Genetic and Epigenetic Regulation of the Gamma Globin Locus Is Associated with Fetal Hemoglobin Levels and Frequency of Pain in Sickle Cell Disease

Abstract 3230

Sickle cell disease (SCD) is caused by a single amino acid substitution of the β-globin chain of hemoglobin, resulting in abnormal red blood cells that occlude vessels, causing tissue damage, pain, and anemia. While all individuals with SCD have the same β-globin mutation, disease severity varies and is correlated with the concentration of retained fetal hemoglobin (HbF) (Akinsheye et al, 2011). After birth, a switch from γ- to β-globin gene expression occurs and HbF (α₂γ₂) is replaced by HbA (α₂β₂). In patients with SCD, increased levels of γ-globin can help compensate for the effects of the β-globin mutation. SCD patients with higher than normal retention of HbF frequently have more benign disease symptoms and longer life expectancy (Platt et al, 1994). Increased HbF expression through the use of hydroxyurea (HU) is a frequently used therapy and the only FDA-approved medication for SCD. However, due to the toxicity and inconsistent efficacy of HU, further understanding regarding the regulation of this “hemoglobin switch” is essential for developing alternative treatments.

DNA methylation is central to the silencing of γ-genes during the switch to β-globin expression (Goren et al, 2006). Methylation marks are established by the DNA methyltransferase DNMT3A, which is recruited to the γ-promoter by the protein arginine methyltransferase, PRMT5 (Rank et al, 2010). We first hypothesized that genetic variation in DNMT3A and PRMT5 contributes to HbF levels. To test this, we performed a genetic association study on 603 unrelated, adult SCD patients. 21 haplotype-tagging SNPs in DNMT3A (n=17) and PRMT5 (n=4) were genotyped in the patients. SNPs were evaluated for association with HbF levels, as well as the occurrence of severe pain and narcotic use, which were used as surrogates for non-hemoglobin effects of HU. HbF levels were measured when the patients were at steady state and not taking HU. Severe pain was defined as a history of hospitalization for pain within a 12 month period. Narcotic use was defined as taking long- or short-acting narcotics to alleviate pain at home. To evaluate associations between the SNPs and clinical outcomes, we used linear and logistic regression (SAS Systems, Cary, NC), while controlling for age, sex, degree of European ancestry and HU treatment. European ancestry was estimated from 4331 genome-wide ancestry informative markers (Ashley-Koch et al, 2011). HbF levels were log-transformed prior to analysis in order to achieve a normal distribution. We corrected for multiple comparisons using the Li and Ji method (Li & Ji, 2005).

Two SNPs in DNMT3A were significantly associated with the occurrence of severe pain and HbF levels after correcting for multiple testing. Additionally, two SNPs in DNMT3A and one in PRMT5 were nominally associated with severe pain, two SNPs in DNMT3A were nominally associated with narcotic use, and one SNP in PRMT5 was nominally associated with HbF levels. These findings demonstrate that genetic variation in key genes responsible for the regulation of DNA methylation at the γ-globin locus is associated with SCD clinical outcomes.

We next hypothesized that the variability in clinical outcomes might be due to epigenetic variation at the γ-globin locus. In order to determine specific DNA methylation patterning of γ-globin genes (HBG1 and HBG2), we performed pyroseqencing subsequent to bisulfite treatment of DNA from a subset of 72 SCD patients, all of whom were HbSS, and 50% of whom were not on HU treatment. Percent methylation at the γ-globin locus was nominally associated with HbF levels among patients not taking HU (p=0.03), but not among patients taking HU. Additionally, we investigated whether SNPs predicted methylation status and identified two SNPs in DNMT3A (rs734693 and rs7583409) to be predictive of methylation status at the γ-globin locus.

These data provide further insight into the complex regulation of the γ-globin locus and suggest that genetic variation in DNMT3A and PRMT5 is associated with clinical outcomes and methylation status in SCD patients. Future studies are needed to further investigate the impact of epigenetic processes as a potential mechanism for HbF expression and induction.