HIF-Mediated and Non-HIF-Mediated Differential Gene Expressions in Sickle Cell Reticulocyte and Their Impact on Clinical Manifestations

Reticulocytosis in sickle cell disease (SCD) is driven by tissue hypoxia from hemolytic anemia and vascular occlusion. Gene expression changes caused by hypoxia and other factors during reticulocytosis may impact SCD outcomes. We detected 1226 differentially expressed genes in SCD reticulocyte transcriptome compared to normal Black controls. To assess the role of hypoxia-mediating HIFs from other regulation of changes of the SCD reticulocyte transcriptome, we compared differential expression in SCD to that in Chuvash erythrocytosis (CE), a disorder characterized by constitutive upregulation of HIFs in normoxia. Of the SCD differentially expressed genes, 28% were shared between CE and SCD and thus classified as HIF-mediated. The HIF-mediated changes were generally in genes promoting erythroid maturation. We found that genes encoding the response to endoplasmic reticulum stress generally lacked HIF mediation.

We then investigated the clinical correlation of erythroid gene expression for the 1226 differentially expressed genes detected in SCD reticulocytes, using clinical measures and gene expression data previously profiled in peripheral blood mononuclear cells (PBMCs) of 157 SCD patients at the University of Illinois at Chicago (UIC). Normal PBMCs contain only a small number of erythroid progenitors, but in SCD or CE PBMCs the erythroid transcriptome is enriched due to elevated circulating erythroid progenitors from heightened erythropoiesis (PMID: 32399971). We applied deconvolution analysis to assess the clinical correlation of erythroid gene expression, using a 16-gene expression signature of erythroid progenitors previously identified in SCD PBMCs. Deconvolution analysis uses the proportion of cell/tissue or specific marker genes (here the erythroid specific 16-gene signature) to dissect gene expression variation in biological samples with cell/tissue type heterogeneity. We correlated, in the 157 UIC patients, erythroid gene expression with i) degree of anemia as indicated by hemoglobin concentration, ii) vaso-occlusive severe pain episodes per year, and iii) degree of hemolysis measured by a hemolysis index. The analysis identified 231 genes associated with at least one of the complications. Increased expression of 40 erythroid specific genes, including 15 HIF-mediated genes, was associated with all three complications. These 40 genes are all upregulated in SCD reticulocytes and correlated with low hemoglobin concentration, frequent severe pain episodes, and high hemolysis index, suggesting that these manifestations may share a relationship to stress erythropoiesis-driven transcriptional activity.

Expression quantitative trait loci (eQTL) contain genetic polymorphisms that associate with gene expression level, which can be viewed as a natural experiment to investigate the causal relations between gene expression change and phenotypic outcomes. To assess the causal effect of erythroid gene expression, we tested association between erythroid eQTL and the clinical manifestations in 906 SCD patients from the Walk-PHaSST and PUSH cohorts. We first mapped erythroid eQTL in the 157 UIC patients, who were previously genotyped by array, applying deconvolution algorithm on the same PBMC data for the 1226 differential genes in SCD reticulocytes, and detected 54 distinct eQTL for 30 genes at 5% false discovery rate. After adjusting for multiple comparisons, we found that the C allele of rs16911905, located in the β-globin cluster and associated with increased erythroid expression of HBD (encodes δ-globin of hemoglobin A ₂), significantly correlated with lower hemoglobin concentration (β=-0.064, 95% CI -0.092 - -0.036, P=6.7×10 ^-6). The C allele was also associated with higher hemolytic rate (P=0.031), less frequent pain episodes (P=0.045), and increased erythroid expression of HBB here encoding sickle β-globin (P=5.1x10 ^-5). The association of the C allele with lower hemoglobin concentration was then validated in 242 patients from the UIC cohort (β=-0.071, 95% CI -0.13 - -0.011, P=0.023), as was the trend of association with higher hemolytic rate (P=0.0031) and less pain episodes (P=0.034).

Our findings reveal HIF- and non-HIF-mediated genes in SCD stress erythropoiesis, and identify novel clinical associations for a HBD eQTL. Our study highlights the correlation of altered erythroid gene expression with SCD hemolytic and vaso-occlusive manifestations.