Tolerance to artificial gastric and intestinal fluids was determined following the method described by Liu et al. (28) with some modifications. The artificial gastric juice was prepared using 3 mg/mL pepsin dissolved in sterile isotonic phosphate-buffered saline (pH 7.4) adjusted to pH 2.5 with 0.1 mol/L hydrochloric acid. The resultant artificial gastric fluid was subsequently filtered via a 0.22-μm membrane (China Hefei Biosharp Biotechnology Co., Ltd.) to ensure sterilization before utilization. The artificial intestinal fluid was prepared with 1.0 g/L of trypsin (Shanghai Yuanye Biological Technology Co., Ltd., Shanghai, China) and 1.8% bile salt (Sinopharm Chemical Reagent Co., Ltd., Shanghai, China) dissolved in sterile PBS and adjusted to pH 8.0 with 0.1 mol/L sodium hydroxide. The resultant artificial intestinal fluid was filtrated using a 0.22-μm membrane filter (China Hefei Biochemical Co., Ltd.) for sterilization before use.

For use, the bacterial pellets obtained from activated selected LAB strains or B. subtilis after centrifugation (5,000g, 10 min, 4°C), were rinsed twice and re-suspended in 5 mL sterile phosphate-buffered saline (PBS) (pH 7.4). For resistance to artificial gastric juice, 0.5 mL containing 8.02 Log CFU/mL counts of each bacterial culture was added to 4.5 mL of artificial gastric fluid (pH 2.5) and incubated at 37°C under anaerobic conditions. Samples were collected at 0 and 3 h, respectively, and the live bacteria were enumerated on the respective culture plates. After the 3 h in the artificial gastric juice, 0.5 mL of each sample was transferred into artificial intestinal juice (4.5 mL), followed by incubation (37°C) under anaerobic conditions. The viable cell counts of the bacteria was determined after 4 and 8 h. The results were reported as colony-forming units per milliliter (Log CFU/mL) from the viable cell count.

The thermotolerant capacity of the selected LAB strains and B. subtilis was determined according to methods by Pérez-Chabela et al. (29) and Tu et al. (30), respectively, with some modifications. Each strain was cultured in 5 mL MRS broth and incubated at 37°C for 18 h. The tubes were immersed vertically in a water bath at three different temperatures, including 45°C to simulate conditions such as sun-drying, warm fermentation or prolonged storage in tropical climates, 85°C to simulate mild thermal processing, and 100°C to simulate boiling, canning, or retort sterilization, for 10 min. After the different thermal treatments, the viable cell counts were determined by serially dilution and plating the appropriate dilutions on MRS agar for LAB strains and LB agar for B. subtilis for 24–48 h at 37°C.

Esterase activity was determined using the method described by Gao et al. (32) with some modifications. Briefly, sample extract (200 μL) was mixed with 80 μL of 25 mM p-NP-butyrate in 1.72 mL of 0.1 M citrate buffer (pH 5.0) and incubated at 37°C for 60 min. Thereafter, 0.2 mL of 0.5 M NaOH was added to terminate the reaction. The p-nitrophenol (p-NP) released was measured at 410 nm using a UV–visible spectrophotometer (Harbor City, CA, USA) compared to a blank prepared without sample extracts. One unit of enzyme activity was defined as amount of enzyme required to release l μmol p-NP from p-NP-butyrate under the above conditions.

β-glucosidase activity was evaluated by quantifying the amount of p-NP liberated from 4-nitrophenyl β-D-glucopyranoside (p-NPG) as substrate in a method described by Liang et al. (33) with some modifications. Briefly, to 100 μL of sample extract, 1.8 mL of 0.05 M sodium acetate buffer (pH 5) was added and incubated at 37°C for 5 min, after which 100 μL of substrate p-NPG was added and incubated at 37°C for 10 min. The reaction was terminated by adding 1 mL of 1 M Na₂CO₃, and the optical density at 400 nm was recorded. One enzyme unit was defined as amount of enzyme required to catalyze production of 1 μmol of p-NP per minute under the above conditions.

The in vitro GAD enzyme activity was determined by directly measuring the GABA produced after GAD reaction as described by Lü et al. (34) with some modifications. Briefly, 1 g of sample extract was incubated with 5 mL of reaction mixture (containing 50 mM sodium phosphate (pH 5.8), 100 mM L-glutamate, 0.2 mM pyridoxal 5′-phosphate) at 40°C for 60 min, then the reaction was terminated in a water bath at 90°C for 5 min. After centrifugation (3,000g, 10 min), it was derivatized with OPA-MCE solution and filtered, and the GABA in the supernatant was assayed. One unit of GAD activity was defined as 1 mM of GABA produced from glutamate per gram of sample extract.

The enzymatic activity of the protease was measured indirectly by the colorimetric reaction of the protease decomposition of casein (substrate) to produce phenolic amino acids (35). One milliliter of 1% tyrosine solution and 1 mL supernatant were preheated for 5 min at 40°C and then reacted in a test tube in a water bath at 40°C for 10 min. Two milliliters of 0.4 mol/L trichloroacetic acid were immediately added successively. It stood for 10 min and was filtered with a slow qualitative filter paper. One milliliter of the filtrate was successively added with 5 mL Na₂CO₃ solution and 1 mL Folin–Ciocalteu reagent, and the absorbance value was measured at 680 nm after 20 min in a water bath at 40°C. Total protease activity was expressed in units (U/mL).

The activity of ascorbate oxidase was measured using the method described by Leong et al. (36). The reaction assay was a mixture of 2,500 μL of sodium phosphate buffer (0.1 M, pH 6.0) containing 0.5 mM EDTA, 400 μL of sample extract, and 100 μL of L-AA substrate solution (0.5 mM) L-AA dissolved in sodium phosphate buffer (0.1 M, pH 5.6) containing 0.5 mM EDTA. The absorbance was measured in a spectrophotometer at 265 nm (ε265 = 14.3 mM⁻¹ cm⁻¹). The blank solution was a mixture of 2,600 μL sodium phosphate buffer (0.1 M, pH 6.0), containing 0.5 mM EDTA, and 400 μL of carrot extract. One unit of AAO activity was defined as the amount of enzyme AAO that catalyzed the oxidation of 1 μmol of L-AA to DHAA per minute at 25°C and pH 6.0.

Vitamin C content was determined following a method described by Uriho et al. (37) with some modifications. Briefly, 20 mL of 4.5% metaphosphoric acid was mixed with 5 g of samples, then centrifuged (6,000g, 15 min, 10°C), and filtered through a 0.22 μm syringe filter. Ten microliter of the supernatant was injected into the HPLC system using a mobile phase consisting of 3, 30 and 970 mL of formic acid, methanol and water, at a flow rate of 1 mL/min, using T3 column (4.6 × 250 mm, 5 μm, Waters, USA), and a 2,998 Photodiode Array (PDA) detector to detect light at a wavelength of 245 nm. The column was maintained at a constant temperature of 35°C.

The TPC was determined in methanol sample extracts using the Folin–Ciocalteu method as described by Omedi et al. (38) with some modifications. Briefly, 5 mL of distilled water, 0.2 mL of methanol sample extracts, and 0.5 mL of Folin–Ciocalteu reagent were combined. After 3 min, 1.0 mL of saturated Na₂CO₃ solution (37%, w/v) was introduced, and the mixture was stirred. The volume was modified using distilled water to 10 mL, and the flask was placed in darkness for 1 h at ambient temperature. Absorbance was recorded at 760 nm using a blank of distilled water in place of the methanol extract. The TPC was determined using a standard curve of gallic acid (0–50 μg/mL) as a reference. Results were presented as milligrams of gallic acid equivalents per gram of sample (mg GAE/g).

The TFC was determined by the AlCl₃–NaNO₂ colorimetric method as described by Zhang et al. (39). Briefly, 150 μL of methanol extracts was reacted with 1 mL NaNO₂ (5%, m/v) for 3 min. Thereafter, 1 mL of 0.27 mol/L aluminum nitrate solution, then 4 mL of 1 mol/L NaOH solution was added into the mixture. Absorbance at 502 nm was recorded after 15 min of reaction at ambient temperature. Rutin was used as the standard. The TFC was expressed as mg of quercetin equivalents (QE) per gram of sample (mg QE/g).

The FAA content in the fermented Cili fruit was determined by chromatographic methods as described by Omedi et al. (40) with some modifications. Briefly, 1 g of sample was added to 25 mL of 5% (w/v) TCA and ultrasound treated for 20 min, followed by centrifugation (15,000g, 30 min, 4°C). The supernatants were then filtered using a 0.22 μm filter before injection into HPLC for analysis. The amino acid profiles were analyzed using liquid chromatography (Ag 1100, Agilent America) with UV detection following OPA/FMOC pre-column derivatization. Chromatographic separation was conducted using the ODS HYPERSIL column (5 μm, 250 mm × 4.6 mm) from Agilent America. The mobile phase A comprised 6.5 g of crystalline sodium acetate dissolved in 1,000 mL of water, to which 225 μL of triethylamine was incorporated and mixed. Acetic acid (5%) was drop-wise added to achieve a pH of 7.2, followed by adding 5 mL of butylene oxide, and the solution was subsequently mixed. Mobile phase B: 6.5 g of sodium acetate crystals were dissolved in 400 mL of water, and 5% acetic acid was drop-wise added to achieve a pH of 7.2. The solution was thoroughly combined with 800 mL of acetonitrile and 800 mL of methanol. The elution gradient for amino acid analysis was as follows: at 0 min, B% was 8; at 20.1 min, B% was 50; and at 24 min, B% was 100. The flow rate was 1.0 mL/min, utilizing a DAD (diode array detector) UV detector with wavelengths ranging from 338 nm to 262 nm (Pro, Hypro), with a column temperature set at 40°C. Amino acids were quantified using an external standard method.

The DPPH radical scavenging capacity was evaluated following the method described by Zhao et al. (41) with some modifications. Briefly, 100 μL of methanolic extracts were added to 1.9 mL of 186 μmol/L DPPH. The mixture was allowed to stand for 30 min at ambient temperature. Absorbance of the mixture was measured using a UV–visible spectrophotometer (UV-1800, Shimadzu, Kyoto, Japan) at a wavelength of 517 nm.

The ABTS radical scavenging activity was assessed following the method described by Yin et al. (42) with some modifications. Briefly, a solution comprising 7.4 mM ABTS and 2.6 mM K₂S₂O₈ was prepared by combining the two substances without light for 12 h. The ABTS + solution was diluted and mixed with either the test sample or buffer at a volume ratio of 7:1. The resultant mixture was kept in darkness for 6 min, after which absorbance was measured at wavelength of 729 nm.

The FRAP was conducted following the method described by Ekumah et al. (43) with some modifications. Briefly, 3.8 mL FRAP reagent (mixture of 10 parts 300 mM sodium acetate buffer (pH 3.6), 1-part 10 mM TPTZ (2,4,6-tripyridyl-s-triazine) in 40 mM HCl and 1-part 20 mM FeCl₃·6H₂O) was added to 0.2 mL test sample. The resulting solution was incubated for 30 min at 37°C, and the absorbance was measured at 593 nm using a UV–vis spectrophotometer (UV-1800, Shimadzu, Kyoto, Japan). The results were expressed in milligram equivalents of FeSO₄ per mL of sample using a FeSO₄ standard curve generated under the same conditions.

The alcohol dehydrogenase (ADH) activity was determined following the Valle & Hoch method (44). Briefly, 1.5 mL pyrophosphate buffer (0.1 M, pH 8.8), 0.1 mL of 0.25 U/mL ADH, 0.5 mL ethanol (11.5%, v/v), and 0.1 mL of fermented Cili fruit sample were mixed at 25°C, followed by addition of 1.0 mL 0.01 M NAD + to initiate the reaction. Absorbance was determined at 340 nm by a microplate reader and was measured again after 15 min.

The hydroxyl scavenging activity was determined using the method described by Zhou et al. (45) with some modifications. For the color reaction of ferric sulfate and salicylic acid, 1 mL 9 mmol/L FeSO₄ solution was mixed with 1 mL 9 mmol/L salicylic acid ethanol solution, and then 8.8 mmol/L H₂O₂ solution (0.1 mL) was added. After shaking, the sample solution (0.1 mL) was added, and absorbance was measured at 510 nm.

The changes in pH and TTA values during co-culture fermentation of Cili fruit was shown in Figures 2A,B. Before fermentation, the pH and TTA values of Cili fruit was 5.24 and 7.95 mL, respectively. In all the co-culture fermented Cili fruits, the pH gradually decreased (except for BsLf at 18 h, where it increased, then decreased thereafter), while the TTA values increased with increase in fermentation time. The increase in pH during the initial 18 h of fermentation in BsLF, was likely due to the higher concentration of MSG (4.68%) added to the medium. In published literature, MSG can act as an alkalinizing agent (glutamate ions bind some free H⁺ ions) which may influence pH dynamics in fermentation systems (55). The subsequent decline in pH throughout the extended fermentation period is likely a consequence of the organic acid production by L. fermentum. Moreover, compared to Bs sample, significantly lower pH (except BsLf) and higher TTA values were observed during co-culture Cili fruit fermentation. This trend was in agreement with the changes in pH and TTA in several LAB fermented substrate systems (38, 40, 41, 52, 56). In LAB fermented substrates, pH altering metabolites such as organic acids are released during fermentation (57). Therefore, the changes increased acidity due to co-culture fermentation of Cili fruit may positively enhance the release of key enzymes such as esterase, β-glucosidase and GAD, and enhance the bioactive composition, in terms of phytochemicals, soluble fiber and GABA of the end product.

The changes in the activities of esterase, β-glucosidase, GAD, protease and ascorbate oxidase enzymes during the co-culture fermentation of Cili fruit was presented in Figures 3A–E. The results revealed that all the three enzyme activities were detected during the fermentation of Cili fruit: an indication that the microbial starters used adequately adapted and synthesized enzymes as processing aids which can bio-transform substrate components into several bioactive metabolites of functional value after fermentation (58, 59). For esterase enzyme, its activity was 0.60 U/mL for all samples at the start of fermentation, but gradually increased with increased fermentation time to 0.79 U/mL for BsLb, 0.80 U/mL for BsLp and BsLf (Figure 3A). However, compared to Bs, the esterase activity was generally higher in the co-culture fermented Cili fruit. In co-culture fermentations of fruit substrates, the use LAB strain starters was found to enhance the esterase activity during fermentation (60). Due to its mode of action (breakdown of ester bonds, releasing fatty acids and alcohols), increased esterase activity is associated with the release of diverse volatile flavor metabolites (32). For β-glucosidase enzyme, its activity gradually increased with increased fermentation time: the highest increase was observed in BsLf (+107.28%), BsLb (+78.52%), BsLp (+74.16%), and Bs (+8.11%) at the end start of fermentation (Figure 3B). In several studies, enhanced β-glucosidase activities resulted in increased bioactive metabolite contents and functional properties in these substrates (61–64). Similarly, the GAD enzyme activity increased during fermentation of Cili fruit, with the highest observed in BsLf (+583.71%), BsLp (+313.18%), BsLb (+271.57%), and least in Bs (+254.67%) samples (Figure 3C). The changes in GAD activity during fermentation could potentially increase the GABA production in the Cili fruit (65). The protease activity increased with the increasing fermentation time (Figure 3D). However, compared with B. subtilis only, B. subtilis co- cultured with LAB exhibited the significantly higher protease activities. Specifically, protease activity increased from 1.04 U/mL to 1.28, 1.61, 1.65, and 1.73 U/mL, respectively, for Bs, BsLb, BsLp, and BsLf (Figure 3D). Our findings, along with a previous study by Ren et al. (66) underscore the potential of co-fermenting B. subtilis and LAB to enhance extracellular protease activity. Moreover, co-culture fermented Cili fruit have reduced the ascorbate oxidase enzyme activity (Figure 3E) and had a stabilizing effect on the vitamin C content of these samples. Furthermore, the ascorbate oxidase activity also showed a continuous increase throughout the fermentation process. However, in the co-cultured Cili fruit, the AAO activity was relatively lower (0.19, 0.25, and 0.18 U/mL for the co-cultures with BsLb, BsLp, and BsLf, respectively) compared to that observed in monoculture fermentation (0.3 U/mL) (Figure 3E). This was due to the production of organic acid by LAB, which leads to a reduction in pH and hence downregulates the activity of ascorbate oxidase in the fermentates (67, 68).

The results of the vitamin C content after co-culture fermentation were presented in Table 3. Before fermentation (NFC), the Vitamin C content was 2,081.74 mg/kg FW, which decreased in Bs (−67.64%), BsLb (−41.50%), BsLp (−64.43%), and BsLf (−53.86%) at the end of fermentation. These findings were in agreement with a recent study where the vitamin C and L-ascorbic acid contents in leaves significantly decreased after aerobic and anaerobic solid-state fermentation (69). The vitamin C decrease may be attributed to the increased activity of ascorbate oxidase enzyme produced by the microbial starters used during fermentation (70). On the other hand, relative to Bs (673.61 mg/kg FW), the vitamin C content was higher in co-culture fermented Cili fruit, especially in BsLb (+80.94%), BsLf (42.60%), and BsLp (+9.93%) (Table 3). This may be attributed to its sensitivity to oxidation (71), as the monoculture fermentation was carried out under aerobic conditions. Conversely, in the co-culture setup, lactic acid bacteria could inhibit the oxygen utilization of B. subtilis, creating an anaerobic environment that helps reduce the oxidation of vitamin C. In an earlier study, organic acids produced during LAB fermentation were found to stabilize the ascorbic acid content in Amaranthus paste after LAB fermentation (68). Hence, the higher acidification rate (Figures 2A,B) in co-culture fermented Cili fruit could have reduced the ascorbate oxidase enzyme activity and had a stabilizing effect on vitamin C content of these samples. The improved vitamin C content of the co-culture fermented Cili fruit may significantly enhance its functional properties (2).

The TPC and TFC content in the different samples was were presented in Table 3. The results showed that compared to NFC (TPC: 153.28 mg GAE/g DW, TFC: 125.51 mg QE/g DW), the TPC and TFC values significantly increased after fermentation. For TPC, the highest increase was found in BsLf (+79.52%), then BsLb (+69.53%), BsLp (+67.02%) and Bs (+63.86%), while the TFC was highest in BsLb (+37.30%), then BsLp (+23.68%), BsLf (+11.97%), and Bs (+5.93%) (Table 3). The increase in TPC and TFC observed in this study was in agreement with several studies where LAB fermentation consistently increased the phytochemical content (TPC, TFC) of the respective substrates (56, 72, 73). Moreover, the TPC and TFC content of lentils significantly increased after co-fermentation Lactobacillus spp. and Bacillus subtilis (74). As shown in Figure 3, the increased activity of esterase (Figure 3A) and β-glucosidase (Figure 3B) enzymes during Cili fruit fermentation could have facilitated the hydrolysis and release of bound phytochemicals, thereby, increasing the TPC and TFC content after fermentation (75). Increased content of phytochemicals in substrates are positively associated with improved antioxidant properties (2).

The composition of 17 FAA in the co-culture fermented Cili fruit was presented in Table 4. The results showed that with the exception of Arg, Try, and Cyst, the content of the remaining 14 FAA generally increased in fermented Cili fruit samples. Subsequently, the total FAA content was highest in BsLf (6,191.39 mg/kg FW), Bs (2,423.87 mg/kg FW), BsLb (2,278.53 mg/Kg FW) and BsLp (2,278.37 mg/kg FW) relative to NFC (1,124.59 mg/kg FW). In many published studies, FAA content has been reported in single and/or mixed LAB strain fermented substrates (76). Similar to the increased GABA content (Figure 1A) due to optimal pH for GAD activity (Figure 3C), the pH during and after fermentation could have favored synthesis of other strain-specific proteolytic enzymes, which degraded proteins and peptides in the Cili fruit into free amino acids during and/or after substrate fermentation (77).