The study has been approved by the Ethics Committee on Human Research of University of Tartu, Estonia (permission no 122/13,22.12.2003). All recruited individuals gave their informed consent prior to their inclusion in the study. Diagnosis and classification of hypertension was carried out by "Practice Guidelines for primary care physicians: 2003 ESH/ESC hypertension Guidelines" Journal of Hypertension 21(10)" pp. 1779–1786. Estonian individuals with primary hypertension (n = 25) were recruited into the study by blood pressure specialists (cardiologist M.V.; nephrologist M.O.) during their ambulatory visit or hospitalization in the Cardiology and Internal Medicine Clinics at the Tartu University Hospital, Estonia. Czech individuals with essential hypertension (n = 25) were recruited from the Cardiology Department of the 2^nd Clinic of Internal Medicine, Faculty Hospital Královské Vinohrady in Prague (coordinated by V.K.). Details of the collection of Czech patients and normotensives are published elsewhere 15. The criteria for the selection of hypertensive individuals for the study:

1. Hypertension diagnosed by a specialist (cardiologist, nephrologist)

2. Grade 2 or severe hypertension at diagnosis (systolic blood pressure, SBP >160 mmHg; diastolic blood pressure, DBP >100 mmHg)

3. Age at diagnosis for men ≤55 and women ≤65 years (to exclude secondary, age-related hypertension)

4. Exclusion of secondary causes for hypertension (such as diabetes, primary renal failure) by a specialist clinician

5. Family background of high blood pressure (to maximize genetic susceptibility)

Hypertension was documented among 1^st degree relatives for 84% (n = 21) of Estonian and 92% (n = 23) of Czech patients and among 2^nd degree relatives for 28% (n = 7) of Estonian hypertensives (for Czech: data not available). Multiple hypertensive family members were reported for 36% (n = 9) of Estonian and 48% (n = 12) of Czech patients.

6. All recruited individuals received antihypertensive treatment prescribed by a specialist. Blood pressure measurements under antihypertensive therapy were documented: SBP mean 142.8 and range 115 – 215 mmHg; DBP mean 88.4 and range 80–105 mmHg).

A classical case-control study maximizes its power by sampling a clinically homogeneous group of patients. The aim of our resequencing and mutation screening study was to obtain exhaustive coverage of PNMT variation in hypertensives. Thus, a high proportion of included patients (Estonians 56%; Czech 92%) exhibited a spectrum of heterogeneous high-blood pressure related organ complications (MI, CAD, stroke, end-stage renal disease).

Normotensives (age at recruitment >35 years) exhibited age-adjusted normal blood pressure based on repeated measurements, lack of family history of essential hypertension, absence of organ damages and had never been prescribed antihypertensive medication. Estonian normotensives (n = 25) were recruited in collaboration with Estonian Blood Centers across Estonia, including only long-term blood donors exhibiting optimal (<120/<80 mmHg)/normal (120–129/80–84 mmHg) blood pressure. Hypertensive and normotensive groups did not differ in Body Mass Index (BMI), a risk factor for hypertension (Mann-Whitney U-test; p > 0.05, two-tailed). BMI ranged for Estonian hypertensives 28.7 ± 6.3, and normotensives 25.2 ± 3.0; and for Czech hypertensives 26 ± 2.3, and normotensives 24.6 ± 2.5.

The sequence of human PNMT gene [GenBank:J03280.1] has been obtained from NCBI GenBank database 16. PCR and sequencing primers for PNMT (Table 1) were designed using the web-based Primer3 software 17. The uniqueness of all the primers was checked using BLAST 18. The PCR primers were designed to cover the entire coding region and parts of 5' and 3' UnTranslated Regions (UTR) (3148 bp); and amplify two overlapping PCR fragments, 1846 bp and 1842 bp respectively. Amplification was performed with 100 ng genomic DNA using Long PCR Enzyme Mix (MBI Fermentas). Conditions for PCR amplifications, product purification, sequencing, sequence contig assembly and polymorphism identification are described in detail elsewhere 19.

Testing of Hardy-Weinberg equilibrium (α = 0.05); estimation of allele and genotype frequencies of identified Single Nucleotide Polymorphisms (SNPs); and comparison of allelic/genotypic as well as promoter haplotype distribution between hypertensives and normotensives (Fisher's exact test; α = 0.05) were implemented with Genepop software (Version 3.4) 20. Promoter haplotypes were predicted by the Bayesian statistical method in the program PHASE 2.1 21.

Sequence diversity parameters were calculated by DnaSP software (Version 4.0) 22. Nucleotide diversity provides a measure of genetic variation that is normalized by the number of sampled chromosomes. We calculated two conventional measures of nucleotide diversity: (i) π, the direct estimate of per-site heterozygosity derived from the observed average pairwise sequence difference and (ii) Watterson's θ 23, the population mutation parameter representing an estimate of the expected per-site heterozygosity based on the number of segregating site (S). Tajima's D statistic (D^T) 24 was calculated to determine if the observed pattern of human PNMT gene diversity is consistent with the standard neutral model. The D^T value is the difference between π and θ estimates. In case of neutrality π equals θ, and thus D^T statistic equals zero. The direction of D^T statistics (either <0 or >0) is potentially informative about the evolutionary and demographic forces that the population has experienced. Significant positive D^T values indicate an excess of intermediate-frequency alleles consistent with either balancing selection or population bottleneck, whereas significant negative D^T values indicate an excess of rare SNPs consistent with either recent directional selection or an increase in population size.

In silico prediction of transcription factor-binding sites (TFBS) for PNMT intron 1 was carried out by MatInspector 2.2 2526 online software. The program identifies TFBS in nucleotide sequences using large library of position weight matrices (PWM) and is based on the information about experimentally defined TFBSs collected in the TRANSFAC database 27. The presence of repetitive sequences was analyzed by RepeatMasker online software 28.

We resequenced the human PNMT genomic sequence (in total 3187 bp) in hypertensive (n = 50) and normotensive (n = 50) individuals originating from two European populations. The analyzed region (from -882 to +2305 relative to ATG) included the entire 5' and 3'UTR regions, three exons (424 bp, 208 bp and 525 bp respectively) and two introns. The identified diversity patterns supported the conservative nature of the PNMT gene. We determined only three rare genic polymorphisms; in contrast, the upstream region harboured four SNPs, two of which (SNP -184 A/G; SNP -390 A/G) are previously characterized common variants (Table 2, Figure 1). Majority of the SNPs were identified in both populations and none was exclusively present in only hypertensive individuals. Compared to a data set of 74 genes 29, characterized by a mean nucleotide diversity parameter π = 0.00080 for European Americans, PNMT gene exhibited approximately three times lower diversity among studied populations including both normotensive and hypertensive individuals (π = 0.00026 – 0.00032; Table 3). As the only identified non-synonymous mutation (Exon 3; SNP +1517; Ala->Thr) was present in both populations for individuals with normal (Major Allele Frequency, MAF = 6%) as well as elevated (MAF = 4%) blood pressure, we exclude this protein variant increasing susceptibility for hypertension.

When we addressed the population differentiation of the identified SNPs by Fisher's exact test, none of the polymorphisms showed either allelic or genotypic differentiation among the Estonians and the Czech, as well as between normotensive and hypertensive individuals in the intrapopulation comparisons (p > 0.05, data not shown). The joint case-control analysis of all samples with an increased test power detected a significant excess of heterozygotes for common promoter region polymorphisms (SNP-184; SNP-390) among the patients (Fisher exact test, p < 0.05; Table 2). There was a non-significant difference between normotensives and hypertensives for the distribution of haplotypes formed from two common 5'UTR polymorphisms, SNP -390 and SNP -184.

However, the most striking feature of PNMT gene is the lack of variation within introns harboring only one rare SNP (Figure 1; intron 1, SNP+360). The introns exhibit >15-fold less diversity compared to analyzed 5'upstream region (π_intron = 0.00002 – 0.00007; π_5'upstream = 0.00099–0.00138; Table 3) and >7-fold reduction in diversity compared exons (π_exon = 0.00050–0.00056). In comparison, the excess of diversity for the 5'upstream region compared to the exons is only 2–2.5 times. To test whether patterns of DNA sequence variation in PNMT fit the expectations under hypothesis of neutrality, we analyzed the data with the Tajima DT neutrality test (Table 3). When the analysis included an entire PNMT gene there was no difference between observed (π) and expected (θ) nucleotide diversity parameters. Notably, intronic sequences formed an exception with the expected variation (θ = 0.00018–0.00021) exceeding the observed values (π = 0.00002–0.00007) up to tenfold. Whereas the lack of diversity in intron 2 spanning only 114 bp is apparently due to functional constraint on neighboring exons, minimal variation and consistent Tajima DT values for intron 1 support an effect of purifying selection on this region.

Alignment of human PNMT and rat Pnmt gene reveals high level of exonic conservation (>82%) at the level of DNA sequence between two species (See additional file 1). In contrast to the expectations based on selection pressure shown for human intron 1, there is no homology between intron 1 of PNMT and Pnmt genes. An AluSp insertion in primate lineage (See additional file 1) has changed the genomic landscape of human PNMT.

In order to explore the hypothesis of the purifying selection acting on human PNMT intron 1, we analyzed in silico the distribution of functionally important putative regulatory elements in this region. Two seminal works 1213 have predicted the presence of a Glucocorticoid Responsive Element (GRE) in the middle of intron 1. In a number of genes, the glucocorticoid (GC) response is enhanced by the binding of other transcription factors to adjacent binding sites, forming Glucocorticoid Responsive Units (GRUs) 30. We explored putative PNMT regulatory elements among the predicted transcription factor binding sites (TFBS) within intron 1 (Figure 2). We focused on TFBS-s either reported (1) to regulate rat Pnmt gene; (2) to form active GRUs; (3) to locate within intronic gene regulatory units; or (4) selected as potential inducers/repressors of epinephrine synthesis.

In rat, Pnmt transcription is synergistically activated by binding of Egr-1 (Early Growth Response 1), AP2 (Activating enhancer binding Protein 2 alpha) and GR (Glucocorticoid-activated Receptor complex) to the upstream promoter 31, whereas Sp1 (Specificity protein 1) and MAZ (Myc-Associated Zinc finger protein) transcription factors may potentially contribute to tissue-specific expression 32. In human PNMT intron 1 we identified multiple binding sites for Egr1, MAZ and Sp1/Sp2 (Figure 2). A shared recognition site for Egr1/MAZ/Sp1 276 bp upstream GRE overlaps with a binding site for another zinc-finger nuclear protein – ZBP-89 (Zinc finger Binding Protein 148). For example, for bovine adrenodoxin gene expressed in adrenal cortex, Sp1/Sp3 confer and ZBP-89 represses basal transcriptional activities 33. An alternative binding site for Sp1/Sp2 (74 bp downstream GRE) is co-localized with binding site for NFκB (Nuclear Factor kappa B), a transcription factor involved mainly in inflammatory and immune responses. There is an overlap also in the function of the two elements since GCs have potent anti-inflammatory and immunosuppressive properties. NFκB activity is antagonized by GCs either directly through inhibiting NFκB binding to DNA or indirectly by binding of the hormone-activated GC receptor complex (GR) to GRE 34.

The most striking feature of human PNMT intron 1 is the abundance of predicted Insulin Responsive Elements [IRE, consensus sequence T(G/A)TTT(T/G)(G/T)]. GRE element is flanked both sides by multiple overlapping perfect IREs (four and two, respectively). Additional four IREs differing by one nucleotide from the consensus [T(G/A)TTT(T/G)C] are located in the vicinity. It has been shown that for a subset of genes insulin inhibits GR activity either via IRE-dependent or IRE-independent mechanisms 35. PNMT intron 1 distal IREs overlap with a predicted binding site for HMGI/Y (High Mobility Group protein isoform I and Y), a mammalian architectural transcription factor that participates in specific protein-DNA and protein-protein interactions that induce both structural changes in chromatin substrates and the formation of stereospecific multiprotein complexes on the promoters/enhancers of genes whose transcription they regulate 36. This factor is also part of nucleoprotein complex controlling human insulin receptor gene transcription and its loss causes insulin resistance and diabetes in human and mice 37.

Among the other reported GRU-belonging cis-acting elements, we identified (i) binding sites for Myc/Max family proteins, reported for the functional GRU of ovine β1-adrenergic receptor gene 38; (ii) two regions of overlapping putative GAGA-boxes shown to increase the rate of GRE-induced gene expression in the proximal enhancer of human carbamoylphosphate synthetase gene 39; (iii) and Estrogen Responsive Element (ERE) 40. Interestingly, ERE within the first intron of PNMT co-localizes with a binding site for RORA (Retinoic acid receptor-related Orphan Receptor α; also called RORα), reported among the key regulators of mammalian circadian gene expression 41. Another putative 'clock controlled element', the binding site for a light-induced transcriptional repressor E4BP4 (mammalian transcription factor E4 Binding Protein 4) 42, was predicted within the cluster of IREs proximal to GRE.