Isolated Changes in Chromatin Accessibility and Enhancer-Promoter Contacts at the β-Globin Locus Distinguish Fetal Hemoglobin Producing F-Cells from a-Cells

Reversal of the developmental switch from fetal (HbF, α ₂γ ₂) to adult (HbA,α ₂β ₂) hemoglobin is an important therapeutic approach for sickle cell disease (SCD) and β-thalassemia. It has been noted since the 1950s that a small number of circulating red blood cells, called F-cells, produce elevated levels of HbF; these cells are resistant to sickling and are present in increased numbers in patients with SCD and those treated with pharmacological HbF inducers such as hydroxyurea. Because successful therapy for SCD requires increasing the number of F-cells, it is imperative to understand how these cells arise. This can potentially occur through a shift towards a global fetal-like program, selective variation in levels of known HbF silencers such as BCL11A or LRF, or through discrete epigenetic changes at the β-globin locus. We previously began to address this clinically important question using a novel experimental approach of sorting cultured primary human erythroblasts into HbF-high (F-cell) and HbF-low (A-cell) populations (Khandros et al, Blood 2020). We showed that surprisingly, F-cells from healthy donor primary erythroid cultures have minimal transcriptional differences with A-cells. Unexpectedly, this was also the case when comparing responders (F-cells) and non-responders (A-cells) to treatment with the HbF inducers pomalidomide and hydroxyurea, and there were no differences in the expression of known HbF regulators. We therefore hypothesize that HbF synthesis in F-cells is determined by epigenetic variation confined to the β-globin locus (and not by global changes in the cell fate or nuclear milieu).

To test this hypothesis, we compared genome wide chromatin accessibility by Assay for Transposase-Accessible Chromatin (ATAC-seq) in differentiation stage-matched F- and A-cells from healthy donor primary erythroid cultures, treated with vehicle, hydroxyurea, or pomalidomide. We observed striking similarities between F- and A-cells: out of 83,295 peaks called across all conditions, a mere five regions of differential accessibility were found, all at the β-globin locus (at the promoters and 3' UTR regions of the HBG1 and HBG2 genes as well as the BGLT3 non-coding RNA and HBBP1 pseudogene). This remarkable similarity in the global chromatin landscape between A- and F-cells cements the notion that these cells are fundamentally the same in terms of developmental and differentiation states, and that local epigenetic variation at the β-globin locus underlies the differences in HbF production. We also found that the gains in ATAC signal at the HBG1/2 genes were the most pronounced in F-cells from pomalidomide treated cultures, consistent with our finding that F-cells that arise following pomalidomide treatment have a higher content of HBG1/2 transcripts per cell. Drug treatments led to a larger number of changes in ATAC-seq peaks, at 123 and 1015 sites for treatment with hydroxyurea or pomalidomide, respectively, compared to vehicle. However, since differences at only 5 ATAC-seq peaks were observed between between F- and A-cells, we infer that the broader changes upon drug treatment are not needed for the phenotypic differences between F- and A-cells.

Since transcription of the β-type globin genes is controlled by developmental stage-specific long-range contacts between the gene promoters and the locus control region (LCR), we determined whether the increase chromatin accessibility at the γ-globin genes in F-cells was associated with enhanced contacts with the LCR. Capture-C experiments revealed increased LCR-HBG1/2 promoter contacts and reduced LCR contacts with the adult HBB and HBD promoters in F-cells vs A-cells, demonstrating that local gains in chromatin accessibility are linked to long-range enhancer promoter contacts. Additionally, we did not detect differences in long-range chromatin contacts at several developmentally regulated genes, including LIN28B and BCL11A, solidifying the idea that γ-globin production in F-cells is specified locally through chromatin accessibility and chromatin architecture.

In sum, our studies demonstrate that in adults, F-cells do not arise through reversion to a fetal like state or variation in expression of any known HbF regulator. Rather these cells reflect highly localized, perhaps stochastic modulation of chromatin architecture at the β-globin locus.