Expression of distinct ERG proteins in rat, mouse, and human heart. Relation to functional I(Kr) channels.

One form of inherited long QT syndrome, LQT2, results from mutations in HERG1, the human ether-a-go-go-related gene, which encodes a voltage-gated K(+) channel alpha subunit. Heterologous expression of HERG1 gives rise to K(+) currents that are similar (but not identical) to the rapid component of delayed rectification, I(Kr), in cardiac myocytes. In addition, N-terminal splice variants of HERG1 and MERG1 (mouse ERG1) referred to as HERG1b and MERG1b have been cloned and suggested to play roles in the generation of functional I(Kr) channels. In the experiments here, antibodies generated against HERG1 were used to examine ERG1 protein expression in heart and in brain. In Western blots of extracts of QT-6 cells expressing HERG1, MERG1, or RERG1 (rat ERG1) probed with antibodies targeted against the C terminus of HERG1, a single 155-kDa protein is identified, whereas a 95-kDa band is evident in blots of extracts from cells expressing MERG1b or HERG1b. In immunoblots of fractionated rat (and mouse) brain and heart membrane proteins, however, two prominent high molecular mass proteins of 165 and 205 kDa were detected. Following treatment with glycopeptidase F, the 165- and 205-kDa proteins were replaced by two new bands at 175 and 130 kDa, suggesting that ERG1 is differentially glycosylated in rat/mouse brain and heart. In human heart, a single HERG1 protein with an apparent molecular mass of 145 kDa is evident. In rats, ERG1 protein (and I(Kr)) expression is higher in atria than ventricles, whereas in humans, HERG1 expression is higher in ventricular, than atrial, tissue. Taken together, these results suggest that the N-terminal alternatively spliced variants of ERG1 (i.e. ERG1b) are not expressed at the protein level in rat, mouse, or human heart and that these variants do not, therefore, play roles in the generation of functional cardiac I(Kr) channels.

Long QT syndrome is an acquired or an inherited disorder that can cause syncope and sudden death resulting from epi-sodic ventricular arrhythmias and ventricular fibrillation (1,2). The characteristic feature identified in surface electrocardiograms of affected individuals is prolongation of the QT interval, consistent with the underlying cause of long QT syndrome being a defect in ventricular repolarization (3,4). One form of inherited long QT syndrome, LQT2, was localized to chromosome 7 (5), and shown to result from mutations (6) in the human ether-a-go-go-related gene, originally referred to as HERG (7). With the identification of additional members of the ether-a-go-go-related gene (ERG) family (8), the terminology HERG1 seems more appropriate (9).
HERG1 encodes a polypeptide with a predicted molecular mass of 127 kDa and a predicted sequence and membrane topology similar to that of other voltage-gated K ϩ channel ␣ subunits (10 -12). Heterologous expression of HERG1 in Xenopus oocytes (10 -12) and in HEK-293 cells (14 -16) reveals voltage-gated K ϩ currents that are similar to the rapid component of delayed rectification, I Kr , in myocardial cells (17)(18)(19)(20)(21)(22). Like I Kr , for example, the currents produced on heterologous expression of HERG are rapidly activating and rapidly inactivating, are K ϩ -selective, and display marked inward rectification at potentials positive to 0 mV (10 -14). The detailed properties of the HERG-induced currents (10 -16), however, are variable and are not identical to those of (endogenous) I Kr characterized in myocardial cells from several species (17)(18)(19)(20)(21)(22), including human (20 -22). Although additional members of the ERG family, ERG2 and ERG3, have been cloned (8), these subunits appear to be nervous system-specific and therefore likely do not contribute to cardiac I Kr . Alternatively spliced variants of HERG1 and MERG1 (the mouse homologue of HERG 1) with unique N termini (referred to as HERG1b and MERB1b), however, have been cloned from heart and postulated to play roles in the generation of functional cardiac I Kr channels (9,23,24). Indeed, coexpression of MERG1b (or HERG1b) with full-length MERG1 (or HERG1) in Xenopus oocytes reveals voltage-gated K ϩ currents that more closely resemble cardiac I Kr than the currents produced on expression of MERG1 (or HERG1) alone (9,23). C-terminal splice variants of HERG1 have also been identified in heart, although the functional significance of these subunits (which do not produce functional channels) is unclear (24,25). HERG1 expressed in HEK-293 cells is differentially glycosylated (14), and it has been demonstrated that some LQT2 mutations result in changes in N-linked glycosylation and that these underlie changes in subunit protein processing and cell surface expression (15). Nevertheless, the functional significance of N-linked glycosylation in the regulation of cell surface expression of HERG1 in cardiac cells and/or in the generation of functional cardiac I Kr channels is unclear. In fact, although ERG1 protein has been documented in ferret heart (26), neither the expression levels nor the distributions of alternatively spliced and/or differentially processed ERG1 proteins have been examined directly to date in the mammalian myocardium.
The experiments here were undertaken to examine ERG1 protein expression in mouse, rat, and human heart. Using antibodies targeted against HERG1, two ERG1 proteins with molecular masses 165 and 205 kDa are detected in rat and mouse atria, ventricles, and brain, and these appear to reflect differentially (N-linked) glycosylated forms of ERG1. In contrast, a single 145-kDa HERG1 protein is identified in immunoblots of human cardiac membrane proteins. In rat, mouse, and human heart, however, no low molecular mass proteins corresponding to the expression of the ERG1b are detected. In rat, ERG1 protein expression and functional I Kr expression are higher in atria than ventricles, whereas in mouse and human, ERG1 expression is higher in the ventricles. The results presented demonstrate that the predominant form of ERG1 expressed in myocardial tissues is the full-length ERG1 protein, suggesting that N-terminal ERG1b splice variants do not play a role in the generation of functional I Kr channels in cardiac myocytes.

MATERIALS AND METHODS
Polyclonal Antibodies against HERG-Peptides corresponding to unique sequences in HERG: (i) residues 174 -188, TARESSVRSG-GAGGA, in the N terminus and (ii) residues 1145-1159, LTSQPL-HRHGSDPGS, in the C terminus, were generated by the Protein Chemistry Laboratory (Washington University Medical Center). A cysteine residue was added to the N terminus of each peptide to allow coupling to the keyhole limpet hemocyanin carrier protein, and the coupled peptides were sent to Caltag (San Francisco, CA) for injection into rabbits. Sera were screened using enzyme-linked immunosorbent assays, and antibodies were subsequently affinity purified using the Im-munoPure Antigen/Antibody Immobilization Kit #2 (Pierce). These antibodies are referred to as N-anti-HERG and C-anti-HERG to denote the location of the amino acid sequence in HERG against which the antibody was generated. Enzyme-linked immunosorbent assays on the affinity purified antibodies revealed that each antibody detected only the peptide against which it was generated; no cross-reactivity was evident.
In addition, a polyclonal antiserum generated against a fusion protein corresponding to the C-terminal 181 amino acids in HERG1 coupled to (histidine-tagged) thioredoxin was obtained from Z. Zhou and C. T. January (University of Wisconsin, Madison, WI). Details of the preparation, purification, and characterization of the antiserum have been published (15). An additional affinity purified anti-HERG antipeptide antibody generated against residues 1118 -1133 in the C terminus of HERG1 (with a tyrosine and a cysteine residue added at the N terminus of the synthetic peptide) was purchased from Alomone Labs (Jerusalem, Israel).
cDNA Constructs-For cloning of the rat homologue of HERG1, RERG1 (27), four PCR 1 primers were designed to amplify the coding region of RERG1 (GenBank TM accession number RNZ96106) from a rat brain cDNA library (CLONTECH). Two primers, RERG1 forward (5Ј-GGAATTCATGCCGGTGCGGAGGGGCCACGTCGCGCCGCAGAACA-3Ј) and RERG1 reverse (5Ј-GGAATTCCTAACTGCCTGGATCTGAGC-CATGTCTGTGCAG-3Ј) flank the entire coding region, whereas RERG1 17 (5Ј-GATGACCAGCAGCAGAATGA-3Ј) and RERG1 3 (5Ј-TCGAGCT-CAGAGC-CTTAACC-3Ј) are located within the coding sequence. PCR amplification of the rat brain library using RERG1 forward and RERG1 17 yielded a 1.2-kilobase product, corresponding to the 5Ј region of the clone. Amplification with RERG1 3 and RERG1 reverse yielded a 2.7kilobase product corresponding to the 3Ј end of the clone and included an overlap region of Ϸ500 base pairs with the first PCR product. The two PCR products were eluted from an agarose gel and joined in all PCR reactions in which the RERG1 forward and reverse primers were combined with the eluted bands. The resulting 3.5-kilobase product was ligated into the pEF6/V5-His TOPO vector (Invitrogen) and amplified by transformation into TOP10 (Invitrogen) competent cells. The orientation and the fidelity of the resulting RERG1 plasmid DNA was confirmed by sequencing both the 5Ј and 3Ј ends of the clone. HERG1 in pBK/CMV was provided by Dr. Gail Robertson (University of Wisconsin). The MERG1 and MERG1b in pBK/CMV cDNAs were obtained from Dr. Barry (29), a quail fibroblast cell line, were cultured in medium 199 (Earle's Salts) supplemented with 10% tryptose phosphate broth, 5% fetal calf serum, 1% dimethyl sulfoxide, 0.15% mycostatin and antibiotics (penicillin and gentamicin) at 37°C in 95% air/5% CO 2 . All media and constituents were obtained from Life Technologies, Inc. QT-6 cells plated on glass coverslips in 12-well (18 mm each) culture plates or in 100-mm culture dishes were transiently transfected using the calcium phosphate precipitation method (30) with cDNA constructs encoding HERG1, HERG1b, MERG1, MERG1b, RERG1, Kv1.2, Kv1.4, Kv2.1, or Kv4.2 (1.2 g of cDNA/18-mm well or 20 g of cDNA/100-mm dish). In each case, a reporter gene construct was also used: either lacZ or GFP in pBK-CMV (0.5 g/18-mm well or 8 g/100-mm dish). For mock transfections, the same amount of the GFP-pBK-CMV or lacZ-pBK-CMV construct was used, and carrier DNA (pSK ϩ ) (1.2 g of cDNA/18-mm well or 20 g of cDNA/100-mm dish) was added in place of the ERG1 and the Kv ␣ subunit cDNAs. Transfection efficiencies (generally 25-35%) were determined either by counting GFP-positive cells, which were visualized directly using an inverted epifluorescence microscope, or by subsequent staining for ␤-galactosidase activity using the chromogenic substrate 5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside (X-gal; Sigma).
Immunohistochemistry-The usefulness of the anti-HERG antibodies for immunohistochemistry was assessed in experiments on ERG1transfected cells and on cryostat sections of adult rat ventricles. All manipulations were performed at 22-23°C. Approximately 48 h after transfections, cells were washed with 0.1 M phosphate-buffered saline (PBS; pH 7.2) and fixed with 4% paraformaldehyde (in PBS). Fixed cells were washed with PBS and incubated for 1 h in blocking buffer I (PBS containing 5% normal goat serum, 0.2% Triton X-100, and 0.1% NaN 3 ). Cells were then incubated overnight at 4°C with anti-HERG antibodies in blocking buffer I. Following washing with PBS, cells were incubated first with a biotinylated goat anti-rabbit secondary antibody (Sigma) diluted 1:200, followed by avidin coupled to biotinylated alkaline phosphatase H (Vector, Burlingame, CA) diluted 1:100 for 1 h each. After washing, cells were exposed to the alkaline phosphatase substrate 5-bromo-4-chloro-3-indoyl phosphate/nitro blue tetrazolium (Bio-Rad); the reaction was monitored under brightfield illumination and stopped by addition of PBS. Cells on coverslips were dehydrated in ethanol, cleared in xylene, and mounted with Krystalon for viewing and photography.
For the preparation of cryostat sections, adult (Harlan Sprague-Dawley) rats (ϳ 0.5 kg) were anesthetized with CO 2 ; hearts were rapidly removed, placed in a tissue bath perfused with Tyrode's solution at 37°C containing 121 mM NaCl, 5 mM KCl, 15 mM NaHCO 3 , 1 mM Na 2 HPO 4 , 2.8 mM sodium acetate, 1 mM MgCl 2 , 2.2 mM CaCl 2 , and 5.5 mM glucose equilibrated with 95% O 2 /5% CO 2 (pH 7.4), and subsequently frozen in isopentane at Ϫ40°C. Small pieces of the ventricles were removed, trimmed, attached to a tissue holder using Histo Prep (Fisher), and placed on the rapid freeze stage of a Microm cryostat (Zeiss). Cryostat sections were cut at 20 m, collected, and air-dried. Dried sections were incubated in blocking buffer II (PBS containing 0.2% Triton X-100 and 0.5% bovine serum albumin (BSA; Sigma)) for 1 h, followed by a 2-h incubation in one of the anti-HERG antibodies in blocking buffer II. After washing with PBS, sections were incubated with a Texas Red-conjugated goat anti-rabbit IgG secondary antibody (Jackson, West Grove, PA) diluted 1:2000 in blocking buffer II. After rinsing with PBS, sections were mounted using ImmunoFluore (ICN, Costa Mesa, CA). For photography, sections were placed on the stage of a Zeiss LSM 410 inverted confocal microscope and viewed using a 63ϫ, 1.4 numerical aperture objective (optical thickness, Ϸ1.0 m); sections were viewed through the entire 20 m in 1-m steps. Texas Red was excited with a helium/neon (543 nm) laser and imaged on a photomultiplier; images were printed on a Kodak XLS8300 high resolution (300 DPI) printer.
Human, rat, and mouse heart membrane proteins were isolated using a protocol similar to that previously described for rat heart (28). Adult human ventricular and atrial tissue samples were from normal donor hearts that were not accepted for transplantation. These hearts were procured by LifeBanc (Cleveland, OH) with consent for research use; one was from a 20-year-old caucasian male, and the other was from a 60-year-old caucasian male. Macroscopic and microscopic examination revealed no evidence of pathology or underlying cardiovascular disease. All procedures were performed at 4°C, and all solutions contained the mixture of protease inhibitors described above (for the brain membrane preparation) as well as 1 mmol/liter benzamidine, 7.9 mol/ liter aprotinin, and 0.15 mol/liter leupeptin). For the preparation of human cardiac membrane proteins, tissue samples (Ϸ1-2 g) from individual donors were processed. For rat and mouse, ventricles and atria from several (2-12) animals were combined. Seven separate rat and three different mouse atrial and ventricular membrane preparations were analyzed. Tissue samples were homogenized at 4°C in 10 ml of TE buffer (containing 10 mmol/liter Tris-HCl and 1 mmol/liter EDTA, pH 7.4) using a Tissuemizer TM . After centrifugation (1000 ϫ g, 10 min), the supernatants were retained, and the pellets were resuspended to original volume in TE buffer, homogenized, and centrifuged. These supernatants were collected, pooled with the original supernatants, and centrifuged (40,000 ϫ g, 10 min). The resulting pellets were resuspended in TE buffer containing 0.6 mol/liter KI, incubated on ice (30 min), centrifuged (40,000 ϫ g, 10 min), and then washed twice with TE buffer. The final pellets were resuspended in TE buffer containing 2% Triton X-100 and incubated on ice (1 h) to solubilize membrane proteins. A final centrifugation (17,400 ϫ g, 30 min) precipitated the insoluble material; supernatants were concentrated using a Centricon 30 (Amicon, Beverly, MA). After completing protein assays, samples were aliquoted and frozen at Ϫ20°C until used.
Western Blots-QT-6 cell extracts (Ϸ40 g of protein), rat and mouse brain membrane proteins (Ϸ35 g), and rat and human atrial and ventricular membrane proteins (Ϸ60 g) were fractionated by SDS-PAGE and then probed by Western blot analysis with the anti-HERG antibodies. QT-6 cells plated on 100-mm tissue culture dishes were placed on ice and rinsed with PBS. Cold TE buffer (0.4 ml) containing 2% Triton X-100 and the same mixture of protease inhibitors used in the heart membrane preparations was then added. Cells were scraped from the dishes, triturated with a 1-ml syringe and an 18-gauge needle, and transferred to an eppendorf tube on ice; after 30 min, the samples were triturated again. Insoluble material was pelleted (17,500 ϫ g, 3 min, 4°C), and the supernatants were again incubated on ice (30 min). Following a second centrifugation, the supernatants were collected and combined. Standard protein assays were completed, and the samples were frozen at Ϫ20°C until used.
QT-6 cell extracts and heart and brain membrane proteins were fractionated on 10% polyacrylamide gels and transferred to Hybond-PVDF membrane (Amersham Pharmacia Biotech). The membranes were washed (3 min) in PBS and incubated (1 h) in blocking buffer III (0.2% I-Block (Tropix, Bedford, MA) in PBS with 0.1% Tween TM 20), followed by an overnight incubation at 4°C with one of the anti-HERG antibodies in blocking buffer I. The next day, membranes were washed (10 min) and then incubated for 1 h at room temperature with alkaline phosphatase-conjugated goat anti-rabbit IgG (Tropix) diluted 1:10,000 in blocking buffer III. After incubation, membranes were washed in blocking buffer III for 15 min and then for 2 min in assay buffer (0.1 mol/liter diethanolamine with 1 mmol/liter MgCl 2 , pH 9.8; Tropix). Bound antibodies were detected using the chemiluminescent substrate CSPD TM (Tropix).
To determine whether MERG1 and/or RERG1 is glycosylated, mouse and rat brain membrane proteins were incubated in the presence of PNGase F (glycopeptidase F; Sigma), an enzyme that cleaves N-linked sugars. Prior to exposure to the enzyme, rat/mouse brain membrane proteins (126 g), prepared as described above, were denatured by boiling for 10 min. After cooling, 6.3 l (1.26 units) of PNGase F was added, and the samples were incubated at 37°C for 24 h. Following the (24 h) incubation, membrane proteins were fractionated, and Western blots were performed as described above with the C-anti-HERG antibody.
Isolation of Adult Rat Cardiac Myocytes-Adult rat cardiac myocytes were isolated using a modified version of a protocol described previously (31). For each experiment, an adult male Harlan Sprague-Dawley rat (250 -275 g) was anesthetized with pentobarbital (0.8 ml, 50 mg/ml). Each heart was rapidly excised and placed in cold incubation buffer containing 118 mM NaCl, 4.8 mM KCl, 1.2 mM MgCl 2 , 1.2 mM KH 2 PO 4 , 0.68 mM glutamine, 11 mM dextrose, 25 mM HEPES, 5 mM pyruvate, and 1 M insulin (pH 7.3). Following cannulation and flushing with 2 ml of cold incubation buffer, the heart was mounted on a Langendorff apparatus and perfused retrogradely through the aorta with 50 ml of warmed (37°C) perfusion buffer containing 118 mM NaCl, 37.5 mM NaHCO 3 , 4.8 mM KCl, 1.2 mM MgCl 2 , 1.2 mM KH 2 PO 4 , 1.0 mM CaCl 2 , 0.68 mM glutamine, 16.5 mM dextrose, and 7.5 mM pyruvate, equilibrated with humidified 95% O 2 /5% CO 2 (pH 7.2). The heart was then perfused with 50 ml of a nominally Ca 2ϩ -free perfusion buffer, followed by perfusion buffer (100 ml) containing Type II collagenase (140 units/ ml; Worthington, Freehold, NJ) and 0.20 mg/ml BSA (Fraction V; Sigma). After 16 min, Ca 2ϩ was added gradually until the free Ca 2ϩ concentration was 1 mM. The flow rate was then increased to 10 ml/min; when the back pressure stabilized (Ϸ30 min), the heart was removed. Atria and ventricles were separated, minced, and transferred to fresh collagenase-containing perfusion buffer in a shaking water bath at 37°C, aerated with humidified 95% O 2 /5% CO 2 ; for the ventricular tissue, the collagenase concentration was increased to 400 units/ml. The secondary incubations were 5 min for atria and 10 min for ventricles, and the tissue pieces were mechanically dispersed by gentle trituration. The resulting suspensions were filtered through gauze, and cells were pelleted by centrifugation at 400 rpm for 1 min. Myocytes were resuspended in incubation buffer containing 0.5% BSA and rinsed with 2 ml of incubation buffer containing 4% BSA. After settling, myocytes were resuspended in incubation buffer with 1 mM Ca 2ϩ at room temperature under 100% O 2 . The yields of rod-shaped atrial and ventricular myocytes were 60 -80%, and cells were used within 6 h.
Electrophysiological Recordings-The nystatin perforated patch, whole cell recording technique (32) was employed to record K ϩ currents from isolated, Ca 2ϩ -tolerant, rod-shaped adult rat atrial and ventricular myocytes (33). The bath solution contained 135 mM NaCl, 5 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 5 mM NaHEPES, 3 mM sodium acetate, 5 mM glucose, and 0.002 mM nifedipine at pH 7.4. Recording pipettes, fabricated from (Corning 8161) borosilicate glass (World Precision Instruments, Sarasota, FL), had tip diameters of 2-3 m and resistances of 2-4 M⍀ when filled with recording solution. The pipette solution contained 100 mM potassium methanesulfonate, 40 mM KCl, 5 mM K 2 EGTA, 2 mM MgCl 2 , and 10 mM HEPES at pH 7.4. Nystatin (Sigma), dissolved in dimethyl sulfoxide (6 mg/100 l), was diluted in pipette solution to 100 g/ml; the nystatin-containing pipette solution was sonicated and used within 3 h. Nystatin-free pipette solution was added to the tip of a recording pipette prior to adding the nystatin-containing solution. Seal resistances were Ն5 G⍀. After establishing the whole cell configuration, Ϯ10 mV steps were applied to allow measurements of cell membrane capacitances and input resistances. Whole cell membrane capacitances and series resistance were compensated (Ն 60 -80%) electronically; voltage errors resulting from the uncompensated series resistance were always Յ4 mV and were not corrected. Experimental data were acquired using an Axopatch 200A amplifier (Axon, Burlingame, CA), interfaced (Digidata 1200; Axon) to an IBM compatible 486/33 microcomputer, and pClamp 6.03 (Axon). Data were acquired at 10 kHz, and current signals were filtered at 1 kHz prior to digitization and storage. For experiments, myocytes in 35-mm culture dishes were placed in a thermal stage controller (Bioptech ⌬T system, Butler, PA), maintained at 35°C, and gassed with 100% O 2 . E-4031 was kindly donated by Esai Co., Ltd (Tsukuba, Japan). Analysis of digitized data was completed using pClamp and Origin (Microcal, Northhampton, MA). Averaged and normalized data are presented as means Ϯ S.E. Statistical significance was assessed using the Student's t test; p values are presented in the text.

Specific C-and N-Anti-HERG Antibodies-
The affinity purified C-and N-anti-HERG antibodies were initially tested for specificity and for potential cross reactivity with other voltagegated K ϩ channel ␣ subunits by immunohistochemistry and Western blot analysis on QT-6 cells transfected with cDNA constructs encoding HERG1, Kv1.2, Kv2.1, or Kv4.2. The immunohistochemical experiments revealed that QT-6 cells transfected with HERG1 cDNA stained when probed with either the C-anti-HERG (Fig. 1A, panel a) or the N-anti-HERG (Fig. 1A, panel b) antibody, whereas no staining was detected in either mock transfected (data not shown) or Kv1.2-transfected (Fig. 1A, panels c and d) cells. Similar negative results were obtained when QT-6 cells transfected with cDNA constructs encoding Kv2.1 or Kv4.2 were probed with the anti-HERG antibodies (data not shown). As demonstrated previously (28), expression of Kv1.2, Kv2.1, or Kv4.2 is readily detected in QT-6 cells expressing these proteins with Kv ␣ subunit-specific antibodies.
In parallel experiments, extracts from QT-6 cells transfected with cDNA constructs encoding HERG1, Kv1.2, Kv2.1, or Kv4.2 and from mock transfected QT-6 cells were fractionated by SDS-PAGE, transferred to PVDF membranes, and blotted with the anti-HERG antibodies. Immunoblots with the C-anti-HERG antibody revealed a single protein band at approximately 155 kDa in extracts of QT-6 cells transfected with HERG cDNA (Fig. 1B, lane a). A protein of the same molecular mass was identified in HERG1-transfected HEK-293 cells with this (and the N-anti-HERG) antibody (14,16), although in HEK-293 cells an additional lower molecular mass (Ϸ135 kDa) protein is also expressed (see "Discussion"). Nothing was detected with the C-anti-HERG antibody in extracts from either mock transfected QT-6 cells (Fig. 1B, lane b) or QT-6 cells transfected with the Kv1.2, Kv2.1, or Kv4.2 cDNA. The 155-kDa protein is also identified in extracts of HERG1-transfected QT-6 cells probed with the N-anti-HERG antibody (Fig. 1B,  lane c) and, as with the C-anti-HERG antibody (Fig. 1B, lane  a), this band is not detected in extracts from either mock transfected QT-6 cells (Fig. 1B, lane d) or QT-6 cells transfected with the Kv1.2, Kv2.1, or Kv4.2 cDNA. The 155-kDa protein was also not seen in Western blots when the antibodies were preincubated with the peptides against which each was generated (data not shown).
Expression of ERG1 and ERG1b Revealed with C-terminal Anti-HERG Antibodies-Subsequent experiments were aimed at determining whether the anti-HERG antibodies could also be used to examine the expression of the alternatively spliced HERG1 variant, HERG1b ( Fig. 2A), which has been postulated to play a role in the generation of functional I Kr channels (9,23), as well as to detect ERG1 protein expression in other species. Preliminary immunohistochemical experiments, similar to those described above (Fig. 1A), revealed that expression of HERG1, MERG1, and RERG1 is reliably detected with the C-anti-HERG and N-anti-HERG antibodies, whereas only the C-anti-HERG antibody can be used to document HERG1b or MERG1b expression (data not shown). To identify the molecular masses of proteins produced on expression of these constructs, Western blot experiments, similar to those described above (Fig. 1B), were completed on extracts from QT-6 cells transfected with cDNA constructs encoding HERG1, HERG1b, MERG1, MERG1b, RERG1, or Kv1.4 and from mock transfected QT-6 cells. Immunoblots with the C-anti-HERG antibody revealed a single 155-kDa protein in extracts of QT-6 cells transfected with the full-length HERG1, MERG1, or RERG1 cDNA constructs (Fig. 2B, lanes a under Herg1, Merg1, and Rerg1), whereas nothing is detected in extracts of cells expressing Kv1.4 (Fig. 2B) or in mock transfected cells (not shown) with this antibody. Expression of the full-length 155-kDa HERG1, MERG1, and RERG1 proteins is also identified in Western blots probed with the anti-HERG C-terminal fusion protein antiserum (15) (Fig. 2C) and with the Alomone Cterminal anti-HERG antibody (Fig. 2D). All three C-terminal anti-HERG antibodies, therefore, identify the same full-length 155-kDa (human, mouse, and rat) ERG1 protein ( Fig. 2 and Table I).
In cells transfected with constructs encoding the truncated HERG1b or MERG1b proteins, the C-anti-HERG antibody also identifies a single protein band, although in this case the molecular mass of the protein is 95 kDa (Fig. 2B, lanes b under  Herg1 and Merg1). Expression of the 95-kDa HERG1b and MERG1b proteins is also identified in Western blots probed with the anti-HERG C-terminal fusion protein antiserum (15) (Fig. 2C, lanes b) and with the Alomone C-terminal anti-HERG antibody (Fig. 2D, lanes b). As with the full-length 155-kDa

ERG1 Expression in Rat, Mouse, and Human Heart
ERG1 proteins, all three C-terminal anti-HERG antibodies identify the same 95-kDa protein (Fig. 2, lanes b), suggesting that all of these antibodies recognize the same ERG1b protein(s) and that they can be used to detect ERG1b expression in tissue. As expected, nothing is detected in Western blots of extracts of QT-6 cells expressing HERG1b or MERG1b probed with the N-anti-HERG antibody (data not shown). No specific bands were detected with the C-terminal anti-HERG antibodies in extracts from either mock transfected QT-6 cells (data not shown) or QT-6 cells transfected with a cDNA construct encoding Kv1.4 (Fig. 2, B-D). In addition, neither the 155-kDa nor the 95-kDa proteins was seen in Western blots when the antibodies were preincubated with the peptides (B, D) or the fusion protein (C) against which each antibody was generated (data not shown).
Two ERG1 Proteins in Rat and Mouse Brain-In Western blots of fractionated rat brain membrane proteins, both the Cand the N-anti-HERG antibodies identify a protein at approximately 165 kDa (Fig. 3, A and C), a molecular mass similar to that of the protein (155 kDa) detected with these antibodies in immunoblots of extracts of ERG1-transfected QT-6 ( Fig. 1) and HEK-293 (14 -16) cells. The 165-kDa protein was not evident when the antibodies were preincubated with the peptides against which each was generated (Fig. 3, A and C, ϩ lanes). In contrast to the findings in QT-6 ( Fig. 1) and HEK-293 (14 -16) cells, however, an additional high molecular mass (205 kDa) protein, was also detected in blots of fractionated rat brain membrane proteins probed with either the C-or the N-anti-HERG antibodies (Fig. 3, A and C). The 205-kDa band was also not seen when the antibodies were preincubated with the peptides against which each (antibody) was generated (Fig. 3, A  and C, ϩ lanes). With the N-anti-HERG (Fig. 3C), however, the intensity of the 205-kDa band is low relative to the 165-kDa band. This intensity difference is not seen with the C-anti-HERG antibody (Fig. 3A), suggesting that the affinity of the N-anti-HERG antibody is greater for the 165-kDa than the 205-kDa ERG1 protein (see "Discussion").
Importantly, no protein bands at Ϸ 95 kDa, as might be expected to result from expression of the N-terminal alternatively spliced ERG1b variant (Fig. 2), are seen in Western blots of fractionated rat brain membrane proteins probed with the C-anti-HERG antibody (Fig. 3A), suggesting that ERG1b is not expressed (at the protein level) in rat brain (see "Discussion"). In Western blots of fractionated mouse brain membrane proteins (Fig. 3B), virtually identical results were obtained, i.e. two proteins at 165 and 205 kDa were routinely identified   (Table I), and there were no prominent low molecular mass bands in these blots (Fig. 3B), as would be expected if MERG1b were expressed. The ERG1 proteins expressed in mouse and rat brain, therefore, are indistinguishable; results similar to those presented in Fig. 3 were obtained in Western blot analyses of seven rat and three mouse brain membrane protein preparations.

ERG1 Expression in Rat, Mouse, and Human Heart
The Two MERG1 and RERG1 Proteins Reflect N-Linked Glycosylation-As noted previously, the predicted molecular mass of full-length HERG is 127 kDa (6,7) and the molecular mass of the protein recognized in HERG1-, MERG1-, and RERG1-transfected QT-6 ( Figs. 1 and 2) and in HERG1-transfected HEK-293 (14 -16) cells by the C-and the N-anti-HERG antibodies is 155 kDa. The finding of two proteins and, particularly, the 205-kDa protein in rat and mouse brain membrane protein preparations, therefore, was unexpected. It is theoretically possible that the 165-and 205-kDa proteins reflect the expression of distinct ERG subfamily genes, splice variants of the same ERG gene or, alternatively, post-translational modifications of the same ERG holoprotein. Although two additional members, ERG2 and ERG3, of the ERG subfamily have recently been cloned from brain (8), the predicted amino acid sequences of the ERG2 and ERG3 proteins are distinct from ERG1, and neither protein would be detected by the N-and C-anti-HERG antibodies. Although the MERG1 and HERG1 transcripts are alternatively spliced in mouse and human heart (9,(23)(24)(25), and this could certainly also occur in (rat and mouse) brain, splicing reduces the size of the ERG1 message and results in the production of lower (rather than higher) molecular mass ERG1 proteins (see Fig. 2). It seemed more likely, therefore, that the finding of two high molecular mass ERG1 proteins in rat and mouse brain reflects post-translational processing.
Examination of the sequence of HERG1 (RERG1 or MERG1) reveals multiple potential sites for post-translational modifications, including several sites for serine-threonine kinase phosphorylation, N-myristoylation, and N-linked glycosylation; no O-linked glycosylation sites are found. The large difference between the 205-kDa (rat and mouse brain) ERG1 proteins detected in the Western blots (Fig. 3) and the predicted molecular mass (127 kDa) (6, 7) of the ERG1 proteins suggested that N-linked glycosylation might be involved. In addition, it has been reported that HERG1 expressed in HEK-293 cells is glycosylated (14 -16). To determine whether ERG1 is (N-linked) glycosylated in vivo, rat and mouse brain membrane proteins were denatured (by boiling) and incubated at 37°C in the presence of PNGase F (see "Materials and Methods"), which cleaves N-linked sugars. Western blot analyses of samples in-cubated with PNGase F for 24 h revealed marked reductions in the intensities of both the 165-kDa and the 205-kDa bands, as well as the appearance of two new bands at Ϸ175 and 130 kDa (Fig. 4). The results obtained on the rat (Fig. 4A) and the mouse (Fig. 4B) brain samples are indistinguishable, consistent with the suggestion above that the rat and mouse ERG1 proteins are the same. Importantly, the 175-kDa and the 130-kDa proteins do not reflect breakdown of the ERG1 protein, as evidenced by the fact that the 165-and 205-kDa bands are readily detected, whereas the 175-and 130-kDa proteins are not, in immunoblots of rat and mouse membrane proteins incubated for 24 h at 37°C in the absence of PNGase F (Fig. 4). Similar results were obtained in Western blots of four rat brain and two mouse brain membrane protein preparations treated with PNGase F. In addition, the results are similar to previous findings on HERG1-transfected HEK-293 (14,15) cells and suggest that the 165-and 205-kDa bands detected in rat and mouse brain reflect differences in N-linked glycosylation of the same (fulllength) ERG1 protein (see "Discussion").
RERG1 Expression in Rat Heart-Western blots of fractionated rat ventricular and atrial membrane proteins with the C-anti-HERG antibody also revealed proteins of 165 and 205 kDa (Fig. 5A); both bands were eliminated when the antibody was preincubated with the C-terminal peptide against which it was generated. The 165-and 205-kDa bands are indistinguishable from those detected in rat brain membrane preparations (Fig. 3A), as evidenced by Western blots of mixed brain and heart samples in which (only) the same two (165-and 205-kDa) proteins are detected (data not shown). Immunoblots of rat heart membrane proteins with the N-anti-HERG antibody revealed only a prominent band at 165 kDa; the 205-kDa ERG1 protein was not evident. Similar results were obtained with the N-anti-HERG antiserum, which reveals intense labeling of the 165-kDa (but not the 205-kDa) protein, as well as several other unidentified proteins (Fig. 5A). The difficulties encountered with detecting the 205-kDa ERG1 protein in brain (see above and Fig. 3C) and heart (Fig. 5A) with the N-terminal (but not the C-terminal) antibody suggests that the N-terminal amino acid sequence against which the antibody was generated is either modified or inaccessible in the denatured (205-kDa) protein (see below and "Discussion"). The Western blots with the C-anti-HERG and N-anti-HERG antibodies also revealed that ERG1 expression is substantially higher in rat atrial than in FIG. 3. Western blots reveal the presence of two distinct ERG1 proteins in rat and mouse brain. Adult rat (A and C) and mouse (B) brain membrane proteins (35 g) were fractionated by SDS-PAGE, transferred to PVDF membranes, and immunoblotted with the C-anti-HERG (A and B) or the N-anti-HERG (C) antibody (diluted 1:500). The antibodies were applied either directly (Ϫ) or after preincubation (؉) with 10 g/ml of the peptide against which the antibody was generated. Two bands at approximately 165 and 205 kDa (arrows) are detected with the C-anti-HERG (A and B) and N-anti-HERG (C) antibodies in brain (see text). Immunohistochemical analysis revealed that ERG1 is also readily detected in the membranes of adult rat ventricular and atrial cells. As illustrated in Fig. 5B, robust membrane (with little or no detectable cytosolic) labeling is evident on cryostat sections of adult rat ventricles probed with both of the anti-HERG antibodies. The subcellular labeling patterns seen with the C-and N-anti-HERG antibodies, however, are distinct (Fig.  5B). In low and high magnification confocal images of sections immunolabeled with the C-anti-HERG antibody (Fig. 5B, panels a and b), for example, staining is evident in the lateral (i.e. the sarcolemmal) membranes and in the T-tubules; the arrows in Fig. 5B, panel b indicate lateral membrane labeling. With the N-anti-HERG antibody, in contrast, only labeling of the T-tubules was revealed (Fig. 5B, panel e). Similar results were obtained on cryostat sections prepared from ventricles isolated from three different animals. The difference in the labeling patterns with the N-and C-anti-HERG antibodies is intriguing, particularly in light of the Western blot data (Fig. 5A), demonstrating a difference between the N-and C-anti-HERG antibodies in the detection of the 205-kDa ERG1 protein (see "Discussion").
I Kr in Rat Atrial and Ventricular Myocytes-The finding of robust expression of the ERG1 protein(s) in adult rat heart was initially surprising for two reasons. First, to our knowledge, there have been no published reports documenting the presence of I Kr in adult rat atrial myocytes. In addition, although I Kr has been reported in adult rat ventricular myocytes (34,35), the densities of the currents are low, particularly when compared with the densities of other voltage-gated K ϩ currents in these cells (36 -38). Electrophysiological experiments were undertaken, therefore, to determine I Kr densities in these cells. Voltage-gated K ϩ current tails at Ϫ30 mV were recorded (see FIG. 6. E-4031-sensitive K ؉ currents, I Kr , in adult rat atrial and ventricular myocytes. B, whole cell K ϩ currents recorded at 35°C from an adult rat atrial myocyte using the nystatin perforated patch technique; the voltage-clamp protocol is displayed in A, and the interpulse interval was 5 s. The holding potential was Ϫ50 mV to inactivate voltage-gated Na ϩ channels, and 2 M nifedipine was added to the bath to suppress voltage-gated Ca 2ϩ currents. The tail currents at Ϫ30 mV in B are replotted in C at a higher gain. In the presence of 5 M E-4031, the slowly decaying (I Kr ) tail currents were suppressed completely (D). I Kr tails were recorded on repolarization to Ϫ30 mV following 200-ms depolarizations from a holding potential of Ϫ50 mV to varying test potentials between Ϫ30 and ϩ50 mV. Currents in the presence of E-4031 were subtracted from the controls, and the peak amplitudes of the E-4031-sensitive tails were determined from single exponential fits to the tail current decays fitted from the end of the test pulse until the current reached a steady-state level (0.5-1.0 s). Peak tail currents determined in individual cells were then normalized to whole cell membrane capacitance, and mean (Ϯ S.E.) peak (I Kr ) tail current densities in adult rat atrial (n ϭ 22) and ventricular (n ϭ 22) myocytes are plotted versus test voltage in E (see text).
FIG. 5. ERG1 protein expression in rat heart. A, rat ventricular (lanes V) and atrial (lanes A) membrane proteins (60 g) were fractionated by SDS-PAGE, transferred to PVDF membranes, and immunoblotted with either the C-anti-HERG antibody (diluted 1:100) or the N-anti-HERG antiserum (diluted 1:500). Similar to the results in brain (Fig. 3), two ERG1 proteins at 165 and 205 kDa were identified with the C-anti-HERG antibody in immunoblots of rat atrial and ventricular membrane proteins; both bands were eliminated when the antibody was preincubated (؉) with 10 g/ml of the peptide against which it was generated. In addition, ERG1 protein expression is higher in rat atrial than in rat ventricular membrane protein preparations. In contrast to the results in brain (Fig. 3), only the 165-kDa ERG1 is detected in rat heart membranes with the N anti-HERG antibody. B, immunofluorescence images of cryostat sections of rat ventricular myocardium exposed to either the C-anti-HERG (panels a-c) or the N-anti-HERG (panels d-f) antibody followed by Texas Red-conjugated goat anti-rabbit IgG. Sections photographed in panels c and f were incubated with the anti-HERG antibodies and 10 g/ml of the peptide against which each antibody was generated. With the C-anti-HERG (panels a and b) antibody, labeling is evident in the lateral membranes of ventricular myocytes and in the T-tubules; the arrows (panel b) indicate lateral membrane labeling. With the N-anti-HERG antibody, no lateral membrane labeling is evident (panels d and e); rather, this antibody appears to label T-tubular membranes exclusively; the arrows (panel e) indicate the absence of detectable plasmalemmal membrane labeling (see text). "Materials and Methods") under control conditions and following exposure to the class III antiarrhythmic, E-4031, a specific blocker of I Kr (17)(18)(19)(20)(21)(22). To evoke I Kr , cells were depolarized to ϩ20 mV from a holding potential of Ϫ50 mV for times ranging from 25 to 575 ms in 50-ms increments (Fig. 6A). These experiments revealed slowly decaying tail currents in adult rat atrial myocytes (Fig. 6, B and C) that were suppressed in the presence of 5 M E-4031 (Fig. 6D), consistent with the expression of I Kr (17)(18)(19)(20)(21)(22). Results similar to those presented in Fig. 6 (B-D) were obtained on 21 other adult rat atrial and on 22 adult rat ventricular myocytes.
The time and voltage-dependent properties of I Kr in adult rat atrial and ventricular cells are indistinguishable. Boltzman fits to the mean normalized peak tail current amplitudes as a function of voltage, for example, yielded V1 ⁄2 values of Ϫ19.4 mV (k ϭ 6.4) and Ϫ18.2 mV (k ϭ 11.1) for I Kr in rat atrial and ventricular myocytes, respectively. To compare I Kr densities in adult rat atrial and ventricular myocytes, the "E-4031-sensitive" tail currents were obtained by subtraction of the currents in the presence of E-4031 (Fig. 6D) from the control records (Fig. 6C). Peak tail current amplitudes at Ϫ30 mV in individual cells were then determined from single exponential fits to the tail current decays in these subtracted records and normalized to the whole cell capacitance. These analyses confirmed that I Kr density is significantly higher in adult rat atrial than in adult rat ventricular myocytes (Fig. 6E). Mean (Ϯ S.E.) peak tail current densities at Ϫ30 mV, for example, were 0.62 Ϯ 0.03 pA/pF (n ϭ 22) and 0.36 Ϯ 0.01 pA/pF (n ϭ 22) in adult rat atrial and ventricular myocytes, respectively. Interestingly, these differences in I Kr density parallel the observed differences in ERG1 protein expression revealed in Western blots (Fig. 5A).
A Distinct Isoform of HERG Is Expressed in Human Heart-In contrast to the findings in rat, Western blot analysis of human atrial and ventricular membrane proteins using the C-anti-HERG antibody revealed a single 145-kDa ERG1 protein (Fig. 7A). The 145-kDa band was eliminated when the C-anti-HERG antibody was preincubated with the peptide against which it was generated (Fig. 7A). In addition, the same band was identified in Western blots of human ventricular and atrial membrane proteins probed with the N-anti-HERG antibody, suggesting that the antibodies are detecting ERG1 protein expression. In Western blots of fractionated mouse heart membrane proteins (Fig. 7B), however, results identical to those in rat heart (Fig. 5A), as well as in mouse (and rat) brain (Fig. 3) were obtained, i.e. two proteins at 165 and 205 kDa were routinely identified. Results similar to those presented in Fig. 7 were obtained in Western blot analyses of five human and three mouse heart membrane protein preparations.
The results presented in Fig. 7A reveal that the molecular mass of the single ERG1 protein (145 kDa) identified in human heart, therefore, is substantially lower than the molecular mass of either of the proteins (165 and 205 kDa) identified with this antibody in rat or mouse heart and brain (compare Figs. 3-5 and 7B), suggesting marked species-specific differences in post-translational processing of ERG1 proteins. In addition, and in contrast to the results in rat (Fig. 5A) heart, the Western data (Fig. 7A) suggest that ERG1 protein expression is higher in human ventricular than in atrial tissue. The expression of the ERG1 proteins is also higher in mouse ventricles than atria (Fig. 7B) (see "Discussion").

DISCUSSION
Generation and Characterization of Specific Anti-HERG Antibodies-The C-and N-anti-HERG antibodies developed here were generated against unique peptide sequences in the C (residues 1145-1159) and N (residues 174 -188) termini, re-spectively, of HERG1 ( Fig. 2A). Experiments completed on HERG1-transfected QT-6 cells revealed that both antibodies are specific for HERG1 and reliably detect HERG1 protein expression using immunohistochemistry or Western blot analysis. Although the sequences of mouse ERG1 (MERG1) and rat ERG1 (RERG1) were not available at the time the peptides were selected for antibody generation, we assumed that the sequences of MERG1 and RERG1 would be very similar to HERG1. Subsequent sequence alignment has indeed revealed that both RERG1 (27) and MERG1 (9, 23) are highly homologous to HERG1; the sequence identity is Ն97% at the amino acid level. In addition, the C termini of RERG1, MERG1, and HERG1, targeted by the C-anti-HERG antibody (corresponding to residues 1145-1159 in HERG1), are (100%) identical ( Fig.  2A). There are several amino acid substitutions in the N-terminal region (corresponding to residues 174 -188 in HERG1) against which the N-anti-HERG antibody was generated (6,7,9,23,25). Nevertheless, 8 of the 15 amino acids (60%) in this region are identical in RERG1 and HERG1, and 10 of the 15 (67%) are identical in HERG1 and MERG1 ( Fig. 2A). More importantly, the results presented here reveal that despite these sequence differences, the N-anti-HERG antibody can be used to detect RERG1 expression (Figs. [2][3][4][5]. Because of the sequence differences (between MERG1/RERG1 and HERG1) and the variable labeling intensity of the 205-kDa RERG1 protein (relative to the 165-kDa protein), however, the C-anti-HERG antibody is preferred for studying ERG1 protein expression. FIG. 7. Different molecular mass ERG1 proteins expressed in human and mouse heart. A, human ventricular (lanes V) and atrial (lanes A) (60 g) membrane proteins were fractionated by SDS-PAGE, transferred to PVDF membranes, and immunoblotted with the C-anti-HERG antibody (1:100) either directly (Ϫ) or after preincubation (؉) with 10 g/ml of the peptide against which the antibody was generated. A single HERG1 protein with an apparent molecular mass of 145 kDa (arrow) is detected in atria and ventricles. B, mouse ventricular (lanes V) and atrial (lanes A) membrane proteins (30 g) were fractionated by SDS-PAGE, transferred to PVDF membranes, and immunoblotted with the C-anti-HERG antibody (1:100) either directly (Ϫ) or after preincubation (؉) with 10 g/ml of the peptide against which the antibody was generated. Similar to the results in rat heart (Fig. 5) and in rat and mouse brain (Fig. 3), two ERG1 proteins at 165 and 205 kDa were identified in mouse heart. In contrast to the results in rat heart (Fig. 4), ERG1 protein expression appears to be higher in ventricular (lanes V) than in atrial (lanes A) tissue in human (A) and in mouse (B) heart. here were performed on adult tissues, it certainly remains a possibility that ERG1b proteins are expressed and contribute to the formation of I Kr channels at other developmental stages.
It has also been suggested that accessory K ϩ channel subunits that are present in heart contribute (with ERG1) to I Kr (40 -42). In AT-1 cells (an atrial tumor cell line), for example, I Kr is attenuated on exposure to antisense oligodeoxynucleotides targeted against the small (130 amino acids), single transmembrane spanning domain accessory K ϩ channel subunit, minK (40). In addition, heterologously expressed minK and HERG1 reportedly coimmunoprecipitate (41). The functional significance of this finding in terms of I Kr is not clear, however, because coexpression of minK with HERG1 results in voltagegated K ϩ currents that are indistinguishable from those produced on expression of HERG1 alone (41). More recently, a homologue of minK, referred to as MiRP1 (minK-related peptide 1), has been identified that also coassembles with HERG1 in Xenopus oocytes (42). In contrast to the results with minK (41), however, coexpression of MiRP1 with HERG1 results in voltage-gated K ϩ currents that are very similar to cardiac I Kr , suggesting that I Kr channels are composed of full-length HERG1 and MiRP1 proteins (42). These observations, together with the results presented here demonstrating only the expression of full-length ERG1 proteins in rat, mouse, and human heart, suggest that it is quite unlikely that ERG1b proteins play a role in the generation of functional cardiac I Kr channels. Further experiments will be necessary to demonstrate directly that ERG1 and MiRP1 coassemble in heart in vivo and to determine the stoichiometry of these subunits in functional I Kr channels.