Epigenetic and molecular profiles of erythroid cells after hydroxyurea treatment in sickle cell anemia

Abstract

Hydroxyurea has been shown to be efficacious for the treatment of sickle cell anemia (SCA), primarily through the induction of fetal hemoglobin (HbF). However, the exact mechanisms by which hydroxyurea can induce HbF remain incompletely defined, although direct transcriptional effects and altered cell cycle kinetics have been proposed. In this study, we investigated potential epigenetic and alternative molecular mechanisms of hydroxyurea-mediated HbF induction by examining methylation patterns within the (G)γ-globin promoter and miRNA expression within primary CD71(+) erythrocytes of patients with SCA, both at baseline before beginning hydroxyurea therapy and after reaching maximum tolerated dose (MTD). Using both cross-sectional analysis and paired-sample analysis, we found that the highly methylated (G)γ-globin promoter was inversely correlated to baseline HbF levels, but only slightly altered by hydroxyurea treatment. Conversely, expression of several specific miRNAs was significantly increased after hydroxyurea treatment, and expression of miR-26b and miR-151-3p were both associated with HbF levels at MTD. The significant associations identified in these studies suggest that methylation may be important for regulation of baseline HbF, but not after hydroxyurea treatment, whereas changes in miRNA expression may be associated with hydroxyurea-mediated HbF induction. This study was registered at ClinicalTrials.gov (NCT00305175).

Figure 1

Induction of HbF levels by hydroxyurea. (A) Linear regression analysis indicates an inverse correlation between age and baseline HbF levels with 95% confidence interval (dotted lines). (B) Paired-sample analysis from individual subjects at baseline (□) and at MTD (Δ) illustrates the increase in HbF with hydroxyurea treatment per patient (gray lines) and mean ± SE (black lines) for each group of samples. (C) A histogram showing the variable HbF-induction in patient samples, with the absolute change in HbF (ΔHbF) between baseline and MTD (*P < .001 baseline HbF compared with MTD by the paired t test).

Figure 2

Methylation of ^Gγ-globin promoter in CD71⁺ NRBCs. (A) CpG sites −54, −51, +5, +16, and +48 of the ^Gγ-promoter are methylated in CD71⁺ NRBCs as quantitated by pyrosequencing of bisulfite treated DNA. (B) Methylation at sites −54, −51, and +16 is significantly decreased at maximum tolerated dose (MTD) of hydroxyurea compared with the baseline (BL) methylation before treatment (P < .05). (C) Data are presented as median with interquartile range. Significant inverse correlations between combined methylation of all 5 sites and HbF were detected at baseline and at MTD (D) by the Pearson correlation coefficient.

Figure 3

Differential miRNA expression in CD71⁺ erythroid cells by microarray and Q-PCR analysis. (A) Heatmap illustrates relative changes in miRNA expression of 10 miRNAs that were significantly different between CD71⁺ cells from 2 individuals without SCA (HbAA), 3 SCA patients before hydroxyurea (HbSS), and 3 SCA patients at hydroxyurea MTD (HbSS(H)). Significant changes were identified in this cross-sectional study by Z-score statistic where P < .05. (B) Graphs illustrate the change in expression of miRNAs 148a, 151-3p, and 494 from baseline to MTD for each patient (gray symbols and lines). Median miRNA expression was significantly higher at MTD compared with baseline for each miRNA as determined by the Wilcoxon signed rank test (*P = .03; **P = .0036; ***P = .0002).

Figure 4

MiRNAs 26b and 151-3p are associated with HbF. (A) Expression of miRNA 26b at baseline and at MTD was inversely associated with HbF levels at baseline and MTD. (B) respectively as identified by the Spearman correlation coefficient. (C) HbF levels and miRNA 151-3p expression in paired samples from baseline to MTD for each patient (represented by individual lines), and analysis of this data with a mixed model indicated a significant association between change in HbF and a change in miRNA 151-3p in response to hydroxyurea treatment (P < .05).