The present sample comprised lymphoblastoid cell lines of 29 individuals and brain samples (frontal cortex) of another 14 individuals. All individuals were Central Europeans. Anonymized brain samples were collected during pathomorphological diagnostics and immediately shock-frozen in liquid nitrogene. Lymphoblastoid cell lines were generated by Epstein-Barr virus transformation of peripheral blood leucocytes propagated according to standard procedures . We extracted gDNAs using standard procedures. Total RNAs were extracted using TRIzol Reagent (Invitrogen). cDNAs were synthesized from total RNAs with random primers and SuperScript III (Invitrogen) following the supplier's protocol. Use of the brain samples was in accordance with the Helsinki Declaration and approved by the local ethics committee [Aerztekammer Rheinland-Pfalz, Decisions 837.103.04 (4261) and 837.073.07 (5608)].
Direct sequencing of human samples
For direct sequencing of gDNAs, PCR was carried out using FastStart Taq DNA Polymerase (Roche) and normal hot start PCR. The primers used bound to exon 3, in particular -18 and +26 bp upstream and downstream of start and stop codon, respectively (forward primer: 5'-TGCCTGGGAAGTTTGAGCTG-3'; reverse primer: 5'-CAGGTTGCCAGGCAGGATG-3'). PCR conditions were as follows: 2 min, 94°C; 35× (40 s at 94°C, 40 s at 65°C, 40 s at 72°C); 5 min, 72°C. Following exonuclease I and shrimp alkaline phosphatase digestion, dye terminator cycle sequencing of the PCR products was performed using the PCR primers and the CEQ DTCS Quick Start Kit (Beckman Coulter). Sequencing products were separated on a Beckman Coulter CEQ 8000 Genetic Analysis System. SCF-files were analyzed by visual inspection using BioEdit .
Quantitative Allele-Specific Expression Analysis of human samples
To more accurately measure the A versus G allele ratios of the A112G (Gly38Ser) variant in gDNAs and cDNAs, Quantification of Allele-Specific Expression by Pyrosequencing (QUASEP) was used . QUASEP allows to accurately quantify the relative amount of one allele to the other, if their sequences differ in at least one base pair. The primer sequences for QUASEP were designed using the Pyrosequencing Assay Design Software (Biotage AB).
PCR products for QUASEP-analyses of the KCNE1 -A/G-SNP rs1805127 were generated from gDNAs and cDNAs of lymphoblastoid cell lines and brain samples using FastStart Taq DNA Polymerase (Roche) and normal hot start PCR with the forward primer 5'-AGAGGGCCTCCAGCTTGC-3' and 5' biotinylated reverse primer 5'-GCAGGGTGGCAACATGTC-3' according to standard protocols. Pyrosequencing was performed on a PSQ™96MA Pyrosequencing System (Biotage) with the PyroGold SQA reagent kit (Biotage) using the forward sequencing primer 5'-CCAGCTTGCCGTCAC-3'. The PSQ™96MA 2.1.1 software (Biotage) was used for data analysis.
In order to assess whether RNA extraction or specific conditions of some of the samples affect the results of QUASEP-analyses of the KCNE1 -SNP rs1805127 we performed analogue QUASEP-analyses of the highly heterozygous A/G-SNP rs11254413 of the housekeeping gene TRDMT1. Hot start PCR was carried out using the forward primer 5'-TGGCTATCCTCTACAAAATGACAA-3' and 5' biotinylated reverse primer 5'-CGGCAGGGTGATATGACTGAT-3'. For pyrosequencing we took the forward sequencing primer 5'-CCTTGGGAGAATATCTAGAA-3'. QUASEP of TRDMT1 -SNP rs11254413 was confined to gDNAs and cDNAs of lymphoblastoid cell lines.
To assess the evolution of KCNE1 across Eutheria, we complemented the human KCNE1 data with gDNA and cDNA data from non-human species (Mammalia, Eutheria), using NCBI and ENSEMBL databases. New sequences were generated from four non-human primates, i.e Bornean orangutan, siamang, crab-eating macaque, and silvered leaf monkey. Total RNA was extracted from testis (crab-eating macaque) and lymphoblastoid cell lines (all others) using TRIzol Reagent (Invitrogen). cDNAs were synthesized from total RNA with SuperScript II reverse transcriptase (Invitrogen), using the reverse gene-specific primer 5'-TCAGGTTGCCAGGCAGGAT-3'. The target cDNA encompassing the entire KCNE1-coding sequence was amplified with the forward gene-specific primer 5'-AGCCAAGGATATTCAGAGGT-3' and the reverse gene-specific primer mentioned above by standard PCR under the following conditions: 3 min at 94°C, 30 × (30 s at 94°C, 30 s at 58°C, 40 s at 72°C), 5 min at 72°C. Direct sequencing was carried out as described for the human samples. In a final step, we completed the dataset with GenBank entries from another 14 eutherian representatives, thus generating a dataset comprising orthologs of human and 18 non-human species (see bold accession numbers in Table 4). The sequences were translated, ClustalW aligned, and re-translated using BioEdit . The non-human sequences were moreover scrutinized for polymorphisms of the codon site ortholog to human KCNE1 38, using ENSEMBL data. We finally checked the respective codon site of non-human KCNE1 sequences for hints of genomic recoding at the mRNA level. Therefore, we compared gDNAs retrieved from ENSEMBL and NCBI with the actual set of cDNAs deposited in the Nucleotide collection and EST databases at NCBI (see Table 4).
Population genetic analyses were carried out using allele frequencies and genotype counts of the geographically distinct human samples covered by HapMap phase II (120 Central Europeans, 90 Han Chinese from Beijing, 88 Japanese from Tokyo, 120 Yoruba from Ibadan/Nigeria). Expected genotype counts were calculated using Fstat 2.9.3 . We engaged the same software to infer the fixation index, Fis. Fis is a measure of the heterozygote deficit within subpopulations. Consequently, positive Fis-values point to a heterozygote deficit, while negative Fis-values indicate a heterozygote excess. Fstat has moreover been used to test for deviations from Hardy-Weinberg equilibrium (400 randomisations). We corrected for multiple testing by strict Bonferroni adjustment, thus lowering the 5% level of significance to 0.0125.
The evolutionary history of the codon position corresponding to codon 38 in human KCNE1 was traced back using ancestral sequences reconstructed by baseml (PAML 3.15 ). The dataset comprised the non-human KCNE1 orthologs plus the two human alleles. The intree used is shown in Figure 1 and represents the commonly accepted phylogeny among mammals [47, 48] and primates .
We furthermore assessed the evolutionary regime acting on KCNE1 as a whole using the ratio of non-synonymous (amino acid altering) to synonymous (silent) nucleotide substitution rates (= dn/ds = ω). Rate ratios > 1 are commonly accepted as a conservative measure of positive selection (= adaptive evolution) while rate ratios = 1 and < 1 are indicative of neutral evolution and negative selection, respectively. To study the evolution of KCNE1 across the human branches (foreground, see Figure 1) in comparison to the remaining phylogeny, we employed the modified version of branch-site model A  of codeml (PAML 4 ). Model A assumes four site classes: site class 0 integrates codon sites that are conserved throughout the entire phylogeny (foreground plus remaining phylogeny) with 0 < ω0 < 1. Site class 1 comprises codon positions that evolve neutrally throughout the entire phylogeny with ω1 = 1. Site classes 2a and 2b include codon sites that undergo positive selection on the foreground, but are under negative selection or neutral evolution across the remaining phylogeny. The model involves four free parameters: the proportion (p0) and the ω estimate (ω0) of site class 0; the proportion estimate of site class 1 (p1) and the ω estimate of the codon sites that evolve under positive selection across the foreground (ω2). The codon frequency was estimated from a 3 × 4 contingency table. Ambiguity data were not removed from the data (cleandata = 0). The Bayes empirical Bayes approach was used to identify codon sites under positive selection along the (human) foreground branches . To test for the significance of the results, the branch-site test of positive selection (test 2 in ) was carried out. Therefore, we compared twice the log likelihood difference between model A with ω2 > 1, estimated from the data, and model A with ω2 = 1 fixed with the critical values from chi-square distribution with 1 degree of freedom.
Motif search was carried out using the PROSITE database as implemented in the PredictProtein server .
Expression of KCNE1 in HEK293 cells and Western blotting
HEK293 cells (ATCC CRL 1573) were grown at 37°C under 5% CO2 in Dulbecco's Modified Eagle Medium (Gibco) supplemented with 10% fetal calf serum. 2.5 * 105 cells were transfected in six well plates (Greiner) with 1 μg or 3 μg of expression plasmids pcDNA3.1-chimpanzee_ KCNE1_GAT, pcDNA3.1-human_ KCNE1_AGT or pcDNA3.1-human_ KCNE1_GGT by calcium phosphate coprecipitation. To assess the extent of N-glycosylation, tunicamycin (Sigma) was added to the medium to a concentration of 5 μg/ml 10 hours after transfection. For Western blotting, whole cell lysates were prepared 24 hours after transfection. Cells were washed three times in phosphate-buffered saline and pelleted by centrifugation at 300 g. The pellet was resuspended in 5 volumes of ice-cold NP-40 (Fluka) lysis buffer consisting of 137 mM KCl, 20 mM HEPES pH 7.9, 1% Nonidet P-40, 5 mM NaF, 10 μg/ml aprotinin, 10 μg/ml leupeptin, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, and incubated on ice for 40 min. The lysate was cleared by centrifugation at 80,000 g for 30 min. Protein concentrations were determined using the Bradford method (Bio-Rad). Cell lysates were separated by SDS-PAGE and proteins were transferred to polyvinylidene difluoride membranes (Millipore) for 20 min at 150 mA, using a Tris-glycine buffer system. After blocking, membranes were incubated with the primary antibody directed against KCNE1 (Calbiochem). As secondary antibody, a 1:5,000 dilution of peroxidase-conjugated donkey, anti-rabbit antibody (GE healthcare) was used. The control antigen corresponds to residues 67 to 129 of human KCNE1 fused to glutathione S-transferase (Calbiochem).
New sequences reported in this manuscript have been submitted to GenBank under accession numbers EF514881 to EF514888.