In this issue of Blood, Qin et al1 show that BMI1 (PCGF4), a polycomb group RING finger protein, is a repressor of fetal hemoglobin (HbF). They present strong evidence that the entirety of the effect of BMI1 is exerted by RNA-binding proteins LIN28B, IGF2BP1, and IGF2BP3.
Elevated HbF in adults moderates the symptoms of sickle cell disease and β-thalassemia and is a highly desirable therapeutic goal.2 The significant discoveries of HbF repressors BCL11A and ZBTB7A (LRF) resulted in encouraging small clinical trials involving gene editing of mobilized patient hematopoietic stem cells ex vivo.3 However, gene therapies are unlikely to be feasible for the vast patient population in less developed parts of the world because of their complexity and cost. Thus, there remains a strong incentive to develop other approaches, such as small-molecule inhibitors. The feasibility of this approach is illustrated by the rational targeting of components of the nucleosome remodeling and deacetylating complex, through which BCL11A and ZBTB7A work.4 Qin et al enlarge the field of potential small-molecule targets by identifying polycomb repressive complexes 1/2 (PRC1/2) as indirect inhibitors of HBG1/2 transcription.
PRC1 and PRC2 are crucial to establishment and maintenance of facultative heterochromatin and have fundamental roles in gene regulation and development.5 PRC1 catalyzes histone H2A lysine 119 monoubiquitylation (H2AK119ub1) through RING1A/B E3 ligase, whereas PRC2 catalyzes histone H3 lysine 27 trimethylation (H3K27me3) through EZH1/2 methyltransferase activity. PRC1 also contains 1 of 6 polychrome group RING finger proteins (PCGF1-6). These are considered canonical (cPRC1) or noncanonical (ncPRC1), depending on whether they contain a CBX protein that recognizes H3K27me3-marked chromatin, making the complex dependent on PRC2 (cPRC1), or instead contain RYBP or YAF1, and bind chromatin independent of PRC2 (ncPRC1). Modestly elevated HBG mRNA was observed after depleting PRC2 subunit EZH2 in primary adult erythroblasts.6 Recently, a pharmacologic inhibitor of PRC2 component EED (FTX-6058) was reported to reduce BCL11A and robustly elevate HbF and is in clinical trials.7 However, our understanding of PRC1/2 involvement in HBG silencing is incomplete.
Qin et al performed a domain-focused screen targeting human E3 ligases in umbilical cord erythroid progenitor-derived HUDEP2 cells, which have low basal expression of HbF. BMI1 (PCGF4, a PRC1 component) was a strong hit with BMI1-edited cells among the highest HbF expressors. BMI1 reduction in HUDEP2 cells and in primary adult erythroblasts robustly activated transcription of the fetal globin genes HBG1 and HBG2, HbF protein, and the percentage of cells with HbF. Thus, BMI1 is revealed as a repressor of HbF in adult erythroid cells.
RNA sequencing confirmed HBG1/2 activation in BMI1-depleted HUDEP2 cells. Interestingly, IGF2BP1 and LIN28B were upregulated in these cells. Overexpression of these RNA-binding proteins had been shown to increase HbF in adult erythroid cells, and interference with BCL11A mRNA translation or turnover was proposed to explain the result.8,9 In the current studies, increased LIN28B and IGF2BP1 correlated with reduced BCL11A protein after BMI1 depletion, and overexpression of BCL11A largely restored HBG silencing. Moreover, combinatorial depletion of BMI1, LIN28B, and IGF2BP1 lowered HbF to essentially basal levels, supporting that upregulation of LIN28B and IGF2BP1 was sufficient to account for HbF activation by BMI1 loss. Unexpectedly, RNA sequencing of primary adult erythroblasts after BMI1 depletion revealed LIN28B was unchanged but IGF2BP1 and IGF2BP3 were upregulated. Overexpression of either IGF2BP1 or IGF2BP3 increased HBG expression, supporting that in BMI1-depleted primary cells, these 2 proteins drive HbF production.
To find direct targets of BMI1, the authors performed CUT&RUN (cleavage under targets and release using nuclease) to localize BMI1 in HUDEP2 cells and in primary adult erythroblasts, and profiled H2AK119ub1 and H3K27me3 in control and BMI1-depleted cells. Notably, there was no enrichment of BMI1 or H2AK119ub1/H3K7me3 at the β-globin locus, consistent with BMI1 acting indirectly to repress HBG1/2. H2AK119ub1 and H3K27me3 overlapped with BMI1 at the CpG islands near the LIN28B and IGF2BP1 promoters in HUDEP2 cells and were reduced on BMI1 loss. The same was true for IGF2BP3 in primary cells, indicating that BMI1 maintains repression of chromatin at these loci in the respective cell systems.
To distinguish whether cPRC1 or ncPRC1 was involved in HbF repression, the authors performed multiplex CRISPR targeting. Indeed, combined depletion of cPRC1 subunits CBX2/4/8 in HUDEP2 cells activated LIN28B and IGF2BP1 and induced a fetal erythroid gene expression pattern, like BMI1 depletion. The authors propose a model in which cPRC1/BMI1/CBX represses LIN28B and IGF2BP1 through H3K27me3 recognition, invoking PRC2 involvement (see figure). Supporting PRC2 participation, after EZH2 depletion in HUDEP2 cells, there was significant HBG1/2 upregulation, and modest upregulation of IGF2BP1 and LIN28B. Furthermore, in differentiating primary erythroid cells, the EZH2 inhibitor, EPZ-6438, reduced bulk H3K27me3 and upregulated IGF2BP1/3 mRNA and protein. Thus, in both cellular backgrounds, cPRC1 and PRC2 repress HbF through developmental silencing of LIN28B, IGF2BP1, and/or IGF2BP3.
RNA binding proteins LIN28B and IGF2BP1/3 link PRC1/2 to HbF repression in adult erythroid cells. cPRC1 and PRC2 components are depicted as differently colored shapes, with the named components that were tested by Qin et al labeled. Silencing of LIN28B and IGF2BP1/3 in HUDEP2 cells or in adult primary cells is depicted by gray type. Depletion of BMI1 and CBX (cPRC1) or EZH2 (PRC2) relieves silencing of these RNA-binding proteins, thereby promoting silencing of BCL11A (gray type) or other events and upregulation of HbF. Professional illustration by Patrick Lane, ScEYEnce Studios.
RNA binding proteins LIN28B and IGF2BP1/3 link PRC1/2 to HbF repression in adult erythroid cells. cPRC1 and PRC2 components are depicted as differently colored shapes, with the named components that were tested by Qin et al labeled. Silencing of LIN28B and IGF2BP1/3 in HUDEP2 cells or in adult primary cells is depicted by gray type. Depletion of BMI1 and CBX (cPRC1) or EZH2 (PRC2) relieves silencing of these RNA-binding proteins, thereby promoting silencing of BCL11A (gray type) or other events and upregulation of HbF. Professional illustration by Patrick Lane, ScEYEnce Studios.
PRCs propagate repressive marks through cell division, prompting the authors to examine the effects of interfering with their activity on HbF induction over time. HUDEP2 cells were induced to differentiate after treatment with either the EZH2 inhibitor EPZ-6438 or the experimental HbF inducer pomalidomide for 3 days or were grown absent the drug for an additional 5 days before induction. Strikingly, HbF induction by EPZ-6438 was sustained even when cells underwent 5 additional days of growth in the absence of the drug before differentiation, but this was not the case for pomalidomide. This potential epigenetic effect has implications for possible future applications of specific therapeutic polycomb group-targeted HbF inducers.
Overall, these new studies provide significant insight into how BMI1 and PRC1/2 function in hemoglobin switching and enlarge the armamentarium of potential therapeutic targets to increase HbF. The data point to BCL11A as a link between BMI1 and HbF repression through BMI1 targets LIN28B and IGF2BP1/3: how these RNA-binding proteins affect BCL11A still has open questions. Epigenetic manipulation by PCR1/2 emerges as a viable therapeutic option to increase HbF. The observation that the effects of EPZ-6438 seem to be sustained after removal of the drug is promising and encouraging.
Conflict-of-interest disclosure: The author declares no competing financial interests.
This feature is available to Subscribers Only
Sign In or Create an Account Close Modal