Heavy water, or deuterium oxide, has long fascinated microbiologists. Because water participates in virtually every biochemical reaction, replacing ordinary hydrogen with its heavier isotope, deuterium, offers researchers an extraordinarily powerful window into cellular activity. It is this property that makes D2O a cornerstone of stable isotope probing, a technique that connects microbial identity with metabolic function in complex ecosystems. Yet deuterium is not an innocent label. Substituting deuterium for hydrogen strengthens chemical bonds and lowers zero-point vibrational energies, producing primary and secondary kinetic isotope effects that slow hydrogen-dependent processes such as proton-coupled electron transfer and hydride transfer. The consequences ripple through reaction kinetics and equilibria, disturbing pathway fluxes and impairing growth. Now, a team at the University of Liverpool, working with collaborators in Toulouse, has delivered the most detailed picture yet of how these effects play out inside living bacteria, revealing why some Escherichia coli strains shrug off isotopic stress while others falter.

The new study, published in the journal Metabolomics, set out to characterise the global metabolic responses of deuterium-susceptible and deuterium-tolerant E. coli isolates and to pinpoint the pathways and metabolites associated with isotopic stress tolerance. Although earlier work had quantified kinetic isotope effects for individual enzymes in vitro, and a landmark proteomic study had shown down-regulation of translational machinery and up-regulation of redox-active enzymes in fully deuterated medium, the small-molecule layer where isotope effects are first expressed had remained largely unexplored. Untargeted metabolomics is ideally suited to filling that gap, because it captures labile cofactors and pathway intermediates that often change before transcriptional or translational regulation comes into play. The researchers also drew on earlier evidence that D2O induces DNA damage responses, including recA activation, and that central carbon fluxes are rerouted towards anaplerosis in 70 percent heavy water, implicating tricarboxylic acid cycle inhibition.

To build their experimental framework, the team grew five uropathogenic E. coli isolates alongside the reference strain K-12 MG1655 in minimal medium supplemented with glucose and casamino acids, adjusted to contain 0, 20, 40, 60 or 80 percent D2O. Growth was monitored over 24 hours using an automated Bioscreen C system with continuous shaking at 37 degrees Celsius. The results split cleanly into two phenotypes. Isolates 99, 133 and 134 proved susceptible: their increases in optical density at 600 nanometres became progressively delayed and more gradual as deuterium concentrations rose, with effects already visible at 20 percent in some isolates. In contrast, isolates 25 and 148 and the MG1655 reference strain showed comparatively modest changes, growing earlier and maintaining higher final densities even at 80 percent heavy water. Importantly, the authors interpret this as gradual, strain-dependent inhibition rather than a single critical tipping point, consistent with earlier reports of concentration-dependent effects dating back to the 1960s.

For metabolomic analysis, the team selected three susceptible and three tolerant strains and cultured them at 0, 40 and 80 percent D2O. Because heavy water altered growth profiles differently among strains, samples were harvested at strain- and condition-specific time points corresponding approximately to the midpoint of the observed optical density increase, rather than at a fixed chronological time, ensuring that all cultures were captured at broadly comparable positions within their exponential growth. Metabolism was arrested by quenching cultures into pre-chilled 60 percent methanol at minus 48 degrees Celsius, a step validated to prevent metabolite leakage. Cells were lysed through freeze-thaw cycles in liquid nitrogen, extracts were normalised to a common biomass equivalent, and pooled quality-control samples were prepared to monitor analytical drift. Five biological replicates were collected for every strain and condition, providing the statistical power needed for downstream testing.

Analytical chemistry was performed on a Thermo Fisher Vanquish ultra-high-performance liquid chromatography system coupled to a Q Exactive Orbitrap mass spectrometer, using a Hypersil GOLD aQ C18 column and a 15-minute gradient. Data were acquired independently in positive and negative electrospray ionisation modes, with full scans at a resolving power of 70,000 and data-dependent MS/MS fragmentation across four precursor ranges to support compound annotation. Raw data were processed in Compound Discoverer 3.3 using a stable isotope labelling workflow optimised for deuterium-based probing, with retention time alignment, five-part-per-million mass tolerance, blank filtering requiring sample signals to exceed blanks fivefold, and deuterium incorporation detection set to a maximum exchange of 25. Annotations were cross-referenced against mzCloud, ChemSpider, the Human Metabolome Database and an in-house spectral library, with a minimum spectral match score of 70 percent, yielding Level 2 assignments under the metabolomics standards initiative.

Stringent quality filtering followed. Features detected in fewer than 75 percent of quality-control occurrences were excluded, as were those with relative standard deviations above 30 percent or blank signals exceeding 10 percent of QC averages. After gap filling with a k-nearest-neighbours algorithm and correction of signal drift using quality-control-based robust LOESS, the processed data were analysed in MATLAB. Principal component analysis confirmed tight clustering of pooled QC samples, testifying to high reproducibility. Principal component-discriminant function analysis, using the first ten principal components and 18 classes defined by deuterium level, phenotype and strain, revealed a clear gradient along the first discriminant function corresponding to D2O exposure, with control samples at one end and 80 percent cultures displaced furthest at the other. Crucially, the authors note that this separation reflects metabolic changes induced by heavy water rather than isotopic enrichment, since feature areas sum monoisotopic and labelled ions. Susceptible and tolerant isolates also formed distinct clusters along the second discriminant function, most visibly at 40 percent D2O.

Two-way analysis of variance with Benjamini-Hochberg false discovery rate correction identified 1,418 significantly altered features in positive mode, including 38 annotated metabolites, and 1,086 in negative mode, including 17, with 11 metabolites detected in both modes. A strain-level sensitivity analysis using linear mixed-effects models confirmed that phenotype-level pooling was justified for the vast majority of annotated metabolites, with only about 6 percent of tests showing significant within-phenotype strain effects. In total, 44 significantly altered metabolites mapped across nine pathway categories, including glutathione metabolism, arginine and proline metabolism, arginine biosynthesis, nicotinate and nicotinamide metabolism, purine and pyrimidine metabolism, glycolysis, pyruvate metabolism, the TCA cycle, and branched-chain amino acid degradation. A complementary network analysis using the KEGG-derived E. coli metabolic network through MetExplore placed 35 mapped metabolites across 92 of 105 network pathways, with 30 pathways significantly enriched by hypergeometric over-representation testing.

The pathway-level story is one of coordinated stress remodelling. Susceptible isolates showed marked decreases in the dipeptides valylphenylalanine and alanyltyrosine at 40 percent D2O, alongside a striking rise in free valine and elevated succinate, whereas tolerant isolates accumulated those dipeptides. Prolylleucine fell sharply only in susceptible strains. Isoleucine increased in susceptible isolates, consistent with impaired downstream catabolism under reduced cofactor availability. In NAD-related metabolism, tolerant isolates maintained higher baseline abundances of nicotinate, nicotinamide, NAD+ and ADP-ribose, all of which declined with increasing deuterium in both groups, hinting that kinetic isotope effects perturb proton-coupled reactions in NAD+ biosynthesis. Polyamine metabolism provided perhaps the most dramatic signal: susceptible isolates showed a roughly one-hundred-fold surge in cadaverine at 40 percent D2O and strong putrescine upregulation, while tolerant isolates changed far more moderately, patterns consistent with impaired polyamine export under heavy water.

Nucleotide metabolism told a parallel tale of energetic strain. Purine intermediates from IMP through XMP, GMP and AMP declined progressively in both phenotypes, with susceptible isolates declining earlier and more steeply, suggesting mobilisation of purine pools to sustain ATP and GTP synthesis and TCA cycle activity. Adenosine behaved distinctly, rising in susceptible isolates at 40 percent D2O while falling in tolerant ones, hinting at differential handling of AMP breakdown. On the pyrimidine side, orotate, UMP, uracil and cytidine monophosphate shifts pointed to bottlenecks in de novo synthesis, reduced RNA turnover and a late shift toward energetically favourable salvage routes under severe stress. Glutathione-associated metabolites and 5-oxoproline changes further supported differential antioxidant responses, although the authors caution that steady-state abundances cannot directly establish flux, redox state or enzyme activity. Taken together, these findings supply a biochemical foundation for understanding isotopic stress tolerance, with direct implications for optimising stable isotope probing, refining deuterium-based analytical tools, and engineering more robust microbial chassis for biotechnology and synthetic biology.

Subject of Research: Metabolic responses of Escherichia coli to deuterium oxide-induced isotopic stress

Article Title: Metabolic profiling of Escherichia coli under deuterium oxide-induced stress

Article References: Shams, S., Roberts, I., Rehuman, A. N. S., Gotts, N., Xu, Y., Ahmed, S., Jourdan, F., Winder, C., Burke, A., Goodacre, R., & Muhamadali, H. (2026). Metabolic profiling of Escherichia coli under deuterium oxide-induced stress. Metabolomics, 22(5), Article 156. https://doi.org/10.1007/s11306-026-02532-3

Image Credits: AI Generated

DOI: 10.1007/s11306-026-02532-3

Keywords: E. coli, deuterium oxide, metabolomics, stable isotope probing, kinetic isotope effects, LC-MS, central carbon metabolism, polyamine metabolism, nucleotide metabolism, glutathione, NAD metabolism, isotopic stress tolerance

Cite Scienmag News
APA MLA Chicago

Copy citation Download RIS

Tags: biochemical pathways disrupted by heavy watercentral carbon metabolismdeuterium isotope effects on biochemical reactionsdeuterium oxideE. coliE. coli strain tolerance to heavy waterglutathioneHeavy water effects on bacterial metabolismimpact of deuterium on microbial growth and pathway fluxinfluence of deuterium on proton transfer reactionsisotope effects on enzyme kinetics and cellular functionisotopic stress in E. coliisotopic stress tolerancekinetic isotope effectsLC-MSmetabolic adaptations to isotopic stressmetabolic profiling of bacteria under isotMetabolomicsNAD+ metabolismnucleotide metabolismpolyamine metabolismstable isotope probingstable isotope probing in microbiologyuse of heavy water in microbial ecology studies