Red blood cell development

We are characterizing many novel genes that are important for terminal stages of erythropoiesis, including gene induction and repression, chromatin condensation, and enucleation. Another major focus is identifying genes and extracellular signals that regulate the self- renewal, proliferation, and differentiation of early (BFU-E) erythroid progenitor cells; extracellular signals include activators of the glucocorticoid and PPARα receptors and oxygen, and inhibition of TGFβ signaling. This work has led to the characterization of several molecules, including several that are FDA-approved drugs for other indications, that show great promise as therapeutics for bone marrow failure disorders and erythropoietin- resistant anemias. This research involves extensive computational analyses of large datasets of gene expression profiles and chromatin modifications generated from cells at different stages of human and mouse red cell development, coupled with high throughput CRISPR screens for genes required during erythroid development.


Erythropoietin (Epo), the principal regulator of red blood cell production, is produced by the kidney in response to low oxygen pressure in the blood. Epo binds to Epo receptors on the surface of committed erythroid CFU-E progenitors, blocking apoptosis (programmed cell death), their default fate, and triggering them to undergo a program of 4 – 5 terminal erythroid cell divisions and differentiation. We showed that the first two cell divisions, concomitant with differentiation from CFU-Es to late basophilic erythroblasts, are highly Epo-dependent; differentiation beyond this stage, involving chromatin condensation, ~1-2 terminal cell divisions, and enucleation, is no longer dependent on Epo but is enhanced by adhesion of the cells to a fibronectin matrix. Following condensation of chromatin and subsequent enucleation reticulocytes (immature red cells) are released into the blood. During this terminal differentiation 400 – 500 genes are induced; over many years we have identified many novel transcription factors and cofactors, splicing protein, chromatin modifying proteins, and enzymes in metabolic pathways including heme biosynthesis, that are essential for red cell formation.

An earlier committed erythroid progenitor, the burst- forming unit erythroid (BFU-E), can divide and generate additional BFU-Es (that is, undergo self- renewal) or diviide and generate later Epo- dependent CFU-E progenitors. Several cytokines and hormones are known to support BFU-E proliferation and formation of CFU-Es, including stem cell factor (SCF, the ligand for the c-kit protein tyrosine receptor) as well as IL-3, IL-6, and IGF-1. However, prior to our research regulation of BFU-E proliferation, self-renewal, and differentiation during basal and stress conditions was not well understood.

We are focusing much effort in this important area because of the clinical observation that many patients with bone marrow failure disorders such as Diamond-Blackfan anemia are helped by glucocorticoids (GCs) rather than by Epo treatment. These patients already have very high Epo levels in the blood, but do not have sufficient Epo-responsive CFU-E cells in the bone marrow to support life.

Culture systems that supports normal expansion and terminal differentiation of human hematopoietic stem/progenitor cells

There are many published cell culture systems for expanding human hematopoietic stem and progenitor cells such that they generate hemoglobin- containing nucleated red cell progenitors. But invariably these fail to undergo normal terminal differentiation, condense their nuclei, and expel the nucleus from the cell. Sherry Lee recently developed a 21-day four- stage culture system that supports synchronized erythroid expansion, terminal differentiation, and enucleation of mobilized human peripheral blood CD34+ stem and progenitor cells. There is a 15,000 to 30,000 fold cell expansion, equal to 14 to 15 cell doublings. At the end of the culture all of the cells were hemoglobinized and over 45% had undergone enucleation. Each enucleated reticulocyte contained ~30 pg hemoglobin, similar to the amount in each human red blood cell. The hemoglobin composition in these cells was the same as in adult human red cells, and the enucleated reticulocytes averaged 6.7 pg hemoglobin, similar to the amount in normal red blood cells and reticulocytes.

More recently Novalia Pishesha, Nai-Jia Huang, and Jiahai Shi have improved this system, using a lower serum concentration supplemented with human plasma. Differentiation is highly synchronized and cell expansion is over 70,000 fold. Over 90% of the cells undergo enucleation and the resulting reticulocytes contain the normal amount of 30 pg hemoglobin per cell. These enucleated cells survive for several days when transfused into immune- compromised mice that accept human cell transplants.

These culture systems enable many studies on terminal differentiation of human erythrocytes. For instance, using lentivirus vectors we have succeeded in expressing one or two foreign genes in well over 90% of the erythroid cells generated in culture. Similarly, we have used lentiviral vectors expressing shRNAs to knock down expression at will of any erythroid gene. This culture system combined with these manipulations allow unprecedented and focused genetic manipulation of red cells and their major proteins, and allow us to do many experiments on drugs and genes and their effects on red cell formation.

Corticosteroids, hypoxia and stress erythropoiesis

In situations of severe loss of red blood cells mammals respond by a process known as stress erythropoiesis (SE), in which there is increased formation of erythroid progenitors. Glucocorticoids (GCs) are known to be very potent enhancers of SE. This stimulatory effect of GCs on SE is utilized in the treatment of Diamond-Blackfan Anemia (DBA), an erythropoietin-resistant congenital red cell aplasia, but severe side effects limit its usefulness.

Findings by a former postdoc, Johan Flygare, indicate that the physiology of SE involves a stimulation of the earlier BFU-E erythroid progenitors, which when activated are able to rescue red cell production in conditions such as DBA, where erythropoietin has little effect. Johan showed that glucocorticoids stimulate self-renewal of early Epo-independent progenitor cells (burst-forming units erythroid or BFU-Es), over time increasing production of colony-forming units erythroid (CFU-E) erythroid progenitors from the BFU-E cells, and enhancing the numbers of terminally differentiated red cells. GCs do not affect CFU-E cells or erythroblasts.

Johan found that glucocorticoids induced expression of ~86 genes more than 2- fold. Computational analyses indicated that, of all transcription factors, binding sites for hypoxia-induced factor 1 alpha (HIF1α) were most enriched in the promoter regions of these genes, suggesting that activation of HIF1α may enhance or replace the effect of glucocorticoids on BFU-E self-renewal. Johan recently established his own research group at the Lund Stem Cell Center in Sweden.

Indeed, two current postdoctoral fellows, Sherry Lee and Xiaofei Gao showed that two clinically-tested specific inhibitors of the prolyl hydroxylase that regulates HIF1α activation synergize with corticosteroids to stimulate both human and mouse BFU-E self renewal. We propose and are testing a physiological model of stress erythropoiesis where increased levels of GCs –systemic stress hormones - and reduced oxygen – local stress - help stimulate self-renewal of the earliest erythroid BFU-E progenitors, increase CFU-E output, and at the same time stimulate terminal differentiation, thus promoting both a rapid and long-lasting increase in red blood cell production. Also, PHI-induced stimulation of BFU-E progenitors represents a conceptually new therapeutic window for treating Epo-resistant anemias.

Identifying potential drugs for Epo-resistant anemias that stimulate BFU-E self-renewal and increase production of red blood cells.

Sherry Lee and Xiaofei Gao next tested whether known pharmaceuticals that are either agonists or antagonists of other nuclear receptors affect BFU-E self-renewal and could potentially be used as new therapeutics for anemias that are not treatable by Epo. Using first the mouse fetal liver BFU-E culture system we developed and then our new ex vivo human CD34+ erythroid culture system, they found that two clinically-tested agonists of the peroxisome proliferator-activated receptor alpha (PPARα), fenofibrate and GW7647, synergize with glucocorticoids to promote BFU-E self-renewal and over time greatly increase red cell production. Genome-wide gene expression analyses both in control and corticosteroid- treated mouse BFU-E cells showed that PPARα occupies many chromatin sites that are in close proximity to those occupied by the glucocorticoid receptor (GR), indicating that the GR and PPARα function cooperatively to regulate gene expression. In particular the GR and PPARα together activate PPARα gene expression, leading to a feed-forward circuit enhancing BFU-E self-renewal.

While PPARα-/- mice show no hematological difference from wild-type mice in both normal and phenylhydrazine (PHZ)-induced stress erythropoiesis, PPARα agonists facilitate recovery of wild-type mice, but not PPARα-/- mice, from PHZ-induced acute hemolytic anemia. The mutant "Nan" (neonatal anemia) mouse has a single amino acid substitution in the erythroid- important transcription factor EKLF (KLF), which abrogates the DNA-binding capacity of EKLF to certain target genes. Heterozygotes (Nan/+) survive with a life-long, intermediate- to- severe hemolytic anemia, displaying many features of hereditary spherocytosis. With the assistance of Russell Elmes, Xiaofei and Sherry showed that both fenofibrate and GW7647 stimulated red cell formation in these mice, raising the level of red cells to almost normal. Both also caused an increase in the numbers of splenic BFU-E progenitors, suggesting that as expected these increase erythroid output via promoting BFU-E self-renewal.

Xiaofei Gao and Sherry Lee also use single cell RNA sequencing technology to dissect the heterogeneity of BFU-E cells. Erythroid progenitor BFU-Es are so-named based on their ability to generate in methylcellulose culture large colonies of erythroid cells that consist of “bursts” of smaller erythroid colonies derived from the later CFU-E Epo- dependent progenitors. “Early” BFU-E cells forming large BFU-E colonies presumably have higher capacities for self-renewal than do those forming small BFU-E colonies. In order to understand the mechanism underlying this heterogeneity, Xiaofei and Sherry conducted single cell transcriptome analysis on BFU-E cells purified from mouse embryos. Their analyses showed that there are two distinct subgroups of mouse BFU-E cells and that the Type III TGFβ receptor (TGFβ RIII) is a potential marker to distinguish “early” and “late” BFU-Es. Expression of TGFβ RIII is correlated with GATA1, a gene encoding an erythroid transcription factor induced during the BFU-E to CFU-E transition. The 10% of the BFU-E population (TGFBR310%lo) expressing the lowest amount of surface TGFβ RIII is indeed enriched for early BFU-Es, and is significantly more responsive to glucocorticoid stimulation as compared to the total BFU-E population. Thus, the TGFBR310%lo BFU-E population presumably represents earlier BFU-Es with maximal capacity for self-renewal. In addition, signaling by the TGFβ receptor kinases RI and RII is increased in TGFBR310%Hi BFU-E cells. Importantly, blocking TGFβ signaling by TGFβ inhibitors increase TGFBR310%Hi BFU-E cells self-renewal and increase total erythroblast production, suggesting the use of this type of drugs in treating EPO untreatable anemias.

Understanding erythroid self-renewal and fate commitment using single-cell RNA sequencing

While the transcriptomes of BFU-Es have been mapped comprehensively, the means by which they balance the decision to self-renew or differentiate still remains to be elucidated. For example, as mentioned, under the influence of corticosteroids, BFU-Es can divide 3-4 times more before differentiating to CFU-Es. Anirudh Natarajan is isolating BFU-Es and monitoring their transcriptomes during differentiation and self-renewal in culture. Specifically, collaborating with the Broad Institute, he is using single-cell RNA-seq to characterize the transcriptomes of BFU-Es, CFU-Es and the transitions in culture with and without addition of corticosteroids. Our first phase using single-cell RNA-seq followed by the use of dimensionality reduction techniques such as PCA and t-SNE reveal heterogeneity in the BFU-E population. However, as judged by RNA sequencing the CFU-E population is largely homogenous reflecting their functional homogeneity observed in MethoCult assays.

In order to capture the transcriptome during the differentiation process, we will use single-cell RNA-seq followed by computational analysis designed for time-series analysis, such as Monocle or Wishbone. These approaches will allow us to exploit the variability of the cells as the progress through the differentiation process and help us generate a fine map of the transcriptional cascades involved. These experiments will help us understand the nature of BFU-E self-renewal and help identify regulators of the fate commitment process.

Pathogenic mutant JAK2 V617F stimulates proliferation of erythropoietin- dependent erythroid progenitors and delays their differentiation by activating non-erythroid signaling pathways

JAK2 is a protein tyrosine kinase activated by the Epo receptor and several other cytokine receptors. JAK2-V617F is a mutant activated JAK2 kinase found in most polycythemia vera (PV) patients and those with other myeloproliferative disorders, namely Essential Thrombocytosis (ET) and Primary Myelofibrosis (PMF). Several years ago we showed that the mutation enables cytokine-independent activation of JAK2 in cells that express a homodimeric cytokine receptor such as the erythropoietin receptor (EpoR) or related receptors including those for thrombopoietin and G-CSF. JAK2-V617F skews lineage determination of hematopoietic stem and progenitor cells towards the erythroid lineage and increases the number of erythroid progenitors. This leads to overproduction of red cells, consistent with the high percentage of erythroid progenitors from most PV patients that express JAK2-V617F. However, in some PV patients JAK2-V617F is found in only 10-30% of erythroid progenitors, implying that JAK2-V617F might also stimulate terminal erythropoiesis after the erythropoietin (Epo) dependent CFU-E stage.

To confirm this hypothesis, Jiahai Shi and Wenqian Hu showed that expression of JAK2-V617F in murine CFU-Es allows then to divide ~6 rather than the normal ~4 times in the presence of Epo, initially increasing the numbers of CFU-Es and delaying cell cycle exit. Expression of genes promoting DNA replication continues in these JAK2-V617- expressing cells for 2 divisions longer than normal. Over time the number of red cells formed from each CFU-E is increased ~4 fold; similar to human PV pathology. JAK2-V617F erythroid progenitors eventually differentiate to normal enucleated cells with an mRNA composition very similar to that of normal mouse reticulocytes. Microarray analyses comparing JAK2 and JAK2-V617F erythroblasts indicate that JAK2-V617F not only activates EpoR-JAK2 signaling pathways, but also transiently induces non-erythroid-signaling pathways. Jiahai showed that purified fetal liver Epo- dependent progenitors express many cytokine receptors additional to the EpoR, as well as Stat1 and Stat3 in addition to Stat5, the only STAT normally activated by the Epo receptor and JAK2. JAK2-V617F triggers activation of Stat1 and Stat3, and inhibition of Stat1 by a drug blocks Jak2 V617F mediated erythropoiesis, but does not affect normal erythropoiesis. This abnormal activation of Stat1 and Stat3 leads to transient induction of many genes not normally activated in terminally differentiating erythroid cells and that are characteristic of other hematopoietic lineages. Jiahai hypothesizes that these non-erythroid-signaling pathways delay terminal erythroid differentiation and permit extended numbers of cell divisions. These results provide a more complete understanding of PV pathogenesis, in particular in patients in with low numbers of JAK2-V617F expressing erythroid progenitors. Jiahai recently established his own laboratory at the City University of Hong Kong.

Current postdoctoral fellow Richard Voit is extending Jiahai’s work from mouse hematopoietic progenitors to investigate the role of JAK2 mutations in the terminal differentiation of primary human CD34+ hematopoietic stem cells. Preliminary work demonstrates that overexpression of JAK2-V617F in CD34+ cells leads to a 2-fold increase in the number of CFU-Es and terminally-differentiated erythrocytes. Ongoing research is focused on investigating the differential gene expression profile of human JAK2-V617F erythrocyte progenitors compared to controls and on the role of JAK2 mutations on megakaryocyte differentiation in vitro.

Transcriptional control of gene expression during terminal erythroid differentiation: Thyroid hormone receptor beta and NCOA4

In the vertebrate, late erythroblasts must undergo terminal differentiation, which involves terminal cell cycle exit and chromatin condensation, to become reticulocytes. In mammals, there is an additional step requiring extrusion of the pycnotic nucleus via an asymmetric cell division. Many aspects of transcriptional regulation of this process remain unknown. Previous RNA sequencing studies on late erythroblasts identified several genes encoding DNA- binding proteins whose expression is up-regulated during terminal differentiation.

AAn effect of thyroid hormone on erythropoiesis has been known for more than a century but the molecular mechanism(s) by which thyroid hormone affects red cell formation have been elusive. Recently Xiaofei and and Sherry demonstrated an essential role of thyroid hormone during terminal human erythroid cell differentiation; specific depletion of thyroid hormone from the culture medium completely blocked mouse and human erythroid differentiation. Genome wide analysis showed that thyroid hormone receptor β (TRβ) occupies many gene loci encoding transcripts abundant during terminal erythropoiesis, including globin genes, and cooperates with GATA-1 and RNA polymerase II (Pol II) to regulate their expression. TRβ agonists stimulated red cell formation in Nan/+ mice, raising the level of red cells to normal; these agonists also accelerated erythroblast differentiation from the BFU-E stage in vitro, likely by reducing BFU-E self renewal.

To identify factors that cooperate with TRβ during human erythroid terminal differentiation, they conducted RNA-Seq in human reticulocytes and identified nuclear coactivator 4 (NCOA4) as a critical regulator of terminal differentiation. Furthermore, Ncoa4-/- mice are anemic in both the embryonic and perinatal periods and fail to respond to thyroid hormone by enhanced erythropoiesis. Genome wide analysis suggested that thyroid hormone promotes NCOA4 recruitment to chromatin regions that are in proximity to Pol II and are highly associated with transcripts abundant during terminal differentiation. Additionally, knocking down NCOA4 interrupted the terminal differentiation in both our human CD34 ex vivo differentiation system and in cultured mouse fetal liver cells. Collectively, their results reveal the molecular mechanism of thyroid hormone function in accelerating terminal red blood cell formation and are potentially useful to treat certain anemias.

Identification and characterization of novel genes essential for mouse terminal erythroid differentiation

Transcription factors essential for terminal erythroid differentiation, including GATA1 and SCL/Tal1, activate and repress multiple downstream genes, including activation of their own expression. Our previous research has analyzed the global expression changes during mouse terminal erythroid differentiation with RNA-seq and created a dataset of more than 500 specifically induced genes. The role of the vast majority of these genes in erythroid differentiation has not been explored. Huan Yang, and Hojun Li are identifying novel terminal erythroid regulatory genes in this dataset via CRISPR/Cas9-based high throughput screening. In collaboration with Dr. John Doetch, Associate Director of Broad Institute's Genetic Perturbation Platform, they are making a retroviral CRISPR library in which guide RNAs are designed with optimized strategy to maximize activity and minimize off-target effects. They will express this CRISPR library in erythroid BFU-E progenitor cells and then use our well-established mouse erythroid differentiation system for screening genes essential for terminal differentiation. They will then characterize novel important genes in both cultured erythroid cells and genetically modified mice.

Transcriptional divergence and conservation of human and mouse erythropoiesis

Mouse models have been used extensively for decades and have been instrumental in improving our understanding of mammalian erythropoiesis. Nevertheless, there are several examples of variation between human and mouse erythropoiesis. In collaboration with Vijay G. Sankaran, a recent postdoc and currently an Assistant Professor at Boston Children's Hospital, Nova Pishesha performed a comparative global gene expression study using publicly available data from morphologically identical stage-matched sorted populations of human and mouse erythroid precursors from early to late erythroblasts. Surprisingly, they found that, at a global level, there is a significant extent of divergence between the species, both at comparable stages and in the transitions between stages. This was especially the case for the 500 most highly expressed genes during development, save for some major transcriptional regulators of erythropoiesis and major erythroid-important proteins. This suggests that the response of multiple developmentally regulated genes to key erythroid transcriptional regulators represents an important modification that has occurred in the course of mammalian evolution. They further developed this compendium of data as a systematic framework that is very practical and useful to understand and study conservation and divergence between human and mouse erythropoiesis as well as to help translate findings from mouse models to potential therapies for human disease.

Translational control of red cell development

Wenqian Hu investigated how mRNA-binding proteins regulate erythroid terminal differentiation. He characterized one such protein, Cpeb4, that is required for terminal erythropoiesis. Specifically, he found that Cpeb4 is dramatically induced during erythroid terminal differentiation by the two erythroid-important transcription factors, GATA1 and Tal1. Knocking down Cpeb4 inhibited this cell differentiation process. Interestingly, Cpeb4 interacts with eIF3, a general translation initiation factor, to repress the translation of a large set of mRNAs in terminal differentiating erythroblasts, including its own mRNA. Thus, transcriptional induction synchronizes with translational repression to maintain Cpeb4 protein within a specific range during terminal erythropoiesis; this precise control of gene expression is required for normal cell differentiation.

Wenqian recently established his own independent laboratory as an Assistant Professor at the Mayo Clinic where he is continuing these investigations. In collaboration with our PhD student Juan Alvarez-Dominguez, he identified a group of mRNA-binding proteins that are specific to erythroid cells and that are dramatically induced during terminal erythropoiesis. Currently, Wenqian is characterizing whether and how these mRNA-binding proteins regulate terminal erythroid differentiation.

Wenqian and Juan also profiled gene expression and ribosome density genome-wide through several stages of erythroid differentiation, and found pervasive translational control of protein production. Translating ribosomes discriminate coding versus noncoding transcripts. Hundreds of protein isoforms are generated via upstream initiation and stop codon read-through, and upstream open reading frames are widely used to attenuate translation of developmentally regulated effectors, including the hemoglobin switch regulator Bcl11A. Strikingly, translation efficiency is up-regulated to sustain synthesis of key proteins whose mRNAs become depleted in terminally differentiated/enucleated cells, including the Tal1 and Rcor1 regulators and translation initiation factors themselves. They found that the erythroid-specific Rbm38 protein is responsible for translational up-regulation of a group of such mRNAs during terminal erythropoiesis by interacting with eIF4G. Inhibition of Rbm38 blocked red cell production and mice lacking Rbm38 developed anemia, highlighting its importance for terminal erythropoiesis. These findings uncover essential roles for translational control in specialized mammalian cell formation.

Dynamics of the nuclear lamina during terminal erythropoiesis

Chromatin condensation during terminal differentiation is accompanied by proportional shrinkage in the size of the nucleus. The nuclear lamina is composed of the fibrous proteins nuclear lamin A/C and nuclear lamin B, and forms a dense fibrillar network on the inside the nuclear envelope that provides structural support to the nucleus. Jiahai Shi and undergraduate Heejo Choi showed that the expression of these three lamins initially increased and then decreased markedly during terminal erythroid development, and that the dynamic expressions of Lamin A/C and Lamin B controls the thickness of the nuclear lamina. Unlike the gradual decrease in nuclear size, the nuclear lamina increases dramatically in the early stages of terminal erythropoiesis, followed by a sudden decrease at the end, as revealed both by western blotting and immunofluorescence confocal microscopy. Jiahai and Hojun Li showed that knock-out of the Lamin A/C gene using CRISPR-mediated gene disruption dramatically decreased proliferation and hemoglobinization of mouse fetal liver-isolated erythroid progenitor cells. These results indicate that enhanced expression of lamin A/C is important for erythroid terminal differentiation. Jiahai is continuing to explore the functional role of the up-regulation of the nuclear lamina in terminal erythropoiesis in his own laboratory as an Assistant Professor at the City University in Hong Kong.

Modeling disorders of erythropoiesis in primary human and mouse erythroid cells

Ongoing work by postdoctoral fellow Hojun Li, is attempting to create models of anemia through genome editing in primary erythroid progenitor cells. Hojun Li, Jiahai Shi, and Jenn Eng developed a method for high efficiency genome editing in mouse fetal liver erythroid progenitor cells using a retroviral vector to deliver the components of the CRISPR-Cas9 nuclease system. They demonstrated that over 50% of the cells in a population of fetal liver erythroid progenitors can be successfully transduced by this vector and that gene knockout results in loss of protein expression. This is the highest to-date reported efficiency of CRISPR-Cas9 mediated gene knockout in isolated primary cells, They have been able to replicate the phenotype of the impaired hemoglobin switch in Bcl11a knockout mice by using CRISPR-Cas9 to disrupt the Bcl11a gene, and have gone on to show developmental phenotypes for knockout of other genes that cannot be knocked down successfully in fetal liver culture. In particular they used this method to demonstrate a novel requirement for Lamins A and C in erythroid differentiation Hojun worked with Jenn to adapt CRISPR-Cas9 genome editing to our human erythroid culture system in order to introduce known mutations associated with human anemias for the purpose of modeling the effects of these mutations on human erythroid development. As noted above, Hojun and Huan Yang are utilizing the CRISPR-Cas9 mediated gene knockout system for high-throughput genetic loss of function screening in primary isolated erythroid progenitor cells.