Phosphorylation-regulated endoplasmic reticulum retention signal in the renal outer-medullary K⁺ channel (ROMK)

Abstract

The renal outer-medullary K⁺ channel (ROMK; Kir1.1) mediates K⁺ secretion in the renal mammalian nephron that is critical to both sodium and potassium homeostasis. The posttranscriptional expression of ROMK in the plasma membrane of cells is regulated by delivery of protein from endoplasmic reticulum (ER) to the cell surface and by retrieval by dynamin-dependent endocytic mechanisms in clathrin-coated pits. The S44 in the NH₂ terminus of ROMK1 can be phosphorylated by PKA and serum- and glucocorticoid-inducible kinase-1, and this process increases surface expression of functional channels. We present evidence that phosphorylation of S44 modulates channel expression by increasing its cell surface delivery consequent to suppression of a COOH-terminal ER retention signal. This phosphorylation switch of the ER retention signal could provide a pool of mature and properly folded channels for rapid delivery to the plasma membrane. The x-ray crystal structures of inward rectifier K⁺ channels have shown a close apposition of the NH₂ terminus with the distal COOH terminus of the adjacent subunit in the channel homotetramer, which is important to channel gating. Thus, NH₂-terminal phosphorylation modifying a COOH-terminal ER retention signal in ROMK1 could serve as a checkpoint for proper subunit folding critical to channel gating.

The posttranscriptional expression of channels and receptors in the plasma membrane of cells is regulated by delivery of protein from endoplasmic reticulum (ER) to the cell surface and by retrieval by endocytic mechanisms. Renal outer-medullary K⁺ channel (ROMK; Kir1.1) (1, 2) is a small-conductance K⁺ channel in apical membranes of the distal nephron that mediates renal K⁺ recycling in the thick ascending limb of Henle and K⁺ secretion in the distal convoluted tubule and collecting duct (3–5). This channel belongs to the inwardly rectifying family (Kir) K⁺ channels that includes the ATP-sensitive Kir6.x or K_ATP channels. ROMK activity is regulated by a wide variety of factors, including changes in cytosolic pH and by several kinases and phosphatases (3). The phosphorylation state of ROMK can affect gating and surface expression. The latter depends on the tyrosine kinase Src (6) and several serine–threonine kinases, including PKC (7, 8), PKA (9–11), WNK4 [with no lysine (K) 4] kinase (12) and serum- and glucocorticoid-activated kinase-1 (SGK1) (10). When ROMK is expressed in Xenopus laevis oocytes, retrieval of channels from the plasma membrane is mediated by a dynamin-dependent endocytic process involving clathrin-coated pits (13) that is regulated by kinases, such as Src, PKC, and WNK4. Endocytosis of ROMK and certain cell-surface receptors can also be modulated by monoubiquitination (14). In contrast, the mechanisms responsible for ROMK channel delivery to the cell surface have not been defined.

PKA and SGK1 phosphorylate an NH₂-terminal serine residue on ROMK and thereby modulate its surface expression (9–11, 15). This phosphorylation site is conserved in all ROMK isoforms (ROMK1–ROMK3), which have varied initial NH₂-terminal sequences (11, 16). Isoforms ROMK1–ROMK3 have similar single-channel properties but are differentially expressed along the nephron (3). PKA also phosphorylates two other serine residues located in the COOH terminus (11), but these sites do not alter surface expression; instead, they affect channel gating (9). Channel activity requires at least two of the three PKA sites to be phosphorylated (9, 11). In X. laevis oocytes, full stimulation by PKA or SGK1 requires accessory scaffolding proteins: a kinase-anchoring protein AKAP (17) and Na⁺–H⁺ exchanger regulatory factor (NHERF)2, respectively (15, 18, 19). The renal excretion of K⁺ is impaired in the sgk1 knockout mouse (20), consistent with a role for SGK1-mediated phosphorylation of the NH₂-terminal serine on ROMK in channel regulation in vivo.

We and others have previously shown that mutation of the NH₂-terminal serine residue to alanine prevents phosphorylation by PKA (11) and SGK1 (10), reduces whole-cell K current when expressed in X. laevis oocytes (10, 11, 21), and lowers channel density in membrane patches (9, 22) but leaves single-channel properties unchanged (9). Mutation of the NH₂-terminal serine residue to aspartate, which mimics the negative charge carried by a phosphate group bound to a serine, has the opposite effect of increasing whole-cell current, channel density, and cell-surface expression (9, 10, 21, 22). The increase in whole-cell current by the serine-to-aspartate mutation is similar to that induced by PKA or SGK1 (9, 10, 15, 18).

Although it is known that phosphorylation of the NH₂-terminal serine in ROMK channels leads to a higher-channel surface expression, the mechanism that increases membrane-bound channel numbers is unknown. PKA-dependent phosphorylation processes have been shown to regulate the surface delivery of certain channels, such as the NMDA receptor (23, 24) and a cardiac Na channel (25). Thus, in the present study, we examined the mechanism by which phosphorylation of the NH₂-terminal serine in ROMK1 modulates surface expression and determined whether this phosphorylation functions as a molecular switch for regulated ER retention.

Methods

Molecular Biology. ROMK1 and ROMK2 cDNA were subcloned into pSPORT-1 vectors (Invitrogen), and the ORF of EGFP-C2 (Clontech) was ligated in-frame to the NH₂ terminus. Site-directed mutagenesis was accomplished by using QuikChange (Stratagene). All constructs were verified by sequencing (The W.M. Keck Facility, Yale University), linearized with Not-1 restriction endonuclease, and used to generate full-length complementary RNA transcripts (mMESSAGE mMACHINE T7 high yield transcription kit; Ambion, Austin, TX). complementary RNA (cRNA) for wild type and the dominant-negative [K44A (26, 27)] forms of dynamin-1 were prepared as described above. Concentrations for coinjection experiments were 5 ng of dynamin-1/5 ng of ROMK per 50 nl of aliquot.

Preparation and Injection of Oocytes. Standard methods were used to express wild-type ROMK and mutant cRNA in stages 5 and 6 X. laevis oocytes (9). In brief, frogs were anesthetized in 0.02% 3-aminobenzoic acid ethyl ester (tricane; titrated to pH 7.4 with 5 mM Hepes using NaOH) for 3–7 min. After partial ovariectomy, oocytes were separated with forceps and then chemically defolliculated by treatment with 2 mg/ml type 1A collagenase dissolved in aCa²⁺-free solution (96 mM NaCl/2mMKCl/2 mM MgCl₂/5mM Hepes, adjusted to pH 7.4 with NaOH). Oocytes were then washed 12 times in ND-96 solution (96 mM NaCl/2 mM KCl/1.8 mM CaCl₂/1 mM MgCl₂/5 mM Hepes, adjusted to pH 7.4 with NaOH). Oocytes were injected with 50 nl of cRNA (5 or 12.5 ng) or water (control) by using an Inject+Matic (Geneva) injector immediately or no later than 12 hours after defolliculation. Oocytes were then incubated in the ND-96 solution containing 500 units/ml penicillin and 500 μg/ml streptomycin (Invitrogen) at 18°C until used for study. Brefeldin A (BFA) was dissolved in 100% ethanol and added to incubation medium to a concentration of 3 μM. Ethanol has no significant effect on ROMK whole-cell currents in oocytes (13).

Western Blot. Total oocyte expression of EGFP–ROMK1 was assessed by Western blotting as described in ref. 14. Three batches of oocytes were injected with 5 ng of cRNA from wild-type EGFP–ROMK1 or S44 mutants, with water-injected oocytes serving as a control. Ten oocytes expressing each construct were homogenized in 100 μl of lysis buffer and 20 μg of total protein loaded per lane. EGFP–ROMK1 bands were resolved by 4–15% SDS/PAGE, transferred to nitrocellulose membranes, and stained with a 1:500 dilution of anti-ROMK antibody (Alomone Labs, Jerusalem) followed by a 1:5,000 dilution of horseradish peroxidase-F(ab′)2 Goat anti-rabbit IgG (Invitrogen). Expression levels of mutants were normalized to wild-type EGFP–ROMK1.

Two-Electrode Voltage Clamp. Microelectrodes were formed from borosilicate glass tubing (Kimble Glass, Vineland, NJ) by pulling to a tip resistance of 0.5–1.5 MΩ using a pipette puller (product no. PP-830, Narishige, Tokyo) and backfilled with 3 M KCl. Currents were measured with an oocyte clamp (product no. OC-725C, Warner Instruments, Hamden, CT) and passed through a low-pass, 100 Hz, eight-pole Bessel filter (product no. 902LPF, Frequency Devices, MA). Data were acquired by using a digital-to-analogue converter (Digidata 1200 series, Axon Instruments) driven by pclamp 6.04 (Axon Instruments). Steady-state currents were analyzed off-line with Clampfit 8.20 (Axon Instruments, CA) and PRISM ⁴ (GraphPad, San Diego). Bath perfusion solution contained 96 mM NaCl, 5 mM KCl, 1 mM MgCl₂, 2 mM CaCl₂, and 10 mM Hepes titrated to pH 7.4 with NaOH. The EGFP tag did not alter whole-cell current profiles (Fig. 5 A and C, which is published as supporting information on the PNAS web site) of wild-type and mutant channels.

Cell-Attached and Giant Inside-Out Patch Clamp. After injection, oocytes were immersed in a hyperosmotic solution for 1–2 min (200 mM N-methyl-d-glucamine/2 mM KCl/1 mM MgCl₂/10 mM EGTA/10 mM Hepes adjusted to pH 7.4 with HCl), and vitelline membranes were removed with forceps. Electrodes (8–9 MΩ for single channel recordings and 0.3–0.6 MΩ for inside-out giant patches) were pulled from borosilicate glass capillaries (Sutter Instruments, Novato, CA) on a Narishige PP-830 puller and polished (Narishige MF-83 microforge). Electrodes were filled with 150 mM KCl/1.0 mM MgCl₂/1.0 mM CaCl₂/5 mM Mes/Tris, adjusted to pH 7.4 with Mes or Tris buffer. Mg²⁺-free bath solutions were used in all experiments and perfused by a multibarrel quick-exchange solution system (Model SF-77B, Warner Instruments). Bath solutions contained 150 mM KCl, 2 mM EDTA, and 10 mM Mes/Tris and were adjusted to pH 7.4, 8.0, 8.5, 9.0, and 9.5 with Mes or Tris. Single-channel currents were recorded in the cell-attached configuration, and macroscopic currents were recorded in the inside-out configuration with a patch clamp amplifier (model L/M–EPC 7, Heka Elektronik, Lambrecht/Pfalz, Germany) and passed through an eight-pole Bessel filter at 1,000 Hz (Warner Instruments). Data were acquired through a digital-to-analogue converter (DigiData 1200 series, Axon Instruments) driven by commercial software (pclamp clampex 8.20, Axon Instruments). Currents were analyzed off-line with clampfit 8.20 and fetchan 6.04 (Axon Instruments) and prism 4 (GraphPad, San Diego). The EGFP tag did not alter single-channel properties (Table 1) of wild-type and mutant channels.

Table 1. Single channel properties of ROMK1 and EGFP–ROMK1.

	ROMK1		EGFP—ROMK1
Channel properties	WT	S44A	S44D	WT	S44A	S44D
Open probability	0.92	0.93	0.91	0.91	0.92	0.91
Single-channel conductance, pS	35.2	31.8	35.4	33.2	34.0	36.3

Open times for each channel were ≈1.5 ms, and closed times for each channel were 19 ms. Single channel conductances were measured between -100 and +40 mV.

Confocal Microscopy. NH₂-terminally tagged EGFP–ROMK1 channels in oocytes were monitored for surface fluorescence 48–72 h after injection with a Zeiss LSM510 confocal laser scanning microscope (objective lens ×10, Nikon) as described previously for ROMK (12) and electroneutral, cation-coupled Cl cotransporters (28, 29). All EGFP–ROMK1 constructs exhibited fluorescence at the oocyte surface and surface fluorescence correlates with protein expression in the plasma membrane (Fig. 5B) (28, 29). The excitation light wavelength was 488 nm, and the emission light wavelength was 505 nm. Fluorescence intensities from confocal images taken at equatorial focal sections were determined by using sigmascan pro 4 (Jandel, San Diego). Oocyte surface EGFP fluorescence provides a reliable determinant of ROMK surface expression (Fig. 5B and Fig. 6, which is published as supporting information on the PNAS web site).

Data Analysis of pH Dose–Response Curves. The pH sensitivity of the ROMK1 S44A/R366A mutant channel was examined in excised giant patches. Observed currents were first fit by using Eq. 1 to determine the theoretical maximum activity. All data were subsequently normalized to this maximal value:

[1]

where I is the total macroscopic current at any pH, I_MIN, and I_MAX are minimum and maximum macroscopic currents, respectively, and n is the Hill coefficient. Data were then fit by using Eq. 2 (least-squares method) to determine I/I_MAX.

[2]

Solutions and Chemicals and Statistical Analysis. All chemicals were purchased from Sigma–Aldrich unless otherwise stated. All data points are means ± SEM. Numbers of oocytes are described as n and are stated in the figure and table legends. For statistical analyses, all data were normally distributed and were compared with a post hoc Student t test. Statistical significance was determined as being P < 0.05.

Results

Ser-44 Modulates Membrane Channel Density. Fig. 1A shows whole-cell currents for each of the EGFP-tagged wild-type and Ser-44 mutant channels, and Fig. 1B summarizes the corresponding EGFP fluorescence levels from the same groups of oocytes. The pattern of surface fluorescence clearly mimics that of the whole-cell current. Because EGFP–ROMK1 single-channel properties were not altered in the S44A and S44D mutants (Table 1), differences in whole-cell currents could not have been due to altered channel function. We also examined whether changes in total expression of the EGFP-tagged ROMK1 constructs might account for the differences in whole-cell current shown in Fig. 1A. A representative Western blot of the EGFP–ROMK1 constructs is shown in Fig. 1D, and expression levels from three batches of oocytes are summarized in Fig. 1C. The overall expression pattern for total protein (Fig. 1C) was similar to those for whole-cell current (Fig. 1A) or surface expression (Fig. 1B) However, S44A current was reduced by ≈90%, whereas expression by Western blot of was reduced by 65%. S44D current was increased by ≈200%, whereas total protein expression was increased by only ≈33%. Thus, the differences in total protein expression cannot fully account for the observed alterations in whole-cell current (Fig. 1A) or channel surface expression (Fig. 1B). In addition, we have previously shown that total ROMK2 channel expression levels are similar among the wild-type and alanine mutations of the serine residues phosphorylated by PKA, including S44 (11). Moreover, others have shown that total expressions of wild-type and S44 mutant hemagglutinin-tagged ROMK1 channels in X. laevis oocytes are equal despite the variations in whole-cell currents (10).

Fig. 1.

Current and surface fluorescence of EGFP–ROMK1 wild-type and S44 mutant channels. (A) Whole-cell currents at +40 mV: wild type = 7.22 ± 1.06 μA; S44A = 0.76 ± 0.17 μA; S44D = 15.1 ± 2.42 μA. (B) EGFP surface fluorescence [expressed in arbitrary units (AU)]: wild type = 4.33 ± 0.98, n = 18; S44A = 1.01 ± 0.26, n = 17; S44D = 19.4 ± 3.57, n = 15. *, Significantly different from wild type. (C) Relative total protein expression levels of EGFP–ROMK1 wild type, S44A, and S44D summarized from three separate batches of oocytes. (D) Representative Western blot of EGFP–ROMK1 constructs with the water-injected oocytes (H₂O) serving as a control.

Phosphorylation of S44 Does Not Inhibit ROMK Endocytosis. The surface expression of ROMK in the oocyte membrane is the balance between delivery and retrieval via dynamin-dependent endocytosis. To distinguish between these processes, we initially examined whether S44 modulates endocytosis. BFA reversibly disrupts Golgi function, hence preventing exocytosis and the delivery of proteins to the plasma membrane (30–32). Thus, in the presence of BFA, loss of whole-cell current is primarily due to endocytosis.

Fig. 2 summarizes the effect of 5 μM BFA on whole-cell currents in groups of oocytes expressing wild-type EGFP–ROMK1 or S44 mutant channels. BFA was added 48 h after cRNA injection when whole-cell currents were stable. Fig. 2A shows S44D currents for 42 h of BFA exposure. In the absence of BFA, S44D currents remain constant (Fig. 2A, open circles along the dashed line), demonstrating that membrane expression of ROMK channels was stable from 48 to 90 h after ROMK1 cRNA injection. In contrast, currents decreased exponentially with BFA treatment (Fig. 2A, filled circles along the solid line), consistent with the endocytosis of ROMK1. The effect of BFA to reduce whole-cell current was reversible because oocytes in which BFA was removed at 72 h increased their current over the next 18 h (Fig. 2A, filled square and arrow). Thus, BFA does not cause a toxic, nonspecific effect in oocytes. Fig. 2B shows that current fell after BFA exposure in wild-type and S44 mutant channels. The current at zero time of BFA treatment was consistent with the differences in channel surface expression shown in Fig. 1. To compare the rates of channel retrieval, currents were normalized and a least-squares one-phase exponential decay was fit to data from individual oocytes. The half-lives for current decay (t_1/2) for each channel type are shown in Fig. 2C and are not significantly different from each other. These results suggest that phosphorylation of S44 does not modulate ROMK endocytosis in oocytes.

Fig. 2.

Effect of S44 mutations on the rates of endocytosis. (A) Time course of changes in normalized whole-cell currents at +40 mV for the EGFP–ROMK1 S44D channel in the absence (–BFA; open circles) or presence (+BFA; filled circles) of 5 μM BFA. In the absence of BFA, current was stable. In contrast, current progressively diminished in the presence of BFA, and the drop in current was reversed after BFA was removed (WASH; filled squares and arrow). +BFA data were fit by a single exponential. All channel constructs behaved similarly. (B) Time courses of currents with BFA. Currents were well fit by single exponentials. (C) Summary of the t_1/2 of current decay for each channel. The t_1/2 was determined from single exponential decays of normalized current. All t_1/2 values were similar.

Phosphorylation of ROMK1 Channels Augments Channel Exocytosis. ROMK channels are known to be internalized by dynamin-dependent, clathrin-coated pit-mediated endocytosis (12, 13). To compare wild-type and mutant S44 ROMK channel delivery rates, clathrin-mediated endocytosis was inhibited by coinjecting EGFP–ROMK1 with the K44A mutant dynamin-1 (12). Dynamin is a GTPase protein essential to clathrin-coated pit-mediated endocytosis (33). This mutation of dynamin prevents GTP hydrolysis and subsequent vesicle budding from the plasma membrane (26, 27). By overexpressing this protein in oocytes, a dominant-negative effect is produced that essentially blocks clathrin-medicated endocytosis. In separate experiments, EGFP–ROMK1 was coinjected with wild-type dynamin to serve as a control. No significant effects on whole-cell currents were observed with the wild-type dynamin (data not shown).

Fig. 3A shows currents over the 24–48 h period after coinjection of oocytes with wild-type or S44 mutant ROMK1 with K44A dynamin. Current progressively increased during this period with all channel constructs. To compare changes in the rates of delivery, currents were normalized to currents at 48 h after injection and rates calculated from the slopes of linear regressions. These data are depicted in 3B and show that the rates of current generation vary in proportion to the surface fluorescence differences as shown in Fig. 1. These results are consistent with phosphorylation of S44 modulating the rate of channel delivery to the plasma membrane.

Fig. 3.

Effect of S44 mutations on the rates of EGFP–ROMK1 channel delivery to the surface. Rates of channel delivery to the oocyte plasma membrane were assessed from the increases in whole-cell current with coexpressed K44A dynamin-1, which inhibits endocytosis of ROMK channels (12, 13). Oocytes were injected alone or coinjected with an equal concentration of mutant (K44A) dynamin-1. (A) Mean currents plotted over a 48-h time course for each channel. (B) The rates of current increases calculated from linear fits to the data values were (expressed in μA/h), for wild type, 0.17 ± 0.01 (n = 3); for S44A, 0.10 ± 0.01 (n = 3); and for S44D, 0.45 ± 0.08 (n = 3). *, Significantly different from WT.

A COOH-Terminal ER Retention Signal Interacts with S44 to Allow Channel Exocytosis. ER retention in Kir6.x (K_ATP) channels (34, 35), GABA_B receptor GB1 subunits (36), and NMDA receptors (23, 24) is mediated by RXR motifs that provide an important quality control mechanism in channel assembly. For K_ATP channels, forward trafficking and subsequent surface delivery depends on the coordinated assembly of Kir6.x and SUR1 subunits that mask individual subunit RXR signals (35). However, in NMDA receptor channels, ER retention mediated by RXR is suppressed by PKA/PKC-mediated phosphorylation processes allowing forward trafficking and surface delivery of this channel. A comparison of the COOH termini of Kir6.2 and ROMK identified putative RXR retention motifs in the NH₂ and COOH termini. The ROMK1 NH₂ terminus contains a triple repeat consisting of RXRXRXR with the terminal R39 (Fig. 7A, which is published as supporting information on the PNAS web site). This RXR repeat sequence is 3 aa upstream of the S44 phosphorylation site. As shown in Fig. 7B, deletion of the NH₂-terimal segment up to R39, ΔN(2–39), did not modulate whole-cell currents in the wild-type channel or the S44A or S44D mutant channels compared with constructs with intact NH₂ termini. Thus, this RXRXRXR segment is not involved in regulation of channel activity by phosphorylation of S44.

In the COOH terminus, the single RXR is shown in Fig. 4A. This ROMK1 RXR motif is 6 aa upstream from the NPXY/F motif, which controls dynamin-dependent endocytosis (12). To assess the role of the COOH-terminal RXR in S44-mediated delivery of ROMK1 channels to the oocyte surface, we mutated R366 to alanine in wild-type and each of the S44 mutants. Data from measurements of whole-cell currents and surface fluorescence in these ROMK1 constructs are summarized in Fig. 4 B and C. The solid bars in Fig. 4 B and C confirm the results in Fig. 1 for this set of experiments. Fig. 4 B and C show that R366A in the wild-type S44 channel increased current and surface fluorescence to the S44D levels. Thus, mutation of the RXR motif in ROMK1 has an effect equivalent to phosphorylation of S44. In addition, the R366A mutation abolished the effect of S44A on surface fluorescence and had no effect on the S44D mutant (Fig. 4C). In other words, R336A produced maximal channel delivery to the surface irrespective of the S44 mutation. Thus, although total expression of the S44A mutant protein is reduced (Fig. 1C), the level of surface expression of this mutant channel can be increased to that for the S44D construct by mutation of the COOH-terminal RXR motif (Fig. 4C).

Fig. 4.

The COOH-terminal RXR ER retention signal is regulated by S44. (A) the RXR sequence in the COOH terminus of ROMK1 is shown with the single-letter amino acid codes, R³⁶⁶AR. Numbering begins from the NH₂-terminal start methionine. NPNF is an NPXY/F endocytosis motif (12), and TQM is the PDZ-1 motif (19). (B) Whole-cell currents at +40 mV. (C) EGFP fluorescence for wild-type channel and S44, R366, and S44/R366 double-mutant channels. (D) Giant patch recording from an oocyte expressing the S44A/R366A channel. The arrow indicates a change from the cell attached to the excised path configuration. Note the current increase with the bath pH change from 7.4 to 9.5. (E) H⁺ concentration dependence of giant patch current for the S44A/R366A channel (n = 3). pH_0.5 for wild type = 6.60 ± 0.01, and pH_0.5 for S44A/R366A = 8.11 ± 0.04. *, P < 0.05 compared with wild type. #, P < 0.05 compared with S44D. The dashed line indicates pH = 7.4.

Although the whole-cell current in the S44D/R366A double-mutant channel was similar to S44D and R366A single-mutant channels (Fig. 4B), the current with the S44A/R366A double mutant was lower than expected from its surface expression (Fig. 4C). Because ROMK channels are gated by cytosolic protons and various mutations have been shown to shift the pH sensitivity of ROMK channels, we examined whether enhanced pH sensitivity could account for the low current in the S44A/R366A double-mutant channel. Fig. 4D shows a representative patch recording from an oocyte expressing the S44A/R366A mutant channel. After patch excision into a bath at pH 7.4, channel activity was low but could be greatly increased by raising the bath pH to 9.5. The pH titrations of the wild-type and S44A/R366A mutant channels are shown in Fig. 4E. The pH curve for the S44A/R366A mutant channel was significantly left-shifted, i.e., exhibited enhanced sensitivity to pH (pH_0.5 for wild-type and S44A/R366A channels were 6.60 ± 0.01 and 8.14 ± 0.27, respectively]. The currents at pH 7.4 are shown by the dashed line in Fig. 7E. The current in wild-type channels is close to 100% of the maximal value, whereas that for the S44A/R366A channel was only ≈25% of the maximal current at pH 9.5. Thus, the enhanced pH sensitivity of the S44A/R366A channels can account for the low current observed in Fig. 4B for the S44A/R366A channel.

Discussion

Potassium secretion mediated by ROMK is critical to the normal handling of Na⁺ and K⁺ by the mammalian kidney, and the regulation of the surface density of ROMK is a major factor in modulating K⁺ secretion (3). Before this study, it was known that phosphorylation of S44 in the NH₂ terminus of ROMK1 (or S25 in ROMK2) by PKA or SGK1 increased the functional channel density in the oocyte plasma membrane (9, 10, 15, 18, 21), but the mechanism was unknown. Here, we report that the regulation of surface expression of ROMK1 results from S44 phosphorylation-induced increases in the rate of surface delivery of the channel. Our results show that the COOH-terminal R³⁶⁶XR modulates ROMK1 surface expression, consistent with this motif being an ER retention signal. Mutation of the COOH-terminal R³⁶⁶XR motif in ROMK1 increased surface expression of ROMK1 channels irrespective of the ability to phosphorylate S44. Importantly, the R366A mutation reversed the reduced surface expression of the S44A mutant channel to the level observed with S44D, which mimics the fully phosphorylated state. These results are consistent with phosphorylation of S44 masking the RXR ER retention signal, thereby stimulating the forward trafficking and surface delivery of ROMK1. Phosphorylation-induced ER export has also been observed for NMDA receptor channels (23, 24), glutamate receptor 5 kainate receptor channels (37), the KDEL receptor that mediates retrograde transport of proteins between the Golgi and the ER–Golgi intermediate compartment (38), and the hH1 human cardiac Na channel (25).

The phosphorylation switch mechanism used by ROMK or Kir1.1 to regulate surface delivery is fundamentally different from that used by Kir6.x (K_ATP) channels to regulate forward trafficking out of the ER. The COOH-terminal RXR ER retention motif that limits surface expression of K_ATP channels is suppressed by interaction with the accessory ABC protein SUR1/2 (sulfonylurea receptor protein) (34, 35) rather than by phosphorylation. The latter interaction in K_ATP channels masks ER retention signals on both proteins, ensuring delivery of properly folded channels with a 4:4 Kir/SUR stoichiometry. In contrast, ROMK channels can be expressed at the plasma membrane without an accessory subunit, and our results provide evidence that this depends on phosphorylation of the NH₂-terminal S44 in ROMK1.

The ability to regulate the forward trafficking of ROMK by phosphorylation-induced suppression of RXR-mediated ER retention would allow kidney epithelial cells to modulate the number of K⁺ channels in the apical plasma membrane in response to physiological demands. It is likely that other proteins participate in this regulatory process. Interaction with PDZ motif-binding proteins have also been reported to regulate the trafficking of glutamate receptors (39, 40) and ROMK (10, 15, 18). NHERF2 is present in nephron segments expressing ROMK and appears to be critical for full appearance of the higher surface expression by SGK1 in cells (15, 18). However, some increase in ROMK surface expression has been observed with SGK1 in oocytes in the absence of NHERF2 (10). NHERF1/2 has been reported to be critical for phosphorylation of the Na⁺–H⁺ exchanger and other transport proteins (41). Thus, it is possible that NHERF2 may be critical to interactions with certain kinases, like SGK1, that are involved in the phosphorylation switch of ER retention of ROMK in cells.

In glutamate (23, 40) and KDEL (38) receptors and the cardiac hH1 Na⁺ channel (25), the critical phosphorylation sites for suppression of RXR-mediated ER retention are just upstream or downstream of the RXR motif; thus, phosphorylation could easily modify the net charge in the vicinity of RXR. In contrast, the NH₂-terminal S44 in ROMK1 is considerably distant in linear sequence to the COOH-terminal R³⁶⁶XR motif. The question arises, then, How could phosphorylation of S44 suppress the RXR ER retention signal? The x-ray crystal structure of the prokaryotic Kir, KirBac1.1, shows that the NH₂ and COOH termini directly interact through hydrogen bonding (42), and this interaction has been suggested to be a component of channel gating (43). Importantly, with proper subunit–subunit folding, the NH₂- and COOH-terminal interaction is between adjacent subunits, and this intermolecular interaction could bring the phosphate on S44 close to the COOH-terminal RXR. Thus, NH₂-terminal phosphorylation modifying a COOH-terminal ER retention signal in ROMK1 could serve as a checkpoint for proper subunit folding critical to channel gating.