Author + information
- Received April 12, 2004
- Revision received June 3, 2004
- Accepted June 7, 2004
- Published online September 15, 2004.
- Li Zhang, MD*,
- G. Michael Vincent, MD, FACC*,†,* (, )
- Marco Baralle, PhD‡,
- Francisco E. Baralle, MD, PhD‡,
- Blake D. Anson, PhD§,
- D. Woodrow Benson, MD, PhD∥,
- Bryant Whiting, BA*,
- Katherine W. Timothy, BS†,
- John Carlquist, PhD*,†,
- Craig T. January, MD, PhD, FACC§,
- Mark T. Keating, MD¶ and
- Igor Splawski, PhD¶
- ↵*Reprint requests and correspondence:
Dr. G. Michael Vincent, Department of Medicine, LDS Hospital, 324 10th Avenue, Suite 130, Salt Lake City, Utah 84103
Objectives The purpose of this research was to determine whether an intronic variant (T1945+6C) in KCNH2is a disease-causing mutation, and if expanded phenotyping criteria produce improved identification of long QT syndrome (LQTS) patients.
Background Long QT syndrome is usually caused by mutations in conserved coding regions or invariant splice sites, yet no mutation is found in 30% to 50% of families. In one such family, we identified an intronic variant in KCNH2. Long QT syndrome diagnosis is hindered by reduced penetrance, as the long QT phenotype is absent on baseline electrocardiogram (ECG) in about 30%.
Methods Fifty-two family members were phenotyped by baseline QTc, QTc maximum on serial ECGs (Ser QTc-max), and on exercise ECGs (Ex QTc-max) and by T-wave patterns. Linkage analysis tested association of the intronic change with phenotype. The consequences of T1945+6C on splicing was studied using a minigene system and on function by heterologous expression.
Results Expanded phenotype/pedigree criteria identified 23 affected and 29 unaffected. Affected versus unaffected had baseline QTc 484 ± 48 ms versus 422 ± 20 ms, Ser QTc-max 508 ± 48 ms versus 448 ± 10 ms, Ex QTc-max 513 ± 54 ms versus 444 ± 11 ms, and LQT2 T waves in 87% versus 0%. Linkage analysis demonstrated a logarithm of odds score of 10.22. Splicing assay showed T1945+6C caused downstream intron retention. Complementary deoxyribonucleic acid with retained intron 7 failed to produce functional channels.
Conclusions T1945+6C is a disease-causing mutation. It alters KCNH2splicing and cosegregates with the LQT2 phenotype. Expanded ECG criteria plus pedigree analysis provided accurate clinical diagnosis of all carriers including those with reduced penetrance. Intronic mutations may be responsible for LQTS in some families with otherwise negative mutation screening.
Hereditary long QT syndrome (LQTS) is caused by over 250 mutations in five genes, four encoding potassium channel subunits KCNQ1(KVLQT1, LQT1), KCNH2(HERG, LQT2), KCNE1(minK, LQT5), KCNE2(MiRP1, LQT6), and the cardiac sodium channel gene SCN5A(LQT3) (1). Mutations of ANKB(2) and KCNJ2(3) have been reported to cause LQT4 and LQT7, respectively. Our recent reports on the ANKBand KCJN2disorders (4,5) have shown that the mean QTc is within the normal range, the majority of patients have normal to borderline QT intervals, and patients in each disorder have other clinical features that are distinctly different from LQT1 to 3. Thus, classification of ANKBand KCNJ2mutations as LQTS disorders is uncertain and remains to be fully elucidated. Genetic screens for mutations in coding regions or flanking intronic sequences of the five LQTS genes have been performed frequently during the past decade. Currently, even after improvements in sequence detection technology, a mutation cannot be identified in about 30% of LQTS families. Many of these patients exhibit characteristic T-wave patterns of the known LQTS genotypes (6), prompting us to believe that mutations in noncoding regions of the known genes may be the cause of LQTS in some of these families.
One such Caucasian family from the LDS Hospital LQTS program was the stimulus for this study. In 1994, the 27-year-old proband experienced a cardiac arrest triggered by activity and excitement, with subsequent documented torsade de pointes (Fig. 1).He had no prior symptoms, but siblings had experienced recurrent syncope. Electrocardiogram (ECG) screening of the nuclear family showed QTc prolongation and typical LQT2 T-wave patterns (Fig. 2)(6) in several members, suggesting a KCNH2mutation. Mutation analysis was performed in one affected family member, but failed to find a sequence variation in the coding sequences of any of the five known LQTS genes. The only sequence variant identified was a single nucleotide substitution, T1945+6C, at the 5′ splice (donor) site of intron 7 in KCNH2. This substitution was not found in 320 ethnically matched normal control subjects.
An analysis of 400 vertebrate genes (7) has yielded consensus sequences near intron-exon boundaries that are essential for correct splicing. The 10 base pair (bp) (−4 to +6) including the GT at the 5′ start of the intron constitute the 5′ (donor) splice site. The G and T at positions +1 and +2, respectively, are highly conserved (100%), and, consequently, sequence alterations at these locations are readily recognized as mutations. Several adjacent nucleotides at the splice donor site are less well-conserved, and the nucleotide at position +6 is a T in ∼50% of the introns.
Intronic KCNH2mutations in LQTS have been infrequently reported. For example, Splawski et al. (1) reported only three (∼4%) such mutations in an analysis of 72 KCNH2mutations. These three mutations altered obligatory (∼100% conservation) nucleotides within donor or acceptor splice sites and, in the absence of functional analysis, were assumed to alter transcript splicing. By contrast, T1945+6C changes a less highly conserved (∼50%) intronic nucleotide of a donor splice site and could not be assumed to alter transcript splicing. Therefore, to determine whether this is a disease-causing mutation or a simple polymorphism, we expanded the pedigree, used enhanced ECG and pedigree criteria for phenotyping, tested the association of genotype with phenotype using linkage analysis, and elucidated the splicing effect and molecular mechanism of the T1945+6C through well-established cellular splicing assay and heterologous expression studies.
A five-generation pedigree was constructed, and 52 blood-related family members were evaluated by history and ECG (Fig. 3).History evaluation included LQTS-related syncope, cardiac arrest, and sudden death. Electrocardiogram analysis was from supine, resting ECGs, and included baseline QTc interval (Bazett's formula) and T-wave pattern (6) on all subjects. During follow-up, we obtained ECGs at yearly clinic visits and at other times as indicated by clinical needs. Twenty-two members received additional ECG evaluation, either bicycle exercise testing (n = 18) and/or serial ECGs (n = 19) (≥2 ECGs/person). From these, the maximum QTc during exercise or recovery (Ex QTc-max) and on serial ECGs (Ser QTc-max) was evaluated. None of the patients was on beta-blockers at time of initial ECG data acquisition, and the phenotyping was performed without knowledge of genotype.
Family members were determined to be phenotypically affected if they met one of the following criteria: 1) a diagnostic QTc interval (≥470 ms in males or ≥480 ms in females) (8) with or without the presence of an LQT2 T-wave pattern, on baseline, Ex QTc-max, or Ser QTc-max ECGs; 2) a QTc of 460 ms with an LQT2 T-wave pattern if other family member(s) met the diagnostic criterion 1; and 3) identification as an obligate gene carrier on family pedigree; members could be classified as an obligate gene carrier if offspring plus other blood relative(s) showed a definite LQT2 phenotype (Fig. 3).
Informed consent (approved by and in accordance with the guidelines of the Institutional Review Board at LDS Hospital of Intermountain Health Care, Inc.) was obtained from all participants before genetic testing.
DNA was collected from buccal swabs and extracted using the QIAamp DNA Mini Kit (Qiagen, Valencia, California). The region containing the T1945+6C intronic variant was amplified at LDS Hospital using previously published oligonucleotides (9). Samples were sequenced by ARUP Laboratories (Associated Regional University Pathologists, Salt Lake City, Utah) using the Applied Biosystems 377 (Foster City, California) deoxyribonucleic acid sequencer with Big Dye Terminator chemistry. The results were read by a single investigator blinded to the phenotype data.
Statistical and linkage analyses
The QTc on baseline ECG, the Ser QTc-max, and the Ex QTc-max were compared between affected and unaffected members using the Mann-Whitney test. The phenotype-genotype agreement was measured as a Kappa statistic. The statistical analyses were performed using SPSS 10.1 for Windows (SPSS Inc., Chicago, Illinois). Using KCNH2-T1945+6C as an allele, we performed linkage analysis. A two-point logarithm of odds (LOD) score was calculated using LIPED software(Rockefeller University, New York, New York) (10), assuming an allele frequency of 0.1% and disease penetrance of 95%.
Functional splicing assays
Normal and mutated (T1945+6C change) sequences of the KCNH2gene were amplified from genomic deoxyribonucleic acid of an individual heterozygous for the mutation using forward primer 5′-ggaattccatatggaatcactgcacctgtcagtgc-3′ and reverse primer 5′-ggaattccatatggaattgtacatctgcgctccagc-3′. The amplification generated a fragment that contained exon 7 with 248 bp of 5′ and 202 bp of 3′ intronic flanking sequences. Both oligonucleotides carry an Nde1 restriction enzyme site in their 5′ ends that was used to clone the product into a modified version of the alpha-globin-fibronectin EDB minigene (11). The splicing assay was performed by transfecting 0.5 μg of each minigene plasmid into 3 × 105He La cells with Qiagen Effectene transfection reagents. Ribonucleic acid extraction and reverse transcription-polymerase chain reaction (RT-PCR) analysis were performed as previously described (11). All of the fragments shown in Figure 4(panel B) were cloned and sequenced to confirm their identity and exclude polymerase errors.
The parental U1 snRNA was pG3U1 (WT-U1), a derivative of PHU1 (12). The variant A>G U1 snRNA, complementary to the nucleotide change observed in the patient, was created by replacing the U1 snRNA 5′ sequence between the sites Bcl1 and BglII in the pG3U1 plasmid with the mutant oligonucleotides. The effect of T1945+6C mutation on ribonucleic acid processing was studied using a well-documented hybrid minigene system (11,13,14).
Functional expression assays
To study the effects of T1945+6C, human embryonic kidney (HEK-293) cells were transiently transfected with either KCNH2-WT cDNA or KCNH2complementary deoxyribonucleic acid containing T1945+6C and intron 7 (KCNH2-T>C). Green fluorescent protein (GFP) was cotransfected with each complementary deoxyribonucleic acid as a marker for successful transfection. The KCNH2-WT cDNA expression vector has been described previously (15), and GFP was expressed with the mammalian expression vector pRK-5 (Clonetech, Mansfield, United Kingdom). The KCNH2-T>C expression vector was generated by amplifying T1945+6C genomic deoxyribonucleic acid, restricting at unique Bgl II and Xho I restriction sites in exons 7 and 8, and subsequently ligating into the WT expression vector. All amplified sequences were confirmed to be error-free through Big Dye sequencing. Current levels were measured from GFP-expressing cells 24 to 48 h after transfection using tight-seal whole-cell voltage clamp recording techniques. Peak currents were assessed as peak tail current (Itail) (Fig. 5,arrows) evoked by depolarizing cells from a holding potential of −80 mV to 20 mV for 5 s followed by repolarization to −50 mV for 2.85 s. Details for both the transfection and recording techniques have been described in detail previously (16).
Phenotype was characterized in the 52 blood-related family members based on medical history, ECG assessment, and pedigree analysis. Twenty-three individuals were considered affected and 29 unaffected (Fig. 3). The QTc on baseline ECG, Ex QTc-max, and the Ser QTc-max were all significantly longer in affected versus nonaffected members (Table 1).A total of 87% of affected members presented typical LQT2 patterns in baseline ECG (Table 1, Fig. 2). No significant gender differences were found in the QTc interval and the T-wave patterns among affected members.
A total of 65% (15 of 23) of affected members were correctly diagnosed by baseline QTc (503 ± 46 ms, range 460 to 610 ms) and the presence of typical LQT2 T-wave patterns in 15 of 15. The other 35% (8 of 23) had a QTc of 449 ± 16 ms (420 to 460 ms) and were classified as “uncertain,” as these values overlap with those of normals. The LQTS diagnosis was confirmed in six of the eight by follow-up assessment: Ser QTc-max (3 ± 1 ECGs/person over 1.3 ± 1.9 years of 467 ± 10, range 460 to 480 ms), and Ex QTc-max (482 ± 11 ms, range 470 to 490 ms) (Fig. 4). All six met diagnostic criterion #2, typical T-wave pattern coupled with a QTc of 460 ms or greater, and most met criterion #1. The remaining two members (II-9 and III-21) (Fig. 3), an 86-year-old female and her daughter, both with QTc of 460 ms and absence of an LQT2 T-wave pattern, were identified as obligate mutation carriers by pedigree analysis. Altogether, therefore, 23 affected members were identified, 21 by the expanded ECG phenotype criteria and 2 as obligate carriers by pedigree analysis.
Sudden death of uncertain cause had occurred in 8.7% (2 of 23) before the LQTS diagnosis in the family, an 18-year-old male (II-3) (Fig. 3) and a 59-year-old female (III-16) (Fig. 3). Four (17%) affected members had syncope, and one had resuscitated cardiac arrest (IV-20) (Fig. 1), before treatment with beta-blockers.
Sequence analyses revealed 22 T1945+6C carriers and 29 noncarriers. An obligate carrier (II-1) (Fig. 3) was phenotyped as affected by baseline ECG (QTc 490 ms with an LQT2 T-wave pattern), but he died at age 85 before genetic testing was available.
Genotype-phenotype correlation: Advantage of the expanded QTc criteria and pedigree analysis
The T1945+6C variant cosegregated with affected status. Linkage analysis yielded a maximum two-point LOD score of 10.22 at recombination fraction θ = 0.0, indicating odds of >1010that the T1945+6C is linked to the disease phenotype.
Phenotype-genotype correlation showed concordance in 21 of 23 carriers and 29 of 29 noncarriers (Kappa = 0.96). Two obligate carriers were identified by pedigree analysis so that the combination of expanded ECG phenotyping criteria plus identification of obligate carriers revealed all 23 mutation carriers (Fig. 3).
Functional effect of the intron 7 mutation
The effect of T1945+6C on RNA processing was studied using a well-tested and documented hybrid minigene system (11,13,14) (Fig. 5, panel A). After transfection and expression of the construct in He La cells, the mRNA derived from the minigene was analyzed for the KCNH2exon 7 splicing pattern by RT-PCR. As shown in Figure 5(panel B, lanes exon 7 WT and exon 7 T>C), the T1945+6C dramatically affected pre-mRNA processing. In contrast with wild-type, T1945+6C processing resulted in three species of RNA that were unambiguously identified by cloning and complete sequencing. The major band (1,302 bp) shows retention of the intronic sequences downstream of the KCNH2exon 7 in the mature messenger ribonucleic acid. A lower level of normally spliced mRNA (637 bp) is also present as well as an additional faint band (239 bp) that results from the skipping of the exon.
Previous studies on a similar mutation in the NF-1 gene (14) showed simple exon skipping but not intron retention. However, our hypothesis was that the same basic molecular mechanism was involved, that is, the lack of recognition of the 5′ splice site in the mutant gene due to an imperfect base pairing with the U1 snRNP that is required for normal intron-exon splicing. Figure 4, panel C, is a schematic illustrating the decrease in complementarity between the mutant and U1 snRNA when compared with wild-type sequence. To functionally test this hypothesis, we simultaneously introduced into the cells the T1945+6C minigene with a variant U1 snRNA complementary to the mutation (Fig. 5, panel C). This resulted in the almost complete rescue of the correct splicing pattern (Fig. 5, panel B, lane exon 7 T>C + U1 A>G).
The effect of intron retention on KCNH2current was examined through heterologous expression of KCNH2-WT and KCNH2-T>C cDNA in HEK-293 cells. Representative current tracings from KCNH2-WT or KCNH2-T>C transfected cells are shown in Figure 6(upper and lower tracings, respectively). All recordings from cells transfected with KCNH2-WT cDNA produced normal KCNH2currents with a mean peak Itailof 938.3 ± 157.6 pA (n = 15) while none of the cells transfected with KCNH2-T>C complementary deoxyribonucleic acid produced currents above the basal noise level (7.9 ± 2.6 pA; n = 20). Thus, complementary deoxyribonucleic acid with a retained intron 7 failed to produce functional channels.
These findings demonstrate that the T1945+6C sequence change in intron 7 of KCNH2is a mutation that causes LQTS with a typical LQT2 phenotype. The functional cellular splicing assay demonstrated that T1945+6C produces a major transcript that retains intron 7. Two minor transcripts include a WT transcript and a transcript resulting from skipping of exon 7. Protein synthesized from the transcript containing intron 7 would be expected to undergo a translational frame-shift at amino acid residue 649 and prematurely truncate 21 amino acid residues further downstream. This would drastically alter the protein's 6th transmembrane domain, completely delete the C-terminus, and most likely render the protein nonfunctional. Whole cell recordings from HEK-293 cells transfected with KCNH2WT or T>C complementary deoxyribonucleic acid support this conclusion. Although speculative, coassembly of mutant and WT subunits could further inhibit the activity of the residual correctly spliced protein as well as protein synthesized from the other allele. Our results also demonstrated that the impaired splicing of T1945+6C was, at least in part, due to imperfect U1 snRNP recognition of the mutant 5′ splice site. This conclusion is supported by the fact that an artificial variant of the U1 snRNA, complementary to the mutant sequence, was able to rescue the correct pre-messenger ribonucleic acid processing.
Mutations in the GT splice donor and AG splice acceptor sites in LQTS genes have been infrequently reported (1,17–20). Because of the high conservation of these consensus sequences, such mutations have been assumed to alter splicing, but, until now, this hypothesis has remained untested. By expanding the genotypic and phenotypic analyses performed in the proband to include a large kindred of 52 individuals, we show that T1945+6C cosegregrates with the LQT2 phenotype. Absence of this nucleotide substitution in 320 ethnically matched control individuals further suggests that it is disease-causing. We have also clearly defined the defective splicing pattern of T1945+6C and provided a mechanism for the molecular dysfunction, making this study the first demonstration that incorrect intron-exon splicing of pre-messenger ribonucleic acid transcripts may result in LQT2. Furthermore, as T1945+6C is a less highly conserved (∼50%) nucleotide position than either the obligatory GT or AG of the donor and acceptor splice sites, this study clearly indicates that mutations of intronic nucleotides other than invariant splice donor and acceptor sites can cause LQTS. Of the 30% of LQTS families in which no mutation has been found with exon screening, the majority have characteristic T-wave patterns of the known genotypes (6). Consequently, in light of our results, it is reasonable to suggest that some of these families may have intronic mutations in the known LQTS genes rather than defects in unidentified genes.
At present, the function of introns is less well-understood than that of exons. Introns are, on average, much larger than exons and constitute about 25% of the human genome (21). They may harbor promoters or other regulatory modules that modify gene expression and might influence penetrance and expressivity of phenotypes (13,22–28). Thus, it may be that, in some cases, intron variations contribute to the phenotypic heterogeneity of LQTS (8,29,30).
The other important finding of this study is the high accuracy of diagnosis of affected family members using the enhanced QTc and T-wave morphology ECG criteria plus pedigree analysis. The phenotypic heterogeneity and reduced penetrance of the QT phenotype in LQTS (8,29,30) can impede accurate clinical diagnosis, leading to a missed diagnosis in about 30% of LQTS patients by baseline ECG. The enhanced ECG criteria and pedigree analyses used in this study correctly identified all the T1945+6C carriers, including those who had baseline normal-to-borderline QTc intervals of 400 to 460 ms, values that overlap those of normals. Large scale studies regarding the diagnostic value of these enhanced criteria in LQTS patients with nondiagnostic QTc intervals are currently underway. In addition to the usefulness of typical LQT2 T-wave patterns for increased diagnostic accuracy, these patterns are also very accurate for genotype prediction (6). The high accuracy provides a strategy for deoxyribonucleic acid screening by indicating which gene to first examine, thus reducing monetary and time costs.
The authors are grateful to Jay W. Mason, MD, and Robert L. Lux, PhD, for their support in genotyping some family members.
Sources of funding: LDS Hospital/Deseret Foundation grant DF400, Salt Lake City, Utah; Telethon Onlus Foundation grant E1038, Trieste, Italy; NHLBI grant R01 HL60723, Madison, Wisconsin; NIH grants R01 HL46401 and SCOR HL52338, Salt Lake City, Utah.
- Abbreviations and acronyms
- Ex QTc-max
- maximum QTc value during the exercise test, either exercise or recovery
- long QT syndrome
- second described variant of LQTS, due to mutations of the KCNH2(HERG) gene
- reverse transcription-polymerase chain reaction
- Ser QTc-max
- maximum QTc value among serial electrocardiograms during follow-up
- Received April 12, 2004.
- Revision received June 3, 2004.
- Accepted June 7, 2004.
- American College of Cardiology Foundation
- Splawski I.,
- Shen J.,
- Timothy K.W,
- et al.
- Zhang L.,
- Probst V.,
- Marec H.L,
- et al.
- Zhang L.,
- Tristani-Firouzi M.,
- Ptacek L.J,
- et al.
- Zhang L.,
- Timothy K.W.,
- Vincent G.M,
- et al.
- Pagani F.,
- Buratti E.,
- Stuani C,
- et al.
- Lund E.,
- Dahlberg J.E.
- Baralle M.,
- Baralle D.,
- del Conte L,
- et al.
- ↵Anson BD, Ackerman MJ, Tester DJ, et al. Molecular and functional characterization of common polymorphisms in HERG (KCNH2) potassium channels. Am J Physiol Heart Circ Physiol 2004;286:H2434–41.
- Berthet M.,
- Denjoy I.,
- Donger C,
- et al.
- Murray A.,
- Donger C.,
- Fenske C,
- et al.
- Castresana J.
- Himes S.R.,
- Tagoh H.,
- Goonetilleke N,
- et al.
- D'Souza I.,
- Schellenberg G.D.
- Fontemaggi G.,
- Gurtner A.,
- Strano S,
- et al.
- Rowntree R.K.,
- Vassuax G.,
- McDowell T.L,
- et al.
- Priori S.G.,
- Napolitano C.,
- Schwartz P.J.