(Received for publication, August 30, 1996, and in revised form, October 15, 1996)
From the Departments of Cell Biology and Inflammation
Research, ¶ Cardiovascular Pharmacology and
Molecular
Biology, Pharmacia & Upjohn, Inc., Kalamazoo, Michigan 49007
The DNA sequence encoding the rat brain inward
rectifier-10 K+ channel was amplified from rat brain RNA
using reverse transcription-polymerase chain reaction and used to clone
the human homolog. Low stringency screening of a human kidney cDNA
library and subsequent DNA sequence analysis identified two related
K+ inward rectifier cDNAs, referred to as
Kir1.2 and Kir1.3, which were derived from
transcription of distinct human genes. Kir1.2 represents
the human homolog of the rat BIRK-10 sequence, whereas Kir1.3 was unique compared with all available sequence data
bases. The genes that encode Kir1.2 and Kir1.3
were mapped to human chromosomes 1 and 21, respectively. Both genes
showed tissue-specific expression when analyzed by Northern blots.
Kir1.2 was only detected in brain kidney and was
detected at high levels in all brain regions examined.
Kir1.3 was most readily detected in kidney and was also expressed in pancreas > lung. Comparative analysis of the
predicted amino acid sequences for Kir1.2 and
Kir1.3 revealed they were 62% identical. The most
remarkable difference between the two polypeptides is that the Walker
Type A consensus binding motif present in both Kir1.1 and
Kir1.2 was not conserved in the Kir1.3 sequence. Expression of the Kir1.2 polypeptide in
Xenopus oocytes resulted in the synthesis of a
K+-selective channel that exhibited an inwardly rectifying
current-voltage relationship and was inhibited by external
Ba2+ and Cs+. Kir1.2 current
amplitude was reduced by >85% when the pH was decreased from pH 7.4 to 5.9 using the membrane-permeant buffer acetate but was relatively
unaffected when pH was similarly lowered using membrane-impermeant
biphthalate. The inhibition by intracellular protons was
voltage-independent with an IC50 of pH 6.2 and a Hill coefficient of 1.9, suggesting the cooperative binding of 2 protons to
the intracellular face of the channel. In contrast, Kir1.3 expression in Xenopus oocytes was not detectable despite
the fact that the cRNA efficiently directed the synthesis of a
polypeptide of the expected Mr in an in
vitro translation system. Co-expression of Kir1.3
with either Kir1.1 or Kir1.2 reduced currents
resulting from expression of these inward-rectifier subunits alone,
consistent with a dominant negative influence on Kir1.1 and
Kir 1.2 expression.
Expression cloning of the ROMK1 (1), IRK (2), and G protein-regulated KGA (3) potassium channels defined a new structural class of potassium (K+) channels that can be regarded as simplified versions of the voltage-gated potassium channels. These potassium channel polypeptides share a homologous H-5 region believed to form an integral part of the K+-selective pore of the channel. In contrast to the predicted structure of the voltage-gated K+ channels, which contain six transmembrane domains and relatively small cytoplasmic domains, these inward rectifier K+ channels are predicted to contain only two transmembrane domains flanking the H-5 segment and a relatively large cytoplasmic COOH-terminal domain. The discovery of these cDNA sequences by expression cloning facilitated isolation and characterization of other members of this family of K+ channels by homology cloning (currently 12 distinct members). Classification of the predicted amino acid sequences of these K+ channels based on homology reveals four distinct subfamilies in which the ROMK (Kir1.1), IRK-1 (Kir2.1), KGA/GIRK-1 (Kir3.1), and Kir 6.1 channels are charter members. The Kir1 family has two members, Kir1.1 (ROMK) and a related polypeptide referred to as BIRK-10 (4), independently cloned from rat brain (BIRK-1 or KAB-2) (5, 6). The IRK family (Kir2) currently contains four distinct members including IRK-1 (Kir2.1) (2, 7, 8), IRK-2 (Kir2.2) (9, 10, 11), IRK-3 (Kir2.3), (12, 13, 14, 15, 16) and BIRK-9 (5). The GIRK family (Kir3) contains four members including GIRK-1 (Kir3.1) (3, 17), GIRK-2 (also referred to as KATP-2) (Kir3.2) (18, 19, 20, 21), GIRK-3 (Kir3.3) (18), and GIRK-4 (also referred to as cardiac KATP-1 and cardiac inward rectifier) (Kir3.4) (22, 23). Finally, a more distantly related K+ channel, referred to as uKATP-1 (Kir6.1), is the charter member of the fourth subfamily. This channel was cloned from rat pancreatic islets (24) and a second member, Kir6.2, was recently described (25).
Although our knowledge of the relationships between these cloned channels and their native counterparts is incomplete, there appear to be common functional features supporting this phylogenetic classification. Many of these polypeptides form inward rectifier K+ channels when expressed in heterologous systems. The Kir2 (IRK) family all form K+ channels that show strong Mg2+- and polyamine-dependent inward rectification. The Kir3 (GIRK) family members are regulated by G proteins (18) and are likely to form heteromultimers that have electrophysiological signatures distinct from the individual polypeptides (23, 26). Finally, the members of the Kir1 and Kir6 subfamilies are regulated by ATP (1, 5, 24, 25) suggesting these polypeptides are candidates for the pore-forming subunit of the KATP channel.
We previously reported the cloning and characterization of multiple isoforms of the human Kir1.1 (ROMK) channel from kidney that are formed by alternative splicing of a common gene (27). To expand our understanding of the function of other members of the Kir1 subfamily in the kidney, we used the BIRK-10 cDNA cloned from rat brain as a probe to clone the human homolog from a kidney cDNA library. In addition to the human homolog of rat BIRK-10, which we refer to as Kir1.2 to be consistent with the emerging nomenclature in this field (28), we also identified Kir1.3, another channel polypeptide that shows the greatest shared identity to Kir1.2 and is unique compared with the most recent release of the GenBank data base. These two human inward rectifier K+ channels were compared by determining their predicted amino acid sequences, their tissue distribution of expression, and their human chromosome assignment. In addition, recombinant Kir1.2 channels prepared by heterologous expression in Xenopus oocytes were characterized by electrophysiological analysis, including the regulation of channel activity by intracellular protons.
Total RNA (10 µg)
isolated from rat brain was reversed transcribed with an oligo(dT)
primer and Moloney murine leukemia virus reverse transcriptase (Life
Technologies, Inc.). A portion of the resulting cDNA was then
amplified in the PCR using a sense oligonucleotide primer (C12-A 5 CGC
TTT GAA TTC ATG ACA TCA GTT GCC AAG GTC TAT TA alignment positions
1-26 in X83858) and an antisense oligonucleotide primer (C12-B 5
CGC
TTT GAA TTC TCA GAC GTT ACT AAT GCG CAC ACT A, alignment positions
1116-1140 in X83858) specific for rat BIRK-10. Preliminary Northern
blot analysis using the rat BIRK-10 cDNA as a probe revealed that
in addition to the brain, BIRK-10 was also expressed in kidney.
Homologous cDNA clones were isolated from a human kidney cDNA
library using radiolabeled rat BIRK-10 DNA exactly as described
previously for human Kir1.1 (27). Fourteen distinct clonal
positives were grouped by a combination of PCR and restriction mapping,
and four representative clones were directly sequenced using the
-DNA templates by cycle sequencing as described previously (27). In
the case of Kir1.2, a genomic clone was also isolated from
a human genomic library in the vector
FIXII (Stratagene,
La Jolla, CA) using one of the partial cDNA clones obtained from
the human kidney cDNA library.
The
tissue distribution of expression of both Kir1.2 and
Kir1.3 was determined by Northern blot analysis using
Multiple Tissue Northern blots purchased from Clonetech (Palo Alto,
CA). For Kir1.2, a 1.3-kb EcoRI fragment
containing the 3 end of the open reading frame and approximately 0.6 kb of 3
-untranslated sequence was used as a probe. For
Kir1.3, a 1.4-kb EcoRI/MunI fragment
containing the entire open reading frame of Kir1.3 was used
as a probe. In either case, the fragments were random prime-labeled
using [
-32P]dATP (>3000 Ci/mmol, Amersham Corp.)
followed by hybridization to nylon membranes containing
poly(A)+ RNAs from various human tissues or human brain
regions as described previously (27). Somatic cell hybrid panel blots
were purchased from Oncor (Gaithersburg, MD) and hybridized to either
the Kir1.2 or Kir1.3 cDNA probes described
above, under the conditions recommended by the manufacturer (50%
formamide, 6 × SSC, 10% dextran sulfate, 1.0% SDS, 50 µg/ml
calf thymus DNA at 50 °C overnight). In all cases, the blots were
washed at high stringency (0.1 × SSC) and labeled bands
visualized by autoradiography.
The open reading frame of Kir1.2
was engineered for expression in Xenopus oocytes using the
PCR. Sense (5 CAG AAG TTA AGT CGA CAT GAC GTC AGT TGC CAA GGT GTA TT
3
) and antisense (5
CAG AAG TTA AGC GGC CGC (T)28 CAG ACA
TTG CTG ATG CGC ACA CT 3
) primers were used to PCR engineer the coding
region of Kir1.2 from the genomic clone for heterologous
expression as described previously (27). The product was digested to
completion with SalI and NotI and cloned into the
plasmid vector pSPORT-1 (Life Technologies, Inc.) to yield
pSPORT/Kir1.2. For Kir1.3, the original
-clone was double digested with EcoRI and MunI
and subcloned into the plasmid vector pBK-CMV to yield
pBK-CMV/Kir1.3. This plasmid was then double-digested with
EcoRI/AccI and directionally cloned into
pSPORT/rROMK1 to introduce the poly(A) tail from the rat ROMK1 cDNA
(1). In each case, cRNA was synthesized from NotI-linearized template using T7 RNA polymerase as described previously (27).
Heterologous expression of the
Kir 1 subunits was performed as described previously (27).
Oocytes were injected with 46 nl of polyadenylated cRNA dissolved in
water at 15, 30, or 100 ng/µl resulting in injection of 0.6, 1.2, or
4 ng of cRNA. Prior to recordings, oocytes were maintained 2-7 days at
20 °C in ND-96 supplemented with gentamycin (50 mg/ml;
Bio-Whittaker, Walkersville, MD) and sodium pyruvate (2.5 mM; Sigma). Electrophysiological recordings were conducted at room temperature. Oocyte resting membrane
potential was measured in ND-96 using 30-80 M glass microelectrodes. For two-microelectrode voltage-clamp recordings, the
voltage-measuring pipette had resistances of 1.5-2.0 M
, whereas the
current-injection pipette had resistances of 0.7-0.9 M
.
Microelectrode pipettes were filled with 3 M KCl. All
recordings were conducted in bath solutions containing 1 mM
MgCl2, 0.3 mM CaCl2, and 5 mM HEPES. For experiments conducted in 50 mM
K+, the solution contained 50 mM KCl and 50 mM NaCl. For lower potassium concentrations, this solution
was mixed with an identical solution in which choline was substituted
for potassium. Pairs of solutions at pH 5.4 and 8.4, containing 50 mM KCl and either 50 mM potassium acetate or 50 mM potassium hydrogen phthalate (biphthalate;
Sigma), were mixed to vary pH. Two-microelectrode
voltage-clamp currents were filtered at 2 kHz and sampled every 100 µs. Current amplitude was measured during the last 4 ms of a 400-ms
test pulse.
A routine query of the GenBank data base using the human
ROMK (Kir1.1) sequence and the BLAST search tool identified
a related K+ channel cDNA from rat brain (BIRK-10). A
human kidney cDNA library was screened at reduced stringency with
the rat BIRK-10 probe to identify the human homolog and related
cDNAs. Comparative analysis of the cDNA sequences revealed that
among the clones that exhibited high homology to rat BIRK-10, none of
them appeared to contain the entire open reading frame. In addition, a
second set of related cDNAs was also identified, and the latter
sequence was novel compared with the GenBank data base (Release 95).
Consistent with the emerging nomenclature in this field, we named these
two new human K+ channel cDNAs Kir1.2 and
Kir1.3 (
nward
ectifier channel), where the ROMK channel is referred to
as Kir1.1.
Using a combination of human kidney cDNA and genomic clones
(Kir1.2) or human kidney cDNA clones
(Kir1.3), the entire coding sequence of these two gene
products was defined. An alignment of the predicted amino acid
sequences of human Kir1.1, Kir1.2, and
Kir1.3 using the CLUSTAL algorithm is shown in Fig.
1. The human Kir1.2 sequence, derived from a
composite of both cDNA and genomic sequences, contains a 1137-bp
open reading frame that encodes a 379-amino acid polypeptide showing
91% shared identity with the rat BIRK-10 amino acid sequence and
represents the human species homolog of rat BIRK-10. The translation
initiation point was defined by alignment with the amino acid sequence
predicted from the rat cDNA (4, 5, 6). There was a second in-frame ATG
in the human genomic sequence 21 base pairs upstream of the initiator
methionine shown in Fig. 1, but transcripts containing this upstream
sequence could not be detected in RNA isolated from either human kidney
or brain using reverse transcription-PCR (data not shown). Consistent
with the predicted structures of other members of this family of
K+ channels, the sequence of human Kir1.2
contains two putative transmembrane domains that flank an H-5 region
forming an integral part of the K+-selective pore. Analysis
of the predicted amino acid sequence of Kir1.2 using the
MOTIF algorithm revealed two canonical Asn-linked glycosylation
acceptor sites (alignment positions 104 and 236), and six consensus
protein kinase C acceptor sites (alignment positions 178, 214, 321, 347, 361, and 374). Also, the predicted Walker Type A consensus binding
motif (GX4GKX7(I/V),
present in the Kir1.1 sequence, is conserved in the
Kir1.2 sequence.
The human Kir1.3 cDNAs predicted an 1125-bp open reading frame that encodes a 375-amino acid polypeptide (Fig. 1) 62% identical to human Kir1.2 and 47% identical to human Kir1.1. Two putative transmembrane domains flanking an H-5 region were also evident in the predicted amino acid sequence of human Kir1.3. Analysis of the predicted amino acid sequence of Kir1.3 using the MOTIF algorithm revealed two canonical acceptor sites for Asn glycosylation (alignment positions 103 and 284), four consensus protein kinase C sites (alignment positions 177, 213, 234, and 346), and two consensus tyrosine kinase acceptor sites near the COOH terminus of the protein (alignment positions 342 and 355). In contrast to the human Kir1.1 and human Kir1.2 sequences, the Walker Type A consensus binding motif GX4GKX7(I/V), is not conserved in the human Kir1.3 sequence.
Tissue Distribution of Expression and Human Chromosome AssignmentThe expression of Kir1.2 and
Kir1.3 in a panel of human tissues was profiled by Northern
blot analysis. Poly(A)+ RNA derived from 9 different
peripheral tissues and 9 brain regions were size-fractionated and
displayed on a solid support. Kir1.2- and
Kir1.3-specific transcripts were visualized by high
stringency hybridization to coding sequence probes derived from either
cDNA, and the results are shown in Fig. 2.
Kir1.2 showed a single major transcript of 5.2 kb that was
expressed at high levels in whole brain, much lower levels in kidney
and a second slightly larger transcript that was unique to brain
(left panel). No expression of Kir1.2
transcripts was detected in the other peripheral tissues examined. The
regional expression of Kir1.2 in brain showed similarly sized transcripts expressed in all regions examined with the highest levels in the corpus callosum (middle panel).
Kir1.3 also showed a limited tissue distribution of
expression (kidney > pancreas lung) and multiple transcript
sizes were detected (right panel). In contrast to
Kir1.2, which was most abundant in the brain, no Kir1.3 transcripts were detected in brain.
The human chromosome assignment of Kir1.2 and
Kir1.3 was determined by Southern blot analysis of
BamHI-digested genomic DNA prepared from a panel of
mouse/human or hamster/human somatic cell hybrids. As shown in Fig.
3, the Kir1.2 probe visualized a 6-kb
fragment on human chromosome 1 (top panel), whereas the Kir1.3 probe hybridized to a >10-kb band on human
chromosome 21 (bottom panel).
Heterologous Expression of Kir1.2 and Kir1.3 in Xenopus Oocytes
The K+ channels
expressed following injection of either Kir1.2 or
Kir1.3 polyadenylated cRNAs into Xenopus oocytes
were studied using standard electrophysiological procedures.
Forty-eight hours postinjection of 4 ng of Kir1.2 cRNA, the
oocyte resting membrane potential was hyperpolarized by 49 mV
(93 ± 3 mV (n = 25)) when compared with
water-injected oocytes (
44 ± 3 mV, n = 16, p < 0.001)). The observed resting membrane potential
of Kir1.2-injected oocytes was close to the calculated
potassium equilibrium potential of
106 mV at 2 mM
extracellular K+ and assuming 130 mM
intracellular K+. Thus, the Kir1.2-injected
cells expressed a potassium-selective ion channel that was not detected
in water-injected cells. In contrast, injection of oocytes with up to 4 ng of polyadenylated Kir1.3 cRNA did not significantly
alter the membrane potential compared with water-injected oocytes
(
49 ± 4 mV (n = 16) versus
44 ± 3 mV (n = 17), respectively).
Whole cell currents recorded by the two microelectrode voltage-clamp
technique from oocytes expressing Kir1.2 are shown in Fig.
4. Oocytes expressing Kir1.2 channels
displayed large inward currents in response to voltage steps negative
of the potassium equilibrium potential (20 mV in 50 mM
K+), and these currents were not detected in water-injected
cells. For example, whole-cell current measured during a test pulse to
140 mV from a holding potential of
20 mV in 50 mM
K+ was
4.4 ± 0.5 (n = 4) and
22.4 ± 4.9 µA (n = 7) in oocytes injected
with 1.2 or 4 ng of Kir1.2 cRNA, respectively, but was only
125 ± 13 nA in water-injected cells (n = 8).
Consistent with the lack of membrane hyperpolarization in oocytes
injected with Kir1.3 cRNA, the whole cell current amplitude
was similar to that measured in water-injected oocytes (
161 ± 27 nA at
140 mV, n = 4). Fig. 4 shows current traces
(panels A-C) and current-voltage curves (panel
D) from a single batch of oocytes injected with either
Kir1.2, Kir1.3, or water. Kir1.2
currents measured during voltage steps positive of the potassium
equilibrium potential were distinctly smaller than those measured
during steps negative of the K+ equilibrium potential,
creating a mild, inward rectification.
Some Kir channel pore-forming polypeptides fail to produce
detectable currents when expressed alone in oocytes. Co-expression of
Kir channel subunits has produced both reconstitution of
channel activity (25, 26), as well as decreases in currents compared with those produced by homomeric co-expression (45). The effect of
co-expression of Kir1.3 cRNA with suboptimal amounts of
either Kir1.1 or Kir1.2 cRNA (0.6 ng) was
determined, and the results are shown in Fig. 5.
Expression of Kir1.1 or Kir1.2 channels alone produced 1.14 ± 0.15 (n = 5) and
1.14 ± 0.34 (n = 5) µA of current, respectively, at a
holding potential of
160 mV. Co-injection of a 6.7-fold excess of
Kir1.3 decreased Kir1.1 and Kir1.2
currents by 54 and 51% to
0.52 ± 0.03 (p < 0.005, n = 5) µA and
0.56 ± 0.09 (p < 0.05, n = 4) µA, respectively.
Similar inhibitory interactions between Kir subunits have
been suggested to result from transmembrane domain-specific formation
of inviable complexes destined for degradation (45).
The Kir 1.2 Channel Forms a K+-selective, Ba2+-sensitive Channel Inhibited by Intracellular Protons
The effect of varying extracellular [K+] on
the current-voltage relationship of Kir1.2 expressing
oocytes was investigated and is shown in Fig.
6A. The dependence of reversal potential on
extracellular K+ was determined (Fig. 6B) and is
similar to that expected for a potassium-selective channel at room
temperature (slope-per-decade change of 55 mV). The membrane
conductance of Kir1.2 was dependent on extracellular
[K+]. The relationship g = C
[K+]oz relates conductance, g,
obtained by fitting the linear portion of the current-voltage curves
(dotted lines), to external potassium concentration.
C and Z were varied to produce the best fit of these data (Fig. 6C). The Z value for
Kir1.2 was 0.54, similar to the values of 0.62, 0.47, and
0.38-0.49 for the cardiac IK1 (28, 29), mouse macrophage IRK1 (3), and
human ROMK isoforms 1-3 (27), respectively. The square root dependence
of membrane conductance on extracellular K+ concentration
is typical of the multi-ion pore of inward rectifier K+
channels.
Application of Ba2+ (0.03 µM to 10 mM) produced a reversible, concentration-dependent inhibition of Kir1.2 current (Fig. 6D, n = 8). The inhibition by Ba2+ was time- and voltage-dependent as expected for the open channel block typical of the effects of Ba2+ on native and cloned inward-rectifier K+ channels (4, 6, 28, 29). The concentration dependence of Kir1.2 inhibition by Ba2+ at three different holding potentials is shown in Fig. 6E. These data were fitted by a simple logistic equation with a Hill coefficient of 2, and the resulting Ki values were plotted versus membrane potential in Fig. 6F. Fit of these data by the Woodhull equation (40) suggests that the Ba2+ binding site senses 46% of the transmembrane electric field, in reasonable agreement with previously published results in BIRK-10 injected oocytes (4). Inhibition of Kir1.2 by Cs+ was also reversible and voltage-dependent, being apparent only at the most negative potentials tested, but was far less effective than block by Ba2+ (n = 6, data not shown). Thus Kir1.2 encodes an inwardly rectifying, Ba2+- and Cs+-sensitive potassium channel.
Reducing pH inhibits the conductance of some native and cloned
K+ channels including ROMK1 (31), the mild inward-rectifier
most closely related to Kir1.2. The effect of changes in pH
on Kir1.2 currents was investigated, and the results are
shown in Fig. 7. When external pH was reduced using the
membrane-permeant buffer, acetate, Kir1.2 currents were
strongly inhibited. Inhibition by protons buffered by acetate reached a
maximum of 87 ± 1.7% at pH 5.9 (Fig. 7A,
n = 5). Alternatively, lowering external pH using the
membrane impermeant buffer, biphthalate, inhibited Kir1.2 currents by 23 ± 8.5% at pH 5.9 (Fig. 7B,
n = 4), suggesting that a low affinity proton-binding
site that weakly inhibits Kir1.2 current is present on the
external face of the channel. Previous measurements using pH-selective
microelectrodes demonstrated that cytosolic pH is reduced in synchrony
with extracellular pH when the bath is buffered with acetic acid,
whereas reducing bath pH with biphthalate does not alter cytosolic pH
(31). These data suggest the proton sensitivity of Kir1.2
is largely determined by a proton-binding site that faces the
intracellular, rather than the extracellular, environment. The less
potent and less efficacious inhibition mediated by extracellular
protons may reflect a distinct, external, proton-binding site. The
effect of pH on the Kir1.2 current is summarized in Fig.
7C in which the fraction of unblocked current is plotted
versus extracellular pH. The data were fitted by a simple
logistic equation giving a pKa and a Hill
coefficient of 6.2 and 1.9 for acetic acid and 5.9 and 2.6 for
biphthalate, respectively. A Hill coefficient of 1.9 for protons
buffered by acetate suggests the intracellular binding site may
accommodate two protons that inhibit cooperatively Kir1.2 currents. As shown in Fig. 7, D-F, inhibition of
Kir1.2 current by acetate-buffered protons was
voltage-independent between 120 and
40 mV. Fig. 7D shows
the current-voltage plots for Kir1.2 currents at various pH
values. In Fig. 7E, the Ki value for
inhibition by protons is plotted versus membrane potential. Fit of these data by the Woodhull equation (Fig. 7F)
revealed this proton-binding site senses only 3% of the transmembrane
electric field. Thus the proton-binding site on the cytosolic face of
Kir1.2 is not part of the membrane-embedded, pore-forming
region.
Alignment of the predicted amino acid sequences for Kir1.1, -1.2, and -1.3 polypeptides with predicted sequences of other cloned inward rectifiers confirms these channel polypeptides represent a subfamily of the inward rectifier family of K+ channels. Previous comparisons did not cluster the ROMK and BIRK10 sequences and hence, they were categorized as Kir 1.1 and Kir 4.1, respectively (45, 46). More recently, these sequences were grouped (5), and we extended this cluster to include the novel sequence encoded by Kir 1.3. Comparative analysis of these subfamily members provides additional information on which the subfamily can be classified. At the genomic level, the members of the Kir1 subfamily are encoded by distinct human genes that reside on three different chromosomes. They all share a similar genomic organization in which the majority of the polypeptide sequence is encoded by a single exon. In human Kir 1.1, a complex pattern of alternative splicing of five distinct exons gives rise to multiple transcripts, some of which add NH2-terminal amino acid extensions onto the core polypeptide (27). In human Kir 1.2, the entire polypeptide sequence presented in Fig. 1 is encoded by a single exon, but we have detected at least one additional exon in an expressed sequence tag derived from human hypothalamus. This exon splices into the genomic sequence 2 bps upstream of the initiator ATG but does not appear to extend the open reading frame (data not shown). Finally, the open reading frame of the human kidney Kir 1.3 cDNA also appears to reside on a single exon by genomic PCR, but we have not verified this result by sequence analysis. More detailed analysis of cDNA and genomic sequences will be required to determine the exact organization of the human Kir 1.2 and Kir 1.3 genes.
The Kir 1.1, Kir 1.2, and Kir 1.3 transcripts show tissue-specific expression in a limited number of tissues. All three genes are expressed in the kidney with Kir1.1 and 1.3 at the highest levels. Co-expression of all three Kir 1 channels in the same tissue suggests that channel diversity may be achieved through heteromultimer formation, and we have provided preliminary data that co-expression of Kir 1.3 with either Kir 1.1 or Kir 1.2 has a dominant negative effect on channel expression. Further experiments devised to compare recombinant heteromultimeric channels to their native counterparts will be required to determine if this is truly a mechanism for generating channel diversity. Kir 1.2 expression is the highest in brain where it is expressed in all brain regions examined. Our results on the tissue distribution of expression of Kir1.2 are in agreement with the expression pattern determined for the rat homolog (5, 6). Higher resolution visualization of rat Kir1.2 transcripts in the brain using in situ hybridization methods has shown a widespread expression throughout the brain with the highest concentrations in the brain stem. Microscopic examination of the staining pattern led to the suggestion that Kir1.2 is expressed in glial cells (6).
Functional analysis by electrophysiological methods of the channel polypeptides prepared by heterologous expression in Xenopus oocytes also showed a number of shared and divergent properties. The most striking of these properties was the robust expression of K+-selective/Ba2+-sensitive pores in the case of both Kir1.1 (27) and Kir1.2, while no functional expression of Kir1.3 could be detected by monitoring either cell membrane potential or whole cell currents. The lack of functional expression with the Kir1.3 channel is consistent with evidence emerging from functional expression of other inward rectifier K+ channels in the oocyte system. Recent studies show that individual members of the GIRK subfamily (Kir3.1, 3.2, 3.3, and 3.4) and members of the Kir6 subfamily (Kir6.1 and 6.2) do not form functional channels when expressed alone in Xenopus oocytes (8, 25, 26). Rather, expression of functional Kir3 K+ channels requires co-expression of at least two members of the subfamily while co-expression of the sulfonylurea receptors are required for the synthesis of functional Kir6.1 and 6.2 channels. We attempted to co-express both human Kir1.1 + Kir1.3 and Kir1.2 + Kir1.3 using suboptimal amounts of Kir1.1 or Kir1.2 paired with what should be saturating amounts of Kir1.3. Co-expression with Kir1.3 reduced currents resulting from homomeric expression of either Kir1.1 or Kir1.2. Similar inhibitory interactions between BIRK-10 and members of the Kir3 subfamily were observed and are suggested to result from the formation of unviable heteromultimeric complexes (45). Preliminary attempts to co-express Kir1.3 with the cystic fibrosis transmembrane regulator cRNA did not lead to expression of functional channels (data not shown).
As suggested by its predicted primary amino acid sequence, Kir1.2 forms a Ba2+- and Cs+-sensitive, mild inward-rectifier K+ channel when heterologously expressed in Xenopus oocytes. The time- and voltage-dependent block of Kir1.2 by Ba2+ and the square root dependence of Kir1.2 membrane conductance on extracellular K+ expected for a multi-ion pore are characteristic of the two-membrane-spanning inwardly rectifying class of K+ channels (4, 6, 28, 29, 31). The mild inward rectification of Kir1.2 is expected on the basis of a glutamate residue at position 159 shared with the rat homolog, BIRK-10 (4, 6). It is interesting to note that mutation of the wild-type asparagine to an aspartic acid residue, shorter than glutamic acid by only one methylene group, confers strong inward rectification on the Kir1.1 channel (36, 37, 38, 39).
Kir1.2 is inhibited by intracellular protons, with a pKa of 6.2. Two other pH-sensitive, weak inward rectifiers have also been reported, Kir1.1, with a pKa of 6.8, and RACTK1, with a pKa of 7.3 (43). Like Kir1.2, intracellular protons are required to inhibit Kir1.1 (31). The human strong inward rectifier, Kir2.3, is also sensitive to protons (42). The proton-binding site of Kir 2.3 appears to be structurally distinct from the Kir1.2 and Kir1.1 proton-binding sites. A series of chimera between Kir2.3 and Kir2.1 (IRK1), a closely related, proton-insensitive homolog, mapped the proton-binding site to the M1-H linker presumed to form the vestibule at the mouth of the channel pore in contact with the extracellular fluid (42). In contrast, the data presented here indicate the Kir1.2 proton-binding site, like that of Kir1.1, faces the cytosol. Kir1.1 (31) and Kir1.2 in the present study were both inhibited more potently and more effectively when pH was buffered by membrane-permeant acetate than when pH was buffered by membrane-impermeant biphthalate. External protons are predicted to have access to an M1-H site if one were present in Kir1.2. Inhibition of Kir1.2 by external protons was weaker and less potent than inhibition by internal protons. Thus, the dominant Kir1.2 proton-binding site appears to be structurally distinct from the proton-binding site of Kir2.3.
A number of studies implicated an acidic residue at position 171 in the predicted transmembrane domain 2 of Kir1.1 and Kir2.1 as the Mg2+ and polyamine binding site responsible for the inward rectification properties of these channels. The relatively acidic pH for Kir1.2 inhibition by protons and the requirement for internal protons to inhibit Kir1.2 suggested the glutamate residue present at this alignment position, residue 159 in Kir1.2, might be the titratable site responsible for the proton sensitivity of this channel. When a protonatable amino acid residue is engineered into this site in Kir1.1, the resulting proton block is highly voltage-dependent, similar to the block by Mg2+ and polyamines binding at that site 32-35 and reflecting the location of this residue deep within the channel pore (41). However, block of Kir1.2 by protons, similar to proton block of wild-type Kir1.1 and Kir2.3, is voltage-insensitive. Thus, Kir1.2 appears to possess a proton-binding site distinct from the cation-binding site of the channel pore. Three distinct proton-binding sites have been currently distinguished among these channels; they are a voltage-independent site mapped to an extracellular domain in Kir2.3, a voltage-dependent site in the channel pore in Kir1.1 (36), and a voltage-independent site that faces the cytosol in Kir1.1 (31) and Kir1.2 (this study).
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) U73191[GenBank], U73192[GenBank], U73193[GenBank].