Multiomic Analyses of Erythropoiesis and Anemia Mechanisms Reveal a Signaling Network with Converging Cytokine and Sphingolipid Inflammatory Pathways

Recent studies from our group have established a paradigm where GATA factors control erythropoiesis by regulating the expression of transporters and biosynthetic enzymes, which dictates intracellular levels of small molecules, including heme, metal ions, and sphingolipids. GATA1 controls the expression of sphingolipid metabolic enzymes, including Degs1 and Sphk1, which confer ceramide homeostasis during erythropoiesis. Disrupting ceramide (Cer) homeostasis impairs erythroid progenitor function and erythroblast survival by antagonizing critical stem cell factor (SCF) and erythropoietin (Epo) signaling via a PP2A-dependent mechanism. While Epo-induced AKT, ERK, and JAK/STAT signaling are attenuated by disrupting Cer homeostasis, the full spectrum of Epo/Cer-dependent signaling events and how such events integrate into the overall signaling milieu of the developing erythroblast in normal and pathological states is unknown.

To comprehensively elucidate the components of the Epo/Cer-instigated signaling network, we performed phospho-proteomics in primary human HSPC-derived erythroblasts with or without short-term (5 min) Epo stimulation under normal conditions and conditions in which Cer disrupts function. Total cellular proteins were digested with lysyl endopeptidase, and phospho-peptides were enriched with titanium dioxide before liquid chromatography-mass spectrometry (LC-MS) analysis. Consistent with the established paradigm, pY694-STAT5A/B, the canonical target of Epo signaling, was the most highly upregulated phospho-peptide in response to Epo, and this upregulation was abolished by Cer. Differential analyses (|log₂FoldChange|>0.5, p<0.05) revealed 268 Epo-regulated phospho-peptides, the majority of which had not been linked to Epo signaling. Of the 268 Epo-regulated phospho-peptides, Cer impaired or facilitated Epo-dependent regulation of 17. Thus, Cer impacts a specialized sector of the Epo signaling network. Of the 17 Epo/Cer-co-regulated phospho-peptides, Epo increased and Cer decreased pY694-STAT5A/B, while 14 were downregulated by Epo and upregulated by Cer, suggesting that Cer antagonizes Epo activity in most of these cases. These peptides include pS109-endosulfine alpha (ENSA), which is decreased 2.4-fold by Epo and increased 1.9-fold when Cer is present with Epo. ENSA, and its paralog cAMP-regulated phosphoprotein 19 (ARPP19), are endogenous regulators of PP2A activity and cell cycle progression. To ask if ENSA and/or ARPP19 mediate Epo signaling, we generated Ensa/Arpp19 double-knockout G1E-ER-GATA1 cells using CRISPR-Cas9 and quantified Epo signaling in WT and double-mutant cells. Depleting ENSA and ARPP19 enhanced Epo-dependent phosphorylation of STAT5, suggesting that ENSA/ARPP19 are new components of the Epo signaling pathway. An in-depth mechanistic analysis is being developed.

Chronic inflammation is often associated with elevated inflammatory cytokines, disrupted ceramide homeostasis, and anemia. Treating primary human HSPCs ex vivo with a cocktail of inflammatory cytokines (TNFα, IFNγ, and IL-6) had little impact on the viability of erythroid progenitors at day 9 but significantly impaired generation and/or survival of glycophorin A (GPA)-positive immature erythroblasts at day 13, suggesting a differential sensitivity of cells at distinct stages of differentiation to inflammation. To elucidate the specific cell types and cadre of genes (and networks) that are hyper- or hypo-sensitive to inflammatory signaling and how sphingolipids contribute to this process, we performed single-cell RNA-sequencing (scRNA-seq) and quantitative lipidomics in primary human erythroid cells at day 9 and day 13 treated with or without the inflammatory cytokine cocktail. We are integrating the lipidomic and scRNA-seq data to establish mechanisms by which inflammation remodels the sphingolipidome, corrupts molecular networks, and dysregulates erythropoiesis at single-cell resolution.

In summary, we established an integrated network involving cytokine signaling, inflammation, and sphingolipids that controls erythropoiesis through multiomic studies in primary human erythroid cells. These datasets are providing unique mechanistic insights into the control of erythropoiesis under normal and pathological conditions and translational applications for anemia of inflammation and potentially of broader impact.

No relevant conflicts of interest to declare.