We investigated associations in a candidate pathway for HCTZ blood pressure response, which controls expression of ENaC. Our study showed no association between common variation in DOT1L, MLLT3, SIRT1, or SGK1 and blood pressure responses to HCTZ that met our pre-defined criteria for significance. In an exploratory analysis, we also evaluated the role of SNPs in this pathway on untreated blood pressure and identified rs12350051 in MLLT3 as being associated with baseline blood pressure in both GERA and PEAR African-Americans. However, in another PEAR and two normotensive cohorts, this association did not replicate.
The strongest pharmacogenetic association with HCTZ response was with rs2269879 in DOT1L, and was only observed in PEAR Caucasians. Results with systolic and diastolic response in GERA were directionally consistent, but nonsignificant. Located in intron 7, rs2269879 was chosen for genotyping as a tagSNP. Upon review in the HapMap CEPH population, the SNP was found to be in perfect linkage with rs8113528 (r2 = 1.0), in intron 3. FastSNP showed the variant A allele at rs8113528 creates a possible binding site for p300, transcriptional co-activator that functions as a histone acetyltransferase. Additionally, UCSC Genome Browser http://genome.ucsc.edu/ indicates rs8113528 exists in an area surrounded by moderate histone acetylation.
Because the association only met the significance threshold in PEAR, and did not replicate in GERA, we cannot rule out that this is a chance finding. One reason for a lack of replication in GERA may be that the effect of this SNP can only be detected using home blood pressure. PEAR was the only study with the home blood pressure phenotype available. We decided a priori to use it as the response phenotype in PEAR because home blood pressure is a more accurate phenotype, as home blood pressure predicts cardiovascular risk better than office blood pressure [25, 26]. In addition, we previously found ambulatory blood pressure measurement, another potentially better predictor of cardiovascular risk, correlated with home blood pressure more than with office blood pressure in a subset of PEAR participants . PEAR home blood pressure entries were averages of multiple measurements spanning at least five days, thus they likely give a better estimate of participants' actual blood pressures. Home blood pressure is also a more precise phenotype, as evidenced by the smaller standard deviations in home systolic blood pressure measurements we observed in PEAR compared with office measurements (Table 1 systolic: P < 0.001, diastolic: P = 0.689). Office measurements, the only blood pressure phenotypes available in GERA, may not possess high enough fidelity to detect this association with rs2269879. Supporting this theory, we observed similar, but much weaker associations with office blood pressure response in PEAR. Perhaps if another large hypertensive cohort, prospectively treated with HCTZ, becomes obtainable for analysis of home blood pressure responses, the association we found could be tested again for replication.
The lack of association we found with HCTZ response suggests that genotyping polymorphisms in this pathway would likely not help predict patient response to thiazide diuretics. The likelihood that common SNP associations were missed with DOT1L, MLLT3, SIRT1, SGK1 and blood pressure associations is low. TagSNPs within 5000 bases of each candidate gene were selected to try to detect any possible cis-regulatory regions. Great effort was spent on identifying pfSNPs in silico for each candidate gene, which were not required to be in the pre-defined gene region for tagSNP development. However, only SNPs with a minor allele frequency of 0.05 were considered for genotyping, so our study cannot rule out very rare SNPs in the candidate genes with large effect sizes affecting blood pressure response. Additionally, our data do not rule out whether or not this pathway plays any role thiazide response. If HCTZ did have some small effect on H3K79 methylation, redundancy in ENaCα regulation  could conceivably overcome the changes in H3K79 methylation and leave behind no measurable change in patient blood pressure response.
Little is known about the effect of this histone H3K79 methylation pathway on blood pressure regulation in humans, so exploratory analyses testing associations in untreated blood pressure phenotypes could also provide valuable information. The SNP that associated and replicated with untreated blood pressure was rs12350051 in MLLT3. It was chosen as a tagSNP, and is located in intron 2, with no linkage to any known functional SNPs. In silico, rs12350051 was not observed in any known miRNA sequences, transcription factor binding sites, exonic splice sites, splice enhancer, or silencer sequences. Because the same blood pressure association was not seen in Caucasians, one possibility could be that this SNP is in high linkage disequilibrium with an undiscovered functional polymorphism in African-Americans.
One would expect using a patient population with a wide range of blood pressures to be the best method to detect genetic associations with untreated blood pressure. So the fact that untreated blood pressure associations were seen in PEAR and GERA is somewhat surprising, as these studies enrolled hypertensives spanning a relatively small blood pressure range. This was one of the reasons we attempted to replicate these findings in normotensive blood pressure ranges not represented in PEAR and GERA. However, the normotensive groups also had a narrow blood pressure range. The fact that no replication was observed in normotensives could be because of this narrow blood pressure range, the fact that they were younger, the differences in study protocols, leading to differences in blood pressure measurement precision, or the sample sizes were too small and lacked the power to detect the effect we observed in the larger hypertensive cohorts. Another possibility is that perhaps the effect of this SNP is easier to detect or only exerts an effect with higher blood pressures.
Our findings are not the first to detect associations between this H3K79 methylation pathway and blood pressure regulation. Dot1 conditional knockout mice were shown to exhibit salt sensitive hypertension . Conversely, mice null for Af17, which has been shown to compete with Af9 for Dot1 binding at the ENaCa promoter,  exhibit renal salt wasting and hypotension with diminished renal H3K79 methylation and renal ENaCα gene expression . Also, Sirtuin-1 deacetylation of endothelial nitric oxide synthase  has been suggested as a possible mechanism for the blood pressure reduction seen during caloric restriction, a well-known inducer of Sirtuin-1 .