Plasma corticosterone levels varied in a dose-dependent manner, as shown in . In the first experiment (experiment 1), mean plasma corticosterone levels in the group receiving 100 μg/ml of corticosterone in their drinking water (“CORT100”) varied from 254.2 to 606.7 ng/ml, and peaked at 606.7±119.2 ng/ml (mean±SEM) during the fourth week of CORT treatment. Plasma CORT levels in the group receiving 50 μg/ml of corticosterone (“CORT50”) varied from 137.2 to 270.0 ng/ml and peaked at 270.0±30.2 ng/ml during the second week. For both treatment groups, plasma CORT returned to baseline levels once treatment ended. The group receiving only vehicle solution without corticosterone (“CONTROL”) had mean plasma corticosterone levels ranging from 31.4 to 64.7 ng/ml for the first 4 weeks with no deviation from that range during the recovery period.
Fig. 1 Plasma corticosterone, expression, and DNAm vs. time by treatment group. Mice were given different doses of corticosterone [100 μg/ml (triangle), 50 μg/ml (square), or vehicle solution (circle)] in their drinking water for 4 weeks (depicted (more ...)
We assessed expression changes in the Fkbp5 gene during CORT treatment and recovery periods. The greatest increase in expression was observed in the CORT100 group (range 66.0% to 202.1%, P≤0.0016). A significant increase in expression was also observed in the CORT50 group (46.7% to 113.7%, P≤0.05). However, expression levels did not show a clear dose-dependent change, except at week 4, where the CORT100 group showed a 56.7% increase in expression over that of the CORT50 group (P=0.045). A time-course change in Fkbp5 expression is shown in .
In contrast, we observed a more pronounced dosedependent decrease in DNAm at the intron 1 region of Fkbp5 (CpG-1) with the greatest overall decrease seen in the CORT100 group (range 20.9% to 34.2%, P<0.0003) and a lesser decrease seen in the CORT50 group (range 8.8% to 18.4%, P≤0.0001), with only minor fluctuations seen in the CONTROL group (). At the CpG-2 position, the greatest decrease in DNAm was observed again in the CORT100 group (range 20.1% to 34.4%, P< 0.0001), with a smaller decrease in the CORT50 group (range 8.1% to 17.2%, P<0.0003; ). Intriguingly, DNAm changes persisted for both treatment groups in the first week of recovery (at week 5), with a return to baseline in the following weeks.
Plasma corticosterone levels determined from blood drawn on a weekly basis show a dynamic range of corticosterone levels among individuals (mean plasma CORT levels from 15.4 to 604.8 ng/ml), as well as intra-individual differences from week-to-week blood draws (e.g., weekly ranges from 204.2 to 1,009.3 ng/ml for the highest exposed CORT-treated mouse and weekly ranges from 7.9 to 25.3 ng/ml for the least varying control mouse). This wide range of CORT levels within a given mouse shows how unreliable a single serum corticosterone level can be as a marker of chronic glucocorticoid exposure, even when glucocorticoids are being administered in the drinking water. We asked whether either of two measures, percent DNAm changes in the intron 1 region or Fkbp5 expression level, could be used to accurately reflect the mean 4-week corticosterone burden imposed on the animals. For both CpG-1 and CpG-2 positions, DNAm measured at week 4 in each animal correlated significantly with the corticosterone burden in the animal as determined by taking an average of plasma CORT levels from weeks 1 to 4. For CpG-1 and CpG-2, the squares of the correlation coefficient were determined to be r2=0.69 (P=2.3×10−9) and r2=0.75 (P=5.9×10−11), respectively (). On the other hand, correlation between the expression levels of Fkbp5 and the mean CORT levels were not as strong as those observed for the two CpG positions (r2=0.34, P=0.00036; ).
Fig. 2 Regression analysis of mean corticosterone levels vs. methylation and expression. A strong correlation was observed between week-4-blood DNAm of Fkbp5 intron 1, CpG position 1 (a) and CpG position 2 (b) vs. mean corticosterone levels from weeks 1–4. (more ...)
Chronic exposure to glucocorticoids is accompanied by physiological and metabolic changes that include significant reductions in the mass of the thymus, spleen, and adrenal glands, as well as an increase in adipose tissue deposition. In the first experiment (experiment 1), we observed a complete atrophy of the thymus tissues and a significant reduction in adrenal gland mass in animals that were treated with either 50 or 100 μg/ml of CORT (). Since the spleen and adipose tissues varied with a wider range of masses, we asked whether these changes correlated with the DNAm values at week 4. We found a significant correlation for both spleen (r2=0.56, P=2.8×10–5) and percent fat (r2=0.51, P=9.0×10–5) when compared to DNAm at CpG-1 (). Similar results were obtained for CpG-2 (spleen: r2=0.60, P= 1.05×10–5;% fat: r2=0.65, P=2.1×10–6; ). We observed that glucose, one of the metabolites commonly elevated in the Cushingoid state, was also moderately correlated with DNAm at CpG-2 (r2=0.48, P=0.0010). In addition, we found a modest but statistically significant relationship between DNAm and the percent of time the mice spent in the closed arms of the elevated plus maze (EPM, r2=0.35, P=0.02 for both CpG-1 and CpG-2), reflecting anxiety-like behavior. The correlation between the two measurements is shown in .
Fig. 3 Organ weights and regression analysis of DNAm vs. spleen mass and percent visceral fat. a Thymus and adrenal glands were weighed for corticosterone-treated (gray and black bars for 50 and 100 μg/ml, respectively) and vehicle-treated (white bars (more ...)
Fig. 4 Regression analysis of DNAm vs. percentage of time spent in the closed arms on the elevated plus maze (EPM). After 4 weeks of CORT treatment, mice were challenged with the EPM to test for anxiety-like symptoms. Mice that exhibit higher anxiety-like symptoms (more ...)
We next sought to extend our results in a second experiment (experiment 2), in which we reasoned that we might be able to: (a) more accurately assess mean CORT burden by performing daily rather than weekly blood draws; and (b) more accurately calculate correlation of DNAm with plasma CORT levels by using a broader range of CORT dose exposures. To this end, we divided 20 mice into five groups of four animals each and gave them different doses of corticosterone in their drinking water (0, 25, 50, 75, 100 μg/ml). Daily plasma CORT levels determined for 25 days were used to establish mean CORT values for each mouse. The range of mean plasma CORT values we achieved was less than that observed for the first experiment (first experiment inter-individual range of 15.4 to 604.8 ng/ml vs. second experiment inter-individual range of 34.1 to 186.3 ng/ml). In the second experiment, the animals drank less water, suggesting that the reduced range of CORT may have been due to the reduced CORT intake via the drinking water. For instance, while the mean daily intake of water for the mice administered 100 μg/ml of CORT was 7.8 ml per mouse in the first experiment, the mean daily intake of water for the same group in the second cohort was only 3.5 ml. The reduction in water consumption in the second cohort may reflect the stress of taking blood samples 5 days per week.
Nonetheless, a strong correlation was again observed between DNAm and the mean CORT levels during the 25-day period for both CpG-1 (r2
=0.60, P=6.6×10−5, Suppl. Fig. 1a
) and CpG-2 (r2
=0.63, P=3.3×10−5, Suppl. Fig. 1b
). We also determined the various organ weights of the second cohort of mice and correlated spleen mass and percent fat to DNAm at week four. Once again, we observed a complete atrophy of the thymus at CORT doses greater than 25 μg/ml and a dose-dependent decrease in adrenal gland mass (Suppl. Fig. 2a
). There was a significant correlation between spleen mass vs. DNAm at CpG-1 (r2
=0.48, P=0.0011) and percent fat vs. DNAm (r2
=0.35, P=0.0076) (Suppl. Fig. 2b and c
, respectively). DNAm at CpG-2 showed similar correlations with spleen mass (r2
=0.48, P=0.0010, Suppl. Fig. 2d
) and percent fat (r2
=0.33, P=0.011, Suppl. Fig. 2e