Abnormal Heart Rate Regulation in GIRK4 Knockout Mice

Neuron, Vol. 20, 103–114, January, 1998, Copyright 1998 by Cell Press

Abnormal Heart Rate Regulation in GIRK4 Knockout Mice Kevin Wickman,*k Jan Nemec,‡k Sandra J. Gendler, § and David E. Clapham†# * Department of Cardiology † Department of Neurobiology Harvard Medical School Children’s Hospital Boston, Massachusetts 02115 ‡ Division of Cardiovascular Diseases Department of Internal Medicine Mayo Clinic Rochester, Minnesota 55905 § Department of Biochemistry and Molecular Biology Mayo Clinic Scottsdale Scottsdale, Arizona 85259

Summary Acetylcholine (ACh) released from the stimulated vagus nerve decreases heart rate via modulation of several types of ion channels expressed in cardiac pacemaker cells. Although the muscarinic-gated potassium channel IKACh has been implicated in vagally mediated heart rate regulation, questions concerning the extent of its contribution have remained unanswered. To assess the role of IKACh in heart rate regulation in vivo, we generated a mouse line deficient in IKACh by targeted disruption of the gene coding for GIRK4, one of the channel subunits. We analyzed heart rate and heart rate variability at rest and after pharmacological manipulation in unrestrained conscious mice using electrocardiogram (ECG) telemetry. We found that IKACh mediated approximately half of the negative chronotropic effects of vagal stimulation and adenosine on heart rate. In addition, this study indicates that IKACh is necessary for the fast fluctuations in heart rate responsible for beat-to-beat control of heart activity, both at rest and after vagal stimulation. Interestingly, noncholinergic systems also appear to modulate heart activity through IKACh. Thus, IKACh is critical for effective heart rate regulation in mice. Introduction The discharge frequency of the sinoatrial node, the dominant pacemaking center in the mammalian heart, is modulated by the tone of the autonomic nervous system. The opposing effects of sympathetic and parasympathetic activity generate beat-to-beat fluctuations in heart rate, a phenomenon termed heart rate variability (HRV; Akselrod et al., 1981; Saul and Cohen, 1994). Pharmacological dissection of HRV in large mammals has revealed correspondence between different autonomic inputs to the heart and particular frequency components of HRV (Akselrod et al., 1981; Saul et al., 1991). For example, the 0.15–0.4 Hz component of HRV in humans is due k These authors contributed equally to this work.

# To whom correspondence should be addressed.

largely to respiratory sinus arrhythmia, a phenomenon related to vagal tone withdrawal during inspiration (Saul et al., 1991). Contribution of this frequency component to HRV can be markedly reduced by parasympathetic blockade, and it is decreased in normal and pathological conditions leading to decreased cardiac vagal tone, such as exercise, heart failure, and autonomic neuropathy (Akselrod et al., 1981; Ewing et al., 1981; Saul et al., 1988; Shin et al., 1995). Since certain frequency components of HRV reflect primarily parasympathetic activity, HRV analysis can reveal abnormalities in heart rate regulation that might not be reflected in mean heart rate, which is determined by the balance between sympathetic and parasympathetic influences. Indeed, HRV is a strong prognostic indicator for patients following myocardial infarction, reflecting decreased vagal tone in this population (Kleiger et al., 1987). Loewi et al. first demonstrated in 1926 that acetylcholine (ACh) secreted from the vagus nerve mediates the decrease in heart rate and contractility (Loewi and Navratil, 1926). After more than 70 years of research, however, the identity of the relevant conductance(s) is still a matter of considerable debate. While early studies indicated that vagal-mediated heart rate regulation was due to ACh activation of a potassium-selective conductance termed IKACh (Giles and Noble, 1976; Noma and Trautwein, 1978; Garnier et al., 1978; Sakmann et al., 1983), several lines of evidence suggest that vagal regulation of heart rate, particularly at low-to-moderate levels of vagal activity, is due to changes in a cAMP-modulated cationic conductance termed If , often referred to as the “pacemaker” current (reviewed by DiFrancesco, 1995a). First, If in rabbit sinoatrial nodal cells was shown to be potently inhibited by ACh (DiFrancesco and Tromba, 1988). Second, concentrations of ACh too low to stimulate IKACh were shown to decrease the firing frequency of isolated sinoatrial nodal cells, acting through adenylyl cyclase inhibition and a decrease in the If current (DiFrancesco et al., 1989). Third, the membrane conductance of pacemaker cells was shown to decrease in response to vagal stimulation or ACh superfusion, as would be predicted if the dominant pacing current was inhibited, rather than stimulated, by ACh (Bywater et al., 1990; Hirst et al., 1992). In addition to IKACh and If, L-type Ca2 1 channels and the sustained inward current (I st) have been described in sinoatrial nodal cells and implicated in mediating the autonomic regulation of cardiac pacing (Noma et al., 1983; Hagiwara et al., 1988; Guo et al., 1995). Unfortunately, the lack of selective pharmacological agents for IKACh and other currents relevant to cardiac pacing has made dissection of the relative contributions of these conductances in vivo impossible. IKACh has been well characterized in terms of distribution in the heart, biophysical properties, and gating regulation (reviewed by Kurachi et al., 1992; Wickman and Clapham, 1995). IKACh is present in the sinoatrial node, atria, atrioventricular node, and possibly Purkinje fibers of mammalian heart. The channel exhibits strong inward rectification, high potassium (K1) selectivity, a unitary

Neuron 104

conductance of z35 pS at physiological K1 concentrations, and a 1–2 ms mean open time. While it is named after the first identified agonist, IKACh is also activated by a variety of neurotransmitters, including adenosine (Hartzell, 1979). IKACh activation by adenosine may be responsible for the profound transient bradycardia and atrioventricular block seen in humans after administration of adenosine (DiMarco et al., 1983; Clemo and Belardinelli, 1986; Kurachi et al., 1986). The neurotransmitters and hormones that activate IKACh all stimulate pertussis toxin–sensitive G proteins, resulting in the dissociation of the G protein complex into Gbg and GTP-bound Ga subunits. Liberated Gbg activates IKACh directly by physical association (Huang et al., 1995; Inanobe et al., 1995; Krapivinsky et al., 1995c). Cardiac IKACh is comprised of two related proteins, termed GIRK1 and GIRK4 (Dascal et al., 1993; Kubo et al., 1993; Krapivinsky et al., 1995a). Functionally similar G protein–gated K1 channels in brain and other tissues are thought to be formed by various combinations of GIRK family subunits (GIRK1, GIRK2, GIRK3, and GIRK4) (Lesage et al., 1994; Duprat et al., 1995; Ferrer et al., 1995; Kobayashi et al., 1995; Kofuji et al., 1995; Karschin et al., 1996; Velimirovic et al., 1996). In heterologous expression systems, both GIRK1 and GIRK4 are required to form the functional IKACh channel; in the absence of the GIRK4 subunit, GIRK1 does not form a functional ion channel (Krapivinsky et al., 1995a, 1995b; Hedin et al., 1996). Therefore, we disrupted the mouse GIRK4 gene in an effort to determine the contribution of this conductance to cardiac function and regulation. Here, we report that GIRK4-deficient mice lack cardiac IKACh but are viable and exhibit no difference from wild-type siblings in terms of mean resting heart rate. Results from pharmacological stimulation of the vagus and adenosine administration in unrestrained conscious GIRK4-deficient mice indicate that IKACh accounts for z50% of the negative chronotropic response to these perturbations. In addition, unlike their wild-type siblings, GIRK4-deficient mice are unable to adjust heart rate on a rapid time scale (,2 s), even at rest. Thus, this study demonstrates the pivotal role of IKACh in heart rate regulation over a substantial range of vagal activity.

Results A 14 kb 129Sv/J mouse genomic DNA fragment (Wickman et al., 1997) containing two GIRK4 coding sequence exons was used to generate a gene targeting construct. The construct contained a neomycin resistance gene (positive selection), a diptheria toxin gene (negative selection), and z8 kb of overall GIRK4 homology (Figure 1A). The first GIRK4 coding sequence exon was targeted because it codes for critical functional domains in GIRK4, including the entire amino terminus, both membrane-spanning domains, the predicted pore-lining sequence, and a large portion of the carboxyl terminus. Two of 480 screened transfected embryonic stem (ES) cell clones exhibited a properly targeted GIRK4 allele (data not shown). Both clones were used to generate chimeric mice, and both yielded male chimeric mice,

Figure 1. Characterization of GIRK4-Deficient Mice (A) GIRK4 targeting strategy. Schematic depiction of the mouse GIRK4 wild-type and targeted alleles and targeting vector. Abbreviations for restriction enzyme sites are as follows: Bg, BglII; B, BamHI; and H, HincII. “(B)” and “(Bg)” identify BamHI and BglII sites blunted during vector construction. The directions of the neomycin (neor) gene and diptheria toxin (DTA) gene transcription are indicated by arrowheads. “Exon1” and “Exon2” denote the first and second GIRK4 coding sequence exons, respectively, and “ATG” and “STOP” indicate the locations of the translation initiation codon and stop codon, respectively. The gray box denotes the 39 flanking probe used for Southern analysis of ES cell and mouse tail DNA. The analogous HincII restriction fragments for the wild-type and targeted GIRK4 alleles are 9.5 and 8.9 kb, respectively. (B) Southern blot of tail DNA from six F2 littermates derived from heterozygous parents (two mice representing each possible genotype: wild-type [WT], heterozygous [HT], and knockout [KO]). Tail DNA samples (z5–10 mg) were digested with HincII restriction enzyme and probed with the radiolabeled 39 flanking probe depicted in (A). (C and D) Northern analysis of total atrial RNA. Ten to fifteen micrograms of total atrial RNA isolated from the mice genotyped in (B) were loaded per lane. The same Northern blot was hybridized with random-primed PCR-generated probes specific for GIRK4 (C) and GIRK1 (D). The diffuse band visible below the 4.4 kb marker in (C) corresponds to the location of the 28S ribosomal RNA band. (E) Western blots of crude atrial membrane proteins from wild-type and knockout mice (two samples each). The blot was probed with a GIRK1 antibody (top), stripped, and reprobed with a GIRK4 antibody (bottom). The GIRK1 antibody recognizes both glycosylated (GIRK1gly) and unglycosylated (GIRK1-core) forms of GIRK1.

which transmitted the GIRK4 mutation to progeny. Pure strain (129Sv/J) and mixed strain (129Sv/J 3 C57Bl6/J) crosses of heterozygous mice were established to generate mice homozygous for the GIRK4 mutation. Viable GIRK4-deficient mice derived from both ES cell clones were identified by Southern blotting and/or a PCRbased screen (Figure 1B). The genotype distribution

Abnormal Heart Rate Regulation in GIRK42/2 Mice 105

Figure 2. Electrophysiological Analysis of Mouse Atrial Myocytes (A) Single-channel recording from a wild-type atrial myocyte. With 10 mM ACh in the pipette, IKACh activity was observed in the cell-attached configuration but declined after patch excision. Channel activity was restored upon addition of GTPgS to the bath. Recording artifacts due to addition of GTPgS to the bath have been deleted. The asterisk indicates the recording segment expanded in the inset. (B) Single-channel recording from a knockout atrial myocyte. No IKACh activity was observed in either the cell-attached or inside-out configuration (n 5 49). In both (A) and (B), holding potential was 280 mV with periodic voltage jumps to 80 mV to test the inward rectification of the observed channels.

from heterozygous intercrosses was as follows: wild type, 26%; heterozygous, 52%; and homozygous null, 22%; based on screening of 350 offspring. There was no significant gender bias for inheritance of the targeted allele (males, 49%; females, 51%). GIRK4-deficient mice appeared normal by all visual criteria. A Northern blot of total RNA isolated from atria from wild-type, heterozygous, and knockout siblings was blotted sequentially with radiolabeled probes specific for GIRK4 (Figure 1C) and GIRK1 (Figure 1D), as well as for GIRK2, GIRK3, and b-actin (data not shown). As expected, GIRK4 mRNA was absent in knockout mice but present in wild-type and heterozygous mice. Interestingly, GIRK1 mRNA levels were constant across all genotypes. No expression of GIRK2 or GIRK3 mRNA was detected by Northern blotting or RT-PCR (data not shown) in atrial tissue for any of the genotypes, arguing that these functional GIRK4 homologs do not compensate for the GIRK4 deficiency in heart. Lack of atrial GIRK4 protein in knockout mice was demonstrated by Western blotting of atrial membrane proteins (Figure 1E, bottom). Furthermore, knockout mouse atria contained significantly decreased (4- to 8-fold) levels of GIRK1 protein as ascertained by a carboxy-terminal–directed GIRK1 antibody (Figure 1E, top). No discernable difference was detected in GIRK1 and GIRK4 protein levels between heterozygous and wild-type mice.

To test whether disrupting the GIRK4 gene eliminated the function of cardiac IKACh, we patch clamped isolated atrial myocytes from wild-type, heterozygous, and knockout mice. Characteristic IKACh activity was observed in 47 of 50 patches from wild-type and heterozygous mice in either the cell-attached or inside-out configuration. The three cells from wild-type mice lacking observable IKACh activity may have been fibroblasts, which comprise a part of all cardiac cultures. The primary biophysical characteristics of mouse cardiac IKACh were indistinguishable from previously described rat and guinea pig atrial IKACh (Logothetis et al., 1987; Kurachi et al., 1992; Wickman et al., 1994); the single channel conductance (measured at 280 mV) was 31 6 1 pS (6SD), and the average open time duration was 2.1 ms. The channel was activated by ACh in the cell-attached configuration and by GTPgS in the inside-out configuration (Figure 2A). No IKACh activity was observed in patches from knockout mice (n 5 49; Figure 2B). Patch vesicle formation cannot explain the lack of observed IKACh activity in knockout cardiomyocytes, as other cardiac potassium channels were observed with characteristic frequencies (I KATP, 30/49; IK1, 4/49). This finding proves that GIRK4 is required for IKACh formation, supports data suggesting that GIRK1 alone is unable to form functional K1 channels (Krapivinsky et al., 1995a, 1995b; Hedin et al., 1996), and argues that the functional homologs GIRK2 and/or

Neuron 106

Figure 3. ECG, HR Fluctuations, and Power Spectra of GIRK4-Deficient Mice (A) One-lead baseline ECG recorded simultaneously from wild-type, heterozygous, and knockout mice. (B) Comparison of beat-to-beat changes in RR intervals (time interval between successive R peaks on an ECG) between wild-type and knockout littermates at baseline. RR intervals, the intervals between successive heart beats (ventricular depolarizations), are plotted against time (s). Note the spontaneous slowing of heart rate in the wild-type animal (bar). (C) Power spectra of the RR signal (HRV spectra) from a wild-type (left) and knockout (right) mouse at rest. HRV in the LF (0.4–1.5 Hz) range is nearly absent in the knockout animal. (D) One-lead ECG from the same mice as in (A), after methoxamine administration. Bradycardia is more pronounced in wild-type and heterozygous mice. (E) Comparison of beat-to-beat changes in RR intervals between wild-type and knockout mice after administration of methoxamine. Note that the RR interval fluctuations in the wild-type mouse contrast dramatically with the steady heart rate of the knockout mouse. (F) Power spectra of the RR signal from a wild-type (left) and knockout (right) mouse after methoxamine administration. Again, HRV in the LF range is nearly absent in the knockout mouse. Due to the large HRV increase in the wild-type mouse, the y-axis was contracted 25-fold.

GIRK3 did not compensate for the GIRK4 deficiency in atrial tissue. Given the absence of cardiac IKACh in GIRK4-deficient atrial myocytes, we turned to analysis of heart function in unrestrained conscious mice using electrocardiogram (ECG) telemetry to determine the physiological significance of this conductance. Surprisingly, resting mean heart rate did not differ between knockout and wildtype mice (647.2 and 646.8 bpm, respectively). At rest, the most striking difference between wild-type and knockout siblings was that wild-type mice often exhibited episodes of mild bradycardia with rapid onset lasting 5–10 s (Figures 3A and 3B). In contrast, the heart rates of GIRK4-deficient mice did not change over time scales shorter than z2 s, with the exception of fast (2–4 Hz) oscillations of low amplitude (1–2 ms). This is reflected in the HRV spectra from wild-type and knockout mice (Figure 3C). Analogous to studies of HRV in large mammals, the spectra were subdivided into three frequency ranges and adjusted 10-fold to account for the faster murine heart rate: high frequency (HF, 1.5–5 Hz), low frequency (LF, 0.4–1.5 Hz) and very low frequency (VLF, ,0.4 Hz) (Bigger et al., 1992). The HRV spectra of knockout mice consisted of two distinct components: a small peak in the HF range (2–4 Hz) and a

steep “1/f-type” component in the VLF range. This 1/ftype component is observed in all HRV spectra, including those of wild-type mice. In contrast to wild-type mice, the HRV spectra of GIRK4-deficient mice had minimal power in the LF range (0.00709 versus 0.0863 ms 2/s, respectively, p , 0.001; Figure 4A). Spectral power in the HF range was also significantly lower in knockout mice (0.0319 versus 0.0780 ms2/s, p , 0.05), whereas the VLF component of HRV did not differ between mice of different genotypes. We tested a panel of pharmacological agents, which selectively target signaling pathways relevant to heart rate regulation. Methoxamine and CCPA (2-chloro, N6cyclopentyl adenosine) were used to test the effects of GIRK4 ablation on vagal-mediated and vagus-independent signaling pathways thought to culminate in IKACh activation. Methoxamine, an agonist of a1-adrenergic receptors, induces bradycardia by causing vasoconstriction, leading to elevated blood pressure and an increase in vagal activity through baroreflex stimulation. Thus, methoxamine indirectly activates the cardiac muscarinic receptors relevant to vagal modulation of heart rate. Baroreflex stimulation is used in clinical practice to assess vagal bradycardia (Eckberg et al., 1971; Farrel et al., 1991). We felt that methoxamine was preferable to

Abnormal Heart Rate Regulation in GIRK42/2 Mice 107

Figure 5. HRV of Wild-Type and GIRK4-Deficient Mice after Atropine Atropine predominantly decreases the LF component of HRV, with the larger decrease observed in wild-type mice (black bars). Interestingly, knockout mice (white bars) still exhibited lower HRV in both HF and LF ranges after atropine. Note that the logarithmic scale on the left applies to the HRV indices, whereas the linear scale on the right applies to heart rate. Error bars indicate 95% confidence intervals, and the asterisk indicates the statistical significance of the difference between wild-type and knockout mice (*p , 0.05).

Figure 4. HRV of GIRK4-Deficient Mice (A) Quantitative comparison of basal HRV and heart rate between wild-type (black bars) and knockout mice (white bars). Note that the logarithmic scale on the left applies to the HRV indices, whereas the linear scale on the right applies to heart rate. Although the mean heart rate is the same in wild-type and knockout mice, LF and HF components of HRV are lower in knockout mice. (B) Comparison of HRV and heart rate between wild-type and knockout mice after methoxamine administration. In wild-type mice, heart rate is lower and HRV is higher in all frequency bands. (C) Comparison of HRV and heart rate between wild-type and knockout mice after CCPA adminstration. As with methoxamine, wild-type mice have lower heart rate and higher HRV in all frequency bands. For all figures, the error bars indicate 95% confidence intervals, and symbols indicate the statistical significance of the difference between wild-type and knockout mice (*p , 0.05, **p , 0.01, †p , 0.001).

direct or indirect muscarinic agonists (ACh, carbachol, bethanechol, organophosphates), since most of these substances applied to the whole animal would likely elicit many noncardiac effects, including bronchial secretion, diarrhea, and hypotension, which would confound our interpretation. CCPA is thought to activate IKACh via stimulation of sinoatrial nodal A1 adenosine receptors. CCPA is preferred in this application over adenosine because it is a selective A1 receptor agonist, it can be administered intraperitoneally, and it exhibits

longer lasting effects. Furthermore, adenosine administration would be likely to stimulate A 2 as well as A1 receptors, leading to hypotension and secondary autonomic changes (Conti et al., 1993). Shortly after methoxamine administration (usually within 2–3 min), wild-type mice developed bradycardia (Figure 3D). While the heart rate of GIRK4-deficient mice also decreased after methoxamine administration, the effect was less than that observed in wild-type siblings (564 6 30 versus 463 6 44 bpm, respectively, p , 0.01; Figure 4B). Following methoxamine administration, wildtype mice also exhibited a dramatic increase in HRV, especially in the HF and LF ranges; the increase in HRV seen in wild-type mice was absent in their knockout littermates (Figure 3E). The LF component of HRV after methoxamine administration was more than 1000-fold lower in knockout mice compared to their wild-type siblings (0.00362 versus 8.04 ms2/s, p , 0.01; Figure 4B). Similarly, CCPA administration caused bradycardia in both wild-type and knockout mice, with the effect being more pronounced in wild-type mice (469 6 65 versus 230 6 33 bpm, knockout versus wild type; p , 0.001). CCPA also led to a large increase in HRV in wild-type mice, particularly in the HF and LF ranges. After CCPA administration, the LF component of HRV was more than 1000-fold lower in knockout mice (0.0272 versus 59.7 ms 2/s, p , 0.001; Figure 4C). Because adjustments in autonomic nervous system output are the primary source of HRV, we tested the effects of parasympathetic (atropine) and sympathetic (propranolol) blockade on heart rate and HRV in wildtype and GIRK4-deficient mice. Compared to saline injection, atropine did not cause significant tachycardia in wild-type (674 and 667 bpm, atropine versus saline) or knockout (661 versus 699 bpm, atropine versus saline) mice. However, the LF component of HRV was decreased after atropine administration compared to saline injection in both wild-type (0.0255 versus 0.115 ms 2/s, p , 0.02; Figure 5) and knockout mice (0.00641 versus 0.0109 ms2 /s, p , 0.01). Interestingly, atropine

Neuron 108

did not completely eliminate the difference between wild-type and knockout mice, since the differences in both LF and HF components of HRV after atropine injection are still significant (LF, 0.00641 versus 0.0255 ms 2/s, p , 0.02; HF, 0.0223 versus 0.0918 ms2/s, p , 0.05; knockout versus wild type). Propranolol administration caused bradycardia in both wild-type and knockout mice (data not shown); the effect on heart rate was profound (z300 bpm decrease) in certain mice regardless of genotype. HRV spectra of these mice were distorted and shifted toward lower frequencies. However, since the profound effects were observed only in certain mice from both genotypes, the confidence intervals were wide, and the difference in heart rate or HRV indices between wild-type and knockout mice after propranolol did not reach statistical significance. Heart rate and HRV indices of heterozygous mice did not differ significantly from that of wild-type mice after atropine, propranolol, methoxamine, or CCPA. On pairwise comparison, the high frequency range of HRV at baseline was higher for heterozygous than for wild-type mice. However, this result is not significant when corrected for multiple comparisons by the Bonferroni method; in contrast, the difference between wild-type and knockout mice is highly significant after Bonferroni correction (p , 0.001). Since both vagal tone and adenosine have been implicated in the control of atrioventricular conduction properties (DiMarco et al., 1983; Clemo and Belardinelli, 1986; Kurachi et al., 1986), we measured the effects of GIRK4 disruption on PR interval duration (time interval between atrial and ventricular depolarization) at baseline and after methoxamine administration. The small difference in PR interval duration between wild-type and GIRK4-deficient mice was of borderline statistical significance (32.1 6 2.6 versus 34.9 6 2.8 ms, knockout versus wild type; p , 0.05). Since this difference disappeared after CCPA administration (37.7 6 4.1 versus 36.9 6 3.6 ms, knockout versus wild type), the importance of this finding is uncertain, but it appears that IKACh is not critical for atrioventricular conduction properties in mice. Furthermore, there was no obvious difference in the frequency of ectopic beats between the wild-type and knockout mice. In .3 hr of analyzed baseline data, no more than 10 ventricular ectopic beats were detected in any mouse. Apart from one wild-type mouse with frequent supraventricular ectopy (.200 ectopic beats over 3 hr), supraventricular ectopic beats were infrequent in mice of all genotypes.

Discussion While the brain probably assembles G protein–gated K1 channels from the entire repertoire of available GIRK subunits (GIRK1, GIRK2, GIRK3, and GIRK4), cardiac IKACh is comprised entirely of associated GIRK1 and GIRK4 subunits (Krapivinsky et al., 1995a). Experiments in heterologous expression systems suggest that the GIRK1 subunit is unable to form a functional channel when expressed alone (Krapivinsky et al., 1995; Hedin et al., 1996). GIRK4 alone, in contrast, does form a G protein–gated K1 channel, albeit with single channel

characteristics dissimilar to any previously reported native conductance (Krapivinsky et al., 1995a, 1995b). Thus, we speculated that disruption of the mouse GIRK4 gene would effectively eliminate all cardiac G protein– gated atrial K1 conductance. Indeed, the absence of IKACh activity in GIRK4-deficient atrial myocytes isolated from mice derived from two independent targeted ES cell clones is strong evidence that GIRK4 is required for IKACh formation. Furthermore, a preliminary attempt at functional IKACh reconstitution in atrial myocytes isolated from GIRK4-deficient mice by transfection of a rat GIRK4 cDNA was successful (data not shown). The lack of IKACh in GIRK4-deficient mice also indicates that GIRK4 functional homologs (GIRK2 and GIRK3) do not compensate for this deficiency in atrial tissue. The level of GIRK1 mRNA was similar in wild-type, heterozygous, and knockout mice, while GIRK1 protein was significantly reduced in knockout mice. Interestingly, virtually no GIRK1 protein was detected using antibodies directed against a short amino-terminal GIRK1 sequence, which readily detect GIRK1 in wild-type atria (data not shown). In contrast, a relatively modest decrease (5-fold) in GIRK1 protein in knockout mice was observed using a carboxy-terminal GIRK1 antibody. This observation suggests that amino-terminal proteolysis of GIRK1 might occur to a greater extent in the absence of GIRK4 and could explain the similar dramatic decrease in GIRK1 protein observed in GIRK2 knockout mice, also assessed using an amino-terminal–directed GIRK1 antibody (Signorini et al., 1997). Regardless of the mechanism, the GIRK2 and GIRK4 knockout studies both suggest the existence of a posttranscriptional mechanism for regulation of GIRK subunit levels. Remarkably, resting heart rate in unrestrained conscious wild-type and GIRK4 knockout mice was virtually identical, indicating that other signaling pathways involved in heart rate regulation can balance the missing IKACh current. Other groups have reported no change in heart rate in mice overexpressing a b-adrenergic receptor kinase 1 (bARK1) inhibitor (Koch et al., 1995), G protein–coupled receptor kinase 5 (GRK5; Rockman et al., 1996), and the b1-adrenergic receptor (Bertin et al., 1993), as well as in b1-adrenergic receptor knockout mice (Rohrer et al., 1996). Thus, mean resting heart rate is determined by several redundant signaling pathways. Elevated basal heart rate has been reported in mice with cardiac overexpression of the b2-adrenergic receptor (Milano et al., 1994) and Gas (Iwase et al., 1996), whereas decreased heart rate was found in mice overexpressing bARK1 (Koch et al., 1995) and in tyrosine hydroxylase knockout mouse embryos (Zhou et al., 1995). However, in some reports the heart rate was measured during anesthesia or control infusion and may not reflect true resting heart rate of the mouse. We report a higher resting mean heart rate (650 bpm) for mice than other groups (450–550 bpm). However, many of these previous studies measured mean heart rate during general anesthesia (Bertin et al., 1993; Koch et al., 1995; Iwase et al., 1996) or in tethered and cannulated mice (Rockman et al., 1996; Rohrer et al., 1996; Desai, 1997). It is conceivable that cannulation of the carotid artery could lead to baroreflex stimulation or that the tethered animals were less active than unrestrained

Abnormal Heart Rate Regulation in GIRK42/2 Mice 109

Figure 6. Mechanisms of Heart Rate Regulation A schematic compilation of heart rate regulation data from mice, rabbits, and guinea pig (DiFrancesco and Tortora, 1991; Hartzell et al., 1991; Freeman and Kass, 1993; Huang et al., 1994; Guo et al., 1995; Noma, 1996). Sympathetic stimulation (norepinephrine) initiates a cAMP cascade, which directly modulates the If conductance and indirectly modulates the activity of several ion channels. Parasympathetic stimulation (acetylcholine) and adenosine counteract the modulatory cAMP pathways via inhibition of adenylyl cyclase as well as directly stimulating IKACh . Abbreviations: NE, norepinephrine; b 1-AR, type 1 b-adrenergic receptor; ACh, acetylcholine; M2, type 2 muscarinic receptor; Ado, adenosine; A1, type 1 adenosine receptor; PKA, protein kinase A; Ga s, G protein a subunit (stimulates adenylyl cyclase); Gai, G protein a subunit (inhibits adenylyl cyclase); Gbg, G protein bg subunit; If, pacemaker current; Ist, sustained inward current; IK, delayed rectifier potassium current; L-Ca 21, L-type calcium channel; and IKACh, muscarinic-gated atrial potassium channel.

mice. Alternatively, it is possible that the implanted transmitters constituted a significant stress for our mice, elevating heart rate relative to that of nonimplanted mice. Other groups that have used implanted telemetry devices have reported resting murine heart rates similar to those of our mice (Ishii et al., 1996; Johansson and Thoren, 1997). One study found that on days 3 and 4 after implantation, resting heart rate was 600–650 bpm. After 1 week, resting heart rate decreased to 500–550 bpm (Johansson and Thoren, 1997). Thus, surgeryrelated stress or other implantation-related factors could have persisted in our mice until day 4 when we acquired the resting mean heart rate data, diminishing a real heart rate difference between wild-type and GIRK4deficient mice. The diminished bradycardic response of the GIRK4 knockout mice to CCPA and methoxamine indicates that IKACh is responsible for approximately half of the heart rate decrease in response to parasympathetic stimulation and to adenosine administration in vivo. We suspect that the residual bradycardic effects of CCPA and methoxamine in knockout mice are due to decreases in depolarizing current(s) (I f , Ist, L-type voltage-gated Ca21 conductance) directly or indirectly modulated by the cAMP pathway (Figure 6). Theoretically, one could dissect the residual cAMP-related cardiac signaling pathways using IKACh-deficient sinoatrial nodal cells and pharmacological tools such as PKA inhibitors in whole-cell experiments. Atrial myocytes are probably not suitable for such assessments since they may not contain the same set of ion channels required for impulse generation and modulation in sinoatrial nodal cells (DiFrancesco, 1995a, 1995b; Guo et al., 1995). Thus, the exact identities of the residual cardiac pacing systems in the GIRK4 knockout mice will require the rigorous electrophysiological characterization of murine sinoatrial nodal cells. HRV analysis was performed in addition to measurement of heart rate, because changes in certain frequency components of HRV can reveal abnormalities in autonomic regulation of cardiac pacing that are not necessarily reflected in changes in mean heart rate measured over minutes or hours. Indeed, the most striking

difference between wild-type and GIRK4-deficient mice is the decreased HRV in the latter. GIRK4 knockout mice have much lower HRV at frequencies above 0.4 Hz. At rest, wild-type mice exhibit occasional rapid-onset episodes of mild bradycardia that last a few seconds, presumably related to a transient increase in vagal activity. While a similar phenomenon has been described previously in rats (Carre et al., 1994), this is not observed in GIRK4-deficient mice. The heart rates of GIRK4-deficient mice change little apart from small-amplitude regular oscillations at 2–4 Hz frequency; when heart rate does change, it occurs gradually over several seconds. The difference in HRV between wild-type and knockout mice is dramatic after vagal stimulation with methoxamine. Methoxamine leads to a major increase in HRV in wild-type but not GIRK4-deficient mice, especially in the LF range. In experiments involving direct vagal stimulation of open-chest dogs, the sinus node responds to vagal stimulation much more rapidly than it does to stimulation of sympathetic fibers (Berger et al., 1989). The cellular response to vagal stimulation is likely to be faster because it utilizes the faster membrane-delimited mechanism involving IKACh, whereas the sympathetic response relies on slower changes in intracellular cAMP levels (Hartzell et al., 1991). The results from this study are consistent with this proposition, because although GIRK4 knockout mice can still decrease their heart rate in response to M2 and A1 receptor stimulation, their mean resting heart rate changes more slowly than that of their wild-type siblings. In contrast to observations in large mammals, atropine did not cause tachycardia in wild-type mice. Atropine did decrease HRV, with the most pronounced effect seen in the low frequency range. Both of these observations in mice have been reported previously (Mansier et al., 1996). Presumably, mice have lower resting vagal tone than large mammals, with most of their vagally mediated HRV concentrated in the LF (0.4–1.5 Hz) range. As expected, the decrease in the LF contribution to HRV was larger in the wild-type animals. Surprisingly however, GIRK4-deficient mice had lower HRV in both LF and HF ranges even after atropine administration.

Neuron 110

Given the high concentration of atropine administered, it is unlikely that this observation is due to the incomplete or slow onset of vagal blockade. A plausible explanation is that IKACh is regulated by receptors other than M2 and that this activity persists in atropinized wild-type mice. Indeed, IKACh can be activated by adenosine, somatostatin and calcitonin gene–related peptide as well as acetylcholine (Hartzell, 1979; Lewis and Clapham, 1989; Kim, 1991). Currently, however, there is no direct evidence that these substances regulate heart rate under normal conditions. The exact origin of the 2–4 Hz oscillations of heart rate in GIRK4-deficient mice is currently unknown. These oscillations are also observed in wild-type animals and persist after atropine administration. Since they persist after atropine administration in knockout mice, the oscillations are probably not vagally mediated. We speculate that they are caused by a direct mechanical stretch of the sinus node caused by intrathoracic pressure changes associated with respiration. This mechanism has been well documented for other mammals, accounting for residual respiratory sinus arrhythmia after complete autonomic blockade (Koizumi et al., 1975; Bernardi et al., 1989; Saul and Cohen, 1994). Indeed, the frequency of these oscillations (2–4 Hz) is consistent with the murine respiratory rate. GIRK4 is expressed almost exclusively in heart atria and pancreas (Ashford et al., 1994; Ferrer et al., 1995; Krapivinsky et al., 1995a). Heart atria is the ideal tissue for studying the role of a G protein–gated K1 channel in vivo, since the atrial G protein–gated K1 channel (I KACh) is comprised exclusively of GIRK1 and GIRK4. In pancreas, the GIRK4 functional homolog GIRK2 is also expressed. Based on results from in situ hybridization, immunohistochemistry, and cloning approaches, GIRK4 exhibits a highly restricted expression pattern in brain (Karschin et al., 1994, 1996; Iizuka et al., 1997; Murer et al., 1997). Furthermore, brain regions thought to express GIRK4 also express the functional homologs GIRK2 and GIRK3. The potential functional compensation provided by these homologs will probably result in more subtle phenotypes with respect to brain and pancreas function in the GIRK4-deficient mice. As was the case in heart, preliminary histology of pancreas and brain revealed no gross morphological differences associated with the GIRK4 deficiency. Studies are currently underway to reveal any functional consequences of the GIRK4 ablation in these tissues. The differences observed between wild-type and GIRK4-deficient mice with respect to cardiac function at rest and after pharmacological manipulation argue strongly that IKACh is involved in heart rate regulation in mice over a substantial range of vagal activity. This conclusion contrasts somewhat with the model proposed by DiFrancesco et al. (1989), which is based largely on electrophysiological data from isolated sinus nodal cells. This group reported that ACh applied to canine sinoatrial nodal cells at concentrations too low to elicit IKACh activity did produce a negative chronotropic response, presumably mediated by inhibition of the cAMP-dependent pacemaker current If. The authors suggested that IKACh does not play a role in heart rate regulation at low-to-moderate levels of vagal activity. In

support of this hypothesis, it was recently demonstrated that IKACh was not required for the negative chronotropic response to carbachol or adenosine seen in cardiomyocytes derived from ES cells lacking Ga i2 or Ga i3 (Sowell et al., 1997). However, the observed difference in HRV between unrestrained GIRK4-deficient mice and wildtype littermates at rest argues that IKACh is important for heart rate regulation even at baseline, when vagal tone is not elevated. The lack of substantial effect of atropine on heart rate supports our contention that vagal tone in unrestrained wild-type mice is low relative to that of large mammals. We suggest that the amount of ACh released from the vagus at baseline is sufficient to modulate both IKACh and If and possibly other relevant conductances. Certainly, this study underscores the benefits of an in vivo model for dissecting complex and convergent signaling pathways. In summary, we conclude that GIRK4 is indispensable for formation of cardiac IKACh, and that IKACh accounts for approximately half of the bradycardia induced by vagal stimulation and adenosine administration. Furthermore, this study demonstrates that IKACh is necessary for fast regulation (,2 s) of the heart rate, both at rest and during baroreflex stimulation; sudden changes in heart rate are virtually absent in GIRK4-deficient mice. Finally, the observation that atropine failed to reduce the HRV of wildtype mice to the level observed in GIRK4-deficient mice suggests that in normal mice at rest, IKACh is also regulated by nonmuscarinic agonists. We conclude that IKACh is a critical effector of the parasympathetic branch of the heart rate regulatory system in mice. Experimental Procedures Generation of Targeted ES Cells Cloning and mapping of the 129Sv/J mouse genomic DNA fragment containing the two GIRK4 coding sequence exons were described previously (Wickman et al., 1997). We replaced a z2.7 kb fragment containing coding sequence exon 1 with the neomycin phosphotransferase gene (neor) under the control of the mouse PGK promoter. The diptheria toxin gene (DTA) driven by the thymidine kinase promoter was included in the targeting vector to enrich for homologous recombinants. 129Sv/J ES cells (passage 10) were transfected with the linearized GIRK4 targeting vector as described (Spicer et al., 1995), and 480 colonies surviving G418 selection (400 mg/ml active constituent, 10 days) were picked and screened by PCR and Southern blotting for homologous recombinants (targeting frequency 5 2/480). Germline-transmitting chimeric mice were generated from both properly targeted ES cell clones. Generation of GIRK4 Knockout Mice C57Bl6/J blastocysts were isolated on day 3.5 and injected with eight to twenty ES cells. Eight to ten injected blastocysts were transferred to each uterine horn of pseudopregnant recipient females (CD1 strain) at 2.5 days post coitum. Chimeric mice were identified on the basis of agouti/chinchilla pigmentation in the coat. Male chimeric mice were crossed with both C57Bl6/J and 129Sv/J females to screen for germline transmission of the targeted GIRK4 allele and to maintain the targeted mutation on inbred (129Sv) and outbred (C57Bl6/129Sv) genetic backgrounds. All mice were genotyped by Southern blotting or by PCR screening using tail DNA prepared as described (Spicer et al., 1995). For electrophysiological experiments on atrial myocytes, F2 siblings of appropriate genotype were mated to produce pups homozygous for either the null or wildtype GIRK4 allele. For the telemetry experiments, three same-sex F2 generation littermates, one representing each genotype (1/1, 1/2, 2/2), comprised a study group and were implanted and monitored simultaneously as described below. All implanted mice were

Abnormal Heart Rate Regulation in GIRK42/2 Mice 111

5–7 months old and weighed at least 22 g. In total, ECG monitoring was performed on seven study groups (four female, three male). Six groups were derived from the same targeted ES cell clone. To confirm the specificity of the gene targeting event, a seventh group derived from a different ES cell clone was tested. Both outbred (C57Bl6/J 3 129Sv/J, three groups) and inbred (129Sv/J, four groups) triplets were analyzed. No gender or strain-associated differences were detected. Northern Blotting Total atrial RNA was isolated from six littermates using TRIzol (GibcoBRL) according to the manufacturer’s specifications. Probes (300–500 bp) specific for GIRK1–4 were prepared by PCR. The GIRK2 fragment was designed to recognize all reported alternatively spliced versions of GIRK2 (Isomoto et al., 1996). Fifteen micrograms of total RNA was loaded per lane, in a 1% GTG agarose gel made in 13 MOPS buffer. The gel was run at 5 V/cm for 5 hr and rinsed for 30 min in diethyl pyrocarbonate–treated (DEPC-treated) H2O to remove formaldehyde. RNA was capillary transferred to Hybond N1 nylon membranes using 103 SSC. The blot was then rinsed for 5 min in 53 SSC, followed by 2 hr baking at 808C. The blot was stained for 1 min in an 0.3 M Na acetate/0.03% methylene blue solution to assess the quality and quantity of the RNA. The blot was probed as described (Wickman et al., 1997). The blot was exposed to BioMax MR X-ray film for 1–5 days. The order of probing and corresponding exposure times was as follows: GIRK4, 3 days; GIRK1, 3 days; GIRK2, 5 days; GIRK3, 5 days; and b-actin, 1 day. Western Blotting Crude atrial membrane proteins were prepared from dissected atrial tissue pooled from three mice (6–8 weeks old) using a protocol modified slightly from Jones et al. (1979). Briefly, samples were thawed in 5 ml ice-cold buffer containing 100 mM NaCl, 20 mM imidazole (pH 7.5), and 1 mM dithiothreitol (DTT). The tissue was homogenized by polytron on ice using three 30 s pulses. After centrifugation at 1000 3 g for 10 min at 48C, KCl was added to a final concentration of 700 mM to solubilize contractile elements, and the samples were rotated for 30 min at 48C. Membranes were pelleted by ultracentrifugation at 50,000 rpm for 30 min using a Beckman Sv-55 rotor. The pellet was resuspended by vigorous pipetting and incubated for 1 hr on ice in a buffer containing 100 mM NaCl, 20 mM HEPES (pH 7.5), 1 mM EDTA, 1 mM DTT, and 1% CHAPS. Insoluble elements were pelleted by centrifugation at 10,000 3 g for 30 min at 48C. Soluble proteins were precipitated using 10% (v/v) trichloroacetic acid. Pellets were washed twice with 500 ml H2O, resuspended in 20 ml of SDS loading buffer, and electrophoresed in an 11.5% polyacrylamide gel. Blots were probed with either affinitypurified GIRK1 (aCsh, 2 mg/ml; Krapivinsky et al., 1995a) or GIRK4 (aCIRN2, 5 mg/ml; Krapivinsky et al., 1995a) antisera, as described. Proteins were visualized using the enhanced chemiluminescence system (ECL, Amersham Life Science) according to the manufacturer’s protocols. Electrophysiology Atrial auricles from neonatal mice (postnatal days 2–7) were microdissected from total heart tissue. Isolated cardiomyocytes were prepared using one of two protocols, the first modified slightly from Davies et al. (1996). Briefly, atrial tissue was digested using two 15 min incubations in buffer containing pancreatin (2.5 mg/ml) and collagenase type I (0.5 mg/ml) at 378C. Dispersed cells were plated onto fibronectin-coated glass coverslips in DMEM supplemented with 10% horse serum and 5% fetal bovine serum, and cultured in 10% CO2 at 378C for 2 days. Myocytes were also isolated using a myocyte isolation kit according to the manufacturer’s specifications (Worthington). Electrophysiological experiments were performed within 48 hr after plating. Pipettes for single-channel recording were pulled from KG-12 glass (Garner Glass) using a P-97 Flaming/Brown puller (Sutter Instrument), coated with HIPEC R-6101 Sylgard (Dow Corning), and fire polished. Pipette resistance was 2–5 MV when filled with K-5 solution (in mM): 118.5 KCl, 5 KOH/EGTA, 2 MgCl2 , and 10 KOH/HEPES (pH 7.2). Cell-attached and inside-out recordings were performed with K-5 in the bath and K-5 plus 10 mM ACh (Sigma) in the pipette. After patch excision, GTPgS (Boehringer

Mannheim) was added to the bath to a final concentration of 20 mM. The holding potential was 280 mV. Currents were recorded with an Axopatch 200A amplifier (Axon Instruments) and stored on videotape after 5 kHz filtering (4-pole low-pass Bessel). The signal was sampled digitally at 50 kHz and stored on computer hard disk using pCLAMP6 (Axon Instruments) software. Although no further filtering was performed on data used for quantitative analysis, data displayed in figures were low-pass filtered at 2 kHz with an 8-pole Butterworth filter prior to digitization. To estimate single channel conductances and mean open times, idealized traces were created from 30 s data segments in the inside-out configuration in 10 patches (“dead time,” 200 ms). Single channel amplitude was fitted with a Gaussian curve using the pCLAMP6 software. The arithmetic average of open time durations was computed using software written in TurboPascal v 7.0 (Borland International). Telemetry Implantable PhysioTel TA10EA-F20 radiotransmitters (Data Sciences International) were used for the ECG telemetry monitoring. Transmitters were implanted under the skin of the back under pentobarbital anesthesia (30–50 mg/kg i.p.) and fixed to the back muscles with Prolene 5-0 sutures. The ECG leads were tunneled under the skin to the ventral aspect of the thorax and sutured to the thoracic muscles in lead II position. Prolene 5-0 was also used to close the incisions. After implantation (day 1), mice were allowed to recover. Twenty-four hours of baseline ECG recording began on the morning of day 4 (10 to 12 a.m.). On day 5, animals were injected i.p. with an 0.9% saline solution (20 ml/kg) as a vehicle control, followed by at least 135 min with atropine sulphate (1 mg/kg; Fluka Chemical). After a minimum of 6 hr, propranolol was administered (20 mg/kg; Fluka Chemical). On day 6, mice were injected i.p. with DMSO (3 ml/kg of 0.1% solution in saline) as a vehicle control. After at least 135 min, mice were injected i.p. with methoxamine (6 mg/kg; RBI). At least 6 hr later, CCPA (2-chloro, N6-cyclopentyl adenosine; Sigma), a selective long-acting adenonsine A 1 receptor agonist, was injected i.p. (0.3 mg/kg in 0.1% DMSO). ECGs were recorded for 15 min before and after each injection. Animals were sacrificed by ether inhalation, and transmitters were explanted and reused after cleaning and sterilization with 2% glutaraldehyde. HRV Analysis The analog signal from the Data Sciences International receiver was digitized with 12-bit precision using a Digidata 1200B board (Axon Instruments) using Axoscope software (Axon Instruments). Sampling frequency was 2 kHz for the 24 hr recordings and 5 kHz for the pharmacologic experiments. Custom-made software was used to detect the R peaks of the ECG signal and to calculate the RR intervals. Mislabeled R peaks were corrected and noisy data segments were replaced manually with a linear spline. Resulting RR data was interpolated linearly at 50 ms intervals to create equidistant points suitable for Fast Fourier Transform (FFT). Three epochs lasting 1 hr each were analyzed from each 24 hr recording (4 to 5 p.m., midnight to 1 a.m., and 8 to 9 a.m.). All 1 hr epochs and 15 min drug recordings were divided into 204.8 s segments, overlapping by 102.4 s. After linear detrending and windowing (3-term Blackman-Harris window [Harris, 1978]), FFT was performed. Mean heart rate and HRV in HF (1.5–5.0 Hz), LF (0.4–1.5 Hz), and VLF (0.015–0.4 Hz) ranges were computed from each data segment and normalized for segment duration. Cutoff frequencies for the HF, LF, and VLF bands were based on those used in human studies multiplied by a factor of 10 (the approximate ratio between murine and human HR [Bigger et al., 1992]). Baseline segments with .10 s of splined data were excluded from analysis. Mean values of heart rate (arithmetic) and HRV indices (geometric) were computed from 102 analyzed baseline data segments and from the seven data segments comprising the 15 min of data before and after each injection. Data from three mice (two 1/1, one 1/2) exhibiting intermittent rhythm disturbances (intermittent A-V block or intermittent junctional rhythm) were excluded from analysis. All other animals exhibited sinus rhythm. In all but the first study group, the investigator performing the ECG monitoring and HRV analysis was blinded to the mouse genotype. Paired t test (littermates from the same group were paired) was used to compare the heart rate and logarithms of the HRV indices of the wild-type

Neuron 112

and knockout mice. Fifty PR intervals were measured manually from a baseline recording acquired at 5 kHz, and the arithmetic mean was calculated for each animal and genotype. Heart rate and HRV of heterozygous mice did not differ significantly from wild-type siblings. Values of p , 0.05 were considered statistically significant. The Bonferroni method was used to correct for multiple comparisons when differences between the whole groups (wild type versus knockout and wild type versus heterozygote) were evaluated. Acknowledgments We would like to acknowledge Dr. Andrew Spicer, Stephanie Munger, Suresh Savarirayan, Dr. Matthew Kennedy, Dr. J. P. Saul, Dr. Marina Picciotto, Luba Krapivinsky, Anita Jennings, Dr. Gary Mitchell, and Drs. Marc and Janice Pfeffer for their technical assistance, kind advice, or for reviewing the manuscript. This work was supported by a National Institutes of Health grant awarded to D. E. C. and S. J. G. and a National Institutes of Health training grant to K. W. All mouse experiments were conducted according to protocols approved by the Children’s Hospital Animal Care and Utilization Committee. Received November 25, 1997; revised December 23, 1997. References

and with exercise: new tools to study models of cardiac disease. Am. J. Physiol. 272, H1053–H1061. DiFrancesco, D. (1995a). Cardiac pacemaker: 15 years of “new” interpretation. Acta Cardiol. 50, 414–427. DiFrancesco, D. (1995b). The onset and autonomic regulation of cardiac pacemaker activity: relevance of the f current. Cardiovasc. Res. 29, 449–456. DiFrancesco, D., and Tromba, C. (1988). Inhibition of the hyperpolarizing-activated current, If , induced by acetylcholine in rabbit sinoatrial node myocytes. J. Physiol. 405, 477–491. DiFrancesco, D., and Tortora, P. (1991). Direct activation of cardiac pacemaker channels by intracellular cyclic AMP. Nature 351, 145–147. DiFrancesco, D., Ducouret, P., and Robinson, R. (1989). Muscarinic modulation of cardiac rate at low acetylcholine concentrations. Science 243, 669–671. DiMarco, J., Sellers, T., Berne, R., West, G., and Bellardinelli, L. (1983). Adenosine: electrophysiological effects and therapeutic use for terminating paroxysmal supraventricular tachycardia. Circulation 68, 1254–1263. Duprat, F., Lesage, F., Guillemare, E., Fink, M., Hugnot, J., Bigay, J., Lazdunski, M., Romey, G., and Barhanin, J. (1995). Heterologous multimeric assembly is essential for K 1 channel activity of neuronal and cardiac G-protein-activated inward rectifiers. Biochem. Biophys. Res. Comm. 212, 657–663.

Akselrod, S., Gordon, D., Ubel, F., Shannon, D., Barger, A., and Cohen, R. (1981). Power spectrum analysis of heart rate fluctuations. Science 213, 220–222.

Eckberg, D., Drabinsky, M., and Braunwald, E. (1971). Defective cardiac parasympathetic control in patients with heart disease. N. Engl. J. Med. 285, 877–883.

Ashford, M., Bond, C., Blair, T., and Adelman, J. (1994). Cloning and functional expression of a rat heart K ATP channel. Nature 370, 456–459.

Ewing, D., Borsey, D., Bellavere, F., and Clarke, B. (1981). Cardiac autonomic neuropathy in diabetes: comparison of measures of R-R interval variation. Diabetologia 21, 18–24.

Berger, R., Saul, J., and Cohen, R. (1989). Transfer function analysis of autonomic regulation: I. Canine atrial rate response. Am. J. Physiol. 256, H142–H152.

Farrel, T., Paul, V., Cripps, T., Malik, M., Bennet, E., Ward, D., and Camm, A. (1991). Baroreflex sensitivity and electrophysiological correlates in patients after acute myocardial infarction. Circulation 83, 945–952.

Bernardi, L., Keller, F., and Sanders, M. (1989). Respiratory sinus arrhythmia in the denervated human heart. J. Appl. Physiol. 67, 1447–1455. Bertin, B., Mansier, P., Makeh, I., Briand, P., Rostene, W., Swynghedauw, B., and Strosberg, A. (1993). Specific atrial overexpression of G protein coupled human beta 1 adrenoceptors in transgenic mice. Cardiovasc. Res. 27, 1606–1612. Bigger, J.J., Fleiss, J., Steinman, R., Rolnitzky, L., Kleiger, R., and Rottmans, J. (1992). Frequency domain measures of heart period variability and mortality after myocardial infarction. Circulation 85, 164–171. Bywater, R., Campbell, G., Edwards, F., and Hirst, G. (1990). Effects of vagal stimulation and applied acetylcholine on the arrested sinus venosus of the toad. J. Physiol. 425, 1–27. Carre, E., Maison-Blanche, P., Mansier, P., Chevalier, B., Charlotte, N., Dakhli, T., Coumel, P., and Swynghedaw, B. (1994). Phenotypic determinants of heart rate variability in cardiac hypertrophy and failure. Eur. Heart J. 15, 58–62. Clemo, H., and Belardinelli, L. (1986). Effect of adenosine on atrioventricular conduction. I. Site and characterization of adenosine action in guinea pig atrioventricular node. Circ. Res. 59, 427–436. Conti, A., Monopoli, A., Gamba, M., Borea, P., and Ongini, E. (1993). Effects of selective A 1 and A 2 adenosine receptor agonists on cardiovascular tissues. Naunyn Schmiedeberg Arch. Pharmacol. 348, 108–112. Dascal, N., Schreibmayer, W., Lim, N., Wang, W., Chavkin, C., DiMagno, L., Labarca, C., Kieffer, B., Gaveriaux, R.C., Trollinger, D., et al. (1993). Atrial G protein-activated K1 channel: expression cloning and molecular properties. Proc. Natl. Acad. Sci. USA 90, 10235– 10239. Davies, M., An, R., Doevendans, P., Kubalak, S., Chien, K., and Kass, R. (1996). Developmental changes in ionic channel activity in the embryonic murine heart. Circ. Res. 78, 15–25. Desai, K.H., Sato, R., Schauble, E., Barsh, G.S., Kobilka, B.K., and Bernstein, D. (1997). Cardiovascular indexes in the mouse at rest

Ferrer, J., Nichols, C., Makhina, E., Salkoff, L., Bernstein, J., Gerhard, D., Wasson, J., Ramanadham, S., and Permutt, A. (1995). Pancreatic islet cells express a family of inwardly rectifying K1 channel subunits, which interact to form G-protein-activated channels. J. Biol. Chem. 270, 26086–26091. Freeman, L., and Kass, R. (1993). Delayed rectifier potassium channels in ventricle and sinoatrial node of the guinea pig: molecular and regulatory properties. Cardiovasc. Drugs Ther. 7 (suppl.), 627–635. Garnier, D., Nargeot, J., Ojeda, C., and Rougier, O. (1978). The action of acetylcholine on background conductance in frog atrial trabeculae. J. Physiol. 274, 381–396. Giles, W., and Noble, S. (1976). Changes in membrane currents in bullfrog atrium produced by acetylcholine. J. Physiol. 261, 103–123. Guo, J., Ono, K., and Noma, A. (1995). A sustained inward current activated at the diastolic potential range in rabbit sinoatrial node cells. J. Physiol. (Lond.) 483, 1–13. Hagiwara, N., Irisawa, H., and Kameyama, M. (1988). Contribution of two types of calcium currents to the pacemaker potentials of rabbit sinoatrial node cells. J. Physiol. 395, 233–253. Harris, F. (1978). On the use of windows for harmonic analysis with the discrete Fourier transform. Proc. Inst. Electrical Electronic Eng. 66, 51–83. Hartzell, H. (1979). Adenosine receptors in frog sinus venosus: slow inhibitory potentials produced by acetylcholine and adenine compounds. J. Physiol. 293, 23–49. Hartzell, H., Mery, P.-F., Fischmeister, R., and Szabo, G. (1991). Sympathetic regulation of cardiac calcium current is due exclusively to cAMP-dependent phosphorylation. Nature 351, 573–576. Hedin, K., Lim, N., and Clapham, D. (1996). Cloning of a Xenopus laevis inwardly rectifying K 1 channel subunit that permits GIRK1 expression of IKACh currents in oocytes. Neuron 16, 423–429. Hirst, G., Bramich, N., Edwards, F., and Klemm, M. (1992). Transmission at autonomic neuroeffector junctions. Trends Neurosci. 15, 40–46.

Abnormal Heart Rate Regulation in GIRK42/2 Mice 113

Huang, C., Slesinger, P., Casey, P., Jan, Y., and Jan, L. (1995). Evidence that direct binding of Gbg to the GIRK1 G protein–gated inwardly rectifying K1 channel is important for channel activation. Neuron 15, 1133–1143. Huang, X., Morielli, A., and Peralta, E. (1994). Molecular basis of cardiac potassium channel stimulation by protein kinase A. Proc. Natl. Acad. Sci. USA 91, 624–628. Iizuka, M., Tsunenari, I., Momota, Y., Akiba, I., and Kono, T. (1997). Localization of a G-protein-coupled inwardly rectifying K1 channel, CIR, in the rat brain. Neuroscience 77, 1–13. Inanobe, A., Morishige, K., Takahashi, N., Ito, H., Yamada, M., Takumi, T., Nishina, H., Takahashi, K., Kanaho, Y., Katada, T., et al. (1995). G beta gamma directly binds to the carboxyl terminus of the G protein–gated muscarinic K1 channel, GIRK1. Biochem. Biophys. Res. Comm. 212, 1022–1028. Ishii, K., Kuwahara, M., Tsubone, H., and Sugano, S. (1996). The telemetric monitoring of heart rate, locomotor activity, and body temperature in mice and voles (Microtus arvalis) during ambient temperature changes. Lab. Anim. 30, 7–12. Isomoto, S., Kondo, C., Takahashi, N., Matsumoto, S., Yamada, M., Takumi, T., Horio, Y., and Kurachi, Y. (1996). A novel ubiquitously distributed isoform of GIRK2 (GIRK2B) enhances GIRK1 expression of the G-protein-gated K1 current in Xenopus oocytes. Biochem. Biophys. Res. Comm. 218, 286–291. Iwase, M., Bishop, S., Uechi, M., Vatner, D., Shannon, R., Kudej, R., Wight, D., Wagner, T., Ishikawa, Y., Homcy, C., and Vatner, S. (1996). Adverse effects of chronic endogenous sympathetic drive induced by cardiac Gsa overexpression. Circ. Res. 78, 517–524.

Krapivinsky, G., Krapivinsky, L., Wickman, K., and Clapham, D. (1995c). Gbg binds directly to the G protein–gated K1 channel IKACh. J. Biol. Chem. 270, 29059–29062. Kubo, Y., Reuveny, E., Slesinger, P., Jan, Y., and Jan, L. (1993). Primary structure and functional expression of a rat G-protein-coupled muscarinic potassium channel. Nature 364, 802–806. Kurachi, Y., Nakajima, T., and Sugimoto, T. (1986). On the mechanism of activation of muscarinic K1 channels by adenosine in isolated atrial cells: involvement of GTP-binding proteins. Pflu¨gers Arch. 407, 264–274. Kurachi, Y., Tung, R., Ito, H., and Nakajima, T. (1992). G protein activation of cardiac muscarinic K1 channels. Prog. Neurobiol. 39, 229–246. Lesage, F., Duprat, F., Fink, M., Guillemare, E., Coppola, T., Lazdunski, M., and Hugnot, J. (1994). Cloning provides evidence for a family of inward rectifier and G-protein coupled K 1 channels in the brain. FEBS Lett. 353, 37–42. Lewis, D., and Clapham, D. (1989). Somatostatin activates an inwardly rectifying K 1 channel in neonatal rat atrial cells. Pflu¨gers Arch. 414, 492–494. Loewi, O., and Navratil, E. (1926). Uber humorale ubertragbarkeit der herznervenwirkung. Pflu¨gers Arch. 189, 239–242. Logothetis, D.E., Kurachi, Y., Galper, J., Neer, E.J., and Clapham, D.E. (1987). The beta gamma subunits of GTP-binding proteins activate the muscarinic K 1 channel in heart. Nature 325, 321–326.

Johansson, C., and Thoren, P. (1997). The effects of triiodothyronine (T3) on heart rate, temperature, and ECG measured with telemetry in freely moving mice. Acta. Physiol. Scand. 160, 133–138.

Mansier, P., Medigue, C., Charlotte, N., Vermeiren, C., Coraboeuf, E., Deroubai, E., Ratner, E., Chevlaier, B., Clairambault, J., Carre, F., et al. (1994). Decreased heart rate variability in transgenic mice overexpressing atrial b 1-adrenoreceptors. Am. J. Physiol. 271, H1465–H1472.

Jones, R., Besch, H.J., Fleming, J., McConnaughey, M., and Watanabe, A. (1979). Separation of vesicles of cardiac sarcolemma from vesicles of cardiac sarcoplasmic reticulum. J. Biol. Chem. 254, 530–539.

Milano, C., Allen, L., Rockman, H., Dolber, P., McMinn, T., Chien, K., Johnson, T., Bond, R., and Lefkowitz, R. (1994). Enhanced myocardial function in transgenic mice overexpressing the beta 2-adrenergic receptor. Science 264, 582–586.

Karschin, C., Schreibmayer, W., Dascal, N., Lester, H., Davidson, N., and Karschin, A. (1994). Distribution and localization of a G protein–coupled inwardly rectifying K1 channel in the rat. FEBS Lett. 348, 139–144.

Murer, G., Adelbrecht, C., Lauritzen, I., Lesage, F., Lazdunski, M., Agid, Y., and Raisman-Vozari, R. (1997). An immunocytochemical study on the distribution of two G-protein-gated inward rectifier potassium channels (GIRK2 and GIRK4) in the adult rat brain. Neuroscience 80, 345–357.

Karschin, C., Dissmann, E., Stuhmer, W., and Karschin, A. (1996). IRK(1–3) and GIRK(1–4) inwardly rectifying K1 channel mRNAs are differentially expressed in the adult rat brain. J. Neurosci. 16, 3559– 3570.

Noma, A. (1996). Ionic mechanisms of the cardiac pacemaker potential. Jpn. Heart J. 37, 673–682.

Kim, D. (1991). Calcitonin-gene-related peptide activates the muscarinic-gated K 1 current in atrial cells. Pflu¨gers Arch. 418, 338–345.

Noma, A., and Trautwein, W. (1978). Relaxation of the ACh-induced potassium current in the rabbit sinoatrial node. Pflu¨gers Arch. 377, 193–200.

Kleiger, R.E., Miller, J.P., Bigger, J.J., and Moss, A.J. (1987). Decreased heart rate variability and its association with increased mortality after acute myocardial infarction. Am. J. Cardiol. 59, 256–262.

Noma, A., Morad, M., and Irisawa, H. (1983). Does the ‘pacemaker current’ generate the diastolic depolarization in the rabbit SA node cells? Pflu¨gers Arch. 397, 190–194.

Kobayashi, T., Ikeda, K., Abe, S., Togashi, S., and Kumanishi, T. (1995). Molecular cloning of a mouse G-protein-activated K1 channel and distinct distributions of three GIRK (GIRK1, 2 and 3) mRNAs in mouse. Biochem. Biophys. Res. Comm. 208, 1166–1173.

Rockman, H., Choi, D., Rahman, N., Akhter, S., Lefkowitz, R., and Koch, W. (1996). Receptor-specific in vivo densensitization by the G protein–coupled receptor kinase 5 in transgenic mice. Proc. Natl. Acad. Sci. USA 93, 9954–9959.

Koch, W., Rockman, H., Samama, P., Hamilton, R., Bond, R., Milano, C., and Lefkowitz, R. (1995). Cardiac function in mice overexpressing the b-adrenergic receptor kinase or bARK inhibitor. Science 268, 1350–1353. Kofuji, P., Davidson, N., and Lester, H. (1995). Evidence that neuronal G-protein-gated inwardly rectifying K1 channels are activated by G beta gamma subunits and function as heteromultimers. Proc. Natl. Acad. Sci. USA 92, 6542–6546. Koizumi, K., Ischikawa, T., Nishono, H., and Brooks, C.M. (1975). Cardiac and autonomic system reactions to stretch of the atria. Brain Res. 87, 247–261. Krapivinsky, G., Gordon, E., Wickman, K., Velimirovic, B., Krapivinsky, L., and Clapham, D. (1995a). The G-protein-gated atrial K 1 channel IKACh is a heteromultimer of two inwardly rectifying K1 channel proteins. Nature 374, 135–141. Krapivinsky, G., Krapivinsky, L., Velimirovic, B., Wickman, K., and Clapham, D. (1995b). The cardiac inward rectifier K 1 channel subunit, CIR, does not comprise the ATP-sensitive K 1 channel, IKATP. J. Biol. Chem. 270, 28777–28779.

Rohrer, D., Desai, K., Jasper, J., Stevens, M., Regula, D.J., Barsh, G., Bernstein, D., and Kobilka, B. (1996). Targeted disruption of the mouse b 1-adrenergic receptor gene: developmental and cardiovascular effects. Proc. Natl. Acad. Sci. USA 93, 7357–7380. Sakmann, B., Noma, A., and Trautwein, W. (1983). Acetylcholine activation of single muscarinic K channels in isolated pacemaker cells of the mammalian heart. Nature 303, 250–253. Saul, J., and Cohen, R. (1994). Respiratory sinus arrythmia, M. Levy and P. Schwartz, eds. (Armonk, NY: Futura Publishing). Saul, J., Arai, Y., and Berger, R. (1988). Assessment of autonomic regulation in chronic congestive heart failure by heart rate spectral analysis. Am. J. Cardiol. 61, 1292–1299. Saul, J., Berger, R., and Albrecht, P. (1991). Transfer function analysis of the circulation: unique insights into cardiovascular regulation. Am. J. Physiol. 261, H1231–H1245. Shin, K., Minamitani, H., Onishi, S., Yamazaki, H., and Lee, M. (1995). The power spectral analysis of heart rate variability in athletes during dynamic exercise. Part I. Clin. Cardiol. 18, 583–586.

Neuron 114

Signorini, S., Liao, Y., Duncan, S., Jan, L., and Stoffel, M. (1997). Normal cerebellar development but susceptibility to seizures in mice lacking G protein–coupled, inwardly rectifying K1 channel GIRK2. Proc. Natl. Acad. Sci. USA 94, 923–927. Sowell, M., Ye, C., Ricupero, D., Hansen, S., Quinn, S., Vassilev, P., and Mortenson, R. (1997). Targeted inactivation of a i2 or a i3 disrupts activation of the cardiac muscarinic K 1 channel, IKACh , in intact cells. Proc. Natl. Acad. Sci. USA 94, 7921–7926. Spicer, A., Rowse, G., Lidner, T., and Gendler, S. (1995). Delayed mammary tumor progression in Muc-1 null mice. J. Biol. Chem. 270, 30093–30101. Velimirovic, B., Gordon, E., Lim, N., Navarro, B., and Clapham, D. (1996). The K 1 channel inward rectifier subunits form a channel similar to neuronal G protein–gated K1 channel. FEBS Lett. 379, 31–37. Wickman, K., and Clapham, D. (1995). Ion channel regulation by G proteins. Physiol. Rev. 75, 865–885. Wickman, K., Iniguez-Lluhi, J., Davenport, P., Taussig, R., Krapivinsky, G., Linder, M., Gilman, A., and Clapham, D. (1994). Recombinant Gbg activates the muscarinic-gated atrial potassium channel IKACh. Nature 368, 255–257. Wickman, K., Seldin, M., Gendler, S., and Clapham, D. (1997). Partial structure, chromosome mapping, and expression of the mouse Girk4 gene. Genomics 40, 395–401. Zhou, Q., Quaife, C., and Palmiter, R. (1995). Targeted disruption of the tyrosine hydroxylase gene reveals that catecholamines are required for mouse fetal development. Nature 374, 640–643.