A Hyperprostaglandin E Syndrome Mutation in Kir1.1 (Renal Outer Medullary Potassium) Channels Reveals a Crucial Residue for Channel Function in Kir1.3 Channels*

Loss of function mutations in kidney Kir1.1 (renal outer medullary potassium channel, KCNJ1) inwardly rectifying potassium channels can be found in patients suffering from hyperprostaglandin E syndrome (HPS), the antenatal form of Bartter syndrome. A novel mutation found in a sporadic case substitutes an asparagine by a positively charged lysine residue at amino acid position 124 in the extracellular M1-H5 linker region. When heterologously expressed in Xenopus oocytes and mammalian cells, current amplitudes from mutant Kir1.1a[N124K] channels were reduced by a factor of ∼12 as compared with wild type. A lysine at the equivalent position is present in only one of the known Kir subunits, the newly identified Kir1.3, which is also poorly expressed in the recombinant system. When the lysine residue in guinea pig Kir1.3 (gpKir1.3) isolated from a genomic library was changed to an asparagine (reverse HPS mutation), mutant channels yielded macroscopic currents with amplitudes increased 6-fold. From single channel analysis it became apparent that the decrease in mutant Kir1.1 channels and the increase in mutant gpKir1.3 macroscopic currents were mainly due to the number of expressed functional channels. Coexpression experiments revealed a dominant-negative effect of Kir1.1a[N124K] and gpKir1.3 on macroscopic current amplitudes when coexpressed with wild type Kir1.1a and gpKir[K110N], respectively. Thus we postulate that in Kir1.3 channels the extracellular positively charged lysine is of crucial functional importance. The HPS phenotype in man can be explained by the lower expression of functional channels by the Kir1.1a[N124K] mutant.

Kir1.1 (ROMK) channels are moderately expressed in many tissues including brain and heart but are present predominantly in kidney (15). Several splice variants of Kir1.1 cDNA have been isolated that give rise to at least three protein isoforms termed Kir1.1a, b, and c (ROMK1-3). They differ in their N-terminal sequence (15) and are differentially expressed along the nephron in the rat kidney (16). Kir1.1 subunits have been reported to associate with other Kir subunits (13) and proteins such as the cystic fibrosis transmembrane regulator, a member of the superfamily of ABC transporters (17). In heterologous expression systems, Kir1.1 yields mildly inwardly rectifying potassium channels that are subject to regulation by phosphorylation (18), intracellular ATP (2), and pH (19).
Kir1.1 channels are of particular functional importance in the kidney. The hyperprostaglandin E syndrome, a renal disorder resulting from impairment of tubular reabsorption, can be caused by either mutations in the furosemide-sensitive Na-K-2Cl cotransporter (NKCC2) or by mutations in Kir1.1 (20 -23). The two types of mutations lead to an impairment of transepithelial ion transport in the thick ascending limb of Henle's loop and have similar but not identical pathophysiological consequences. This renal disorder mimics long term furosemide treatment (24) and is characterized in the fetus by excessive saluresis and polyuria leading to polyhydramnios and premature birth. After birth, affected infants also suffer from the typical patterns of impaired tubular reabsorption in the thick ascending limb of Henle's loop (24,25). Characteristically, the strong stimulation of prostaglandin E 2 release re-sults in further aggravation of saluretic polyuria, secretory diarrhea, vomiting, mediation of fever, osteolysis, and failure to thrive (24). This renal disorder is inherited in an autosomal recessive manner and affects 1 in 50,000 -100,000 newborns (26).
Expression studies of several mutations in KCNJ1 revealed an almost complete loss of Kir1.1 channel function (23). In the thick ascending loop, this defect prevents luminal potassium recycling with secondary inhibition of the furosemide-sensitive Na-K-2Cl cotransporter, thereby disrupting electrogenic chloride reabsorption (27). The molecular mechanisms leading to nonfunctional Kir1.1 channels have not been defined in detail but may include incorrect targeting, abnormal pH regulation, misfolding, or occlusions of the permeant pathway. In the present report, we describe the clinical and functional analysis of a novel HPS mutation, N124K, 3 that affects an extracellularly localized residue of Kir1.1. A positively charged lysine at the equivalent position is present in none of the known Kir subunits except Kir1.3. Like Kir1.1, Kir1.3 is expressed predominately in kidney but also in lung and pancreas (13). After heterologous expression and functional characterization of Kir1.1a and Kir1.3 channels, we demonstrate that these lysine residues are responsible for impaired channel function in mutant Kir1.1a and for low macroscopic currents in Kir1.3 channels.

Mutation Analysis by Single-strand Conformation Polymorphisn
Analysis and DNA Sequencing-Genomic DNA was extracted from peripheral leukocytes. Aberrant band patterns for the KCNJ1 gene were sought by means of single-strand conformation polymorphism analysis (28) using the same primers as in a previous study (21). PCR was performed in a 20-l volume containing 50 ng of genomic DNA, 1.5 mM MgCl 2 , 5 mM Tris, pH 8.3, 50 mM KCl, 10 pmol of each primer, and 1.0 units of Taq polymerase. After an initial step at 94°C for 5 min, PCR was conducted for 30 cycles with denaturation at 94°C for 45 s, annealing at 55°C for 30 s, and extension at 72°C for 45 s. The reaction was completed with a final elongation step at 72°C for 10 min. Amplified products were separated using the CleanGel DNA analysis kit (Amersham Pharmacia Biotech) with the Multiphor II electrophoresis system (Amersham Pharmacia Biotech). Migration was performed at 18 watts constant power at 15°C for 1 h. The band patterns were visualized by the silver-staining method. Direct sequencing was performed after reamplification of the remaining PCR product using 5Ј-Cy5-labeled primers on an automated laser flourescence express sequencing system (Amersham Pharmacia Biotech) following the protocols provided by the manufacturer.
Reverse Transcription Amplification and Cloning of Human Kir1.1a (ROMK1) cDNA-Total cellular RNA from human kidney was a kind gift of Dr. Martin Kömhoff (Marburg, Germany). 2 g of the isolated RNA was reverse-transcribed with 30 units of M-MuLV reverse transcriptase (Fermentas) and 80 units of RNase inhibitor (Boehringer Mannheim) using a random hexanucleotide mixture, as described by the manufacturer. After 45 min at 42°C, reverse transcriptase was inactivated by incubation at 75°C for 5 min. For the amplification of the entire coding region of Kir1.1a, two PCR primers were chosen. The sense-primer (5Ј-CTTTCTGCAGCCATGAATGCTTCCAGTCGGAA-3Ј) contained the Kir1.1a start codon and a PstI restriction site, and the antisense-primer (5Ј-GAAAGTCGACTGTTACATTTTGGTGTCATCT-GTT-3Ј) contained the stop codon and a SalI restriction site. PCR reactions were performed with 2.5 units of AmpliTaq Gold Polymerase (Applied Biosystems) for 30 cycles (1 min, 94°C; 1 min, 52.5°C; 2 min, 72°C). The resulting 1.2 kilobase PCR product was digested with SalI and PstI and cloned in the COS-7 cell expression vector pSV Sport1. To verify the cloning procedure, DNA sequencing was performed using the ABI prism dye termination method and an ABI prism 377 DNA sequencer (Applied Biosystems).
Isolation of a Kir1.3 Clone from a Guinea Pig Genomic Library-A guinea pig FIX II genomic library (Stratagene) was plated on the Escherichia coli strain XL1-blue MRA (P2). A 605-bp fragment from a human Expressed Sequence Tag clone (GenBank accession number T94781/T94029, IMAGp998J11108, kindly supplied by the Resource Center of the German Human Genome Project, Max-Planck-Institute for Molecular Genetics, Berlin) was isolated after digestion with EcoRI and XhoI and labeled with digoxigenin-11 dUTP using a commercial kit (Boehringer Mannheim). Replicate nylon membrane filters representing about 5 ϫ 10 6 independent clones were screened at reduced stringency (2 ϫ SSC (1 ϫ SSC, 0.15 M NaCl and 0.015 M sodium citrate), 0.1% SDS, 60°C) using a standard protocol provided by Boehringer Mannheim. Pure clones were isolated after two rescreenings under the same conditions. DNA preparations (Qiagen, Hilden) were analyzed by restriction digestion and Southern blotting. Subsequently a 2188-bp BamHI fragment showing a strong positive signal in the Southern analysis was subcloned in the pBluescript SKϩ vector and sequenced using the ABI method as described above. For construction of expression vectors containing gpKir1.3, a PCR reaction was performed on the cloned gpKir1.3, amplifying only a 1179-bp fragment harboring the entire coding region. Sense (5Ј-CCCTAGGCGGCCGCAAGAATCG-GGAG-3Ј) and antisense (5Ј-GGCGCGGCCGCGAGACACCCTGGTG-TCA-3Ј) primers were designed for amplification. Primer internal NotI restriction sites were used for subcloning (see below).
Expression and Electrophysiological Analysis-Wild type (WT) and mutant Kir1.1a and Kir1.3 cDNAs were subcloned into the expression vector pSVSport1 (Life Technologies, Inc.) for expression in COS-7 cells. For expression in Xenopus laevis oocytes, cDNAs were subcloned into the polyadenylating transcription vector pSGEM (a gift from Dr. M. Hollmann, Göttingen). Capped run-off poly(A ϩ ) cRNA transcripts from linearized cDNA were synthesized, and ϳ6 ng was injected in defolliculated oocytes. Oocytes were incubated at 19°C in ND96 solution (96 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 5 mM HEPES, pH 7.4) supplemented with 100 g/ml gentamicin and 2.5 mM sodium pyruvate and assayed 48 h post-injection. Two-electrode voltage-clamp measurements were performed with a Turbo Tec-10 C amplifier (npi, Tamm, Germany) and sampled through an EPC9 interface (Heka Electronics, Lambrecht, Germany) using PULSE/PULSEFIT software (Heka) on a Macintosh computer, and data analysis was performed with IGOR software (WaveMetrics, Lake Oswego, Oregon). For rapid exchange of external solutions, oocytes were placed in a small-volume perfusion chamber with a constant flow of low K ϩ or high K ϩ solution (96 mM KCl, 2 mM NaCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 5 mM HEPES, pH 7.4).
Confluent COS-7 cells (ATCC number CRL1650) were transfected with 0.4 -1.0 g/ml Kir cDNA using LipofectAMINE and Opti-MEM I (Life Technologies, Inc.) following the manufacturer's protocol. Wholecell recordings were performed at room temperature 48 -72 h posttransfection in a bath solution consisting of 135 mM NaCl, 5.4 mM KCl, 1.8 mM CaCl 2 , 1 mM MgCl 2 , 10 mM glucose, 5 mM HEPES, pH 7.4. Patch pipettes were pulled from borosilicate glass capillaries (Kimble Products, Sussex, UK), Sylgard©-coated (Dow Corning, Corning, NY) and heat-polished to give input resistances of 4 -6 megaohms. The pipette recording solution contained 140 mM KCl, 2 mM MgCl 2 , 1 mM EGTA, 1 mM Na 2 ATP, 100 M cyclic AMP, 100 M GTP, and 5 mM HEPES, pH 7.3. Currents were recorded with an EPC9 patch clamp amplifier (Heka) and low pass-filtered at 2.9 kHz. Stimulation and data acquisition were also controlled by PULSE/PULSEFIT software. Series resistance of cells was routinely compensated by Ͼ80%, resulting in a maximum voltage error of 1-2 mV. Data are presented as mean Ϯ S.D. (number of cells).

RESULTS
Case Report and Mutational Analysis in KCNJ1-Replacement of an asparagine by a lysine at position 124 (N124K) in Kir1.1a (KCNJ1) was found in a heterozygous state in a sporadic case of hyperprostaglandin E syndrome (HPS). The pregnancy of the affected infant's mother was complicated by polyhydramnios and premature birth after 28 weeks of gestation. The child developed hypokalemic alkalosis, hyposthenuria, and hypercalciuria with subsequent nephrocalcinosis. Singlestrand conformation polymorphism analysis revealed two aberrant bands in the KCNJ1 gene of the patient. Direct sequencing demonstrated (i) a heterozygous base exchange of a thymine 232 adenine at base position 372 resulting in a missense mutation (N124K) and (ii) the insertion of an additional cytosine at base position 1055-58, causing a frameshift mutation with an altered C-terminal amino acid sequence (starting at His-354) and a premature stop codon at amino acid position 362 in the mutated protein. Cosegregation analysis identified N124K as the maternal, and the frameshift, as the paternal allele. Thus, the clinical diagnosis of hyperprostaglandin E syndrome was confirmed genetically.
Heterologous Expression of Wild Type Kir1.1a and Mutant Kir1.1a[N124K] Channels-To investigate the functional consequences of the novel HPS mutation N124K, WT, and mutant Kir1.1a channels were heterologously expressed in both Xenopus oocytes and mammalian COS-7 cells. As demonstrated earlier (23), WT Kir1.1a cRNA/cDNA in both systems gave rise to robust inwardly rectifying K ϩ currents with properties typical of weakly rectifying Kir channels (2). With the extracellular K ϩ concentration ([K ϩ ] e ) raised to 96 mM, Kir1.1a current amplitudes in oocytes averaged 50.6 Ϯ 8.0 A (n ϭ 9) at Ϫ100 mV membrane potential (Fig. 1A). When mutant Kir1.1a-[N124K] channels were expressed in oocytes, the kinetics of the macroscopic current were indistinguishable from WT Kir1.1a, but amplitudes were dramatically reduced to ϳ8% (3.9 Ϯ 1.2 A; n ϭ 5) under the same recording conditions (Fig. 1A). Expression of Kir1.1a[N124K] in COS-7 cells resulted in a similar reduction in amplitude as compared with WT Kir1.1a (135 Ϯ 24 pA, n ϭ 3 versus 1.12 Ϯ 1.1 nA, n ϭ 33). To estimate possible effects in heterozygous individuals, oocytes were injected with equal cRNA amounts of Kir1.1a and Kir1.1a-[N124K] (total amount equal to that used before). Surprisingly, coinjected oocytes averaged only 1.75 Ϯ 0.6 A (n ϭ 5), i.e. current amplitudes were also only a fraction of WT currents. This indicated a dominant-negative effect of the mutant Kir1.1a[N124K] subunits, and in heterozygous individuals, renders a total rescue of channel function by a simple heteromerization of WT and mutant subunits quite unlikely (see "Discussion").
Cloning and Sequence Analysis of a Guinea Pig Kir1.3 Subunit-It is also conceivable that in analogy to other Kir channels, kidney Kir1.1 subunits coassemble with other subfamily members expressed in the same cells. A second subfamily member, Kir1.2 (K AB -2; Kir4.1), is also expressed in the kidney and is functionally similar to Kir1.1a (13,14). In addition, using an uncharacterized Expressed Sequence Tag sequence (IMAGp998J11108), we isolated a novel full-length clone with strong homology to Kir1. 1 (30), respectively (Fig. 2). As phylogenetic analysis shows, both human and guinea pig sequences are more closely related to Kir1.1/Kir1.2 than to any other Kir subunit. Thus, we consider the novel sequence to represent a species ortholog of the human Kir1.3 (hKir1.3) and term it gpKir1.3 following a new emerging nomenclature for Kir channels. The main sequence variations between the two species were located in the first 20 amino acids in the N-terminal region, which were only ϳ45% similar. Sequence comparison showed 62.7% identity to Kir1.2 and 47.0% identity to Kir1.1 subunits (Fig. 2). Identity scores to members of other Kir subfamilies were significantly lower (Ͻ40%). Further analysis of gpKir1.3 recognized a C-terminal SNV motif also found in Kir1.2 subunits, indicating a putative interaction site with PDZ-domain proteins (31). An ATP binding site (Walker A motif), present in Kir1.1 and Kir1.2 (2,14), was absent from gpKir1.3.
Northern analysis of tissue total RNA identified strong expression of gpKir1.3 transcripts primarily in guinea pig kidney (data not shown). Thus gpKir1.3 was a likely candidate to associate with Kir1.1 subunits in renal Kir channels. hKir1.3 subunits isolated from human kidney remained uncharacterized since individually they failed to express in Xenopus oocytes (13). We noticed that in contrast to any other known Kir subunit, both hKir1. A (n ϭ 7), i.e. by a factor of ϳ6 (Fig. 1B), without noticeably changing other biophysical properties of macroscopic currents (see below). Robust whole-cell currents were also obtained after transfection of COS-7 cells with mutant Kir1.3[K110N] cDNA (968 Ϯ 405 pA, n ϭ 12), but only minor currents (100 -150 pA) were observed in few cells transfected with WT Kir1.3.
As shown from ramp and voltage-jump responses in varying concentrations of [K ϩ ] o , both WT and mutant Kir1.3[K110N] channels were highly selective for K ϩ ions with large amplitudes negative to the K ϩ Nernst potential E K . Measured Nernst (zero current) potentials were Ϫ83 mV for 5 mM, Ϫ67 mV for 10 mM, Ϫ43 mM for 25 mM, and Ϫ8 mV for 100 mM [K ϩ ] e (Fig. 3A), which was in perfect agreement with E K as predicted from the Nernst equation and followed [K ϩ ] e with a slope of approximately 54 mV per decade (Fig. 3B). These data indicated that the conductance was predominantly carried by K ϩ ions as expected for Kir channels. Other functional characteristics of gpKir1.3 were more similar to Kir1.2 channels. Unlike weakly rectifying Kir1.1, but not unexpected from the presence of a crucial negatively charged residue (E157) in the second transmembrane segment (Fig. 2) that is directly involved in the binding of Mg 2ϩ or polyamines (32), Kir1.3 rectified more strongly and displayed only moderate outward currents (Fig.  3C). Interestingly, both in Kir1.1 and in Kir1.3 channels, the HPS mutation and reverse HPS mutation, respectively, did not change the rectification properties of the channels. Rectification of the Kir1.3 I-V relation remained identical with either 2 mM Mg 2ϩ or 0 mM Mg 2ϩ , 100 M EGTA in the internal solution (data not shown) as has been shown for Kir2.1 channels (33). As a glutamate at position 157 may not suffice for total rectification as is the case in Kir2 channels (36), additional residues are likely involved in the rectification mechanism.
When analyzing the block by extracellular Ba 2ϩ , we found that in some aspects the two subfamily members Kir1. A and B). In contrast, Kir1.1a and mutant Kir1.1a[N124K] were blocked by 1 mM Ba 2ϩ with a unique rapid on and off rate ( ON ϭ 124 Ϯ 74 ms; OFF ϭ 230 Ϯ 71 ms; n ϭ 6). A concentration-response analysis of macroscopic steady-state currents determined in oocytes revealed that WT Kir1.3 channels (K i ϭ 16.8 M) were ϳ5-fold more sensitive to Ba 2ϩ than Kir1.1a channels (K i ϭ 84.9 M; Fig. 5A and B). Interestingly, replacement of the negatively charged residue in Kir1. 3 ; Fig. 4A), introducing a lysine in the HPS Kir1.1a[N124K] channels lowered the Ba 2ϩ sensitivity (K i ϭ 282 M) by a factor of Ͼ3 (Fig. 4B). A homologous residue in Kir2.1 channels had been implicated before in affecting Ba 2ϩ sensitivity (35), suggesting that this site is in the vicinity of the outer channel pore and likely interacts with cationic channel blockers in a subunit-specific manner.
Single Channel Analysis of Kir1.1a and Kir1.3 Channels-After cRNA injections in oocytes of both WT and mutant Kir1.3 channels, single-channel activity was recorded in the cell-attached configuration with 140 mM K ϩ in the pipette (Fig. 6). As deduced from current responses to hyperpolarizing voltage pulses, the unitary slope conductance of WT Kir1.3 channels was ␥ ϭ 26 Ϯ 1.7 pS (n ϭ 6). A similar elementary conductance was found in Kir1. 3[K110N] channels (␥ ϭ 27.5 Ϯ 1.8 pS; Fig.   6A). These values were only slightly smaller than the unitary conductances of Kir1.1 (30 pS (2)) and the high conductance state of Kir1.2 (4.1) channels (36 pS (14)). As reflected in the macroscopic currents (Fig. 3C), the current-voltage relationship of single Kir1.3 channels was strongly rectifying. Further quantitative analysis revealed an open probability of WT Kir1.3 channels of p o ϭ 0.52 Ϯ 0.05 (n ϭ 6) at a membrane potential of Ϫ100 mV and a value in the same range for mutant Kir1.3 [K110N] channels (p o ϭ 0.48 Ϯ 0.04; n ϭ 6; Fig. 6B). Thus the product p o ⅐ ␥ as a measure of average current flux through single channels was virtually identical between WT and mutant Kir1.3 channels. The ratio of macroscopic currents, however, was ϳ1:6, suggesting that current potentiation in Kir1.3[K110N] channels was mainly due to elevated expression of functional channels (Fig. 6C). The equivalent analysis performed for WT and mutant Kir1. Heteromerization of Kir1.1a and Kir1.3 Channels-With respect to a potential heteromerization between Kir1.3 and Kir1.1 subunits in the kidney, Shuck et al. (13) reported that hKir1.3 inhibited expression of both Kir1.1 and Kir1.2 channels by ϳ50%, which was interpreted as being caused by the formation of unviable heteromeric channel complexes. Analysis of current amplitudes and using the above-mentioned criteria of current rectification and Ba 2ϩ block oocyte coinjection experiments with WT and mutant Kir1.1a and gpKir1.3 subunits revealed different results for gpKir1.3 in our experiments. When Kir1.1a cRNA was injected together with Kir1.3 cRNA (1:10 dilution to titrate amplitudes), macroscopic current amplitudes were not significantly different from those induced by Kir1.1a cRNA injection alone (Fig. 7A). However, in contrast to channels solely composed of Kir1.1 subunits, current properties after coinjections were dominated by Kir1.3, i.e. they were uniformly strongly rectifying and showed a Ba 2ϩ block with a slow OFF rate and high Ba 2ϩ affinity (K i ϭ 14. position in Kir2.1 subunits was implicated in being responsible for functional differences between the human and the chicken ortholog of the subunit (35). The exchange of a negatively charged residue (Glu-125 in human Kir2.1) to an uncharged residue (Gln-125 in chicken Kir2.1) reduced the single channel conductance and lowered the sensitivity to Ba 2ϩ block (35), suggesting that this extracellular site might be in close vicinity to the conduction pathway in the channel. Second, in the novel Kir1.3 channels, a positively charged lysine (Lys-110) aligns with Asn-124 in Kir1.1a subunits. Since initial attempts failed to elicit currents from individually expressed hKir1.3 subunits (13), we suspected a similar mechanism to underlie the loss of function HPS mutation Kir1.1a [N124K].
Indeed Kir1.1a[N124K] currents were strongly decreased and Ba 2ϩ sensitivity lowered Ͼ3-fold compared with WT channels. By analogy, the reverse HPS mutation K110N introduced in gpKir1.3 gave rise to a strong macroscopic current amplitude ϳ6-fold larger than for WT gpKir1.3 channels. Thus, exchange of a positively charged by an uncharged residue near the extracellular pore region of both Kir1.1a and Kir1.3 channels was accompanied by a strong rise of macroscopic current amplitudes. With both single channel conductance and open channel probability of Kir1.1a[N124K] and gpKir1.3 WT channels virtually unchanged by the mutation, the underlying molecular mechanism for this impaired function was found to be primarily due to altered expression of functional channels. We suspect that the extracellular Lys-110 residue in Kir1.3 channels (sporadically mutated in HPS-Kir1.1) may be of crucial importance for channel assembly.
Furthermore the Lys-110 residue may be important in determining the functional properties of the pore. The exposed extracellular localization of these residues open the possibility for a direct influence of permeant and nonpermeant ions or pH on the regulation of channel open probability. The HPS and reversed HPS mutation in Kir1.1a and gpKir1.3 subunits, respectively, did neither noticeably affect single channel conductance nor open channel probability. Mutant and WT gpKir1.3 channels had a higher sensitivity to Ba 2ϩ compared with Kir1.1 channels, and the typically slow unbinding of Ba 2ϩ may result from the blocking cation being trapped in the pore. In contrast, as reflected by the fast Ba 2ϩ unblocking kinetics in Kir1.1a[N124K] mutant channels, the negatively charged lysine substantially lowered the sensitivity to Ba 2ϩ. . This indicates that although the exact function of these residues slightly varies between Kir channels, this position at the outer channel pore may electrostatically interact with blocking cations.
Our coexpression studies of Kir1.1a and Kir1.3 have shown that the current amplitude was dominated by Kir1.1a (it was unchanged compared with expression of Kir1.1a alone and much larger than Kir1.3 alone), whereas the shape of the current voltage relation and the kinetics of Ba 2ϩ block were dominated by Kir1.3. This strongly suggests that heteromultimers of Kir1.1a and Kir1.3 can form in vitro and possibly in vivo as well. We found a dominant-negative effect of the extracellular lysine residue on composite currents when a subunit with lysine (Kir1.1a[N124K] or gpKir1.3) was coexpressed with a subunit with an asparagine at the respective position (Kir1.1a, Kir1.3[K110N]). From these findings one might expect that individuals expressing the HPS mutation heterozygously would suffer from impairment of tubular reabsorption like homozygous HPS patients. This is apparently not the case. There are at least two explanations for this surprising finding. (i) If a single subunit carrying a lysine is sufficient to prevent the formation of a functional tetrameric protein and subunits assemble in a random manner, then a dominant-negative effect is expected since statistically only one in 16 channels would lack a lysine-carrying subunit. Nevertheless, if the K ϩ recycling mechanism is not saturated, one of 16 channels might still be sufficient for epithelial cell function. (ii) The dominantnegative effect of Kir1.1a[N124K] in the presence of WT Kir1.1a might be exaggerated in vitro as compared with in vivo conditions. This could be due, for example, to the lack of some physiologically important interaction partner in the expression system. It should be mentioned that dominant-negative suppression of another potassium channel (KvLQT1) has recently been shown in patients with the autosomal recessive Jervell and Lange-Nielson syndrome, although the dominant-negative regulation by mutant KvLQT1 was much less pronounced (7). Rescue of channel function of heterozygous Kir1.1a[N124K] by Kir1.3 is unlikely since coexpression of these two subunits also yielded low current amplitudes. Similar dominant-negative effects have also been found by coexpression of other HPS mutations with Kir1.1a WT. 4 To date the functional role of Kir1.3 channels is still elusive because heterologous expression of Kir1.3 alone yields only small currents. On the other hand, coexpression of Kir1.3 and Kir1.1a results in large currents and generates K ϩ channels with distinct properties that clearly differ from channels obtained by expression of Kir1.3 alone. Because Kir1.1 and Kir1.3 are both strongly expressed in the kidney, it is tempting to speculate that the functional form of Kir1.3 may be a heteromultimer with Kir1.1a and/or possibly other channels of the Kir family. It should be noted that Kir1.3 mRNA has also been found in pancreas and lung in the adult and in several fetal tissues including brain and may thus, in connection with other subunits, also play a role in the developing central nervous system.