Single-Cell Transcriptomic Analysis of Bone Marrow from Humanized Sickle Cell Mice Reveals a Subpopulation of Highly-Expressing S100A8/A9 Mesenchymal Stem Cells

Sickle cell disease (SCD) is the most common monogenic red cell disorder worldwide, caused by a mutation in the β-globin gene leading to the production of the pathologic hemoglobin S (HbS). To date, therapeutic strategies to limit severity of clinical manifestations of SCD are unsatisfactory. Hematopoietic stem cell (HSC) transplantation is the only curative option for SCD, but patient age and availability of suitable donors deeply affect the number of patients with access to this treatment. Gene therapy is an attractive novel curative option for patients with SCD. To better understand the bone marrow (BM) niche, which is believed to play a crucial role in cell therapies (Jones & DeBaun, 2021; Pincez et al, 2021; Park et al 2019), non-hematopoietic cells in SCD BM need further investigation. Here, we used humanized SCD mice (‘Townes’ mice, SS) and healthy control mice (AA) to explore non-hematopoietic BM cells at single-cell resolution. Since previous single cell studies have shown that erythroid cells (TER119+) and BM neutrophils (Ly6G+) are the most abundant cells in BM (Zhong et al. 2022), we removed TER119+ cells and neutrophils using MACS-based negative selection, followed by sorting of Ly6G+ neutrophils to obtain 7-AAD- Ter119- Ly6G- cells. Single-cell RNA sequencing (scRNA-seq) was performed using the Chromium Next GEM Single Cell 3ʹ Gel Bead and Library kit v3.1 (Dual Index). The sequencing was carried out using the NovaSeq 6000 SP Reagent Kit v1.5 (Illumina, California, USA). Cell type annotation was done manually by ScType (Ianevski et al., 2022) with a manually curated database (Tabula muris consortium, 2018, Fievet et al, 2023, Kim et al., 2022). Our dataset was downsampled to 2,898 cells for each condition (AA/SS), generating 16 different cell clusters, twelve of which were non-hematopoietic cell subsets. Among them, we observed increased expression of endothelial cells with sinusoidal hallmarks in SS mice when compared to AA animals (cluster 11). This agrees with our previous study showing an abnormal bone marrow vascular niche in SS mice (Park et al 2020). We found that cluster 0 was larger in SS mice than in AA animals. Cluster 0 was annotated as progenitor cells, which contains a subset of MSCs. These showed down-regulation of genes involved in oxidation-related cell repairing mechanism(s) (Atf3, Fosb), in modulation of the quiescent state of MSCs (Angp1) and in osteogenesis (Thbs1). Osteogenic mechanisms were further supported by the downregulation of Fn1 and Runx1, in agreement with our previous work on reduced osteoblastogenesis in SS mice (Dalle Carbonare et al 2015; Park et al 2020). Cluster 0 MSCs display up-regulation of Ebf1, S100A8, S100A9, Lcn2 genes, supporting the immunoregulatory (Ebf1) and cell proliferation/self-renewal (S100A8/9, Lcn2) potential of MSCs in SS mice compared to AA animals. We noted the presence of an MSCs subset highly expressing S100A8 and S100A9 genes. Indeed, S100A8 and S100A9 are involved in different processes such as neo-angiogenesis, inflammatory response and cell self-renewal events. We carried out gene network analysis (Gene Ontology and KEGG pathways implemented in the STRING database) showing a high-level interaction of S100A8/9 between themselves, and with Lcn2 and p53 (Trp53bp2). Previous papers have shown upregulation of S100A8/9 in myeloid leukemia cells and in plasma/BM from patients with MPN (myeloproliferative neoplasms) turning to myelofibrosis (Min et al 2020; Leimkuhler et al 2021). To better understand S100A8/9 in SCD, we analyzed plasma S100A8/9 heterodimers in mice (n= 10) and S100A8 homodimers in patients (n= 21) with SCD by ELISA. We found a significant increase in plasma S100A8/9 in both SS mice and patients with SCD compared to healthy controls. In conclusion, the single cell analysis of TER119-LyG- BM cells confirmed our previous report on the abnormal BM vascular niche and impaired osteoblastogenesis in SCD mice. We identified a novel subset of MSCs highly expressing S100A8/9 genes, associated with increased plasma S100A8/9 in SS mice. Our data indicates a perturbation of the BM microenvironment in SCD involving MSC-derived S100A8/9, potentially networking with Lcn2 and Trp53bp2. We propose that S100A8/9 could serve as a novel biomarker for BM niche pathology in SCD. Furthermore, S100A8/9 provides a druggable target that could be explored for reducing BM pathology in SCD.