To identify specific cellular microenvironments affected by spaceflight, we combined the techniques of spatial transcriptomics (ST; 10X Genomics Visium) and single-nucleus multiomics (snMultiomics; gene expression and chromatin accessibility; 10 Genomics Single Cell Multiome ATAC+Gene Expression) on mouse brain. In total, we analyzed three brains from mice euthanized on-board of the International Space Station (ISS; F1, F2, F3) and three brains from ground control mice (G1, G2, G3) that were kept under matched conditions (see Animals in Methods). For each sample, we isolated nuclei from one hemisphere for snMultiomics analysis and cryo-sectioned the other hemisphere for ST analysis with the focus on the hippocampal region (Fig.1). <\/p>\n
Overview of the study workflow where brains from International Space Station (ISS; Flight mice) and ground control mouse groups (Ground control mice) were split into the two hemispheres for Spatial Gene Expression Analysis (Spatial Transcriptomics or ST) and Single Nuclei Multiomics analysis (snMultiomics). <\/p>\n
As a first step, we ensured that the morphological and RNA quality of the samples was suitable for our experimental workflow given that the spaceflown samples had undergone a specific preservation approach17, which was also used for the corresponding ground control animals (see Animals in Methods). We measured the RNA integrity number (RIN) for each sample and found that it was 9.15 on average (Supplementary Fig.1A). Furthermore, we performed a tissue optimization experiment confirming that both RNA integrity and tissue morphology was of sufficient quality for ST analysis (see Visium Spatial Gene Expression technology and sequencing in Methods; Supplementary Fig.1B). <\/p>\n
To dissect the alterations induced by spaceflight at the single-nucleus level, we performed a snMultiomics analysis on hemispheres of three spaceflown (F1, F2, F3) mice and two out of three ground controls (G2, G3), obtaining RNA expression profiles (RNA-seq) and chromatin accessibility (ATAC-seq) information from the same nucleus. <\/p>\n
In total, we isolated 21,178 nuclei across the spaceflight and control samples with an average of 3140 unique transcripts (Unique Molecular Identifier or UMI) per nucleus (i.e., from snRNA-seq) and 9217 peaks per nucleus (i.e., from snATAC-seq) (Fig.2A, B; Supplementary Fig.1C) and an overall high gene expression correlation between the spaceflight and ground control samples (r=0.95, p<0.05; Fig.2C). By integrating snRNASeq and snATAC-seq data and performing a joint clustering analysis, we identified 18 snMultiomics clusters (Fig.2D; Supplementary Fig.2). <\/p>\n
A Distribution of UMIs per nucleus in the entire snRNA-seq dataset. nUMI\/nuclei: number of UMIs detected in each nuclei. B Distribution of peaks per nucleus in the entire snATAC-seq dataset. nPeaks\/nuclei: number of peaks detected per nuclei in the multiomics dataset. C Correlation between flight (y-axis) and ground control (x-axis) single nuclei multiomics samples (Pearsons correlation coefficient, r=0.95; p<0.05) shown as a scatter plot. This is a two-sided Pearson correlation test with 95% confidence intervals performed on the average expression (log(1+avgUMI)). avgUMI: average UMI counts per spot. D UMAP of single nuclei multiomics data and cluster annotations. E 11 functional multiomics clusters categories represented by their marker genes. F Distribution of UMIs per spot for the whole spatial transcriptomics (ST) dataset. nUMI\/spot: number of UMIs detected per spot in the ST dataset. G Distribution of unique genes per spot for the whole spatial transcriptomics (ST) dataset. nGenes\/spot: number of genes detected per spot in the ST dataset. H Correlation between flight (y-axis) and ground control (x-axis) ST samples (Pearsons correlation coefficient, r=0.99; p<0.05) shown as a scatter plot. This is a two-sided Pearson correlation test with 95% confidence intervals performed on the average expression (log(1+avgUMI)). avgUMI: average UMI counts per spot. <\/p>\n
Next, we leveraged previously reported marker genes in the literature (see Gene and cluster annotation in Methods for details) to identify 11 macro categories for the 18 snMultiomics clusters (interchangeably referred to as multiomics clusters in the next sections) according to their functions (Fig.2E; Supplementary Data1, 2). The majority of clusters were related to neurogenesis, neuronal activity and synaptic transmission, distinguished by differences in neurotransmitters (GABAergic, glutamatergic, dopaminergic) and based on gene expression patterns, tentatively associated with neuronal locations in hypothalamus, striatum, cortex and hippocampus. <\/p>\n
We identified a total of 825 differentially expressed genes (DEGs) between spaceflown and ground control samples across all multiomics clusters (Supplementary Data3). The majority of these 825 DEGs were involved in neuronal development (multiomics clusters 9, 11), axonal or dendritic outgrowth (multiomics cluster 9), and synaptic transmission (multiomics cluster 4), including specifically GABAergic synaptic transmission (multiomics cluster 11). <\/p>\n
Comparison of 825 spaceflight multiomics DEGs to the 629 significant DEGs (Spaceflight vs Ground Control; p-value<0.05) from the bulk RNAseq data of the same mice brains from the same NASA mission (RR-3), indicated 11 shared genes (p-value=0.01582549, hypergeometric distribution test; see Gene overlap test in Methods; Supplementary Data4). Out of these 11 overlapping genes, only 2 genes (Gabra6, and Kctd16) showed the same directional change in both the datasets indicating that the majority of spaceflight effects are cell type-specific and emphasizing the need for cell-specific analysis of central nervous system responses to spaceflight. <\/p>\n
We also compared these 825 spaceflight DEGs with spaceflight DEGs reported in a total of 11 other datasets processed by NASA OSDR including mass spectrometry and RNA-seq data collected from different organs of BALB\/c and C57BL\/6J mice strains. This comparison revealed a total of 461 overlapping DEGs (p-value<0.05) across all the 11 datasets combined (refer to Supplementary Data5 for a detailed list of overlapping genes and the resulting p-value from the hypergeometric distribution test performed for each dataset). <\/p>\n
To investigate spaceflight-induced CNS alterations at a spatial level, we performed ST analysis on the other brain hemispheres from 3 flight (F1, F2, F3) and 3 ground control mice (G1, G2, G3). We collected two coronal sections from each brain hemisphere containing hippocampus, somatosensory cortex, striatum, amygdala and corpus callosum. <\/p>\n
In total, we captured 14,630 genes across 29,770 spots after filtering and detected 10,884 UMIs\/spot and 3755 genes\/spot on average (Fig.2F, G; Supplementary Fig.3A, B) and found a high overall gene expression correlation between spaceflight and ground control tissue sections (r=0.99, p<0.05; Fig.2H). Unsupervised clustering analysis of spot information identified 18 distinct spatial clusters (further referred as ST clusters) (Fig.3A, B; Supplementary Data6), which presented a clear separation between the cortical top (ST cluster 1) and bottom layers (ST cluster 9), as well as other major structures, including hippocampus (with separation of CA1, CA3, and dentate gyrus in ST clusters 10, 8 and 11 respectively), thalamus (ST cluster 5), striatum (ST clusters 0, 14), hypothalamus (ST cluster 2), pituitary (anterior and posterior; ST cluster 2), corpus callosum (ST cluster 12) and cerebral peduncles (ST cluster 4) (Fig.3C). Key functions of the markers (Supplementary Data7) that were shared by numerous ST clusters include neurogenesis, neuronal development, axonal growth and synaptogenesis, indicating that ST cluster analysis is dominated by neuronal gene expression. <\/p>\n
A Clustering of spatial transcriptomics data, cluster annotations and spatial location of clusters visualized on flight and ground control mouse brain sections. B Marker genes for each ST cluster visualized as dotplot. C Spatial distribution of 3 genes (Wfs1 for CA1 region of hippocampus, Dkk3 for CA1 and CA3 hippocampal region and Prox1 for Dentate gyrus) in three flight (left column) and three ground control (right column) ST sections. D Significantly different pathways (p<0.05) between flight and ground control in ST cluster 9 (Cortical neurons, bottom layers). E Visualization of number of clusters identified by single-nuclei multiomics and their proportions in each ST cluster (x-axis; 017). Only multiomics clusters with higher proportions (>10%) are displayed in the barplot. F Cell type proportions mapped to spatial coordinates on three ground control (top row) and three flight (bottom row) mouse brain sections (Synaptic transmission I or multiomics cluster 1; Myelination or multiomics cluster 3; Neuronal activity, Synaptic transmission III or multiomics cluster 15). <\/p>\n
Next, we investigated how spaceflight influences gene expression at the spatial level and identified a total of 4057 DEGs in 7 out of 18 ST clusters (Supplementary Data8). The majority of DEGs were involved in neuronal development, synaptogenesis and synaptic plasticity, and neurodegeneration, including 21 DEGs in hippocampal CA3 neurons. The most pronounced change in gene expression due to spaceflight was observed in cortical neurons (bottom layers; ST cluster 9) which showed 3208 DEGs (1808 upregulated, and 1400 downregulated) with similar functions related to neuronal development and synaptic transmission in somatosensory, motor and visual cortex. Consensus pathway analysis18 highlighted neurodegeneration-associated pathways in cortical neurons (bottom layers; ST cluster 9) including protein misfolding and abnormal protein clearance, indicating potential similarities with neurodegenerative diseases characterized by protein misfolding and accumulation, such as Parkinsons disease19,20 (Fig.3D). <\/p>\n
To infer the spatial distribution of the clusters identified by multiomics, we performed spot deconvolution analysis on matching ST dataset using Stereoscope21 (which corrects for biases arising from different experimental techniques before calculating celltype proportions probabilities) (Fig.3E; refer to Supplementary Figs.46 for detailed visualizations of multiomics cluster proportions in ST dataset). The deconvolution analysis revealed similarities based on the assigned functional annotations between several multiomics and spatial data clusters, for instance, synaptic transmission (multiomics cluster 1 matched with ST clusters 0 and 2), myelination (multiomics cluster 3 matched ST clusters 4 and 12), and neuronal activity (multiomics cluster 15 matched ST cluster 5) (Fig.3F; Supplementary Figs.7, 8; Supplementary Data9). This comparative analysis suggested the effects of spaceflight on synaptic transmission specifically in cortex (including both neurons and astrocytes, as revealed by snRNA-seq data that allowed cell type separation) and on dopaminergic neuron development specifically in striatum (Supplementary Data9). <\/p>\n
To assess the effects of spaceflight on the cell-cell interaction level, we performed a ligand-receptor analysis on two multiomics clusters that showed among highest number of differentially expressed genes in response to spaceflight, i.e., multiomics clusters 4 (Astrocytes), and 11 (GABAergic Synaptic Transmission). We found 4 significantly upregulated interactions (Fig.4A), including adhesion molecule pairs, EGFR (epidermal growth factor receptor) pairs, and VEGFA (vascular endothelial growth factor). These ligand-receptor interactions have previously been shown to be involved in cellular development in the CNS. EGFR22, is involved in neuronal development, including axonal outgrowth. Meanwhile, VEGFA23,24 primarily regulates angiogenesis though it can also play a role in hippocampal neurogenesis, and astrocyte-produced VEGFA has previously been demonstrated to regulate neuronal NMDA receptor activity23,24,25. Interestingly, we found that spaceflight widely increased VEGFA_GRIN28 interactions between multiomics cluster pairs related to astrocytes and synaptic transmission, i.e., 4-11 (Astrocytes-GABAergic Synaptic Transmission). No ligand-receptor interactions in these clusters were significantly downregulated. <\/p>\n
A Dotplot showing the differentially expressed ligand receptor pairs found by CellPhoneDB between two interacting multiomics clusters (4 and 11) which are affected by spaceflight. These clusters showed the largest number of spaceflight DEGs, and four LR pairs were found significantly upregulated in these interactions. The null distribution of the mean expression of the LR pairs was estimated by employing a random permutation approach. The mean expression of the interacting LR molecule pairs are indicated by the dot colors and the dot sizes represent the p-values which refers to the enrichment of the LR pair in the interacting multiomics clusters. Scales for both dot size and color are presented below the plot. B Accessibility differences for motifs Atoh1, Zic1, and Zic2 in multiomics cluster 4 of flight mice and ground control mice. Spaceflight results in reduced accessibility of these motifs in flight samples. Two-sided Chi-square test statistic was used for differential testing with FDR correction (fdr <0.05). C Accessibility differences for motifs Pou5f1, and Sox2 in multiomics cluster 11 of flight and ground control mice. Spaceflight results in increased accessibility of these motifs in flight samples. Effects of spaceflight shown by increased accessibility of these motifs in flight samples. Two-sided Chi-square test statistic was used for differential testing with FDR correction (fdr <0.05). D (left) adjusted p-value of differential interactions found by MISTy in intraview (cell type and pathway activity colocalization) occuring only in flight (blue; n=3 individual ST flight mouse samples) or in controls (red; n=3 individual ST ground control mouse samples), tiles with black border identify statistically significant changes, (middle) correlation of MAPK pathway activity and Neurovasculature abundance, and mapped on Visium slide for two samples (right). Two-sided Students t tests with BenjaminiHochberg multiple testing correction was used to determine the differential interactions. E adjusted p-value of differential interactions found by MISTy in paraview (cell type and pathway activity in local neighborhood) occuring only in flight (blue; n=3 individual ST flight mouse samples) or in controls (red; n=3 individual ST ground control mouse samples), tiles with black border identify statistically significant changes. Two-sided Students t tests with BenjaminiHochberg multiple testing correction was used to determine the differential interactions. F Pearson correlation of Glis3 activity (left) containing vascular endothelial cells and MAPK activity (n=6 individual ST mouse samples, 3 flight, 3 ground controls), and their respective activities in Visium slides (4 plots on the right). Two-sided Students t-tests with BenjaminiHochberg multiple testing correction was used to determine the changes in correlation. G Pearson correlation of Lef1 activity (left) within spots containing vascular endothelial cells and MAPK activity, and their respective activities in Visium slides (4 plots on the right). Two-sided Students t tests with BenjaminiHochberg multiple testing correction was used to determine the changes in correlation. multiomics cl: multiomics cluster. The boxplots in D, F, and G show the median as a central line, the box boundaries denote the first and third quartiles and the whiskers extend to the most extreme point in the range within 1.5 times the interquartile range from the box. <\/p>\n
We also extended the ligand-receptor analysis to the ST dataset using SpatialDM26. We applied SpatialDM on each ST brain section to identify spatially co-expressed LR pairs and found a total of 1260 LR pairs (Supplementary Fig.9; refer to Supplementary Data10 for a detailed list of LR pairs with corresponding z-scores across each ST section). Differential testing between the two conditions (flight and ground control) for the observed 1260 LR pairs revealed a total of 134 differential LR pairs (differential p-value<0.1; Supplementary Data11). <\/p>\n
To investigate the effects of spaceflight on transcription factors (TFs), we performed motif analysis on snATAC-seq peaks from the single nucleus multiomics data, which revealed spaceflight-mediated differences in TF activity in several multiomics clusters (Supplementary Data12), especially 4 (Astrocytes), and 11 (GABAergic Synaptic Transmission). <\/p>\n
Spaceflight was associated with reduced accessibility of motifs Zic1, Zic2 and Atoh1 in multiomics clusters 4 (Astrocytes)27,28 (Fig.4B). Meanwhile, increased accessibility of motifs Pou5f1 and Sox2 in multiomics cluster 11 (GABAergic Synaptic Transmission) might indicate reduced neuronal differentiation in spaceflight29,30,31 (Fig.4C). In addition to neuronal effects, motifs Pparg, Rxra and Nr2f6, which collectively inhibit immune responses, showed decreased accessibility in telencephalon interneurons (multiomics cluster 11), suggesting increased inflammatory responses in space32,33,34, and possible circadian dysregulation35,36,37,38,39. <\/p>\n
Local environments of cell types may affect the functional responses to spaceflight represented by changes in signaling pathways. We compared key signaling pathways in adjacent locations based on the spatially-resolved cell type deconvolution results from Stereoscope analyzed using the Multiview intercellular SpaTial modeling framework (MISTy)40. This tool allowed us to investigate the relationships between cell type proportions in each ST spot and activities of 14 pathways inferred by decoupler-py and PROGENy41,42. Specifically, the MISTy models predict cell type abundance in a spot based on an intraview (features in the same spot) and paraview (weighted sum of the features in the neighboring spots; weights decreasing with distance). Either cell type abundances or pathway activities were selected as features for the model, and a separate model was built for each sample and cell type. To analyze the effects of spaceflight, the models were subsequently aggregated into flight and ground control groups. <\/p>\n
Based on cell type abundances, we did not find any significant changes in cell type colocalization (which would occur during tissue restructuring or lesion formation) between flight and ground controls, similar to our previous finding of no significant changes in cell type abundance in deconvolution results (Supplementary Figs.7 and 8). <\/p>\n
In contrast, changes in signaling pathways were associated with individual cell types. Cell abundance in neurovasculature (multiomics cluster 12) colocalized with decreased MAPK signaling in spaceflight (Fig.4D). Similarly, signaling changes in local neighborhood (MISTy paraview) of several other cell types were found in spaceflight samples (Fig.4E): (1) less negative correlation of EGFR signaling and glutamatergic neuronal cells; (2) more negative correlation of MAPK and cholinergic, monoaminergic and peptidergic neurons; (3) increased TGFbeta signaling in the vicinity of GABAergic interneurons; (4) reduced WNT signaling in class II glutamatergic neurons. <\/p>\n
To assess the downstream effects of these changes, we built a tissue-specific gene regulatory network (GRN) from the multiomics data using CellOracle43 and used it to predict TF activities in spatial data and computed the Pearson correlation between TF and signaling activities for the dysregulated pathways in spots containing the cell types identified above. The network suggested that the decrease in MAPK signaling in spaceflight increases activity of the transcription factor Glis3 and reduces Lef1 in neurovasculature, respectively (Fig.4F, G). <\/p>\n
Gene Set Enrichment Analysis (GSEA) on the ST data using metabolic pathways indicated spaceflight-mediated inhibition of the oxidative phosphorylation pathway, especially Complex I signaling (Fig.5A, Supplementary Data13), as well as pathways related to glycolysis\/gluconeogenesis (Supplementary Fig.10), fructose and mannose metabolism (Supplementary Fig.11) and arachidonic acid metabolism (Fig.5B). Analysis of multiomics data was consistent with spaceflight-mediated reduction in these pathways together with fatty acid synthesis (Fig.5C; Supplementary Data14). Deficits in glycolysis and oxidative phosphorylation are consistent with previously reported mitochondrial impairments caused by spaceflight44, while, arachidonic acid is primarily produced by astrocytes and suggests astrocyte dysfunction as a potential target for future spaceflight CNS studies. <\/p>\n
A Heatmap showing fold change differences (log2FC) between flight and ground control samples in oxidative phosphorylation pathway in both ST and multiomics datasets. There is a spaceflight-mediated inhibition seen for this pathway that is consistent across the two datasets. Two-sided Wilcoxons rank-sum test was done with FDR adjustment. B Heatmap showing fold change differences (log2FC) between flight and ground control samples in Arachidonic acid metabolism pathway in both ST and multiomics datasets. There is a deficit for this pathway seen in spaceflight samples in both the datasets. Two-sided Wilcoxons rank-sum test was done with FDR adjustment. C Heatmap showing fold change differences (log2FC) between flight and ground control samples in Fatty acid synthesis pathway in both ST and multiomics datasets. There is a spaceflight-mediated reduction observed for this pathway in both the modalities. Two-sided Wilcoxons rank-sum test was done with FDR adjustment. multiomics cl: multiomics cluster. <\/p>\n
In order to validate our findings on the spaceflight affected processes in mouse brain, we performed single molecule Fluorescence In situ Hybridization (smFISH) using the RNAscope technology for two genes of interest (Adcy1 and Gpc5) in five brain sections: 3 flights, 2 ground controls (Supplementary Fig.12) from a comparative set of mice (see Methods). We observed significant upregulation in the expression of both genes in spaceflight samples, confirming our findings from the ST data and multiomics data analysis (Supplementary Data3 and 8, Supplementary Fig.13AC). Adcy1 was particularly upregulated in the hippocampus and associated with changes in neuronal activity (ST clusters 8, 11), while Gpc5 was upregulated in astrocytes (multiomics cluster 4). <\/p>\n
<\/p>\n
Here is the original post: <\/p>\n
Spatially resolved multiomics on the neuronal effects induced by spaceflight in mice - Nature.com<\/a><\/p>\n","protected":false},"excerpt":{"rendered":" To identify specific cellular microenvironments affected by spaceflight, we combined the techniques of spatial transcriptomics (ST; 10X Genomics Visium) and single-nucleus multiomics (snMultiomics; gene expression and chromatin accessibility; 10 Genomics Single Cell Multiome ATAC+Gene Expression) on mouse brain. In total, we analyzed three brains from mice euthanized on-board of the International Space Station (ISS; F1, F2, F3) and three brains from ground control mice (G1, G2, G3) that were kept under matched conditions (see Animals in Methods) Continue reading