Functional and developmental expression of a zebrafish Kir1.1 (ROMK) potassium channel homologue Kcnj1


  • L. Abbas and S. Hajihashemi contributed equally to this work.

Corresponding author S. J. White: Department of Physiology, Ross University School of Medicine, Commonwealth of Dominica, West Indies.  Email:


Non-technical summary  Due to the conservation of developmental pathways and genetic material over the course of evolution, non-mammalian ‘model organisms’ such as the zebrafish embryo are emerging as valuable tools to explore causes and potential treatments for human diseases. Ion channels are proteins that form pores and help to establish and control electrical gradients by allowing the flow of ions across biological membranes. A diverse range of key physiological mechanisms in every organ in the body depends on the activity of ion channels. In this paper, we show that a potassium-selective channel that underlies salt reabsorption and potassium excretion in the human kidney is also expressed in zebrafish in cells that are important regulators of salt balance. Disruption of the channel's expression in zebrafish leads to effects on the activity of the heart, consistent with a role for this channel in the control of potassium balance in the embryo.


Abstract  The zebrafish, Danio rerio, is emerging as an important model organism for the pathophysiological study of some human kidney diseases, but the sites of expression and physiological roles of a number of protein orthologues in the zebrafish nephron remain mostly undefined. Here we show that a zebrafish potassium channel is orthologous to the mammalian kidney potassium channel, ROMK. The cDNA (kcnj1) encodes a protein (Kcnj1) that when expressed in Xenopus laevis oocytes displayed pH- and Ba2+-sensitive K+-selective currents, but unlike the mammalian channel, was completely insensitive to the peptide inhibitor tertiapin-Q. In the pronephros, kcnj1 transcript expression was restricted to a distal region and overlapped with that of sodium–chloride cotransporter Nkcc, chloride channel ClC-Ka, and ClC-Ka/b accessory subunit Barttin, indicating the location of the diluting segment. In a subpopulation of surface cells, kcnj1 was coexpressed with the a1a.4 isoform of the Na+/K+-ATPase, identifying these cells as potential K+ secretory cells in this epithelium. At later stages of development, kcnj1 appeared in cells of the developing gill that also expressed the a1a.4 subunit. Morpholino antisense-mediated knockdown of kcnj1 was accompanied by transient tachycardia followed by bradycardia, effects consistent with alterations in extracellular K+ concentration in the embryo. Our findings indicate that Kcnj1 is expressed in cells associated with osmoregulation and acts as a K+ efflux pathway that is important in maintaining extracellular levels of K+ in the developing embryo.


chloride channel


hours post fertilization


inwardly rectifying potassium channel


morpholino antisense oligonucleotide


sodium–chloride cotransporter


sodium–potassium–chloride cotransporter


protein data bank


renal outer medullary potassium channel


thick ascending limb of Henle's loop


two-electrode voltage clamp




In mammals, the kidney is the major organ underlying K+ homeostasis, which is determined via regulated secretion of K+ by cells of the distal tubule and collecting duct. In addition, cells in the thick ascending limb of Henle's loop (TAL) secrete potassium into the tubular lumen, a process that underpins both the primary dilution of tubular fluid and the concentration of the urine during water deprivation (Hebert et al. 2005). In the collecting duct, secretion of K+ determines urinary potassium excretion, but in the TAL, K+ secretion enables recycling of K+ between the cell and lumen, coupling the action of transporters and channels that underlie reabsorption of NaCl by this nephron segment (Wang et al. 1990). Potassium secretion in the distal nephron occurs predominantly via a class of ion channels known as inward rectifiers (Kir channels). Isoforms of the Kir1.1 (ROMK) family are expressed in the apical membranes of cells of the TAL, the distal and connecting tubule, as well as the cortical and outer medullary collecting duct (Xu et al. 1997).

The zebrafish, Danio rerio, is a small freshwater teleost that has emerged as a powerful model of a number of human diseases and has great potential for studies of vertebrate integrative physiology (Briggs, 2002; Shin & Fishman, 2002). However, details of the functions of identified channels and transporters in ion transport by zebrafish embryonic renal tubules are largely unknown. The experiments described in this study were designed to determine the basic properties and sites of expression of a ROMK-like channel of the zebrafish as a first step in assessing the suitability of Danio as a model with which to investigate the pathogenesis of renal ion transport-related diseases of humans.


Work with animals

All experiments with Xenopus laevis and Danio rerio were carried out under the guidelines described by Drummond (2009) and approved by legislation and local institutional animal welfare committees as indicated.

Cloning of kcnj1

A database search identified a partial clone in a zebrafish EST library with similarity to human Kir1.1 (kcnj1: Genbank Accession No. BF157829). In this study we follow the nomenclature guidelines for zebrafish genes (, so that gene and transcript are all lowercase and italicized (i.e. kcnj1), and the protein not italicized, and with the first letter uppercase (i.e. Kcnj1). The clone was sourced from the Integrated Molecular Analysis of Genomes and their Expression (IMAGE) Consortium (No. 3815817) and was provided in the plasmid pME18S-FL3. The full length clone was sequenced and the open reading frame was subcloned into the vector pTLN (Lorenz et al. 1996).

Sequence analysis

The genome sequence database of zebrafish (Ensembl genome, was used for BLAST (basic local alignment search tool) searching. The deduced protein amino acid sequence and splicing information of kcnj1 was obtained from VEGA (

Structural modelling

The query-template alignments were produced with Praline-TM alignment software built especially for transmembrane proteins (Pirovano et al. 2008). These alignments were then input to Modeller in PIR format. Structural models were created using Modeller 9v7 (Sali & Blundell, 1993) employing the ‘MyModel’ method to allow tetrameric proteins of the correct geometry (as indicated by the symmetry transformations within the template protein data bank (PDB) files) to be created. Both Kcnj1 and rat ROMK2 (rROMK2) models were based upon a Kir3.1-KirBAC1.3 chimera protein (Nishida et al. 2007), the structure of which has been solved by X-ray crystallography (PDB code 2qks) and has the highest sequence identity (35%) with both Kcnj1 and rROMK2. Three models were created of each channel. The lowest energy conformation of each channel was then selected to undergo further rounds of energy minimization via 200 cycles of steepest descent, performed with Swiss-pdbviewer, implementing the GROMOS 43B1 force field.


We determined the properties of macroscopic currents by conventional two-electrode voltage clamp (TEVC: Leipziger et al. 2000). Oocytes were obtained from mature female Xenopus laevis killed humanely using a procedure in accordance with the Home Office Guidance on the Operation of the Animals (Scientific Procedures) Act of 1986. Briefly, frogs were immersed in anaesthetic 0.2% (w/v) tricaine methanesulfonate (supplemented with 5 mm Hepes, pH 7.80) until unconscious. The animals were then killed by decapitation and destruction of the spinal cord. Standard protocols were followed for the isolation and care of oocytes (Leipziger et al. 2000).

For expression studies, defolliculated stage V/VI oocytes were injected (50 nl) with 1.6 ng of capped (mMESSAGE mMACHINE SP6 kit, Ambion) kcnj1 mRNA, 1 ng of rat ROMK2 mRNA (positive control), or an equal volume of water. Injected oocytes were incubated at 18°C in OR3 solution containing 6.85 g l−1 of Leibovitz L-15 cell culture medium plus 10,000 units ml−1 penicillin G sodium and 10,000 μg ml−1 streptomycin sulphate (Invitrogen), and 5 mm Hepes at pH 7.5. After 48–72 h, TEVC was performed using a Gene-Clamp 500B amplifier (Axon Instruments, Union City, CA, USA). Data were acquired using Clampex (pCLAMP, Axon Instruments, version 6) on an IBM compatible PC (Gateway) equipped with an analog-to-digital interface (Digidata 1200, Axon Instruments). The initial holding potential was maintained at −50 mV, and this was ramped from −120 to +100 mV in steps of 20 mV, and held for 50 ms at each test voltage and between each step voltage returning to –50 mV. For each recording this protocol was repeated 5 times and the signals averaged. Currents were determined mid-way through the time-independent phase of the current trace. Unless stated otherwise, experiments were performed at room temperature in amphibian Ringer solution (ND96) containing (in mm): NaCl (96), KCl (2), CaCl2 (1.8), MgCl2 (1), Hepes (5) at pH 7.5.

Zebrafish embryos

The work with zebrafish described in this paper was carried out under UK Home Office Licence regulations and approved by the University of Sheffield Project Applications and Amendments Committee. Wild-type adult zebrafish (AB strain) were maintained on a 14 h light–10 h dark cycle and embryos obtained from mass matings. Developing embryos were maintained at 28.5°C in E3 medium, containing (in mm): NaCl (5), KCl (0.2), CaCl2 (0.3), MgCl2 (0.3), KH2PO4 (0.05) and Na2HPO4 (0.29), pH 7.0–7.2. Embryos were staged by standard criteria (Kimmel et al. 1995) in hours post fertilization (hpf).

In situ hybridization of kcnj1 and other transcripts

Embryos were fixed in 4% paraformaldehyde/PBS for 3 h at room temperature and stored in methanol at −20°C. In situ hybridisation for kcnj1, the zebrafish orthologue of the renal chloride channel ClC-Ka (NM_200382), Na+/K+-ATPase a1a.4 (AY008376), Barttin (XM_001332980) and the thiazide-sensitive sodium chloride cotransporter NCC (NCCT) (NM_001045080) was carried out according to established protocols (Oxtoby & Jowett, 1993) using digoxigenin- or fluorescein-labelled RNA probes (all reagents from Roche). Double-labelled embryos were incubated overnight with anti-fluorescein-AP antibody 1:3000 in blocking solution at 4°C before washing and staining with Fast Red. The first antibody was inactivated with 0.1 m glycine-HCl, pH 2.2 in 0.1% Tween-20 at room temperature for 10 min. After washing, anti-digoxygenin-AP antibody 1:3000 was applied overnight and the embryos washed. The colour reaction was carried out in the dark in 4.5 μl ml−1 4-nitro blue tetrazolium chloride and 3.5 μl ml−1 5-bromo-4-chloro-3-indolyl-phosphate (BCIP) in 0.1 m Tris pH 9.5, 50 mm MgCl2, 0.1 m NaCl, and 0.1% Tween (PBTw). Single staining was performed essentially as above, omitting the fluorescein-specific steps. Staining reactions were quenched with PBTw washes followed by 2 h of 4% paraformaldehyde fixation and the embryos dehydrated through a glycerol series to 90% for mounting and photography.


Embryos were fixed in Dent's solution (80% methanol, 20% DMSO) for 2 h at room temperature, followed by washing in PBTw and dehydration through a progressive (25% to 100%) methanol series. Following rehydration, embryos were blocked in 10% bovine serum albumin and incubated with a 1:1000 dilution of an anti-Nkcc monoclonal antibody (T4: Developmental Studies Hybridoma Bank, Iowa) overnight at 4°C. Samples were washed and incubated in a 1:200 dilution of anti-mouse peroxidase-conjugated secondary antibody (Sigma) overnight. Embryos were washed and the colour reaction developed using a diaminobenzidine (DAB) kit (Vector Labs) followed by a glycerol series for mounting and photography.

Morpholino antisense injections

Morpholino oligonucleotides (Gene Tools, LLC) were dissolved in water to a stock concentration of 10 mg ml−1. Sequences were: kcnj1 morpholino: 5′-CTCTCTCAAGGAGCGAGTCATCTTA-3′; mismatch control morpholino: 5′-CTgTCTgAAGGAcCGAcTCATgTTA-3′. Injections were carried out using a microinjection rig (Narishige) into the yolk of 1–8 cell stage wild-type embryos (0.5–2 nl per injection).

Heart rate measurements

Heart rates in beats per minute were determined in unanaesthetized 24 hpf and 48 hpf embryos by measuring the time (by stopwatch) taken for 15 ventricular contractions observed under a stereomicroscope. Determinations were performed in triplicate for each embryo.

Data presentation and analysis

Unless otherwise indicated, data are presented as means ±s.e.m. Comparisons were made between means by ANOVA or Student's unpaired t test as appropriate and values are presented as significant when P was <0.05 (n.s. signifies non-significance).


Identification of homologues

Analysis of the nucleotide sequence: ( confirmed that kcnj1 contained a 1113 bp open reading frame preceded by 172 5′ untranslated bases encoding a 370 amino acid protein (Kcnj1) of molecular mass 42,486 Da. Clustal-W alignments with all known vertebrate Kir1.1 sequences, and a phylogenetic comparison of Kcnj1 with other inward rectifiers, demonstrated that Kcnj1 showed greater similarity with the Kir1.1 family than with other inward rectifiers (see Fig. S1A in Supplemental material, available online only). A comparison with human (h) Kir1.1 isoforms showed that Kcnj1 shares greater than 55% identity with Kir1.1 proteins (Fig. S1B in Supplemental material). Amino acid alignments of Kcnj1, rROMK2 (used as a positive control for the functional studies) and the template protein 2qks are shown in the online Supplemental material (Fig. S2). Kcnj1 contains a number of potentially phosphorylatable residues, some of which are shared with those of rROMK2. In contrast to the mammalian channel, Kcnj1 lacks an N-linked glycosylation site (N98 in rROMK2: G in Kcnj1). G98 is part of an insertion LWQNPPPGH at residues 91–99 in Kcnj1 that is not shared either by rROMK2 or 2qks (Supplemental Fig. S2). The predicted secondary structure of Kcnj1 is shown in Fig. 1A and, as expected, is consistent with a two transmembrane spanning protein characteristic of Kir channels. The predicted structural models of Kcnj1 and rROMK2 are shown in Fig. 1B (Kcnj1) and C (rROMK2).

Figure 1.

A, sequence of Kcnj1 with predicted secondary structures. Block arrows: β sheets. Waved blocks: α helices. The predicted helical transmembrane segments are waved boxes with diagonal line shading. Continuous lines indicate loop regions; dotted lines indicate parts of the sequence that could not be structurally modelled. Residues in red indicate predicted insertion in loop sequence in Kcnj1. Bold, underlined residues are those in disallowed areas of the Ramachandran plot. Bold, boxed residues are energetically unstable according to the GROMOS force field. Lower panels show multimeric structural models for Kcnj1 (B) and rROMK2 (C), with one subunit in each structure coloured blue for clarity.

Functional properties of Kcnj1

At an external [K+] of 2.00 mm, Xenopus laevis oocytes injected with rROMK2 displayed a resting membrane potential (Vm) of −96.00 ± 1.36 mV (n= 16 oocytes from three separate animals). Similarly, in oocytes expressing Kcnj1, Vm was −92 ± 1.52 mV (n= 16). Under the conditions applied, the predicted reversal potential for a K+-selective membrane was −96.0 mV, indicating that Kcnj1 is a potassium-selective channel. Water-injected oocytes displayed a Vm of −26.20 mV (n= 13). We determined current–voltage relationships of Kcnj1 by TEVC using the voltage protocol shown in Fig. 2A. From currents recorded from oocytes expressing Kcnj1 (Fig. 2B) and rROMK2 (Fig. 2C) we calculated Ba2+-sensitive current by subtraction of current measured at each potential in the presence of Ba2+ from the total current at each potential in the absence of Ba2+. This showed that both channels displayed weak inwardly rectifying currents under these experimental conditions (Fig. 2D and E). However, expression of current was consistently lower for Kcnj1 than for rROMK2 (Fig. 2F).

Figure 2.

Currents expressed in Xenopus oocytes
A, the voltage protocol used. B and C, examples of current recordings from an oocyte expressing Kcnj1 (B) and an oocyte expressing rROMK2 (C) in the absence (upper traces) or presence (lower traces) of 5 mm BaCl2. Horizontal dotted lines indicate zero current. Current values determined at each voltage at the points indicated by the dashed vertical lines were used to plot current–voltage (IV) curves for Kcnj1 (D) and rROMK2 (E): filled circles, −Ba2+; open circles, +Ba2+; inverted triangles, the difference (Ba2+ sensitive). F, mean current at 0 mV for control (H2O-injected) oocytes and oocytes expressing Kcnj1 or rROMK2 (*P < 0.01, n= 13–16 oocytes from at least three separate animals).

Both channels were inhibited by Ba2+ in a dose-dependent manner (Fig. 3). For Kcnj1 (Fig. 3A), the calculated Ki for Ba2+ was 2.42 ± 0.16 mm, which was similar to rROMK2 (Fig. 3B; 2.22 ± 0.53 mm: n.s., n= 9 in each case). As reported earlier for rat ROMK1 (Löffler & Hunter, 1997), Ba2+ inhibition of rROMK2 was voltage dependent. However, for Kcnj1 the voltage dependency was not evident.

Figure 3.

Voltage dependence of Ba2+ inhibition on expressed currents
Currents were determined using the same voltage protocol as shown in Fig. 2A in the absence and presence of 0.1, 1, 2.5, 5.0 and 10 mm BaCl2 and at different clamp voltages for Kcnj1 (A) and rROMK2 (B). Normalised current I/Imax (where I is the current in the presence of Ba2+ and Imax the current in the absence of Ba2+) is plotted against the log of the extracellular [Ba2+] expressed in μm for the data obtained at holding potentials of +100 mV (triangles), 0 mV (squares) and −120 mV (circles). Values are presented as mean ±s.e.m. (*P < 0.001, Kcnj1 vs. rKir1.1b; n= 9 in both groups). At each of the holding potentials the dose–response curve was calculated for individual oocytes using the modified Hill equation inline image, where IC50 is the concentration of Ba2+ producing 50% inhibition, n the Hill coefficient and with the min value constrained to 0 for the fitting. Where an error bar is not visible, the value is smaller than the size of the symbol.

pH sensitivity

Mammalian ROMK isoforms are exquisitely sensitive to changes in intracellular pH, a property that involves a non-titratable lysine residue (K61). The equivalent residue in Kcnj1 is an isoleucine (see online Supplemental material, Fig. S2). We therefore determined the pH sensitivity of Kcnj1 by TEVC of oocytes expressing Kcnj1 in which intracellular acidification was induced by superfusing with external solutions of varying pH containing 3 mm sodium butyrate (Leipziger et al. 2000). Acidification produced a significant reduction in Kcnj1-dependent current, which was fully reversible (Fig. 4A). The calculated half-maximal pH inhibition was 6.91 ± 0.001 (n= 8, Fig. 4B). Thus, the zebrafish channel, despite lacking the K61 residue, showed marked pH dependency within the physiological range.

Figure 4.

Effect of pHi on activity of Kcnj1
A representative trace (A) showing the effect of changes of pHi on total current in an oocyte expressing Kcnj1. Throughout the experiment the membrane potential was clamped to −50 mV. The black rectangle indicates application of 5 mm Ba2+. During the period indicated by the hatched bar the oocyte was exposed to ND96, together with 3 mm butyrate at pH shown, as indicated by the open bars before returning to ND96. B shows the relationship between normalized K+ current (I/Imax) in oocytes (n= 8) expressing Kcnj1 against pHi. For each oocyte, I/Imax was plotted as a function of calculated pHi and the data were fitted using the modified Hill equation, I= 1/[1 + ([H+]/K0.5)n], where I is the normalized current, K0.5 is the [H+] concentration value for half-maximal channel inhibition, and n is the Hill coefficient. The cytosolic pH required for half-maximal current inhibition of Kcnj1 was 6.91 ± 0.01, with a Hill coefficient of 6.5 ± 0.02. Where an error bar is not visible, the value is smaller than the size of the symbol.

Effects of tertiapin-Q

Tertiapin-Q (TPNQ) is a toxin derived from honey bee venom that inhibits the G-protein-gated inward-rectifier K+ (GIRK1/4) and ROMK channels with nanomolar affinities (Jin & Lu, 1998, 1999; Jin et al. 1999; Frindt et al. 2009). We tested the ability of TPNQ to inhibit Kcnj1. Exposure of rROMK2 to TPNQ resulted in a dose-dependent inhibition of channel activity in the nanomolar concentration range (Fig. 5B, C and D) with a Ki of approximately 2 nm. In complete contrast, Kcnj1 was insensitive to TPNQ at concentrations as high as 1 μm (Fig. 5A, C and D). The alignments of the TPNQ binding regions for Kcnj1 and rROMK2 (generated by Praline-TM) as well as the predicted structural features that may underlie the difference in the sensitivity of the two channels to the toxin are shown in Fig. 5E (see Discussion).

Figure 5.

Effects of tertiapin-Q (TPNQ)
I–V relationships for individual oocytes expressing Kcnj1 (A) and rROMK2 (B) are plotted in the absence (ND96: filled squares) and presence of 10 nm (filled circles) and 100 nm (filled triangles) TPNQ. Currents were recorded at an external K+ concentration of 10 mm. C, superimposed, time-matched current recordings obtained at a membrane potential of −60 mV from individual oocytes expressing Kcnj1 or rROMK2 in the absence or presence of increasing concentrations of TPNQ. D, the mean normalized current (I/I0) in the presence of increasing concentrations of TPNQ for Kcnj1 and rROMK2 (n= 5 in each group). I/I0 was calculated as the ratio of current (I) to that measured in the absence of blocker (I0). Curves were fitted using a standard inhibitor–response equation (GraphPad Prism 5). Where an error bar is not visible, the value is smaller than the size of the symbol. E, alignment of the TPNQ binding regions of Kcnj1 and rROMK2 showing residues (underlined) in rROMK2 that influence binding of, and inhibition by, TPNQ (Jin & Lu, 1999); corresponding residues in Kcnj1 are in bold. Residues in blue correspond to the predicted loop sequence in Kcnj1 not present in rROMK2. The predicted structures of the pore opening (only two subunits shown in each case for simplicity) on the extracellular side of Kcnj1 (F) and rROMK (G) are shown with residues in orange (including side chains) indicating those residues known to bind TPNQ in rROMK2, and the corresponding residues in Kcnj1. The predicted additional loop region in Kcnj1 is highlighted in blue (F).

Developmental expression of kcnj1

In order to predict the likely function of Kcnj1 in vivo, we established both the time course, expressed as hours post fertilization (hpf), and tissue expression patterns of kcnj1 transcripts in zebrafish embryos and 120 hpf (5 day) larvae by RT-PCR and in situ hybridization (Figs 6 and 7). By RT-PCR, kcnj1 mRNA from whole embryos was detectable from the earliest post fertilization stages (Fig. 6A), suggesting a significant maternal pool of kcnj1 mRNA. Transcripts were present up to 120 hpf, the latest time point studied. Using in situ hybridisation, expression of kcnj1 was undetectable at stages before 12 hpf (data not shown), but transcripts were strongly expressed by 24 hpf and thereafter up to 120 hpf.

Figure 6.

Developmental expression patterns of kcnj1
A, RT-PCR detection of kcnj1. Lane 1, 1 kb ladder; 2, <2 hpf; 3, 2–4 hpf; 4, 7 hpf; 5, 14–16 hpf; 6, 18–20 hpf; 7, 26 hpf; 8, 50 hpf; 9, 72 hpf; 10, 96 hpf; 11, 120 hpf; 12, H2O control; and 13, 1 kb ladder. B, lateral view showing distribution of kcnj1 mRNA by in situ hybridisation in a 24 hpf embryo, showing expression in the integumentary epithelium (black arrows) and pronephric duct (white arrow: anterior to the left). Scale bar is 200 μm. C, anterior-dorsal view showing no kcnj1 expression; D, kcnj1 sense control (anterior to the left). Scale bar (C and D), 100 μm. Representative images (B, C and D) of 2–4 independent experiments.

Figure 7.

Expression of kcnj1 in relation to other epithelial channels and transporters
AE (ventral views, anterior to the top), pronephros (28–32 hpf). A, immuno-localisation of Nkcc protein; B, double in situ hybridisation showing location of kcnj1 (blue) and a1a.4 (red); C, ClC-Ka (blue); D, Barttin (blue); E, NCCT (blue). The scale bar applies to A–E and is 50 μm. F and G, kcnj1 expression in non-renal cells using double in situ hybridisations showing location of kcnj1 (blue) and a1a.4 (red). F, high magnification (×60) of integumentary cells (black arrow) and pronephros (out of focus; white arrow) at 28 hpf; G, double-labelled cells (black arrow) in gill primordia (120 hpf). Scale bar applies to F and G and is 100 μm. Images are representative of 2–4 independent experiments.

In non-renal tissues, we found that transcripts of kcnj1 and of the α1a.4 subunit of the Na+/K+-ATPase (a1a.4) were colocalized in cells of the yolk sac and integument, particularly in ventrolateral regions of the embryo (Figs 6B and 7F) but were absent from anterior-dorsal regions (Fig. 6C). All controls were negative (Fig. 6D). From 72 hpf, both transcripts became detectable in a population of cells in gill primordia also expressing a1a.4 (Fig. 7G).

In the pronephros, expression of kcnj1 was confined to a region of the duct mid-segment (Figs 6B and 7B). In the pronephros, we further defined the expression of kcnj1 relative to other markers, homologues of which in the mammalian kidney identify the TAL and distal tubule. The expression of kcnj1 (Fig. 7B) overlapped with that of transcripts of the chloride channel ClC-Ka (Fig. 7C; Kieferle et al. 2005), the ClC-Ka/b accessory subunit ‘Barttin’ (Fig. 7D; Estévez et al. 2001) and with Nkcc protein (Fig. 7A), a Na+- and K+-dependent electroneutral Cl cotransporter (Gamba et al. 1994). The expression of ClC-Ka and Barttin transcripts extended along the remainder of the pronephric duct throughout the whole of the mid and late duct as far as the cloaca and overlapped with transcripts of the thiazide-sensitive cotransporter NCCT (Gamba et al. 1994). Expression of NCCT transcripts did not overlap with those of kcnj1, but were located posterior to the site of expression of both kcnj1 and Nkcc (Fig. 7E). kcnj1 transcripts were only detectable in cells that also expressed a1a.4 in the pronephros, skin and gill and were not detectable in any other tissue up to and including 120 hpf (the latest time point investigated).

Effects of knockdown of Kcnj1

To determine the functional significance of Kcnj1 in vivo, we injected a morpholino antisense oligonucleotide (MO) directed against the predicted translational start site of kcnj1. Injection of MO caused retention of the kcnj1 transcript in the nucleus, suggesting that the ATG MO may disrupt normal processing of the primary transcript (Fig. 8A and B). Knockdown of translation produced no consistent effects on embryo morphology, nor on embryo mortality (data not shown); however, knockdown was associated with effects on cardiac function. At 24 h post fertilization, wild-type (n= 13) and control-injected embryos (n= 13) displayed heart rates of 90.0 ± 2.0 and 91.0 ± 2.5 beats min−1, respectively. At this time point, heart rates of MO-injected embryos (n= 15) were higher than both control groups (107.1 ± 2.8 beats min−1; P < 0.05 in both cases). Over the following 24 h, wild-type and control-injected embryonic heart rates increased significantly to 156.6 ± 3.8 beats min−1 and 152.0 ± 4.0 beats min−1, respectively. Over the same time period, MO-injected embryonic heart rate did increase significantly (P < 0.05 compared to 24 hpf), but at 48 hpf they were significantly lower (125.1 ± 3.3 beats min−1) compared to either control group (P < 0.05 in both cases).

Figure 8.

Effects of knockdown of translation of Kcnj1
In situ hybridisations of kcnj1 in control-injected (A) and morpholino (MO)-injected (B) embryos, both 28–32 hpf. In the control-injected embryo, kcnj1 transcripts (blue stain) show a typical cytoplasmic distribution. In the MO-injected embryo, expression is present but is now in rounded structures (arrowheads), indicating retention in cell nuclei. Scale bar for A and B is 50 μm. C, heart rates in beats min−1 at 24 hpf (left) and 48 hpf (right). Open and filled circles: non-injected and control MO-injected, respectively; filled triangles, MO-injected embryos. The mean values are indicated by the horizontal lines over the symbols. Some symbols represent identical heart rates in two or more embryos.


Given the current status of Danio rerio as a model of a number of human diseases, it is surprising that so little is known regarding the expression and functions of K+ channels in this organism. To date, five Kir channel genes have been identified in zebrafish (Sprague et al. 2006), of which only two have been characterized: kcnj11l (Kir6.3), which is expressed in brain (Zhang et al. 2006), and kcnj13 (Kir7.1), which is expressed in melanophores and where loss of function of the channel gives rise to the jaguar/obelisk mutant (Iwashita et al. 2006). In comparison to our detailed knowledge of K+ channel expression and function in mammalian epithelia, virtually nothing is known about the function of such channels in freshwater fish such as Danio (Perry et al. 2003).

Structure and function of Kcnj1 channel protein

Our sequence and phylogenetic analysis suggests that Kcnj1 is an orthologue of mammalian Kir1.1b rather than Kir1.1a or Kir1.1c, since the extended amino terminal sequence characteristic of both a and c isoforms is absent from the zebrafish protein (Fig. S1B in online Supplemental material). Structural models of both ROMK2 and Kcnj1 were created using PDB template 2qks, resolved in 2007. A previous structural model of ROMK1 (Haider et al. 2007) was itself based upon a model that is less accurate than when a resolved structural template is used. The new models described in this paper will, thus, provide a more accurate structural context with which to interpret the functional results.

Heterologous expression of Kcnj1 in oocytes produced barium-sensitive and pH-dependent K+-selective currents. These basic functional properties suggest that in vivo, Kcnj1 acts as a K+ efflux pathway. Given that Kcnj1 shares 55% similarity with Kir1.1, it is not surprising that a number of functionally important phosphorylation motifs are conserved between zebrafish and the well-characterized mammalian channels (Hebert et al. 2005). The role of these motifs in the function of Kcnj1 will require further study. However, some sequence differences are noteworthy since they suggest that Kcnj1 may differ from the mammalian channels in some important regulatory features. We found that the expressed current magnitude observed for Kcnj1 was always lower than that of rKir1.1b, despite injection of equivalent amounts of RNA. This may be due to the lack of an N-linked glycosylation site in Kcnj1. Kir1.1 possesses a single glycosylation site (N98), disruption of which is associated with reduced current (Schwalbe et al. 1995). The corresponding residue in Kcnj1 is isoleucine (Fig. S1 in Supplemental material). The physiological significance of lack of glycosylation of Kcnj1 is currently unclear. Secondly, although Kcnj1 was inhibited by Ba2+, the characteristic voltage dependency of blockade was not observed. The reasons for this are presently unclear, but may be resolved by future studies utilising mutagenesis of selected pore residues in Kcnj1. Thirdly, Kcnj1 was inhibited by intracellular acidification. The pH dependency of Kir1.1 was first reported (Fakler et al. 1996) to be due to the presence of a titratable lysine residue (K61 in Kir1.1b). In Kcnj1, the corresponding residue is an isoleucine. Despite this, the zebrafish channel was pH sensitive over a similar range to that reported for mammalian isoforms (Fakler et al. 1996). A putative fugu Kir1.1 orthologue is also pH sensitive, but lacks a lysine at the corresponding position (Rapedius et al. 2006). For mammalian Kir1.1 isoforms, pH sensitivity is allosterically modulated by external K+ concentration. Elevated external K+ shifts the IC50 for pH to more alkaline values, so that Kir1.1 activity is enhanced as external K+ rises (Doi et al. 1996). The sensitivity to potassium is conferred by the presence of a valine (V) at position 140. Substitution of V140 by threonine abolishes K+ sensitivity of the channel (Schulte et al. 2001). Interestingly, V140 is conserved in all vertebrate Kir1.1 proteins except those of fugu and zebrafish. The fugu channel contains a threonine at position 140 and lacks sensitivity to external K+ (Rapedius et al. 2006). The equivalent residue in Kcnj1 is a similar uncharged hydrophilic residue (serine) and therefore the zebrafish channel is likely also to be insensitive to external K+, though this remains to be determined experimentally.

We determined the sensitivity of Kcnj1 to the bee venom toxin TPNQ and rather surprisingly, found that in contrast to rROMK2, Kcnj1 was insensitive to the toxin. However, our structural modelling does provide a potential explanation for this finding. A number of residues around the pore region of ROMK channels have been investigated for their influence on both binding and inhibition by TPNQ and are shown in Fig. 5E alongside the corresponding alignment for Kcnj1. Of particular importance to TPNQ binding to ROMK is the presence of a phenylalanine residue: F127 (Jin et al. 1999). The corresponding residue in Kcnj1 is an asparagine (N), and the predicted effects of this substitution on the pore region on the extracellular side of the membrane are shown in the structures in Fig. 5F. Moreover, our modelling suggests that the insertion in Kcnj1 of the motif LWQNPPPGH between the outer and the pore helices may form an additional loop structure that, whilst it will be flexible, has the ability to be raised (as shown in our energetically favourable confirmation) and potentially hinder movement of the toxin into a binding position that is less sterically hindered in ROMK2 (Fig. 5F and G). This additional feature, in Kcnj1, together with the absence of the critical phenylalanine residue at position 127, may be what hinders binding, and channel block, by the toxin.

Expression of kcnj1 in the pronephros

Fish do not possess a loop of Henle and therefore cannot produce concentrated urine (Smith, 1951). However, freshwater species dilute the tubular fluid in an analogous fashion to the mammalian diluting segment, a function attributed to the TAL. The zebrafish pronephros is essentially a linear version of the mammalian nephron. Spatial and temporal expression of transporter and channel genes is dependent on both signalling by retinoic acid and the ‘caudal’ (cdx) transcription factors (Wingert et al. 2007). We observed overlapping expression of kcnj1 transcripts with that of Nkcc, a Na+–K+–2Cl cotransporter protein homologous to the mammalian loop diuretic-sensitive cotransporter expressed in the TAL. In the pronephros, Nkcc (presumed Nkcc2) protein is expressed in the apical membrane (Abbas & Whitfield, 2009), suggesting a similar function to its mammalian counterpart. Expression of kcnj1 and Nkcc overlapped with that of the Cl channel-encoding gene ClC-Ka, and for the first time is shown also to overlap with the ClC-Ka/b accessory subunit Barttin. This distribution is consistent with the cells of this segment representing the diluting segment in the embryonic and larval kidney and corresponds to the ‘distal early’ segment of the pronephros identified by Wingert and co-workers (Wingert et al. 2007). Thus the role of Kcnj1 at this site is likely to be similar to that in the mammalian TAL, i.e. to provide an efflux pathway for recycling K+ across the apical membrane via Nkcc and to maintain the membrane potential that provides the driving force for exit of Cl across the basolateral membrane via ClC-Ka, thereby generating a lumen-positive potential difference and therefore providing a favourable driving force for paracellular reabsorption of cations.

We did not observe overlapping expression of kcnj1 and Ncct (Slc12a3). However, this may be due to the time point used in our study, since in older embryos, kcnj1 expression has been reported to extend to the ‘late distal’ segment, overlapping with that of Ncct (Wingert et al. 2007). This suggests that the remainder of the pronephros at this stage of development is a functional equivalent of the mammalian distal tubule. In mammals, the KCNJ1 gene undergoes alternate splicing, generating three major polypeptides that are differentially expressed along the TAL, early and late distal tubule, initial connecting tubule and cortical and outer medullary collecting duct (Lee & Hebert, 1995). Secretion of K+ via presumptive heteromeric channels consisting of Kir1.1 subunits in these segments determines final urinary excretion of potassium. However, our results suggest that this is unlikely to be the case in the zebrafish. Analysis of the kcnj1 genomic sequence reveals that there are three transcripts of kcnj1 (kcnj001–kcnj003), but each codes for an identical peptide of 370 amino acids, suggesting that transcript expression of the same channel protein is regulated both spatially and temporally, as we have observed in this study. Secondly, kcnj1 transcripts were absent from the ‘late distal’ segment of the pronephros as delineated by Ncct.

Expression of kcnj1 in ionocytes

We found that kcnj1 was expressed in a set of cells in the integument and gill primordia that also express the a1a.4 isoform of the Na+/K+-ATPase. In fish, prior to the development of kidney and gill function, the integumentary epithelium is a key osmoregulatory organ (Varsamos et al. 2005). Three types of ion-transporting cells (‘ionocytes’) have so far been identified in both integument and gill that are involved in Na+ and Ca2+ absorption as well as H+ secretion. These ionocytes are known as NCCT, NaR and HR cells, respectively (Pan et al. 2005; Lin et al. 2006; Hsiao et al. 2007; Wang et al. 2009). It follows, therefore, that it is of interest to determine the cell type in which Kcnj1 is expressed. It would appear that the cells that express kcnj1 that we have identified are distinct from the other ionocyte types described previously. There are three known transcripts of Na+/K+-ATPase a isoforms associated with specific types of ionocyte: a1a.1 (in NaR cells), a1a.2 (in NCCT cells), and a1a.5 (in HR cells). A proposed fourth type of ionocyte, the a1a.4-expressing cell has hitherto not been linked with any known function, since the expression of a1a.4 does not overlap with other known transporter or channel proteins (Liao et al. 2009). Here we have shown that kcnj1 is only expressed in cells also expressing a1a.4, tentatively identifying these cells as this fourth type of ionocyte, which we propose to term ‘KS’ (for K+-secreting) cells (Fig. 9).

Figure 9.

Suggested models of ion transport mechanisms in ionocytes of zebrafish integument and gill
KS cells are likely K+-secreting cells. Details of NCCT, NaR and HR cells adapted from Liao et al. 2009. For further details see the Discussion. Copyright declaration: this is an unofficial adaptation or translation of an article that appeared in a publication of the American Physiological Society. The American Physiological Society has not endorsed the content of this adaptation or translation, or the context of its use.

Disruption of Kcnj1 function

In humans, the interrelationship between channels of the ROMK family and electrolyte homeostasis is demonstrated by the autosomal-recessive disease known as Bartter's syndrome type II (Simon et al. 1996). Inhibition of K+ secretion in the TAL caused by loss-of-function mutations in KCNJ1 leads to ineffective salt reabsorption, elevated renal salt loss and disturbed acid–base balance (Hebert, 2003). We knocked down the function of Kcnj1 by morpholino antisense oligonucleotide injection and measured heart rate as a reporter of disturbed K+ homeostasis. Excitability of cardiac muscle and pacemaker activity is determined by the transmembrane K+ concentration gradient and both hypo- and hyperkalaemia can affect heart rate (Mangoni & Nargeot, 2008). We found that at 24 hpf, heart rate was elevated in comparison to controls, whose heart rates were within the range reported by others (Baker et al. 1997), but by 48 h MO-injected embryos were bradycardic. In the absence of measurements of extracellular K+ concentration, we cannot determine whether the effects on heart rate were due to increased or decreased extracellular K+. However, it is of interest that in humans, mild to moderate hyperkalaemia is associated with increases in heart rate but higher levels of hyperkalaemia cause impaired myocardial responsiveness and reduced heart rate (Kahloon et al. 2005).

In summary, we have demonstrated that the zebrafish orthologue of human Kir1.1 is expressed in organs concerned with osmoregulation during embryonic and larval development. The properties of the channel and the consequences of knocking down its function suggest that it plays a role in electrolyte transport in epithelia of the kidney, integument and gill and thereby helps to maintain extracellular potassium levels in developing zebrafish embryos.


Author contributions

Experiments were performed in Sheffield and Leeds. L.A. performed morpholino injections, localisation studies and RT-PCR, analysed data and contributed to writing the manuscript. S.H. performed electrophysiological characterisations, analysed data and contributed to drafting the manuscript; L.F.S. performed structural modelling and contributed to drafting the manuscript; G.J.C. and T.L.W. were responsible for cloning and developing PCR strategies and sequencing. Both contributed to initial drafting of the manuscript and revisions. G.J.C. also performed electrophysiological characterisations and analysed data. T.S.M. performed electrophysiological (TPNQ) experiments, analysed data and contributed to drafting the manuscript. T.T.W. provided reagents, designed experiments, performed localisation experiments and with S.J.W. wrote the manuscript. S.J.W. conceived the project, designed and performed experiments, analysed data and with T.T.W. was responsible for writing and revising the manuscript. All authors have approved the final version of the manuscript and their contributions.


This work was supported in part by grants to S.J.W. from Kidney Research UK and the University of Leeds Strategic Development Fund, and to T.T.W. from the MRC (G0300196). S.H. was a graduate student supported by the Iranian Ministry of Health and Medical Education and Arak University of Medical Sciences. We thank Kate Hammond and Sarah Peacock for contributing to the early stages of this work, Dr Robert Levenson (Dept Pharmacology, Penn State University) for supplying the Na+/K+-ATPase a1a.4 clone and Dr Malcolm Hunter (IMSB, Leeds University) for comments on an earlier version of the manuscript. Finally, we fondly acknowledge Steven Hebert for pioneering the modern field of inward rectifiers, and for his unmatched cooperative and enthusiastic approach to physiology.

Authors’ present addresses

S. Hajihashemi: Department of Physiology, Medical Faculty, Arak University of Medical Sciences, Taleghani Street, Sardasht, Arak, Iran.

T. L. Ware: Department of Biology, Salem State University, 352 Lafayette Street, Salem, MA 01970, USA.