Els. A principal component analysis (PCA) revealed 5 components 
of which the first one explained 79 of the total variance of all CpG residues. A sum score for methylation was then calculated by multiplying the values for each individual CpG residue by its factorial loading obtained in the first component of PCA. These weighted values were then summed up to create a weighted sum score, denoted as METsum throughout this study. The mean values for METsum in 
the high and low work K162 web stress groups were 25.7610.6 (95 confidence interval 21.2?0.1) and 50.7621 (95 confidence interval 42.0?9.4) respectively (see Table S1). Based on confidence limits, the Calciferol site differences between the two work stress groups were evident. The mean METsum in the total sample was 38.4620.8. In order to determine the effect size between work stress groups, we calculated the Cohen’s d for the difference between means [(50.7?5.7)/20.8 = 1.2]. The effect size of 1.2 indicates that the difference is notable. We then performed a multifactorial analysis of covariance to assess the association of METsum (as the dependent variable) with work stress environment, burnout, work control, work demand, age, and 5-HTTLPR (as explanatory variables). When the main effects of all explanatory variables were included into the model, only work stress was statistically significant and independently (F = 5.7; p = 0.022) associated with METsum (see Table S1). However, when non-significant effects were eliminated one by one from the model, it was found that in the final model both workStress Affects Serotonin Transporter Methylationstress (p,0.0001) and MBI-GS (p = 0.038) were independently associated with methylation levels. Together they explained 43 of the total variance of METsum (R2 = 0.43, p,0.0001). When the effect of work stress environment was adjusted, burnout was positively associated with methylation levels. In other words, in stratified analyses separately in high and low work stress environments, higher burnout scores were associated with higher methylation levels (Figure 3). Because of a small number of observations, these stratified analyses did not reach statistical significance (p = 0.062 in high work stress environment and p = 0.065 in low work stress environment).Table 2. Distribution of the 5-HTTLPR genotype.Genotype N ( ) Mean burnout scores (SD)La/La 21 (39.6 )La/S 19 (35.8 )S/S 9 (17.0 )0.889 (60.764) 1.02 (60.604) 0.951 (60.499) 6.05 (64.94) 36.5 (622.3) 6.89 (64.34) 49.9 (625.3)Mean depression scores (SD) 6.48 (66.32) METsum (SD) 35.3 (616.2)doi:10.1371/journal.pone.0045813.t5-HTTLPR GenotypeNext we examined the distribution of 5-HTTLPR polymorphism and its association with methylation and MBI scores. Mean METsum levels for the La/La, La/S and S/S genotypes were 35.3, 36.5 and 49.9, respectively (Table 2). There was no significant association between 5-HTTLPR and METsum (p = 0.25), 5-HTTLPR and work stress (p = 0.067) or 5-HTTLPR and burnout (p = 0.50). We then tested interactions for the genotype and work stress or burnout in a multivariable model but found no statistically significant evidence for these interactions (p = 0.58 and p = 0.082, respectively). Effect sizes between La/La and S/S, La/S and S/S, and La/La and La/S were 0.73, 0.57, and 0.06 and their minimal detectable effect sizes were 1.03, 10.09, and 1.02, respectively.DiscussionOur initial results showed that DNA methylation levels in the promoter region of SLC6A4 varied between high and low work stress env.Els. A principal component analysis (PCA) revealed 5 components of which the first one explained 79 of the total variance of all CpG residues. A sum score for methylation was then calculated by multiplying the values for each individual CpG residue by its factorial loading obtained in the first component of PCA. These weighted values were then summed up to create a weighted sum score, denoted as METsum throughout this study. The mean values for METsum in the high and low work stress groups were 25.7610.6 (95 confidence interval 21.2?0.1) and 50.7621 (95 confidence interval 42.0?9.4) respectively (see Table S1). Based on confidence limits, the differences between the two work stress groups were evident. The mean METsum in the total sample was 38.4620.8. In order to determine the effect size between work stress groups, we calculated the Cohen’s d for the difference between means [(50.7?5.7)/20.8 = 1.2]. The effect size of 1.2 indicates that the difference is notable. We then performed a multifactorial analysis of covariance to assess the association of METsum (as the dependent variable) with work stress environment, burnout, work control, work demand, age, and 5-HTTLPR (as explanatory variables). When the main effects of all explanatory variables were included into the model, only work stress was statistically significant and independently (F = 5.7; p = 0.022) associated with METsum (see Table S1). However, when non-significant effects were eliminated one by one from the model, it was found that in the final model both workStress Affects Serotonin Transporter Methylationstress (p,0.0001) and MBI-GS (p = 0.038) were independently associated with methylation levels. Together they explained 43 of the total variance of METsum (R2 = 0.43, p,0.0001). When the effect of work stress environment was adjusted, burnout was positively associated with methylation levels. In other words, in stratified analyses separately in high and low work stress environments, higher burnout scores were associated with higher methylation levels (Figure 3). Because of a small number of observations, these stratified analyses did not reach statistical significance (p = 0.062 in high work stress environment and p = 0.065 in low work stress environment).Table 2. Distribution of the 5-HTTLPR genotype.Genotype N ( ) Mean burnout scores (SD)La/La 21 (39.6 )La/S 19 (35.8 )S/S 9 (17.0 )0.889 (60.764) 1.02 (60.604) 0.951 (60.499) 6.05 (64.94) 36.5 (622.3) 6.89 (64.34) 49.9 (625.3)Mean depression scores (SD) 6.48 (66.32) METsum (SD) 35.3 (616.2)doi:10.1371/journal.pone.0045813.t5-HTTLPR GenotypeNext we examined the distribution of 5-HTTLPR polymorphism and its association with methylation and MBI scores. Mean METsum levels for the La/La, La/S and S/S genotypes were 35.3, 36.5 and 49.9, respectively (Table 2). There was no significant association between 5-HTTLPR and METsum (p = 0.25), 5-HTTLPR and work stress (p = 0.067) or 5-HTTLPR and burnout (p = 0.50). We then tested interactions for the genotype and work stress or burnout in a multivariable model but found no statistically significant evidence for these interactions (p = 0.58 and p = 0.082, respectively). Effect sizes between La/La and S/S, La/S and S/S, and La/La and La/S were 0.73, 0.57, and 0.06 and their minimal detectable effect sizes were 1.03, 10.09, and 1.02, respectively.DiscussionOur initial results showed that DNA methylation levels in the promoter region of SLC6A4 varied between high and low work stress env.
