlates with the strength of Kir-PIP₂ interactions: the higher the affinity of the channel to PIP₂, the weaker the modulatory effects (26). However, our further observations did not support the hypothesis that interfering with Kir-PIP₂ interactions may be solely responsible for the sensitivity of Kir2 channels to cholesterol. First, with the exception of the 3 residues (L222, N216, and K219), multiple other mutations that have been shown earlier to affect PIP₂ sensitivity of the channels (arginine 228 and lysines 182, 185, and 187) had no effect at all on the sensitivity of the channels to cholesterol. Comparative analysis of the 2 effects also shows no correlation: L222I, R228Q, and K219Q were shown earlier to have very similar effects on the strength of Kir2.1-PIP₂ interactions (23, 25), whereas the impact of the K219Q mutation on cholesterol sensitivity of the channels is much weaker than that of L222I, and R228Q has no effect at all. Similarly, for the corresponding positions to 222 and 228 in Kir2.1 for Kir2.3, whereas a PIP₂-sensitive mutation I214L increased the sensitivity of the channels to cholesterol, another PIP₂-sensitive mutation R220Q had no effect. There was also no clear correlation between the effects of different cholesterol conditions on Kir2.1 current densities and on neomycin-induced current rundown. Furthermore, lack of any effect of cholesterol enrichment on current rundown suggests that changes in Kir-PIP₂ interactions are unlikely to underlie cholesterol-induced suppression of the current. Finally, lack of an effect following a decrease in PIP₂ availability on cholesterol sensitivities of Kir2.1/2.3 channels suggests further that cholesterol modulates Kir2 activity independently of PIP₂.

An alternative possibility of how cholesterol sensitivity of the channels may depend on a cytoplasmic domain is through control of channel targeting to cholesterol-rich versus cholesterol-poor membrane domains. Indeed, partitioning into lipid rafts is proposed to be a major mechanism that underlies sensitivity of ion channels to cholesterol (1, 14). Generation of cholesterol-insensitive ion channel mutants allows the testing of this hypothesis directly. It is important to note that, even though the exact nature, size, density, and molecular composition of raft domains are still controversial, numerous studies have shown that cholesterol distribution in the membrane is heterogeneous and that it is concentrated in cholesterol-rich and sphingomyelin-rich membrane domains (i.e., membrane rafts) (32, 33). Typically, membrane rafts are identified as low-density membrane fractions separated by sucrose gradient. Using this method, we have shown recently that Kir2 channels are distributed between low-density (i.e., cholesterol-rich) and high-density membrane fractions, with a significant portion (50%–60%) appearing in the low-density fractions (29). Thus, if mutations of the residues in the CD loop of the channels were to prevent/interfere with Kir2.1 targeting to cholesterol-rich membrane domains, it could lead to the loss of cholesterol sensitivity of the channels. However, our observations show that this is not the case. Despite a complete loss of cholesterol sensitivity of Kir2.1-L222I mutants, these mutants still partition into the low-density membrane fractions. These observations also indicate that partitioning into lipid rafts is not sufficient to render ion channels cholesterol-sensitive. Moreover, the observation that cholesterol depletion results in similar re-distributions of Kir2.1 WT and L222I mutant between the raft and the non-raft fractions demonstrates further that mis-targeting of the channels is not responsible for the loss of cholesterol sensitivity in the Kir2.1-L222I mutant.

Instead, we propose that the residues of the CD loop are involved in “docking” of the C terminus of Kir2.1 to the inner leaflet of the membrane and facilitating its interaction with membrane cholesterol. In this model, when a channel is in the docking configuration, it may interact with cholesterol, which in turn is proposed to stabilize the channel in a closed state. Thus, increasing membrane cholesterol results in a decrease in Kir current density. In contrast, a loss of the docking configuration prevents the C terminus from interacting with cholesterol, in which case the channels become cholesterol-insensitive. It is important to note, however, that in the framework of this hypothesis, the critical residues of the CD loop do not necessarily interact with cholesterol directly. Alternatively, it is possible that their role is to maintain the channels in a docking conformation state that allows cholesterol to bind to another part of the channel. Finally, it is also possible that residues in the CD loop facilitate the hydrophobic interaction between the TM domains of the channel and the lipid core of the plasma membrane. It is impossible to unequivocally discriminate between these possibilities in the absence of a crystal structure of the whole channel with and without cholesterol. However, our experiments provide some additional insights into the structural determinants of the sensitivity of Kir2.1 to cholesterol. First, it would seem that the most likely part of the channel to bind cholesterol directly would be the outer TM helix that faces the lipids of the plasma membrane. However, the observations showing that mutating the lipid-facing residues of the outer TM helix has no effect on cholesterol sensitivity of the channels suggest that this is not the case and that it is unlikely that cholesterol binds to these residues. Moreover, these observations also suggest that the hydrophobic interactions of these residues with the membrane lipids are also not critical for the sensitivity of the channels to cholesterol.

Another clue comes from the finding that L222I mutation has a strong dominant-negative effect on the sensitivity of the channels to cholesterol. Although several possible mechanisms may underlie this effect, the most straightforward interpretation of this observation is that more than one subunit is needed to bind a cholesterol molecule, which in turn would imply that cholesterol binds at the interface between the subunits. This interpretation would be consistent with L222 that is positioned close to the interface between the adjacent subunits to play a direct role in cholesterol binding.

Our earlier studies have shown that cholesterol-induced suppression of Kir2 channels should be attributed to a decrease in the number of active channels on the membrane. We proposed, therefore, a “silencing” hypothesis suggesting that cholesterol-Kir interactions strongly stabilize the channels in the closed state, rendering them virtually inactive, or silent (3, 4). Consistent with this hypothesis, cholesterol depletion results in a delay in neomycin-induced current rundown because stabilizing the channels in an open state is expected to stabilize Kir-PIP₂ interactions. However, it is not quite clear why stabilizing the channels in a closed state does not de-stabilize Kir-PIP₂ interactions. In the framework of the silencing hypothesis, loss of cholesterol sensitivity is expected to result in an increase in the number active channels and increase in Kir current density. This is exactly the case for Kir2.1-L222I and Kir2.1-N216D mutants. For both mutants, current densities were significantly higher than that of Kir2.1-WT channels even though the levels of surface expression of these mutants and the WT channels were similar. However, this was not the case for the third mutation that decreased the sensitivity of the channels to cholesterol: K219Q, which generated smaller currents. Smaller currents were also observed for all other mutations of charged PIP₂-sensitive residues. We therefore suggest that the loss of the electrostatic interaction with PIP₂ in these mutants is the dominant cause for the decrease in the current density, as was shown earlier (23, 25, 26). In contrast, we suggest that the dominant effects of L222I and N216D substitutions is the loss of channel-cholesterol interactions.

In summary, identification of a specific region critical for the sensitivity of Kir2 channels to cholesterol provides insight into the structural requirements of the sensitivity of ion channels to cholesterol. Furthermore, identification of a mutant that has a dominant-negative effect specific for cholesterol sensitivity of Kir2.1 channels opens numerous possibilities to investigate the role of cholesterol-induced Kir suppression in cholesterol-induced injury of different cell types, including endothelial cells, smooth muscle cells, and cardiomyocytes that express Kir2.1.

Methods

All experimental procedures are described in detail in SI Methods. Briefly, cell culture of CHO cells, cholesterol modulation, and transfection protocols are performed as described previously (4). Kir2.x constructs were co-transfected with EGFP using Lipofectamine as described before. Kir2.1 (mouse) and Kir2.3 (human) with the HA tags inserted into the extracellular domains of the channels are a gift of Carol Vandenberg (University of California, Santa Barbara, CA). Point mutations were generated by Pfu mutagenesis and confirmed by DNA sequencing. All primers are shown in Table S1. Electrophysiological recordings were performed in standard whole-cell configuration (34) on cells that were either depleted or enriched with cholesterol compared with control cells. Mean peak current densities (−97mV) were compared for different cell populations. In all of the experiments, control and cholesterol-treated cells were recorded on the same day. Immunostaining and flow cytometry were performed using mouse monoclonal anti-HA antibody. The images were acquired using a Zeiss Axiovert microscope and flow cytometry analysis was carried out by FACS analyzer. Isolation of membrane fractions was performed using a density gradient non-detergent method, as described earlier (29).

Modeling.

Homology model of Kir2.1 was created by combining the crystallographic structure of the cytosolic domain of Kir2.1 (PDB accession number 1U4F) (35) with the TM domain of the chimera of the cytosolic domain of Kir3.1 and the TM domain of KirBac1.3 (Protein Data Bank accession number 2QKS) (36). Further details of all experimental procedures are described in SI Methods.

Statistical Analysis.

All data points are shown as mean ± SEM (*P < 0.05 by Student t test). All of the experiments were performed on multiple cells in 3 to 5 independent cell populations.