Erythropoiesis is a complex, tightly regulated process. Altered erythroid production leads to various types of anemia and hemoglobin disorders including thalassemia syndrome, inherited bone marrow failure, chronic anemia, polycythemia vera, and sickle cell anemia (Zhang Y. et al., 2022). Billions of people are affected by various erythrocyte-related diseases worldwide, posing a significant public health challenge (Srivastava et al., 2022). Erythrocyte transfusion is indispensable and irreplaceable not only for patients with anemia and hemoglobin disorders, but also for modern medical practices, such as cancer treatment and surgery (Lee et al., 2023). However, the whole process, from blood collection to clinical use, consumes a significant amount of resources, and there are many problems to solve, such as the spread of pathogenic microorganisms, the cost of the whole process, and approximately 118.5 million units of blood collected globally each year, which are insufficient to meet the annual global demand of over 300 million units for blood transfusions (Christaki et al., 2019). An aging population and emerging risks of viruses and pathogens exacerbate the problem of supply shortages. The development of alternative blood transfusion products helps mitigate erythrocyte shortages and transfusion limitations. Each unit of transfused blood contains a trillion erythrocytes. However, obtaining such large quantities of erythrocytes in vitro involve overcoming several challenges. The bottleneck of low adult β-hemoglobin expression and nucleation efficiency remains to be resolved. The regulatory mechanism during erythroid differentiation will help to further study and ultimately identify the molecular pathways involved in the pathological process of erythroid diseases and provide great promise for transfusion medicine and novel cell-based therapies.

Developing erythroid precursors absorb exceptionally large amounts of iron to accommodate increased synthesis of heme (Ryu et al., 2017; Philpott, 2020). Iron could be released from ferritin, where most of the cellular iron is stored through a process known as “ferritinophagy” (Dowdle et al., 2014; Mancias et al., 2014). Ferritinophagy is an important step for iron release before mitochondrial iron import and heme biosynthesis (Nilsson et al., 2009; Mancias et al., 2015). Ferritinophagy initiates ferroptosis by promoting iron accumulation. Ferroptosis is an active mode of cell death, defined as catalytic Fe(II)-dependent regulated necrosis accompanied by lipid peroxidation (Toyokuni et al., 2023). Ferroptosis involves embryonic hematopoiesis, particularly erythropoiesis in rats (Zheng et al., 2021). Ferroptosis is observed in the extraembryonic endodermal component of the visceral yolk sac, which induces blood precursors, and in embryonally nucleated erythrocytes that, disappear in enucleated erythrocytes (Zheng et al., 2021). Ferroptosis inhibitors significantly delay erythrocyte enucleation. Additionally, ferroptosis may mediate B-cells differentiation (Chen et al., 2022). Abnormal ferroptosis damages the development erythrocytes, leading to erythropoiesis suppression and anemia (Chen et al., 2022). Ferroptosis has new therapeutic potential for blood cell-related diseases. In many patients with heart failure, inflammation, and oxidative stress lead to an iron-deficient state, which can limit erythropoiesis in erythroid precursors (Packer, 2022). Increasing the levels of cytosolic Fe (2+) available to the mitochondria enables the synthesis of heme from erythroid precursors (Packer, 2022). However, the genes significantly related to ferroptosis in erythropoiesis are unknown, and whether ferroptosis genes can be used as biomarkers to discriminate between early and late differentiation is yet to been reported.

CD34⁺ cells isolated from umbilical cord blood (UCB) serve as a valuable model system to study gene regulation of erythropoiesis. This study obtained data on the erythrocyte differentiation (ED) of human umbilical cord blood-derived CD34⁺ cells from the Gene Expression Omnibus (GEO) database for bioinformatic analysis to screeng differentially expressed genes (DEGs) between early and late differentiation. Subsequently, the differentially expressed ferroptosis-related genes (DE-FRGs) were acquired. Their ability to diagnose the differentiation stage, biological functions, and regulatory networks was analyzed. We also identified hub genes and predicted their association with immune cell infiltration. Our findings explain the potential role of ferroptosis in erythropoiesis and may provide new directions for treating of erythrocyte-related diseases.

Anemia is a worldwide health issue and a common complication of many diseases, such as heart failure and tumors. Transfusion is a crucial treatment. However, blood supply shortages and transfusion-related risks have become global issues of major concern due to donor-related issues. Despite the substantial efforts made in vitro generation of erythrocytes using various stem cells, most protocols have low efficiency in generating sufficient functional red blood cells (RBCs), including stable β-globin expression and enucleation. Therefore, understanding the erythropoiesis mechanism during development is necessary. Erythrocytes, organelle-free cells packaged with iron-containing hemoglobin, are the end-product of a complex hierarchy of hematopoietic progenitors that become progressively restricted to the erythroid lineage. This stepwise differentiation process requires large amounts of iron for hemoglobin synthesis. Ferroptosis is involved in various physiological and pathological processes, including erythropoiesis.

In this study, we explored 753 genes involved in ferroptosis and identified 25 DE-FRGs between the early and late ED stages. Enrichment analysis indicated that DE-FRGs were primarily involved in the JAK-STAT, MAPK, PI3K-AKT-mTORC1, WNT, and NOTCH signaling pathways. Ten hub DE-FRGs were obtained based on the PPI analysis. Furthermore, we created mRNA-miRNA and mRNA-TF constructs. The infiltration levels of immune cells differed significantly during ED. Finally, we established a signature using six DE-FRGs related to ED stages through LASSO regression analysis. The RT-qPCR validation results suggested that the six DE-FRGs are potential signature genes for erythropoiesis. This study revealed that ferroptosis may provide a promising strategy for optimizing in vitro erythrocyte hematopoiesis and treating erythrocyte diseases.

This study identified the molecular signatures of ferroptosis in ED of human umbilical cord blood-derived CD34⁺ cells. We obtained 10 hub DE-FRGs with significant interactions (IL1B, HMOX1, DPP4, ATF3, YAP1, CDH2, CHAC1, DDR2, GDF15, and IFNA21). Similar to the results of our study, IL1B, HMOX1, GDF15, YAP1, and DPP4 have been reported to play roles in ferroptosis and erythroid differentiation. IL1B interacted with five FRGs. IL-1β might promote hypoferremia or decreased iron availability by stimulating the expression of hepcidin, which could impair erythropoiesis (Nemeth et al., 2004). IL-1 receptor antagonists reduced the incidence of anemia (Dinarello, 2005; Vallurupalli et al., 2020). Therefore, IL-1β may be decreased to increase iron availability and promote erythroid differentiation. Autophagy inhibits excess iron-induced ferroptosis and subsequently increases IL-1Β (Su et al., 2021). HMOX1, the inducible enzyme that catabolizes the degradation of heme into iron, plays a major role in the clearance of senescent and damaged RBCs, systemic iron homeostasis, and erythropoiesis (Starzyński et al., 2013; Kim et al., 2018). The discrepancy in sensitivity to ferroptosis can be determined and regulated by HO-1 (Li et al., 2023). The oxidative stress also promoted HO-1 translocation to mitochondria, leading to mitochondrial iron overload (Chen Y. et al., 2023). GDF-15, a bone marrow-derived cytokine, suppresses the iron regulator hepcidin in vitro and is significantly increased in patients with ineffective erythropoiesis (Liu et al., 2021). GDF-15 is the mitochondrial metabolism related markers. Those with iron overload had higher GDF15 levels compared with non-iron overload patients (Huang et al., 2019). YAP1, an erythroid regulator, coordinates metabolic status with the proliferation of erythroid progenitors to promote stress erythropoiesis (Hao et al., 2019). YAP/TAZ play an important role in erythroid maturation and enucleation (Damkham et al., 2022). Hippo/YAP1/c-Jun axis regulated iron metabolism (Zhou et al., 2023). YAP1 deficiency boosted mitochondrial dysfunction and the ferrous iron accumulation (Zhang J. et al., 2022). DPP-4 decreases erythropoietin (Epo) activity by cleavage, negatively regulates colony-stimulating factor activity and stress hematopoiesis (Broxmeyer et al., 2012; Ou et al., 2013), and improves the responsiveness to erythropoiesis-stimulating agent (Hasegawa et al., 2021). In our study, we predicted the differential expression of these genes in the early and late stages of differentiation and analyzed the pathways and immune infiltration associated with these genes in subsequent analyses. The transcription factors ATF3 and YAP1 play multiple roles in shaping ferroptosis sensitivity through either transcription-dependent or transcription-independent mechanisms (Dai et al., 2020). As there is still a lack of evidence on whether FRGs contribute to the stage of ED, a logistic regression model was constructed to distinguish between five key genes (DPP4, CDH2, CHAC1, and DDR2). The validation of the six key genes met the trend of bioinformatic analysis. However, a few studies have investigated the mechanisms of action of other genes involved in ED.

Furthermore, functional enrichment analysis indicated that the DEGs were primarily involved in the JAK-STAT, MAPK, PI3K-AKT-mTORC1, WNT, and NOTCH signaling pathways. Hypoxia promotes erythropoiesis by increasing Epo production. Although Epo is the principal regulator of erythroid progenitors, signaling from the Epo-receptor activates several pathways, including the JAK/STAT, ras/raf/MAP kinase, and PI3K/Akt cascades, to promote cell survival, proliferation, and differentiation (Ingley et al., 2004). JAK/STAT, MAPK, and PI3K are considered the main signaling pathways that play significant roles in fetal hemoglobin induction (Rahim et al., 2013). The JAK-STAT signaling pathway regulates certain TFs (GATA1, GATA2, SPI1, and RUNX1) involved in ED regulation, development, and maturation (Liu et al., 2017). Erythrocyte proliferation and survival are also associated with the activation of the JAK-STAT pathway (Cokic et al., 2012). The level of IL-1β was decreased via inactivating of JAK/STAT signaling pathway (Ni et al., 2023). The PI3K-AKT signaling pathway regulates the Epo-induced survival, proliferation, and maturation of early erythroid progenitors (Lopez et al., 2011). The reduction of Pi4ka inhibits myeloid and erythroid cell differentiation in vitro and promotes anemia in vivo through a mechanism involving the deregulation of AKT, MAPK, and JAK-STAT signaling pathways (Ziyad et al., 2018). Ferroptosis can be reducd through the MAPK signaling pathway (Chen W. et al., 2023). We previously reported the mTOR signaling pathway played roles during human umbilical cord blood-derived CD34⁺ cell erythropoiesis in vitro (Liu et al., 2020). The inhibition of mTORC1 sensitizes cells to ferroptosis (Zhang et al., 2021). The components of the JAK-STAT-NF-κB signaling pathway are DNA hypomethylated and upregulation, targeting key genes for erythropoiesis. The activation of the Wnt/beta-catenin signaling attenuates cellular lipid ROS production and subsequently inhibits ferroptosis. The beta-catenin/TCF4 transcription complex directly binds to the promoter region of GPX4, a peroxidase that suppresses ROS-triggered ferroptosis resulting in suppressing ferroptosis (Wang et al., 2022). Although Wnt/β-catenin is dispensable for steady-state erythropoiesis, its activity is essential for stress erythropoiesis in response to bone marrow injury and anemia (Krimpenfort and Nethe, 2021). The canonical WNT signaling pathway also inhibits the expansion and/or survival of primitive erythrocytes (Paluru et al., 2014). GATA2 contributes to the inhibition of the canonical WNT signaling pathway, thereby permitting progenitors to exit the cell cycle and commit to a hematopoietic fate. Subsequently, the activation of the non-canonical WNT signaling pathway plays a role in enabling progenitors to differentiate into mature RBCs (Mimoto et al., 2015). The activation of the NOTCH signaling pathway leads to the inhibition of differentiation of immature precursors, suggesting important roles for the NOTCH signaling pathway in hematopoiesis (Sugimoto et al., 2006). Our results expand our understanding of the mechanisms underlying erythropoiesis. However, broader validation is needed to improve our understanding of the mechanisms of ferroptosis in the ED of human umbilical cord blood-derived CD34⁺ cells.

Erythropoiesis is a complex and sophisticated multistage process regulated by TFs and miRNAs. This study also established TF- and miRNA-target gene networks. The importance of several TFs in erythropoiesis has been unequivocally demonstrated by cell-based ex vivo assays, as well as in knockout mouse models and rare patients with anemias (Kerenyi and Orkin, 2010). Members of the GATA TF family, GATA1 and GATA2, play crucial roles in regulating lineage-restricted genes during erythroid differentiation (Moriguchi and Yamamoto, 2014). GATA-2 is essential for the maintenance and proliferation of immature hematopoietic progenitors, whereas GATA-1 is essential for the survival of erythroid progenitors and terminal differentiation of erythroid cells (Ohneda and Yamamoto, 2002). Similarly, in our differentiation system, GATA1 expression gradually increased and then decreased, whereas GATA2 expression gradually decreased following cell maturation (Liu et al., 2020). Several TFs, such as GATA-1 (Vakoc et al., 2005) or BCL11A (Sankaran et al., 2008), have been described as being required for the transcriptional switch from γ-globin to β-globin expression. BCL11A plays a novel regulatory role in erythroid differentiation, maturation, and globin production (Luanpitpong et al., 2022). MiRNAs play key roles in erythropoiesis and control the expression of several TF genes involved in erythroid differentiation and hemoglobin gene expression (Jafari et al., 2019). MiRNA-146a is significantly more abundant in reticulocytes obtained from adults than those from umbilical cord blood and inhibits γ globin expression (Azzouzi et al., 2011). MiR‐24 promotes terminal differentiation (Drak Alsibai and Meseure, 2018). Our findings provide a resource for future studies aimed at elucidating the roles of TFs and miRNAs in erythropoiesis.

The blood and immune systems develop during early embryogenesis. Immune-erythroid cells are coupled with dual erythroid and immune regulatory networks and play immunomodulatory roles throughout human ontogenesis by actively interacting with various immune cells (Xu et al., 2022). Two transcriptional regulation programs are co-activated in immune erythroid cells: one is centered on the GATA1, MYC, and MYB regulons to ensure normal erythroid differentiation, and the other is dependent on the GATA2, FOS, and JUN regulons to instruct immunomodulatory activity. The mRNA-TF interaction network in our study included GDF15, GATA1, MYC, MYB, GATA2, FOS, and JUN. A negative correlation was found between GDF15 and the abundance of the 12 immune cell types of infiltration in samples with high and low FRG scores. GDF15 was associated with the abundance of five immune cell types between the early and late differentiation stages. Therefore, we speculated that GDF15 is an important gene associated with ED, ferroptosis, and immune invasion. Our results showed that some immune cells differed significantly during erythropoiesis. Erythroid progenitor cells play an important role in the regulation of immune responses and tumor progression. CD71+-nucleated erythroid cells play immunosuppressive roles in orchestrating immune response (Grzywa et al., 2021). RBCs affect immune cell functions. For example, they inhibit T-cell proliferation via direct cell-cell contact and affect dendritic cell functions (Bernard et al., 2010; Yi et al., 2015). Moreover, RBCs serve as critical immune sensors through the surface expression of nucleic acid–sensing toll-like receptor 9 (Lam et al., 2021). Endothelial progenitor cell (EPC) differentiation is a promising strategy to reduce cancer-induced immunosuppression and tumor-promoting effects of EPCs (Zhang H. et al., 2022).

Ferroptosis is associated with the immune response. ssGSEA showed that the infiltration abundance of 12 immune cell types was significantly different between high and low FRG scores. Ferroptosis affects immune cells in two ways. Immune cells require sufficient amounts of iron for their proliferation and for mediating their effector function. Ferroptosis affects the number and functions of immune cells. In contrast, ferroptotic cells are recognized by immune cells and trigger various inflammatory or specific responses (Chen et al., 2021). Ferroptosis regulates the activity and function of cytotoxic (CD8⁺) and helper T (CD4⁺) cells. Ferroptosis mediates neutrophil recruitment and neutrophil extracellular trap formation (NETosis). In T cells, ferroptosis induces a novel synergy between immunotherapy and radiotherapy. Additionally, ferroptosis may mediate B cell differentiation, antibody responses, and lymphoma. Ferroptosis regulates T- and B-cell immunity, which are involved in infectious diseases. Chronic immune activation in the setting of malignancy alters systemic iron homeostasis and directs iron fluxes into myeloid cells, which may result in iron-restricted erythropoiesis and anemia (Pfeifhofer-Obermair et al., 2018). Tumor-induced erythroid progenitor cells eventually differentiate into myeloid cells that exert robust immunosuppressive functions (Long et al., 2022). Consistent with our results, an interaction was found between the immune status, ferroptosis, and erythropoiesis. However, the exact mechanisms underlying these interactions require further study.

Considering the individual differences and technical limitations, we integrated three gene expression profile datasets to obtain a larger sample size and improve the accuracy of our conclusions. However, the number of patients included in this study was relatively small. Another apparent limitation was that we did not perform in vitro or in vivo experiments to validate the underlying mechanisms and their correlation between immune cells and hub genes, although we verified that six DE-FRGs changed markedly during ED stages. Therefore, we will continue to improve this aspect in the future.

In our study, 25 hub DE-FRGs were identified for erythropoiesis, which could be potential targets for subsequent research. Functional annotations were performed to elucidate the processes and pathways involving these genes. Ten hub DE-FRGs were obtained according to the PPI analysis. This study also predicted certain target miRNAs and TF that may be related to the pathophysiological process of erythropoiesis via ferroptosis. In addition, our analysis confirmed differences in immune infiltration during erythropoiesis. Finally, we established a signature using six DE-FRGs related to erythrocyte differentiation stages by LASSO regression analysis. The six DE-FRG were validated to be potential signature genes for erythropoiesis. Our findings may lay the groundwork for future clinical applications of ex vivo production of functional human reticulocytes for transfusion of human umbilical cord blood-derived CD34⁺ cells. This study could be important for understanding erythropoiesis and hematologic disorders whose etiology is related to impaired erythroid differentiation and hemoglobinopathies.