Residues at the Outer Mouth of Kir1.1 Determine K-Dependent Gating

Abstract

Three residues (E132, F127, and R128) at the outer mouth of Kir1.1b directly affected inward rectifier gating by external K, independent of pH gating. Each of the individual mutations E132Q, F127V, F127D, and R128Y changed the normal K dependence of macroscopic conductance from hyperbolic (Km = 6 ± 2 mM) to linear, up to 500 mM, without changing the hyperbolic K dependence of single-channel conductance. This suggests that E132, F127, and R128 are responsible for maximal Kir1.1b activation by external K. In addition, these same residues were also essential for recovery of Kir1.1b activity after complete removal of external K by 18-Crown-6 polyether. In contrast, charge-altering mutations at neighboring residues (E92A, E104A, D97V, or Q133E) near the outer mouth of the channel did not affect Kir1.1b recovery after chelation of external K. The collective role of E132, R128, and F127 in preventing Kir1.1b inactivation by either cytoplasmic acidification or external K removal implies that pH inactivation and the external K sensor share a common mechanism, whereby E132, R128, and F127 stabilize the Kir1.1b selectivity filter gate in an open conformation, allowing rapid recovery of channel activity after a period of external K depletion.

Introduction

Diet-dependent variations in luminal (extracellular) potassium affect renal K secretion in the thick ascending limb (TALH), connecting tubule, and cortical collecting duct (CCT) via the weak inward rectifier Kir1.1 (ROMK) that is present at the apical membrane of these nephron segments (1–4). Previous studies in both tubules and heterologous expression systems have demonstrated external K modulation of Kir1.1 both by changes in single-channel conductance (g_K) (5–7) and by K-dependent pH-gating (8–12).

Normally, internal acidification closes the Kir1.1 cytoplasmic-side gate at the bundle crossing of the inner transmembrane helices (12–14). If this occurs at reduced luminal [K], the channel enters an inactivated state that can only be reversed by raising external K (6,8,9,12,15–17) or by adding and removing extracellular blockers (Ba, Cs, TPNQ) (9). The results of these studies suggest that Kir1.1 inactivation involves a K-induced conformational change at either the selectivity filter or the outer mouth of the channel.

Regulation of channel function by external K is not unique to Kir1.1. In the strong inward rectifier Kir2.1, extracellular K modulates single-channel conductance (18–21), surface-charge screening (18), activation kinetics (22), and polyamine rectification (23,24). Potassium-dependent channel activation has also been demonstrated for Kir4.1 (25) and for Kir4.2 (26). Direct activation of Kir3.1/Kir3.4 channels by external K (22,27) has been proposed as an alternative mechanism to K-induced changes in rectification.

In Kv and KcsA potassium channels, C-type inactivation involves structural changes at either the selectivity filter or the extracellular mouth of the channel (28–36). An increase in extracellular K inhibits C-type inactivation, probably by affecting occupancy of a K binding site at the extracellular side of the channel (36–40).

In addition to the effect of external K on single-channel conductance, pH-gating, and C-type inactivation, extracellular K may also gate inward rectifier channels by directly interacting with the outer mouth of the channel. In this study, we examine this possibility for the Kir1.1b inward rectifier. Our results suggest that three specific residues at the outer mouth of Kir1.1 (E132, R128, and F127) are required for maintaining channel activity at the luminal K concentrations normally found in the renal distal nephron. These residues may stabilize a selectivity filter gate, preventing closure of the channel after reduction of external K.

Abbreviated Methods

Detailed methods are presented in the online Supporting Material.

We mutated residues near the outer mouth of Kir1.1b (E132, F127, and R128) that were likely to interact with the selectivity filter, as well as residues whose negative side chains might be expected to interact with extracellular K (E104, E92, and D97). Only those mutations that produced significant whole-cell currents above background in Xenopus laevis were investigated. Whole-cell currents and conductances were measured in intact oocytes bathed in permeant acetate buffers to control their internal pH, as previously described (41), using a two-electrode voltage clamp with 16 command pulses of 30-ms duration between −160 mV and +100 mV, centered around the resting potential. There was no dependence on external pH. Single-channel currents were sampled at 5 kHz, filtered at 900 Hz, and recorded on oocytes expressing either wt-Kir1.1b or Kir1.1b mutant channels.

Results

Dependence of single-channel conductance (g_K) and open probability (P_o) on external K

Wild-type Kir1.1b single-channel inward conductance was a hyperbolic function of external K (Fig. 1) that could be approximately described by Michaelis-Menten kinetics with K_m = 8 ± 1 mM and g^max of 51 ± 1 pS. Single-point mutations at selected Kir1.1b outer-mouth residues, E132(Q), E104(S or A), or F127(V), or mutation of negatively charged residues (E132Q, E92A, D97V, E104S)-Kir1.1b preserved the hyperbolic K dependence of single-channel conductance (Fig. 1), with relatively minor effects on K_m and g^max (Table 1). In the experiments of Fig. 1, single-channel g_K responded immediately to changes in bath [K], which could be accomplished within 10 s (see Methods).

Figure 1.

Single-channel inward conductances as a function of external (pipette) K concentration for wt-Kir1.1b, E132Q, E104S, E104A, F127V, and the quadruple mutant E132Q,E92A,D97V,E104S, obtained in cell-attached patches on oocytes with zero Ca and Mg. Fitted K_m and g^max parameters for the external K dependence are given in Table 1. N = 3–6 patches for each channel type.

Table 1.

Potassium dependence of inward single-channel conductance (Fig. 1)

Channel	gmax	Km
wt-Kir1.1b	50.7 ± 1	7.7 ± 1
E132Q	58.0 ± 1	9.7 ± 1
E104S	51.3 ± 1	5.1 ± 1
E104A	55.2 ± 1	7.7 ± 1
F127V	40.0 ± 2	33.1 ± 4
QASV	46.8 ± 2	8.9 ± 2

Inward single-channel conductance data were fit to a Michaelis-Menten saturable function: g(K) = g^max/(1 + K_m/[K]), where g^max (pS) denotes maximum inward single-channel conductance and K_m (mM) is the half-maximal [K]. QASV denotes the four-point mutant (E132Q,E92A,D97V, E104S-Kir1.1b) in which negative surface charges at the outer mouth of the channel have been replaced.

In all of the constructs studied, channel open probability (P_o) was close to 0.9 for all channels examined, consistent with previously reported P_o values for Kir1.1b (6). Elevating external (pipette) K from 10 mM to 100 mM increased P_o by only 5%, provided the pipette contained 1 mM EDTA to chelate trace divalents (Table 2). Examples of inward current records at different external (pipette) K concentrations are shown in Fig. S1, Fig. S2, Fig. S3, and Fig. S4 in the Supporting Material. External K had minimal effect on P_o, mean open time, and mean closed time, all of which were similar to wt-Kir1.1b (P_o = 0.9; open time, 20 ms; closed time, 1.2 ms). Although it was possible to record whole-cell currents from R128Y-Kir1.1b, the extremely low single-channel K conductance of this mutant (g_K < 1.5 pS in 100 mM K) prevented determination of g_K(K) for R128Y-Kir1.1b (42).

Table 2.

Potassium dependence of open probability, P_o

Channel	Po(Kext = 10 mM)	Po(Kext = 100 mM)
wt-Kir1.1b	0.83 ± 0.01	0.91 ± 0.01
E132Q	0.85 ± 0.01	0.90 ± 0.01
F127V	0.84 ± 0.03	0.88 ± 0.01

Open probability (P_o) was determined in three to five cell-attached patches on oocytes bathed in 100 mM K (2 mM Ca and 1 mM Mg) at pH 8.6 maintained with HEPES acetate buffer. The external pipette solution (K_ext) contained either 10 mM or 100 mM [K], plus 1 mM EDTA and no divalents. Currents were recorded at a patch potential of −100 mV (inside negative).

Maintaining the pH gate in an open conformation

The pH-dependent gate of Kir1.1 channels is located at the bundle crossing of inner transmembrane helices and is closed by internal acidification, with wild-type Kir1.1 channels having a pKa between 6.6 and 7 (6,12,17). We reconfirmed the wt-Kir1.1b pKa of 6.7 ± 0.01 and measured the pKa of the individual mutants used in this study during progressive acidification of individual oocytes (Fig. 2 and Table 3). For the channels indicated in Fig. 2, the pKa varied between 6.6 and 6.8, except for the double mutant K61M-E132Q whose pKa of 6.0 was significantly lower, consistent with the effect of the K61M mutation (K80M-Kir1.1a) on pKa (8). The Hill coefficients for the channels of Fig. 2 averaged 4.5 ± 0.3, except for the K61M-E132Q double mutant (Table 3).

Figure 2.

pH titrations for wt-Kir1.1b and mutants during progressive internal acidification of oocytes. Fitted titration parameters are given in Table 3. N = 4–5 oocytes for each channel type.

Table 3.

pH gating of wild-type Kir1.1b and mutant channels (Fig. 2)

Channel	pKa	Hill slope
wt-Kir1.1b	6.7 ± 0.01	5.2 ± 0.3
E132Q	6.8 ± 0.01	5.8 ± 1.0
E132Q-K61M	6.0 ± 0.1	1.2 ± 0.3
E92A	6.6 ± 0.1	3.3 ± 0.5
E104S	6.6 ± 0.1	4.5 ± 0.5
F127V	6.7 ± 0.1	4.4 ± 0.3
F127D	6.8 ± 0.1	4.1 ± 0.4
R128Y	6.8 ± 0.04	4.3 ± 1.0

Titrations were performed in external solutions of100 mM K, 2 mM Ca, and 1 mM Mg. However, similar curves were obtained in the absence of external divalents (not shown). Data for each channel type were pooled from four to five oocytes.

For the protocols discussed in the subsequent sections (except in the case of Fig. 10), oocyte internal pH was maintained at 7.8 with external permeant acetate buffers, to guarantee that the cytoplasmic-side pH gate at the bundle crossing remained in an open conformation for all channels investigated.

Figure 10.

pH inactivation of E132Q, R128Y, F127V, and F127D (Kir1.1b) mutants by internal acidification in 100 mM external K, 2 mM Ca, and 1 mM Mg. wt-Kir1.1b, K61M-Kir1.1b, and K61M-E132Q are shown for comparison.

Dependence of Kir1.1b whole-cell conductance (G_K) on external K

In wild-type Kir1.1, both whole-cell (G_K) and single-channel conductance (g_K) were hyperbolic (Michaelis-Menten) functions of external K, with similar values of K_m (6 ± 2 mM for whole-cell (Fig. 3) versus 8 ± 1 mM for single-channel (Fig. 1)). An example of raw data depicting the K dependence of whole-cell conductance is shown in Fig S5 for the wild-type Kir1.1b channel. The progressive increase in linear slope conductance results in a hyperbolic relationship between wt-Kir1.1b normalized G_K and external K (Figs. 3 and 4). Return of extracellular K from 500 mM to 1 mM K retraces the hyperbolic K dependence of wild-type G_K (Fig. 3, brown dashed curve). This is consistent with wt-Kir1.1 G_K being primarily determined by the K dependence of single-channel conductance, g_K (Fig. 1).

Figure 3.

Normalized wt-Kir1.1b (red) and E132Q (blue) whole-cell conductance as a function of extracellular K. Dashed lines indicate the return phase from 500 mM external K to 1 mM K. Whole-cell conductances were normalized to G_K in 500 mM K for each oocyte. Curves were fit to a Michaelis-Menten function and straight line was fit to a linear regression through the origin. All data were obtained with 0 Ca, 0 Mg + 1 mM EDTA at internal pH 7.8. Fitted parameters are in Table 4. N = 4–5 oocytes for each channel type.

Figure 4.

Hyperbolic (Michaelis-Menten) dependence of whole-cell conductance (G_K) on extracellular K for wt-Kir1.1b and selected mutants (of Kir1.1b). G_K(K) values were normalized to the conductance in 500 mM K for each oocyte. All data were obtained with 0 Ca, 0 Mg + 1 mM EDTA at internal pH 7.8. Fitted parameters are in Table 4. N = 4–5 oocytes for each channel type.

However, replacing the negative side chain at E132 with uncharged Gln (E132Q-Kir1.1b) altered the whole-cell G_K(K) from hyperbolic to linear (Fig. 3, blue curve). Raw data depicting the K dependence of E132Q whole-cell conductance are shown in Fig. S6. The linear K dependence of E132Q (Fig. 3, blue line) contrasts with the hyperbolic K dependence of E132Q single-channel conductance (Fig. 1). Once E132Q channels were activated by 500 mM external K, they remained activated until external K was reduced below 50 mM (Fig. 3, cyan dashed line). Both wt-Kir1.1b and E132Q-Kir1.1b exhibited very little change in open probability (P_o) between 10 mM K and 100 mM K (Table 2).

Mutations that alter whole-cell K dependence, G_K(K)

Mutation of any of three side chains near the outer mouth of Kir1.1b (E132, F127, or R128) altered the external K dependence of G_K from hyperbolic to linear (Fig. 5), although single-channel conductance remained a hyperbolic function of external K, at least for E132Q and F127V (Fig. 1). In addition, the E132Q mutation had a dominant effect on G_K(K), since it linearized the K dependence of the quadruple mutant (E132Q,E92A,D97V,E104S) even though the E92A and E104S individual mutants had a hyperbolic K dependence (Fig. 4). Since elevation of external K raised the P_o of E132Q and F127V by only 5% (Table 2), the linear G_K(K) of these mutants (Fig. 5) cannot be explained by K-dependent changes in P_o but suggests recruitment of channels from a dormant to an active state. Not all mutations near the selectivity filter altered G_K(K). For example, Kir1.1b mutants Q133E, E104S, E92A, F127Y, and F127W, near the outer mouth of the channel, had a hyperbolic K dependence similar to that of wt-Kir1.1 (Fig. 4). The K_m values for these mutants are summarized in Table 4.

Figure 5.

Whole-cell conductance as a function of external K for mutants (of Kir1.1b) showing a linear dependence of G_K on extracellular K. G_K(K) were normalized to the conductance in 500 mM K for each oocyte, and data were fit to linear functions constrained to pass through the origin. All data were obtained with 0 Ca, 0 Mg + 1 mM EDTA at internal pH 7.8. N = 4–5 oocytes for each channel type.

Table 4.

Potassium dependence of whole-cell G_K conductance (Figs. 3 and 4)

Channel	Gmax	Km
wt-Kir1.1b	0.9 ± 0.04	5.8 ± 2
wt-Kir1.1b (retn)	1.0 ± 0.04	4.6 ± 1
E132Q	NA	NA
E132Q (retn)	0.9 ± 0.05	4.5 ± 2
E104S	0.9 ± 0.04	16.6 ± 4
E92A	1.0 ± 0.03	5.3 ± 1
Q133E	1.0 ± 0.04	8.8 ± 2
F127W	1.0 ± 0.03	10.6 ± 2
F127Y	0.9 ± 0.05	10.5 ± 3

Normalized whole-cell conductance data (near the reversal potential) were fit to a Michaelis-Menten saturable function, G(K) = G^max/(1 + K_m/[K]), where G^max denotes the normalized maximal conductance and K_m (mM) is the half-maximal [K]. Values were obtained under 0 Ca, 0 Mg conditions during increasing external [K], except for return (retn), which were obtained during a subsequent decrease in external K.

The time course of K-dependent recovery from external K removal

The difference in K-dependent channel activation between E132Q, F127V, R128Y (Fig. 5), and wt-Kir1.1b (Fig. 4) could arise from differences in the interaction of external K with the outer mouth of the channel. To test this hypothesis, we compared the time course of G_K(K) recovery after partial and complete K removal. Switching the oocyte bath from 100 mM K to nominally zero K reduced (but did not eliminate) outward G_K in wt-Kir1.1b, E132Q, R128Y, and F127V (Fig. 6). Mutations at R128 were also investigated, because previous results suggested an intersubunit bridging between E132 and R128 in Kir1.1b (9).

Figure 6.

Normalized whole-cell conductance during removal and return of external K without 18-Crown-6 ether. All data were obtained with 2 mM Ca, 1 mM Mg in the bath and alkaline internal pH 7.8. Fitted parameters for the recovery phase during return of K are given in Table S1. N = 4–5 oocytes for each channel type.

Subsequent return of 100 mM external K restored 95% of the initial whole-cell conductance in both wt and mutant channels within 5 min (Fig. 6 and Table S1). The rapid phase of the G_K increase in Fig. 6 was similar for wt-Kir1.1b, E132Q, R128Y, and F127V and was consistent with the concentration-dependent increase in single-channel conductance (Fig. 1), which occurs as fast as the increase in bath [K]. The slow phase of G_K recovery (Fig. 6) was similar for wt-Kir1.1b, E132Q, and R128Y and only slightly longer for F127V (Table S1). This slow time course reflects slow reactivation of channels after return of bath K.

The biphasic time course of G_K recovery is actually more evident after complete removal of external K by 18-Crown-6 polyether in either the presence (Fig. 7) or absence (Fig. 8) of external divalents. Return of external K in the presence of Ca and Mg restored channel activity in wt-Kir1.1b, F127Y, F127W, E92A, Q133E, and E104S, although the speed of recovery varied with each of the mutants (Fig. 7, Table S2). However, return of external K after the same exposure to 18-Crown-6 polyether restored <40% of channel activity in E132Q, R128Y, F127V, and F127D (Fig. 7). These were the same four mutants that showed a linear dependence of channel activity on external K (Fig. 5). Raw data used to construct the time courses of wt-Kir1.1b and E132Q during K removal and return (Fig. 7) are shown in Fig. S7 and Fig. S8, respectively.

Figure 7.

Normalized whole-cell conductance during external K removal with 18-Crown-6 ether, followed by K return. Data were obtained on 5–7 oocytes/channel type, with 2 mM Ca, 1 mM Mg in the bath and alkaline internal pH 7.8. Fitted parameters for the recovery phase are given in Table S2.

Figure 8.

Normalized whole-cell conductance during external K removal with 18-Crown-6 ether, followed by K return. Data were obtained on five to seven oocytes/channel type, with 0 Ca, 0 Mg + 1 mM EDTA in the bath and alkaline internal pH 7.8. Fitted parameters for the recovery phase are given in Table S3.

Both E132Q and the double mutant K61M-E132Q displayed the same incomplete (<40%) recovery of channel activity after K chelation by 18-Crown-6 polyether (Fig. 7). This confirms that the effect of complete K removal is not mediated by pH inactivation, since the K61M mutation removes pH inactivation (12,17). An additional point elucidated in Fig. 7 is that although a small (F127V) or a negatively charged (F127D) side chain inhibited channel recovery, conservative replacement of Phe at 127 by other aromatics (Tyr and Trp) allowed complete recovery of activity after K removal.

In the experiments of Figs. 6 and 7, the external media contained 2 mM Ca and 1 mM Mg. Previous results indicated that extracellular Ca interfered with K access to the outer mouth of Kir1.1 and impeded recovery from inactivation (10). Consequently, we surmised that E132Q, R128Y, and possibly F127V or F127D, would show greater recovery of channel activity in the absence of external divalents. This was indeed the case for E132Q and R128Y, which recovered >90% of their channel activity in 0 Ca, 0 Mg (Fig. 8 and Table S3), but not for F127V and F127D, which still recovered <40% even in divalent-free media (Fig. 8, magenta and black curves). However, F127V and F127D did recover most of their initial channel activity after a 5-min exposure to 400 mM external K, implying that the recovery process for these mutants depended on external K (Fig. 9).

Figure 9.

High (400 mM) external K rescues mutant channel activity in 0 Ca, 0 Mg + 1 mM EDTA after chelation of external K by 18-Crown-6 ether at alkaline internal pH 7.8. N = 5–7 oocytes/channel type.

Previously, we reported that salt-bridging between the P-loop residues E132 and R128 in Kir1.1b stabilized the active state of the channel during intracellular acidification, thereby preventing Kir1.1 inactivation during pH gate closure of the channel (9). In this study, we report that each of the four mutants E132Q, R128Y, F127V, and F127D (Kir1.1b) inactivate after internal acidification in 100 mM external K, 2 mM Ca, and 1 mM Mg (Fig. 10). This result is analogous to that represented in Fig. 7, in which these same mutants (E132Q, R128Y, F127V, and F127D) failed to recover their original channel activity after depletion of external K, even though the cytoplasmic-side pH gate remained open throughout the K cycle. The similarity between these two observations suggests that external K depletion alone can mimic the pH inactivation of E132Q, R128Y, F127V, and F127D, most likely by closure of a putative gate at the selectivity filter.

In contrast to the dominant effect of E132Q on K61M-E132Q recovery from K removal (Fig. 7), the K61M mutation in the E132 background prevents pH inactivation (Fig. 10), consistent with previous reports on the effect of K80M-Kir1.1a (8).

Discussion

Kir1.1 is modulated in three ways by external K. First of all, low extracellular K inactivates Kir1.1 by preventing the cytoplasmic-side pH gate from reopening after an internal acidification (6,11,15,17). This is similar to the K dependence of C-type inactivation seen with KcsA and Shaker K_v channels (29,30). Second, external K modulates single-channel Kir1.1 g_K in a manner consistent with saturation of K permeation, resulting in a hyperbolic K dependence of g_K (Fig. 1).

In this study, we describe a third effect of external K on Kir1.1 channel activity that is independent of pH gating and appears to involve direct modulation of Kir1.1 by extracellular K. This unilateral effect of external K on channel activity (NP_o) can be quantified by the simple relation NP_o(K) = G_K(K)/g_K(K). If both whole-cell (G_K) and single-channel (g_K) conductances have a similar hyperbolic dependence on external K (namely, (1/(1 + K_m/[K])), NP_o will be approximately unchanged (equal to G_max/g_max) as [K] is varied. This was the case for wt-Kir1.1b and the single-point mutants E104S, E92A, Q133E, F127W, and F127Y (Fig. 4). For these channel types, the K dependence of G_K and channel activity could be explained by the K dependence of single-channel conductance, g_K.

However, any of four separate mutations (E132Q, R128Y, F127V, and F127D) at the outer mouth of Kir1.1b (Fig. 11) cause G_K to become linearly dependent on external K (Fig. 5). This can be interpreted as an increased requirement for external K to maintain activity. This is most easily appreciated by considering external K concentrations >100 mM. In this range, wild-type single-channel and whole-cell conductances are both constant, implying a constant density of open channels, since P_o is essentially unchanged (Table 2). In contrast, the whole-cell conductance of E132Q, R128Y, F127V, and F127D increases with increasing [K] (>100 mM), whereas single-channel conductance and P_o are approximately constant, implying a K-induced increase in the number of open channels (N). This is consistent with wt-Kir1.1 being maximally activated at 100 mM external K, whereas the mutant channels have a low level of activity that can be increased by increasing extracellular K. This also indicates that in these mutants, channel regulation involves an external site different from the site that causes saturation of single-channel conductance (7).

Figure 11.

Homology model showing three of four wt-Kir1.1b subunits, viewed from the external face of the channel, indicating the outer mouth (space-filled) residues (E132, R128, and F127) involved in modulating gating of the channel by external K. Previous studies indicated an intersubunit bridge between E132 and R128. Small orange spheres (S1–S4) are K ions in the selectivity filter, and S₀ and S_ext are putative extracellular K ions in close association with the outer mouth of the channel. Location of PIP2 binding was inferred from the site of PIP2 binding in Kir2.2 (44).

Cell-attached recordings made on oocytes expressing wt-Kir1.1b, E132Q, or F127V indicated very little effect of extracellular K on the normally high P_o of these channels (Table 2, Fig. S1, Fig. S2, Fig. S3, and Fig. S4) and no time dependence. For wt-Kir1.1b, the K-dependent increase in G_K(K) could be explained by the dependence of single-channel conductance (g_K) on external (pipette) K. In contrast, the linear K dependence of mutants like E132Q and F127V (Fig. 5) could not be explained by either the hyperbolic increase in g_K or the 5% increase in P_o associated with a 10-fold increase in external [K]. This implies the existence of electrically undetectable channels residing in the membrane that have P_o values close to zero and can be recruited into the active state by increasing [K].

We also examined these same mutants (E132Q, R128Y, and F127D and F127V) during removal and return of extracellular K with the pH gate in an open conformation (pH_i >7.8). In contrast to wt-Kir1.1, these mutants did not recover their whole-cell conductance after extensive removal of external K by 18-Crown-6 polyether (Fig. 7). Since single-channel conductance (g_K) would be expected to respond immediately to the change in external K (Fig. 1), and the solution exchange is complete within 10 s, the fractional contribution of g_K to the increase in G_K during recovery can be estimated by G_K(t = 10 s)/G_K(t = 40 min), where t is the time after return of 100 mM K to the external solution (Fig. 7, Table S1, and Table S2). For channels showing complete, or nearly complete, recovery, g_K contributed an average of 25 ± 6% to the increase in G_K during return of 100 mM K (Fig. 7 and Table S2). This is not significantly different from the average of 27 ± 6% (Table S2) seen with those mutants (E132Q, E132Q-K61M, R128Y, F127D, and F127V) that recovered <40% of their original whole-cell conductance. In both cases, the remaining 75% increase in G_K reflects the increase in channel activity (gating) produced by elevation of external K (Fig. 7). However, it is not clear why the above mutants recover some, but not all, of their original G_K after K removal.

The same protocol in the absence of extracellular divalent cations allowed >95% recovery of E132Q and R128Y whole-cell conductance (Fig. 8 and Table S3). This effect of divalent cations is consistent with their previous effect on pH inactivation (10). The implication is that in the absence of external Ca and Mg, extracellular K binds more tightly to the outer mouth of the channel, where it facilitates recovery of channel activity. In the absence of divalents, the maximal contribution of g_K to the increase in G_K was similar for channels showing complete recovery (24 ± 3%) and mutants F127V and F127D (29 ± 4%), showing only a 40% recovery of G_K (Fig. 8 and Table S3).

The importance of F127-Kir1.1b for K-sensitivity seems to depend on having an aromatic side chain at this locus, since Tyr (F127Y) or Trp (F127W) mutants recovered in a manner similar to that of wild-type after removal of external K (Figs. 7 and 8). This could occur if the aromatic side chain at 127-Kir1.1b interacted directly with external K, increasing the affinity of the K sensor. Alternatively, it might form a complex with the other key residues adjacent to the selectivity filter to stabilize the active state.

Since divalent-dependent pH inactivation (9,10) and divalent-dependent external K-activation both involve the same outer-mouth residues (E132, R128, and F127), these two processes may share a common mechanism. In previous studies, mutation of either E132 or R128 (Kir1.1b) resulted in closure of the selectivity-filter gate by internal acidification (9). The pH inactivation of these mutants could be largely reversed by solutions with high (200 mM) K, indicating a clear involvement of external K with the pH inactivation process (9). In this study, inactivation of F127V and F127D could also be reversed by transient elevation of external K (Fig. 9).

The role of E132, R128, and F127 in inactivation together with their close proximity to the selectivity filter reinforces the idea of a gate at the selectivity filter (28–32) that not only interacts with the pH gate at the bundle crossing of inner transmembrane helices but can be independently opened and closed by external K. The three residues (E132, R128, and F127) may act in concert to render Kir1.1b resistant to external K depletion, thereby allowing the channel to remain conductive even when external K drops to extremely low levels. This could be physiologically important in maintaining a near-normal cell potential in low-K environments and in rapid recovery of K permeation after return of external K.

We surmise that each of these three residues stabilizes the active state during periods of K removal, possibly by preventing partially hydrated K ions from leaving the external S₀ and S_ext sites at the outer mouth of the channel. Mutation of even one of these outer-mouth side chains destabilizes the putative selectivity-filter gate so that either cytoplasmic pH-gate closure or external K removal inactivates the channel, causing inward and outward K permeation to approach zero.

From a physiological standpoint, the concerted action of the E132, R128, and F127 side chains guarantees that the renal K channel, Kir1.1b, remains active throughout a wide range of luminal K concentrations normally found in the distal nephron, at the same time still allowing channel closure when K is very low, minimizing urinary K loss during periods of low dietary K. This type of gating by extracellular K, at normal intracellular K, may be fundamentally different from the selectivity-filter collapse that occurs during bilateral K removal (43).