Dnmt3a-Mutant Haematopoietic Stem and Progenitor Cells Induce Mesenchymal Stromal Cell Senescence through the Production of Alarmins S100A8/A9

Abstract

Clonal haematopoiesis (CH) is an age-related condition occurring due to somatic genetic variation in haematopoietic stem and progenitor cells (HSPCs), most frequently observed in the gene DNA methyltransferase DNMT3A. CH can be a precursor condition for haematologic malignancy, but the mechanisms driving progression of CH to malignancy are incompletely understood. Given that malignant cells are understood to reprogram their bone marrow (BM) microenvironment to create a self-reinforcing niche, we hypothesized that HSPCs carrying a CH-associated mutation in DNMT3A have microenvironment-remodelling properties that promote their clonal advantage. Using single-cell RNA-seq, we comprehensively captured bone and BM cells of haematopoietic and non-haematopoietic lineages in Dnmt3a R878H/+ mice in which the Dnmt3a mutation is present in the haematopoietic but not the non-haematopoietic compartments. In the non-haematopoietic fractions, we identified enrichment of a cellular senescence program, including transcriptional signatures of cell cycle arrest and expression of the canonical senescence-associated gene Cdkn1a (p21), in BM mesenchymal stromal cells (MSCs). The MSC cellular senescence phenotype caused by Dnmt3a-mutant haematopoiesis was confirmed at the protein level by increased expression of senescence-associated beta galactosidase (SA-β-gal) and the anti-apoptotic proteins BCL-2 and BCL-xL. In ex vivo co-culture assays, primary wild-type MSCs cultured with Dnmt3a-mutant HSPCs versus control HSPCs had increased SA-β-gal, Cdkn1a (p21), and Cdkn2a (p16) expression. Moreover, transwell assays demonstrated that MSC senescence induced by Dnmt3a-mutant HSPCs is cell contact-independent. To undercover the mechanisms driving MSC senescence, we analyzed soluble factors produced by Dnmt3a-mutant HSPCs. Conditioned mediafrom Dnmt3a-mutant HSPCs had significantly elevated levels of S100A8/A9, IL-6 and TNFα, which we confirmed were also elevated in BM fluid of Dnmt3a-mutant mice. In our single cell RNA-seq data, S100a8 and S100a9 were the top significantly increased genes in Dnmt3a-mutant versus control HSCs, making S100A8/A9 our top candidates for functional studies. We tested the direct effect of recombinant S100A8/A9 on primary wild-type MSCs and found that this was sufficient to induce SA-β-gal. Ongoing studies using the S100A8/A9 inhibitor Tasquinimod are examining the dependency of Dnmt3a-mutant hematopoiesis on alarmin production ex vivo and in vivo. Given that S100A8/A9 are heavily involved in the pathogenesis of acute inflammation and myeloproliferative neoplasms, induction of senescence through the alarmin axis is a mechanistic link between aging, clonal hematopoiesis and haematologic malignancy.