Graphical abstract graphic file with name fx1.jpg [33]Open in a new tab Highlights * • We integrate spatial transcriptomic and spatial metabolic in human injured brain * • Neuronal loss is spatially associated with lipid peroxidation in injured brain * • Spatial multi-omics reveal metabolic heterogeneity of injured brains * • Spatial multi-omics shows the link between metabolic change and local gene expression __________________________________________________________________ Zheng et al. highlight the complex transcriptomic regulation and metabolic alterations in the injured brain with spatial multi-omics, enabling the design of reagents to target specific genes in the human brain for functional analysis. Introduction The cellular constitution of brain is of extreme diversity.[34]^1^,[35]^2^,[36]^3^,[37]^4^,[38]^5 Single-cell molecular assays, especially transcriptomic measurements by RNA sequencing (RNA-seq), have facilitated the identification of cell types in neurological diseases.[39]^6 Although single-cell transcriptomic datasets are able to locate hundreds of neuronal and non-neuronal cell types across the central nervous system,[40]^1^,[41]^6 anatomical and physiological (metabolic) information is missing. Recent advances in spatial multi-omics present an opportunity for linking a function-based transcriptomic[42]^7 and metabolic[43]^8 atlas that facilitates progress across modalities and obtains functional information. Herein, we intend to compose an atlas of functional types across the human injured brain by integrating spatial omics approaches. We selected human patients with traumatic brain injury (TBI) owing to its relatively emergent surgical necessity and obtainable tissues in the brain contusion area. Although neuron and non-neuron cell types have been previously identified in human brains,[44]^9 no spatial transcriptomic and metabolic analyses have been performed and further explored. Our integrated atlas is a case study of the expansive potential and the technical limitations of spatial multi-omics methods for comprehensive brain-wide analysis. 10× Genomics Visium with spatial transcriptome can locate the gene expression information of different cells in the tissue to their original spatial position[45]^10 so as to directly observe the difference of gene expression in different functional regions in the tissue. Spatial metabolics, as a new type of molecular imaging technology, can directly obtain a large number of known or unknown molecular structures, content, and spatial distributions of endogenous metabolites.[46]^8 Compared with other imaging methods (such as fluorescence imaging, radioactive labeling imaging, etc.), this technology does not need chemical or radioactive labeling and complex sample pretreatment.[47]^11 It has the outstanding advantages of high specificity, high throughput, and spatial information retention. Mass spectrometry imaging technology can realize the qualitative and quantitative localization of thousands of metabolites in biological tissues.[48]^12 Combined with bioinformatics analysis, it has developed into a spatial metabolomics method that can find different metabolites in situ from biological tissues and identify their biological functions.[49]^8 The simultaneous application of spatial transcriptome and spatial metabolism, the two advanced technologies at present, can well reveal the spatial composition of the functional map in tissue, the heterogeneous distribution of cell groups, and the expression differences of genes in different positions, and obtaining complex and complete temporal and spatial expression maps of genes, which can be used in cancer, immunity development, and other research fields, has very important research value. This would bridge the gap in spatial location in previous TBI transcriptomic and metabolic studies. Based on these, our data suggest that patients with severe TBI have increased lipid peroxidation and myo-inositol levels (metabolic heterogeneity) compared with patients with moderate TBI, which might contribute to the poor response to comprehensive therapies. Results Spatial transcriptomics reveal distinct molecular regionalization of injured brains To explore the cellular landscape in TBI, we collected brain contusion tissues from patients with TBI with surgery for single-cell RNA sequencing (RNA-seq) based on 10× Genomics. Cells were subjected to t-distributed stochastic neighbor embedding (tSNE) analysis, and 10 clusters in two groups were obtained ([50]Figure 1A). The cell annotation identifies ten clusters, named B cells, endothelial cells, granule cells, microglia, mural cells, neutrophils, natural killer (NK) cells, oligodendrocytes, and T cells ([51]Figure 1B). Furthermore, while studying signaling pathways, KEGG analysis revealed that the upregulated genes in severe TBI were mainly enriched in infection and immune statuses including Th1, Th2, and Th17 differentiation and antigen processing and presentation ([52]Figure 1C), while the downregulated genes were enriched in neuroinflammation, such as tumor necrosis factor (TNF) signaling, interleukin-17 (IL-17), HIF-1, and nuclear factor κB (NF-κB) signaling ([53]Figure 1D). Figure 1. [54]Figure 1 [55]Open in a new tab The differences in cell subsets of injured brain in patients with TBI (A and B) t-SNE plot shows the 10 cell subsets in all samples (A) and cell annotations in ten clusters (B). (C) KEGG pathways in upregulated DEGs in severe TBI. (D) KEGG pathways in downregulated DEGs in severe TBI. As we found different cell clusters between the moderate and severe brain injuries, this led us to characterize the transcriptomic landscape of the injured brain, by processing fresh surgical samples for spatial transcriptome (ST) using the Visium (10× Genomics) platform. Upon filtering out the mitochondrial protein-coding genes, the resulting dataset consists of 4,000 individual spots, with an average of ∼2,500 genes and ∼7,500 unique transcripts per spot. First, we deconvolved the spatial transcriptomic dataset using uniform manifold approximation and projection (UMAP) for dimension reduction to visualize the clusters[56]^13 ([57]Figure 2A). We identified 10 basic structural transcriptomic landscapes that were histologically discernible. Analysis of the top contributing genes for each factor confirmed the identity of these clusters and their mixed signatures ([58]Figure 2B). Figure 2. [59]Figure 2 [60]Open in a new tab The cluster analysis for spatial transcriptomics and metabolic (A) UMAP reduction to visualize the clusters. The x axis in the left panel represents the first and second principal components of UMAP dimensionality reduction, respectively. Each point in the panel represents a spot. The spots of different groups are distinguished by different colors. The right panel reflects the specific distribution of different groups in the slice position of each sample, and the color corresponds to the left panel. (B) Heatmap plot for spatial marker genes. The x axis is the spot number, and the y axis is the marker gene. The yellow color indicates higher expression, while the purple color indicates lower expression. (C) K-means map after UMAP for spatial metabolic data in each sample. (D) PLS-DA in multiple statistical comparison to differentiate sample groups. The six plots indicate one negative and one positive ion mode in each sample. To better visualize the molecular regionalization both across the moderate and severe TBIs, we added the glioma and meningioma samples as references ([61]Figures 2A and 2B), which are described in the [62]STAR