Metabolomic Study Tracks How a Fermented Milk Beverage Holds Its Flavor Over 21 Days of Cold Storage

A fermented milk beverage made with the lactic acid bacterium Lactiplantibacillus plantarum LP-315 as the sole starter culture retained viable cell counts above 1 × 10⁷ CFU/mL, comparatively stable acidity, and high sensory scores across 21 days of refrigerated storage, according to a new time-resolved metabolomic study. The research team prepared the beverage from whole milk preheated to 65 °C, mixed with granulated sugar at a sugar-to-milk mass ratio of 6.5:93.5, homogenized at 20 MPa, and heat-treated at 95 °C for 30 minutes. After cooling to 37 °C, the milk was inoculated with LP-315 at an initial concentration of 1 × 10⁷ CFU/mL and fermented until the pH reached 3.8.

The fermented milk base was then blended with sugar solution and purified water at a mass ratio of 25:12:63, adjusted for pH, and homogenized at 18 MPa to yield the finished drink. Samples were collected before fermentation, at the end of fermentation, and after 1, 7, 14, and 21 days of storage at 4 °C. The physicochemical measurements told a story of controlled change.

During fermentation, pH fell from roughly 6.2 to 3.8 while titratable acidity climbed from about 38 to 170 °T, and viable cells multiplied from 1 × 10⁷ to more than 1 × 10⁸ CFU/mL. Once the beverage was formulated and chilled, pH declined progressively and titratable acidity rose only moderately, consistent with limited post-acidification. Although viable counts drifted downward, they stayed above 1 × 10⁷ CFU/mL at day 21.

For comparison, the authors note a separate study of a lactic acid beverage fermented with L. plantarum NJAU-01 in which viable counts fell to 4.40 × 10⁶ CFU/mL after 21 days, though differences in strain, formulation, and processing complicate direct comparisons. Metabolomic profiling revealed that fermentation itself was by far the largest chemical transition. Comparing the beginning and end of fermentation, 418 non-volatile metabolites met the differential abundance criteria—254 more abundant and 164 less abundant at the end of fermentation—the largest number in any pairwise comparison.

Principal component analysis separated the pre-fermentation and post-fermentation samples cleanly, while the day 1 sample stood apart from days 7, 14, and 21, which clustered more closely together. Tightly clustered quality-control samples supported the reproducibility of the chromatographic and mass spectrometric analyses throughout the sequence. During storage, the pace of change was uneven.

The day 7 versus day 1 comparison contained 254 differential metabolites, including 98 amino acids and related metabolites, 38 organic acids and derivatives, 31 glycerophospholipids, 20 carbohydrates, and 20 nucleotide-related compounds. By contrast, only 43 differential metabolites distinguished day 14 from day 7—the smallest count of any comparison—before activity picked up again between days 14 and 21, when 53 metabolites rose and 8 fell. Across the full storage span from day 1 to day 21, 265 metabolites differed, including 160 higher and 105 lower at the end.

Pathway enrichment assigned these differential metabolites predominantly to glycerophospholipid, glycerolipid, starch and sucrose, propanoate, glutathione, tryptophan, branched-chain amino acid degradation, riboflavin metabolism, and oxidative phosphorylation pathways, with glycerophospholipid metabolism standing out for both the number of metabolites and the strength of enrichment. Within individual chemical classes, the trajectories were compound-specific rather than uniform. L-phenylalanine, L-tyrosine, L-cysteine, and L-proline were comparatively abundant at day 1 and generally decreased by day 7, while L-tryptophan, L-arginine, L-ornithine, and L-valine increased at later time points or followed non-monotonic patterns.

This selective remodeling of the free amino acid pool matters for flavor because aromatic amino acids serve as precursors of compounds such as phenylacetaldehyde, 2-phenylethanol, and phenylacetic acid, which carry floral, honey-like, and sweet notes, while branched-chain amino acids can yield malty, fruity, and fermented aromas through transamination and decarboxylation. Carbohydrates and organic acids showed similarly staged behavior. UDP-D-galactose, trehalose, and lactulose were comparatively high at day 1 and then declined, whereas lactose, mannose, mannitol, and maltotriose increased or fluctuated during middle and late storage.

Among organic acids, lactic, pyruvic, succinic, α-ketoglutaric, oxaloacetic, and malic acids peaked in relative abundance at day 1 and then generally decreased, while citric acid ran counter to the trend, remaining low at day 1 and consistently higher from day 7 onward. Lactic acid is the principal end product of carbohydrate fermentation by L. plantarum and drives the beverage's acidity, and the authors suggest that the concurrent variation in lactose and UDP-D-galactose reflects temporal shifts in lactose- and galactose-related metabolic pools during cold storage. On the volatile side, the GC–MS analysis detected 637 volatile compounds, dominated by hydrocarbons (135), heterocyclic compounds (103), and esters (77), with ketones, alcohols, and aromatic compounds also widely represented.

Fermentation again produced the biggest shift: 85 differential volatile compounds distinguished the end of fermentation from the start, with 50 decreasing and 35 increasing. Early storage brought changes to 28 compounds between days 1 and 7, the day 7-to-14 interval saw only six, and late storage revived with 23 compounds changing between days 14 and 21. Cumulatively, 55 volatile compounds differed between days 1 and 21, including declines in several xylene isomers, aldehydes, and δ-dodecalactone, and increases in ethyl nonanoate, 2-heptanone, 2-octanone, and iridomyrmecin.

Separately, the authors report that several furan derivatives and related compounds increased in abundance around day 7, a stage they associate with potentially strengthened sweet and caramel-like odor attributes. To estimate which of these volatiles actually matter for aroma, the team calculated relative odor activity values, which weigh each compound's estimated concentration against its odor threshold. The aroma-active compounds fell into five odor categories—fruity, floral, citrus, sweet, and wine-like—with fruity odorants forming the largest group.

Within the sweet category, 2-methyltetrahydrofuran-3-one showed the highest rOAV by a wide margin, while (E)-2-octen-1-ol, dodecenenitrile, and nonanal led the citrus group and (Z)-2-octen-1-ol, decanal, and linalool ranked highest among floral-associated compounds. Crucially, most leading odorants showed only modest numerical differences between days 1 and 21, suggesting broad retention of the calculated aroma profile across the storage period, even as individual esters, aldehydes, lactones, ketones, and terpenoids followed their own distinct temporal patterns. An integrated correlation analysis then linked the non-volatile and volatile datasets, associating aromatic amino acid metabolism with floral, honey-like, and sweet volatiles; carbohydrate and organic acid metabolism with furan and ketone compounds tied to caramel-like, roasted, and ripened odors; and lipid-related transformations with esters and aldehydes contributing fruity and creamy attributes.

The authors are careful to note the limits of this framework: the statistical associations do not establish precursor–product relationships or metabolic causality, and the rOAV indices rest on semiquantitative concentrations that are subject to compound-dependent extraction and instrumental response differences. Confirming the actual aroma contributions would require targeted quantification with authentic standards, stable isotope tracing, and sensory recombination and omission tests. Still, the overall picture is striking: the beverage's storage stability appears to arise not from metabolic quiescence but from a dynamic balance in which LP-315 shifts into a low-level, selective metabolic state—intensively remodeling carbohydrates, organic acids, and amino acids early in storage, then tapering—while limiting excessive post-acidification and preserving the major sweet, floral, citrus, and fruity aroma characteristics through the full 21 days.

Subject of Research: Metabolomic characterization of refrigerated storage stability and aroma development in a Lactiplantibacillus plantarum LP-315–fermented milk beverage

Article Title: Integrated metabolomic characterization of refrigerated storage stability and aroma development in a Lactiplantibacillus plantarum LP-315–fermented milk beverage

Article References: Jiao, Y., Liu, L., Yuan, J., Wang, J., Chen, Y., & Zha, M. (2026). Integrated metabolomic characterization of refrigerated storage stability and aroma development in a Lactiplantibacillus plantarum LP-315–fermented milk beverage. Food Chemistry: X, 40, Article 104674. https://doi.org/10.1016/j.fochx.2026.104674

Image Credits: AI Generated

DOI: 10.1016/j.fochx.2026.104674

Keywords: Lactiplantibacillus plantarum, fermented milk beverage, metabolomics, refrigerated storage, aroma compounds, lactic acid bacteria, post-acidification, odor activity value, GC-MS, UPLC-MS, food flavor chemistry, probiotics