Introduction

Lactic acid bacteria (LAB) play a crucial role in the food industry, where they primarily ferment carbohydrates into lactic acid for food preservation purposes. LAB also break down proteins and fats, yielding organic acids, amino acids, and flavor compounds that enhance the quality, safety, and nutritional value of food (Yang et al., 2024). LAB have traditionally been used for natural leavening, improving fermentation control, enhancing product quality, and extending shelf life while conserving energy (Aguirre-Garcia et al., 2024). The rise in demand of fermented foods is linked to increased health awareness and decrease in disease occurrence (Zdziobek et al., 2023). Approximately 20 strains of LAB, considered generally recognized as safe (GRAS), are used in fermentation because they play a vital role in medical and food applications by modulating gut microbiota and improving metabolic health, thereby shaping the future of functional foods (Yang et al., 2024). They produce organic acids that extend shelf life of bread, inhibit spoilage, and improve various properties, including rheology, mineral bioavailability, protein digestibility, flavor, and glycemic index (Șerban et al., 2021; Woo et al., 2020). Lactiplantibacillus plantarum, including the strain ATCC 8014, can effectively ferment various cereal substrates, including wheat, barley, sorghum, millet, and rice, making it a widely used LAB in production of non-dairy food products. It provides significant technological and nutritional benefits, converting cereals into more nutritious, safer, and tastier food products (Arya et al., 2025; Paventi et al., 2024). It also produces bioactive compounds, contributing to the preservation of food and human health (Cui et al., 2021; Șerban et al., 2023).

Millet is an essential, gluten-free cereal crop mainly grown in Asia and Africa. It originated in Africa and is cultivated in Saudi Arabia’s Jazan region (El-Hashash et al., 2023; Food and Agriculture Organization [FAO], 2021). Nutritionally, it is a rich source of calories, dietary fiber, protein, vitamins, and minerals. It is also valued for its beneficial fatty acids and bioactive compounds, such as polyphenols, making it a popular functional food ingredient (Amadou, 2022; Goudar et al., 2023). While pearl millet offers a good amount of calories, it is also gluten-free and rich in dietary fiber, non-starch polysaccharides, polyphenols, proteins, fatty acids, minerals—including higher levels of iron, zinc, and calcium—and vitamins, such as vitamin E, riboflavin, thiamine, and niacin. It also contains significant amounts of unsaturated fatty acids such as linoleic acid, linolenic acid, and oleic acid as well as saturated fatty acids, such as palmitic acid and stearic acid (Amadou, 2022). Additionally, pearl millet is a good source of bioactive compounds, such as phenolics, flavonoids, total anthocyanins, and -polymeric tannins (Goudar et al., 2023).

Fermentation is the most cost-effective and energy--efficient method for improving food preservation, and various changes occur in the nutritional and functional properties of grains (Balli et al., 2023; Suma and Urooj, 2017). Fermentation is performed using specific starter cultures (controlled fermentation), intrinsic microorganisms, or spontaneous processes (Balli et al., 2023). Compared to traditional fermentation, immobilized bacteria fermentation helps to control the process and enhance yield (Yang et al., 2024). Utilizing specific strains with desired properties as starter cultures in controlled fermentation is a cost-effective and straightforward approach to improve the nutritional value and health benefits of cereal substrates (Sidari et al., 2020; Yépez et al., 2019). Sour dough/paste, a fermented mixture of flour, water, and LAB, is popular among bakers for its ability to improve bread quality (Șerban et al., 2023). Moreover, fermentation produces various metabolites, including antioxidants and vitamins, which enhance their bioactivity or introduce new exogenous bioactive compounds, thereby increasing the bioactivity of fermented flour and providing health benefits (Melini et al., 2019). Furthermore, millet fermentation enhances the availability of calcium, iron, phosphorus, and zinc (Yousaf et al., 2021).

In fermentation, using an inoculum of single or mixed microorganisms, the culture medium must be sterilized. Therefore, autoclaving, a wet-heat sterilization method, effectively eliminates microorganisms by using pressurized steam, making it much more effective than dry-heat sterilization (Agrawal, 2024; Jayashree et al., 2024). However, sterilization can enhance the fermentability of solid substrates by modifying nutrients such as polysaccharides, thereby improving solid-state fermentation (Zhao et al., 2015). Nevertheless, employing a high--temperature, short-duration method during sterilization may lessen nutritional value while simultaneously eradicating bacteria (Mann et al., 2001). Accordingly, a comprehensive understanding of how L. plantarum interacts with the components of millet flour during fermentation, especially following various autoclaving methods used to sterilize paste mix, is necessary (Șerban et al., 2023). However, the importance of pearl millet lies in its natural composition, which provides a nutrient-rich environment that LAB utilizes, promoting strong microbial growth and activity (Adebo et al., 2022; Mudau and Adebo, 2025). L. plantarum-fermented pearl millet is shown to improve antioxidant capacity, bioactive compounds, and nutrient content, positioning it as a promising functional food for nutritional enhancement. Research indicates that its enhanced properties can lead to health benefits, including anti-inflammatory effects (Gabriele et al., 2024; Srivastava et al., 2024). Fermented pearl millet offers a cost-effective, health-promoting solution for communities that rely on millet as a staple food (Srivastava et al., 2021). On the other hand, autoclaving used to sterilize pearl millet slurry before inoculation with L. plantarum is a crucial pre-processing step that significantly affects the substrate’s physicochemical properties by quickly releasing fermentable sugars; this impacts the bioaccessibility of compounds metabolized by L. plantarum, thereby influencing the efficiency of fermentation process (Cui et al., 2021; Zheng et al., 2023). This is reflected in enhanced starch and carbohydrate digestibility, reduced antinutritional factors, improved mineral accessibility, and improved microbial safety and quality of the end fermented product (Chu et al., 2025; Yehuala et al., 2025). While extensive research exists on moist autoclaving treatments (i.e., autoclaving the mixture of flour and water) (Moiseenko et al., 2024; Ntsamo et al., 2020), understanding how different autoclaving methods affect the substrate and, in turn the fermentation process by L. plantarum is a significant knowledge gap. To address this, the novelty of this study lies in pretreating a millet flour–water mixture (1:4, w/v) or flour and water separately to autoclaving (121°C; 15 psi) for 15 min before fermentation. The first type of autoclaving is known as mixed-mode autoclaving (MMA), while the second is referred to as single-mode autoclaving (SMA). After cooling, the Lactiplantibacillus (LPB) inoculum was added aseptically. Therefore, this study aims to assess the impact of SMA and MMA treatments on the ability of fermenting L. plantarum to enhance the nutrient content, bioactive compounds, and antioxidant activity of millet flour.

Materials and Methods

Chemicals

Agar powder and de Man, Rogosa, and Sharpe (MRS) broth were obtained from EMDA International Trading Establishment (EITE; Riyadh, Saudi Arabia). The following supplies were purchased from Sigma-Aldrich (St. Louis, MO, USA): Folin–Ciocalteu reagent, gallic acid standard, quercetin standard, and tannic acid standard. 2,2-Diphenyl-1-picrylhydrazyl (DPPH) was purchased from Merck (Darmstadt, Germany).

Millet flour preparation

Millet grains were purchased from Al-Baha market in Saudi Arabia. The grains were cleaned of debris, washed, and left to dry at room temperature. They were then milled into powder, passed through a 60-mesh sieve, and stored at 4°C for further use.

Fermentation process

Preparation of L. plantarum ATCC 8014 inoculum

Lactiplantibacillus plantarum ATCC 8014 was procured from the microbiology laboratory of College of Food and Agriculture Sciences, King Saud University, Saudi Arabia. The stock strain of L. plantarum, preserved in 10 mL of MRS broth with glycerol at -80°C, was inoculated into MRS broth and incubated at 37°C for 48 h. Grown culture, 1 mL, containing approximately 10⁸ colony-forming unit (CFU)/mL, was inoculated into 500-mL sterilized conical flasks containing 100 mL of sterilized MRS broth and incubated at 37°C for 48 h. The cells of L. plantarum were centrifuged (2,300 ×g; 4°C; 10 min) and washed for three times with sterile water to remove any remaining medium. The total number of vegetative cells and spores was counted by using the colony count technique according to ISO 7932 (International Organization for Standardization [ISO], 2004). To enumerate L. plantarum, a serial dilution in sterile sodium chloride (0.85%) solution and the nutrient agar pour plate method were used; the plates were incubated at 37°C for 24 h. The CFU/mL was counted with a laser colony scanner model 500A.

Fermentation of millet flour

A 1-L Manifor fermentor (Lambda Laboratory Instruments, Sihlbruggstrasse 105, Switzerland), equipped with dissolved oxygen (DO), temperature, and pH probes, was used to ferment millet flour. L. plantarum served as an inoculum. The fermentor received a specified weight of millet flour, 490 mL of deionized water, and 10 mL of inoculum (L. plantarum, 10⁸ CFU/g) to create fermentation mixture. The millet flour–water ratio was 1:4. Before adding the inoculum (1 mL; 10⁸ CFU/mL of L. plantarum), the flour and water were sterilized separately in an autoclave (121.1°C; 15 psi; 15 min) in a process called SMA, or they were sterilized together, referred to as MMA, a commonly used autoclaving method for sterilization; thus, it was chosen to serve as a control. Subsequently, the fermentation process was conducted at a constant agitation speed of 120 rpm and 37°C for various durations (12, 24, 36, 48, 60, and 72 h). The DO level was maintained above a set saturation point (e.g., >10%) using a computer-controlled feedback system that adjusts agitation speed, airflow rate, and oxygen level in the incoming gas mixture based on real-time DO sensor data (Zheng and Pan, 2019). After completion of fermentation, the pH was measured, the sample was removed from fermentor, and then heated in a boiling water bath for 15 min to halt fermentation. The samples were dried in an oven at 50°C, packed in sample bags, and stored at -20°C for later analysis. For safety, we fermented millet flour samples from 6 h to 72 h, each in a separate conical flask, to eliminate the possible risk of contamination during withdrawing of samples at desired time intervals (Figure 1).

Figure 1. Overview of the pearl millet flour fermentation process using L. plantarum ATCC 8014. Notes. SMA: single-mode autoclaving; MMA: mixed-mode autoclaving. a: raw; b: raw millet flour sterilized with MMA; c: 12-h SMA-fermented millet; d: 24-h SMA-fermented millet; e: 36-h SMA-fermented millet; f: 48-h SMA-fermented millet; g: 60-h SMA-fermented millet; h: 72-h SMA-fermented millet; C: 12-h MMA-fermented millet; D: 24-h MMA-fermented millet; E: 36-h MMA-fermented millet; F: 48-h MMA-fermented millet; G: 60-h MMA-fermented millet; H: 60-h MMA-fermented millet.

Proximate composition analysis

The proximate composition of millet flours was determined using AOAC methods without modifications (AOAC, 2023). Moisture content was measured by drying overnight at 105°C (method 930.15), ash content was determined by incineration at 600°C (method No. 942.05), and fat content was estimated using the ether extraction Soxhlet method (No. 966.06). Total nitrogen percentage was determined by the Micro Kjeldahl acid digestion/distillation method (No. 984.13), and crude protein was calculated as N% × 6.25. Although a more accurate method exists for measuring protein from amino acids, it has not yet been validated by AOAC, leading to the continued use of the 6.25 conversion factor (Krul, 2019). The carbohydrate content was calculated using the difference method (Pehrsson et al., 2015).

Determination of contents of glucose and fructose

Glucose and fructose levels in fermented samples were determined by high-performance liquid chromatography with a refractive index detector (HPLC-RID; Shimadzu, Kyoto, Japan) using Kirsten’s method with some modifications (Weiß and Alt, 2017). The procedure involved mixing 5 mL of HPLC-grade water with 0.5 g of sample, vortexing for 1 h, and then filtering the mixture through a 0.45-µm cellulose acetate syringe filter. Standard stock solutions of glucose and fructose at different concentrations were prepared and used for external calibration. The HPLC system, equipped with an LC-NH2 column (150 × 4.6 mm, 5 µm), an RID-10A detector, an LC-10AB binary pump, and an SIL-20A autosampler (Shimadzu), was used to analyze fructose and glucose. Separation conditions were as follows: a mobile phase of 90% aqueous acetonitrile (HPLC grade) with an isocratic flow rate of 1 mL/min, a column temperature of 85°C, an injection volume of 1 μL (diluted 10-fold) for sample extracts and standards, and a 50-min run time. Additionally, a spiked sample was used to ensure the accuracy and quality of results.

Determination of nutritional and heavy metal levels

A mixture of 0.5 g of sample, 0.5 mL of HCl, and 9.5 mL of HNO₃ was digested in a microwave adjusted as follows: step 1: 50°C, 2 min, and 1,000 W; step 2: 30°C, 3 min, and 0 W; step 3: 189°C, 25 min, and 1,000 W; step 4: 150°C, 1 min, and 0 W; step 5: 180°C, 4 min, and 1,000 W; step 6: 180°C, 15 min, and 1,000 W (ETHOS UP, 2024). For analysis, the sample (diluted 50-fold with 2% HNO₃) was injected in an ICP-MS-2030. Key parameters included a sampling depth of 5 mm, an output power of 1.2 kW, a plasma gas flow rate of 9.0 L/min, an auxiliary gas flow rate of 1.10 L/min, a spray chamber temperature of 5°C, and a carrier gas flow rate of 0.70 L/min. Calibration curves were plotted using element standards, allowing the calculation of sample element concentrations, expressed in mg/kg (ppb × 0.001) for nutritional elements and µg/kg (ppb) for heavy toxic elements.

Determination of TPC, TFC, TTC, and DPPH scavenging activity

Preparation of methanol extract

A total of 1 g of fermented millet flour and 20 mL of 95% methanol were placed in a conical flask and wrapped in aluminum foil. The flask was shaken for 4 h and then filtered using Whatman filter paper No. 42, followed by filtration through a 0.2-µm membrane filter. The filtrate was packed in amber glass bottles and stored at 5°C until used. This filtrate was used to analyze TPC, TFC, TTC, and DPPH scavenging rate.

Total phenolic content

The TPC of millet methanol extract was determined according to the method reported by Mouhoubi et al. (2023). A volume of 100 µL of extract was mixed with 100 µL of Folin–Ciocalteu reagent. After 5 min, 300 µL of 20% sodium carbonate solution was added to the mixture. Next, the sample was incubated in a dark place for 30 min at ambient temperature. Then absorbance of the sample was measured at 765 nm using a ultraviolet (UV)-visible spectrophotometer (UV-3200; Shimadzu). The TPC was calculated using a linear regression equation: y = 0.0049x - 0.038 (R² = 0.9981), derived from a gallic acid standard curve created with varying concentrations (25–400 µg/mL). The TPC findings are expressed as mg gallic acid equivalent (GAE)/g dry weight (dw).

Total flavonoid content

The TFC of sample methanol extract was determined using the method described by Hussain et al. (2024, 2025). A 0.5-mL aliquot of the extract was mixed with the same volume of 2% aluminum chloride (AlCl₃) solution. After 1 h, the wavelength was read at 420 nm. The TFC was quantified using the regression equation: y = 0.0031x - 0.0294 (R² = 0.9975). This equation was derived from a quercetin calibration curve prepared using different concentrations ranging from 25 to 400 µg/mL. The TFC results are expressed as mg of quercetin equivalents (QE)/g of sample (mg QE/g dw).

Total tannin content

The TTC of methanol extract was determined by following the method described by Rodrigues et al. (2007), with minor changes. A total of 0.1 mL of extracted sample was added to a 2-mL Eppendorf tube containing 1.5 mL of milli-Q water and 0.1 mL of Folin–Ciocalteu phenol reagent for 8 min. Then, 0.3 mL of 35% Na₂CO₃ solution was added to the mixture for neutralization. Next, the mixture was shaken well and kept in the dark at room temperature for 20 min. Absorbance of the mixture was recorded at 700 nm. To determine TTC, a calibration curve was created using various concentrations of tannic acid standard, and the following equation was used: y = 0.005x - 0.016 (R² = 0.9992). The TTC results are expressed as mg tannic acid equivalents (TAE)/g dw.

DPPH radical scavenging rate

The methanol extract of fermented millet was evaluated for its antioxidant activity using the DPPH scavenging test, as described earlier (Alshammari et al., 2025). Briefly, 200 µL of fermented millet sample was mixed with 2 mL of 0.1-mM DPPH solution in ethanol. The mixture was vortexed and left in the dark at room temperature for 30 min before measuring its absorbance at 517 nm. Methanol was also evaluated for its DPPH scavenging rate as a blank. The DPPH scavenging rate of the extract was calculated using the following formula:

DPPH scavenging activity (%) = ([A₀ – A_s]/A₀) × 100,

where A₀ is the absorbance of the blank, and A_s is the absorbance of the sample.

Untargeted LC-MS/MS analysis of the composition of pearl millet flour

The analysis of the composition of SMA- and MMA-pretreated pearl millet, followed by fermentation for 72 h, was conducted using the liquid chromatography-tandem mass spectrometry (LC-MS/MS) system. The analytes were separated using a Phenomenex Luna C18 PS column (100 Å, 100 × 2.1 mm, 1.7-μm particle size) with gradient conditions involving water with 0.01% formic acid (solution A) and acetonitrile (solution B), both modified with 10-mM ammonium acetate, at a flow rate of 0.4 mL/min. The column oven was 40^oC, and the injection volume was 10 µL. Mass spectrometry analysis used the X500R Quadrupole time-of-flight (QTOF) system with electrospray ionization in positive mode and sequential window acquisition of all theoretical mass spectra (SWATH) acquisition. Time-of-flight mass spectrometry (TOF-MS) scans were conducted from 50 to 1,000 Da with specific voltage and gas parameters (Sun et al., 2017). For detection and identification, the LC-MS/MS data were processed using the MZmine 2.53 software (Boutet et al., 2022; Chambers et al., 2012).

Statistical analysis

The data were presented as means ± standard deviation (SD) from triplicate experiments. One-way analysis of variance (ANOVA) and two-way ANOVA were employed to identify significant differences between treatment groups. Tukey’s test (one-way ANOVA) and pairwise comparisons (two-way ANOVA) were used to assess significant differences between mean values (p < 0.05). IBM SPSS Statistics, version 28.0 (IBM Corporation, NY, USA), was used for statistical analysis.

Results and Discussion

Viability growth dynamics of L. plantarum ATCC 8014 during millet fermentation under various autoclaving regimes

Assessing the kinetic behavior of microorganisms is crucial for planning industrial production, understanding responses to environmental variables, and designing the most technologically efficient reactors (Üçok and Sert, 2020).

In Table 1, the growth data show the population changes of L. plantarum ATCC 8014 during a 72-h fermentation, using SMA- and MMA-pretreated millet flour as substrates. A control MRS substrate, simulating ideal laboratory conditions, was used as a baseline for comparison. All groups began with statistically similar, high inoculum levels (~1.00–1.03 × 10⁸ CFU/g), confirming a consistent starting point. First, a lag/log phase (lag/early exponential, 0–24) at which an adaptation period occurs, followed by exponential growth. It has been reported that the lag phase for L. plantarum in cereal dough generally happens between 3 h and 4 h under conditions of 30–37°C and pH 5.5–6.5, immediately followed by the late exponential (Log) phase, and then the death phase (Popova-Krumova et al., 2024; Śliżewska and Chlebicz-Wójcik, 2020).

In detail, the most notable observation is the superior early growth with SMA treatment. This is evident in the higher population count for SMA (3.53 × 10⁸) at 0–12 h, which increased by 249.5%, compared to the initial bacterial count (1.01 × 10⁸). In contrast, increments of 70% and 99.03% in bacterial count were observed for MRS control and MMA-fermented samples, respectively, during the same period. At 12–24 h of fermentation, compared to the initial inoculum count, the population of SMA-treated millet increased by 71%, followed by MMA-treated millet, which increased by 51%, and then MRS, with an increase of 42.2%.

Differences in the lag/early exponential phases observed in fermented millet substrates, particularly those treated with SMA, may result from variations as to how the flour matrix is affected. This difference probably results from the Maillard reaction and other heat-induced chemical changes. MMA (mixing flour and water during sterilization) may cause more extensive non-enzymatic Maillard reactions or starch gelatinization, which temporarily reduces nutrient bioavailability for fermenting bacteria, leading to a slightly more extended lag phase because the bacteria need to adapt to a more complex or nutrient-depleted environment (Pontonio et al., 2020; Xiang et al., 2019).

Moreover, SMA reduces direct contact and reactions between flour components and water at high temperatures. This helps to maintain the native structures of starches, sugars, and proteins, making them more accessible to enzymatic breakdown and to L. plantarum during the early stages of fermentation (Verni et al., 2019). Additionally, SMA pretreatment may enhance starch gelatinization compared to the MMA method, as MMA’s inconsistent hydration can result in partial gelatinization. SMA, by using pre-sterilized flour and sterile water, ensures a uniform gel that enhances the accessibility of glucose and dextrins (Chiodetti et al., 2024). This efficient starch hydrolysis is crucial for LAB growth (Pontonio et al., 2020).

Furthermore, at 24–48 h (late exponential phase), L. plantarum in all samples reveals rapid growth during 24–36 h of fermentation, increasing by 293.8% in MRS control, 256.5% in MMA-pretreated millet, and 106.94% in MMA-pretreated millet. The second part of the exponential phase (36–48 h) shows slower growth, as shown by 74.6%, 71.95%, and 51.64% increase in bacterial counts in MRS control, MMA-, and SMA-pretreated samples, respectively. Although L. plantarum grew faster in MRS control during the exponential phase, the bacterial counts of SMA- and MMA-pretreated fermented samples reached a peak of 1.9 × 10⁹ CFU/g, compared to 1.65 × 10⁹ CFU/g in the control. These findings aligned with the changes observed in glucose and fructose during the fermentation of millet samples (Figure 3).

Figure 2. Changes in pH in millet flour following L. plantarum ATCC 8014 fermentation. Note: 0r: raw millet flour; 0MMA: raw millet flour sterilized with MMA (mixed-mode autoclaving: sterilizing flour and water together; control); SMA: single-mode autoclaving.

Figure 3. Glucose and fructose contents in autoclave-treated pearl millet, followed by fermentation with L. plantarum ATCC 8014. Means ± SE (n = 3) were statistically analyzed using one-way ANOVA. According to Tukey’s test (p < 0.05), mean values with different superscript alphabets differ significantly. Total mean values (n = 24) were analyzed using a two-way ANOVA. According to pairwise comparison (p < 0.05), the total mean values marked with (*) or (^#) are statistically different. Notes: 0r: raw millet flour; 0MMA: raw millet flour sterilized with MMA; MMA: mixed-mode autoclaving: sterilizing flour and water together; control; SMA: single-mode autoclaving. Lowercase and uppercase alphabets indicate significant differences among the mean values of glucose and fructose in millet flour during fermentation of samples, respectively. Error bars indicate the standard error (SE) of mean values.

Finally, the stationary/death phase (48–72 h) exhibits a significant reduction in cell viability (41–46.6%) across all samples (48–60 h), with a more pronounced decrease noted after 60 h of fermentation (73–87.4%). As shown, bacterial death occurred more rapidly in millet samples compared to the control. This is probably due to environmental stressors, including the production of bacteriocins and organic acids, and a decrease in pH (Anumudu et al., 2024; Yoon et al., 2024).

Interestingly, despite decrease in counts at 72 h, L. plantarum levels in all fermented SMA- and MMA-treated millet samples remain above the original inoculum, with values of 2.21 × 10⁸ and 1.29 × 10⁸, respectively, compared to 2.63 × 10⁸ for the control MRS. However, higher bacterial death in MMA samples could be due to faster depletion of glucose and fructose than in SMA, consistent with the findings of simple sugar analysis (Figure 3; Supplementary Table S2). In general, L. plantarum exhibits higher acid tolerance than other LABs (Paramithiotis, 2025); therefore, the millet samples retained moderately viable cells after 72 h of fermentation. This is supported by the analysis of pH values of the studied samples (Figure 2).

Besides, in addition to lactic acid, LAB, including L. plantarum, can produce other antimicrobial compounds, including acetic acid, ethanol, and bacteriocins, under specific conditions. In the nutrient-depleted and acidic environment, metabolic pathways can undergo significant changes. The production of acetic acid, which is more toxic to microbial cells than lactic acid, might increase. Besides, L. plantarum is known to produce bacteriocins (e.g., plantaricin), which are protein-based compounds that inhibit the growth of closely related bacterial strains (Ismael et al., 2024; Zhao et al., 2025). While primarily targeting competitors, at high cell densities, these bacteriocins can also cause auto-inhibition, a phenomenon known as “chemical warfare,” where some cells undergo autolysis (self-digestion). The release of hydrolytic enzymes (autolysins) degrades the cell wall (Kawai et al., 2023), leading to the observed decrease in CFU. Nevertheless, autolysis, followed by apoptosis, is believed to be a survival strategy for the population, as lysed cells release intracellular nutrients that are absorbed by the remaining live cells, helping the culture survive longer (Allocati et al., 2015; Mohiuddin et al., 2021). A sharp decline between 60 h and 72 h is probably due to a combination of ongoing acid stress and substantial autolytic activity. The above data suggest that the sterilization method (SMA vs. MMA) had a slight but significant impact, especially during the early- to mid-fermentation phases.

In summary, the impressive growth potential of L. plantarum in both SMA- and MMA-pretreated millet, reaching ~1.0 × 10⁹ CFU/mL, demonstrates that millet flour is an effective substrate for LAB fermentation, supporting the trend toward plant-based fermented foods and functional components (Das et al., 2022).

Changes in pH of millet flour during L. plantarum fermentation

As shown in Figure 2, raw millet flour has a pH of 6.9, which decreases to 6.56 after MMA (i.e., sterilized raw millet flour, 0MMA), indicating the release of organic acids into the medium. Autoclaving can increase the acidity of cereal flour by releasing adhered phenolic acids through hydrolysis in the high-temperature and high-moisture environment (Rico et al., 2020; Solmaz et al., 2025). Fermentation of both SMA and MMA samples results in a gradual decrease of pH over time, reaching its lowest point at 72 h (pH = 3 and 4, respectively). The most rapid decline in pH occurred between 12 h and 36 h, corresponding to the period of the most vigorous bacterial growth and sugar acidification (Table 1). Lactic acid, a byproduct of sugar metabolism, accumulates during L. plantarum fermentation of cereals, lowering the medium’s pH (Adisa et al., 2024; Moiseenko et al., 2024). Additionally, it has been reported that autoclaving can render a more favorable substrate, thereby -enhancing LAB fermentation and lowering pH (Isembart et al., 2025).

It is also worth noting that fermented SMA-treated millet samples have lower pH values than fermented MMA-treated millet samples, indicating a higher presence of L. plantarum in SMA, compared to MMA. This is probably due to the greater effect of moist heat and pressure exerted by MMA on millet components, mainly starch, which may lead to the formation of resistant starch. It is discovered that autoclaving causes resistant starch type III to form via retrogradation, a process in which short-chain amylose molecules bond via hydrogen bonds after cooling (Chuwech et al., 2023). As observed previously, SMA and MMA treatments resulted in a decline in pH from 6.9 in the raw sample to 3.0 (SMA) and 4.0 (MMA) after 72 h of fermentation. However, it is reported that millet fermentation resulted in higher acidification with milder initial heat treatments (Janiszewska-Turak et al., 2024). Homofermentative L. plantarum primarily produces lactic acid, which aids in food preservation and flavor, but can inhibit bacterial growth at high concentrations (Popova-Krumova et al., 2024). A notable decrease in pH in SMA, compared to MMA, may be due to high accumulation of lactic acid, which enters bacterial cells and dissociates at neutral pH, releasing protons and lowering internal pH. Accordingly, cells expend energy to pump out protons via adenosine triphosphate (ATP)-dependent transporters to maintain homeostasis, which is called “uncoupling effect.” This effect and subsequent acidification lead to energy depletion as ATP is diverted, reducing enzyme activity and potentially causing cell death when pH balance is disrupted (Martinez et al., 2012; Sionek et al., 2024). Our data show that the population decreased by 41% between 48 h and 60 h, clearly indicating severe acid stress.

Overall, the MMA treatment produced less acid than the SMA treatment because its harsher processing created more resistant starch, which fermented less effectively. In contrast, the SMA treatment created a much more acidic environment, which was toxic to L. plantarum, leading to greater apoptosis than MMA by the end of fermentation.

Proximate composition of millet flour fermented with L. plantarum ATCC 8014

The proximate composition of L. plantarum-fermented millet flours subjected to different autoclaving methods (SMA and MMA) is detailed in Table 2 and Supplementary Table S1. The levels of ash, fat, protein, and carbohydrates in the raw millet flour were 1.89, 7.76, 6.19, and 84.23 g/100 g dw, respectively. Ash levels significantly increased to 2.25 g/100 g dw, while carbohydrate and protein levels remained unchanged at 85.07 and 5.68 g/100 g dw, respectively, in the sterilized raw millet flour (0MMA), compared to the untreated flour. Meanwhile, fat levels decreased to 7.00 g/100 g dw (p < 0.05) (Supplementary Table S1).

Regardless of the fermentation method, using two different autoclaving modes, SMA and MMA, prior to L. plantarum fermentation resulted in varying macronutrient contents. In SMA, the levels of fat and protein (total means of 5.96 and 6.95 g/100 g dw, respectively) were significantly higher (p < 0.05) than those in MMA. Conversely, ash and total carbohydrates were higher in MMA, with total mean values of 2.58 and 85.54 g/100 g dw, respectively (Table 2). The differences observed between SMA and MMA in changes in macronutrient suggested that they exhibit distinct autoclaving effects, reflected in their unique interactions with millet flour components. Previous studies reported an increase in total carbohydrate level of cereals after autoclaving (Faridah et al., 2022; Udensi et al., 2010).

Besides, regardless of the autoclaving mode used, the ash content increases significantly from a total mean of 2.25 g/100 g dw for the 0MMA sample as fermentation time progresses, reaching a total mean of 2.88 g/100 g dw at 48 h. It then decreases with further fermentation, reaching its minimum level at 72 h (a total mean of 2.25 g/100 g dw), and leveled off at the initial 0MMA value (p < 0.05). The fat content of 0MMA (7.0 g/100 g dw) is significantly higher than all fermented samples (12–72 h), with the 72-h sample showing the lowest value (p < 0.05). The protein content of the 0MMA sample increases significantly in all fermented samples, with the highest values observed at 24 h and 36 h (7.53 g/100 g dw and 7.22 g/100 g dw, respectively). Furthermore, the protein content of the remaining fermented samples does not differ significantly from that of the 36-h sample (p < 0.05). The carbohydrates content remains unaffected by fermentation for 36 h but decreases after 60 h, then increases after 72 h (the total mean of 88.29 g/100 g dw; p < 0.05) (Table 2).

As shown in Supplementary Table S1, both SMA and MMA demonstrate distinct behavior during L. plantarum fermentation. Specifically, L. plantarum fermentation has a more pronounced impact on MMA than on SMA, as indicated by the magnitude of changes in macronutrients. In particular, MMA shows higher increases/decreases in ash, fat, protein, and carbohydrates as fermentation progresses than SMA. This could result from the moist heat and pressure applied by MMA to millet components, which enhance the millet matrix’s suitability for both growth and activity of L. plantarum.

Furthermore, increase in protein levels in millet after fermentation is probably due to the production of single-cell protein by the LAB culture (Yadav et al., 2016), or the synthesis of specific amino acids, which subsequently elevates protein levels after the fermentation process (Uwaegbute et al., 2000). In contrast, decrease in protein level is potentially due to changes in dry matter, which is probably the result of the hydrolysis of sugars and fats (Dangal et al., 2024).

Reduced fat content in both SMA and MMA during fermentation could be due to the enhanced lipase activity produced by L. plantarum. It is reported that lipases could degrade triglycerides into fatty acids and glycerol, which are then metabolized for energy, potentially increasing nutrient bioavailability or altering fat content (Djorgbenoo et al., 2023; Jan et al., 2022; Ogodo et al., 2019).

The rate of increase in carbohydrate content, particularly between 48 h and 72 h of fermentation, is slightly higher than in SMA (Supplementary Table S1). The rise in carbohydrates could be due to L. plantarum depleting dry matter during the hydrolysis and metabolism of oligosaccharides, simple sugars, and lipids. In this study, the calculated carbohydrate by difference includes crude fiber. Another explanation for increase in carbohydrates after L. plantarum fermentation is that because pearl millet contains 2.6–4.0% crude fiber, a lignin–carbohydrate complex (Abdalla et al., 1998), laccases discovered in L. plantarum (Olmeda et al., 2021), could probably -contribute to breaking down lignocellulose molecules (Bao et al., 2022; Zhou et al., 2023), partially releasing cellulose and hemicellulose from the complex, thereby increasing the carbohydrate content.

Overall, the results indicated that MMA showed slightly better preservation of macronutrients prior to fermentation, compared to SMA, although variations were observed in fat and protein levels as well.

Sugar analysis

In this study, sucrose, glucose, and fructose were analyzed in millet samples. However, sucrose was not detected in these samples. As shown in Figure 3, glucose levels in 0MMA and raw millet (0r) are statistically similar (86.81 and 94.11 mg/kg dw, respectively; p > 0.05). In contrast, fructose levels are significantly higher in 0MMA than in raw millet (73.18 and 59.92 mg/kg dw, respectively; p < 0.05). This suggests that autoclaving may partially hydrolyze specific oligosaccharides, such as sucrose or maltose, into their constituent monosaccharides, glucose and fructose. This is supported by the fact that autoclaving can hydrolyze about 10% of sucrose (Druart and Wulf, 1993). Furthermore, regardless of the fermentation time, data in Supplementary Table S2 reveal that SMA has a significantly (p < 0.05) higher glucose level (total mean of 192.57 mg/kg) than MMA (total mean of 53.79 mg/kg), indicating the difference between the two autoclaving modes in their interactions with substrate material (see Section 3.1).

Regarding the impact of fermentation, two distinct phases are observed in glucose and fructose levels in millet samples. An initial phase (0–36 h), aligning with the lag–log exponential stage, and a second phase (36-72 h) aligning with the log–stationary–death stage (Figure 3; Supplementary Table S2). The SMA-treated millet shows a significantly higher glucose level (181.58 mg/kg dw) after 12 h of fermentation compared to that of 0MMA (control: 86.81 mg/kg dw), while it remains unchanged in MMA-treated millet (71.72 mg/kg dw; Figure 3). Increasing the fermentation time from 12 h to 36 h resulted in a gradual increase in glucose content from 86.81 to 388.25 mg/kg dw in SMA samples, after which it declined. MMA samples exhibit a similar decreasing trend as that of SMA-treated samples, with glucose reaching zero level after 36 h of fermentation (Figure 3). After 48 h of fermentation, glucose levels decreased in both SMA and MMA (p < 0.05), with SMA showing a gradual decline to its lowest level at 72 h, while glucose in MMA is entirely depleted after 48 h. Conversely, fructose in SMA and MMA behaved differently during L. plantarum fermentation compared to glucose, as its levels consistently decreased in MMA samples over 12–72 h of fermentation, reaching zero by 72 h. In contrast, L. plantarum fermentation of SMA-treated millet causes fructose to drop to zero after 24 h (p < 0.05; Figure 3).

Generally, fructose level in the MMA sample, regardless of fermentation, is significantly higher (p < 0.05) than in the SMA sample. Fermentation of millet, regardless of the autoclaving method used, exhibited a significant and gradual decrease (p < 0.05) in fructose levels with increasing incubation time, reaching a minimum at 72 h (Supplementary Table S2).

However, L. plantarum strains demonstrated superior carbohydrate utilization compared to most LABs, attributed to their sugar biosynthesis pathways and sugar-specific phosphate transport systems, enabling metabolism of a range of sugars, such as glucose and fructose (Zhao et al., 2023). Increase in glucose in fermented SMA-treated millet samples could be due to the breakdown of complex sugars into simpler forms through starch hydrolysis (Mohapatra et al., 2024). In contrast, gradual decrease of glucose by L. plantarum is probably attributed to the fact that LAB are characterized by more robust carbohydrate utilization system, commonly discovered in most LAB strains, which supports cellular growth (Mohapatra et al., 2024). These results coincided with the change in population count over 72-h fermentation of autoclaved-pretreated millet flour (Section 3.1). However, decrease in fructose during fermentation could be due to its consumption through the glycolysis pathway (Krahn et al., 2021) or its conversion to mannitol by L. plantarum (De Vuyst et al., 2009; Krahn et al., 2021). Generally, the severe nutrient limitation is a primary driver of the transition into the death phase (Section 3.1), as the cells no longer have the energy to maintain viability (Li et al., 2023).

Finally, millet fermentation, following SMA and MMA, variably affects glucose and fructose levels, suggesting that sterilization and microbial processes interact to influence these levels.

Analysis of nutritional and toxic heavy element levels

For the quality control (QC) of the ICP-MS-2030 used for element analysis, a certified multi-element solution containing 36 inorganic standard elements (10 mg/L each of Ag, Al, As, Ba, B, Ca, Cd, Ce, Co, Cr, Cu, Dy, Er, Eu, Fe, Gd, Ho, K, La, Li, Lu, Mg, Mn, Na, Nd, Ni, P, Pb, Rb, Se, Sm, Sr, Tl, Tm, V, and Zn) and a separate 10 mg/L Hg standard were used for calibration. Spike samples were analyzed to assess matrix effects and method accuracy, yielding recovery rates of 80–120%. The method was validated for heavy toxic metals with limits of detection (LOD) and quantitation (LOQ) values for Fe (0.33 and 1.125 ppb), Na (19.97 and 66.58 ppb), Mn (1.54 and 5.25 ppb), Zn (0.22 and 0.744 pb), Ca (32.57 and 108.56), K (19.44 and 64.81 ppb), Mg (2.26 and 7.55), As (2.80 and 9.34 ppb), Cd (0.54 and 1.81 ppb), Hg (1.54 and 5.14 ppb), Pb (1.09 and 3.63 ppb), and Sb (3.50 and 11.66 ppb), in accordance with ISO (2004, 2016) standards (ISO and International Electrotechnical Commission [IEC], 2017).

The raw millet flour contains both nutritional macro--elements and microelements as well as heavy toxic metals (Tables 3 and 4; Supplementary Tables S3 and S4). K, Mg, Na, Ca, Cr, Cu, Fe, Ni, Mn, and Zn have levels of 3432.5, 1011.1, 127.4, 201.8, 1.23, 6.3, 1.04, 3.59, 19.8, and 46.2 mg/kg dw, respectively. Additionally, it contains heavy toxic metals such as As, Cd, Hg, Pb, and Sb, with levels of 6.62, 66.87, 33.74, 16.59, and 1,256.62 µg/kg dw, respectively. All nutritional elements and heavy toxic metals in millet flour remain unchanged, compared to 0MMA, except for K, Na, and Sb, which showed significantly increased or decreased levels, respectively (p < 0.05; Supplementary Table S3). Regardless of fermentation, SMA resulted in significantly higher total mean levels of Cr, Cu, Fe, Cd, Hg, and Sb compared to MMA, and significantly lower mean levels of Mn, Ca, Mg, and Na (p < 0.05), while Ni, As, and Pb remained unchanged (p > 0.05) (Tables 3 and 4). This suggests that MMA and SMA treatments may interact differently with the millet flour matrix (Figure 3), leading to varied mineral content. Previous studies have reported fluctuations in the mineral content of cereals and legumes after autoclaving (Karaçoban et al., 2023; Kemal et al., 2025).

The mineral content results of millet flour after L. plantarum fermentation, regardless of SMA and MMA, are shown in Table 2. Total mean levels of several nutritional minerals, including Cu, Mn, Ca, K, Mg, and Na, increased significantly during the fermentation period of 12–36 h (p < 0.05). After that, their levels either decreased gradually until 72 h (Mg), fluctuated (Na) (p < 0.05), or remained stable (p > 0.05). In contrast, other nutritional elements varied differently over the 72 h of fermentation. For example, Cr gradually decreased significantly until 24 h of fermentation, then remained stable for rest of the period. While the total mean levels of Zn and Ni remained unchanged over the entire 72 h, but Fe showed variable changes (Table 3; Supplementary Table S3). It is worth noting that decrease in Mg during later fermentation stages may result from its uptake by L. plantarum as an essential cofactor for microbial enzymes involved in glycolysis and nucleic acid stability (Sun et al., 2025). Considering that adults can consume, on average, 1.5 servings of cooked millet (1 serving = 1 cup; 174 g) (Gleim, 2025), raw and autoclaved/fermented samples can supply Fe, Mn, Zn, Cu, Cr, and Ni that meet the recommended dietary allowance (RDA) or adequate intake (AI) daily requirements.

In Table 4, SMA shows significantly higher levels of Hg and Cd (total means of 33.17 and 1,307.64 µg/kg dw, respectively) and lower Cd levels (total mean: 71.81 µg/kg dw) than MMA (p < 0.05), while As levels remain similar (p > 0.05). These results confirmed the varied impacts of SMA and MMA on millet components. Moreover, regardless of the autoclaving method, the fermentation of autoclave-pretreated pearl millet with L. plantarum does not reduce the levels of analyzed heavy metals. Instead, it causes significant increase in Cd, Pb, and Sb (p < 0.05), while As and Hg levels remain unchanged (Table 4). Although cereals provide essential nutrients, they also contain toxic heavy metals that can harm humans. The European Regulation (EU) 2021/1323 and the US Food and Drug Administration (FDA) have set maximum permissible limits (MPL) for these harmful substances (Mititelu et al., 2025). Except for Sb, all toxic heavy metals in raw and autoclaved-millet fermented samples are below the MPLs (Table 3 and Supplementary Table S4). Previous studies have reported contamination of millets with toxic heavy metals (Omeje et al., 2021).

According to the above-mentioned data, variations in millet’s mineral composition after autoclaving pretreatment, followed by L. plantarum fermentation, could be due to changes in dry matter content, as microbial metabolism alters millet composition by consuming energy-rich components.

In summary, processing methods for pearl millet flour significantly influence its mineral content. SMA and MMA result in varying changes in flour. At the same time, fermentation generally boosts levels of beneficial minerals but also raises certain toxic heavy metals, mostly to non-harmful levels.

Analysis of TPC, TFC, TTC, and DPPH scavenging activity

The raw millet sample exhibits TPC, TFC, TTC, and DPPH scavenging activity of 2.81 mg GAE/g dw, 2.20 mg QE/g dw, 2.04 mg TAE/g dw, and 76.82%, respectively (Table 5 and Supplementary Table S5). As observed, 0MMA has significantly lower TPC (2.73 mg GAE/g dw) and higher TFC (2.67 mg QE/g dw) than raw millet (p < 0.05), while TTC is similar in both cases (Supplementary Table S5). This decline in TPC of millet after autoclaving could be due to the breakdown of heat-sensitive phenolic compounds or their transformation into unextractable forms, as they form insoluble complexes with macromolecules (Wang et al., 2022). Moreover, Rahman et al. (2021) reported that thermal treatment boosts flavonoid levels by releasing bound forms, but the effect depends on plant matrix, temperature, and duration. Furthermore, regardless of fermentation period, SMA samples have significantly higher TPC than MMA samples (total mean of 3.12 and 2.94 mg GAE/g dw, respectively), while TFC and TTC showed similar values (Table 5). This suggests that these molecules react differently under two different autoclaving modes (121°C, 15 psi, 15 min). However, the effect of autoclaving on food varies depending on factors such as material, duration, temperature, and the type of phenolic compounds present. The impact of autoclaving on TPC, TFC, and TTC varies across studies, with decrease, increase, or no significant change reported (Balyatanda et al., 2024; Kemal et al., 2025).

Regarding the impact of fermentation, TPC increases gradually and significantly after 12 h (total mean of 2.92 mg GAE/g dw), reaching a peak between 36 h and 48 h, then gradually declines from 48 h to 72 h (p < 0.05; Table 5). At 12–24 h, fermentation does not (p < 0.05) alter TFC significantly in 0MMA (total mean of 2.67 mg QE/g dw). However, 24-h fermentation results in a significant increase in TFC, peaking at 36 h and remaining nearly stable until 72 h. After 36 h, L. plantarum fermentation increases significantly (p < 0.05) the TFC of 0MMA to 3.87 mg QE/g dw, which remained almost unchanged until 72 h of fermentation (Table 5).

For TTC, results show that regardless of the sterilization method used, 24-h fermentation significantly reduces the TTC of 0MMA from 2.00 to 1.70 mg TAE/g dw, and further to 1.56 mg TAE/g dw after 36 h (p < 0.05), remaining unchanged until the end of the 72-h fermentation period (p > 0.5) (Table 4). However, controlled fermentation can partially restore or enhance phenolics, with the degree of recovery depending on the initial autoclaving intensity and fermentation time (Khalil et al., 2025; Liangyu et al., 2023). Jan et al. (2022) investigated the effects of autoclaving (at 121°C for 15 min), followed by fermentation of Finger millet flour for 12, 24, and 36 h on phenolic compounds. They discovered that the TPC of millet increased with fermentation period, while tannins decreased (by 53.6–56%), aligning with our findings.

Generally, gradual increase in TPC and TFC after 72 h of L. plantarum fermentation could be due to fermentation reactions, such as decarboxylation, hydrolysis, microbial oxidation, reduction, esterification, and release of conjugated phenols (Alshammari et al., 2025). It is discovered that LAB produce -β--glucosidase and decarboxylase. These enzymes can facilitate the release and transformation of phenolic compounds from the plant cell wall during pomace fermentation, thereby boosting TPC (Kalinowska et al., 2023). In addition to improving TPC, L. plantarum metabolizes phenolic compounds into non-phenolic ones, such as converting protocatechuic acid (a hydroxybenzoic acid) into catechol through decarboxylation (Pulido-Mateos et al., 2024). This may occur when the TPC of millet flour decreases after more than 60 h of fermentation (Supplementary Table S5). Reduction in antinutritional factor TTC after L. plantarum fermentation of pearl millet flour could be due to partial hydrolysis by tannase, a tannin acyl hydrolase produced by LAB, which reduces tannin polymerization, thereby lowering its content and anti-nutritional properties (Yang et al., 2023). This decrease in tannin content was observed previously in finger millet fermentation, with a greater reduction with increased fermentation period (Endalew et al., 2024). An increase in TPC and TFC after cereal fermentation by Lactobacillus spp., including L. plantarum, has been reported previously (Srivastava et al., 2024; Yang et al., 2024).

Post-autoclaving fermentation with L. plantarum has been reported to enhance DPPH antioxidant activity by releasing bound phenolics and flavonoids and by forming new antioxidant metabolites (Mutshinyani et al., 2020).

The DPPH scavenging rate of raw millet is 76.82%, which significantly decreased to 63.4% in raw millet treated with MMA (0MMA) (p < 0.05; Supplementary Table S5). Moist heat and pressure during autoclaving may reduce antioxidant molecules’ ability to neutralize free radicals (Chuwech et al., 2023). However, SMA exhibits a lower DPPH scavenging rate than MMA (77.51% and 78.32%, respectively), regardless of fermentation period (p < 0.05; Table 5). The high moist heat, probably occurring during MMA treatment (with a millet flour–moisture ratio of 1:4) may create a more effective heat-transfer environment in MMA than in SMA, which involves autoclaving of millet flour and water separately before combining them. This could enhance the activity of specific antioxidant molecules and, in turn, improve the overall antioxidant activity (Popoola, 2022).

Research indicates that pressure and thermal treatment can alter the integrity of macromolecular structures, thereby affecting polyphenol dissolution rates and antioxidant efficacy (Zheng et al., 2023). Furthermore, although SMA sample has a significantly higher TPC than MMA sample, the DPPH activity is significantly higher in MMA sample (Table 5). This rise in TPC may primarily result from the release of bound phenolics and the formation of Maillard reaction products (Carciochi et al., 2016). Horvat et al. (2020) discovered that the DPPH scavenging rate decreases because the most potent, heat-sensitive antioxidants may be destroyed, and the newly available phenolics often become inactive through binding or transformation into less effective compounds. Moreover, Zhang et al. (2017) discovered that extrusion cooking of highland barley increased total phenolic content while decreasing DPPH activity. They attributed this to decrease in free phenolic acids, such as vanillic acid and syringic acid, which form insoluble complexes that reduce antioxidant activity. Han et al. (2023) stated that the impact of thermal processing on cereal antioxidant activity depends on a balance, as it may generate new antioxidant Maillard reaction products while also destroying native phenolics. The result is either an increase, a decrease, or no change in antioxidant activity, depending on specific processing conditions and cereal variety.

Additionally, regardless of the autoclaving method, the DPPH scavenging rate of 0MMA (63.40%) increases by 79.08% after 12 h of fermentation (p < 0.05). This rate continues to rise, reaching its peak at 36 h, and then slightly declines as fermentation progresses to 72 h (77.10%) (p < 0.05) (Table 5). As observed, L. plantarum fermentation of millet flour after MMA treatment shows better DPPH scavenging activity than SMA (Table 5; Supplementary Table S5). This is likely due to the effectiveness of moist heat from the MMA treatment, which enhances L. plantarum’s ability to produce low molecular phenolic metabolites with increased bioactivity, such as (+)-catechin, (–)-epicatechin, and others (Pulido-Mateos et al., 2024). Studies indicate that L. plantarum produces enzymes, such as β-glucosidase, during fermentation that hydrolyzes phenolic glycosides into aglycones with radical-scavenging properties, thereby enhancing the antioxidant activity of the fermented product (Paventi et al., 2025). The decrease in DPPH scavenging activity after 36 h of fermentation in millet samples could result from the conversion of phenolic compounds into non-phenolic ones, as mentioned above (Pulido-Mateos et al., 2024). Additionally, an increase in lactate in the medium can cause cellular acidic stress, leading to a linear decrease in the growth rate of L. plantarum, which negatively impacts all metabolic activities (Giraud et al., 1991). This is evident in the reduction of TPC and DPPH scavenging rates after 72 h of fermentation in millet samples (Table 5; Supplementary Table S5).

In conclusion, the 24–36-h fermentation period often marks a turning point, during which levels of phenolics, flavonoids, and tannins as well as antioxidant activity undergo significant changes. This corresponds to the findings of increased phenolics and DPPH activity at 38.86 h of fermentation by L. plantarum (Srivastava et al., 2024). Additionally, higher concentrations of flavonoids, total phenols, and improved radical scavenging activity were observed when using L. plantarum and other strains (Yang et al., 2024).

Untargeted LC-MS/MS profiles of processed pearl millet samples

Although untargeted LC-MS/MS is an effective method for identifying active ingredients in complex natural mixtures, it has several limitations. Owing to issues such as ionization bias and combined effects, results depend heavily on data processing. They can be influenced by false positives (detecting inert chemicals) and false negatives (missing potent and low-concentration compounds; Caesar et al., 2019). Nevertheless, despite possible quantitative errors from using the percentage of total area, we used these data to preliminarily indicate that our fermented samples contained bioactive compounds, possibly from millet or microbes, which may contribute to their DPPH radical scavenging activity (Section 3.6) as well as some contaminants that could be impacted by autoclaving/fermentation processes.

A total of 18 bioactive compounds and contaminants were identified in SMA-pretreated fermented pearl millet samples after 72 h, compared to 15 in MMA-pretreated samples under the same conditions. These compounds mainly originate from millet, microbial sources, and contaminants. The SMA- and MMA-pretreated fermented samples share several compounds, such as indoxacarb, 3-carboxy-1-hydroxypropyl-thpp, methotrexate, absinthin, and metconazole (Tables 6 and 7; Supplementary Figures F1 and F2). The profile of bioactive compounds changed significantly over time. Some beneficial phytochemicals (e.g., salicin 6-phosphate and absinthin) and contaminants (e.g., insecticides, such as indoxacarb) decreased or were completely metabolized during the 72-h L. plantarum fermentation (Acebrón et al., 2017; Wei et al., 2022). Conversely, some compounds (e.g., inulicin and solasodine) are only detected later in fermentation (72 h), potentially released from conjugated forms by microbial enzymes (Kumar et al., 2019). It has been reported that fermentation of millet flour with LAB can yield probiotic-rich products and enhance bioactive compounds (Tomar et al., 2025). Moreover, fermentation effectively reduced levels of pesticides and fungicides (e.g., indoxacarb, cyfluthrin, and metconazole), with most disappearing within 60–72 h (Armenova et al., 2023).

In conclusion, fermentation with L. plantarum modifies the bioactive profile of SMA- and MMA-pretreated pearl millet by increasing compound diversity and decreasing beneficial phytochemicals while effectively degrading pesticide and fungicide contaminants over 72 h.

Conclusions

This study demonstrates that SMA and MMA pretreatments affect millet matrix and L. plantarum fermentation, revealing their links to bacterial growth, sugar consumption, nutritional content, and antioxidant and bioactive compound production through a detailed systems-level analysis. Both pretreatments achieved the same peak cell density but differed in their effects. The data showed that SMA was favored for fast startup processes, such as producing fermented beverages or quick acidification, due to its rapid growth and acid production. MMA is more appropriate for developing functional ingredients that require higher antioxidant activity and mineral bioavailability, despite a slower initial fermentation. This shows that sterilization method is more than just a technical step; it is a key factor in developing effective fermentation processes.

However, the study has some limitations, including the method of carbohydrate measurement, which is calculated “by difference” and may not accurately reflect metabolizable carbohydrates. In addition, explanations for starch gelatinization, nutrient bioavailability, and compound release rely on indirect data, and direct measurements of resistant starch and key enzyme activities were absent. Moreover, LC-MS/MS data comparisons based on percentage of total area are prone to ionization bias, affecting the accuracy of concentration measurements.

Future research should assess the impact of autoclaving pretreatments on millet’s metabolic and structural changes, evaluating starch and enzyme activity, quantifying metabolic products, and analyzing structural changes using techniques such as scanning electron microscope (SEM) and X-ray diffraction (XRD). Results should be validated through pilot- and larger-scale trials, as well as consumer sensory evaluations, to determine product quality and acceptability.