Transcriptome-wide changes in response to spaceflight

To understand the impact of spaceflight to skin and tissue microenvironment, paired 4mm skin punch biopsies from Inspiration4 crew members upper arms were used for pathology evaluation and spatial transcriptomics profiling (Fig.1a and Supplementary Fig.1). In total, 95 ROIs were collected across 16 slides for processing, with the GeoMx whole transcriptome profiling probe set (18,422 probes). Based on imaging we selected four region types of interest, including the outer epidermis, inner epidermis, outer dermis, and the vasculature (OE, IE, OD, and VA). We also performed a skin histopathology analysis from the biopsied samples, which showed no significant abnormalities or changes in tissue morphologies or gross architecture (Supplementary Fig.2).

a Experimental design and workflow with representative tissue staining images (created with BioRender.com), b Uniform Manifold Approximation and Projection (UMAP) of all ROIs collected, c Volcano plot of overall post- vs. pre-spaceflight DEGs (using DESeq2 method), d Pathway enrichment analysis comparing DEGs from pre- and post-spaceflight skin tissues, visualizing normalized enrichment scores of MSigDB Hallmark pathways, and e Cell proportion comparisons between pre- and post-spaceflight samples (ns non-significant, *p0.05, **p0.01, ***p0.001, and ****p0.0001 by Wilcoxon test, two-sided; boxplot shows median/horizontal line inside the box, the interquartile range/box boundaries, whiskers extending to 1.5 times the interquartile range, and outliers as individual points outside the whiskers; exact p values are included in the Source Data). Source data are provided as a Source Data file.

From GeoMx spatial transcriptomics analysis, unsupervised clustering of all ROIs showed large clustering around compartmental identities. Slight shifts in response to spaceflight, and batch effects from both technical and biological replicates were not apparent after normalization (Fig.1b and Supplementary Fig.3a). Differential gene expression analysis comparing post-spaceflight to pre-spaceflight samples found significant upregulation in 95 genes (log2FC>0 and q value<0.05 by DESeq2) including ARHGAP31, GALNT9, CPNE2, NMB, GPR50, CLDN2, OOSP2, and downregulation in 121 genes (log2FC<0 and q value<0.05 by DESeq2) such as AP3B1, LMNA, COL6A2, VIM, HLA-B, PPP1CB, PABPC1 (Fig.1c and Supplementary Data1). Furthermore, proteins associated with cell junctions and extracellular matricesparticularly those from vimentin (VIM) and keratin (KRT) familywere the primary transcripts lost based on the DEG analyses.

Pathway analysis of these differentially expressed genes (DEGs) revealed statistically significant enrichment in kirsten rat sarcoma viral oncogene homolog (KRAS) signaling pathways, while transcripts associated with cell junctions and protein (i.e., apical junction, unfolded protein response) decreased (Fig.1d and Supplementary Data2). From expression levels, cell type composition for each ROI was estimated and compared across timepoints. We also observed statistically significant decreases in the cell type associated gene signatures of the major skin cell types and immune cells (e.g., melanocyte, pericyte, fibroblast, and T cells) (Fig.1e).

We then investigated region-specific expression changes across pre- and post-spaceflight samples for each ROI type label (OE, IE, OD, and VA). OE and IE regions were selected based on and corresponds to stratum granulosum and spinosum/basal, respectively. OD ROIs were selected by capturing a minimum of 200 cells inside of the basal cell layer (therefore mostly papillary layer), while VA ROIs were collected based on epithelial (FAP) and fibroblast (SMA) staining (Fig.1a). We observed transcripts specific to each ROI label and timepoint (Supplementary Fig.3b, c).

For each ROI type, differential gene expression analyses were performed comparing postflight samples relative to preflight samples (Fig.2a and Supplementary Data1). For example, we found that the decrease in transcripts related to fibroblast and junction genes (e.g., DES, ACTA2, TLN1, TAGLN) specifically near the vasculature sites (VA). Loss of KRT14 as well as other keratin family transcripts (KRT1, 5, and 10) were found predominantly in the dermal layer (OD). Taking the intersections of these DEGs to identify unique and overlapping genes across ROI types, we confirmed that most of the gene overlaps occur within ROI types that are relatively close to each other (i.e., VA and OD) (Fig.2b). In particular, changes in AP3B1, a transcript related to granule formation, cytokine production, and inflammatory responses, were found in multiple comparisons (overall, OE, and OD) and was orthogonally validated with another technology, RNA scope (Supplementary Fig.4ac)23. In the inner layers of the tissue (OD and VA), we found overlapping DEGs related to stress and growth factor associated pathways, such as COL6A2, CRKL, HLA-B.

a Volcano plot showing DEGs by ROI typesOE, IE, OD, and VA respectively; the number of DEGs were determined by cutoffs of adjusted p value<0.1 and |log2FC|>0.5 (using DESeq2 method), b UpSet plots comparing the intersections of region-specific DEGs, c Hallmark, non-germline gene set enrichment analysis across four ROI types; NES Normalized Enrichment Scores; Arrow indicates tissue locations, where OE is the outermost layer and VA is the innermost layer. Source data are provided as a Source Data file.

Gene set enrichment analysis (GSEA) revealed the consistent increase of KRAS signaling and inflammatory responses across all regions while specific immune pathways such as Interferon alpha and gamma response showed positive enrichment only in epidermal regions (OE and IE) (Fig.2c and Supplementary Data2). Pathways such as DNA repair, apoptosis, and UV response, reactive oxygen species were enriched only in the OE. We observed downregulation in genes involved with mitochondrial metabolism (e.g., myc target genes and oxidative phosphorylation) across all regions, particularly stronger in IE and OD ROIs. Also, the myogenesis pathway and EMT-related genes showed stronger decrease in enrichment scores in the VA ROIs, underscoring the region- and layer-specific responses to spaceflight. Comparing the pathway-level changes to blood sequencing datasets from the same mission and previous mission (NASA Twin Study, although with different duration of exposure), we found consistent changes in pathways such as KRAS signaling, epithelial-to-mesenchymal transition, and angiogenesis (Supplementary Fig.4d)5.

In addition to differential analyses, we also found that the marker genes reported to be specific to each skin layer and cell type corresponded to the expression levels in each ROI type and were consistent with the previous findings (Supplementary Fig.3b, c)24,25,26. Based on the reference datasets, deconvolved cell type abundances were compared across ROI types and timepoints (Supplementary Fig.5a). We found a loss of melanocyte related gene signatures specifically in the middle layers (IE and OD), not in the outermost region (OE) or vascular region deeper in the dermal layer (VA). On the contrary, fibroblast related gene expressions were decreased across all regions except for the outermost epidermal layer (OE). Although fibroblast is an unanticipated cell type in the epidermis ROIs, decreased fibroblast signature could indicate loss or damage of cellular and matrix interactions, consistent with previous reports highlighting the role of fibroblasts with epidermal regeneration (Supplementary Fig.5b, c)27,28.

To investigate the phenotypic impact of spaceflight, we then focused on genes and pathways related to skin barrier formation, disruption, and regeneration. From the pathway analysis, we found enrichment changes in apical junction, UV stress response, hypoxia, and TGF signaling (Fig.2c and Supplementary Data2). Specifically, we observed a decrease in filaggrin (FLG) expression, a gene related to skin barrier function and plays a crucial role during epidermal differentiation by controlling interactions across cytoskeleton components, in postflight relative to preflight samples29. The decrease of FLG was the most evident in the OE region (Supplementary Data1). Related to this observation, we also observed decreases in transcripts such as HAS1, HAS2, HAS3, OCLN, CLDN, TGM2 in the OE region (Fig.3a).

a Gene expression changes of interest, b fold change of proportions in post-flight samples relative to pre-flight samples, by compartments, c cell type correlation matrix changes. Black boxes represent undetermined spots (due to minimal cell counts); boxes with X marks represent correlations that did not pass statistical testing (p value<0.05, Pearson correlation, two-sided). Source data are provided as a Source Data file.

The decrease in protein production and response potentially are connected to decrease in keratinocyte and increase in immune signatures (potentially related to interactions with T cells and fibroblasts) in OE region ROIs (Fig.3b)30. Although weaker, the IE region shows a similar trend of cell proportion fold changes. Specifically, among fibroblast populations we also found that gene signatures of reticular fibroblast increased in postflight samples while there were no statistically significant changes in papillary fibroblast, suggesting disruptions in regeneration processes (Supplementary Fig.5b, c)31,32. Taking co-occurrence of the proportion changes, cellular interactions within the ROIs were estimated. While cluster disruption was relatively minimal, an increase in melanocyte-macrophage interactions were found in the epidermis (OE and IE) ROIs (Fig.3c). In addition, expression changes related to vascular and lymphatic endothelial cells and pericytes varied across the skin regions. The most pronounced cell signature changes were seen in the OE and VA compartments. In the OE compartment, we observed an increase in signatures related to lymphatic endothelial cells, potentially indicating the changes in the skins vascular and immune system (Fig.3b). While blood and lymphatic capillaries are not typically found in the epidermis, these adaptations may be suggestive of a wound-healing phenotype with the skin, which is supported by our results showing increased damage, inflammation, apoptosis, ROS, hypoxia, angiogenesis, TGF-beta expression, etc., in the epidermis (Fig.2c)33,34. On the other hand, in the VA compartment, there was an increase of gene signatures related to blood endothelium and decrease in lymphatic endothelium, also associated with vascular remodeling events.

To test whether immune activation and epithelial barrier disruption can be explained with external environmental change, we performed metagenomics and metatranscriptomics analysis on the skin swabs collected right before biopsies (Supplementary Fig.6a). After assignment of taxonomic labels to DNA sequences, we identified 826 bacterial and 9819 viral species with non-zero counts from metagenomics analysis, and 88 bacterial and 1456 viral species from metatranscriptomics analysis (Supplementary Data3). From PCA analysis, no major clustering was observed, although post flight samples were located closer to one another in the PCA space (Fig.4a). The shifts of the samples were mostly from species from Staphylococcus and Streptococcus family, along the PC2 axis. Slight decrease in overall numbers of bacterial and viral species was observed in postflight samples relative to preflight, with one exception of C003 in metagenomics data and of C004 in metatranscriptomics data (Fig.4b). Gross comparison of bacterial species by family showed decreased abundance in Actinobacteria (e.g., Actinomyces sp000220835) while increased abundance in Firmicutes/Bacillota (e.g., Peptoniphilus C/B) and Proteobacteria/Pseudomonadota (e.g., Caulobacter vibrioides, Sphingomonas carotinifaciens, Roseomonas mucosa/nepalensis) (Fig.4c, d and Supplementary Fig.6b). When grouped into genus, several species, including Cutibacterium (e.g., Cutibacterium acnes/avidum/modestum/porci), Mycobacterium (e.g., Mycobacterium paragordonae, Mycobacterium phocaicum), and Pseudomonas (e.g., Pseudomonas aeruginosa/nitroreducens) showed statistically significant decrease (p values<0.05). Several species including Streptococcus (e.g., Staphylococcus capitis, Streptococcus mitis BB) and Veillonella (e.g., Veillonella atypica/parvula/rogosae) showed significant increase (Fig.4d). Also, species under the Staphylococcus genus, such as staphylococcus capitis/epidermidis/saprophyticus showed slight decrease while the relative abundances were highly variable across biological replicates.

a PCA across all metagenomic and metatranscriptomic (bacterial and viral reads) relative abundance features and all crew members pre- and post-flight, b Total number of bacterial and viral species with nonzero counts, c Relative abundances by sample and timepoint, grouped by family, d Changes in relative abundance before and after spaceflight, grouped by genus; statistically significant or previously reported microbes are visualized (two-sided Wilcoxon test across four crew members was performed to compare means between pre- and post-flight samples and to obtain p values, and error bars represent the standard error of the mean), and e Correlation across relative abundance of bacterial phyla identified by metagenomics data and known barrier/immune genes associated with skin diseases and disruptions. Source data are provided as a Source Data file.

Changes of bacterial species were then linked to skin gene expression profiles, especially dermatitis-related genes (i.e., STAT3, STAT5B, FLG, CDSN, and ADAM17) previously associated with Staphylococcus species, as Staphylococcus aureus-dependent atopic dermatitis have been reported to activate immune system and reduce microbe diversity35,36,37 (Fig.4e and Supplementary Fig.6c). When subsetting previously reported bacterial species and associated genes, we found Staphylococcus species show an inverse relationship with JAK1 (Fig.4e). In particular, Staphylococcus correlates closely to FLG, SPINK5, and DSG1, all of which are related with epithelial barriers (stratum corneum and junctional barriers)38. Also, microbes belong to Carnobacteriaceae, Lactobacillaceae, Nanosynbacteraceae, and Weeksellaceae families showed high correlation with both barrier and immune genes (CDSN, DSP, DSG1, SPINK5, FLG, and JAK1), whereas common skin microbes from Dermatophilaceae and Dermabacteraceae families showed no correlation. Although larger sample size is needed, it is possible that skin microbiome disruptions, such as those observed in these bacterial families, also contribute to barrier disruption and immune activation during short-term spaceflight.

In addition, from alignment to known viral assemblies we found statistically significant decrease in abundance of reads associated with those from Uroviricota (i.e., Fromanvirus, Acadianvirus, Armstrongvirus, Amginevirus, Bixzunavirus) and Naldaviricetes (i.e., Alphabaculovirus), and increased abundance of reads associated with those from Negarnaviricota (i.e., Almendravirus, Orthotospovirus) and Cossaviricota (i.e., Betapapillomavirus, Betapolyomavirus) (p values<0.05). Virome changes are limited by the depth of the sequencing and skin virome knowledge, however we also report relative abundances of both bacterial and viral species (Supplementary Data3). To explore microbiota-skin interactions, we also identified potential associations between microbiome shifts from metagenomics/metatranscriptomics data and human gene expression from skin spatial transcriptomics data; these included associations were with viral phyla (i.e., Uroviricota, Cressdnaviricota, Phixviricota), which is a potentially interesting area to explore as more crew samples are collected. (Supplementary Fig.6d, e and Supplementary Data3).

To investigate immune changes that occur beneath the epidermis we also examined changes in immune cells in the profiled vascular regions vs. PBMCs. We saw overall decrease of T cells and increase of macrophage DCs in VA ROIs (Fig.3b), indicating an immune-epidermis interaction. Related to this, we also observed increased cytokines and inflammatory signals including IL4, IL5, and IFNG in the inner regions (VA and OD ROIs) of the tissue (Fig.5a)39,40. As a confirmation, we observed that these specific cytokines are also shown to be increased in cytokine assays from the crew members serum samples (Fig.5b). To compare immune change observations from VA ROIs to system-wide immune system changes, we performed leveraged 10X multiome sequencing (dual snRNA and ATAC sequencing from each cell) on timepoint-matched PBMCs from the crew members (Supplementary Fig.7a). We analyzed 151,411 cells across 9 gross cell types and performed differential expression analysis (Supplementary Fig.7b, c). Overall, we observed fluctuations of T-cells across timepoints, consistent to the observations from skin spatial transcriptomics data (Fig.5c, d and Supplementary Fig.7d). Among 555 DEGs from multiome samples and 446 DEGs from GeoMx VA ROIs, 12 overlapping DEGs were found (both log2FC>0.1 and p values<0.01, DESeq2), including ATP11A, CEP85L, CEPT1, DMXL1, DOP1A, EVI5, GSAP, MDFIC, SENP7, TBCK, VAV3, and VPS13C (Fig.5c and Supplementary Fig.7c). Several of these genes are related to cellular metabolism and cytosolic transports. In particular, VAV3, one of signaling adapters in NK/T cell activation, has been previously reported to be associated with atopic dermatitis onset41,42,43. While all these overlapping DEGs were temporary in PBMCs, i.e., upregulated in the immediate postflight samples (R+1 timepoint) and returned to pre-flight expression levels, the chromatin accessibility of these genes stayed slightly longer, up to R+45 timepoint (Fig.5d).

a Notable cytokine changes and locations from (a) skin transcriptomics data by region and, b cytokine assay from serum samples (sig. Indicates overall statistical significance of the cytokine levels in the postflight samples relative to the preflight samples, where red indicates significantly increased, and green means stable/no change; two-sided Wilcoxon test was done with the p value cutoff of 0.05), c Comparison of DEGs between PBMC multiome data and spatial transcriptomics data from VA ROIs, d Dot plots visualizing mRNA transcript expression levels (left) and gene activity score from ATAC signals (right), where preflight samples were collected 44 days before launch (L-44) and postflight samples were collected 1, 45, and 82 days post return (R+1, R+45, and R+82, respectively), e Flight and cell type specific gene signature enrichment in spatial data by timepoint and ROI types, f gene signature enrichment analysis using gene signatures built from skin disease-related gene expression profiles; two-sided Wilcoxon test across four crew members and 95 ROIs was performed to obtain p value, where *p0.05 and **p0.01, and error bars represents standard deviation of the mean. Source data are provided as a Source Data file.

Finally, we derived cell type- and spaceflight-specific gene signatures from the multiome data, to examine any enrichment in the GeoMx samples (using single-sample gene set enrichment analysis, or ssGSEA approach) (Fig.5e). Most of the immune cell specific postflight DEGs enrichments were near the innermost ROIs (OD and VA), except for T cells (both CD4+ and CD8+), which showed enrichment in the postflight OE ROIs. While it was previously reported that spaceflight stressors change the immune system, increased enrichment of the T cells in the epidermal region correlates with activated T cell activity and connects to inflammatory responses and barrier disruptions44,45,46,47,48. Lastly, we found that these increased T cell signatures in the OE region may not have direct connection to Th17 T cells or psoriasis, rather have closer connection to the antigen-associated and lymphatic T cells infiltrated from inner layers of the skin (Fig.5f)49,50. Also, the ssGSEA analysis using skin disease-associated gene signatures showed a slight increase in melanoma signatures. The slight increase can be explained with previous observations throughout this manuscript, including increase in cell death, immune activation, and stress response (Supplementary Fig.7e, f), but more research is needed to prove the direct connection or causality of gene expression shifts.

Here is the original post:

Spatial multi-omics of human skin reveals KRAS and inflammatory responses to spaceflight - Nature.com