Single cell spatial transcriptomics reveals age-dependent cellular organization of mouse bone marrow

Abstract Hematopoiesis occurs in the complex microenvironment of the bone marrow. This includes interactions between hematopoietic cells themselves as well as interactions with non-hematopoietic cells such as stromal cells, endothelial cells, adipocytes, bone cells, and neural cells, as well as extracellular matrix materials. Given this complex milieu, it is critical to study hematopoietic regulation in the context of its native tissue architecture. However, a lack of methods to simultaneously identify cell types at high granularity and examine their spatial organization, inter-cellular signaling, and cellular states has impeded efforts to further understand hematopoietic regulation in health and disease. Recently, a suite of spatial transcriptomic technologies has emerged enabling such an analysis. Among these, Multiplex Error Robust Fluorescence In Situ Hybridization (MERFISH) is a leading platform. MERFISH is a massively multiplexed single-molecule RNA imaging method built upon single-molecule fluorescence in situ hybridization (smFISH). MERFISH multiplexes smFISH by assigning to each RNA a unique binary barcode physically encoded into a library of nucleic acid encoding probes, which is read-out through sequential rounds of smFISH measurements on the same sample. Critically, MERFISH utilizes error-robust and error-correcting encoding methods which allow this technique to image hundreds to thousands of RNAs with a high detection efficiency, a low false positive rate, and a large dynamic range. MERFISH has been applied to multiple tissues such as the brain and intestine yielding novel insights into tissue organization and signaling. A major obstacle to applying MERFISH, or other RNA-based methods, to the blood system has been the high RNase activity of hematopoietic tissues, resulting in poor RNA integrity. Here we present a new method of bone marrow preparation that prevents RNA degradation and enables high quality bone marrow MERFISH data acquisition. With this method and a 1,022 gene library we spatially mapped ~1.5 million cells from 28 mouse femoral cross sections, identifying 53 distinct cell clusters. This dataset embeds well with published dissociated single cell RNAseq datasets, verifying capture of all expected cell types. Importantly, it also includes cell types that do not dissociate well such as mesenchymal stromal cells, osteoclasts, eosinophils, mature megakaryocytes, and rare hematopoietic stem cells. With this dataset, we quantified the frequency and distance of all cell-cell contacts and find that bone marrow microenvironments are well described by a statistical model of cell contact frequency. We find that two cell types—sinusoidal endothelial cells (SECs) and mesenchymal stromal cells (MSCs)—exist in a spectrum of transcriptional states that can be defined by a pseudotime axis. The spatial intersection of high pseudotime SECs and MSCs coincides with HSC niches, indicating that specific signaling from a subset of these cells facilitates hematopoiesis. Conversely, the intersection of low pseudotime SECs and MSCs coincides with increased density of natural killer, dendritic, T-, and plasma B-cells, indicating that subsets of these cells facilitate immune microenvironments. To understand the factors that regulate these microenvironments, we interrogated our MERFISH measurements and integrated scRNAseq datasets and find a series of candidate factors secreted by MSCs and SECs that may facilitate bone marrow niche formation. To understand the restructuring of bone marrow during aging, we compared MERFISH measurements of femoral cross sections from 8-, 40-, and 90-week mice. In aged mice, we find a structural breakdown of hematopoietic organization correlating with mouse age. These results are consistent with age-associated chronic inflammation and provide a step toward understanding aging related bone marrow dysfunction. Overall, this study provides new insights into bone marrow single cell spatial organization and a new tool that should greatly facilitate further understanding bone marrow niches in health and disease.