A novel inwardly rectifying K⁺ channel, Kir2.5, is upregulated under chronic cold stress in fish cardiac myocytes

Inward-rectifier potassium (Kir2) channels conduct large inward currents at membrane potentials negative to the K⁺ reversal potential but, due to voltage-dependent block by Mg²⁺ and polyamines, permit only limited outward current at depolarised membrane potentials, thus generating a typical voltage dependence of the inward-rectifier current(I_K1) (Matsuda et al.,1987; Vandenberg,1987; Ficker et al.,1994; Lopatin et al.,1994; Fakler et al.,1995; Kurata et al.,2006). The small outward current is, however, physiologically important in stabilising resting membrane potential close to the K⁺equilibrium potential and in regulating the duration of the cardiac action potential by phase-3 repolarisation(Shimoni et al., 1992; Lopatin and Nichols,2001).

The Kir2 subfamily has four known members (Kir2.1–2.4), three of which (Kir2.1–2.3) are expressed in the heart. Kir2.1 is the dominating subunit in the mammalian heart, although Kir2.2 and Kir2.3 channels are variably expressed depending on the animal species and cardiac chamber(Liu et al., 2001; Preisig-Müller et al.,2002; Zobel et al.,2003; Dhamoon et al.,2004; Hume and Uehara,1985; Wang et al.,1998; Melnyk et al.,2002). Each homotetrameric Kir2 channel has distinct kinetics,conductance and sensitivity to polyamines, enabling the regulation of I_K1 density and inward rectification by variable expression and coassembly of the Kir2 subunits(Périer et al., 1994; Takahashi et al., 1994; Ishihara and Yan, 2007). Human diseases due to Kir2 mutations and knockout animal models of the Kir2 channels have demonstrated a vital role of I_K1 in normal cardiac function (Plaster et al.,2001; Zaritsky et al.,2001). Furthermore, expression and function of the cardiac I_K1 is altered in hypoxia, ischaemia and disease states(Piao et al., 2007; Liu et al., 2007; Ten Eick et al., 1992)indicating that Kir2 channels are plastic entities and are probably involved in cardiac remodelling in pathophysiological conditions.

Unlike mammalian hearts, the hearts of ectothermic vertebrates are naturally exposed to large temperature changes, which impose special requirements for cardiac ion channel function to maintain proper excitability and to prevent cardiac arrhythmias. For example, crucian carp (Carassius carassius L.) can tolerate temperatures between 0 and 38°C(Horoszewicz, 1973) and in their natural environment face seasonal temperature changes of over 20°C(Vornanen and Paajanen, 2004). Owing to its excellent thermal tolerance, crucian carp heart is an interesting subject for testing the plasticity of vertebrate cardiac phenotype and its molecular basis, in particular the responses of cardiac ion channels to temperature change. The objective of this study was to examine the contribution of different Kir2 subunits to putative temperature-induced changes in the cardiac I_K1. Three Kir2 channel genes were found in crucian carp heart, two of them being homologues of the mammalian Kir2.1 and Kir2.2 channel genes. The third one was a new, previously unknown member of the Kir2 subfamily and was designated as ccKir2.5(cc for Carassius carassius). The expression of ccKir2.5 was strongly increased in the cold-acclimated fish (4°C), suggesting that the novel ccKir2.5 is intimately involved in cardiac adjustment to low temperature conditions by increasing the density of I_K1.

Crucian carp (Carassius carassius L.; 20–50 g in body mass, N=46) were caught from a small lake and in the lab were reared in temperature-controlled 500 l stainless steel tanks at either 4°C (cold acclimation) or 18°C (warm acclimation) for a minimum of 4 weeks. Fish were stunned by a sharp blow to the head and killed by cutting the spine, and the organs needed in experiments were prepared. All experiments were carried out with the consent of the national committee for animal experimentation.

Atrium, ventricle and brains and a piece of gill, muscle, liver and kidney were homogenised under liquid nitrogen and RNA was extracted with TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. DNA was extracted from liver by the method of Sambrook et al.(Sambrook et al., 1989). The quality and quantity of RNA and DNA were monitored by agarose gel electrophoresis and UV spectrophotometry, respectively.

The open reading frames (ORFs) for crucian carp Kir2.1, Kir2.2 and Kir2.5 and a 437 bp fragment for crucian carp DnaJA2 gene were cloned by reverse transcriptase-PCR (RT-PCR). First strand cDNA synthesis was carried with RNAseH⁺ (Finnzymes, Espoo, Finland) using random hexamers or oligo(dT) primers. Degenerative primers designed to the conserved regions of mammalian Kir2.1, Kir2.2 and Kir2.3 genes were used to get partial cDNA clones of crucian carp Kir2 genes(Table 1). In spite of several trials, no Kir2.3 products were found in the crucian carp heart. The first fragment of an unidentified Kir gene was obtained with degenerative primers to mammalian Kir2.1. New primers for further cloning were designed for each clone on the basis of sequences obtained. PCR was performed using a PTC-200 DNA Engine Cycler (MJ Research, Waltham, MA,USA) and conditions described previously(Hassinen et al., 2007). Oligo(dT) primers and a 3′-RACE kit (Invitrogen) were used to clone the 3′ end of the ccKir2.2 and ccKir2.5 gene, respectively(Table 1). Genome Walker kit(Clontech, Palo Alto, CA, USA) was used for cloning the 5′ ends. Four genomic libraries from crucian carp DNA were constructed and used as templates in PCR as previously described (Hassinen et al., 2007). Finally, the sequences were confirmed by cloning the whole coding region of the ccKir2.1, ccKir2.2 and ccKir2.5genes.

All PCR products were analysed by agarose gel electrophoresis, extracted from the gel by Qiaex II gel extraction kit (Qiagen, Valencia, CA, USA) and cloned into the pGEM-T Easy vector (Promega, Madison, WI, USA). DNA sequencing was conducted using the ABI PRISM BigDye Terminator cycle sequencing kit v2.0(Applied Biosystems, Foster City, CA, USA) and the reactions were analysed using ABI PRISM 310 Genetic Analyzer (Applied Biosystems).

The whole coding sequences of ccKir2 proteins, zebrafish Kir proteins and all known mouse Kir proteins were aligned by ClustalW and a phylogenetic tree was constructed in ClustalX(http://www.clustal.org/download/current/)using the neighbour-joining method. Sequence positions containing gaps in any of the Kir genes were ignored from the analysis. KirBac3.1 was used as an outgroup member and the analysis was performed with 1000 bootstrap replicates. Graphical presentation of the tree was produced in Treeview(http://taxonomy.zoology.gla.ac.uk/rod/treeview.html/).

Atrium, ventricle and brains and a piece of gill, muscle, liver and kidney were pooled from several fish for RNA sample preparation (N=3 for both acclimation groups). First strand cDNA synthesis was performed from DNAse-treated RNA (Hassinen et al.,2007) and DNA contamination was tested by a control cDNA synthesis containing all other reaction components except the RT enzyme. Quantitative RT-PCR was performed using Chromo4 Continuous Fluorescence Detector (MJ Research) under previously described conditions(Hassinen et al., 2007). Primers were designed to the non-conserved N- or C-termini of the ccKir2.1, ccKir2.2 and ccKir2.5 and to the cloned region(nucleotides 121–557; GenBank accession number EU191947) of the crucian carp DnaJA2 (Table 2). DnaJA2 was used as a reference gene because it is more stable in thermal acclimation than the conventional reference genes(Vornanen et al., 2005). Transcript abundance of the ccKir2 genes was normalised to the DnaJA2 expression level.

ORF sequences for the putative ion channel-forming genes ccKir2.1,ccKir2.2 and ccKir2.5 were subcloned into the pcDNA3.1/Zeo (+)vector (Invitrogen) for expression in a COS-1 cell line(Hassinen et al., 2007). Electrophysiological experiments were made 48–72 h after transfection.

Ventricular myocytes were enzymatically isolated(Vornanen, 1997) and used within 8 h of isolation. Whole-cell voltage clamp experiments were done using an EPC9 patch clamp amplifier (HEKA Instruments Inc., Lambrecht/Pfalz,Germany), a PC-16 solution exchanger (Bioscience Tools, San Diego, CA, USA)and Pulse acquisition software (HEKA Instruments). External saline solution contained (mmol l^–1): 150 NaCl, 5.4 KCl, 1.8 CaCl₂, 1.2 MgCl₂, 10 glucose and 10 Hepes (pH 7.6). Patch pipettes were pulled from borosilicate glass (Garner, Claremont, CA,USA) and filled with K⁺-based electrode solution (mmol l^–1: 140 KCl, 1 MgCl₂, 5 EGTA, 4 MgATP and 10 Hepes at pH 7.2). For experiments with ventricular myocytes, tetrodotoxin (0.5μmol l^–1, Tocris Cookson, Bristol, UK)(Haverinen et al., 2007),nifedipine (10 μmol l^–1, Sigma, Helsinki, Finland),glibenclamide (10 μmol l^–1, Sigma) and E-4031 (1 μmol l^–1, Alomone labs, Jerusalem, Israel) were added to the extracellular solution to block Na⁺, Ca²⁺, ATP-sensitive K⁺ and delayed-rectifier K⁺ (I_Kr)current, respectively. The mean resistance of the pipettes was 2.52±0.07 MΩ.

Single-channel properties of the cloned ccKir2 channels were recorded at room temperature (21±1°C) in inside-out configuration with an EPC-9 amplifier and Pulse software and analysed with TAC, TACFit (Bruxton Corporation, Seattle, WA, USA) and SigmaPlot 6.0 (SPSS, Chicago, IL, WA)software (Paajanen and Vornanen,2004). The pipette solution was composed of (mmol l^–1): 134 KCl, 1.8 CaCl₂, 2 MgCl₂, 10 glucose and 10 Hepes adjusted to pH 7.6 with KOH ([K⁺]=141 mmol l^–1). The composition of the bath solution in the inside-out experiments was (mmol l^–1): 140 KCl, 2 EGTA, 1 EDTA and 5 Hepes adjusted to pH 7.6 with KOH. EGTA and EDTA were included to prevent endogenous Ca²⁺-dependent currents of the COS-1 cells. Inside-out patches with any outward current were excluded from further analyses.

Native Kir2 channels of fish ventricular myocytes were measured in inside-out and cell-attached configurations. Inside-out experiments were made under the same experimental conditions as the experiments with the cloned channels. Cell-attached recordings were conducted at 11°C (temperature in the middle of the acclimation temperatures of 4 and 18°C) by using the same external saline in the bath as in the whole-cell experiments. Pipettes were pulled (PP-83 puller, Narishige, Tokyo, Japan) from thick-walled borosilicate glass (Garner), coated with Sylgard (WPI, Stevenage, UK), fire polished on a microforge (MF-83, Narishige) and filled with the same high K⁺ solution that was used for the inside-out patches. The mean resistance of the pipettes was 25.1±1.8 MΩ.Because low external K⁺ causes a voltage offset, the Nernst potential of K⁺ions (–80 mV) was added to the membrane voltage.

All single-channel recordings were sampled at 4 kHz and low-pass filtered at 2 kHz. Single-channel conductance was determined by applying 5 s square pulses from –120 to +80 or –200 to –20 mV in 20 mV increments every 10 s for inside-out and cell-attached patches, respectively. Distributions of open and closed times were obtained from 20 to 120 s recordings at –100 mV. Open and closed time analyses were performed on patches that had only a single open current level. Open and closed times were detected with time-course fitting, and probability density functions (pdf)were analysed (TACFit) from idealised data with the log-likelihood method on log(event times). Histograms of single-channel conductance were constructed from cell-attached and inside-out recordings of the endogenous inward-rectifier channels from cold- and warm-acclimated fish to see the natural variation of conductance levels.

Differences between mean current values from warm- and cold-acclimated carp ventricular I_K1 were assessed by Student's t-test, whereas mean current values of the channels encoded by the cloned ccKir2.1, ccKir2.2 and ccKir2.5 genes were evaluated by using one-way analysis of variance. Differences in Kir2 transcript abundances were tested using Student's t-test. A P value of 0.05 was regarded as the limit of statistical significance.

Mammalian cardiac Kir2 channels are composed of Kir2.1, Kir2.2 and Kir2.3 subunits. Three partial cDNAs for Kir2 genes were also obtained from crucian carp heart using degenerative primers in RT-PCR. Two of the completely sequenced ORFs showed high predicted amino acid sequence identity with mammalian (about 83% and 77%) and fish (Oncorhynchus mykiss) (90% and 80%) Kir2.1 and Kir2.2 channels, respectively, suggesting that they are homologues of the vertebrate Kir2.1 and Kir2.2 genes(Fig. 1). Accordingly, these crucian carp Kir genes were designated as ccKir2.1(EU182582) and ccKir2.2 (EU182583). Surprisingly, the third clone of the crucian carp Kir channel genes had a relatively low predicted amino acid sequence identity (59%) with the mammalian Kir2.3 channel and with other members of the Kir2 subfamily (68%, 71% and 56% for Kir2.1, Kir2.2 and Kir2.4,respectively). Even lower identity values were obtained with another six Kir subfamilies (Fig. 1B). These findings suggest that Kir2.3 is not expressed in the crucian carp heart, and the third gene is a member of the Kir2 subfamily, even if it is not homologous to any known Kir2 gene. We assumed it to be a novel gene of the vertebrate Kir2 subfamily and designated it as ccKir2.5 (EU182584).

ORFs for ccKir2.1, ccKir2.2 and ccKir2.5 consisted of 1284, 1314 and 1269 bp, coding for 427, 437 and 422 amino acids, respectively(supplementary material Fig. S1). To establish evolutionary relationships between the ccKir2 channels and the known vertebrate Kir channels, we constructed a phylogenetic tree with members from all seven Kir subfamilies. Phylogenetic analysis showed that ccKir2.1 and ccKir2.2 are, indeed,homologues of Kir2.1 and Kir2.2, respectively(Fig. 2). From the known Kir family members, the closest relative of ccKir2.5 was Kir2.2 with an 81%bootstrap value. Importantly, ccKir2.5 shares a common ancestor with other Kir2 genes, providing strong evidence that ccKir2.5 is a new member of the Kir2 subfamily. Taken together, crucian carp heart expresses ccKir2.1, ccKir2.2 and ccKir.2.5, but not a homologue of vertebrate Kir2.3.

All vertebrate Kir proteins consist of a conserved pore-forming region (P)flanked by two transmembrane alpha helices (M1 and M2). These functional domains were identified from all cloned ccKir2 proteins and were more conserved than the cytoplasmic N- and C-termini(Fig. 1A, supplementary material Fig. S2). In mammalian Kir2.1, Mg²⁺ binds to serine S165(numbering according to mouse Kir2.1)(Fujiwara and Kubo, 2002),whereas glutamates E224 and E299 function as an intermediate binding site for polyamines before entering to the pore-blocking site D172(Xie et al., 2003). All these residues also exist in ccKir2.5, ccKir2.2 and ccKir2.1, suggesting a similar inward-rectification mechanism for mammalian and fish Kir2 channels. In contrast, some variability appeared in amino acids important for Ba²⁺ binding. In mammalian Kir2.1, glutamate E125 between M1 and P binds Ba²⁺ and facilitates its movement to the plugging site threonine T141 of the narrow pore region(Alagem et al., 2001). Threonine T141 is conserved in all mammalian and fish cardiac Kir2 channels. In contrast, glutamate E125 exists in mammalian Kir2.1, but not in mammalian Kir2.2 and Kir2.3 channels. In crucian carp, the glutamate E125 is replaced by other amino acids, not only in ccKir2.1, but also in ccKir2.5 and ccKir2.2. Yet, the cloned ccKir2.5 and ccKir2.2 channels were about 10 times more sensitive to Ba²⁺ than ccKir2.1. While this agrees with the finding that mammalian Kir2.2 is 5–10 times more sensitive to Ba²⁺than Kir2.1 (Liu et al., 2001; Preisig-Müller et al.,2002), it strongly suggests that other residues in addition to E125 are important for Ba²⁺ binding.

Transcript abundance of ccKir2 subunits was determined by quantitative PCR in seven tissues of the cold-acclimated crucian carp. All three ccKir2 genes were expressed to some extent in heart, brain, gill,kidney, liver and skeletal muscle (Fig. 3). ccKir2.5 was a major ccKir2 channel component in atrium(59.1±2.1%) and ventricle (65.6±3.2%) of the heart and in the skeletal muscle (88.6±3.4%), while in other tissues it was weakly expressed (<25%), suggesting that it is a muscle-specific isoform.

The majority (∼90%) of COS-1 cells transfected with ccKir2genes and a separate eGFP vector had a large inward-rectifying current with a slope conductance of 100±15 nS, which reversed direction near the equilibrium potential of K⁺ ions (–73.3±1.0 mV). The current was reversibly blocked by 0.1–0.3mmoll^–1[Ba²⁺]_o (Fig. 4A), indicating that the cloned genes encoded functional inward-rectifying K⁺ channels of the Kir2 family.

When expressed in COS-1 cells, ccKir2.5 was almost 10-times more sensitive to [Ba²⁺]_o (K_d, 2.43±0.37μmol l^–1) than the ccKir2.1 channel (22.25±5.37μmol l^–1; P<0.05), and it also rectified much more strongly (z=2.40±0.08, V_1/2=–85.62±3.50 mV) than ccKir2.1(z=1.73±0.04, V_1/2=–54.20±0.57 mV; P<0.05; Fig. 4B–D). With regard to Ba²⁺ sensitivity, ccKir2.5 current was similar to the current generated by ccKir2.2 (3.48±0.90 μmol l^–1; P>0.05), but more strongly rectifying than ccKir2.1(V_1/2=–76.29±2.21 mV; P<0.05). Accordingly, at the whole-cell level ccKir2.5 differs from the other ccKir2 channels with regard to either Ba²⁺ sensitivity or inward rectification, or both.

Even at the single-channel level, ccKir2.5 channels were functionally closer to ccKir2.2 than to ccKir2.1 channels(Fig. 5). Single-channel conductance of ccKir2.5 (44.19±1.89pS) was 3 times as large as that of ccKir2.1 (14.62±1.95 pS; P<0.05), but only 1.6 times as large as the conductance of the ccKir2.2 (27.54±5.3 pS; Fig. 5B). The three channels also had variable single-channel kinetics. The mean open time at –100 mV was 45.73±10.89, 22.93±9.11 and 9.21±1.63 ms for ccKir2.5, ccKir2.2 and Kir2.1, respectively. Furthermore, ccKir2.5 had 4.17 times as high an open probability as ccKir2.2 (0.401±0.056 vs0.096±0.005; P<0.05; Fig. 5E), and is therefore likely to contribute more to the whole-cell IK1 than ccKir2.2. Taken together,the functional properties of ccKir2.5 are strikingly different from those of ccKir2.1 and closer to, although still distinct from, those of the ccKir2.2.

Adjustment of cardiac ion channel function to changing temperature conditions is probably vital for proper excitability of ectothermic hearts. In crucian carp ventricular myocytes, acclimation to cold (4°C) increased the slope conductance of I_K1 (707±49 vs1001±59 pS pF^–1, between –120 and –80 mV; P<0.05) and the density of both inward and outward I_K1 (Fig. 4C), indicating positive thermal compensation of the ccKir2 channel system. To resolve this change at the molecular level, transcript expression and electrophysiological properties of the cardiac ccKir2 channels were examined.

ccKir2.5 and ccKir2.2 were clearly the main ccKir2 subunits in the crucian carp heart, forming together over 99% of all ccKir2 transcripts. ccKir2.1 accounted for less than 0.8% of the transcripts, which was considered to be physiologically insignificant. Thermal acclimation for 4 weeks had a striking impact on ccKir2 mRNA expression. In cold-acclimated fish, ccKir2.5 was the dominating isoform representing 59.1±2.1% and 65.6±3.2% of the total ccKir2 transcripts in atrium and ventricle, respectively. In contrast,in warm-acclimated fish hearts ccKir2.2 was the main component, accounting for 83.6±1.6% of atrial and 77.7±1.7% of ventricular Kir2 mRNA,respectively (Fig. 6). These findings indicate that in cold-acclimated carp the major part of the ccKir2.2 transcripts is replaced by ccKir2.5, suggesting that ccKir2.5 is the cold-adapted isoform and is probably important for acclimation of the heart to low temperatures. Interestingly, the total amount of ccKir2 transcripts was slightly higher in warm- than cold-acclimated crucian carp heart, both in atrium and ventricle (P<0.05).

Ba²⁺ sensitivity of I_K1 was similar in cold-and warm-acclimated fish, which is not unexpected considering the similar Ba²⁺ sensitivities of ccKir2.5 and ccKir2.2(Fig. 4B). The voltage dependence of inward rectification of I_K1 was also similar in cold- and warm-acclimated carp ventricular myocytes(–79.60±0.85 vs –81.74±1.72), but I_K1 had a slightly shallower slope of rectification(z=2.5±0.1 vs 3.0±0.1; P<0.05) in cold- than warm-acclimated fish. The rectification properties of I_K1 from cold- and warm-acclimated fish do not conform well to the rectification characteristics of the cloned ccKir2 channels and the transcript levels of ccKir2.5 and ccKir2.2 in ventricular myocytes,suggesting that temperature acclimation must also modify the regulation of Kir2 channel function.

Single-channel currents with similar characteristics to the currents of cloned ccKir2 channels were observed in crucian carp ventricular myocytes(Fig. 5)(Paajanen and Vornanen, 2003). Channels with ccKir2.5 characteristics (the slow I_K1) were found in 8 out of 19 and 1 out of 16 cell-attached patches from cold- and warm-acclimated fish, respectively, whereas ccKir2.2-type channels (the fast I_K1) were found in 9 out of 19 and 15 out of 16 cell-attached patches from cold- and warm-acclimated fish, respectively. These frequencies are in line with mRNA expression levels of ccKir2.5 and ccKir2.2 in warm- and cold-acclimated fish. Single-channel amplitudes and mean open times of slow I_K1 and fast I_K1channels of ventricular myocytes correspond well with the values of cloned ccKir2.5 and ccKir2.2 channels (Fig. 5D,E), suggesting that they are composed of ccKir2.5 and ccKir2.2 subunits, respectively. Although Kir2.5 channels were more frequent in cold-than warm-acclimated fish hearts, the mean conductance of the endogenous Kir2 channels did not differ between acclimation groups (P>0.05; Fig. 5B). A small amplitude(∼14 pS) current (small I_K1) was found only in a few cell-attached patches (Fig. 5C). Similarly, the probability of finding a small amplitude channel was low in inside-out patches (in 6 out of 100 and 5 out of 52 patches from cold- and warm-acclimated fish, respectively). These findings suggest that the small ccKir2.1-like channels have low expression levels in ventricular myocytes of cold- and warm-acclimated crucian carp.

The background I_K1 current, formed by Kir2 channels, is involved in stabilisation of resting membrane potential and in late phase-3 repolarisation of action potential in cardiac myocytes(Shimoni et al., 1992; Lopatin and Nichols, 2001). In mammalian hearts, this task is accomplished by Kir2.1, Kir2.2 and Kir2.3 channels, Kir2.1 being the dominant isoform(Lopatin and Nichols, 2001; Liu et al., 2001; Wang et al., 1998). We have shown here that in crucian carp ventricular myocytes too, three inward-rectifier channels are involved in the formation of I_K1, two of them being homologues of the known mammalian cardiac Kir2.1 and Kir2.2 channels. The third gene of the crucian carp cardiac Kir assembly seems to belong to the Kir2 subfamily, even though it is not clearly homologous to any of the known Kir2 genes, and accordingly is proposed to be a new, fifth member for the vertebrate Kir2 subfamily, ccKir2.5. Indeed, the phylogenetic analysis showed that ccKir2.5 has a common ancestor with all four known Kir2 genes (Kir2.1, Kir2.2, Kir2.3, Kir2.4), thus providing firm evidence that it belongs to the Kir2 subfamily. Interestingly, a predicted protein (XM_001335914) sharing 98.3% similarity with ccKir2.5 was found from the zebrafish sequence database, suggesting the Kir2.5 channels might be more generally expressed in different fish species. In contrast, no homologues of ccKir2.5 were found in mammalian genomes. Thus, it is likely that the novel ccKir2.5 is unique to ectothermic vertebrates.

All vertebrate Kir2 channels rectify strongly due to a high-affinity voltage-dependent block by free polyamines and Mg²⁺, and thus in a given cell type the magnitude of the physiologically important outward current is primarily determined by the sensitivity of Kir2 channels to polyamines(Dhamoon et al., 2004; Panama and Lopatin, 2006). Although all three cardiac ccKir2 channels of the crucian carp heart have a distinct negative slope conductance and they completely rectify at 0 mV,notable differences exist between the channel isoforms in terms of inward rectification. ccKir2.5 is clearly the strongest inward rectifier followed by ccKir2.2 and ccKir2.1. Among the mammalian cardiac Kir2 channels, Kir2.1 is the weakest and Kir2.2 the strongest inward rectifier(Dhamoon et al., 2004; Panama and Lopatin, 2006). Kir2.3 is intermediate between Kir2.1 and Kir2.2, but it rectifies incompletely. Evidently, ccKir2.5 is the strongest inward-rectifier K⁺ channel of the vertebrate heart and passes little outward current. Consequently, its membrane potential stabilising and repolarising effects are relatively weak.

The present findings indicate that under chronic thermal stress the phenotype of crucian carp cardiac I_K1 is changed by a compensatory increase in the density of I_K1 in the cold. In the chronic cold, ccKir2.5 transcripts were strongly upregulated,suggesting that the ccKir2.5 isoform might be important in producing the cold-acclimated phenotype of the cardiac I_K1. Concomitantly with the increased ccKir2.5 expression, transcripts of ccKir2.2 were strongly suppressed, suggesting the possibility that the cold-induced increase in the cardiac I_K1 was obtained by an isoform shift from ccKir2.2 towards ccKir2.5. But how is the cold-induced increase in I_K1 achieved by the strongly rectifying ccKir2.5 channels,especially when the total amount of ccKir2 transcripts is simultaneously reduced? Two single-channel properties of ccKir2.5 could be contributing. First, open probability and mean open time of ccKir2.5 are 4.17 and 4.96 times, respectively, as large as those of ccKir2.2. Second, ccKir2.5 has a larger single-channel conductance than ccKir2.2. Thus, in spite of the strong voltage-dependent block by polyamines, ccKir2.5 channels might allow more outward current than ccKir2.2 channels, mainly because they stay longer in the open state and because they have a larger conductance than ccKir2.2 channels. While the larger slope conductance of I_K1 in cold-acclimated crucian carp is consistent with this, the rectification of the endogenous I_K1 does not match the rectification properties of the cloned ccKir2.2 and ccKir2.5 channels and the expression levels of these channels in cold- and warm-acclimated fish. Rectification of I_K1 in cold-acclimated fish is weaker than would be assumed on the basis of the high ccKir2.5 expression level. Therefore, other mechanisms in addition to ccKir2 isoform change must be involved in temperature-dependent regulation of the crucian carp cardiac I_K1. The present results do not provide an explanation for the divergence in rectification properties between cloned and endogenous currents, but it can be speculated that in cardiac myocytes ccKir2.2 and ccKir2.5 channels are located in different membrane compartments and therefore might face different free polyamine concentrations and/or compositions. A different distribution of Kir2.1 and Kir2.3 channels between cholesterol-rich and cholesterol-poor membrane domains has recently been suggested(Tikku et al., 2007), raising the possibly that regulation of Kir2 isoforms may differ depending on their subcellular location.

Although isoform change of ccKir2 channels does not alone explain temperature-induced changes in the I_K1 of the crucian carp heart, it probably has a central role in thermal modification of the IK1. Recently, we showed that Kir2.2 is upregulated by warm acclimation in another fish species, the rainbow trout (Hassinen et al., 2007). Together these studies suggest that temperature-induced changes in Kir2 isoforms may be a common way for temperature acclimation of the cardiac I_K1 in fish. Evidently, ccKir2.2 is a warm-adapted and ccKir2.5 a cold-adapted isoform of the cardiac inward-rectifying K⁺ channel and the cardiac phenotype of I_K1 is determined by their relative abundance.

It is interesting that temperature compensation of I_K1density was produced by an isoform shift and not by parallel upregulation of all three Kir2 channel isoforms in the cold. The potential benefit of isoform switching in comparison to a simple increase in the number of the ccKir2 population remains unexplained, especially since the electrophysiological properties of ccKir2.2 and ccKir2.5 do not radically differ (with the exception of mean open time and open probability). It can be speculated that there might be some constraints on genome function that would prevent upregulation of ccKir2.2 and ccKir2.1 isoforms at low temperatures. In this regard it is interesting that rainbow trout heart, which responds to constant cold by a decrease in the density of the I_K1, does not express Kir2.5 (Vornanen et al.,2002; Hassinen at al.,2007). Thus, it is possible that cold-induced compensation of cardiac I_K1 is dependent on the expression of the Kir2.5 isoform, and species that lack Kir2.5 or cannot express it in the heart are unable to upregulate I_K1 in the cold. Future research should find out how widely the novel Kir2.5 isoform is distributed among fish species and other ectotherms, and to clarify its importance in thermal adaptation of heart and muscle tissues.

In conclusion, a new member for the Kir2 subfamily of the inward-rectifying K⁺ channels has been isolated and characterised from the heart of crucian carp. Increased expression of the novel ccKir2.5 channel, at the expense of the ccKir2.2 isoform, is assumed to contribute to the cold-induced increase of atrial and ventricular IK1, which partly compensates for the depressive effects of low temperature on the current that maintains negative resting membrane potential and accelerates the phase-3 repolarisation of the cardiac action potential. These electrophysiological effects may be necessary to limit action potential prolongation and stabilise membrane potential to prevent cardiac arrhythmias in the cold winter waters.

The technical assistance of Anita Kervinen and Riitta Pietarinen during this study is appreciated. This study was supported by a research grant from the Academy of Finland to M.V. (project nos 210400 and 119583). M.H. was supported by the Biological Interactions graduate school.