From Medical Herb to Functional Food: Development of a Fermented Milk Containing Silybin and Protein from Milk Thistle

Milk thistle is a traditional medicinal herb. Silybin is a medicinal component found in the seed coat of milk thistle, which has liver-protective and anti-cancer properties. Conventional studies have focused on the extraction of silybin with organic reagents, which was inapplicable to the food industry. This study aims to develop a fermented milk containing silybin and protein from the milk thistle seeds. A three step procedure was developed, comprising homogenization of milk thistle seeds, NaHCO3 heat treatment, and microbial fermentation. The silybin was characterized by high performance liquid chromatography, and the protein was quantified and electrophorized. It was found that the homogenization step was essential for the preparation of protein, and the NaHCO3 heat treatment was the crucial step in obtaining silybin. The optimal NaHCO3 treatment settings were 1% NaHCO3, 60°C, and 3 h, and the optimal strains for microbial fermentation were L131 (Rummeliibacillus stabekisii) and RS72 (Lactobacillus plantarum). The silybin yield in the fermented milk reached 11.24–12.14 mg/g seeds, accounting for 72.6–78.4% of the total silybin in the milk thistle seeds, and the protein yield reached 121.8–129.6 mg/g seeds. The fermented milk had a slightly sweet yoghurt-like flavor and could be used as a dietary supplement for silybin and protein.


Introduction
Milk thistle (Silybum marianum) is an annual or biennial herbaceous plant belonging to Asteraceae family with a 2000-year history of use [1]. Silymarin is a mixture of flavonolignans contained in the coats of milk thistle seeds, which has a variety of pharmacological activities [2]. Silymarin can prevent liver cell degeneration, promote liver purification, improve detoxification, and also promote the repair of damaged liver cells [3][4][5]. Silymarin mainly scavenges free radicals through anti-lipid peroxidation, increases the level of glutathione (G-SH) in the liver, and therefore exerts a hepatoprotective effect [6,7]. Chemical analysis reveals that silymarin is composed by silybin, isosilybin, silydianin, and silychristin [8]. Silybin and isosilybin account for 60-70% of silymarin. Silybin can significantly inhibit the expression of IL-2, IL-4, IFN-γ, TNF-α in the liver, reduce the levels of alanine aminotransferase and aspartate aminotransferase, and inhibit apoptosis in hepatocyte [6,7,9]. Therefore, the purified silybin and the crude silymarin compounds are important pharmaceutical materials obtained from milk thistle seeds.

Extraction with Organic Solvents
Two previous methods used for silybin extraction with organic solvents were performed. Milk thistle seeds were ground into fine powder (particle diameter < 1 mm) using a Scientz-48 grinder (Xinzhi Biological Technology Co., Ltd., Ningbo, China) at 50 Hz for 4 min. Subsamples of ground seed powder were treated by n-hexane overnight at a ratio of 1:10 (w/v) to remove lipid in the seeds and obtain the defatted seed powder. Method 1 was conducted with methanol solvent following Ren et al. [13]. Briefly, seed powder or defatted powder were soaked in methanol overnight at a ratio of 1:5 (w/v) and then ultrasonically extracted for 45 min at 30 • C (40 kHz, 100 W) using a G-020S ultrasonic cleaner (Geneng Cleaning Equipment Co., Ltd., Shenzhen, China). The mixture was centrifuged to withdraw the supernatant and repeated for the ultrasonic extraction until the two sectors of supernatant were combined and quantified for the silybin using HPLC analysis. Method 2 was conducted with ethanol solvent following Jia et al. [22]. Briefly, seed powder or defatted powder were mixed with ethanol in a 1:15 w/v ratio and then sonicated (40 kHz, 120 W) at 60 • C for 60 min using the G-020S ultrasonic cleaner (Geneng). The mixture was centrifuged to obtain the supernatant and was further quantified for silybin using HPLC assay. The seed powder and defatted seed powder were extracted in duplicate for each method with 4 g seeds as starting materials.

Design of the Three Step Procedure
A three step procedure was used to prepare the fermented Silybum milk ( Figure 1). Firstly, each gram of milk thistle seeds was added with 50 mL water and ground by a colloidal mill (JM50; Jiadeyihai Instruments Ltd., Wenzhou, China) for 5 min. The mixture was centrifuged (4000 rpm, 10 min) to obtain 40 mL supernatant of solution I (Soln I). Secondly, the solid-liquid residue of the first step was added with 40 mL NaHCO 3 solution, and treated under various conditions of temperature, time, and concentration of NaHCO 3 . The resulting solution was centrifuged (4000 rpm, 10 min) to obtain 40 mL supernatant of solution II (Soln II). Thirdly, the solid-liquid residue of the second step was added with 20 mL NaHCO 3 solution and 1 g sucrose and further sterilized. The solution was then supplied with 3 mL of bacterial culture and fermented overnight. The resulting solution was centrifuged (4000 rpm, 10 min) to obtain the 20 mL supernatant of solution III (Soln III). Finally, the solutions I, II, and III were combined and sterilized to obtain the fermented Silybum milk. Figure 1 shows the volume of solution used per gram seeds, and 4 g milk thistle seeds were used as the starting material in each experiment.

Bacterial Fermentation
The bacterial fermentation was the third step in the three step procedure. To optimize the third step, the second step of the NaHCO 3 treatment was set at 60 • C, 3 h, 1% NaHCO 3 (0.01 g/mL) according to the optimal condition identified in Section 2.4. The bacterial culture was obtained with a density > 10 8 CFU/mL after overnight culturing at 30 • C and 190 rpm rotation speed. The optimization of the third step was carried out as follows: (1) using RS72 strain, fermented in 0, 0.5 and 1% NaHCO 3 solution for 12 h, 1 d, 2 d, and 3 d, respectively; (2) using 1% NaHCO 3 solution and 12 h fermentation time, inoculated with the bacterial strain of L131, RS72, RS102, RS103, and NJZ, respectively, associated with the blank control (BK, without bacterial inoculation); and (3) finally, performing a big scale treatment, so that 100 g seeds were processed instead of 4 g engaging with BK, L131, and RS72 condition, respectively. The silybin concentrations in the Solns I, II, and III were quantified using the HPLC, and each condition was conducted in duplicate.

Protein Analysis
The soluble protein contents in the Solns I, II, and III following Section 2.5 (2) were quantified and profiled through gel electrophoresis. The quantification used the Bradford Protein Assay kit (Sangong Bioengineering Co., Ltd., Shanghai, China) following the manufacturer's instructions. The gel electrophoresis was performed using the denaturing protein electrophoresis method (SDS-PAGE) with 5% concentrated gel and 8% separation gel [23,24]. Then, 4 µL of liquid sample was loaded on the gel and electrophoresed at 180 V for 2 h in a JY-SCZ2+ chamber (Junyidongfang Equipment Co., Ltd., Beijing, China) with PowerPac TM HC electrophoresis instrument (BIO-RAD, Hercules, CA, USA).

HPLC Analysis of Silybin Content
The silybin component in the sample was determined by a SPD-16 HPLC platform (Shimadzu Instrument Manufacturing Co. Ltd., Suzhou, China) equipped with a Zorbax Eclipse XDB-C18 Column (250 mm× 4.6 mm, 5 µm; Agilent Technologies, Santa Clara, CA, USA) following Zhao et al. [25] with modifications. Briefly, dual mobile phases were used in the detection with water as phase A and methanol as phase B. The HPLC was programmed as follows: 0~8 min, phase B increased from 50% to 70%; 8~10 min, phase B increased from 70% to 80%; 10~10.5 min, phase B increased from 80% to 95%; 10.5~12 min, phase B kept at 95%; 12~12.1 min, phase B reduced from 95% to 50%; and 12.1~15 min, phase B kept at 50%. Column temperature was 20 • C; flow rate was 1.2 mL/min; UV detection wavelength was 288 nm. The liquid sample was diluted with methanol, filtered through a 0.45 µm membrane, and examined through the HPLC with 20 µL volume. Silybin standard was prepared as 1 mg/mL in methanol, serially diluted with a 10-fold gradient, and profiled by the HPLC. A standard curve was constructed with the integrated area and standard concentration. The liquid sample was compared with the standard for the HPLC profile and the silybin content was identified through the retention time and quantified using the integrated area and the standard curve.

Statistical Analysis
The silybin yields derived from different conditions were compared by one-way analysis of variance (ANOVA) using multcomp package in the R program (ver3.4.1) [26]. Differences were considered significant at p < 0.05.

HPLC Analysis
There were double peaks in the HPLC profile of the silybin standard, corresponding to silybin A and silybin B (Figure 2A), which is consistent with previous studies [8,27].   Figure 2D), respectively. Furthermore, the peak intensity indicated that the concentration of silybin content ranked as Soln II > Soln III > Soln I ( Figure 2).   Figure 2D), respectively. Furthermore, the peak intensity indicated that the concentration of silybin content ranked as Soln II > Soln III > Soln I ( Figure 2).

Extraction with Organic Solvents
The silybin yield from Method 1 was 10.84 ± 1.35 mg/g seeds for the defatted seeds and 15.48 ± 0.35 mg/g seeds for the normal seeds ( Figure 3). The silybin yield resulted from Method 2 was 9.18 ± 0.43 mg/g seeds for the defatted seeds and 10.3 ± 1.01 mg/g seeds for the normal seeds ( Figure 3). Both methods showed that the normal seeds received higher yield than the defatted seeds, which suggested that the oil extraction step by nhexane may cause silybin loss. Comparing Method 1 and Method 2, Method 1 obtained a higher yield than Method 2, which could be partially explained by the solvent difference, that the methanol used in Method 1 was more efficient than the ethanol used in Method 2 for the silybin extraction purpose [13]. Several studies have reported the yield of silybin or silymarin with an organic extraction method: Zhang et al. [11] applied an ultrasoundassisted ethanol reflux method and achieved an extraction yield of 16.4 mg/g for silymarin; Ruan et al. [28] performed an ultrasound-assisted ethanol and ammonium sulfate biaqueous extraction and obtained a silybin yield of 10.18 mg/g; Sun et al. [10] utilized ethyl acetate for the extraction and used ethanol for the recrystallization, and obtained a silybin yield of 10.9 mg/g; Wang et al. [12] performed a silybin extraction with ultrasound-assisted ethanol method and reported the yield as 16.02 mg/g. The silybin yield of 15.48 ± 0.35 mg/g seeds (Method 1, normal seeds) in this study was aligned with previous reports. Therefore, the current milk thistle seeds were considered embracing 15.48 ± 0.35 mg/g silybin content.
Foods 2023, 12, x FOR PEER REVIEW for silymarin; Ruan et al. [28] performed an ultrasound-assisted ethanol and amm sulfate biaqueous extraction and obtained a silybin yield of 10.18 mg/g; Sun et utilized ethyl acetate for the extraction and used ethanol for the recrystallization, a tained a silybin yield of 10.9 mg/g; Wang et al. [12] performed a silybin extractio ultrasound-assisted ethanol method and reported the yield as 16.02 mg/g. The yield of 15.48 ± 0.35 mg/g seeds (Method 1, normal seeds) in this study was aligne previous reports. Therefore, the current milk thistle seeds were considered emb 15.48 ± 0.35 mg/g silybin content.

NaHCO3 Thermal Treatment
The silybin yield in Soln I was 0.434 mg/g seeds as detected by the HPLC. The yield in Soln II exhibited significant difference among varied NaHCO3 thermal con ( Figure 4). The heating time comparison revealed that 2 h was superior to the othe under 60 °C and 4% NaHCO3 (p < 0.001), that the silybin yield reached 8.75 mg/g ( Figure 4A). The heating temperature comparison indicated that 60 °C outperform other temperatures under 4 h and 4% NaHCO3 (p < 0.001), and the silybin yield ac 4.35 mg/g seeds ( Figure 4B). The NaHCO3 concentration assay suggested that 8% N was better than the other concentrations under 60°C and 4 h (p < 0.001), that the yield gained 6.33 mg/g seeds ( Figure 4C). Therefore, the heating time of 2 h, heatin perature of 60 °C, and NaHCO3 concentration of 8% were the optimal conditions preliminary determination of NaHCO3 thermal conditions. However, the advantag °C was clear in the thermal condition comparison, whereas the heating time and N concentration could be further optimized.

NaHCO 3 Thermal Treatment
The silybin yield in Soln I was 0.434 mg/g seeds as detected by the HPLC. The silybin yield in Soln II exhibited significant difference among varied NaHCO 3 thermal conditions ( Figure 4). The heating time comparison revealed that 2 h was superior to the other times under 60 • C and 4% NaHCO 3 (p < 0.001), that the silybin yield reached 8.75 mg/g seeds ( Figure 4A). The heating temperature comparison indicated that 60 • C outperformed the other temperatures under 4 h and 4% NaHCO 3 (p < 0.001), and the silybin yield achieved 4.35 mg/g seeds ( Figure 4B). The NaHCO 3 concentration assay suggested that 8% NaHCO 3 was better than the other concentrations under 60 • C and 4 h (p < 0.001), that the silybin yield gained 6.33 mg/g seeds ( Figure 4C). Therefore, the heating time of 2 h, heating temperature of 60 • C, and NaHCO 3 concentration of 8% were the optimal conditions in the preliminary determination of NaHCO 3 thermal conditions. However, the advantage of 60 • C was clear in the thermal condition comparison, whereas the heating time and NaHCO 3 concentration could be further optimized.
was better than the other concentrations under 60°C and 4 h (p < 0.001), that the silybin yield gained 6.33 mg/g seeds ( Figure 4C). Therefore, the heating time of 2 h, heating temperature of 60 °C, and NaHCO3 concentration of 8% were the optimal conditions in the preliminary determination of NaHCO3 thermal conditions. However, the advantage of 60 °C was clear in the thermal condition comparison, whereas the heating time and NaHCO3 concentration could be further optimized. The second round of NaHCO 3 optimization focused on the possibility of using longtime treatment under 60 • C with low concentrations of NaHCO 3 . The time groups of 1-5 h showed similar silybin yield of 5.65-7.69 mg/g seeds under 60 • C and 2% NaHCO 3 condition, which were all significantly higher than the 0 h group (1.19 mg/g seeds, Figure 5A). The 1 h and 3 h group obtained silybin yield of 6.73 ± 0.73 and 6.53 ± 0.04 mg/g seeds, respectively. The higher reproducibility of the 3 h than the 1 h group made the 3 h the optimum condition. Furthermore, the concentration groups of 0.5%-6% NaHCO 3 exhibited a similar silybin yield of 6.23-7.98 mg/g seeds under 60 • C and 3 h heating, which were all significantly higher than the 0% NaHCO 3 group of 3.04 mg/g seeds ( Figure 5B). The 0.5% and 1% NaHCO 3 group reached a silybin yield of 7.98 ± 1.04 and 6.88 ± 0.40 mg/g seeds, respectively. Additionally, the higher reproducibility of 1% than 0.5% NaHCO 3 condition made the 1% NaHCO 3 the optimal condition. Therefore, the optimal condition for NaHCO 3 thermal treatment was established as 60 • C, 3 h, and 1% NaHCO 3 with the silybin yield of 6.88 ± 0.40 mg/g seeds in the Soln II.
The current study used a step-by-step optimization procedure for the silybin yield improvement, i.e., from temperature condition to the duration time and NaHCO 3 concentrations, respectively, instead of using the response surface methodology (RSM) [29]. The reasons were as follows: (1) the temperature was found to be the critical parameter for the silybin extraction with NaHCO 3 treatment, since high temperature, e.g., 80 • C and 100 • C, caused great degradation of silybin in the Soln II ( Figure 4B). Although temperatures over the 60-80 • C range may lead to a higher yield than 60 • C, the risk of thermal decomposition of silybin was also raised [30]. Therefore, the current strategy used the mild thermal condition of 60 • C instead of further optimizing the temperature to reduce the risk of silybin thermal degradation. (2) The thermal duration times and NaHCO 3 concentrations resulted in an optimal range instead of certain points for a comparable silybin yield ( Figure 5). Therefore, the RSM strategy may lead to a bias that not only one condition could result in a superior yield. Instead, the reproducibility of the extraction procedure was important, that the 3 h duration time and 1% NaHCO 3 were selected due to the stable silybin extraction performance. (3) The simplicity and feasibility of the technical parameters were of significance since the current method aimed to be applied in the food industry. The parameters of 60 • C, 3 h, and 1% NaHCO 3 were easy and feasible in a food factory, which fulfilled the application needs. (4) Finally, the silybin that remained in the Soln II residue could be further extracted in Soln III.
were all significantly higher than the 0% NaHCO3 group of 3.04 mg/g seeds ( Figure 5B). The 0.5% and 1% NaHCO3 group reached a silybin yield of 7.98 ± 1.04 and 6.88 ± 0.40 mg/g seeds, respectively. Additionally, the higher reproducibility of 1% than 0.5% NaHCO3 condition made the 1% NaHCO3 the optimal condition. Therefore, the optimal condition for NaHCO3 thermal treatment was established as 60 °C, 3 h, and 1% NaHCO3 with the silybin yield of 6.88 ± 0.40 mg/g seeds in the Soln II. The current study used a step-by-step optimization procedure for the silybin yield improvement, i.e., from temperature condition to the duration time and NaHCO3 concentrations, respectively, instead of using the response surface methodology (RSM) [29]. The reasons were as follows: (1) the temperature was found to be the critical parameter for the silybin extraction with NaHCO3 treatment, since high temperature, e. g. 80 °C and 100 °C, caused great degradation of silybin in the Soln II ( Figure 4B). Although temperatures over the 60-80 °C range may lead to a higher yield than 60 °C, the risk of thermal decomposition of silybin was also raised [30]. Therefore, the current strategy used the mild thermal condition of 60 °C instead of further optimizing the temperature to reduce the risk of silybin thermal degradation. (2) The thermal duration times and NaHCO3 concentrations resulted in an optimal range instead of certain points for a comparable silybin yield (Figure 5). Therefore, the RSM strategy may lead to a bias that not only one condition could result in a superior yield. Instead, the reproducibility of the extraction procedure was A few studies have also used alkaline solution for the silybin extraction from the defatted milk thistle seeds. Ren et al. [13] reported that the treatment with 0.5 mol/L NaOH resulted in a silybin yield of 2.32 mg/g. Li et al. [17] found that 2 mol/L NaOH extraction associated with ultrasonic treatment resulted in a silymarin yield of 10.47 mg/g. The current study used 1% NaHCO 3 , i.e., 0.119 mol/L NaHCO 3 , which was milder than the previous 0.5 mol/L or 2 mol/L NaOH solution, whereas it achieved a comparable yield of silybin (6.88 ± 0.40 mg/g seeds). This high extraction efficiency could be derived from four reasons. Firstly, the current study utilized a three step extraction procedure, for which a large amount of seed protein was collected and removed in the first homogenization step, which allowed the NaHCO 3 to mainly interact with the ground seed coats instead of the seed protein in the second step. Secondly, the NaHCO 3 interacted with the phenolic hydroxyl group of silybin, which enhanced the solubility of silybin. Thirdly, the current procedure utilized a heating method incorporated with NaHCO 3 treatment, which facilitated the saponification of seed fatty acids. Fourthly, the NaHCO 3 improved the disruption of plant cell walls, which facilitated the release of silybin.

Microbial Fermentation
The NaHCO 3 concentration in the broth and the fermentation time had significant effects on the silybin yield in the Soln III. The 1% NaHCO 3 broth resulted in a higher silybin yield than the 0% and 0.5% NaHCO 3 broth (p < 0.001, Figure 6). The 12 h fermentation time achieved higher silybin yield than the 1 d, 2 d, and 3 d time (p < 0.001, Figure 6), which suggested that bacteria may utilize silybin during the long-time fermentation. Engaging with the RS72 Lactobacillus plantarum strain, the fermentation results indicated that the 1% NaHCO 3 broth and 12 h fermentation time were the optimal conditions for the third step, and the silybin yield achieved 1.50 ± 0.19 mg/g seeds in Soln III under this condition ( Figure 6C). lybin yield than the 0% and 0.5% NaHCO3 broth (p < 0.001, Figure 6). The 12 h fermentation time achieved higher silybin yield than the 1 d, 2 d, and 3 d time (p < 0.001, Figure 6), which suggested that bacteria may utilize silybin during the long-time fermentation. Engaging with the RS72 Lactobacillus plantarum strain, the fermentation results indicated that the 1% NaHCO3 broth and 12 h fermentation time were the optimal conditions for the third step, and the silybin yield achieved 1.50 ± 0.19 mg/g seeds in Soln III under this condition ( Figure 6C).  Using the fermentation condition of 1% NaHCO 3 broth and 12 h time, varied bacterial strain resulted in different silybin yields in the fermentation step (Figure 7). The silybin yields in Soln III were found as follows: BK (3.16 ± 0.34 mg/g seeds), L131 (3.83 ± 0.96 mg/g seeds), RS72 (2.07 ± 0.42 mg/g seeds), RS102 (1.22 ± 0.67 mg/g seeds), RS92 (1.69± 0.14 mg/g seeds), and NJZ (0.82 ± 0.07 mg/g seeds). The silybin yields in the Soln I and Soln II were 0.549 and 7.34 mg/g seeds, respectively. Therefore, the integral silybin yields following the three step procedure were achieved as: BK (11.05 mg/g seeds), L131 (11.71 mg/g seeds), RS72 (9.96 mg/g seeds), RS102 (9.10 mg/g seeds), RS92 (9.57 mg/g seeds), and NJZ (8.70 mg/g seeds) ( Figure 7A). Except for the BK, the Soln III that resulted from the bacterial fermentation had a slightly sweet taste and yoghurt-like flavor, while the Soln III of BK exhibited a slightly bitter taste without a special flavor. The BK, L131, and RS 72 were used for scaling-up processing (100 g) due to the relative higher silybin yields than other strains. The silybin yields for Soln I, II, and III following the scaling-up processing were found as 0.528, 8.69, and 2.27 mg/g seeds for the BK (i.e., 4.5%, 75.6%, 19.8% of the total yield), 0.548, 8.89, and 2.70 mg/g seeds for the L131 (i.e., 4.5%, 73.2%, 22.3% of the total yield), and 0.570, 9.03, and 1.65 mg/g seeds for the RS72 (5.1%, 80.3%, 14.6% of the total yield), respectively. The total yields of silybin resulted from the scaling-up processing were 11.49 ± 1.93 mg/g seeds for the BK, 12.14 ± 0.78 mg/g seeds for the L131, and 11.24 ± 0.54 mg/g seeds for the RS72, respectively ( Figure 7B).
The five bacteria (L131, RS72, RS102, RS92, NJZ) used in this study are all food fermentation strains, which are commonly used in the food industry for dairy, rice, and vegetable fermentation [21]. The silybin yield in Soln III suggested the L131 and RS72 were optimal bacterial strains for the fermentation step. The RS102, RS92, and NJZ strain may consume silybin during the fermentation ( Figure 7A). Although the silybin yield of L131 and RS72 did not improve significantly compared with BK, the liquid flavor did upgrade a lot, being slightly sweet and with a yoghurt-like taste. For both the small scale (4 g) and big scale (100 g) processing, the silybin yield in the Soln I, II, and III ranked as Soln II > Soln III > Soln I (Figure 7), and the second NaHCO 3 thermal step achieved the highest silybin yield, accounting for 73-80% of the total yield, while the bacterial fermentation step accounted for 15-22% of the total yield, and the homogenization step represented 4-5% of the total yield. The scaling-up processing did not lower the silybin yield, whereas elevating the processing efficiency probably due to the reduction of operational or material loss, which suggests the current three step method could be scaled up. The final silybin yield derived from L131 or RS72 fermentation achieved 12.14 ± 0.78 mg/g seeds and 11.24 ± 0.54 mg/g seeds, respectively ( Figure 7B, 100 g scale). These yields were comparable to the yield of 10.9-16.4 mg/g in the previous reports engaging with chemical extraction methods [10][11][12][13]. The current milk thistle seeds were found to contain 15.48 mg/g silybin using the methanol extraction method (Section 3.2). The total silybin yields of the L131 and RS72 procedure accounted for 78.4% and 72.6% silybin contained in the milk thistle seeds, respectively, which suggests the current three step processing strategy achieved good efficiency for the silybin extraction through food processing technology. the three step procedure were achieved as: BK (11.05 mg/g seeds), L131 (11.71 mg/g seeds), RS72 (9.96 mg/g seeds), RS102 (9.10 mg/g seeds), RS92 (9.57 mg/g seeds), and NJZ (8.70 mg/g seeds) ( Figure 7A). Except for the BK, the Soln III that resulted from the bacterial fermentation had a slightly sweet taste and yoghurt-like flavor, while the Soln III of BK exhibited a slightly bitter taste without a special flavor. The BK, L131, and RS 72 were used for scaling-up processing (100 g) due to the relative higher silybin yields than other strains. The silybin yields for Soln I, II, and III following the scaling-up processing were found as 0.528, 8.69, and 2.27 mg/g seeds for the BK (i.e., 4.5%, 75.6%, 19.8% of the total yield), 0.548, 8.89, and 2.70 mg/g seeds for the L131 (i.e., 4.5%, 73.2%, 22.3% of the total yield), and 0.570, 9.03, and 1.65 mg/g seeds for the RS72 (5.1%, 80.3%, 14.6% of the total yield), respectively. The total yields of silybin resulted from the scaling-up processing were 11.49 ± 1.93 mg/g seeds for the BK, 12.14 ± 0.78 mg/g seeds for the L131, and 11.24 ± 0.54 mg/g seeds for the RS72, respectively ( Figure 7B). The five bacteria (L131, RS72, RS102, RS92, NJZ) used in this study are all food fermentation strains, which are commonly used in the food industry for dairy, rice, and vegetable fermentation [21]. The silybin yield in Soln III suggested the L131 and RS72 were optimal bacterial strains for the fermentation step. The RS102, RS92, and NJZ strain may consume silybin during the fermentation ( Figure 7A). Although the silybin yield of L131 and RS72 did not improve significantly compared with BK, the liquid flavor did upgrade a lot, being slightly sweet and with a yoghurt-like taste. For both the small scale (4 g) and big scale (100 g) processing, the silybin yield in the Soln I, II, and III ranked as Soln II > Soln III > Soln I (Figure 7), and the second NaHCO3 thermal step achieved the highest silybin yield, accounting for 73-80% of the total yield, while the bacterial fermentation step accounted for 15-22% of the total yield, and the homogenization step represented 4-5% of the total yield. The scaling-up processing did not lower the silybin yield, whereas elevating the processing efficiency probably due to the reduction of operational or

Protein Analysis
Soluble protein was detected in all the Soln I, II, and III samples, with the concentration ranked as Soln I > Soln II > Soln III ( Figure 8A). The Soln I contained 1.51 mg/mL soluble protein, the Soln II obtained 1.24 mg/mL of soluble protein, and the Soln III had 0.50-0.98 mg/mL soluble protein ( Figure 8A). The bacterial fermentation resulted in more protein released in the Soln III than the BK condition (p < 0.05), and the L131 (0.59 mg/mL), RS72 (0.98 mg/mL), RS102 (0.72 mg/mL), RS92 (0.94 mg/mL), NJZ (0.65 mg/mL) all showed a higher protein concentration than the BK (0.50 mg/mL). The PAGE gels aligned with the protein quantification results, showing that the Soln I sample contained more protein bands than Soln II and Soln III with the protein molecular weight ranging from 0-116 kDa ( Figure 8B). The Soln III samples of RS72, RS92, and L131 showed more protein bands than the BK with wide distribution of protein size. Considering every gram of milk thistle seeds resulted in 40 mL Soln I, 40 mL Soln II, and 20 mL Soln III, the soluble protein yields derived from three step processing were found as 60.4 mg/g (Soln I), 49.6 mg/g (Soln II), and 11.8 mg/g (L131) or 19.6 mg/g (RS72) (Soln III), respectively. In total, the three step processing method retrieved 121.8 and 129.6 mg/g soluble protein for the L131 and RS72 fermentation, respectively. There was approximately 332-360 mg/g total protein contained in the milk thistle seeds [14][15][16], among which 58% protein was soluble. Therefore, the current three step procedure achieved 33.8-39.0% of the total protein in seeds, accounting for 58.3-67.3% of the total soluble protein, which highlights that the current method is efficient in the extraction of seed protein.

Summary of the Three Step Procedure
The current study developed a three step procedure for the preparation of a fermented milk containing silybin and protein from the milk thistle seeds. The procedure included the homogenization of milk thistle seeds, NaHCO3 heat treatment, and microbial fermentation (Figure 9). The optimal conditions for the NaHCO3 treatment were 1% Na-HCO3, 60 °C, and 3 h time, and the optimal fermentation strains for the microbial fermentation were L131 (Rummeliibacillus stabekisii) and RS72 (Lactobacillus plantarum). The silybin yield in the fermented milk achieved 11.24-12.14 mg/g seeds, accounting for 72.6-78.4% of the total silybin in seeds, and the soluble protein yield reached 121.8-129.6 mg/g seeds, accounting for 33.8-39.0% of the total protein in seeds. The fermented milk had slightly sweet taste and a yoghurt-like flavor, which could be used for a dietary supplement of silybin and protein.

Summary of the Three Step Procedure
The current study developed a three step procedure for the preparation of a fermented milk containing silybin and protein from the milk thistle seeds. The procedure included the homogenization of milk thistle seeds, NaHCO 3 heat treatment, and microbial fermentation ( Figure 9). The optimal conditions for the NaHCO 3 treatment were 1% NaHCO 3 , 60 • C, and 3 h time, and the optimal fermentation strains for the microbial fermentation were L131 (Rummeliibacillus stabekisii) and RS72 (Lactobacillus plantarum). The silybin yield in the fermented milk achieved 11.24-12.14 mg/g seeds, accounting for 72.6-78.4% of the total silybin in seeds, and the soluble protein yield reached 121.8-129.6 mg/g seeds, accounting for 33.8-39.0% of the total protein in seeds. The fermented milk had slightly sweet taste and a yoghurt-like flavor, which could be used for a dietary supplement of silybin and protein. Figure 9. Summary of the three step procedure for the preparing of the Silybum-fermented milk. The milk contained the silybin and protein from the milk thistle seeds. The silybin and protein yields were derived from the average data gained through the scaling-up processing of the L131 and RS72 fermentation.

Conclusions
The current study achieved a fermented Silybum milk containing silybin and protein.
The preparation method included three steps, namely homogenization of milk thistle seeds, NaHCO3 heat treatment, and microbial fermentation treatment. The homogenization step was important for the preparation of the protein, and the NaHCO3 heat treatment was the key procedure in obtaining silybin. Under optimal conditions, the silybin yield in the fermented milk reached 11.24-12.14 mg/g seeds, accounting for 72.6-78.4% of the total silybin in seeds, while the soluble protein yield achieved 121.8-129.6 mg/g seeds, accounting for 33.8-39.0% of the total protein in seeds. The current study provides a novel strategy for the food processing of medical herbs. Further studies can be carried out for the human health effect of the fermented milk and also for the preservation technology of the fermented milk in long-term storage.   The milk contained the silybin and protein from the milk thistle seeds. The silybin and protein yields were derived from the average data gained through the scaling-up processing of the L131 and RS72 fermentation.

Conclusions
The current study achieved a fermented Silybum milk containing silybin and protein.
The preparation method included three steps, namely homogenization of milk thistle seeds, NaHCO 3 heat treatment, and microbial fermentation treatment. The homogenization step was important for the preparation of the protein, and the NaHCO 3 heat treatment was the key procedure in obtaining silybin. Under optimal conditions, the silybin yield in the fermented milk reached 11.24-12.14 mg/g seeds, accounting for 72.6-78.4% of the total silybin in seeds, while the soluble protein yield achieved 121.8-129.6 mg/g seeds, accounting for 33.8-39.0% of the total protein in seeds. The current study provides a novel strategy for the food processing of medical herbs. Further studies can be carried out for the human health effect of the fermented milk and also for the preservation technology of the fermented milk in long-term storage.

Conflicts of Interest:
Haijun Bao, Qingan Wang and Nan Wang was employed by the company Yingkou Chenguang Extraction Equipment Co., Ltd. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.