Fermentation method outweighs geographic origin in shaping kefir's in vitro anti-obesity potential and probiotic properties

Abstract

Kefir is a traditional fermented dairy product containing diverse microbial consortia with potential health benefits. However, systematic comparisons of fermentation methods and grain origins on kefir's functional properties remain limited. This study selected kefir grains from eight geographic regions worldwide and prepared kefir fermented milk using both kefir grain fermentation (KGF) and back-slopping fermentation (BSF) methods. The in vitro anti-obesity potential and probiotic characteristics were evaluated, and multidimensional analysis was employed to elucidate the influence of different fermentation methods and regional sources on functional properties. Results demonstrated that the fermentation method exerted a stronger influence than geographic origin. KGF generally yielded kefir with higher antioxidant capacity, stronger antimicrobial activity, and safer probiotic profiles, whereas BSF favored lactic acid bacteria proliferation and triglyceride reduction. These findings emphasize that changes in fermentation strategy reshape the microbial community structure, thereby influencing the functional quality and probiotic properties of kefir. Consequently, KGF is recommended as the optimal strategy for producing kefir with enhanced health-promoting properties and superior safety.

Highlights

• •Fermentation method dominated kefir functionality, surpassing grain geographic origin.
• •KGF enhanced probiotic safety and anti-obesity activity compared with BSF.
• •KGF supernatants showed stronger antioxidant and pathogen inhibition capacities.
• •BSF favored LAB growth and triglyceride reduction but lacked E. cloacae inhibition.
• •TOPSIS scoring confirmed KGF as superior for health-promoting kefir production.

1. Introduction

Kefir is a complex fermented dairy product with a consumption history spanning several millennia. It harbors diverse microbial populations, including lactic acid bacteria (LAB), acetic acid bacteria, and yeasts, which confer distinctive organoleptic properties and probiotic characteristics. Regular kefir consumption has been associated with multiple health benefits, encompassing anti-cancer (Fatahi et al., 2021); anti-obesity (Bourrie et al., 2022); antidepressant (Noori et al., 2025); anti-allergic (Baars et al., 2025); hepatoprotective (Cui et al., 2024); hypoglycemic; and gut microbiota modulation (Coco et al., 2023). Kefir grains; the traditional inoculum for kefir fermentation; manifest as ivory-white or white irregular gelatinous particles comprising symbiotic microbial consortia of LAB; acetic acid bacteria; and yeasts. They are distributed worldwide across Europe; Asia; Africa; and the Americas. Although kefir grains collected from different geographic regions demonstrate similarity at the core microbial level; substantial variations exist among minor and low-abundance species; reflecting the influence of geographic location on microbial community structure (Walsh et al., 2023). During the kefir fermentation process; the dynamic succession of the microbiota is profoundly modulated by the fermentation substrate; which directly dictates the functional and organoleptic consequences. For instance; in kefir-fermented vegetable juices; the succession of specific microflora is strongly correlated with the evolution of distinct volatile flavor compounds (Wang et al., 2024). Similarly; co-culture fermentation of water kefir within dark tea infusions drives specific microbial succession; selectively enriching beneficial LAB (notably Liquorilactobacillus and Lacticaseibacillus); which dynamically modulates metabolic pathways and flavor development (Gao et al., 2025).

Obesity is defined as excessive adipose tissue accumulation resulting from energy intake-expenditure imbalance and constitutes a global health concern affecting all age demographics (Sakers et al., 2022). Beyond lifestyle interventions and pharmacological treatments; functional fermented foods possessing anti-obesity potential offer safe; drug-free alternatives for obese populations. Kefir; as a functional fermented food containing complex beneficial microbial communities and bioactive compounds; has been demonstrated to exert anti-obesity effects through gut microbiota improvement (Tarakci et al., 2022); reduction of plasma cholesterol and triglyceride concentrations (Bourrie et al., 2022); modulation of bile acid metabolism (Gao et al., 2021); and amelioration of insulin resistance and sensitivity (Mutalub et al., 2025).

The geographic origin of kefir grains and fermentation methodology influence both product characteristics and functional properties of kefir. Kefir grains obtained from different geographic locations exhibit considerable variation in morphological appearance, fermentation characteristics, and microbial composition. Consequently, the resulting kefir products display notable differences, including disparities in functional properties, with grains collected from certain regions potentially presenting food safety concerns (González-Orozco et al., 2023). Kefir grain fermentation (KGF) and back-slopping fermentation (BSF) represent the two principal methods for kefir production. KGF directly utilizes kefir grains for fermentation and constitutes the traditional approach; whereas BSF employs kefir produced from kefir grains as the seed culture (Garofalo et al., 2020). Generally; kefir prepared by KGF exhibits superior functional properties compared to commercial kefir. Although commercial starter cultures yield more standardized products; the health benefits of kefir have been found to be exclusively associated with consumption of products manufactured using complex microbial communities derived from kefir grains or kefir. BSF is postulated to recapitulate the microbial; nutritional; and physicochemical characteristics (Kim et al., 2018) as well as functional properties (Bourrie et al., 2021) of traditional kefir; however; comprehensive comparative investigations remain lacking. Beyond milk kefir; the profound impact of the fermentation modality (KGF versus BSF) has been recognized in other fermented matrices. For example; water kefir cultivated in fruit and vegetable juices can be propagated via either direct grain fermentation or the back-slopping method. Studies indicate that the ultimate compositional; functional; and sensory profiles of these beverages are significantly modulated not only by their inherent microbiological characteristics but specifically by the chosen fermentation modality (de Almeida et al., 2025).

Given the importance of microbial origin, structure, and interactions in functional fermented foods, systematic evaluation is required to clarify whether the fermentation method or geographic origin exerts the dominant influence on kefir's functional properties. Therefore, this study selected kefir grains from eight geographic regions worldwide and prepared kefir using both KGF and BSF. The resulting products were assessed for functional indicators relevant to obesity-related metabolism (α-glucosidase and pancreatic lipase inhibition, antioxidant capacity, cholesterol and triglyceride degradation) and probiotic traits (gastrointestinal tolerance, adhesion, bile salt hydrolase activity, antibiotic resistance, hemolytic characteristics, and nitrate reduction). Multivariate analyses, including principal component analysis (PCA), correlation analysis, and Technique for Order Preference by Similarity to Ideal Solution (TOPSIS), were employed to determine the relative contributions of fermentation method and geographic origin to kefir's functional and probiotic properties.

2. Materials and methods

The kefir grains utilized in this investigation (Supplementary Table S1, Supplementary Fig. S1), along with Enterobacter cloacae CMCC(B) 45301 and Staphylococcus aureus ATCC 6538, were acquired from and maintained in the collection of Provincial Key Laboratory of Probiotics and Deep Processing of Dairy Products (Yangzhou University).

2.1. Activation and preservation of kefir grain

Kefir grains maintained under cryogenic storage were activated by inoculation into sterile whole milk containing 3.2% (w/v) protein and 4.0% (w/v) fat at an inoculation ratio of 3 g/100 mL. Fermentation was conducted in gas-permeable fermentation vessels at 25 °C for 24 h under static conditions to replicate traditional household fermentation practices. Following fermentation, kefir grains were recovered using sterile household mesh strainers that had been rinsed with sterile water and subjected to ultraviolet sterilization. The collected grains were washed twice with double-distilled water before being transferred to fresh milk for subsequent cultivation cycles. After two consecutive activation cycles, the kefir grains were considered suitable for experimental use or long-term storage. For extended preservation, kefir grains were suspended in 80% (v/v) sterile glycerol and maintained at −80 °C in an ultra-low temperature freezer (Blasche et al., 2021).

2.2. Preparation and storage of freeze-dried kefir starter culture

Kefir grains were activated following the procedures described above. The activated kefir grains were subsequently inoculated into sterile whole milk and fermented statically at 25 °C for 24 h to produce kefir fermented milk. The resulting fermented milk was thoroughly mixed with a modified cryoprotectant solution at a 1:1 ratio, dispensed into sterile vials, and pre-frozen overnight at −80 °C. The samples were then subjected to vacuum freeze-drying (13 mTorr, −86 °C, 48 h). Upon completion of lyophilization, the freeze-dried kefir starter culture was stored in an ultra-low temperature freezer at −80 °C. The modified cryoprotectant formulation was referenced from Lv et al. (2021).

2.3. KGF and BSF

KGF: Activated kefir grains were inoculated into sterile whole milk at 3% (w/v) and fermented statically at 25 °C for 24 h. Upon completion of fermentation, the fermented milk was centrifuged at 4500 rpm for 10 min at 4 °C. The supernatant was collected and filtered through a 0.22 μm sterile filter to remove residual bacterial cells. The pellet was washed twice with sterile phosphate-buffered saline (PBS; Solarbio). The resulting cell-free supernatant and kefir microbiota were temporarily stored at −4 °C for subsequent analysis.

BSF: The initial viable counts of the freeze-dried kefir starter powders ranged from 7.50 to 7.93 log CFU/g for LAB and 4.82 to 5.51 log CFU/g for yeasts (detailed counts for each group are provided in Supplementary Table S2). The freeze-dried kefir starter culture was reconstituted in whole milk at a volume equal to half that of the cryoprotectant used prior to lyophilization. The reconstituted culture was then inoculated into whole milk at 3% (v/v) and fermented statically at 25 °C for 24 h for activation. The activated fermented milk was subsequently re-inoculated into fresh whole milk at 3% (v/v) and fermented under identical conditions. This specific fermentation batch represents the second generation (subculture) of the back-slopping process. To ensure propagation consistency and standardized microbial profiling, all subsequent in vitro functional assays and safety evaluations in this study were strictly conducted using these second-generation BSF samples. Following fermentation, the cell-free supernatant and kefir microbiota were prepared using the same procedures described above.

2.4. Enumeration of LAB and yeasts

1 mL of kefir or 1 g kefir grain was added to 9 mL of sterile PBS and thoroughly vortexed before serial dilution. For each dilution, 1 mL of liquid was plated in triplicate on selective media. LAB were enumerated on de Man, Rogosa and Sharpe medium (MRS; Hopebio, Qingdao, China) agar plates supplemented with 0.4 g/L nystatin (Solarbio) to inhibit yeast growth. Yeasts were enumerated on Rose Bengal Chloramphenicol Agar (RBCA; Hopebio), where chloramphenicol inhibited LAB growth. LAB were cultivated under anaerobic conditions at 37 °C, while yeasts were cultivated at 25 °C. Colony counts were performed at 48 h and 96 h, respectively. Results were expressed as the logarithm of colony-forming units per milliliter of kefir (log CFU/mL).

2.5. In vitro α-glucosidase inhibition assay

25 μL of cell-free kefir supernatant was mixed with 25 μL of α-glucosidase solution (1.5 U/mL; Macklin, Shanghai, China) and incubated at 37 °C for 10 min. Subsequently, 25 μL of 1 mM p-nitrophenyl-α-D-glucopyranoside (pNPG; Macklin) was added, and the mixture was incubated for an additional 15 min. The reaction was terminated by adding 100 μL of 1 M sodium carbonate solution (Sinopharm). The reaction mixture was incubated in a boiling water bath for 5 min, cooled to room temperature, and the absorbance was measured at 405 nm. PBS was used in place of the cell-free supernatant as the blank control. The α-glucosidase inhibition rate was calculated using the following formula:

$$\mathrm{\alpha }-\text{Glucosidase Inhibition}\left(\%\right)=\left(\frac{{\mathrm{OD}}_{\mathrm{con}405\mathrm{nm}}-{\mathrm{OD}}_{\exp 405\mathrm{nm}}}{{\mathrm{OD}}_{\mathrm{con}405\mathrm{nm}}}\right)\times 100\%$$

where ${\mathrm{OD}}_{\mathrm{con}405\mathrm{nm}}$ represents the absorbance of the control group and ${\mathrm{OD}}_{\exp 405\mathrm{nm}}$ represents the absorbance of the experimental group.

2.6. In vitro pancreatic lipase inhibition assay

The reaction mixture was prepared by combining 50 μL of cell-free kefir supernatant, 100 μL of 1 M Tris-HCl buffer (pH 8.5), and 50 μL of pancreatic lipase solution (1 U/mL dissolved in 1 M Tris-HCl buffer; Macklin). The mixture was incubated at 37 °C for 25 min. Subsequently, 50 μL of 5 mM p-nitrophenyl palmitate (p-NPP; Macklin) dissolved in dimethyl sulfoxide (DMSO) was added, and the incubation was continued at 37 °C for an additional 25 min. Absorbance was recorded at 412 nm using a microplate reader. PBS served as the blank control, and the pancreatic lipase inhibition rate (%) was calculated using the following formula:

$$\text{Pancreatic lipase inhibition rate}\left(\%\right)=\left(\frac{{\mathrm{OD}}_{\mathrm{con}412\mathrm{nm}}-{\mathrm{OD}}_{\exp 412\mathrm{nm}}}{{\mathrm{OD}}_{\mathrm{con}412\mathrm{nm}}}\right)\times 100\%$$

where${\mathrm{OD}}_{\mathrm{con}412\mathrm{nm}}$represents the absorbance of the blank control group and ${\mathrm{OD}}_{\exp 412\mathrm{nm}}$ represents the absorbance of the experimental group.

2.7. In vitro inhibitory effects against E. cloacae

E. cloacae CMCC(B) 45,301 was activated in Luria-Bertani (LB) broth (Hopebio) and cultured at 37 °C for 12–24 h. Under aseptic conditions in a biosafety cabinet, 15 mL of LB agar was poured into disposable plates and allowed to solidify. A separate portion of LB agar was pre-warmed to 55 °C, inoculated with bacterial suspension (10⁷ CFU/mL), and gently mixed to avoid bubble formation. 5 mL of inoculated agar was poured onto the surface of the solidified agar plates and gently swirled to ensure uniform distribution. After solidification, four-unit Oxford cups (Hopebio) were placed on the agar surface and pressed for 5 s. 100 μL cell-free kefir supernatant was added to each cup, with sterile PBS serving as the control. The plates were allowed to stand at 37 °C for 2 h, then incubated under anaerobic conditions for 12–24 h. The appearance of inhibition zones was observed, and their diameters were measured in millimeters.

2.8. Antioxidant capacity assay

The antioxidant properties of cell-free kefir supernatant were evaluated using three indicators: DPPH eliminating ability, total antioxidant capacity (T-AOC assay), and lipid peroxidation inhibition capacity. DPPH eliminating ability and total antioxidant capacity were determined using commercial kits (JianCheng, Nanjing, China). Lipid peroxidation inhibition capacity was assessed using the thiobarbituric acid (TBA) method (Wei et al., 2022). Specifically, 1 mL of linoleic acid (Macklin), 0.5 mL of 10 g/L ferrous sulfate (Sinopharm), 0.2 mL of 10 g/L ascorbic acid (Sinopharm), and 0.5 mL of cell-free kefir supernatant were mixed and incubated at 37 °C for 12 h. Subsequently, 1 mL of the reaction mixture was analyzed for lipid peroxidation using a TBA-based malondialdehyde (MDA) detection kit (JianCheng). PBS served as the blank control, and absorbance was measured at 532 nm to characterize lipid peroxidation levels.

$$\text{Lipid peroxidation inhibition rate}\left(\%\right)=\left(\frac{{\mathrm{OD}}_{\mathrm{con}532\mathrm{nm}}-{\mathrm{OD}}_{\exp 532\mathrm{nm}}}{{\mathrm{OD}}_{\mathrm{con}532\mathrm{nm}}}\right)\times 100\%$$

where ${\mathrm{OD}}_{\mathrm{con}532\mathrm{nm}}$represents the absorbance of the blank control group and ${\mathrm{OD}}_{\exp 532\mathrm{nm}}$represents the absorbance of the experimental group.

2.9. In vitro cholesterol degradation capacity

A cholesterol (Sinopharm)–ethanol (Sinopharm) solution was prepared, sterilized by membrane filtration, and added to MRS broth to achieve a final cholesterol concentration of 0.1 g/L. The kefir microbiota was inoculated into the cholesterol–MRS broth at 3% (v/v) and cultivated under anaerobic conditions at 37 °C for 24 h. The control group was prepared by inoculating an equal volume of sterile PBS. Following fermentation, the cholesterol–MRS broth was centrifuged at 10,000 rpm for 10 min at 4 °C, and the supernatant was collected. Cholesterol concentration in the supernatant was determined using a commercial total cholesterol assay kit (JianCheng). The cholesterol degradation rate (%) was calculated using the following formula:

$$\text{Cholesterol degradation rate}\left(\%\right)=\left(\frac{{\mathrm{C}}_{\mathrm{con}}-{\mathrm{C}}_{\exp }}{{\mathrm{C}}_{\mathrm{con}}}\right)\times 100\%$$

where C_con represents the cholesterol concentration in the control group, and C_exp represents the cholesterol concentration in the experimental group.

2.10. In vitro triglyceride degradation capacity

A uniform mixture was prepared by thoroughly mixing 15 mL of 2% (v/v) polyvinyl alcohol (PVA; Sinopharm, Shanghai, China) aqueous solution with 5 g of triglyceride (Sinopharm). The triglyceride–PVA mixture was added to MRS broth at 3% (v/v), and the pH was adjusted to 6.5 prior to sterilization at 121 °C. After cooling, the kefir microbiota was inoculated into the triglyceride–MRS broth at 3% (v/v) and cultivated under anaerobic conditions at 37 °C for 72 h. Following incubation, the culture was centrifuged at 10,000 rpm for 10 min at 4 °C, and the supernatant was collected. Triglyceride concentration in the supernatant was determined using a commercial assay kit (JianCheng).

The control group was prepared by inoculating an equal volume of sterile PBS into the triglyceride–MRS broth. The triglyceride Lowering rate (%) was calculated using the following formula:

$$\text{Triglyceride degradation rate}\left(\%\right)=\left(\frac{{\mathrm{C}}_{\mathrm{con}}-{\mathrm{C}}_{\exp }}{{\mathrm{C}}_{\mathrm{con}}}\right)\times 100\%$$

where C_con represents the triglyceride concentration in the control group, and C_exp represents the triglyceride concentration in the experimental group.

2.11. BSH activity assay

MRS agar was supplemented with 0.37% (w/v) calcium chloride (Sinopharm), 2% (w/v) sodium thioglycolate (Sinopharm), and 0.5% (w/v) taurodeoxycholic acid (TCA; Sinopharm) or glycodeoxycholic acid (GCA; Sinopharm). Sterile filter paper discs (6 mm diameter) were placed on the agar surface, and 20 μL of kefir microbiota suspension was inoculated onto each disc. The plates were incubated under anaerobic conditions at 37 °C for 72 h and photographed to record the presence of precipitation zones around the filter paper discs. The formation of precipitates surrounding the discs indicated BSF activity of the microbiota (Cai et al., 2024).

2.12. Tolerance of kefir microbiota to simulated gastrointestinal fluids

Simulated gastric fluid (SGF) was prepared by adding 3 g/L pepsin (Macklin) to sterile PBS and adjusting the pH to 2.5 with 0.1 mol/L hydrochloric acid (Sinopharm), followed by sterilization through a 0.22 μm sterile filter. Simulated intestinal fluid (SIF) was prepared by adding 1 g/L trypsin (Sinopharm) and 3 g/L porcine bile salts (Sinopharm) to sterile PBS, adjusting the pH to 8.0 with 0.1 mol/L sodium hydroxide, and sterilizing through a 0.22 μm filter. After washing, the kefir microbiota was sequentially incubated with SGF (2 h) and SIF (4 h) under anaerobic conditions at 37 °C. Following incubation, viable counts of LAB were determined by plate counting.

$$\text{Survival rate}\left(\%\right)=\left(\frac{{\mathrm{N}}_1}{{\mathrm{N}}_0}\right)\times 100\%$$

where N₀ represents the initial viable count at the beginning of the experiment (0 h), and N₁ represents the viable count detected after incubation in SGF or simulated gastrointestinal fluids.

2.13. Adhesion capacity of kefir microbiota

Human colorectal adenocarcinoma cell line Caco-2 (Caco-2 cells; Procell Life Science, Wuhan, China) in the logarithmic growth phase were seeded into 24-well plates (well diameter 16 mm, effective membrane area 1.9 cm²) at a density of 4.5 × 10⁵ cells per well and cultured at 37 °C with 5% CO₂ until reaching 90% confluence. Cell growth was monitored using an inverted microscope. Activated kefir microbiota were centrifuged at 4000 rpm for 20 min, washed twice with sterile PBS, and the supernatant was discarded. The bacterial pellet was resuspended in Dulbecco's Modified Eagle Medium (DMEM; Beyotime, Shanghai, China) supplemented with 10% (v/v) fetal bovine serum (Beyotime), and the bacterial concentration was adjusted to 10⁷–10⁸ CFU/mL. The culture medium in the 24-well plates was removed, and the wells were washed twice with sterile PBS. 1 mL of the prepared bacterial suspension was added to each well and incubated at 37 °C with 5% CO₂ for 2 h. After incubation, the culture medium was discarded, and the wells were gently washed three times with PBS buffer to remove excess non-adhered bacterial cells. Each well was treated with 0.5 mL of 0.25% trypsin (Beyotime) for 1–2 min, and digestion was terminated by adding 0.5 mL of complete DMEM. The adherent cells were detached by repeated pipetting to form a uniform single-cell suspension. 100 μL of the bacterial suspension co-incubated with Caco-2 cells was serially diluted to appropriate concentrations, plated on MRS agar, and cultured under anaerobic conditions at 37 °C for 48 h to determine viable counts. The adhesion rate was calculated using the following formula:

$$\text{Adhesion rate}\left(\%\right)=\left(\frac{{\mathrm{N}}_{2\mathrm{h}}}{{\mathrm{N}}_{0\mathrm{h}}}\right)\times 100\%$$

where N_0h represents the viable count before co-incubation with Caco-2 cells, and N_2h represents the viable count after 2 h of co-incubation with Caco-2 cells.

2.14. Antibiotic resistance testing

The Kirby-Bauer disc diffusion method was employed to assess the antibiotic susceptibility of kefir microbiota. The antibiotic types and concentrations of the antimicrobial susceptibility discs (Binhe, Hangzhou, China) are presented in Supplementary Table S3. Due to the mixed nature of the microbial community, heterogeneous inhibition patterns (concentric zones) were frequently observed. In such cases, the diameters of all distinct inhibition zones were measured and classified independently according to the criteria. Therefore, a single sample could be recorded as possessing resistant, intermediate, and susceptible subpopulations for a specific antibiotic, reflecting the physiological diversity of the consortium.

2.15. Hemolytic activity test

The kefir microbiota was streaked onto Columbia blood agar plates (Hopebio) and incubated inverted at 37 °C for 48 h before observing color changes around individual colonies. Staphylococcus aureus ATCC 6538 served as the control strain. The appearance of a greenish zone around colonies indicated α-hemolysis (positive), a clear zone indicated β-hemolysis (positive), and no color change indicated γ-hemolysis (negative), signifying the absence of hemolytic activity.

2.16. Nitrate reductase activity assay

The kefir microbiota was inoculated into nitrate reductase detection medium (Hopebio) and cultured at 37 °C for 5 days. Detection was performed using a commercial nitrate reductase assay kit (Hopebio). Staphylococcus aureus ATCC 6538 served as the positive control strain. A red color reaction in the medium indicated nitrate reductase activity (positive), while no color change indicated the absence of nitrate reduction capacity (negative).

2.17. Statistical analysis

Unless otherwise stated, all experiments were performed in triplicate as biological replicates, and error bars represent standard deviation. Comparisons between two group means were performed using the least significant difference (LSD) test, while multiple comparisons among multiple groups were conducted using one-way analysis of variance (ANOVA) followed by Waller-Duncan test. Statistical significance was defined as P < 0.05. Data processing was performed using SPSS software (version 25.0, IBM, Armonk, NY, USA). PCA and correlation analysis were conducted in Origin 2025 (OriginLab, Northampton, MA, USA), with data preprocessed using the mean-centering method. Pearson correlation analysis was employed for correlation assessment. TOPSIS comprehensive evaluation was performed online using SPSSAU (https://spssau.net/), with equal weights assigned to all indicators.

3. Results

3.1. Fermentation kinetics of kefir

During kefir fermentation, LAB and yeasts constitute the dominant microbial populations, and changes in their viable counts reflect the dynamic characteristics of microbial communities under different fermentation methods. As shown in Fig. 1a and d, under KGF, the viable counts of LAB increased rapidly to 6.15–7.00 log CFU/mL within 3–9 h. Several groups (DE1, RU, NZ, UZB) exhibited a transient decline at 12 h, while other groups continued to grow, reaching 7.72–8.14 log CFU/mL at 24 h. In contrast, under BSF, all groups demonstrated consistent growth patterns, attaining peak values of 8.41–8.78 log CFU/mL at 12 h. As illustrated in Fig. 1g, kefir samples from different origins exhibited limited variation within the same fermentation method; however, the fermentation method exerted significant effects on certain groups. Notably, the AZ group exhibited significantly higher viable counts of LAB under BSF compared to KGF (an increase of approximately 1.6 log CFU/mL, P < 0.001).

Fig. 1.

Changes in viable counts of LAB, yeasts, and pH during kefir fermentation. a–c, Viable counts of LAB (a), yeasts (b), and pH (c) changes under KGF. d–f, Viable counts of LAB (d), yeasts (e), and pH (f) changes under BSF. g–i, Viable counts of LAB (g), yeasts (h), and pH (i) at the fermentation endpoint for kefir samples from different origins and regions. The x-axis represents the origin of kefir grains (see Supplementary Table S1). green bars indicate KGF, and red bars indicate BSF Different lowercase letters denote significant differences among groups within the same fermentation method (p < 0.05, Tukey's multiple comparison test); asterisks indicate significant differences between the two fermentation methods for the same origin (ns, p > 0.05; *, p ≤ 0.05; **, p ≤ 0.01; ***, p ≤ 0.001; ****, p ≤ 0.0001). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

The growth pattern of yeasts differed from that of LAB (Fig. 1b, e, h). Under KGF, yeast counts reached 5.60–6.72 log CFU/mL at 24 h, whereas under BSF, most groups remained below the detection limit during the first 9 h. By 24 h, the US group exhibited the highest yeast count (5.38 log CFU/mL), while the UZB group showed the lowest (3.46 log CFU/mL). Overall, kefir produced by KGF contained higher yeast populations. The pH declined over time under both fermentation methods (Fig. 1c, f, i); however, the pH decreased more rapidly under BSF, resulting in a lower endpoint pH compared to KGF. This difference was most pronounced in the UZB, AZ, CN, and DE2 groups (p < 0.01).

3.2. In vitro obesity-related functional indicators and antioxidant capacity of cell-free kefir supernatants

To comprehensively evaluate the in vitro anti-obesity potential of kefir, the α-glucosidase inhibitory activity, pancreatic lipase inhibitory activity, and antimicrobial activity against E. cloacae of cell-free kefir supernatants were investigated. Additionally, the antioxidant properties of kefir were assessed through DPPH eliminating ability, total antioxidant capacity (T-AOC assay), and lipid peroxidation inhibition capacity.

As shown in Fig. 2a, the α-glucosidase inhibition rates of cell-free kefir supernatants ranged from 64.80% to 91.63%. Under KGF, the DE1 group exhibited the highest inhibition rate, while under BSF, the highest value was observed in the UZB group. Overall, the fermentation method exerted limited influence on α-glucosidase inhibition rates; only in the DE1 and AU groups did supernatants prepared by KGF demonstrate significantly higher (p < 0.001) inhibition rates than those prepared by BSF. As illustrated in Fig. 2b, pancreatic lipase inhibition rate across all groups ranged from 41.92% to 50.95%, with no significant differences (p > 0.05), indicating that neither the fermentation method nor the kefir grain origin substantially affected this parameter. Fig. 2c displays the antimicrobial activity against E. cloacae. Under BSF, only the AZ and CN groups exhibited inhibitory activity, whereas supernatants prepared by KGF demonstrated inhibitory effects in all groups, with significantly higher (p < 0.05) activity compared to BSF.

Fig. 2.

In vitro obesity-related functional indicators and antioxidant activities of cell-free kefir supernatants. a, α-Glucosidase inhibition rate. b, Pancreatic lipase inhibition rate. c, Antimicrobial activity against E. cloacae. d, DPPH eliminating ability. e, Total antioxidant capacity (T-AOC assay). f, Lipid peroxidation inhibition rate.

Regarding antioxidant capacity (Fig. 2d, e), all supernatants exhibited robust DPPH eliminating ability and total antioxidant capacity. Except for the UZB group, supernatants prepared by KGF demonstrated significantly superior (p < 0.005) DPPH eliminating ability compared to those prepared by BSF. Similarly, for total antioxidant capacity, the RU, AZ, CN, and DE2 groups showed a consistent trend favoring KGF over BSF. As shown in Fig. 2f, the lipid peroxidation inhibition capacity of cell-free kefir supernatants varied considerably, with some groups even displaying pro-oxidant effects. Overall, the inhibitory effect under BSF was weaker, while the supernatant from the AU group prepared by KGF exhibited the best performance, with an inhibition rate of 51.47%.

3.3. In vitro obesity-related functional indicators of kefir microbiota

The cholesterol degradation capacity and triglyceride degradation capacity of microbiota reflect their anti-obesity potential, which are inversely correlated with host weight gain and energy storage. As shown in Fig. 3a, kefir microbiota from all groups demonstrated cholesterol degradation capacity. Under KGF, differences in cholesterol degradation ability among groups were minimal, whereas under BSF, all groups except the US and AU groups (< 45%) exhibited degradation rates exceeding 60%. Furthermore, in the US, AZ, and AU groups, cholesterol degradation rates under KGF were significantly higher (p < 0.05) than those under BSF. As illustrated in Fig. 3b, triglyceride degradation rates under KGF ranged from 24.66% to 55.04%, while those under BSF ranged from 46.95% to 62.52%. Statistical analysis revealed that the RU, US, and AZ groups exhibited significantly higher (p < 0.05) triglyceride reduction rates under BSF compared to KGF. Fig. 3c presents the evaluation of BSH activity. Results demonstrated that kefir microbiota exhibited hydrolytic activity toward TCA but not toward GCA. Additionally, the extent of precipitation zones indicated that microbiota prepared by KGF displayed stronger BSH activity.

Fig. 3.

In vitro obesity-related functional indicators of kefir microbiota. a, Cholesterol degradation rate. b, Triglyceride degradation rate. c, BSH activity, including hydrolytic activity toward TCA and GCA. The appearance of white precipitates around filter paper discs indicates BSH activity.

3.4. Probiotic characteristics of kefir microbiota

As shown in Fig. 4a, after 2 h of SGF digestion, the survival rates of LAB varied significantly among groups. Several groups (the US group under KGF, and the UZB, AZ, AU, and CN groups under BSF) exhibited survival rates exceeding 100%, indicating a highly robust tolerance of these specific kefir microbiota to the acidic gastric environment. Conversely, the UZB, AU, CN, and DE2 groups under KGF, as well as the RU group under BSF, displayed survival rates below 50%. The overall trend indicated that kefir microbiota prepared by BSF demonstrated superior gastric tolerance compared to those prepared by KGF. As illustrated in Fig. 4b, after 6 h of sequential simulated gastric and intestinal fluid digestion, the survival rates of LAB declined substantially. Except for the CN group under KGF, which maintained a survival rate of 13.21%, all other groups exhibited survival rates below 5%, indicating that most kefir-derived LAB possess limited viability under prolonged gastrointestinal exposure. Fig. 4c displays the adhesion capacity of kefir LAB to Caco-2 cells. Only the RU and US groups under KGF exhibited adhesion rates exceeding 15% (15.06% and 30.80%, respectively), while all other groups displayed adhesion rates below 10%.

Fig. 4.

Gastrointestinal tolerance, Caco-2 cell adhesion ability, and safety profiles of kefir microbiota. a, Survival rate of LAB after 2 h of SGF digestion. b, Survival rate of LAB after 6 h of sequential SGF and SIF digestion. c, Adhesion rate to Caco-2 cells. d, Circular heatmap of antibiotic resistance and number of resistant inhibition zones (green: susceptible; yellow-green: intermediate; red: resistant). e, Hemolytic activity test results (PC, positive control). f, Nitrate reductase activity test results (red: positive; white: negative; PC, positive control). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 4d illustrates the antibiotic resistance profiles of kefir microbial communities. Since kefir microbial communities comprise multiple strains, phenotypic heterogeneity was observed, where the same antibiotic produced multiple inhibition zones corresponding to different subpopulations. The heatmap visualizes this diversity, displaying simultaneous resistance, susceptibility, or intermediate responses within single groups. However, from a safety perspective, the bar chart focuses on the frequency of resistance. It indicates that, except for the RU and DE1 groups, microbiota prepared by KGF exhibited fewer resistant inhibition zones compared to those prepared by BSF. Fig. 4e demonstrates that kefir microbiota from all origins and fermentation methods lacked hemolytic activity, meeting probiotic safety requirements. Fig. 4f shows that all microbiota prepared by BSF tested positive for nitrate reductase activity, whereas under KGF, only the NZ and UZB groups were positive.

3.5. Multidimensional analysis of in vitro anti-obesity potential and probiotic properties of kefir

As shown in Fig. 5a, PCA effectively distinguished kefir samples prepared by different fermentation methods. Samples prepared by KGF were predominantly distributed on the right side of PC1, while those prepared by BSF clustered on the left side of PC1. The contribution directions of functional indicators revealed that viable counts of LAB, triglyceride reduction rates, and pancreatic lipase inhibition rates were more strongly associated with BSF samples, whereas antioxidant-related indicators (DPPH eliminating ability, total antioxidant capacity) and other probiotic characteristics were more closely associated with KGF samples. Fig. 5b further revealed correlations among different functional indicators. Viable counts of LAB exhibited significant negative correlations with yeast counts (r = −0.81, p < 0.0005), DPPH radical scavenging capacity (r = −0.81, p < 0.0001), and inhibition zone diameter against E. cloacae (r = −0.81, p < 0.0001). Conversely, yeast counts showed significant positive correlations with α-glucosidase inhibition rate (r = 0.61, p < 0.01), DPPH eliminating ability (r = 0.69, P < 0.005), and inhibition zone diameter against E. cloacae (r = 0.81, p < 0.001). Regarding anti-obesity properties, α-glucosidase inhibition rate was negatively correlated with pancreatic lipase inhibition rate (r = −0.59, p < 0.01) but positively correlated with DPPH eliminating ability (r = 0.86, p < 0.001) and total antioxidant activity (r = 0.52, p < 0.05). These results suggest that antioxidant capacity may be closely associated with anti-obesity potential.

Fig. 5.

PCA, correlation, and TOPSIS-based evaluation of in vitro anti-obesity potential and probiotic characteristics of kefir under different fermentation methods. a, PCA showing the distribution of samples prepared by different fermentation methods. b, Correlation analysis of functional indicators (Pearson correlation coefficients; circle size indicates correlation strength, color indicates positive or negative correlation, asterisks indicate significance levels). c, TOPSIS scores based on probiotic characteristics. d, TOPSIS scores based on anti-obesity characteristics, equal weights were assigned to all indicators in the TOPSIS comprehensive evaluation.

TOPSIS scoring results are presented in Fig. 5c and d. For probiotic characteristics, AU-KGF, CN-KGF, and RU-KGF ranked in the top three; for anti-obesity characteristics, CN-KGF, AZ-KGF, and US-KGF performed best. Notably, CN-KGF exhibited outstanding performance in both categories, demonstrating comprehensive advantages. When average scores were calculated by grouping according to fermentation method, KGF outperformed BSF in both probiotic and anti-obesity characteristics. Supplementary Fig. S2 further validated this trend, with CN-KGF, AU-KGF, and RU-KGF ranking in the top three for comprehensive scores.

4. Discussion

The microbial architecture of kefir grains exhibits a distinctive spatial stratification: microbial density progressively decreases from the outer to the inner layers, a phenomenon attributed to limited nutrient availability and lower pH in the core regions (Bourrie et al., 2016). Predominant bacterial genera; including Lactobacillus; Lactococcus; and Acetobacterium; are localized on the external surface of kefir grains; whereas the yeast genus Saccharomyces is distributed throughout both outer and inner layers (Plessas et al., 2017). In terms of absolute abundance; eukaryotic microorganisms constitute less than 2% of the kefir microbial community; with the remainder predominantly comprising bacterial taxa (Walsh et al., 2023). Consistent with these findings; kefir fermented using the KGF method maintained a consistent differential of 1–2 log CFU/mL between LAB and yeast populations throughout the fermentation process. In the KGF model; the microbial consortium within the grains remains stable while the milk matrix undergoes sequential colonization (Blasche et al., 2021). Conversely; in BSF; the microbial community exists in a planktonic state without sequential colonization dynamics. Consequently; BSF demonstrates enhanced stability in fermentation kinetics; with kefir microbial communities from geographically distinct origins maintaining considerable similarity throughout the fermentation process (Fig. 1d–f). Nevertheless; the microbial community structure of kefir grains is significantly influenced by geographic provenance; exhibiting substantial variation in fermentation proliferation patterns and viable counts of both LAB and yeast (Fig. 1a–c). It is crucial to note that the comparison between KGF and BSF in this study was evaluated at the macroscopic process level rather than based on microbial equivalence. These fundamentally divergent initial microbial architectures (a solid grain matrix versus a liquid planktonic starter) naturally lead to the substantial differences in viable counts observed during fermentation. Under standardized external conditions; the resulting functional differences realistically reflect the systemic outputs of these distinct industrial modalities; representing the inherent physiological manifestations of each fermentation strategy. Beyond these single-batch systemic outputs; the long-term dynamic equilibrium of these two methods also diverges significantly. Previous investigations have demonstrated that bacterial and fungal communities in kefir grains from diverse sources remain relatively stable during 2–4 months of continuous subculturing (Wang et al., 2020). However; in BSF; each successive fermentation cycle alters the microbial composition of kefir; disrupting its dynamic equilibrium (Garofalo et al., 2020; Kim et al., 2018) The transition from KGF to BSF resulted in increased viable counts of LAB, decreased yeast populations, and reduced pH values—findings concordant with other published studies (Garofalo et al., 2020; Kim et al., 2018). These observations may be attributed to differential intra-community interactions and lactose metabolic activities between solid-phase (kefir grains) and liquid-phase (milk) environments (Garofalo et al., 2020). While our viable count data effectively reflect these broad population shifts; the lack of comprehensive microbial profiling (such as 16S rRNA and ITS sequencing) represents a limitation of the current study. Precisely characterizing the specific taxonomic alterations at the genus and species levels remains an essential step to mechanistically validate these structural changes. Generally; pH reduction in kefir correlates primarily with elevated LAB populations. Higher pH values are considered favorable for maintaining desirable sensory attributes; including flavor profile and mouthfeel; throughout the product shelf life (Gao & Zhang, 2019). Commercial fermented dairy products typically exhibit pH values ranging from 3.9 to 4.2; with consumers generally rejecting products with lower pH. A pH of 4.3 is recognized as one of the optimal chemical characteristics for kefir (Gao & Zhang, 2019). In the present study, kefir fermented using the KGF method demonstrated pH characteristics more likely to be accepted by consumers.

Excessive energy absorption by the body constitutes a primary etiological factor in obesity. α-Glucosidase, a hydrolytic enzyme involved in starch and glycogen metabolism, catalyzes reactions that lead to postprandial hyperglycemia. Inhibition of α-glucosidase activity attenuates the rate of starch degradation into glucose, thereby reducing and delaying intestinal glucose absorption to regulate energy intake (Sharma et al., 2023). Pancreatic lipase; a triacylglycerol acylhydrolase synthesized and secreted by the pancreas; plays a pivotal role in triglyceride digestion; and its inhibition contributes to the modulation of lipid metabolism. Metabolites produced by probiotics; including exopolysaccharides and bioactive peptides (Muganga et al., 2015); have been demonstrated to inhibit α-glucosidase or pancreatic lipase activity; thereby exerting anti-obesity potential. E. cloacae is recognized as a pathogenic bacterium strongly associated with obesity; capable of producing endotoxins in the intestinal tract and inducing obesity in germ-free mice (Gu et al., 2022). In the present study; kefir cell-free supernatants exhibited α-glucosidase inhibition rates ranging from 64.80% to 91.63%; pancreatic lipase inhibition rates of 41.92% to 50.95%; and demonstrable inhibitory activity against E. cloacae. The fermentation method exerted no significant influence on α-glucosidase or pancreatic lipase inhibition rates in most experimental groups. However; supernatants prepared via BSF predominantly lacked inhibitory activity against E. cloacae. Previous investigations have reported that cell-free fermentation supernatants from 28 LAB strains exhibited α-glucosidase inhibition rates exceeding 30% (Wei et al., 2022); while another study examining viable-cell-containing fermentation broths demonstrated α-glucosidase inhibition ranging from 12.19% to 99.16% (Tan et al., 2025). Compared with single-strain fermentation products; kefir; as a multi-microbial consortium; may confer advantages in α-glucosidase inhibition. The capacity to suppress E. cloacae growth represents a widely employed metric for evaluating the anti-obesity potential of probiotics (Li et al., 2023); with multiple studies confirming that both LAB cultures and fermentation supernatants possess such inhibitory properties (Li et al., 2023; Tan et al., 2025). The acidogenic capacity of LAB is conventionally regarded as the principal mechanism underlying pathogen growth inhibition. In the current study, kefir produced via BSF exhibited lower pH values (Fig. 1i). However, its E. cloacae inhibitory capacity remained inferior to that of KGF (Fig. 2c). This discrepancy may be attributed to the production of specific antimicrobial compounds during KGF, such as bacteriocins, bioactive peptides, and exopolysaccharides (specifically kefiran) (González-Orozco et al., 2022; Hasheminya & Dehghannya, 2020), or potentially the synergistic interplay of these agents (Gut et al., 2022). Further experimental validation is required to substantiate this hypothesis. Reactive oxygen species generated through oxidative stress represent a significant contributing factor to obesity and other chronic metabolic diseases; consequently; antioxidant activity has emerged as a critical parameter for assessing the anti-obesity properties of fermented foods (Hashemi et al., 2021). The present investigation evaluated three antioxidant indices of kefir cell-free supernatants: DPPH eliminating ability; total antioxidant capacity (T-AOC assay); and lipid peroxidation inhibition. Overall trends demonstrated that KGF surpassed BSF across all three antioxidant parameters; with particularly pronounced superiority in DPPH radical scavenging activity. Notably; kefir fermentation supernatants did not invariably exhibit lipid peroxidation inhibitory effects; certain BSF groups even demonstrated pro-oxidant activity; potentially indicating the generation of hydroxyl radicals or other reactive oxygen species that promote lipid peroxidation (Suryani et al., 2024).

The beneficial effects of kefir on human health derive from two principal components: the non-microbial fraction (supernatant) containing diverse bioactive metabolites, and the microbial consortium (solid fraction) possessing probiotic attributes. Previous investigations have demonstrated that only whole kefir containing viable microorganisms, as opposed to cell-free or heat-treated preparations, significantly reduces hepatic cholesterol and triglyceride levels in obese mice, thereby underscoring the indispensable role of the kefir microbial community in anti-obesity activity (Bourrie et al., 2022). Comprehensive characterization of the probiotic properties of kefir microbial communities facilitates understanding of how process conditions alter functional characteristics through modulation of microbial composition; and provides guidance for designing synthetic microbial consortia in industrial applications. Excessive dietary intake of cholesterol and triglycerides induces dyslipidemia; whereas reduction of their absorption contributes to the amelioration of obesity and associated metabolic syndrome (Zhang et al., 2019). For instance; LAB isolated from water kefir samples exhibited cholesterol degradation rates up to 69.9% and triglyceride degradation rates ranging from 37.24% to 59.27% (Li et al., 2023). In the study by Sharma et al. (2019); Lactobacillus plantarum demonstrated the highest cholesterol degradation rate of 52.27%. Research conducted by Wei et al. (2022) reported maximum cholesterol degradation rates of 52.27% for L. plantarum and 63.08% for Lactobacillus fermentum; with triglyceride degradation reaching 46.3%. In the present investigation; kefir microbial communities achieved cholesterol degradation rates up to 84.29% (Fig. 3a) and triglyceride degradation rates up to 62.52% (Fig. 3b); both exceeding values obtained with single-strain fermentations. The lipid-reducing capacity of microorganisms is conventionally attributed to assimilation; adsorption; and co-precipitation mechanisms (Wang et al., 2021). Kefir prepared via KGF has been shown to reduce body weight and plasma cholesterol levels in obese mice; whereas commercial kefir with lower microbial community richness failed to exhibit anti-obesity effects (Demir, 2020). Furthermore; removal of either yeasts or lactobacilli from the kefir microbial community resulted in loss of beneficial effects on cholesterol metabolism (Bourrie et al., 2021); indicating that synergistic interactions between LAB and yeasts play a critical role in lipid reduction. BSH activity represents another important mechanism whereby intestinal microorganisms hydrolyze conjugated bile salts into free bile salts; which subsequently co-precipitate with cholesterol; thereby reducing intestinal absorption and energy intake. In the current study; kefir microbial communities demonstrated hydrolytic activity toward taurine-conjugated bile salts but not glycine-conjugated bile salts (Fig. 3c). Overall; kefir microbial communities demonstrated anti-obesity potential through cholesterol assimilation; triglyceride reduction; and bile salt hydrolysis; though distinct fermentation methods yielded differential functional characteristics. KGF proved more conducive to cholesterol assimilation and BSH activity; whereas BSF exhibited superior performance in triglyceride reduction. Moreover; relative to conventional fermented dairy products such as yogurt; which is typically inoculated solely with Lactobacillus delbrueckii subsp. bulgaricus and Streptococcus thermophilus; kefir exhibits a markedly superior functional profile. While standard yogurt demonstrates baseline probiotic properties; kefir displays an enhanced therapeutic phenotype characterized by elevated antioxidant capacity; hypolipidemic effects; and stronger digestive enzyme inhibitory potential (Qaisrani et al., 2025; Uzkuç et al., 2025; Yilmaz et al., 2024). This augmented potency is primarily attributed to the intricate symbiotic synergy between the bacterial and yeast microbiota within kefir. Such a complex microbial consortium facilitates the synthesis of a diverse repertoire of bioactive peptides and kefiran, metabolites that are largely absent in conventional fermented dairy matrices. Consequently, kefir functions not merely as a probiotic delivery vehicle but as a distinct functional food matrix possessing superior metabolic regulatory capabilities. Specifically, regarding the fermentation methods, KGF proved more conducive to cholesterol assimilation and BSH activity, whereas BSF exhibited superior performance in triglyceride degradation. This functional divergence suggests potential for precision nutrition applications: KGF-based products may be more suitable for managing hypercholesterolemia, while BSF-fermented kefir could be specifically targeted toward populations with hypertriglyceridemia or those seeking to control dietary lipid absorption.

The tolerance of probiotics to extreme gastrointestinal environments, including gastric acid, bile salts, and digestive enzymes, constitutes a prerequisite for their biological functionality. Gastric fluid tolerance reflects the survival rate of probiotics during gastric transit, whereas intestinal fluid tolerance serves as a critical indicator of their capacity to exert beneficial effects in the intestinal tract. The gastrointestinal fluid tolerance of kefir microbial communities was significantly influenced by the source of kefir grains. SIF resulted in a substantial decline in LAB viability, primarily attributable to the bactericidal action of bile salts. Notably, the CN group fermented by the KGF maintained a survival rate of 13.21% after 6 h of exposure to simulated gastrointestinal fluids (Fig. 4a, b), significantly exceeding that of other groups. Furthermore, the pronounced viability of the kefir microbiome within SGF signifies robust acid resistance. This enhanced tolerance is primarily attributable to its intricate symbiotic consortium, which effectively potentiates the acid resilience of the constituent lactic acid bacteria (Oana et al., 2023). The adhesion capacity of probiotics to intestinal epithelial cell surfaces represents a pivotal criterion for their ultimate colonization potential. Huang et al. (2013) isolated L. plantarum strains from kefir grains that exhibited maximum adhesion rates to Caco-2 cells of 10.5%. Plessas et al. (2020) reported that Lactobacillus paracasei ARG4; isolated from kefir grains; demonstrated an adhesion rate of 6.1% ± 0.14% to HT-29 cells (The HT-29 human colorectal adenocarcinoma cell line). These findings are comparable to the adhesion rates observed for most kefir microbial communities in the present study (Fig. 4c). Under the KGF method; the RU and US groups achieved adhesion rates of 15.06% and 30.80%; respectively (Fig. 4c); significantly surpassing those of other groups. Evidence suggests that yeast populations within kefir microbial communities enhance the probiotic potential of LAB. In Caco-2 cell models; co-cultivation of yeasts with LAB improved the aggregation capacity of strain combinations and increased the adhesion rate of LAB (Xie et al., 2012).

Clinical studies have further confirmed that kefir fermented products prepared from kefir grains can be safely administered to critically ill patients to improve intestinal health (Gupta et al., 2024). Nevertheless; considering the variability in kefir grain sources and fermentation protocols; the final products may potentially harbor pathogenic bacteria (Cufaoglu & Erdinc, 2023); necessitating systematic safety evaluation. In this study; antibiotic susceptibility; hemolytic activity; and nitrate reductase activity were selected as safety assessment parameters. All kefir microbial communities exhibited no hemolytic activity (Fig. 4e). However; with respect to antibiotic resistance and nitrate reductase activity; KGF demonstrated superior safety profiles compared to the BSF. When evaluating the clinical relevance of antibiotic resistance in probiotic consortia; it is imperative to distinguish between intrinsic and acquired resistance (Li et al., 2020). The widespread resistance to vancomycin; ciprofloxacin; trimethoprim-sulfamethoxazole; and streptomycin observed in our study is recognized as intrinsic to most Lactobacillus species (Campedelli et al., 2019); posing minimal risk of horizontal gene transfer. Conversely; resistance to antibiotics such as tetracycline; clarithromycin; and ampicillin is typically acquired and transmissible. The microbiota prepared by BSF exhibited a higher frequency of resistant zones to these higher-risk antibiotics compared to KGF (Fig. 4d; Supplementary Table S4); suggesting that BSF may facilitate the survival of microbial subpopulations harboring acquired resistance determinants. This vulnerability of the BSF method is further corroborated by the nitrate reductase assay. While KGF samples were predominantly negative; BSF samples universally exhibited positive nitrate reductase activity. Nitrate reduction in dairy fermentations is primarily associated with opportunistic pathogens or spoilage organisms; such as Enterobacteriaceae (Cufaoglu & Erdinc, 2023). The liquid planktonic environment and continuous subculturing in BSF likely dilute antimicrobial metabolites specific to the grains (e.g.; kefiran and bacteriocins); creating a permissive niche that may favor the persistence and enrichment of trace opportunistic microbes capable of nitrate reduction over successive cycles (Garofalo et al., 2020). These specific microbes are postulated to be intrinsically present at subdominant levels within the initial complex consortium. In contrast, the solid grain matrix in KGF provides a highly competitive and protective microenvironment that effectively suppresses such contaminants. Consequently, while our community-level assays provide a valuable macroscopic evaluation, they cannot precisely attribute specific resistance or enzymatic traits to individual microbial species. Future investigations employing single-strain isolation coupled with whole-genome sequencing are necessary to precisely identify the genetic basis of these safety profiles. In summary, the fermentation method fundamentally shapes the safety profile of kefir, with KGF demonstrating markedly superior food safety and microbial control compared to BSF.

To comprehensively elucidate the influence of fermentation methods on the in vitro anti-obesity potential and probiotic characteristics of kefir, this study employed PCA, correlation analysis, and TOPSIS scoring to conduct multidimensional assessments of the functional properties of kefir produced under different fermentation conditions. The results indicated that kefir prepared by the BSF method exhibited superior performance in triglyceride degradation capacity. Conversely, the KGF outperformed the BSF across other in vitro obesity-related functional indicators, including cholesterol degradation rate, α-glucosidase inhibition rate, E. cloacae inhibitory capacity, DPPH eliminating ability, total antioxidant capacity, and lipid peroxidation inhibition capacity, as well as probiotic characteristics such as gastrointestinal fluid tolerance and adhesion capacity of LAB. This distinct functional trade-off, where BSF favors triglyceride reduction but lacks the robust pathogen inhibition of KGF, is fundamentally rooted in their divergent microbial microenvironments. The superior triglyceride degradation in BSF is likely driven by its planktonic state and significantly higher LAB biomass, which facilitates efficient lipid assimilation. In contrast, the strong inhibition of E. cloacae in KGF relies on the solid grain matrix, which serves as a protective niche for the localized accumulation of grain-specific antimicrobial compounds that are otherwise diluted or diminished during the continuous liquid sub-culturing of BSF. Additionally, samples produced by KGF harbored higher yeast populations, whereas BSF samples contained more abundant LAB. Integrating the TOPSIS comprehensive scores based on probiotic and anti-obesity characteristics (Fig. 5c, d), KGF demonstrated greater potential for producing probiotic kefir products that align with consumer health expectations.

5. Conclusions

This study systematically compared KGF and BSF across kefir grains from eight geographic origins. The results indicated that the fermentation method exerted a stronger influence than geographic origin on kefir's functional and probiotic properties. In general, KGF produced kefir with superior in vitro anti-obesity potential (e.g., cholesterol degradation up to 84.29%, and α-glucosidase inhibition up to 91.63%) and stronger probiotic characteristics (e.g., Caco-2 cell adhesion up to 30.80%, enhanced bile salt hydrolase activity, and safer profiles). Conversely, BSF favored lactic acid bacteria proliferation (up to 8.78 log CFU/mL) and triglyceride reduction (up to 62.52%). Building upon our in vitro findings, future research should integrate multi-omics approaches with in vivo trials to comprehensively evaluate the safety profiles and decode the systemic mechanisms of kefir's health benefits. Ultimately, this study provides a scientific basis for optimizing industrial kefir production, highlighting KGF as the recommended strategy for developing high-quality functional dairy products.

CRediT authorship contribution statement

Qingyang Ji: Writing – original draft, Project administration, Investigation, Formal analysis, Conceptualization. Hengxian Qu: Writing – review & editing, Supervision. Qiming Li: Supervision, Project administration, Funding acquisition. Ruixia Gu: Writing – review & editing, Project administration, Funding acquisition, Conceptualization.

Funding sources

This work was supported by the Sichuan Province Science and Technology Plan [2023YFN0101], Sichuan S&T Achievements Transformation Demonstration Project [2024ZHCG0078], and Key Laboratory of Probiotics and Dairy Deep Processing of Yangzhou [YZ2020265].

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary material

Data availability

No data was used for the research described in the article.