Cell, Voi. 80, 795-803, March 10, 1995, Copyright © 1995 by Cell Press A Molecular Basis for Cardiac Arrhythmia: HERG Mutations Cause Long QT Syndrome Mark E. Curran,*t Igor Splawski,*t Katherine W. Timothy,~ G. Michael Vincent,t§ Eric D. Green,# and Mark T. Keating*t$11 *Department of Human Genetics tEccles Program in Human Molecular Biology and Genetics $Cardiology Division flHoward Hughes Medical Institute University of Utah Health Sciences Center Salt Lake City, Utah 84112 §Department of Medicine Latter Day Saints Hospital Salt Lake City, Utah 84037 #National Center for Human Genome Research National Institutes of Health Bethesda, Maryland 20892 Summary To identify genes involved in cardiac arrhythmia, we investigated patients with long QT syndrome (LQT), an inherited disorder causing sudden death from a ventricular tachyarrhythmia, torsade de pointes. We previously mapped LQT loci on chromosomes 11 (LQT1), 7 (LQT2), and 3 (LQT3). Here, linkage and phys- ical mapping place LQT2 and a putative potassium channel gene, HERG, on chromosome 7q35-36. Single strand conformation polymorphism and DNA sequence analyses reveal HERG mutations in six LQT families, including two intragenic deletions, one splice-donor mutation, and three missense mutations. In one kin- dred, the mutation arose de novo. Northern blot analy- ses show that HERG is strongly expressed in the heart. These data indicate that HERG is LQT2 and suggest a likely cellular mechanism for torsade de pointes. Introduction Long QT syndrome (LQT) is an inherited cardiac disorder that causes syncope, seizures, and sudden death, usually in young, otherwise healthy individuals (Ward, 1964; Ro- mano, 1965; Schwartz et al., 1975). Most LQT gene carriers manifest prolongation of the QT interval on electrocardio- grams, a sign of abnormal cardiac repolarization (Vincent et al., 1992). The clinical features of LQT result from epi- sodic cardiac arrhythmias, specifically torsade de pointes, named for the characteristic undulating natu re of the elec- trocardiogram in this arrhythmia. Torsade de pointes may degenerate into ventricular fibrillation, a particularly lethal arrhythmia. Although LQT is not a common diagnosis, ven- tricular arrhythmias are very common; more than 300,000 Americans die suddenly every year (Kannel et al., 1987; Willich et al., 1987), and in many cases the underlying mechanism may be aberrant cardiac repolarization. LQT, therefore, provides a unique opportunity to study life- threatening cardiac arrhythmias at the molecular level. A molecular basis for LQT was not previously known. In 1991, we discovered tight linkage between autosomal dominant LQT and a poiymorphism at H-RAS (Keating et al., 1991a). This discovery localized an LQT gene (LQT1) to chromosome 1 lp15.5 and made genetic testing possi- ble in some families. Autosomal dominant LQT was pre- viously thought to be genetically homogeneous, and the first seven families that we studied were linked to 1 lp15.5 (Keating et al., 1991b). In 1993, however, several labora- tories, including our group, identified families that were not linked to chromosome 11 p15.5 (Benhorin et al., 1993; Curran et al., 1993a; Towbin et al., 1994). In 1994, we identified two additional LQT loci, LQT2 on chromosome 7q35-36 (nine families) and LQT3 on chromosome 3p21-24 (three families) (Jiang et al., 1994). Since three families in our study remained unlinked, at least one more LQT locus exists. This degree of heterogeneity suggests that distinct LQT genes may encode proteins that interact to modulate cardiac repolarization and arrhythmia risk. Since LQT is associated with abnormal cardiac repolar- ization, genes that encode ion channels (or their modu- lators) are reasonable candidates. H-RAS, which was lo- calized to chromosome 1 lp15.5, was excluded as a candidate for LQT1 based on direct DNA sequence analy- ses (unpublished data) and by linkage analyses (Roy et al., 1994). A skeletal muscle chloride channel (CLCN1; Koch et al., 1992) and a cardiac muscarinic-acetylcholine receptor (CHRM2; Bonner et al., 1987) became candidates for LQT2 based on their chromosome 7q35-36 location, but subsequent linkage analyses have excluded these genes (Wang et al., submitted). Warmke and Ganetzky (1994) identified a novel human cDNA, the human ether-a-go-go-related gene (HERG). HERG was localized to human chromosome 7 by polymer- ase chain reaction (PCR) analysis of a somatic cell hybrid panel (Warmke and Ganetzky, 1994). The function of the protein encoded by HERG is not known, but it has pre- dicted amino acid sequence homology to potassium chan- nels. HERG was isolated from a hippocampal cDNA library by homology to the Drosophila ether-a-go-go (eag) gene, which encodes a Ca2+-modulated potassium channel (Bruggeman et al., 1993). HERG is not the human homolog of eag, however, their products sharing only -50% amino acid sequence homology. In this study, we provide evidence indicating that HERG is LQT2. First, we identified and characterized new LQT families and showed that all were linked to markers on chromosome 7q35-36, confirming the location of LQT2. Second, we mapped HERG to chromosome 7q35-36, mak- ing HERG a candidate gene for LQT2. Third, we demon- strated that HERG is strongly expressed in the heart. Fi- nally, we identified six HERG mutations associated with LQT; one of these mutations arose de novo.
`
`1 of 9
`
`PENN EX. 2067
`CFAD V. UPENN
`IPR2015-01836
`
`

`
`Cell 796 K1956 D7S505 978636 6 2 IK5i6) HERG5 1 1 I~1) HERG3 1 1 1~3i2i D78483 11 I~1) +3 2+ 261~6 + +i I11111 ii31 ~ ~ i3)+ 1 1 111~I i (-~ III I" 3) I: !1 (- 2) K2287 I D78505 ( 1- 1~ 3~ D78636 (- 15)1] 2 7 HERG5 (- 10)11 1 1 HERG3 (- 12)II 1 1 D78483 (" l~j I 2 3 ':~3 33a 1~ 3 5~ 3~2 6 3 1 211Si7 3 61i512 ;5i211 4 2 1 [i0i 1 2 2 liOl 1 i011 I 2 2 2 1 li2} 1 2 11!211 !2i 1 12 2 + 31~3 2 21L!J 2 i~214 3 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 3 1~3~2i~ 3 3 23 33 2 3 2 3 33 3 3 1~ 3 li i i'~ I 2 2 5!21511i5 27 17 27 1 7 1732 32 6 3 1~1 23 1 1 0~110i2i0 11 21 11 21 2121 21 2 21 i0! 1O 2 1 211,2+2,2 11 21 11 21 2121 211 1 , i21, 2 3 311~ 4, 33 43 33 43 4 3 2 2 22 2 2~2~ 2 2 2 4 3 4 K2595 I III IV D78505 D7$636 HERG5 HERG3 D78483 /- 1 2 [~q Th! 1hi 22[ 213~ (- 1) 1~ 1 2 (- 4)Jiljl 1 {- 1}1i111 1 (- 2)1i2t2 2 (~3 2 1 2 1 1 li2i 4 1 4ili 1 1 1111 +223212i 5 i3~ 11 2)l(1 b)i (1 1)1(1 !1)', (2 2)1(2 i2)i (2 2)J(4 -o0(5 2 3 4 12 12 22 13 1 1 21 1 1 1 1 1 1 22 22 22 31 2.i~ 24 4 22 3 13 11 2 22 21 K2596 I D78505 D78636 HERG5 HERG3 D7S483 II K2600 I 313~23 1 2 078505 D78636 2 7 J~31 4 HERG5 2 1 Jill 1 HERG3 1 1 1121 1 D7S483 ++1~3 JL-J ( ( ( ( ( III III 1 2 2 4i 3 i2! 6 2i 2!1 1 li 1 i2 2 2i 4 {2! 5 i! 1,1+,22 7 ~3il 6 2 lt2i11 1 33 3 3 i2{ ~ 3 ii~l 3 3 1 2 2 1131 i3j 2 i3 2 2 1 2 2 111illi2!1 2 2 1 1 221 i2i 1 12 1 1 4 2 1 !4j 14~ 2 2 2 8 41713+ 3 3 2 lJljli 4 ' 21' !21 ± 2i213 3 4 !3j 8 7 4~ 14 Figure 1. Pedigree Structure and Genotypic Analyses of Five LQT Families Individuals showing the characteristic features of LQT, including prolongation of the QT interval and history of syncope, seizures, or aborted sudden death, are indicated by closed circles (females) or closed squares (males). Unaffected individuals are indicated by open circles or open squares. Individuals with an equivocal phenotype or those for whom phenotypic data are unavailable are shown as stippled. Circles or squares with a slash denote deceased individuals. Haplotypes for polymorphic markers linked to LQT2 are shown under each individual. These markers include (centromere to telomere) D78505, D78636, HERG 5-11, HERG 3-8, and D7S483 (Gyapay et al., 1994; Wang et al., submitted). Haplotypes cosegregating with the disease phenotype are indicated by a box. Recombination events are indicated with a horizontal black line. Informed consent was obtained from all individuals or from their guardians in accordance with local institutional review board guidelines. Haplotype analyses indicate that the LQT phenotype in these kindreds is linked to markers on chromosome 7q35-36. Results LQT2 Is Linked to Markers on Chromosome 7q35-36 To determine the relative frequency of the three known LQT loci (LQT1, LQT2, and LQT3), we performed linkage analyses in families with this disorder. Five LQT families were identified and phenotypically characterized (Figure 1). These families were unrelated and of varying descent, including Mexican (Spanish), German, English, and Dan- ish. In each case, an autosomal dominant pattern of inheri- tance was suggested by inspection of the pedigree. Af- fected individuals were identified by the presence of QT prolongation on electrocardiograms and, in some cases, a history of syncope or aborted sudden death. No patients had signs of congenital neural hearing loss, a finding asso- ciated with the rare autosomal recessive form of LQT, or other phenotypic abnormalities. Genotype analyses with polymorphic markers linked to the known LQT loci sug- gested that the disease phenotype in these families was linked to polymorphic markers on chromosome 7q35-36 (Figure 1). The maximum combined two-point Iod score for these five families was 5.13 at D78636 (~) = 0.0; Table 1). When combined with our previous studies (Jiang et al.,
`
`2 of 9
`
`PENN EX. 2067
`CFAD V. UPENN
`IPR2015-01836
`
`

`
`HERG Mutations Cause Long QT Syndrome 797 Table 1. Maximum Pairwise Lod Scores, Recombination Fractions for Linkage of LQT2 with HERG, and Polymorphic Markers on Chromosome 7 Families from Families Studied Present Study to Date LOCUS Z=,~ 0 Zm~ 0 D7S505 4.40 0.0 22.91 0.009 D7S636 5.13 0.0 26.14 0.00 HERG 3-8 0.11 0.0 6.34 0.00 HERG 5-11 3.55 0.0 9.64 0.00 D7S483 2.48 O.0 22.42 0.00 Markers are shown in chromosomal order (centromere to telomere) (Gyapay et al., 1994). The first column (families from present study) indicates combined Iod scores for the five families described in this study. The second column (families studied to date) indicates com- bined Iod scores from the five families studied here, as well as the nine families from our previous study (Jiang et al., 1994). Zm~ indicates maximum Iod score; e indicates estimated recombination fraction at Zmax, $1 $2 $3 Hitt4 NBD $4 $5 S 6 Intron 1 ~625bp Pore <- 4- 10 11 ~325bp NBD ,','! m--" ~1100bp Sca~p Figure 2. Partial Genomic Structure of HERG and Location of PCR Primers Used in This Study Regions encoding predicted membrane-spanning domains ($1-$6), the pore domain, and the nucleotide-binding domain (NBD) are indi- cated. The DNA sequences for the intron-exon boundaries are these: intron I, 5'-AGGAGgtgggg._ccccagCTGATC-3'; intron II, 5"TGG- CTgtgagt...ccccagCCCTC-3'; intron Ill, 5'-CCTGGgtatgg...ctccagG- GAAG-3'. 1994; Wang et al., submitted), the maximum combined two-point Iod score for the 14 chromosome 7-1inked fami- lies was 26.14, also at D7S636 (0 = 0.0; Table 1). Haplo- type analyses were consistent with our previous studies, placing LQT2 between D7S505 and D7S483 (Figure 1; Wang et al., submitted), localizing this gene to chromo- some 7q35-36. HERG Maps to Chromosome 7q35-36 HERG was previously mapped to chromosome 7 (Warmke and Ganetzky, 1994). To test the candidacy of this gene, we refined the localization of HERG using two physical mapping techniques. First, we mapped HERG on a set of yeast artificial chromosome (YAC) contigs constructed for chromosome 7 (Green et al., 1994). HERG was localized to the same YAC as D7S505, a polymorphic marker that was tightly linked to LQT2 (Table 1). Second, we mapped HERG to chromosome 7q35-36 using fluorescent in situ hybridization (FISH) with a P1 genomic clone containing HERG (data not shown). To determine whether HERG was genetically linked to the LQT locus, we used single strand conformation poly- morphism (SSCP) analyses to identify polymorphisms within HERG and performed linkage analyses in the chro- mosome 7-1inked families. Two aberrant SSCP conform- ers were identified in DNA samples from patients and con- trols using primer pairs 5-11 and 3-8 (Figure 2). These conformers were cloned and sequenced. One abnormal conformer resulted from a C to T substitution at position 3 of codon 489 (cDNA nucleotide 1467; observed hetero- zygosity of 0,37). The second abnormal conformer re- suited from an A to G substitution at position 3 of codon 564 (cDNA nucleotide 1692; observed heterozygosity of 0.44). Neither substitution affected the predicted amino acid sequence of HERG. HERG polymorphisms were used for genotypic analyses in chromosome 7-1inked families (see Figure 1; data not shown). No recombination events between HERG and LQT were identified in any of these families. The maximum combined Iod score for the 14 families was 9.64 (0 = 0.0; Table 1). These data indicate that HERG is completely linked to LQT2. HERG Intragenic Deletions Associated with LQT in Two Families To test the hypothesis that HERG is LQT2, we used SSCP analyses to screen for mutations in affected individuals. Since the genomic structure of HERG was unknown, oligo- nucleotide primer pairs were designed from published (Warmke and Ganetzky, 1994) HERG cDNA sequences (Figure 2; Table 2). In most cases, single products of ex- pected size were generated. For primer pairs 1-10, 6-13, and 15-17, however, products of greater than expected size were obtained, suggesting the presence of intronic sequences. To examine this possibility, we cloned and sequenced these larger products. DNA sequence analy- ses identified three introns at positions 1557/1558, 1945/ Table 2. HERG PCR Primers Name Position Sequence 1L 1147-1166 2 L 1291-1312 3 L 1417-1437 4 L Intronl 5 L 1618-1636 6 L 1802-1823 7 R 1446-1426 8 R 1527-1503 9 R Intronl 10 R 1643-1623 11R 1758-1736 12 R Intron II 13 R 2034-2016 16 R Intron III 15 L 2259-2278 14 L 2214-2233 17 R 2550-2529 GACGTGCTGCCTGAGTACAA TTCCTGCTGAAGGAGACGGAAG ACCACCTACGTCAATGCCAAC TGCCCCATCAACGGAATGTGC GATCGCTACTCAGAGTACG GCCTGGGCGGCCCCTCCATCAA CACCTCCTCGTTGGCATTGAC GTCGAAGGGGATGGCGGCCACCATG TACACCACCTGCCTCCTTGCTGA GCCGCGCCGTACTCTGAGTAG CAGCCAGCCGATGCGTGAGTCCA GCCCGCCCCTGGGCACACTCA CAGCATCTGTGTGTGGTAG GGCATTTCCAGTCCAGTGC CCTGGCCATGAAGTTCAAGA GCACTGCAAACCCTTCCGAG GTCGGAGAACTCAGGGTACATG All primers are shown in 5' to 3' direction. Sense strand oligonucleo- tides are indicated with an L and antisense oligonucleotides are indi- cated with an R. cDNA sequence was obtained from the GenBank data base; nucleotide numbering begins with the initiator methionine.
`
`3 of 9
`
`PENN EX. 2067
`CFAD V. UPENN
`IPR2015-01836
`
`

`
`Cell 798 A ® K2287 I --.m Normal ~.170 i0p <"14310p GATC TGGTTCCTCATCGACATGGTGGCCGCCATCCCCTTCGACCTGCTC w F L[~ D M v A A i P F]D L L B K2595 =~_o66 Normal Deletion 4- GATC GATC Normal GTCATCTAC~CGGCTG~TTCACA VTYTAVFT CCCTACTC~CT~CTTCCTGC~ PYSAAFLL Deletion GTCATCTACC~CTGTCTTCACAC VIYIRLSSH CCTACTCGGCTGCCTTCCTGC~ PTRLPSC.J Figure 3. HERG Intragenic Deletions Associated with LQT in Two Families (A) Pedigree structure of K2287, results of PCR amplification using primer pair 3-9, and the effect of the deletion on the predicted structure of HERG protein are shown. Note that an aberrant fragment of 143 bp is observed in affected members of this kindred, indicating the presence of a disease-associated intragenic deletion. DNA sequence of normal and aberrant PCR products defines a 27 bp deletion (AI500-F508). This mutation causes an in-frame deletion of nine amino acids in the third membrane-spanning domain ($3). Deleted sequences are indicated. (B) Pedigree structure of K2595 is shown. Deceased individuals are indicated by a slash. The results of SSCP analyses using primer pair 1-9 are shown beneath each individual. Note that an aberrant SSCP conformer cosegregates with the disease in this family. DNA sequence shows a single base pair deletion (A1261). This deletion results in a frameshift followed by a stop codon 12 amino acids downstream. The deleted nucleotide is indicated with an arrow. 1946, and 2398(cid:127)2399 of the cDNA sequence (Figure 2). These boundaries were confirmed by direct DNA sequenc- ing of HERG genomic clones containing HERG (data not shown). To facilitate SSCP analyses, we designed addi- tional primers to intronic sequences. As indicated previously, SSCP analyses using primer pair 3-8 identified an A to G polymorphism within HERG (cDNA nucleotide 1692). Analysis of kindred 2287 (1<2287) using this SSCP defined a pattern of genotypes consistent with a null allele (see Figure 1). Possible explanations for these findings included multiple misinheritances, a possi- bility not supported by our previous genotypic analyses, DNA sample errors, base pair substitutions, or a deletion. To test the hypothesis that the genotypic data were due to a small deletion, we repeated PCR analyses of K2287 using a new primer pair (3-9) flanking the previous set of primers. These experiments identified two products of 170 bp and 143 bp in affected members of K2287 (Figure 3A). By contrast, only a single product of 170 bp was observed in unaffected members of this kindred. Furthermore, only the 170 bp band was seen in DNA samples from more than 200 unaffected individuals (data not shown). The 143 bp and 170 bp products were cloned from affected individ- ual 11-2. Direct sequence analyses of the aberrant PCR product revealed the presence of a 27 bp deletion begin- ning at position 1498 (AI500-F508). This deletion disrupts the third membrane-spanning domain (S3) of HERG. To test further the hypothesis that HERG is LQT2, we continued SSCP analyses in additional kindreds, SSCP using the primer pair 1-9 identified an aberrant conformer in affected individuals of K2595 (Figure 3B). Analyses of more than 200 unaffected individuals failed to show this anomaly (data not shown). The normal and aberrant con- formers were cloned and sequenced, revealing a single base deletion at position 1261 (A1261). This deletion re- sults in a frameshift in sequences encoding the first mem- brane-spanning domain ($1), leading to a new stop codon within 12 amino acids. The identification of intragenic dele- tions of HERG in two LQT families suggests that HERG mutations can cause LQT. Three HERG Point Mutations Associated with LQT To identify additional HERG mutations in LQT2, we contin- ued SSCP analyses in linked kindreds and sporadic cases. Three aberrant SSCP conformers were identified in af- fected members of K1956, K2596, and K2015 (Figure 4). In each case, the normal and aberrant conformers were cloned and sequenced. In K1956, a C to T substitution at position 1682 was identified. This mutation results in substitution of valine for a highly conserved alanine at
`
`4 of 9
`
`PENN EX. 2067
`CFAD V. UPENN
`IPR2015-01836
`
`

`
`HERG Mutations Cause Long QT Syndrome 799 A K1956 4- Normal Missense 4-C/T B Normal K2596 4- Missense 4-TIC C K2015 Sporadic LQT NA QQ Normal Affected GATC GATC -C/G GATC GA TC GATC GATC GTG -~ V GAC -~ D GCG AAC \/ \/ L I AHWL D I L I NFR - $5 D • $2 • K1956 L I V H W L K2596 D I L I D F R H-Erg L I A H W L H-Erg D I L I N F R M-Eag L A A H W K M-Eag D I V L N F H R-Eag L A A H W M R-Eag D I V L N F H Eag L V A H W L Eag D I V L N F H Elk L A A H W L Elk D I L L N F R c 5'-CAT CCT GG // gtatggg-3' Figure 4. HERG Point Mutations identified in Three LQT Kindreds Pedigree structure of K1956 (A), K2596 (B), and K2015 (C) are shown. Below each pedigree, the results of SSCP analyses with primer pair 5-11 (K1956), primer pair 1-9 (K2596), and primer pair 14-16 (K2015) are shown. Aberrant SSCP conformers cosegregate with the disease in each kindred. DNA sequence analyses of the normal and aberrant conformers reveals a C to T substitution at position 1682 in K1956. This mutation results in substitution of valine for a highly conserved alanine residue at codon 561 (A561V). Analyses of K2596 reveals an A to G substitution at position 1408 (T to C substitution on the antisense strand is shown). This mutation results in substitution of aspartic acid for a conserved asparagine in the second transmembrane domain (N470D). Analyses of K2015 reveals a G to C substitution (C to G substitution on the antisense strand is shown). This mutation occurs inthe splice-donor sequence of intron III. Coding sequences are shown in uppercase and intronic sequences are lowercase. Note that the G to C substitution disrupts the splice-donor site (HERG, M-eag, elk [Warmke and Ganetzky, 1994]; R-eag [Ludwig etal., 1994]). codon 561 (A561V), altering the fifth membrane-spanning domain ($5) of the HERG protein (Figure 4A). In K2596, an A to G substitution was identified at position 1408. This mutation results in substitution of aspartic acid for a con- served asparagine at codon 470 (N470D), located in the second membrane-spanning domain ($2) (Figure 4B). In K2015, a G to C substitution was identified. This substitu- tion disrupts the splice-donor sequence of intron III, affect- ing the cyclic n ucleotide-binding domain (Figure 4C). None of the aberrant conformers were identified in DNA samples from more than 200 unaffected individuals (data not shown). De Novo Mutation of HERG in a Sporadic Case of LQT To substantiate that HERG mutations cause LQT, we used SSCP analysis to screen for mutations in sporadic cases. Primer pair 4-12 identified an aberrant conformer in af- fected individual II-1 of K2269 (Figure 5). This conformer was not identified in either parent or in more than 200 unaffected individuals. Direct DNA sequencing of the ab- errant conformer identified a G to A substitution at position 1882. This mutation results in substitution of serine for a highly conserved glycine at codon 628 (G628S), altering the pore-forming domain. Genotype analysis of this kin- dred using nine informative short tandem repeat polymor- phisms confirmed maternity and paternity. The identifica- tion of a de novo mutation in a sporadic case demonstrates that HERG is LQT2. HERG Is Expressed in the Heart HERG was originally identified from a hippocampal cDNA library (Warmke and Ganetzky, 1994). To determine the tissue distribution of HERG mRNA, we isolated partial cDNA clones and used them in Northern blot analyses. Northern blot analyses showed strongest hybridization to heart mRNAs, with faint signals in brain, liver, and pan- creas (Figure 6). Nonspecific hybridization was also seen in lung, possibly due to genomic DNA contamination. The size of the bands observed in cardiac mRNA was consis-
`
`5 of 9
`
`PENN EX. 2067
`CFAD V. UPENN
`IPR2015-01836
`
`

`
`Cell 800 K2269 Normal Missense ~-C/T GATC GATC AGC -~ S GGC \/ S VG F GNV S I Pore =' K2269 SVGFSNVS H-Erg SVGFGNVS M-Eag SVGFGNIA R-Eag SVGFGNIA Eag SVGFGNVA Elk SVGFGNVS Shaker TVGYGDMT Figure 5. De Novo Mutation of HERG in a Sporadic Case of LQT Pedigree structure of K2269 and SSCP analyses (primer pair 14-16) showing an aberrant conformer in a sporadic case of LQT. DNA se- quence analyses identified a G to A substitution at position 1882 of the cDNA sequence (C to T substitution on the antisense strand is shown). Note that this mutation results in the substitution of a serine for a highly conserved glycine residue at codon 628 (G628S). This amino acid sequence is known to be critical for potassium ion selec- tivity. tent with the predicted size of HERG. Two bands of -4.1 and 4.4 kb were identified, possibly due to alternative splic- ing or the presence of a second related mRNA. These data indicate that HERG is strongly expressed in the heart, consistent with its involvement in LQT. Discussion We conclude that mutations in HERG cause the chromo- some 7-1inked form of LQT. Several lines of evidence sup- port this conclusion. First, we used linkage analyses to map an LQT locus (LQT2) to chromosome 7q35-36 in 14 families. Second, we used physical and genetic mapping to place HERG in the same chromosomal region as LQT2. Third, we demonstrated that HERG is expressed in the heart. Fourth, we identified intragenic deletions of HERG associated with LQT in two families. Fifth, we identified four HERG point mutations in LQT patients. Finally, one of the point mutations arose de novo and occurs within a highly conserved region encoding the potassium-selective pore domain. Our data suggest a likely molecular mechanism for chro- mosome 7-1inked LQT. Although the function of HERG is "G :~ ~ ~. =, ~ ~ ~:a. KB --9.5 --7.5 --4.4 --2.4 -- 1.35 Figure 6. Northern Blot Analysis of HERG mRNA Showing Strong Ex- pression in the Heart A Northern blot (poly[A] ÷ RNA, 2 p~g/lane; Clontech) was probed using an HERG cDNA containing nucleotides 679-2239 of the coding se- quence. Two cardiac mRNAs of ~ 4.1 and 4.4 kb are indicated. Back- ground in mRNA extracted from lung was high, but no specific bands were identified. not yet known, analyses of its predicted amino acid se- quence indicates that it encodes a potassium channel subunit Potassium channels are formed from four ~ sub- units (MacKinnon, 1991), either as homo- or heterotetra- mers (Covarrubias et al., 1991). These biophysical obser- vations suggest that combination of normal and mutant HERG ~ subunits could form abnormal HERG channels. This raises the possibility that HERG mutations have a dominant negative affect on potassium channel function. The mutations that we identified are consistent with a dominant negative mechanism (Figure 7). Two mutations result in premature stop codons and truncated proteins (A1261 and the splice-donor mutation). In the first case, only the amino terminus and a portion of the first mem- brane-spanning domain ($1) remain. In the second, the carboxyl end of the protein is truncated, leaving all mem- brane-spanning domains intact. HERG contains a cyclic nucleotide-binding domain near the carboxyl terminus, and in both mutations this domain is deleted. In another mutation, an in-frame deletion of nine amino acids disrupts the third membrane-spanning domain (AI500-F508). Two missense mutations also affect membrane-spanning do- mains, A561V in the $5 domain and N470D in S2. Both mutations affect amino acids conserved in the eag family of potassiu m channels and likely alter the secondary struc- ture of the protein. The de novo missense mutation G628S occurs in the pore-forming domain. This domain is highly conserved in all potassium channel a subunits. This muta- tion affects a conserved amino acid that is of known impor- tance for ion selectivity. When this substitution was intro- duced into Shaker H4, potassium ion selectivity was lost (Heginbotham et al., 1994). As discussed above, these mutations could induce the loss of HERG function. Our data have implications for the mechanism of ar- rhythmias in LQT. Two hypotheses for LQT have pre- viously been proposed (Schwartz et al., 1995). One sug- gests that a predominance of left autonomic innervation causes abnormal cardiac repolarization and arrhythmias. This hypothesis is supported by the finding that arrhyth-
`
`6 of 9
`
`PENN EX. 2067
`CFAD V. UPENN
`IPR2015-01836
`
`

`
`HERG Mutations Cause Long QT Syndrome 801 S1 S2 $3 $4 $5 Pore $6 Extracel[ular Frameshift Intracellular domain Splice error Figure 7. Schematic Representation of the Predicted Topology of the Protein Encoded by HERG and the Location of LQT-Associated Muta- tions For simplicity, not all amino acids are shown. SSCP analyses identified HERG mutations in 4 of the 14 chromosome 7-linked kindreds. mias can be induced in dogs by removal of the right stellate ganglion. In addition, anecdotal evidence suggests that some LQT patients are effectively treated by 13-adrenergic blocking agents and by left stellate ganglionectomy (Schwartz et al., 1995). The second hypothesis for LQT- related arrhythmias suggests that mutations in cardiac- specific ion channel genes (or genes that modulate cardiac ion channels) cause delayed myocellular repolarization. Delayed myocellular repolarization could promote reacti- vation of L-type Ca 2÷ channels, resulting in secondary de- polarizations (January and Riddle, 1989). These second- ary depolarizations are the likely cellular mechanism of torsade de pointes arrhythmias (Surawicz, 1989). This hy- pothesis is supported by the observation that pharmaco- logic block of potassium channels can induce QT prolon- gation and repolarization-related arrhythmias in human and animal models (Antzelevitch and Sicouri, 1994). The discovery that one form of LQT results from mutations in a cardiac potassium channel gene supports the myocellular hypothesis. The presence of a cyclic nucleotide-binding domain in HERG suggests a mechanism for the link between altered autonomic nervous activity and arrhythmias in LQT. 13-Adrenergic receptor activation increases intracellular cAMP and enhances L-type Ca 2÷ channel function. Cyclic AMP may also activate HERG, thereby increasing net out- ward current and accelerating the rate of myocellular repo- larization. Dominant negative mutations of HERG might interrupt the normal modulation of HERG function by cAMP, thereby permitting a predominant effect on L-type Ca 2+ channel function. The resulting imbalance would in- crease the likelihood that enhanced sympathetic tone could induce Ca 2÷ channel-dependent secondary depolar- izations, the probable cellular mechanism of torsades de pointes. 13-Adrenergic blocking agents could act by inter- rupting the effect of cAMP on L-type Ca 2÷ channels, possi- bly explaining the beneficial effects of 13 blockers in some LQT patients. The relative frequency of the three LQT loci is not yet known. In this study, we identified five new families with autosomal dominant LQT, and all were linked to chromo- some 7. This brings the total number of chromosome 7-1inked families to 14. To date, we have linked seven families to chromosome 11 (LQT1), 14 families to chromo- some 7 (LQT2), and three families to chromosome 3 (LQT3), and three families remain unlinked (Keating et al., 1991a, 1991b; Jiang et al., 1994). Although preliminary, these data suggest that LQT2 is a common form of inher- ited LQT. This work may have important clinical implications. Re- cently, presymptomatic diagnosis has been possible in large families using linkage analysis. Most cases of LQT are sporadic, and therefore genetic testing using linkage analysis is not feasible. Continued mutational analyses of LQT2 will facilitate genetic testing for this form of LQT. Identification and characterization of genes responsible for other forms of LQT will be necessary for the develop- ment of generalized diagnostic tests. Improved diagnostic capacity may enable rational therapy. For example, chro- mosome 7-1inked LQT patients may respond to potassium channel activators, like pinacidil. In summary, our results provide the molecular mecha- nism of a life-threatening cardiac arrhythmia. Future ex- periments will define the specific function of HERG and its role in LQT. In addition, we will continue to search for LQT genes; these genes will provide further insight into cardiac repolarization and repolarization-related arrhythmias. Experimental Procedures Identification and Phenotyping of LQT Kindreds LQT kindreds were ascertained from medical clinics throughout North America. Phenotypic criteria were identical to those used in our previ- ous studies (Keating et al., 1991a, 1991b; Keating, 1992; Curran et al., 1993a; Jiang et al., 1994). Individuals were evaluated for LQT based on the QT interval corrected for heart rate (QTc; Bazette, 1920) and for the presence of syncope, presence of seizures, and aborted sudden death. Informed consent was obtained from all individuals or from their guardians in accordance with local institutional review board guidelines. Phenotypic data were interpreted without knowledge of genotype, Genotypic and Linkage Analyses Amplification and detection of microsatellite markers were performed as described previously (Curran et al., 1993a; Jiang et al., 1994). Poiy- morphic markers used in this study were D7S483, D7S636, and D7S505 (Gyapay et al., 1994). Pairwise linkage analysis was per- formed using the MLINK program of the LINKAGE software package (Lathrop et al., 1985). In accordance with our previous studies, we assumed a penetrance of 0.90 and a disease gene frequency of 0.001. Gene frequency between males and females was assumed to be equal. Isolation of HERG Genomic and cDNA Clones HERG probes were generated using the products of PCRs with human genomic DNA and primer pairs 1-10, 6-13, and 15-I 7. These products were cloned, radiolabeled to high specific activity, and used to screen a human genomic P1 library (Sternberg, 1990). Positive clones were purified, characterized, and used for FISH and DNA sequence analy- ses. To isolate HERG cDNA clones, genomic probes containing HERG coding sequences were used to probe 106 recombinants of a human hippocampal cDNA library (Stratagene). A single clone containing -2.2 kb of HERG coding sequenc