A Biological Model of the Ageing Metabolome Reveals Potential Clinically Relevant Biomarkers

Significance of the topic

Ageing alters cellular metabolism through mechanisms such as mitochondrial dysfunction, genomic instability and senescence, producing measurable shifts in the metabolome that can serve as biomarkers of biological age and disease risk. Robust untargeted metabolomics workflows that combine high-resolution chromatography with data-independent MS/MS acquisition and interoperable data-analysis pipelines accelerate discovery of such biomarkers from complex biological matrices including cell extracts and culture media.

Study objectives and overview

This application study used serially passaged human foreskin fibroblasts (HFF-1) as an in vitro model of cellular ageing (comparison of passage 3 versus passage 20) to:

• implement an untargeted HILIC-based LC–QTOF metabolomics method with DIA-MS/MS acquisition;
• establish an end-to-end data workflow that is interoperable with widely used open-source tools (MS-DIAL, MetaboAnalyst) as well as vendor library-search software (LabSolutions Insight Explore);
• identify metabolic features and putative metabolites associated with the senescent state in cells and spent media.

Methodology and instrumentational details

Samples and experimental design:

HFF-1 cell extracts and corresponding culture media were obtained at early (passage 3) and late (passage 20) culture passages. Six biological samples per group were processed; cell and media extracts were analysed separately and sample order was randomized within batches.

Chromatography and injection:

UHPLC system: Nexera X2.

Column: Shim-pack Velox HILIC (2.1 × 100 mm, 2.7 µm) at 40 °C.

Mobile phases: A = water + 10 mM ammonium formate + 0.1% formic acid; B = acetonitrile:water (92:8) + 10 mM ammonium formate + 0.1% formic acid. A HILIC gradient from 98% B to 10% B and back was used, with a variable flow profile (0.3 mL/min baseline, 0.4 mL/min for short segment). Injection volume was 0.5 µL, autosampler at 4 °C. Rinse solution included IPA/MeOH/MeCN/H2O with 1% formic acid.

Mass spectrometry and acquisition:

Instrument: Shimadzu LCMS-9030 Q-TOF.

Ionization: ESI, both positive and negative modes (interface temperatures and voltages adjusted per polarity). TOF-MS range m/z 60–1000. Data-independent acquisition (DIA) used 27 MS/MS windows covering m/z 40–1000 with ~35 Da precursor isolation widths and collision energy spread 5–55 V. Total MS + MS/MS cycle time was ~0.991 s, enabling dense MS/MS coverage across chromatographic peaks.

Data processing and analysis workflow:

Feature detection and alignment: MS-DIAL was used directly on LabSolutions raw files (no conversion) to detect chromatographic features and perform deconvolution of DIA MS/MS spectra. Positive and negative mode data were processed separately and later combined for statistics.

Feature filtering and statistics: Exported feature tables were filtered to retain features present in ≥80% of samples in at least one group (passage 3 or 20). Within MetaboAnalyst, missing values were imputed by replacing with 1/5 of the minimum positive value, and interquartile range (IQR) filtering (40%) removed near-constant variables. Differential features were identified by volcano-plot analysis using fold-change >2 and p < 0.05 (passage 20 vs passage 3).

Identification and annotation: LabSolutions Insight Explore performed automated library searching against local and external libraries (MassBank, HMDB, NIST), supporting MSI level 1 (match to authentic standard) and level 2 (match to spectral database) annotations. Up to five libraries could be searched simultaneously and MS1, retention time and DIA–MS/MS spectra were used for assignment.

Main results and discussion

The combined untargeted workflow identified a panel of metabolites that changed significantly between early and late passage cells and media. Key findings include:

Cell extracts: several metabolites were elevated in passage 20 cells, including polyamines (N8-acetylspermidine, spermidine), certain lysophosphatidylcholines (LPC 16:0, LPC 18:0), carnitine conjugates (palmitoyl- and stearoyl-carnitine), serine and nucleobases/nucleosides (adenosine, adenine). Taurine was identified (MSI level 1) and found to be significantly decreased in passage 20 cell extracts versus passage 3, indicating perturbation in sulfur/amino-acid metabolism associated with senescence.

Culture media: media from passage 20 showed increases in metabolites such as sn-glycerol-3-phosphate, ribose-5-phosphate, inosine, uracil, alpha-ketoglutaric acid and nucleobases/nucleosides, reflecting altered secretion, turnover and extracellular metabolite composition from senescent cells.

Biological interpretation: the observed shifts point to altered energy metabolism, nucleotide turnover, lipid remodeling and polyamine metabolism during replicative senescence. The concordance between intracellular and media signatures supports the potential of media metabolites as non-invasive readouts of cellular ageing in culture models.

Practical benefits and applications of the method

The demonstrated workflow offers several practical advantages:

• DIA-MS/MS acquisition provides comprehensive MS/MS coverage for all detected features, facilitating robust spectral matching without precursor-driven selection bias.
• HILIC separation complements polar metabolite coverage, improving detection of amino acids, nucleotides, small organic acids and polar lipids.
• Interoperability with MS-DIAL and MetaboAnalyst enables use of widely adopted, validated open-source tools for feature detection and statistics, while LabSolutions Insight Explore supports flexible library searching and MSI-compliant identification.
• Filtering thresholds and statistical settings are clearly defined, supporting reproducibility in untargeted discovery workflows.

These characteristics make the approach suitable for biomarker discovery studies, phenotype screening in cell models, and exploratory metabolomics where broad coverage and spectral annotation are priorities.

Used instrumentation

Instruments and key consumables reported in the study:

• UHPLC: Shimadzu Nexera X2.
• Column: Shim-pack Velox HILIC (2.1 × 100 mm, 2.7 µm).
• Mass spectrometer: Shimadzu LCMS-9030 Quadrupole Time-of-Flight (Q-TOF).
• Software: MS-DIAL (feature detection and deconvolution), MetaboAnalyst (statistical analysis), LabSolutions Insight Explore (library searching and identification).

Future trends and potential applications

Opportunities to extend and apply this workflow include:

• Targeted follow-up: transition candidate markers to targeted, quantitative assays (e.g., LC–MS/MS MRM) for validation in larger sample cohorts and biological fluids.
• Multi-omic integration: combine metabolomics with transcriptomics/proteomics to map mechanistic links between metabolic shifts and gene/protein regulation during ageing.
• Improved spectral libraries and community standards: adoption of expanded, well-curated MS/MS libraries and updated MSI guidelines will improve annotation confidence and cross-study comparability.
• High-throughput adaptations: automation of extraction and sample handling plus optimized chromatographic cycles for larger cohort studies.
• Clinical translation: exploring secreted media metabolites in plasma/urine for minimally invasive biomarkers of ageing and age-related disease risk.

Conclusions

The study presents an end-to-end untargeted metabolomics workflow combining HILIC LC separation, DIA-MS/MS on a high-resolution Q-TOF, open-source feature processing (MS-DIAL), statistical analysis (MetaboAnalyst) and library-driven identification (LabSolutions Insight Explore). Applied to an HFF-1 replicative senescence model, the workflow identified multiple intracellular and extracellular metabolites associated with late passage cells, including decreased taurine and increased polyamines, LPCs and acyl-carnitines. The approach is well suited for discovery-stage biomarker identification and provides a reproducible, interoperable template for further validation and extension to larger or clinical studies.

References

1. Spicer R, Salek R, Steinbeck C. A decade after the metabolomics standards initiative it's time for a revision. Sci Data. 2017;4:170138.