The experimental design hinged on a carefully controlled hypoxia challenge. After an eight-week acclimation period, adult males of each species were assigned to either normoxic conditions (20.9% oxygen) or simulated high-altitude hypoxia (10% oxygen for 48 hours) delivered through a normobaric oxygen chamber at sea-level pressure. Following treatment, skeletal muscle from the left hindlimb was rapidly dissected, flash-frozen in liquid nitrogen and stored at −80 °C. Metabolites were then extracted by homogenizing tissue in prechilled 80% methanol with 0.1% formic acid, vortexing, and centrifuging at 15,000 g and 4 °C. The resulting supernatants were diluted and injected into a high-performance liquid chromatography system coupled to a QTRAP 6500+ mass spectrometer, operated in both positive and negative ionization modes using multiple reaction monitoring. Rigorous quality control—including pooled QC samples, total ion chromatogram overlap analysis, Pearson correlation and principal component analysis—confirmed the stability of the data, with the median coefficient of variation across all identified metabolites in QC samples just 8.26%.

In total, the team identified 721 metabolites spanning 38 chemical classes in the three species’ skeletal muscle. Amino acids and their derivatives dominated the catalog at 20.67% of detected compounds, followed by organic acids (15.12%), nucleotides (10.68%), fatty acyls (7.21%) and carbohydrates (5.69%). Using partial least squares discriminant analysis, the researchers found clear separation between normoxic and hypoxic samples in each species, with models explaining between roughly 47% and 57% of the variance. When they screened for differentially accumulated metabolites (DAMs), defined by a variable importance in projection score of at least 1.0 and an absolute log2 fold change of at least 1.0, striking species-specific patterns emerged. The Qinghai vole showed 9 significantly upregulated and 57 downregulated DAMs under hypoxia; Brandt’s vole had 32 upregulated and 54 downregulated; and the Kunming mouse showed a reversal of this trend, with 71 upregulated and only 16 downregulated.

For the high-altitude specialist N. fuscus, the metabolic signature pointed away from glucose. Several glycolysis intermediates and glucose-related metabolites—3-phosphoglyceric acid, 2-phosphoglyceric acid, gluconolactone, D-gluconic acid and L-gulono-1,4-lactone—were downregulated under hypoxia, suggesting reduced glucose utilization. Cysteine metabolism pathway metabolites such as L-cysteine, cystine and S-sulfo-L-cysteine, which can feed pyruvate synthesis, were also suppressed. In contrast, 3-methyl-2-oxobutanoic acid was upregulated, potentially channeling carbon into succinyl-CoA and the tricarboxylic acid cycle. Paired with gene set enrichment analysis showing significant upregulation of oxidative phosphorylation (normalized enrichment score 1.89) and the pentose phosphate pathway, and earlier transcriptomic evidence that genes like Acsl6, Gpat4, Acot4 and Ndufb7 promote fatty acid beta-oxidation and electron transport, the picture is coherent: the plateau vole leans on fatty acids, not glucose, to keep its muscle powered when oxygen is thin. Notably, bile acids such as taurohydroxylic acid and taurochenodeoxylic acid showed strong intercorrelations in this species, hinting at signaling roles through the TGR5 receptor in regulating muscle oxygen and energy consumption.

Brandt’s vole, a species from habitats below 2,000 meters, told a different story. Under hypoxia, its muscle accumulated glycolytic intermediates—D-glucose 6-phosphate, D-fructose 6-phosphate, dihydroxyacetone phosphate, 2-phosphoglyceric acid and phosphoenolpyruvic acid—while the gene Pygm, encoding muscle glycogen phosphorylase, the rate-limiting enzyme that initiates glycogen breakdown, was upregulated. Genes supporting the electron transport chain, including Ndufb8 (a complex I subunit) and Coa8 (a cytochrome c oxidase assembly factor that protects complex IV from oxidative degradation), were also elevated, implying that oxidative phosphorylation in this species continues but depends on a glycolysis-fed supply of substrate. Meanwhile, the pentose phosphate pathway shifted into gear: the oxidative-stage intermediate 6-phospho-D-glucono-1,5-lactone and nonoxidative intermediates such as sedoheptulose 1,7-bisphosphate and D-erythrose 4-phosphate rose, alongside upregulation of Rpia and Rpe. This route generates NADPH, which keeps cellular glutathione reduced—a crucial antioxidant reserve—and the team found that upregulated Nmnat2 and Nmrk2 feed NADP+ production through the salvage pathway, supporting pentose phosphate flux.

The Kunming mouse, a low-altitude laboratory strain, fared worst across the board. Its muscle showed upregulation of Gaa (alpha-glucosidase) and Gck (hexokinase), pointing to enhanced glycolysis, but fatty acid beta-oxidation was clearly suppressed: Acsl6 was upregulated while Cpt1a, which encodes the rate-limiting mitochondrial fatty acid transporter and is a known target of hypoxia-inducible factor inhibition, was downregulated. More tellingly, several electron transport chain genes—Nd1, Foxred1, Oxa1l and Cox1—were downregulated, indicating that complex I assembly and cytochrome c oxidase function falter under hypoxia. The upregulation of NADH itself fits this picture: prior work has shown that NADH accumulates when respiration is impaired and that this buildup inhibits isocitrate dehydrogenase, further stalling the TCA cycle. With fatty acid oxidation, the TCA cycle and oxidative phosphorylation all damped, mouse muscle essentially falls back on anaerobic glycolysis, sacrificing efficiency for survival.

The transcriptomic layer of the analysis, reprocessed with DESeq2 from previously published RNA-sequencing data (available under BioProject PRJNA993829), revealed 1,523 differentially expressed genes in N. fuscus, 1,978 in L. brandtii and 1,683 in M. musculus under hypoxia versus normoxia. For the Qinghai vole, upregulated genes were significantly enriched in pathways involving hemoglobin and heme synthesis—Alas2, the rate-limiting enzyme of heme production; Uros, a central hub of heme synthesis; and the globin genes Hba and Hbb—likely boosting oxygen transport capacity and potentially triggered by hypoxia-inducible factor 1-alpha. Brandt’s vole also upregulated Alas2 and Hbb, while the mouse showed no differentially expressed oxygen-transport genes at all, underscoring the plateau specialist’s distinctive strategy. Downregulated genes in N. fuscus were enriched in complement and coagulation cascades and platelet activation, hinting at suppressed inflammatory and thrombotic activity.

Beyond fuel choice, the three species diverged in how they defended their tissues. The Qinghai vole showed a coordinated vasodilatory program: upregulation of 3-(3-hydroxyphenyl) propionic acid, which lowers arterial blood pressure and scavenges free radicals, and downregulation of NG, NG-dimethyl-L-arginine and L-arginino-succinate, both inhibitors of nitric oxide synthase, thereby freeing nitric oxide to dilate blood vessels and improve oxygen delivery. Upregulated Trpc6 raises intracellular calcium, further enhancing nitric oxide production, and 3-methyl-2-oxobutanoic acid was found to inhibit HIF-1α proline hydroxylase, promoting accumulation of the master hypoxia transcription factor. Brandt’s vole, by contrast, mounted an antioxidant defense: upregulation of Gstm1 (glutathione-S-transferase), L-ascorbate and spermine within glutathione metabolism, plus elevated UDP-N-acetylglucosamine and cholesteryl sulfate, all of which protect cells from reactive oxygen species—consistent with its evolutionary history of intermittent hypoxia exposure.

The mouse, once again, showed the most stressed profile. Downregulation of anti-inflammatory polyunsaturated fatty acids such as alpha-linolenic acid, 12(S)-HETE and 15-HETE—which normally support angiogenesis and vascular endothelial growth factor signaling—accompanied the observed suppression of arachidonic acid metabolism. Accumulation of indoxylsulfuric acid, a uremic myotoxin, threatens to induce reactive oxygen species, raise inflammatory cytokines like tumor necrosis factor-alpha, deplete glutathione and drive muscle atrophy gene expression, aligning with downregulation of the antioxidant gene Oxr1. Some compensatory responses were visible, however: upregulated 3-hydroxyisovaleric acid may limit protein degradation, succinic anhydride shows antihypoxic activity that reduces lactic acid buildup, and kynurenic acid from tryptophan metabolism scavenges free radicals.

Taken together, the study proposes that regulated fatty acid oxidation may be a hallmark metabolic strategy of plateau-dwelling mammals, echoing earlier findings in high-altitude deer mice (Peromyscus maniculatus), while lowland species default to glycolysis and mid-altitude species blend aerobic glucose oxidation with enhanced antioxidant machinery. The authors caution that their quasi-targeted metabolomics platform, while sensitive and quantitative across a broad panel of compounds drawn from a database of over 2,200 animal-source metabolites, does not capture the entire metabolome, and that the species-specific patterns warrant further experimental verification. Even so, by integrating 721 metabolites with thousands of differentially expressed genes across shared KEGG pathways, the work establishes a rigorous framework for future studies of transcriptional–metabolic coupling and could inform strategies for protecting human muscle function at altitude or in hypoxic disease states.