Spatial Transcriptomic Profiling of Bone Marrow Vascular Niches Unveils an Endothelial Stress Signature in Myeloproliferative Neoplasms

Introduction: Bone marrow endothelial cells (BMECs) are not only key contributors to normal hematopoiesis but also play an important role in the pathogenesis of myeloid diseases, as in the case of Philadelphia-negative myeloproliferative neoplasms (MPNs), where vascular complications, such as thrombosis and angiogenesis, are a mayor clinical challenge. Single cell technologies have contributed to understand the transcriptional heterogeneity of the hematopoietic MPN malignant clones (O'Sullivan, J. et al., Blood, 2023) and is starting to shed light on niche stromal cells. However, the spatial crosstalk between hematopoietic cells and BMECs at specific vascular niches and its impact on the pathogenesis of MPNs remain to be investigated. Our aim was to evaluate the contribution of vascular niches to the MPN marrow microenvironment by performing for the first-time a spatial transcriptomic analysis on bone marrow (BM) biopsies from MPN patients. Methods: We combined spatial context and transcriptomics to perform the high-plex GeoMx digital spatial profiling (DSP) approach (Nanostring) using the human Whole transcriptome atlas on twelve distinct EDTA-decalcified formalin fixed paraffin embedded (FFPE) BM biopsies. Tissue samples from three healthy donors (HDs) and nine MPN patients [three essential thrombocythemia (ET), two primary myelofibrosis (PMF), two advanced PMF and two myelofibrosis (MF) post-ET] were distributed on four different slides. Seven patients were JAK2V617F positive, one CALR and one triple negative MPN. Among JAK2V617F patients, two were treated with the JAK2 inhibitor Ruxolitinib. Regions of Interest (ROIs) with similar cell density were selected throughout the biopsies based on two parameters: spatial location with areas near the bone (vascular-endosteal) or far from it (vascular-medullar) and specific cell surface markers for our cells of interest (BMECs, megakaryocytes and monocytes). Subsequently, Next Generation Sequencing (NGS) with Illumina technology was employed and the data obtained underwent thorough analysis using various bioinformatics tools. Results: We spatially profiled 112 vascular-enriched ROIs in total (nine per sample) from MPN patients and HDs. After alignment on HG38 human genome and gene quantification, a gene regulation network of vascular ROIs was built by co-expression examination using the WeiGhted Correlation Network Analysis (WGCNA) algorithm that considers the heterogeneity and spatial distribution of the ROIs. Supervised machine learning analysis of all ROIs identified a specific gene profile of BM endosteal and medullar areas correlating with vascular and endothelial labeling. We next performed two in-depth analysis of all the vascular-enriched ROIs. First, we compared HDs to naïve JAK2 positive MPN patients which unveiled 105 differentially expressed genes between these two groups. Next, we analyzed Ruxolitinib-treated against naïve JAK2 positive MPN patients, revealing this time a compelling inhibition of gene expression in MPN vascular-enriched niches. These results strongly suggest an important inflammatory process that possibly alters the endothelial response in MPN vascular niches. Gene Ontology Biological Process (GOBP) analysis further supported these observations, as it highlighted genes involved in crucial endothelial functions (migration, sprouting, angiogenesis), stress signaling (JNK and MAPK activation) and inflammatory pathways (TNF/NF-kappaB, IL-1R1, TGFb) in MPN vascular-enriched niches compared to those in HDs. Conclusions: We examined for the first time the EC status within the BM microenvironment in MPNs using spatial transcriptomics. Our results revealed distinct gene expression profiles in BM vascular-enriched ROIs between MPN patients and HDs, indicating an activated gene signature specific of MPNs. Analysis of Ruxolitinib-treated MPN patients further highlighted endothelial stress signaling processes associated with pro-inflammatory pathways. Examination of these gene profiles will serve to identify BMEC markers for diagnostic and treatment of vascular complications in myeloid diseases in combination with current treatments.