The NASA Twins Study: A multidimensional analysis of a year-long human spaceflight

Extensive multisystem changes occur in spaceflight

Data were generated from fresh (FR) and ambient return (AR) whole blood or PBMCs that were either directly processed and fractionated into subpopulations (T cells: CD4 and CD8; B cells: CD19; and lymphocyte depleted: LD) (fig. S1) or processed directly from frozen PBMCs. For fractionated subpopulations, the inflight collections were confounded by ambient return (AR specimens were collected on the ISS and returned to Earth via Soyuz capsule at room temperature). To simulate this effect, independent samples were generated to mimic FR and AR sample collection conditions and used for correction. Unsupervised analysis of gene expression profiles segregated TW’s inflight from preflight and postflight samples, as well as from HR specimens, for all cell populations examined (fig. S2, A to E). Similar expression changes were observed for all cell types assessed (Fig. 1B). Differentially expressed genes (DEGs, DESeq2 multivariate negative binomial model, q < 0.01) were identified across all samples and cell types (n = 9317) with HR as the control and a particular focus on the differences that persisted postflight. In the LD cell population, of the 481 DEGs found inflight, compared with preflight, there were 112 (23.2%) that remained differentially expressed postflight (Fig. 1B and table S2). Across all sorted cell types, only 811 of the inflight DEGs remained disrupted postflight. Concomitant with an overall shift in mitochondrial RNA (mtRNA) levels (see below), far more DEGs (n = 8564, DESeq2 multivariate negative binomial model, q < 0.01) were observed in the 6- to 12-month inflight period compared with HR than in the 0- to 6-month inflight period (n = 1447) in PBMCs. This large (6×) shift in overall gene regulation upon exposure in TW to a year-long spaceflight suggests that overall gene regulation may be more affected by longer-duration spaceflight.

To assess physiological and immune signaling changes, the metabolome, proteome, and cytokine complement were measured using mass spectrometry and antibody-based methods. Targeted and untargeted metabolomics and proteomics were performed on plasma and urine samples, whereas cytokine profiling was performed only on plasma samples (fig. S3 and table S1). Untargeted analysis was focused on a high-confidence set of 719 metabolites (out of 4162 metabolic features) identified in plasma using a broad-spectrum liquid chromatography–mass spectrometry (LC-MS) metabolomics platform (18). Linear models demonstrate significant differences [false discovery rate (FDR) < 0.05] in the abundance of 245 metabolites among pre-, in-, and postflight samples between TW and HR, including those specific to the earlier or later halves of the flight period as well as those associated with time spent on the ISS (Fig. 1C and table S3). Additionally, 39 metabolites were significantly different (FDR < 0.05) between TW and HR at baseline (table S3). Metabolites indicative of genotoxic stress, inflammation, and altered amino acid metabolism were increased in TW inflight (table S3). Targeted metabolomic analysis using gas chromatography–tandem mass spectrometry (GC-MS/MS) was performed in plasma and urine to quantify metabolites involved in the tricarboxylic acid (TCA) cycle, glycolysis, and amino acid, fatty acid, ketone body, and pyrimidine metabolism. Of the 81 targeted metabolites, 60 and 66 metabolites were detected in plasma and urine, respectively (table S3). No significant changes were observed in plasma metabolites, but 32 urinary metabolites were significantly altered inflight for TW [FDR: P < 0.05, analysis of variance (ANOVA)]. To assess the epigenetic contribution to the changes observed, whole-genome bisulfite sequencing (WGBS) was performed on a subset of CD4 and CD8 samples from each subject. Global DNA methylation changes in TW were within the range of variation seen in HR throughout the study. However, ontology analysis of genes ranked on the basis of epigenetic discordance, as quantified by average promoter Jensen-Shannon distance (JSD) between samples, revealed distinct pathways enriched in TW when comparing preflight with inflight samples (Fig. 1D and tables S4 and S5). For example, in CD8 cells, enrichment in regulation of neutrophil activation or granulation was observed in TW [q < 0.01, hypergeometric test (HGT)], but not in HR (Fig. 1D). Similarly, in CD4 cells, enrichment in platelet aggregation (q = 0.0002, HGT) was observed only in TW.

To identify complex changes spanning diverse analyte classes that occur over time, the different data types were combined and scaled, and c-means clustering analysis was performed. Although this analysis (Fig. 1E) revealed that physiological levels for a large number of analytes were unaffected by spaceflight (Fig. 1E; clusters 1 and 2), several notable patterns were identified. These included a cluster of features, including cognitive performance, that for TW increased inflight and declined sharply upon return before reaching baseline (Fig. 1E, cluster 3; and Fig. 1F); this cluster was enriched for analytes involved in central carbon metabolism [q < 0.0002; integrated molecular pathway level analysis (IMPaLA) combined pathway analysis, Fisher’s method]. Integrating gene expression changes with the other analytes revealed additional complex biochemical trajectories; for example, gene expression changes associated with neutrophil regulation were enriched (q < 1 × 10⁻¹²; IMPaLA combined pathway analysis, Fisher’s method) in cluster 25 (fig. S5), illustrating an increase very early inflight followed by a steady decline, until increasing again postflight.

Telomeres lengthened during spaceflight

Telomeres are repetitive features of chromosomal termini essential for maintaining genomic integrity; they protect physical DNA ends from degradation and prevent them from triggering inappropriate DNA damage responses (DDRs). Telomere length shortens with cell division and thus with age, as well as with a variety of lifestyle factors, such as stress, and environmental exposures, including air pollution and radiation. Here, average telomere length was evaluated pre-, in-, and postflight (DNA; PBMCs) using quantitative real-time polymerase chain reaction (qRT-PCR) (19, 20). Consistent with a strong genetic component (21), HR and TW had similar telomere lengths at baseline (preflight P = 0.942; one-way ANOVA), and telomere lengths for HR remained relatively stable for the duration of the study (Fig. 2A). Most notable was a significant increase in telomere length during flight for TW (14.5%), as compared with his preflight and postflight measures as well as with those of HR (P = 0.048, 0.0003, and 0.0073, respectively; one-way ANOVA). TW’s increased telomere length was observed at all inflight time points assessed [flight day (FD) 14 to FD334; fig. S6A], as well as in sorted CD4, CD8, and LD cells, but not in CD19 cells (Fig. 2B and fig. S6B). These results are consistent with recently reported cell type–specific responses to factors that contribute to telomere length regulation (22). Notably, telomere length shortened rapidly upon TW’s return to Earth, within ~48 hours [FD340 ambient return to R+0 (R+ days post return); fig. S6B] and stabilized to near preflight averages within months.

Additionally, metaphase chromosomes were prepared from pre-, in-, and postflight samples (stimulated T cells; ambient whole blood), and thousands of individual telomeres were evaluated with telomere fluorescence in situ hybridization (Telo-FISH) (23, 24), an independent measure that showed the same trends observed with qRT-PCR (fig. S6C). Importantly, Telo-FISH cell-by-cell analyses also enabled generation of telomere length distributions that confirmed HR’s relatively stable telomere lengths throughout the study (Fig. 2C and fig. S6D). Consistent with TW’s increased average telomere length during spaceflight, temporal shifts toward increased numbers of longer telomeres and concurrent decreases in the numbers of shorter telomeres were also observed (fig. S6D); for example, there was a significant shift in TW’s inflight (FD259) telomere length distribution as compared to his preflight (L−162; L− days before launch) distribution (P < 2.2 × 10⁻¹⁶; Mann-Whitney U test) (Fig. 2D). After returning to Earth (R+190), TW’s telomere length distribution shifted back toward increased numbers of shorter telomeres, more closely resembling his preflight distribution. However, an increased number of signal-free chromosome ends, indicative of complete loss and/or critically short telomeres (i.e., below our level of resolution), was also apparent (Fig. 2D). Spaceflight-specific shifts in telomere length dynamics were also supported by gene expression results, in that when inflight, TW’s down-regulated DEGs were significantly enriched in packaging-of–telomere ends and telomere-maintenance pathways, in both CD4 and CD8 cells [gene set enrichment analysis (GSEA), all q ≤ 0.001] (table S2). Similarly, changes in DNA methylation of the gene promoter for TERT, which encodes the regulatory subunit of telomerase, were observed when comparing preflight with the late inflight time point in both CD4 and CD8 T cells (JSD rank 248 and 32, respectively); by contrast, for HR, TERT was not highly ranked (428 or greater) (table S5).

Telomerase is the reverse transcriptase capable of de novo addition of species-specific telomeric repeats onto the ends of newly replicated chromosomes (25). Telomerase activity was evaluated pre-, in-, and postflight (PBMCs, ambient) with the qRT-PCR telomeric repeat amplification protocol (TRAP) (24, 26). In general, TW had lower levels of telomerase activity than HR throughout the study, whereas levels for the two individuals were similar preflight and postflight (fig. S6E). Unfortunately, telomerase activity was lost in all inflight samples collected onboard the ISS. Given the sensitivity of enzymatic activity to sample handling conditions, we speculated that unavoidable transit time and/or temperatures associated with ambient return (via Soyuz) destroyed telomerase activity. To test this hypothesis, we compared ground control samples that had been in transit ≤24 hours with those that had been in transit ≥48 hours, mimicking the transit time from the ISS to Johnson Space Center. Loss of telomerase activity was observed in all blood samples experiencing transit times ≥48 hours (P = 0.002; Student’s t test); however, telomere length and DNA integrity were unaffected (fig. S6, F to H).

DNA damage responses during spaceflight

Chromosome aberrations were analyzed in the same samples as above to evaluate potential telomere-related instability (e.g., fusions) as well as to assess DDRs to ionizing radiation (IR) exposure during spaceflight (27), specifically galactic cosmic ray (GCR)–induced cytogenetic damage, as has been done previously in blood lymphocytes of astronauts (28). Here, high-resolution directional genomic hybridization (dGH) paints (29) for chromosomes 1, 2, and 3 were used to enable simultaneous detection of interchromosomal translocations and intrachromosomal inversions (Fig. 2E). Baseline frequencies of these structural variants were similar for the two subjects, with inversions being more frequent than translocations (Fig. 2F). TW’s inversion frequencies increased at a greater rate than translocations, consistent with inflight IR exposure, particularly to high linear energy transfer (LET) space radiation (30, 31). Indeed, a physical dosimeter dose of 76.18 milligrays and an effective dose of 146.34 millisieverts were recorded by NASA. Also consistent with IR exposure, genes whose expression was altered inflight were significantly enriched in pathways related to DNA damage responses in LD, CD4, CD8, and PBMCs [generalized linear model (GLM), all q ≤ 0.002; table S2]. Furthermore, TW’s inversion frequencies continued to rise postflight, indicative of instability and possibly reflective of IR-induced DNA damage to stem cell compartments. Although TW’s translocation frequencies were variable, translocations were also increased inflight and postflight compared with preflight, additional evidence of inflight radiation exposure and the persistence of GCR-induced cytogenetic damage (32).

DNA methylation patterning during spaceflight

We also examined DNA methylation in FR and AR specimens (Fig. 1A and table S1). Genome-wide DNA methylation levels were measured in CD4 and CD8 lymphocytes isolated from both subjects at four time points throughout the study: one preflight, two inflight (early and late), and one postflight. Principal components analysis (PCA) revealed a distinction in global DNA methylation between the two cell types and between subjects for the CD4 samples (Fig. 3A). Moreover, TW’s inflight samples were distinct from preflight and postflight samples for both cell types. Subtle differences in both genome-wide mean methylation levels (MML) and normalized methylation entropy (NME; a measure of methylation stochasticity) were detected inflight for TW, which returned to baseline postflight (Fig. 3B). Although for HR, the direction of these changes was consistent in CD4 and CD8 cells, it was reversed for TW, with a decrease in MML at FD259 for CD4 cells and an increase in CD8 cells. Although MML and NME are not necessarily predictive of one another (33), an inverse relationship between genome-wide medians of MML and NME is evident across all samples. Importantly, the magnitude of changes in both MML and NME was greater for HR and did not fully return to baseline. These data indicate that genome-wide DNA methylation changes detected in CD4 and CD8 cells from TW were within the range of variation seen in HR.

These results identify only limited epigenetic discordance in global methylation and entropy levels. Therefore, we quantified both global and local epigenetic discordances between samples by calculating JSDs among all relevant pairwise comparisons within a cell type. Peaks of JSD indicate observable changes in the probability distributions of methylation levels within genomic regions, even if changes in mean methylation levels are small. Genome-wide JSD analysis revealed only subtle changes when comparing preflight with inflight or postflight samples within a subject (Fig. 3C). By contrast, we observed greater JSD changes in comparisons between subjects, particularly in CD4 cells (fig. S7A), consistent with our observation of higher MML and NME changes in that cell type (Fig. 3B). Similar to the results associated with methylation level and entropy, the magnitude of JSD changes was slightly greater for HR than for TW, again indicating that global methylation changes in TW were within the range observed for HR.

Although the observed genome-wide changes were minimal, we hypothesized that local changes at promoters of genes involved in environmental adaptation could be particularly affected in TW during flight. We therefore ranked genes according to their epigenetic discordance, as quantified by the JSD, and used these ranked lists for gene ontology (GO) enrichment analysis (table S4). When comparing inflight with preflight samples from TW, we observed enrichments for genes involved in the response to platelet-derived growth factor (PDGF) and T cell differentiation and activation pathways in both CD4 and CD8 cells (Fig. 3D). Genes involved in platelet aggregation (q = 0.028, HGT), regulation of ossification (q = 0.009, HGT), and cellular response to ultraviolet-B (UV-B) (q = 0.039, HGT) were also enriched in these comparisons, specifically in CD4 cells. By contrast, analysis of CD8 cells revealed enrichment in genes involved in the somatostatin signaling pathway (q = 0.016, HGT) and positive regulation of superoxide anion generation (q = 0.016, HGT). Importantly, for all time point comparisons, the most enriched GO terms for TW preflight versus inflight were not enriched in HR (Fig. 1D).

Immune response and inflammation during spaceflight

Differential gene expression analysis indicated that many immune-related pathways were significantly changed inflight across all cell types (GLM, q < 0.001), including the adaptive immune system, innate immune response, and natural killer cell–mediated immunity (table S2). Additionally, DNA methylation analysis revealed epigenetic discordance (indicated by JSD peaks) for TW in the promoters of NOTCH3, a well-known regulator of T cell differentiation (34), and SLC1A5/ASCT2, which promotes T cell activation (35), between preflight (L−162) and inflight samples (FD259 for NOTCH3 and FD76 for SLC1A5) in CD4 and CD8 cells, respectively (Fig. 3E). This epigenetic discordance was associated with a peak of increased promoter methylation (from 10.7 to 20.2%) and entropy (from 41.4 to 71.3%) during flight and was observed in TW, but not in HR nor between HR and TW in preflight samples (fig. S7, B and C), and for SLC1A5, was associated with differential expression (down-regulation) (table S2).

To further evaluate T cell and immune functionality in spaceflight, we assessed cytokine data for inflammation signatures. Although none of the cytokines assayed were significantly different at baseline between TW and HR, 50 of the 62 cytokines assayed were differentially abundant between pre-, in-, and postflight in TW. We observed three notable signatures of inflammation. The first signature was composed of cytokines whose levels rose in TW’s plasma in the days after return and continued to be increased 6 months after return. These return-associated cytokines included CSF2 (GM-CSF), HGF, IL10, IL17A, IL18, LEP (leptin), CD40LG (CD40L), LTA (TNFB), VEGFD, NGF, FGF2 (FGFB), IFNA2 (IFNA), IFNB1 (IFNB), IFNG, SERPINE1 (PAI1), RETN, ICAM1, VCAM1, TNFSF10 (TRAIL), and VEGFA (VEGF), among others (Fig. 4A and table S6). The second signature was composed of cytokines that were at relatively high levels in TW preflight, continued to be high inflight, and decreased after return. These cytokines included BDNF, EGF, CXCL5 (ENA78), CXCL1 (GROα), IL2, IL6, LIF, PGDF-BB, TGFA, and TNF (TNFA), with several of them involved in mediating inflammation, cell growth, and cell proliferation, as well as tumor proliferation and vascularization (Fig. 1E and table S6). Of note, IL2 stimulation in CD8 cells activates Hippo signaling (36), and we observed enrichment in this pathway (q = 0.033, HGT) when comparing average promoter JSD (a measure of methylation discordance) from TW’s preflight with postflight samples in both CD4 and CD8 cells (table S4). The third signature of inflammation was seen immediately after landing (R+0 and R+4), with up to eightfold changes for CCL2 (MCP-1), C-reactive protein (CRP), and IL1RN (IL1RA; fig. S8, A to C; ANOVA, q < 0.001; table S8). Notably, CCL2 levels were increased at R+0, whereas CRP levels rose at R+0 and continued to rise at R+4 days. IL1RA levels only peaked at our R+4 data collection point. All three measures returned to baseline levels by the R+36 collection. During both inflight and postflight periods in TW, we detected increased levels of lysophospholipids containing proinflammatory omega-6 20:4 fatty acid (arachidonic acid) (37), together with a decrease in lysophospholipids containing anti-inflammatory omega-3 20:5 fatty acid (eicosapentaenoic acid) (38), suggestive of an increased inflammatory status (Fig. 4B and table S3).

Finally, we performed a vaccination response experiment to compare the effect of an influenza immunization in the spaceflight environment with that on Earth on the T cell receptor (TCR) repertoires in both subjects with the 2015–2016 trivalent unadjuvanted flu vaccination. Samples were obtained from TW and HR before and ~10 days after a trivalent flu immunization at three time points over the duration of the study. TW was sampled both on Earth (years 1 and 3) and on the ISS (year 2). For samples collected on Earth, repertoires for CD4 and CD8 T cells were separated, whereas frozen PBMCs from the ISS were expected to be a mixture of CD4 and CD8 repertoires. Vaccination-responsive clones were identified by matching complementarity-determining regions (CDR3s) pre- and postvaccination for each subject and cell type, with z scores as measures of change in a one-tailed approach. There were no significant differences (FDR > 0.05) in the percentage of unique CD4 or CD8 T cell receptor sequences (vaccination-responsive clones) that were noted as recruited by vaccination at 6 months inflight from the total PBMC fraction when compared with preflight and postflight responses in TW (Fig. 4C and fig. S9).

Taxonomic and functional microbial shifts during spaceflight

Changes over time in the gastrointestinal (GI) microbiota were examined with shotgun metagenome sequencing of genomic DNA extracted from serial fecal swabs of both subjects over the course of the study (tables S1 and S10). At the taxonomic level of genus, 1364 to 1641 taxa were detected in microbiome samples of both subjects. Although the richness of the fecal microbiome of HR was consistently higher than that of TW (P < 0.05, paired Wilcoxon signed-rank test, genus level) for the duration of the study, there were no significant differences between subjects in the Shannon index (SI; an index of diversity) (P = 0.82, paired Wilcoxon signed-rank test, genus level). Within TW, there was no decrease in richness or SI in inflight samples relative to preflight and postflight samples [median genera detected 1405.5 versus 1405; median SI (log₂), 2.66 versus 2.42; Fig. 5A].

Fecal microbial communities from both subjects were dominated by bacteria from the phyla Firmicutes and Bacteroidetes, together representing >96% of all annotated sequences recovered. Actinobacterial and Proteobacterial sequences represented 1.97 and 1.27% of annotated sequences, respectively. The ratio of sequences derived from Firmicutes to those derived from Bacteroidetes (F/B ratio) ranged from 0.72 to 5.55. The F/B ratio in TW was higher in inflight samples relative to combined pre- and postflight samples (median F/B ratio 3.21 versus 1.45) but returned to preflight levels postflight (a maximum F/B ratio of 4.60 was observed in TW inflight; Fig. 5A). Yet, all TW F/B ratios were in the range of 0.01 to 5.96 observed in a healthy human cohort (39). Although HR had an increased F/B ratio at the second preflight sampling [ground day (GD)-56; GD- days before TW launch; F/B ratio of 5.55], this increased ratio was not sustained over time and was low throughout the remainder of the analysis period (range of F/B ratios was 0.72 to 1.76; Fig. 5A). Fecal microbiomes at all sampling time points from both subjects had high relative abundance of bacteria from the genus Bacteroides (ranging from 15.4 to 61.0% of reads annotated to the taxonomic level of genus).

When examining diversity among samples (beta diversity), fecal microbial communities of the subjects were significantly different [analysis of similarity (ANOSIM) sample statistic R = 0.587, P = 0.001] and remained distinct over time (Fig. 5B and fig. S10A). The microbial community structure observed in the four inflight TW samples was significantly different from the pre- and postflight samples combined (ANOSIM, R = 0.438, P = 0.016; Fig. 5B). No such effect was observed in HR samples (ANOSIM, R = −0.05, P = 0.579). Similar flight-associated trends were identified in targeted analyses of fungal and viral community structures (fig. S10B) and in analyses of functional gene content across all taxa (Fig. 5C). At the highest resolution of functional gene annotation (SEED function), the subjects had significantly different functional gene content (ANOSIM, R = 0.456, P = 0.003). For TW, the functional gene content observed in the four inflight samples was significantly different from the preflight and postflight samples (ANOSIM, R = 0.55, P = 0.016; Fig. 5C). No such change over time was observed for HR (ANOSIM, R = −0.094, P = 0.762).

To evaluate the extent of microbial community shifts in flight, we identified specific taxonomic groups and gene function features that were significantly altered during the inflight period in both subjects. Specifically, 0.5 to 4.5% of the detected microbial taxa—including 36 species, 13 genera, 8 families, 4 orders, 3 classes, and 3 phyla—were differentially abundant (FDR < 0.05), when inflight samples were compared with the combined pre- and postflight periods (Fig. 5D). However, we found a smaller number of altered microbial groups at each taxonomic level in HR for the same comparison (Fig. 5D). Similar observations were made when analyzing functional gene content. The spaceflight environment significantly (FDR < 0.05, EdgeR) altered the abundance of a moderate fraction (2 to 23%) of gene functional categories at each level in TW inflight compared with the combined preflight and postflight periods. The number of altered gene functional categories was smaller for HR during the inflight-equivalent period (Fig. 5E). Overall, TW’s inflight samples had high similarity in taxonomic composition and functional gene content, as assessed with Bray-Curtis similarity (Fig. 5F). The level of similarity within TW’s inflight samples was slightly higher than the inflight-equivalent samples from HR for both taxonomic and functional gene annotations (P = 0.03 and 0.31, paired Wilcoxon signed-rank test, respectively). Although TW had more similar preflight and postflight microbial communities than HR, these differences were not significant (P = 0.16 and 0.13, paired Wilcoxon signed-rank test, for taxonomy and functional gene annotations, respectively; Fig. 5F).

Finally, we examined metabolomic data for corroboration with the microbiome signatures. We observed several small-molecule markers of microbial metabolism in the untargeted LC-MS metabolomics data, including phenyls, secondary bile acid metabolites, and indole containing compounds, some of which are produced solely by microbial metabolism. Some metabolites, such as 3-indole propionic acid, have anti-inflammatory effects and were observed at lower levels in TW throughout the duration of the study (Fig. 5G and table S3).

Mitochondria-related adaptations during spaceflight

Our integrated analyses identified a number of mitochondria-related changes at the genomic and functional levels related to the 1-year mission. Genic annotation of RNA sequencing (RNA-seq) data (40, 41) revealed higher levels of mtRNA inflight as compared with preflight and postflight (Fig. 6A and fig. S11A). The time spent on the ISS was correlated [coefficient of determination (R²) = 0.84] with increased mtRNA inflight (P = 3.6 × 10⁻⁶, ANOVA). Mitochondrial genome coverage was consistent across all samples (fig. S11B) and validated by qPCR in DNA from blood plasma and qRT-PCR in RNA from PBMCs, supporting this finding further (fig. S11, C to E). We also conducted an extracellular flux assay that monitors real-time changes in the mitochondrial oxygen consumption rate (OCR, a measure of mitochondrial oxidative phosphorylation) by treating muscle cells (L6 cell line) with plasma from the astronauts. The results can be characterized as twofold, whereby the plasma can (i) increase basal respiration (Fig. 6B) and (ii) decrease the spare reserve adenosine triphosphate (ATP) capacity (Fig. 6C). These trends show an increase in ATP-linked respiration, decrease in nonmitochondrial respiration, and no change in maximum respiration (Fig. 6, D and E), which may also reflect the changes in mtRNA levels observed during flight.

Additionally, a significantly increased level of lactic acid (P = 0.018; ANOVA) was detected in TW’s urine inflight, which returned to baseline postflight. The highest lactic acid levels were observed in TW during the first 15 days and final 14 days of spaceflight (Fig. 6F). GC-MS/MS urine metabolomics data showed an increase in the lactic acid/pyruvic acid ratio (Fig. 6G), indicating a shift from aerobic to anaerobic metabolism. A similar trend in lactic acid levels, as observed in the urine, was also observed in plasma, although only the lactic acid/pyruvic acid ratio was significantly altered in plasma at an FDR of 0.05 (fig. S12, A to C). Lactic acid production promotes higher extracellular acidification rates (42) that, in turn, are associated with increased expression of glycolysis-related genes. We also observed a trend for increased levels of plasma TCA cycle intermediates in the inflight period, such as citric acid and malic acid (FDR < 0.10), compared with preflight and postflight levels (fig. S12, D and E).

Given the previously noted changes in mitochondrial respiration, we examined TW’s data for other related signatures. TW exhibited altered gene expression patterns in pathways related to oxygen and reactive species metabolic processes, mitochondrial transport, hypoxia, and apoptotic mitochondrial changes (GSEA, q < 0.0001) (table S2). Although differential expression of these genetic pathways is observed in response to hypoxic stimuli in a rodent model (43), the partial pressure of oxygen levels on ISS were near sea-level values (fig. S13).

Cardiovascular changes during spaceflight

Spaceflight-specific changes in cardiovascular structure and function were observed similar to those previously reported (44, 45) in astronauts during long-duration spaceflight, including a cardiac output increase of 10% (average of spaceflight measurements) (Fig. 7A) and a moderate decrease in systolic and mean arterial pressure (table S7) relative to the upright posture on Earth. Moreover, carotid artery distension developed early in flight during both systole and diastole and persisted throughout the mission (Fig. 7A). Concomitantly, carotid intima-media thickness increased in TW from preflight to inflight and remained thicker 4 days after landing (Fig. 7B). In addition, indices of inflammation, including cytokines [e.g., IL1A, IL1B, IL2, and prostaglandin F2α (PGF2α)] and chemokines, were increased during spaceflight in TW, whereas they remained unchanged in HR across all time points measured (Fig. 7, C and D; fig. S8; and table S6). Biomarkers of oxidative stress were not consistently increased [e.g., oxidized low-density lipoprotein (LDL), myeloperoxidase, and 8-hydroxy-2′-deoxyguanosine (8-OHdG)] in TW during spaceflight (table S8). However, targeted proteomics identified increased levels of two proteins that may be associated with observed changes in vascular wall dimensions. Collagen alpha-1(III) chain (COL3A1) and collagen alpha-1(I) chain (COL1A1) proteins in TW’s inflight urine were increased compared with preflight levels, and these values returned to baseline levels postflight (Fig. 7, E and F). We also observed increased COL1A1, from measurements of two separate COL1A1 peptides, by untargeted proteomic analyses (Fig. 7F). This finding is consistent with enrichment of expression levels of genes in the collagen-related pathways [including but not limited to GO term “collagen trimer” and Reactome pathway “collagen formation” in PBMCs (DESeq2, q < 0.001)] (table S2). We also observed an increase in the ratio of plasma levels of apoliprotein B (APOB; a major constituent of LDL particles) to apolipoprotein A1 [APOA1; a major constituent of high-density lipoprotein (HDL) particles] during the last 6 months inflight (late inflight; after FD180) in TW, compared with preflight and early inflight (ANOVA, P = 0.003, Fig. 7G). These levels decreased postflight and, then, were not significantly different from preflight levels. No significant increase was seen in HR, indicating an association of the APOB/APOA1 ratio with the long duration of the mission.

Body mass, nutrition, and hydration during spaceflight

Baseline and inflight values for HR and TW can be found in table S8. Similar to observations in other astronauts (46), TW lost 7% of his body mass during flight, whereas HR gained ~4% of his body mass over the course of the study (Fig. 8A). TW had decreased urine volume (table S8) and increased urinary aquaporin-2 (AQP2) inflight compared with preflight, which both returned to baseline postflight (Fig. 8B). Serum sodium concentrations (Fig. 8B) and osmolality (Fig. 8C) followed urine AQP2 levels (Fig. 8D) for the duration of the study. TW’s energy intake was about 66% of predicted requirements, and beyond the nominal vitamin D supplements provided to all astronauts, he did not report taking vitamin or mineral supplements during the mission. In TW, proteomic analysis revealed a decrease in urine angiotensinogen inflight relative to pre- and postflight that is correlated with the decline of body mass inflight (Fig. 8E).

Bone turnover markers for both breakdown (N-telopeptide; Fig. 8F) and formation (bone-specific alkaline phosphatase; Fig. 8G) increased 50 to 60% in TW during the first 6 months, as previously reported during 4- to 6-month spaceflights (13, 47, 48). However, these markers decreased in the last 6 months of the mission until immediately before landing. Similar trends were observed for other resorption (e.g., serum CTX-β) and formation markers (e.g., P1NP and osteocalcin; table S8), supporting these data. Bone formation and volume of resistive exercise before the blood draw followed similar trends (Fig. 8H), that is, when exercise volume was lower in the second half of the mission (due to injury), bone formation was correspondingly lower, as expected.

Neuro-ocular changes during spaceflight

Spaceflight resulted in a cephalad fluid shift and ocular changes in TW that were not present in HR. Internal jugular vein cross-sectional area, forehead tissue thickness, cardiac filling, stroke volume, and cardiac output increased, despite an average decrease in mean arterial pressure when compared with values associated with the seated position on Earth (Fig. 9, A and B). Increases in subfoveal choroidal thickness (primary vascular supply of the outer retina) and peripapillary total retinal thickness were observed, indicating retinal edema formation (Fig. 9, C and D). Moreover, although TW exhibited choroidal folds during preflight testing before this mission, the severity of the choroidal folds increased during spaceflight, as quantified by number and depth. These ocular structure parameters were unchanged in HR for the duration of the study, and choroidal folds were not present at the initiation of these data collection sessions (Fig. 9, A to D).

Of relevance to these ocular changes, untargeted proteomics revealed a decrease in urine leucine-rich alpha-2-glycoprotein (LRG1) levels in TW inflight relative to those pre- and postflight (Fig. 9E). LRG1 has been reported to play a role in retinal vascular pathology (49). The predisposition to develop these ocular changes during spaceflight has also been associated with B-vitamin status and the presence of specific single-nucleotide polymorphisms (SNPs) (50). Risk alleles in five SNPs predict incidence of ophthalmic issues (49), and six of the nine risk alleles are present in the twins (50). Furthermore, serum folate, a B vitamin that is often low in astronauts who experienced ophthalmic changes during flight (51), was also low in both TW and HR (Fig. 9F).

Cognitive performance

Continuous high levels of astronaut cognitive performance are critical for mission success. Study subjects performed a computerized cognitive test battery (Cognition) (52) throughout all phases of the mission (Fig. 10A). TW’s accuracy and speed scores were measured for the different phases of the mission [preflight, inflight months 1 to 6 (early flight), inflight months 7 to 12 (late flight), and postflight] (Fig. 10B). TW took more risk than HR on the balloon analog risk test (BART) during all phases of the mission, more so inflight compared with preflight and postflight. Compared with HR, TW did not exhibit significant deficits in either speed (−0.7 SD, P = 0.3518; Student’s t test) or accuracy (−0.2 SD, P = 0.6678; Student’s t test) across cognitive domains preflight. TW’s cognitive speed increased across all cognitive domains early inflight relative to preflight. During the same period, spatial orientation and motor praxis accuracy increased (>1 SD), whereas accuracy of visual object learning and abstract matching decreased (>1 SD) relative to HR. There was no significant decrease in performance from early to late inflight, with the following exceptions: (i) decline (>1 SD) in emotion recognition task (ERT) and digit symbol substitution task (DSST) speed and (ii) decline (>2 SD) in abstract matching (AM) accuracy (P = 0.0280; Student’s t test; TW versus HR late flight, Fig. 10C). Performance trends across the study on key accuracy and speed metrics for all 10 tests in the Cognition battery are shown in fig. S14.

Most notably, cognitive speed decreased for all tests except for the DSST, and accuracy decreased for all domains except for spatial orientation postflight. TW’s cognitive efficiency (a combination of speed and accuracy across cognitive domains) was similar pre- and inflight relative to HR but was significantly lower postflight (P = 0.0016, Student’s t test). This postflight decline in cognitive performance persisted up to 6 months postflight in both speed and accuracy domains (Fig. 10C).