Hydroxyurea Induces Genome-Wide Epigenetic Changes In Sickle Cell Disease

Hydroxyurea (HU) is currently the only pharmacologic agent widely used to ameliorate the symptoms of sickle cell disease (SCD). The clinical effects of HU are diverse, including the well-known increase in levels of fetal hemoglobin (HbF), effects on leukocyte and platelet counts, and down-regulation of red cell adhesion despite increased expression of some adhesion receptors. However, the precise mechanisms by which HU exerts its ameliorative and pleiotropic actions are not well understood. HU is primarily thought to inhibit DNA replication and cause cell cycle arrest due to inhibition of ribonucleotide reductase (Yarbro, 1992). This mechanism is likely how HU exhibits its ameliorative effects in myeloproliferative conditions. But it is not as clear how this action results in an increase in the percentage of red blood cells that express a large percentage of HbF (Platt, 2008). Previous evidence suggests that HU may induce epigenetic (specifically, hypermethylation) changes to DNA (Nyce, 1989). Thus, we hypothesized that HU may increase HbF levels and cause other systemic changes through epigenetic mechanisms.

To test the hypothesis that HU usage is associated with alterations in DNA methylation, we examined DNA samples from 24 adult patients with SCD, 12 of whom were taking HU at the time of DNA collection and 12 of whom were not. DNA from each individual was pre-treated with bisulfite (Zymo Research) and assessed for methylation levels at 27,578 CpG sites in 14,495 genes using the Illumina HumanMethylation27 BeadChip. One sample (off HU) failed to undergo a successful bisulfite DNA conversion and was subsequently removed from analysis. The relative levels of methylation (β) were calculated as the ratio of methylated probe signal to total locus signal intensity. Linear regression (PROC GLM, SAS version 9.1.3, Cary, NC) was used to test for differences in methylation (β) as a function of HU usage, controlling for sex and age.

On average, patients using HU had higher mean levels of methylation genome-wide compared with patients not taking HU, although the difference was not statistically significant, likely due to the small number of patients examined. Using the Benjamini-Hochberg false discovery correction for multiple testing and setting a stringent corrected p-value threshold of 0.05, we identified 247 out of the approximately 27,000 CpG sites that were differentially methylated as a function of HU usage. Eight CpG sites met a very stringent Bonferroni correction and represented a functionally diverse set of genes, including ones that encode a phosphatase, influence neural outgrowth, and play a role in vertebral development. CpG sites meeting the false discovery correction (n=247) were subsequently subjected to gene network and ontology analysis using DAVID pathway software (http://david.abcc.ncifcrf.gov/) to determine if specific biological pathways were statistically enriched for differences in DNA methylation. Several of these sites were associated with pathways involved in cell growth, senescence and differentiation, as might be expected given the known effects of HU on hematopoiesis. However, the TGFβ pathway was also significantly represented in this subset of CpG sites (pathway analysis p<0.05).

Consistent with findings by Nyce (1989), we observed a trend in which HU usage was associated with hypermethylation, so that patients taking HU had on average more methylation genome-wide than patients not taking HU. Genetic variation in the TGFβ pathway has been implicated in the occurrence of several SCD complications, including pulmonary hypertension (Ashley-Koch et al., 2008), risk for stroke (Sebastiani et al., 2005), leg ulcers (Nolan et al., 2006), bacteremia (Adewoye et al., 2006), and priapism (Elliott et al., 2007). Therefore, identification of this pathway as affected by HU is especially interesting. In summary, we have evidence that DNA epigenetic differences occur in sickle cell patients as a result of HU usage and are associated a wide variety of gene pathways, consistent with the diverse array of clinical and laboratory changes observed in patients on HU. Future studies will include confirmation of the epigenetic differences in the specific genes implicated among a larger cohort of SCD patients on and off HU, as well as functional studies of these genes.

No relevant conflicts of interest to declare.