Abstract
The anemias of myelodysplastic syndrome (MDS) and Diamond Blackfan anemia (DBA) are generally macrocytic, always reflect ineffective erythropoiesis, and yet result from diverse genetic mutations. To delineate the shared mechanisms that lead to the death of maturing erythroid cells, we studied single marrow cells from MDS-5q patients, DBA patients, and normal individuals with CITE-seq (antibody barcoding of cell surface proteins) and RNA sequencing. After testing several approaches, we opted to align transcriptomes with SCVelo pseudotime analysis, which includes incompletely processed mRNAs, since these are highly prevalent and regulated during erythropoiesis. We identified 38,296 erythroid lineage cells then plotted CITE-seq protein quantitation data versus the velocity pseudotime to identify BFU-E through orthochromatic erythroblasts based upon the sequential CD117, CD36, CD71, and GlyA expression levels, using well-established criteria.
The analysis identified two distinct transcriptional trajectories. These diverge during the CFU-E stage, following upregulation of the transferrin receptor, CD71, and initiation of iron import, and right after upregulation of ALAS2, the first and rate-limiting step of heme synthesis. The normal (healthy) trajectory (trajectory A) contains cells at all stages, including the late stages of erythroid maturation, and there is high expression of mitochondrial oxidative metabolism and erythroid differentiation pathway genes. In contrast, the second trajectory (trajectory B) ends abruptly near the start of terminal erythroid differentiation as defined by cell surface proteins, and highly expresses gene pathways associated with excess heme and with apoptosis. There is loss of mitochondrial gene expression, premature loss of nuclear gene expression (increased % mitochondrial genes), and elevated levels of BAX and other apoptosis markers, implying that this is a death pathway. The significant increase in expression of genes in the heme-responsive dataset defined by Liao et al (Cell Rep. 2020. 31:107832) in trajectory B cells along with the pathway analysis, suggests that DBA and MDS-5q erythroid cells died from heme toxicity, a finding supportive of our earlier work (Yang et al. Sci Transl Med 2016. 8:338ra67). Cells along each trajectory were present in all samples but at different frequencies: 57±4% of MDS-5q, 73±2% of DBA, and 24±12% of normal cells followed trajectory B. That 24±12% of normal cells followed trajectory B was an unexpected finding, and indicated that heme might regulate the ineffective vs. effective differentiation of normal, as well as DBA and MDS-5q cells. As a corollary, drugs or interventions decreasing heme synthesis might modulate trajectory choice and be therapeutically beneficial. Also notably, fate decisions occur well before cell death, allowing time for therapeutic intervention.
In addition, since we can independently track the fate and transcriptomes of cells with an intact chromosome 5q vs. cells with chromosome 5q deletion, these experiments provide insight into why anemia occurs early in MDS-5q (i.e., at presentation) often only 50-75% of the marrow is replaced by neoplastic cells. The analysis revealed both 5q- and 5q+ cells show early upregulation of HIF1A targets, ROS response, cell cycle regulation and spliceosome pathways and down regulation of erythroid differentiation and mTOR signaling pathways, indicating both 5q- and 5q+ cells share common pathologies. Additionally, the 5q- cells upregulate p53 pathways. The fact that the 5q+ cells are not normal implies that the presence of neoplastic 5q- cells influence the erythroid differentiation of the 5q+ cells. That they too have transcriptomic findings attributable to excessive heme may suggest a cell-extrinsic role for heme toxicity, which has led us to study heme processing by the central macrophage of erythroblastic islands (EBIs). Initial studies suggest EBI macrophage dysfunction contributes to the poor differentiation of 5q+ cells. This abstract includes data available on the medRxiv preprint server. https://doi.org/10.1101/2022.08.01.22278133
Disclosures
No relevant conflicts of interest to declare.
Author notes
Asterisk with author names denotes non-ASH members.
