Portal de Periódicos da CAPES

A K+ Channel Splice Variant Common in Human Heart Lacks a C-terminal Domain Required for Expression of Rapidly Activating Delayed Rectifier Current

1998; Elsevier BV; Volume: 273; Issue: 42 Linguagem: Inglês

10.1074/jbc.273.42.27231

1083-351X

Sabina Kupershmidt, Dirk J. Snyders, Adam Raes, Dan M. Roden,

Cardiomyopathy and Myosin Studies

We have cloned HERGUSO, a C-terminal splice variant of the human ether-à-go-go-related gene (HERG), the gene encoding the rapid component of the delayed rectifier (IKr), from human heart, and we find that its mRNA is ∼2-fold more abundant than that for HERG1(the originally described cDNA). After transfection of HERGUSO in Ltk − cells, no current was observed. However, coexpression of HERGUSO with HERG1 modified IKr by decreasing its amplitude, accelerating its activation, and shifting the voltage dependence of activation 8.8 mV negative. As with HERGUSO, HERGΔC (a HERG1 construct lacking the C-terminal 462 amino acids) also produced no current in transfected cells. However, IKr was rescued by ligation of 104 amino acids from the C terminus of HERG1 to the C terminus of HERGΔC, indicating that the C terminus of HERG1 includes a domain (≤104 amino acids) that is critical for faithful recapitulation of IKr. The lack of this C-terminal domain not only explains the finding that HERGUSO does not generate IKr but also indicates a similar mechanism for hitherto-uncharacterized long QT syndrome HERG mutations that disrupt the splice site or the C-terminal. We suggest that the amplitude and gating of cardiac IKrdepends on expression of both HERG1 and HERGUSO. We have cloned HERGUSO, a C-terminal splice variant of the human ether-à-go-go-related gene (HERG), the gene encoding the rapid component of the delayed rectifier (IKr), from human heart, and we find that its mRNA is ∼2-fold more abundant than that for HERG1(the originally described cDNA). After transfection of HERGUSO in Ltk − cells, no current was observed. However, coexpression of HERGUSO with HERG1 modified IKr by decreasing its amplitude, accelerating its activation, and shifting the voltage dependence of activation 8.8 mV negative. As with HERGUSO, HERGΔC (a HERG1 construct lacking the C-terminal 462 amino acids) also produced no current in transfected cells. However, IKr was rescued by ligation of 104 amino acids from the C terminus of HERG1 to the C terminus of HERGΔC, indicating that the C terminus of HERG1 includes a domain (≤104 amino acids) that is critical for faithful recapitulation of IKr. The lack of this C-terminal domain not only explains the finding that HERGUSO does not generate IKr but also indicates a similar mechanism for hitherto-uncharacterized long QT syndrome HERG mutations that disrupt the splice site or the C-terminal. We suggest that the amplitude and gating of cardiac IKrdepends on expression of both HERG1 and HERGUSO. long QT 2 long QT syndrome human and mouse ether-à-go-go-related gene rapidly activating component of delayed rectifier potassium current cyclic nucleotide binding polymerase chain reaction nucleotide(s). The long QT syndrome is a disorder characterized by prolongation of the QT interval as a result of unusually slow repolarization of the cardiac action potential (1Roden D.M. Lazzara R. Rosen M.R. Schwartz P.J. Towbin J.A. Vincent G.M. the SADS Foundation Task Force on LQTS Circulation. 1996; 94: 1996-2012Crossref PubMed Scopus (568) Google Scholar, 2Ackerman M.J. Clapham D.E. N. Engl. J. Med. 1997; 336: 1575-1586Crossref PubMed Scopus (312) Google Scholar). One variant, LQT2,1 is caused by mutations in the human ether-à-go-go-related gene (HERG). Heterologous expression of HERG results in a current with the physiologic and pharmacologic properties of IKr (3Sanguinetti M.C. Jiang C. Curran M.E. Keating M.T. Cell. 1995; 81: 299-307Abstract Full Text PDF PubMed Scopus (2152) Google Scholar, 4Snyders D.J. Chaudary A. Mol. Pharmacol. 1996; 49: 949-955PubMed Google Scholar). These properties include strong inward rectification that is now recognized to be determined by the ultra-rapid inactivation that the channel undergoes with depolarization (4Snyders D.J. Chaudary A. Mol. Pharmacol. 1996; 49: 949-955PubMed Google Scholar, 5Wang S.M. Morales M.J. Liu S.G. Strauss H.C. Rasmusson R.L. FEBS Lett. 1996; 389: 167-173Crossref PubMed Scopus (66) Google Scholar, 6Smith P.L. Baukrowitz T. Yellen G. Nature. 1996; 379: 833-836Crossref PubMed Scopus (668) Google Scholar, 7Spector P.S. Curran M.E. Zou A. Keating M.T. Sanguinetti M.C. J. Gen. Physiol. 1996; 107: 611-619Crossref PubMed Scopus (377) Google Scholar, 8Yang T. Snyders D.J. Roden D.M. Circ. Res. 1997; 80: 782-789Crossref PubMed Scopus (152) Google Scholar) and by sensitivity to IKr-specific methanesulfonanilide blockers, such as dofetilide (4Snyders D.J. Chaudary A. Mol. Pharmacol. 1996; 49: 949-955PubMed Google Scholar, 9Spector P.S. Curran M.E. Keating M.T. Sanguinetti M.C. Circ. Res. 1996; 78: 499-503Crossref PubMed Scopus (274) Google Scholar). Importantly, HERG mRNA is detected at high abundance in human heart (10Curran M.E. Splawski I. Timothy K.W. Vincent G.M. Green E.D. Keating M.T. Cell. 1995; 80: 795-803Abstract Full Text PDF PubMed Scopus (1996) Google Scholar), and IKr has been recorded in human atrial and ventricular myocytes (11Wang Z. Fermini B. Nattel S. Cardiovasc. Res. 1994; 28: 1540-1546Crossref PubMed Scopus (217) Google Scholar, 12Li G.R. Feng J. Yue L. Carrier M. Nattel S. Circ. Res. 1996; 78: 689-696Crossref PubMed Scopus (423) Google Scholar). LQT2-associated HERG mutations in the pore of the channel display reduced or no IKr (13Sanguinetti M.C. Curran M.E. Spector P.S. Keating M.T. Proc. Natl. Acad. Sci. U. S. A. 1996; 93: 2208-2212Crossref PubMed Scopus (392) Google Scholar) and, when expressed with wild-type cRNA, a dominant negative effect has been observed, implying that heteromeric (wild-type + mutant) channel assembly occurs. Heteromeric assembly of physiologically occurring splice variants has also been suggested by the finding that an N-terminal splice variant of the human gene (or of the murine homolog mERG) itself resulted in no current in Xenopus oocytes, but when coexpressed with wild-type cRNA, IKr with rapid activation and deactivation (typical of that observed in heart) was observed (14Lees-Miller J.P. Kondo C. Wang L. Duff H.J. Circ. Res. 1997; 81: 719-728Crossref PubMed Scopus (177) Google Scholar,15London B. Trudeau M.C. Newton K.P. Beyer A.K. Copeland N.G. Gilbert D.J. Jenkins N.A. Satler C.A. Robertson G.A. Circ. Res. 1997; 81: 870-878Crossref PubMed Scopus (251) Google Scholar). The initial identification of HERG mutations as a cause of LQT2 included a mutation in the 3′-region of the gene, which is predicted to result in disruption of a splice site and truncation of the C-terminal region of the protein (10Curran M.E. Splawski I. Timothy K.W. Vincent G.M. Green E.D. Keating M.T. Cell. 1995; 80: 795-803Abstract Full Text PDF PubMed Scopus (1996) Google Scholar). However, it is not clear how this mutation disrupts IKr function because the membrane-spanning segments, as well as the N-terminal region, which has been reported to be necessary for assembly of multiple K+channel subunits into functional channels (16Babila T. Moscucci A. Wang H. Weaver F.E. Koren G. Neuron. 1994; 12: 615-626Abstract Full Text PDF PubMed Scopus (98) Google Scholar, 17Li X. Xu J. Li M. J. Biol. Chem. 1997; 272: 705-708Abstract Full Text Full Text PDF PubMed Scopus (86) Google Scholar), are present. One possible mechanism is disruption of a putative cyclic nucleotide binding (CNB) domain within which the splice site is located. This domain was identified with the initial cloning of HERG (18Warmke J.W. Ganetzky B. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 3438-3442Crossref PubMed Scopus (866) Google Scholar), but its functional role has not yet been elucidated, and cardiac IKr has been reported to be unaffected by protein kinase A-dependent intracellular signaling (19Sanguinetti M.C. Jurkiewicz N.K. Scott A. Siegl P.K.S. Circ. Res. 1991; 68: 77-84Crossref PubMed Scopus (283) Google Scholar). Because this domain is disrupted in the C-terminal LQT2 mutation, it is conceivable that it plays a role in recapitulation of IKr. However, a recent report of the genomic structure of HERG and analysis of additional LQT2 families raises the possibility that a more C-terminal region is actually required. This report included seven new mutations predicted to result in C-terminal truncations of HERG (20Vesely M.R. Duggal P. London B. Wattanasirichaigoon D. Beggs A.H. Circulation. 1997; 96: I56Crossref PubMed Scopus (154) Google Scholar), with a number 3′ to the CNB domain; the one most 3′ was predicted to result in truncation of the 3′ 158 amino acids of the HERG protein. Thus, it remains unknown whether the CNB domain is required for recapitulation of IKr. Recent studies with other potassium channels have raised the possibility that C-terminal regions are required for recapitulation of current, through mechanisms that have yet to be firmly established (21Ludwig J. Owen D. Pongs O. EMBO J. 1997; 16: 6337-6345Crossref PubMed Scopus (58) Google Scholar, 22Daram P. Urbach S. Gaymard F. Sentenac H. Cherel I. EMBO J. 1997; 16: 3455-3463Crossref PubMed Scopus (116) Google Scholar, 23Biervert C. Schroeder B.C. Kubisch C. Berkovic S.F. Propping P. Jentsch T.J. Steinlein O.K. Science. 1998; 279: 403-406Crossref PubMed Scopus (935) Google Scholar). In this report, we describe the isolation of a HERGC-terminal splice variant, which we call HERGUSO, and its localization and functional characterization. These studies demonstrated that IKr can be recapitulated even when the putative CNB domain is disrupted and have led to the identification of a 104-amino acid domain, present in the C terminus of HERG but absent in HERGUSO, which appears essential for the recapitulation of IKr in a heterologous expression system. This finding not only points to another potential mechanism for regulation of IKr in vivo but also explains how C-terminal mutations in HERG can cause LQTS. 1 × 106 plaques of a human ventricular cDNA library (a gift of Drs. Michael Tamkun and Sarah England) in λZap II (Stratagene, La Jolla, CA) were screened using standard techniques (24Sambrook J. Fritsch E.F. Maniatis T. Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Press, Cold Spring Harbor, NY1989Google Scholar) under high stringency conditions. The probe consisted of a PCR fragment that covers nucleotides 1766–3469 of the originally reported HERG, which here we termHERG 1. In this report, nucleotide numbering starts at the translational start site of HERG. RNA was isolated for quantitation from samples of normal human left ventricle using the acid-phenol method of RNA extraction of Chomczynski and Sacchi (25Chomczynski P. Sacchi N. Anal. Biochem. 1987; 162: 156-159Crossref PubMed Scopus (63189) Google Scholar). In order to determine whether HERGUSO mRNA was expressed in the human heart, we designed two probes, both overlapping the splice site at nt 2398, which we used in RNase protection analyses. A 104-nt probe for cyclophilin was used as an internal standard to normalize for equal RNA concentration across the lanes. RNase protection was performed as described previously (26Yang T. Kupershmidt S. Roden D.M. Circ. Res. 1995; 77: 1246-1253Crossref PubMed Scopus (73) Google Scholar). The gels were dried and exposed to x-ray film as well as to a PhosphorImager (Molecular Dynamics) to quantitate relative band intensities. Clone pHERG1 was constructed by excising the HindIII/EcoRI fragment of HERG1 from the pSP64 vector (a gift from Mark Keating, University of Utah) and ligating it via its EcoRI site at the 3′-end into the EcoRI-digested mammalian expression vector pSI (Promega, Madison, WI) followed by treatment with Klenow and a second blunt-end ligation step. Clone pHERGUSO was constructed by first ligating the 3′-terminal XhoI/EcoRI fragment of phage isolate Φ-D (Fig. 1) into the XhoI/EcoRI-digested pSI vector to generate clone pSKF104. The EcoRI site corresponds to the 3′-terminus of the isolate Φ-D, whereas the XhoI site corresponds in location to the XhoI site at nt 2089 of HERG1. The resulting clone SKF104 was then digested at its 5′-end with XhoI and reconstituted at the N terminus using the HindIII/XhoI fragment of HERG1, which consists of the 2089 5′-most nt of the HERG1 coding sequence. The initial Xho ligation was followed by treatment with Klenow and a second blunt-end ligation step to circularize the plasmid generating pHERGUSO. Clone pHERGΔC was constructed by excising theHindIII/XhoI fragment of HERG1 from the pSP64 vector and ligating its 3′-end into XhoI-digested pSI followed by creation of blunt ends at the 5′-end using Klenow and a second, blunt-end ligation step. To control for possible mutations during growth in bacterial cells or other cloning artifacts, plasmid pSKH40 was created by excising the XhoI/EcoRI fragment of pHERGUSO and replacing it with theXhoI/EcoRI fragment of HERG1 in pSP64. This replaces the HERGUSO 3′-terminus with that of HERG1, in effect reconstituting the original pHERG1 clone. Clone pHERGΔC+104 was created by generating a PCR fragment corresponding to nt 3052–3366 of HERG with XhoI linkers at the termini and a stop codon (TAG) incorporated into the reverse primer. The forward primer was 5′-GGACTCGAG C CCACCCCCAG CCTCCTCAAC ATCCC-3′. The reverse primer was 5′-CTACTCGAG C TACGGGGGCA GCTCCTCACA CGCCATG-3′. XhoI recognition sites are italicized and boldface, and the stop codon is double underlined. The PCR fragment was digested with XhoI followed by gel purification and ligation into XhoI-digested pHERGΔC. Resulting clones were sequenced to identify those containing the PCR fragment in the correct orientation. The reading frame of the PCR fragment is conserved. All plasmids described were partially sequenced across vector/insert junctions and across junctions generated by ligation steps and were further characterized by restriction pattern analysis. Ltk −cells were cultured in DMEM supplemented with 10% horse serum. The cells were transfected using the LipofectAMINE method according to the manufacturer's instructions (Life Technologies, Inc.). Green fluorescent protein was coexpressed to assess the transfection efficiency and to identify expressing cells for voltage-clamp analysis (27Chalfie M. Tu Y. Euskirchen G. Ward W.W. Prasher D.C. Science. 1994; 263: 802-805Crossref PubMed Scopus (5508) Google Scholar). The transfection mixture included 1–10 μg of HERG isoforms or chimeras (in pSI), 1 μg of green fluorescent protein/pRcCMV, and 12–30 μl of LipofectAMINE reagent in 0.5 ml of serum-free DMEM for 6 h, after which the standard medium was restored. The cells were removed from the dish 24 h later by a brief trypsinization, washed twice in standard medium, and stored for use within the next 12 h. Parallel nontransfected cultures or cells transfected with green fluorescent protein alone (mock transfection) served as controls. Whole cell voltage clamp (28Hamill O.P. Marty A. Neher E. Sakmann S. Sigworth F.J. Pfluegers Arch. 1981; 391: 85-100Crossref PubMed Scopus (15148) Google Scholar) was used to record currents in transfected cells displaying green fluorescence, as described previously (4Snyders D.J. Chaudary A. Mol. Pharmacol. 1996; 49: 949-955PubMed Google Scholar). Experiments were performed at room temperature (21–23 °C) and sampled at 1–10 kHz after antialias filtering at half or less of the sampling frequency. Specific protocols are specified in the text and figure legends. The cycle time between pulses was ≥20 s to accommodate pulse durations and slow IKr activation (especially below +10 mV). HERGactivation kinetics were slow below −20 mV (see under "Results"), and only passive linear leak (>1 GΩ) was observed during short (100 ms) depolarizations between −90 and −40 mV; therefore, least squares fits to these data were used for passive leak correction. Mono- and multiexponential functions were fit to activating current using least squares methods. The intracellular pipette filling solution contained 110 mmKCl, 10 mm HEPES, 5 mm K4BAPTA, 5 mm K2ATP, and 1 mmMgCl2, adjusted to pH 7.2 with KOH (final intracellular [K+], ∼145 mm). The bath solution contained 145 mm NaCl, 4 mm KCl, 1.8 mmCaCl2, 1 mm MgCl2, 10 mm HEPES, and 10 mm glucose, adjusted to pH 7.35 with NaOH. Dofetilide was provided by Pfizer Central Research. All other compounds were obtained from Sigma. Three clones, designated Φ-D, Φ-L, and Φ-M, were isolated by screening the human heart library. Isolate Φ-L had a partial cDNA insert that corresponded to nucleotides 3064–3886 of HERG1, i.e. the 3′ -terminal 138 amino acids plus the 3′-untranslated region of HERG1 (18Warmke J.W. Ganetzky B. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 3438-3442Crossref PubMed Scopus (866) Google Scholar) (Fig. 1). Φ-M and Φ-D also contained partial cDNAs. The 5′-end of Φ-M was located at HERG1 nt 1267, and that of Φ-D was located at HERG1 nt 813. Both clones were identical to HERG1 up to HERG1 nt 2398, the previously reported splice site (10Curran M.E. Splawski I. Timothy K.W. Vincent G.M. Green E.D. Keating M.T. Cell. 1995; 80: 795-803Abstract Full Text PDF PubMed Scopus (1996) Google Scholar). Following the splice junction in both isolates were 750 nt of a novel sequence that we designated USO. Because clone Φ-M appears to be a subset of clone Φ-D, we restricted our subsequent analysis to Φ-D. The reading frame of HERG1 as it continues into the USO stays open for another 264 nt (88 amino acids). USO appears to be partially made up of sequences previously thought to be intronic because a sequence corresponding to a previously reported intronic primer (10Curran M.E. Splawski I. Timothy K.W. Vincent G.M. Green E.D. Keating M.T. Cell. 1995; 80: 795-803Abstract Full Text PDF PubMed Scopus (1996) Google Scholar), located between exons 9 and 10 in the mouse gene (15London B. Trudeau M.C. Newton K.P. Beyer A.K. Copeland N.G. Gilbert D.J. Jenkins N.A. Satler C.A. Robertson G.A. Circ. Res. 1997; 81: 870-878Crossref PubMed Scopus (251) Google Scholar), is located 30 nt 3′ to the splice junction at position 2398. Probe E75–1 is predicted to protect a fragment of 228 nt if the HERG1-type splice junction is represented within the mRNA and a fragment of 154 nt if a splice event other than HERG1 (e.g.HERGUSO) occurs (Fig. 2). Conversely, probe E114 is predicted to protect a 191-nt fragment if HERGUSO mRNA is present and a smaller (154 nt) fragment if another splice variant is present. Fig. 2 shows that HERGUSO is present at ∼2-fold higher levels in human heart than HERG1 (range, 1.7–2.1-fold, n = 4). Similar findings were obtained using human left ventricle, atrium, midmyocardium, uterus, cultured human lymphocytes, and, on longer exposure, brain. Further evidence that HERGUSO mRNA is expressed in other tissues is provided by the results of a Basic Local Alignment Search Tool (BLAST) (29Altschul S.F. Gish W. Miller W. Myers E.W. Lipman D.J. J. Mol. Biol. 1990; 215: 403-410Crossref PubMed Scopus (70731) Google Scholar) search of GenBankTMusing the USO sequence. This identified two expressed sequence tags from human fetal liver/spleen (R07336 and R07383) matching HERGUSO. R07383 is 395 nt and matches the HERGUSO open reading frame starting 42 nt 3′ to the 2398 splice junction and extending beyond the HERGUSO stop codon, whereas R07336 is 311 nt and matches HERGUSO in its 3′-untranslated region. When pHERG1 was transfected (Fig. 3, left panels), typical IKr was observed. A brief hyperpolarizing pulse to remove inactivation (4Snyders D.J. Chaudary A. Mol. Pharmacol. 1996; 49: 949-955PubMed Google Scholar, 5Wang S.M. Morales M.J. Liu S.G. Strauss H.C. Rasmusson R.L. FEBS Lett. 1996; 389: 167-173Crossref PubMed Scopus (66) Google Scholar, 6Smith P.L. Baukrowitz T. Yellen G. Nature. 1996; 379: 833-836Crossref PubMed Scopus (668) Google Scholar, 7Spector P.S. Curran M.E. Zou A. Keating M.T. Sanguinetti M.C. J. Gen. Physiol. 1996; 107: 611-619Crossref PubMed Scopus (377) Google Scholar, 8Yang T. Snyders D.J. Roden D.M. Circ. Res. 1997; 80: 782-789Crossref PubMed Scopus (152) Google Scholar) resulted in a characteristic augmentation of current, which then rapidly reinactivated, at positive potentials (Fig. 3 C, left). By contrast, no current was observed, even after a hyperpolarizing pulse, following transfection with pHERGUSO alone. When pHERG1 and pHERGUSO were cotransfected in equimolar concentrations, IKr with amplitude and gating properties very similar to those observed with transfection of pHERG1 alone was observed. As shown in Fig. 3 (right panels), when a 10-fold excess of pHERGUSO was transfected with pHERG1, the current remained IKr-like. However, it is apparent from Fig. 3 (right) that the current in the latter cotransfection experiments was smaller than that observed with pHERG1alone and displayed faster and more negative voltage dependence of activation. These findings are summarized in Fig. 4. The midpoint of activation was shifted 8.8 mV negative (Fig. 4 A). After depolarizing clamp steps from −50 to −40 to +60 mV (Fig. 3, A and B), activating currents were 3–4-fold larger with pHERG1 alone (Fig. 4 B). When inactivation was removed by a brief hyperpolarization (Fig. 3 C), current with pHERG1 alone was 513 ± 137 pA/pF (n = 6) versus 203 ± 40 pA/pF (n = 7, p = 0.03) with cotransfection. Fits to activating current required biexponential functions in the −10 to +30 mV range, and a monoexponential function was sufficient in the +40 to +60 mV range. In either case, it is apparent from Fig. 4 C that activation was consistently ∼3–5-fold faster with cotransfection over all voltages tested. By contrast, the time constants for deactivation were similar in the two sets of experiments. The fact that no current was observed when pHERGUSO was expressed alone was somewhat unexpected, because the protein domains of the N terminus and the transmembrane-spanning region, including the voltage sensor and pore region, are all intact. To guard against the possibility of sequence rearrangements during the passage of plasmid constructs in bacterial hosts resulting in changes that would account for the lack of current observed with expression of pHERGUSO and pHERGΔC, we conducted experiments to rescue HERG1 from pHERGUSO. A full-length HERG1 plasmid, pSKH40, was generated from pHERGUSO by replacing the HERGUSO C terminus with the C terminus from HERG1. Expression of pSKH40 resulted in IKr. We next considered the possibility that the protein domain encoded by the USO might exert an inhibitory effect on channel expression or function. To test this idea, we generated a C-terminal truncation mutant of HERG1 that lacks the C-terminal 461 amino acids, including both the CNB domain and the USO. This construct, pHERGΔC (Fig. 1), terminates 32 amino acids past the S6 domain at a unique XhoI site (nt 2089) of HERG1. However, with transfection of pHERGΔC, no currents were recorded. We therefore concluded that a protein domain encoded by the region 3′ of the splice junction at nt 2398 (i.e. within residues 800–1159) of HERG1 is required for the functional expression of IKr. The CNB domain homolog present in HERG1 is truncated in HERGUSO and completely missing in HERGΔC. Our next step was therefore to determine whether pHERGΔC+104, a construct totally lacking this domain but preserving the N terminus (Fig. 1), could rescue expression of IKr. AfterLtk − cells were transfected with pHERGΔC+104, a current with most features of IKr was recorded (Fig. 5). Although the kinetics of activation and deactivation as well as the V1/2 appeared to be somewhat different from those of HERG1 and the tails were somewhat smaller, the expressed current still activated and deactivated rapidly (Fig. 5 A) and displayed prominent inward rectification (Fig. 5 B) and a large very rapidly inactivating current at positive potentials after a hyperpolarizing step (Fig. 5 C). The current was suppressed by 1 μm dofetilide. We have identified here a 3′-splice variant of HERG that is more abundantly expressed in human heart and other tissues than the originally reported mRNA. However, expression of the variant in mammalian cells did not give rise to potassium current, although coexpression with HERG1 showed a striking effect on the amplitude and activation of IKr, suggesting that HERGUSO can coassemble with HERG1 to modify its gating. The mechanism underlying the failure of HERGUSOalone to generate current is indicated by the experiments with HERGΔC and HERGΔC+104, which demonstrate that a domain near the 3′-end of the mRNA, absent in HERGUSO, is required for recapitulation of IKr. When HERG1 and HERGUSO were coexpressed in equimolar concentrations, there was no obvious effect on HERG1-mediated current (IKr). The finding that mRNA transcripts encoding HERGUSO were much more abundant than those for HERG1 suggested testing the effect of excess HERGUSO as a next step in these experiments. This approach resulted in currents that demonstrated gating properties quite different from those seen with expression of HERG1 alone. This finding strongly suggests that the two isoforms coassemble to form heterotetrameric channels, although direct biochemical evidence of such coassembly in heterologous systems, and ultimately in native tissues, will be required to further test this postulate. We reconstituted our partial HERGUSO cDNA with the N terminus from HERG1 because Spector et al. (7Spector P.S. Curran M.E. Zou A. Keating M.T. Sanguinetti M.C. J. Gen. Physiol. 1996; 107: 611-619Crossref PubMed Scopus (377) Google Scholar) have previously shown that a deletion of amino acids 2–354 of HERG1 did not affect activation or fast inactivation of the resulting IKr, although the voltage dependence of activation and inactivation were shifted by +20 to +30 mV, and the rate of deactivation was much faster with the truncation mutant than with the wild-type channel (7Spector P.S. Curran M.E. Zou A. Keating M.T. Sanguinetti M.C. J. Gen. Physiol. 1996; 107: 611-619Crossref PubMed Scopus (377) Google Scholar). Similar results with a N-terminal deletion mutant, from amino acids 2–373, have also been reported by Schonherr and Heinemann (30Schonherr R. Heinemann S.H. J. Physiol. 1996; 493: 635-642Crossref PubMed Scopus (263) Google Scholar). We consider that experiments using heterologous expression approaches that demonstrate only small changes in ion channel function cannot be interpreted with any confidence. Therefore, we used a 1:10 HERG1:HERGUSO ratio when experiments with the 1:1 ratio did not produce an obvious change current amplitude or gating. If we assume that the two RNAs are translated with equal efficiency and have equal opportunity to assemble with other subunits, the incidence of homotetrameric HERG1 complexes formed by the 1:10 ratio of RNAs will be vanishingly small (114,641, or <0.01%). Thus, the result in Fig. 3 B indicates that heterotetrameric HERG1/HERGUSO channels can recapitulate IKr (or that the assumptions of equivalent translation and synthesis are incorrect). Given the assumptions inherent in interpreting this sort of experiment, we believe that the most appropriate interpretation is that expression HERGUSOmodifies HERG1 amplitude and gating in the human heart but that the extent of this interaction in situ remains uncertain. HERGΔC+104 resulted in typical IKr in the absence of the CNB domain, thereby establishing that the CNB domain is not required for IKr expression. Furthermore, C-terminal variants lacking the 104-amino acid domain at positions 1018–1122 (HERGΔC and HERGUSO) expressed no IKr, whereas the HERGΔC+104 rescue construct, which included this domain at the 3′-end, did. The requirement for this domain provides an explanation for the presumed defect in repolarizing current in LQT2 mutations reported in the 3′-end of the gene, at the nt 2398 splice site and beyond (10Curran M.E. Splawski I. Timothy K.W. Vincent G.M. Green E.D. Keating M.T. Cell. 1995; 80: 795-803Abstract Full Text PDF PubMed Scopus (1996) Google Scholar, 20Vesely M.R. Duggal P. London B. Wattanasirichaigoon D. Beggs A.H. Circulation. 1997; 96: I56Crossref PubMed Scopus (154) Google Scholar). The function of the 104-amino acid domain that we have implicated as essential for recapitulation of IKr is not yet understood. One possibility is that it is required to allow for the assembly for HERG protein tetramers. One previous report (17Li X. Xu J. Li M. J. Biol. Chem. 1997; 272: 705-708Abstract Full Text Full Text PDF PubMed Scopus (86) Google Scholar) has implicated a protein domain from the N terminus of HERG in the assembly of HERG homotetramers. However, N-terminal truncations of HERG do result in current (7Spector P.S. Curran M.E. Zou A. Keating M.T. Sanguinetti M.C. J. Gen. Physiol. 1996; 107: 611-619Crossref PubMed Scopus (377) Google Scholar), indicating that the N-terminal domain is not required, at least not on all four subunits of the presumed tetrameric channel. Recent studies with rat eag (21Ludwig J. Owen D. Pongs O. EMBO J. 1997; 16: 6337-6345Crossref PubMed Scopus (58) Google Scholar), with the plant potassium channel AKT1 (22Daram P. Urbach S. Gaymard F. Sentenac H. Cherel I. EMBO J. 1997; 16: 3455-3463Crossref PubMed Scopus (116) Google Scholar), and with KCNQ2 (23Biervert C. Schroeder B.C. Kubisch C. Berkovic S.F. Propping P. Jentsch T.J. Steinlein O.K. Science. 1998; 279: 403-406Crossref PubMed Scopus (935) Google Scholar) have implicated regions of the respective C termini in subunit-subunit interactions that are crucial to generation of potassium current. The same may thus hold true for the C-terminal 104-amino acid domain that we have identified in HERG1, although the specific function (e.g.assembly or trafficking) is uncertain. We cannot rule out the possibility that this domain interacts with other proteins; indeed, we and others have suggested that the minK protein interacts with HERG to increase IKr (26Yang T. Kupershmidt S. Roden D.M. Circ. Res. 1995; 77: 1246-1253Crossref PubMed Scopus (73) Google Scholar, 31McDonald T.V., Yu, Z. Ming Z. Palma E. Meyers M.B. Wang K.W. Goldstein S.A. Fishman G.I. Nature. 1997; 388: 289-292Crossref PubMed Scopus (308) Google Scholar). Finally, the effect of HERGUSO may be exerted not at the protein level but at the RNA level; production of the alternatively spliced HERGUSOisoform could regulate the amount of and/or the processing and translation of HERG1 mRNA. This could be accomplished by competition for necessary transcription factors that are present in limiting amounts or competition for components of the splicing machinery. Such competition could then serve to limit the amount of functional HERG1 protein produced and might assume importance in acquired diseases associated with perturbations of cardiac repolarization. In this regard, it is of interest that splice variants of mRNA encoding glutamate transporters, including an intron-retention event, have recently been reported in amyotrophic lateral sclerosis (32Lin C.L.G. Bristol L.A. Jin L. Dykes-Hoberg M. Crawford T. Clawson L. Rothstein J.D. Neuron. 1998; 20: 589-602Abstract Full Text Full Text PDF PubMed Scopus (589) Google Scholar). Along these lines, in preliminary studies, we have found that transcripts encoding HERG1 mRNA are unchanged in hearts from patients with advanced cardiac failure but that those encoding HERGUSO are decreased by ∼50% (33Choy A.M.J. Kupershmidt S. Lang C.C. Pierson Jr., R.N. Roden D.M. Circulation. 1996; 94: I164Google Scholar). The expert advice provided by Al George during development of this project is greatly appreciated. We also appreciate the superb technical assistance of Jason Eck and Nancy Sugg.