Heme-Regulated Inhibitor (HRI) Activates Transcription Factor ATF4 to Promote BCL11A Transcription and Fetal Hemoglobin Silencing

Reactivation of fetal hemoglobin in adult red blood cells benefits patients with sickle cell disease and β-thalassemia. BCL11A is one of the predominant repressors of fetal γ-globin transcription and stands as an appealing target for therapeutic genome manipulation. However, pharmacologic perturbation of BCL11A function or its co-regulators remains an unmet challenge. Previously, we reported the discovery of the erythroid-enriched protein kinase HRI as a novel regulator of γ-globin transcription and found that HRI functions in large part via controlling the levels of BCL11A transcription (Grevet et al., Science, 2018). However, the specific mechanisms underlying HRI-mediated modulation of BCL11A levels remain unknown.

To identify potential HRI-controlled transcription factors that regulate BCL11A, we performed a domain-focused CRISPR screen that targeted the DNA binding domains of 1,447 genes in the human erythroid cell line HUDEP2. Activating transcription factor 4 (ATF4) emerged as a novel γ-globin repressor. Prior studies reported that ATF4 production is under positive influence of HRI. Specifically, HRI phosphorylates translation factor EIF2α which in turn augments translation of ATF4 mRNA. As expected, HRI deficiency reduced ATF4 protein amounts in HUDEP2 and primary erythroid cells. We further found that the degree of γ-globin reactivation was similar in ATF4 and HRI-depleted cells.

ATF4 ChIP-seq in both HUDEP2 and primary erythroblast identified 4,547 and 3,614 high confidence binding sites, respectively. Notably, we did not observe significant enrichment of ATF4 binding or even the presence of an ATF4 consensus motif at the γ-globin promoters, suggesting that ATF4 regulates the γ-globin genes indirectly. However, ATF4 specifically bound to one of the three major BCL11A erythroid enhancers (+55) in both cell types. This was the sole binding site within the ~0.5Mb topologically associating domain that contains the BCL11A gene. Eliminating this ATF4 motif via CRISPR guided genome editing lowered BCL11A mRNA levels and increased γ-globin transcription. Capture-C showed that ATF4 knock-out or removal of the ATF4 site at the BCL11A (+55) enhancer decreased chromatin contacts with the BCL11A promoter. Forced expression of BCL11A largely restored γ-globin silencing in cells deficient for ATF4 or lacking the ATF4 motif in the BCL11A (+55) enhancer.

An unexplained observation from our prior study was that HRI loss did not significantly lower Bcl11a levels in murine erythroid cells. Therefore, we mutated the analogous ATF4 motif in the Bcl11a enhancer in the murine erythroid cell line G1E. Unlike in human cells, Bcl11a mRNA synthesis was decreased only very modestly, and there was no effect on the murine embryonic globin genes whose silencing requires Bcl11a. This suggests that the species specific regulation of BCL11A by HRI results from divergent functional roles of ATF4 binding at the BCL11A (+55) enhancer.

In sum, our studies uncover a major pathway that extends linearly from HRI to ATF4 to BCL11A to γ-globin. Moreover, these results further support HRI as a pharmacologic target for the selective regulation of BCL11A and γ-globin.