**Background:** Stress that disrupts normal lifestyle can alter the human microbiota, and extreme stress may impact physiology and endurance. The human stress response involves the endocrine system producing corticosteroids and can increase gut permeability, potentially allowing the gut microbiota to influence the hypothalamic-pituitary-adrenal axis. Previous studies, such as the Mars-500 isolation study, showed changes in gut microbiota taxonomic composition within the first 14 days but reversion to initial composition after the study ended. The Trans-Antarctic Winter Traverse (TAWT) expedition aimed to complete a 2000-mile crossing of Antarctica in winter (March to September 2013), with ambient temperatures below -50°C and three months of complete darkness. The expedition was halted in June 2013 due to crevasse fields after traversing 300 km, and the five-man team remained stationary for three months before a two-month return traverse. This study presents the first multi-site investigation of the human microbiota (saliva and stool) and metabolome (saliva, stool, and plasma) response to long-term environmental and physiological stress analogous to space travel.
**Methods:** Five male participants (mean age 37, range 28-54) provided stool and saliva samples at monthly intervals over eight months, with baseline samples collected pre-expedition. Plasma samples were collected during the expedition only. Samples were stored at ambient temperatures (≈-25 to -40°C) during the expedition and transferred to -80°C storage upon return to the UK. Metataxonomic sequencing of the V3-V4 region of the 16S rRNA gene was performed using Illumina MiSeq platform. Bacterial load was measured by 16S rRNA gene quantitative PCR. Metabolite fingerprinting was conducted using flow-infusion electrospray ionisation mass spectrometry (FIE-MS) and Fourier transform infrared spectroscopy (FTIR). Statistical analyses included PERMANOVA for beta-diversity, Shannon diversity for alpha-diversity, and FDR-corrected p-value thresholds of <0.05 for univariate features.
**Key Results:** For saliva, significant activity-related differences were observed in beta-diversity between baseline and expedition samples (R²=0.31, p<0.001), with participant differences also significant (R²=0.40, p<0.001). Salivary bacterial load (16S rRNA gene copy number) showed significant increases during the expedition compared to baseline (p<0.001). Shannon diversity was significantly higher in expedition samples compared to baseline. For stool microbiota, no significant activity-related differences were found (R²=0.09, p=0.186), but significant individual differences were maintained throughout (R²=0.56, p<0.001). At the OTU level, only Streptococcus and Prevotella melaninogenica in saliva and Ruminococcaceae family in stool were significantly altered by expedition activity. In total, 32 OTUs in saliva and 33 OTUs in stool showed significant differences between participants. FTIR fingerprinting showed no significant differences between activity periods across all three biofluids at the multivariate level, though univariate analysis revealed activity-related differences in saliva within the alcohol/primary amide region (3500-3400 cm⁻¹), primary amide region (≈1690 cm⁻¹), and N-O stretching region (1550-1500 cm⁻¹). Metabolite fingerprinting (FIE-MS) showed that individual differences substantially outweighed activity-associated differences across all biofluids. In negative ion detection mode, 282 features were significantly different between participants (3 in plasma, 279 in stool), while only 8 features in stool significantly differed between activity periods. Tentative identifications included homovanillic acid sulphate and a potential dietary biotransformation product. Multi-omic correlation analysis revealed no correlations between same-taxon features in matched saliva and stool samples, but correlations between different taxonomic groups were observed. Stool samples showed the greatest number of correlating features between metataxonomic and metabolite fingerprinting data.
**Clinical Implications:** The study demonstrates that extreme environmental and physiological stress differentially affects microbial communities across body sites, with saliva showing greater susceptibility to perturbation than stool. The increased bacterial load and diversity in saliva may indicate impaired immune status, potentially increasing infection risk during extreme stress or isolation. The resilience of the stool microbiota despite significant dietary changes (freeze-dried meals) suggests that individual differences in gut microbiota are maintained under extreme conditions. The minimal impact on metabolite fingerprints across all biofluids suggests that participants possessed a high degree of physiological endurance, indicating that metabolomic profiling could serve as a predictor of endurance potential for long-term expeditions. The study highlights the need for multi-omic approaches beyond microbiota profiling alone when monitoring health during long-duration space travel or terrestrial analogues. Limitations include the small sample size (n=5), lack of a non-stressed control group, absence of baseline plasma samples, and inability to measure immune markers due to logistical constraints.