INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

POTASSIUM SECRETION in the mammalian cortical collecting tubule (CCT) is primarily controlled by a mildly inward rectifying, potassium channel (SK), located at the apical membrane of CCT principal cells (5, 15). Expression cloning resulted in the putative identification of the SK channel with the ROMK family of mild inward rectifiers, consisting of the following three splice variants: ROMK1 (K_ir1.1a; see Ref. 8), ROMK2 (K_ir1.1b; see Ref. 18), and ROMK3 (K_ir1.1c; see Ref. 1). ROMK2 lacks the first 19 amino acids of ROMK1, whereas ROMK3 contains a 7-amino acid extension.

Potassium channels in the inward rectifier superfamily (ROMK, IRK, GIRK) are thought to consist of four identical subunits surrounding a central pore (7, 17). Hydropathy analysis of the primary structure of these subunits suggests a common motif, consisting of two putative membrane spanning domains (M1 and M2) separated by a relatively short loop thought to be associated with the permeation path or pore region. Both the COOH-terminal and NH₂-terminal segments of the subunits are devoid of long hydrophobic stretches and are presumed to be cytoplasmic.

Despite similarities in structure among members of the IRK superfamily, the ROMK channels display a much greater sensitivity to intracellular pH (pH_i) than do the strong inward rectifiers like IRK1 (4). Reductions in pH_i from 7.4 to 6.8 greatly decrease the open probability (P_o) of native SK channels (15), as well as ROMK channels expressed in Xenopus oocytes (10, 11, 14). This marked sensitivity of the ROMK family to pH_i may be important for K homeostasis during different metabolic states. Specifically, the clinical observation that metabolic alkalosis is often accompanied by enhanced K secretion and hypokalemia could partly be explained by augmentation of luminal CCT potassium permeability by increases in pH_i (13, 16).

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Clones. The primary structure of the ROMK2 clone (GenBank accession no. L29403) has been previously described (18). The clone was obtained from a cDNA library made from rat kidney poly(A)⁺ RNA in the plasmid vector pSPORT. To obtain RNA from the clone, plasmid DNA was purified with the Qiagen anion-exchange column system (Qiagen, Chatsworth, CA). Plasmid DNA was linearized with Not I and transcribed in vitro with T7 RNA polymerase in the presence of the GpppG cap using mMESSAGE mMACHINE kit (Ambion, Austin, TX). Synthetic RNA was dissolved in water and stored at 70°C before use. Mutations were made using the overlap extension method (9). Nucleotide sequences were checked using the dideoxy chain termination method with a Sequenase kit (US Biochemical, Cleveland, OH).

Oocytes. Stage V-VI oocytes were obtained by partial ovariectomy of female Xenopus laevis (Xenopus-I, Ann Arbor, MI), anesthetized with tricaine methanesulfonate (1.5 g/l, adjusted to pH 7.0). Oocytes used for the whole cell, two-electrode voltage clamp (TEVC) experiments were defolliculated by incubation with 2 mg/ml collagenase type II and 2 mg/ml hyaluronidase type II (Sigma Chemical, St. Louis, MO) for 60 min at 23°C and stored overnight at 19°C in modified Leibovitz L-15 medium (18). On the next day, healthy oocytes were selected for RNA injection.

Oocytes destined for patch-clamp experiments were defolliculated by incubation with 4 mg/ml collagenase type II and 4 mg/ml hyaluronidase type II (Sigma Chemical) for 60 min at 23°C and then exposed to hypertonic media (460 mosmol/kgH₂O) for 15 min. Only those oocytes exhibiting a clear separation between vitelline and plasma membranes were selected, returned to solutions of normal osmolarity, and saved overnight at 19°C in Leibovitz L-15 medium (2, 3). This modification in the oocyte preparative technique significantly reduced adhesions between the vitelline and plasma membranes and greatly enhanced the probability of forming high-resistance seals at the time of patching (3). Oocytes used for either TEVC or patch-clamp were injected with 0.5 to 1 ng cRNA and incubated at 19°C in Barth's solution supplemented with Leibovitz medium for 1-3 days before measurements were made.

Electrical. Whole oocyte current-voltage (I-V) relationships were obtained, in intact oocytes, using a TEVC with 3 M KCl-filled current and voltage electrodes, as previously described (2). Currents were recorded for 50 ms at each voltage, using a pulse protocol in which membrane potential (V_m) was stepped by 10 mV from +60 mV to 140 mV, interspersed with a return to the resting V_m. The oocytes reserved for single-channel measurements were again subjected to a hypertonic shrinking solution, thereby allowing the vitelline membrane to be easily removed. Single channels were studied using the patch-clamp technique as previously described (2, 18).

The stripped oocytes were placed in a small Lucite chamber where the bath solution could be exchanged by gravity perfusion. All measurements were made at room temperature. Methods for data acquisition and analysis have been described in detail elsewhere (6, 12). Pipettes were pulled from hematocrit tubing on a three-stage puller and were coated with Sylgard prior to use. Currents were recorded with either List EPC-7 or Dagan 8900 patch-clamp amplifiers and stored, unfiltered, on videotape. For analysis, current records were replayed from videotape, sampled at 5 kHz, low-pass filtered at 1 kHz, and analyzed using an Atari-based data acquisition system and TAC software (Instrutech, Mineola, NY).

Solutions. In the TEVC experiments, initial measurements of resting potential were performed with (in mM) 105 NaCl, 5 KCl, 1 MgCl₂, 2 CaCl₂, and 5 N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid (HEPES), adjusted to pH 7.4 with NaOH. Those experiments examining the effects of internal oocyte pH on macroscopic conductance utilized high-K bath solutions consisting of (in mM) 55 KCl, 55 potassium acetate, 1 MgCl₂, and 2 CaCl₂, buffered with 10 HEPES, and adjusted to a final pH between 6.3 and 8.2 with KOH. As previously described (14), internal oocyte pH could be controlled with membrane-permeable potassium acetate solutions. However, this method of changing oocyte pH required about a 20-min equilibration period for each new pH, and the measured internal pH was always significantly lower than bath pH (see below).

In the patch-clamp experiments, oocytes were bathed in a high-K bath solution, consisting of either 110 mM KCl or 55 mM KCl + 55 potassium acetate (for the pH studies) and 2 mM CaCl₂, buffered with 10 HEPES. Pipettes were filled with either 110 mM KCl or 55 mM KCl + 55 mM potassium acetate (for the pH studies) and 1 mM MgCl₂, buffered with 10 mM HEPES. In the patch experiments, MgCl₂ was omitted from the bath solution, because it accelerates rundown in excised patches, but it was retained in the pipette solution, because the single-channel conductance of ROMK is Mg²⁺ dependent. All other chemicals were obtained from Sigma Chemical.

Intracellular pH measurements. pH_i was measured with pH-selective microelectrodes in a small sample of oocytes to determine how well the membrane-permeable acetate-buffered solutions controlled pH_i. Glass capillaries (model GC 200 F-10; Warner Instruments, Hamden, CT) were pulled on a horizontal microelectrode puller (model P-97; Sutter Instruments, San Francisco, CA). pH-selective electrodes were silanized by exposure to 40 µl of bis(dimethylamino)-dimethylsilane (catalog no. 14755, Fluka Chemical) in a covered glass container at 200°C, and the tips were coated with hardened Sylgard (Dow Corning) to reduce electrical noise. The tip of the pH electrode was filled (from the back) with Hydrogen ionophore I-Cocktail B (catalog no. 95293, Fluka) and then back-filled with pH 7 solution of 0.04 M KH₂PO₄, 0.023 M NaOH, and 0.015 M NaCl and inserted into an Ag/AgCl half-cell electrode holder. The pH electrodes had resistances of ~5 × 10¹¹  and were calibrated using commercial standard pH 6 and pH 8 solutions. The mean slope of the electrodes used in this study was 56.2 ± 0.6 mV/pH unit. Voltage electrodes were pulled as above but filled with 3 M KCl and had tip resistances of 30-60 M. The voltage due to pH_i was obtained by electronically subtracting the potential recorded by the voltage microelectrode from the voltage recorded by the pH microelectrode using a high-impedance electrometer (model FD 223; World Precision Instruments, Sarasota, FL). Intracellular voltage and pH were continuously displayed with an online computer system. The pH_i measurements reported in this study were conducted at Yale Univ. Dept. of Physiology in the laboratory of Dr. W. F. Boron, with the expert assistance of Dr. G. Cooper and Dr. M. Romero.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Control of oocyte pH with acetate solutions. To determine the time course and effectiveness of membrane-permeable, acetate-buffered solutions at controlling oocyte pH, pH-selective (resin-filled) microelectrodes were used to measure pH_i in a small group of oocytes having either high or low K conductance (see MATERIALS AND METHODS). When initially placed in a 2 mM K-Ringer solution, those oocytes expressing large amounts of wild-type ROMK2 (WT-ROMK2) channels for several days had generally more negative resting potentials (90 ± 5 mV) and slightly lower initial pH_i (7.13 ± 0.03) than low K conductance, water-injected oocytes (73 ± 6 mV, initial pH_i = 7.32 ± 0.02).

Immersion in 55 mM KCl + 55 mM potassium acetate depolarized both high and low K conductance oocytes to resting potentials near zero mV. Progressive exposure of these oocytes to potassium acetate-chloride solutions buffered at pH values between 8.2 and 6.3 decreased oocyte pH_i from 7.2 to 6.1 where steady-state pH_i values were generally obtained at 20-25 min after a change in external pH. These results are summarized in Table 1 for the five bath pH values tested on either ROMK2- or water-injected oocytes from three frogs.

In all oocytes, steady-state pH_i was always lower than the bath pH, although the degree of deviation depended on extracellular pH. Furthermore, exposure to acetate solutions always decreased pH_i below its resting value, as initially measured in 2 mM K Ringer solution. It was not possible to alkalinize the oocytes with extracellular potassium acetate solutions.

Effect of pH on single-channel ROMK2 currents. Results of the present study confirm the pH sensitivity of ROMK at both the single channel and whole cell level. Expression of ROMK2 in Xenopus oocytes yielded channels with high open probability and mild inward rectification as previously described (2). In cell-attached patches with one channel in the patch, intracellular acidification (via permeant acetate buffers) consistently reduced P_o.

A complete time course illustrating the effect of lowering pH_i on ROMK2 is given in Fig. 1. The cell-attached patch contained only one channel, and the oocyte was maintained at 100 mV relative to the pipette throughout the 20-min recording. In this experiment, extracellular pH was changed from 7.4 to 6.8 using the membrane-permeable acetate buffer described (MATERIALS AND METHODS). According to the pH_i measurements (Table 1), this would correspond to a steady-state change in oocyte pH_i from 6.8 to 6.4. Although pH_i was not directly measured in this particular experiment, the decline in pH_i roughly paralleled the decrease in channel activity. Comparison of the time course of pH_i and inward current is described below. About 10 min after the change in bath pH from 7.4 to 6.8, there was a noticeable increase in long closures characterized by the appearance of a third closed state in the kinetic scheme (Fig. 2B). This effect could be reversed by returning the bath pH to 7.4. In five single-channel recordings (each on a separate oocyte), the average lifetime of this third long-closed state was 2.6 ± 0.5 s (Table 2).

View larger version (36K): [in this window] [in a new window]   Fig. 1.   Effect of changing internal pH from 7.4 to 6.8 on wild-type ROMK2 (WT-ROMK2) currents recorded from a single channel in a cell-attached patch on an oocyte depolarized by 55 mM KCl + 55 mM potassium acetate. Oocyte was maintained at 100 mV relative to pipette throughout the changes in internal pH, which were accomplished with a membrane-permeable acetate buffer. Bath pH values of 7.4 and 6.8 correspond, respectively, to intracellular pH (pHi) values of 6.8 and 6.4. Downward deflections from the closed state (dotted line) denote inward currents. Circled events denote subconductances, appearing before channel shutdown and shortly after reactivation.

View larger version (24K): [in this window] [in a new window]   Fig. 2.   Effect of decreasing pHi on closed-time histograms of WT-ROMK: analysis of cell-attached patch, containing a single channel (records shown in Fig. 1). Oocyte was maintained at 100 mV relative to pipette throughout the changes in internal pH, which were accomplished with a permeant buffer. A: closed-time distribution for initial recording at a bath pH of 7.4 (pHi = 6.8), showing 2 discrete closed states with time constants of 1.6 ms and 85 ms and open probability (Po) = 0.8. B: 6 min after changing external, acetate-buffered solution to pH 6.8 (pHi = 6.4). Note appearance of very long closed state; Po = 0.2. C: recovery of channel activity 3 min after return of bath to pH 7.4 (pHi = 6.8), showing 2 closed states with time constants of 1.5 ms and 81 ms. Note disappearance of very long closed state. D: single-channel conductance (39 pS in this experiment) was unaffected by changes in pH.

In five experiments at a bath pH of 7.4 and a V_m of 100 mV, the average open probability for WT-ROMK2 was P_o = 0.76 ± 0.02. Reduction of bath pH to 6.8 decreased P_o in all five cases. Analysis of cell-attached current records 30 s preceding shutdown of the channel yielded the kinetic data shown in Table 2, middle. This decrease in open probability to P_o = 0.38 ± 0.11 (n = 5) could be completely explained by the appearance of a very long closed state (Fig. 2B). In five recordings, return to a bath pH of 7.4 (steady-state pH_i = 6.8) caused a disappearance of the this long closed state and restored the open probability to P_o = 0.72 ± 0.06.

As indicated by the example of Figs. 1 and 2, the period of long closures persisted for ~20-30 s until a complete closure occurred, producing an apparent cessation of channel activity. It was possible to restore channel activity if bath pH were returned to 7.4 within about 45 s of channel shutdown. Allowing channels to remain for more than 45 s in the closed state at low pH_i prevented reactivation upon return to bath pH 7.4. In addition, the long closures prior to complete shutdown were often preceded by the random appearance of discrete substates (circled events in Fig. 1). These also appeared at the beginning of channel recovery following return to a bath pH of 7.4. These substates were too few in number to study, and their significance is not understood.

The pH-dependent shutdown of the channel did not alter the lifetime of either the open state or the two shorter closed states. This is illustrated in Fig. 2, and the kinetic data are summarized in Table 2. Single-channel conductance also remained the same during the pH-dependent shutdown of the channel (Fig. 2D). The decrease in P_o at low pH can be attributed to the appearance of a long closed state.

Channels in excised (inside-out) patches had a similar, but somewhat faster, response to alterations in cytoplasmic-side pH (Fig. 3) than was observed with cell-attached patches. ROMK2 channel activity in the excised patch of Fig. 3 was completely blocked by a bath pH of 6.4, although this required 47 s after the start of the bath solution change. Since bath exchange was complete within 10 s, pH shutdown of the channel is not immediate, even in excised patches. Furthermore, channel activity in the excised patch did not recover until more than a minute after return of the bath pH to 7.4 (Fig. 3).

View larger version (26K): [in this window] [in a new window]   Fig. 3.   Effect of changing cytosolic side pH from 7.4 to 6.4 on WT-ROMK2 currents recorded from a single channel in an excised (inside-out) patch in a high-K bath (55 mM KCl + 55 mM potassium acetate). Cytoplasmic side of the membrane was maintained at 100 mV relative to pipette throughout the changes in cytoplasmic-side pH. Downward deflections from the closed state (dotted line) denote inward currents.

Similar to the results with cell-attached patches, a complete absence of channel activity for more than 45 s at low pH prevented recovery, even after a return of bath pH to 7.4. In no case did a reduction in pH cause significant changes in current amplitude, although the rundown that often occurred with excised patches sometimes reduced current amplitude before complete disappearance of channel activity. The reason for this is not known, but, to avoid confusion, average kinetic parameters were computed only for cell-attached patches.

pH_i dependence of inward rectifier macroscopic currents. Macroscopic whole cell currents from WT-ROMK display a marked sensitivity to pH_i (14), analogous to the pH sensitivity of the native small K channel in native rat CCT (15). Our studies confirmed this steep dependence of inward conductance on pH_i (solid line of Fig. 4). For WT-ROMK2 the average pK_a was 6.5 ± 0.01 (8 oocytes), and the Hill coefficient was 5.2 ± 0.6, indicating that permeation through this channel is turned on and off over a relatively narrow range of pH.

View larger version (16K): [in this window] [in a new window]   Fig. 4.   pH dependence of WT-ROMK2, IRK1, and K61M mutants. Ordinate: normalized inward conductance (at 80 mV) of oocytes expressing either WT-ROMK2 (, solid line), IRK1 (, dotted line) or K61M (, dashed line). Abscissa: pHi controlled with extracellular potassium acetate solutions (see text). Conductances at different values of internal pH were normalized to the maximum inward conductance for that oocyte. Data were fit to a modified Hill equation (see text).

1

1

View larger version (13K): [in this window] [in a new window]   Fig. 5.   Comparison of macroscopic current and intracellular pH (pHi) time course for WT-ROMK2. For both curves the bath pH was changed from 7.4 to 6.8 using potassium acetate solutions. This produced a decline in steady-state pHi from 6.8 to 6.4 in 20 min. Solid line () are the measured normalized macroscopic inward currents at 60 mV during the change in bath pH. This was fit to a single exponential with a time constant of 5.6 ± 0.2 min1. Dashed line (), time course for pHi during this same period. This was fit to a single exponential with a time constant of 7.6 ± 0.6 min1.

pH_i dependence of T51 mutant macroscopic currents. Both IRK and ROMK channels possess a highly conserved, mildly hydrophobic area (designated the "Q" region) that is located immediately NH₂ terminal to the first transmembrane segment (M1) of both inward rectifiers (Fig. 6). The "Q regions" of the inward rectifier superfamily show a high degree of homology. The boldface residues of Fig. 7 indicate amino acids that are identical to those of ROMK2. A cursory inspection of Fig. 7 reveals that many of the nonidentical residues in the same column have similar properties. Finally, the hydrophobicity and proximity of the Q region to the M1 transmembrane segment raises the possibility that this region might be involved in the permeation process.

View larger version (16K): [in this window] [in a new window]   Fig. 6.   Hydropathy analysis of the predicted amino acid sequence of ROMK2, based on the Kyte-Doolittle algorithm. Abscissa: amino acid sequence number. Ordinate: positive numbers denote increasing hydrophobicity. M1 and M2, putative membrane spanning region; P, pore region; and Q, area of mild hydrophobicity, explored in the present treatment.

View larger version (35K): [in this window] [in a new window]   Fig. 7.   Homologies within the "Q" regions of ROMK2 and the inward rectifier family. Boldface letters indicate residues identical to those of ROMK2. Boxed residues depict the locus of the T51E and T52E mutations. Numbers 44, 51, and 59 refer to amino acid locations along the ROMK2 sequence.

View larger version (19K): [in this window] [in a new window]   Fig. 8.   pH dependence of WT-ROMK2 and T51 mutants. Ordinate: normalized inward conductance (at 80 mV) of oocytes expressing either WT-ROMK (WT, solid line, ) or mutants T51E (dashed line, ) or T52E (dotted line, ), or T52K (broken line, ). Conductances at different values of internal pH were normalized to the maximum inward conductance for that oocyte. Data were fit to a modified Hill equation (see text). Abscissa: pHi controlled with extracellular potassium acetate solutions. Relationship between internal and external pH is given in Table 1.

View larger version (30K): [in this window] [in a new window]   Fig. 9.   K currents through a single T51E mutant ROMK2 channel, expressed in an oocyte, depolarized by 110 mM KCl (pH 7.4). A: cell-attached patch with oocyte potential of 200 mV relative to pipette. Downward deflections from the closed state (dotted line) denote inward currents. Inward single-channel conductance was 16 pS. B: closed time distribution for T51E, showing 2 discrete closed states with time constants of 1.0 and 25 ms. C: open time distribution for T51E currents, showing 1 open state with time constant of 5.2 ms. Po (200 mV) = 0.74.

View larger version (38K): [in this window] [in a new window]   Fig. 10.   K currents through a single WT-ROMK2 channel, expressed in an oocyte, depolarized by 110 mM KCl (pH 7.4). A: cell-attached patch with oocyte potential of 200 mV relative to pipette. Downward deflections from the closed state (dotted line) denote inward currents. Inward single-channel conductance was 33 pS. B: closed time distribution for WT-ROMK, showing 2 discrete closed states with time constants of 0.9 and 17 ms. C: open time distribution for WT-ROMK, showing 1 open state with time constant of 4.8 ms. Po (200 mV) = 0.72.

View larger version (18K): [in this window] [in a new window]   Fig. 11.   Comparison of single-channel rectification between T51E and WT-ROMK2. Cell-attached patches on oocytes expressing either T51E mutant (dashed line, ) or WT-ROMK (solid line, ). Pipette and bath solutions consisted of 110 mM KCl buffered to pH 7.4. Pipette also contained 1 mM MgCl2. Mean inward single-channel conductance for T51E was 17 ± 1 pS (n = 9). Mean inward single-channel conductance for wild type was 34 ± 1 pS (n = 8).

View larger version (16K): [in this window] [in a new window]   Fig. 12.   Normalized single-channel current-voltage (I-V) relations for T51E and wild type. Cell-attached patches on oocytes expressing either T51E mutant (dashed line, ) or WT-ROMK (solid line, ). Pipette and bath solutions consisted of 110 mM KCl buffered to pH 7.4. Pipette also contained 1 mM MgCl2. Data are from Fig. 11, where T51E currents have been scaled by the WT/T51E ratio of inward conductances at 80 mV.

View larger version (15K): [in this window] [in a new window]   Fig. 13.   T51E enhances macroscopic inward rectification. Comparison of the 2-electrode I-V relations for the mutant T51E and WT-ROMK2. All currents were normalized to maximum inward current at 80 mV. Bathing solution contained 110 mM KCl (pH = 7.4), and the average oocyte resting potential was zero mV.

View larger version (14K): [in this window] [in a new window]   Fig. 14.   Voltage dependence of open probability for T51E and wild type. Cell-attached patches on oocytes expressing either T51E mutant (dashed line, ) or WT-ROMK2 (solid line, ). Pipette and bath solutions consisted of 110 mM KCl buffered to pH 7.4. Pipette also contained 1 mM MgCl2.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Modulation of macroscopic pH sensitivity. Patch-clamp studies on cut-open rat CCT initially established the regulation of secretory SK potassium channels by pH_i (15). This dependence on pH_i was also observed with ROMK channels expressed in oocytes and studied either at the single channel (10, 11) or the whole cell level (14). In the latter experiments, simultaneous measurement of macroscopic current and pH_i indicated a steep dependence of ROMK1 conductance on internal pH, involving a highly cooperative process with four H⁺ binding sites and a pK_a of 6.8 (14). This is in contrast to the pH insensitivity of the strong inward rectifiers like IRK1 (4).

Recently, it was discovered that a single lysine (K61 on ROMK2, K80 on ROMK1), NH₂ terminal to the first hydrophobic segment (M1), is necessary for conferring pH_i sensitivity to ROMK (4). This result was confirmed in the present study, where replacing the lysine at position 61 with methionine (K61M) abolished the pH_i sensitivity of ROMK2 between 7.2 and 6.2. In addition to the critical residue K61, a second site (position 51 on ROMK2; 70 on ROMK1) was also found to affect pH sensitivity of the channel. This might indicate a physical proximity between positions 51 and 61 (on ROMK2) since changing the charge of residue 51 shifted the apparent pK_a of the channel. Replacing the uncharged threonine by a negatively charged glutamate shifted the pK_a in an alkaline direction by 0.5 pH units, from 6.5 ± 0.01 (WT) to 7.0 ± 0.02 (T51E). Replacing threonine by a positively charged lysine shifted the pK_a in an acidic direction by 0.5 pH units, from 6.5 ± 0.01 (WT) to 6.0 ± 0.02 (T51K). However, the modulatory effects of position 51 were not observed after removal of the critical lysine (K61); i.e., the double mutant T51E/K61M had the same insensitivity to pH as the single-point mutant (K61M, ROMK2). This suggests that the T51 locus modulates the pH sensitivity of the K61 site rather than functioning as an additional pH sensor.

Since the pK_a of lysine in free solution is 10.5, it was suggested that residue K61 confers pH sensitivity to the channel in the physiological range because the local chemical environment within the protein shifts the pK_a of lysine from 10.5 to ~6.5, (4). How residue T51 modulates this pH sensitivity is not known. One hypothesis consistent with our results is that the mildly hydrophobic region at T51 is sufficiently close to the hydrophobic region of the first transmembrane segment (M1) to physically interact with K61, which is just NH₂ terminal to M1. The interaction could be electrostatic in nature, since positive and negative charges had opposite effects.

We believe that intracellular rather than extracellular pH modulates both the K61 and the T51 sites on ROMK2. We observed no obvious effect on ROMK macroscopic inward currents when bath pH was changed with membrane-impermeable KCl-HEPES solutions. A lack of effect of extracellular pH was also reported for ROMK1, where changes in bath pH via membrane-impermeable biphthalate had no effect on macroscopic inward currents (14).

Effect of pH on single-channel currents. The mechanism whereby decreases in cytoplasmic-side pH shut down channel activity is not well understood. In cell-attached patches, no ostensible change in channel kinetics occurs until ~10-15 min after reducing bath pH to 6.8. Presumably, this is the time required for the interior of the oocyte to reach a critical internal pH of ~6.4. At this pH_i a very long closed state appears (Fig. 2B), which eventually becomes a complete closure. For reasons we do not understand, complete channel closure at low pH_i for more than 45 s seems to interfere with reactivation upon return to a pH 7.4 bath solution. Although channels may continue to remain closed for a variable time after return to normal bath pH, the time spent at low pH seems to be a critical factor for restoration of channel activity.

Recordings from excised patches support the hypothesis that lowering cytoplasmic-side pH initiates very long closures, followed by complete cessation of channel activity. In excised patches, the decline in channel activity is slower than the decline in bath pH. It is unclear whether this represents a delay in diffusion of new solution to the patch of membrane within the pipette or simply a delayed effect of pH on channel gating.

Upon return of the bath to pH 7.4, ROMK2 channel activity recovered with a latency that varied between about a minute for excised patches to more than 15 min for cell-attached recordings. The delay for the cell-attached experiments presumably represents the time required to elevate oocyte pH back to normal with the membrane-permeable acetate solutions. However, a delay in recovery of channel activity was also observed in excised patches, suggesting that the reversal of the inhibited state is slow compared with the actual change in pH.

Exposure of either excised or cell-attached patches for longer than 45 s prevented restoration of channel activity following a return to pH 7.4. The relatively narrow time window (45 s) beyond which it became impossible to reactivate individual single channels differs from the reversibility of WT-ROMK macroscopic conductance measured with the TEVC. In these latter experiments, inward conductance essentially returned to control values even after a 10-min exposure to a pH 6.8 bath. The reason for this difference in microscopic vs. macroscopic reversibility is not known.

Single-channel properties of the T51 mutants. The position of residue T51 (ROMK2) relative to the first transmembrane segment raised the possibility that this site might also interact with the K permeation path of the channel. In fact, the T51E mutation reduces apparent single-channel conductance by 50% and dramatically increases open-channel noise. These changes are consistent with an interaction between residue 51 and the permeation path of the channel.

The persistence of T51E-specific open-channel noise at all voltages and over a range of pH_i from 6.1 to 7.4 seems to rule out cytoplasmic proton block as a cause of the increased noise level. However, it is still possible that the T51E mutation introduces a fast-flicker open state that is beyond the resolution of the data acquisition system. In this case, T51E could appear to have a much reduced conductance when records are filtered at 1 kHz.

Although the reason for the decrease in apparent conductance produced by T51E is not known, it is unlikely that it arises from a simple electrostatic effect, since replacing threonine with a positively charged basic lysine residue (T51K) did not produce an opposite effect on conductance. In fact, the single-channel conductance of T51K was indistinguishable from that of WT-ROMK2 (Table 4). Furthermore, when the T51 residue was replaced by histidine (which should be less protonated than lysine at pH = 7.4), single-channel conductance actually increased (Table 4).

We have no hypothesis for the dramatic effect of the glutamate substitution (T51E) on single-channel voltage gating (Fig. 14). Other amino acid substitutions at this position, T51H for example, leave P_o virtually unchanged from wild type (Table 4). A simple electrostatic mechanism seems an even less likely explanation for changes in voltage gating, since the T51K mutant (which introduces a positive charge at position 51) also decreases the P_o of the channel, in a manner similar to that seen with (the negatively charged) T51E.

Comparison with previous studies. The present study confirms previous findings on the marked dependence of ROMK current on pH_i (4, 10, 11, 14, 15). We have also confirmed that the lysine residue (K61 on ROMK2, K80 on ROMK1), NH₂ terminal to the first putative transmembrane segment (M1), is primarily responsible for conferring a steep pH sensitivity to ROMK that is not seen with IRK1 (4).

The principal difference between our pH dependence of WT-ROMK and that reported by both Tsai et al. (14) and Fakler et al. (4) is the value of apparent pK_a for ROMK. In our studies the apparent pK_a for ROMK2 was 6.5 ± 0.01. This is significantly lower than either the pK_a of 6.8 reported for ROMK1 by Tsai et al. (14) or the pK_a of 6.9 reported for ROMK2 by Fakler et al. (4). The precise reason for this disagreement is not known but probably arises from the larger discrepancy between intracellular and extracellular pH measured in our study, compared with previous reports (14).

In our experiments (Table 1), decreasing bath pH from 7.4 to 6.8 reduced oocyte pH from 6.8 to 6.4 compared with the decrease in pH_i of 7.2 to 6.7 reported by Tsai et al. (14). Furthermore, we were never able to alkalinize oocytes above their resting pH (about 7.2) using acetate-buffered solutions, so that it was only possible (with this technique) to study oocytes with an internal pH less than or equal to 7.2.

The disparity in pK_a values is of interest, since it determines whether ROMK2 would be regulated by normal variations in pH_i. The results of our study suggest that ROMK2 is probably not significantly regulated by pH under physiological conditions, since it is unlikely that renal cells would normally possess a pH_i below 6.7. Hence, only those cells with impaired acid extrusion processes would have a sufficiently low pH_i to shut down the secretory ROMK channel. This might serve as a protective mechanism to conserve intracellular K during energetically restricted conditions in which pH_i regulation was compromised.

Summary. The present study confirms that a lysine residue (K61 on ROMK2; K80 on ROMK1) is the primary locus of the pH sensitivity present in ROMK but not in IRK1. However, a specific residue in a mildly hydrophobic region of ROMK (T51 on ROMK2; T70 on ROMK1) appears to modulate this pH sensitivity. Replacing the neutral threonine at this site with a negatively charged residue shifts the apparent pK_a in an alkaline direction, whereas substitution with a positively charged residue shifts the pK_a in an acid direction. Substitutions at position 51 also significantly affected single-channel conductance and voltage-dependent gating of the channel. In contrast, mutations in a neighboring residue T52 (T71 on ROMK1) had no effect on pK_a or single-channel conductance. This is consistent with a model in which T51 is sufficiently close to the primary pH site (K61) to alter ROMK pH sensitivity while also exerting effects on the permeation pathway in the pore region.