Integration of single-nucleus and spatial transcriptomics reveals the molecular landscape of the human anterior hippocampus

The hippocampus (HPC) is a critically important brain region with characteristic cytoarchitectural organization that supports its unique capacity for structural and functional plasticity. Recent studies in the human post-mortem brain have generated molecular profiles for human hippocampal cell types at single-cell resolution. However, these profiles do not retain information about spatial positioning of cells, and do not capture extra-nuclear transcripts. Given the tight correlation between spatial structure and function in the HPC, and the particular importance of transcripts localized to the synaptic compartment, molecular profiling technologies with the capability to address these gaps in our knowledge would be highly useful. Thus, we created paired spatially-resolved transcriptomic and single nucleus profiles across the major subfields of the anterior human HPC from neurotypical control brain donors. We used the 10x Genomics Visium Spatial Gene Expression platform, which combines spatially-resolved transcriptomics with high resolution histological images to generate spatial transcriptomic maps of the anterior human HPC. To characterize the organization of the HPC, we scored the tissue blocks and used multiple arrays to fully capture all major subfields (CA1-4, dentate gyrus, and subiculum) in N = 10 neurotypical donors, both males and females (150917 spots from 36 capture areas). To provide cell-type specific information, we performed single-nucleus RNA sequencing (snRNA-seq) on adjacent tissue sections from the same donors (87095 nuclei from 10 donors). We used spatially-informed, data-driven clustering methods to identify the spatial domains, and compared these results with those obtained from histologically-derived annotations of canonical subfields. To identify molecular signatures of each domain, we perform differential expression analysis to identify unique gene expression profiles for each domain. We spatially registered all snRNA-seq clusters to add anatomical context, and performed spot-level deconvolution analysis to identify predicted cell-type composition of each gene expression spot. We performed spatial enrichment analysis of differentially expressed genes from neurodevelopmental disorders in which hippocampal function is impaired, including schizophrenia and major depression. Finally, we created an interactive web application to display gene expression and clustering information overlaid on tissue morphology, to enable exploration of this data resource by the larger scientific community. We generated large-scale, transcriptome-wide, spatially-resolved molecular profiles of the human anterior HPC for N=10 neurotypical adult donors. We provide a rich data resource to the neuroscience community. These data integrate single nucleus and spatial information to enable better understanding of the molecular landscape of the human HPC.