Increased Production of Ginsenoside Compound K by Optimizing the Feeding of American Ginseng Extract during Fermentation by Aspergillus tubingensis

The ginsenoside compound K (C-K) is widely used in traditional medicines, nutritional supplements, and cosmetics owing to its diverse pharmacological activities. Although many studies on C-K production have been conducted, fermentation is reported to produce C-K with low concentration and productivity. In the present study, addition of an inducer and optimization of the carbon and nitrogen sources in the medium were performed using response surface methodology to increase the C-K production via fermentation by Aspergillus tubingensis, a generally recognized as safe fungus. The optimized inducer and carbon and nitrogen sources were 2 g/l rice straw, 10 g/l sucrose, and 10 g/l soy protein concentrate, respectively, and they resulted in a 3.1-fold increase in the concentration and productivity of C-K (0.22 g/l and 1.52 mg/l/h, respectively) compared to those used before optimization without inducer (0.071 g/l and 0.49 mg/l/h, respectively). The feeding methods of American ginseng extract (AGE), including feeding timing, feeding concentration, and feeding frequency, were also optimized. Under the optimized conditions, A. tubingensis produced 3.96 mM (2.47 g/l) C-K at 144 h by feeding two times with 8 g/l AGE at 48 and 60 h, with a productivity of 17.1 mg/l/h. The concentration and productivity of C-K after optimization of feeding methods were 11-fold higher than those before the optimization (0.22 g/l and 1.52 mg/l/h, respectively). Thus, the optimization for the feeding methods of ginseng extract is an efficient strategy to increase C-K production. To our knowledge, this is the highest reported C-K concentration and productivity via fermentation reported so far.

and high energy consumption. To overcome these disadvantages, biological methods, including enzyme conversion, cell conversion, and fermentation, have been utilized [10]. Although enzyme transformation shows higher conversion and productivity, it is less economical than fermentation because it requires enzyme purification, which results in the loss of enzymes during the purification from cells [11]. On the other hand, fermentation is relatively cost-effective because it does not require enzyme purification steps and can use both intracellular and extracellular enzymes.
In this study, we optimized the inducer and carbon and nitrogen sources in the fermentation medium to increase C-K production by Aspergillus tubingensis, a generally recognized as safe (GRAS) fungus. The concentration and productivity of C-K were also increased by optimizing the feeding with American ginseng extract (AGE) during fermentation.

Fungal Strain, Media, and Fermentation Conditions
The fungus A. tubingensis KCTC 14166 (Korean Collection for Type Cultures, Republic of Korea) was used for C-K production [20]. A. tubingensis was incubated on potato dextrose agar for 14 days. Thereafter, distilled water mixed with 0.1% (v/v) Triton-X100 was added to the plate. Spores were collected by raking agar surface using a spreader and the obtained spore solution was filtered using two layers of gauze and diluted to 1.0 × 10 6 spores/ml. The number of spores was counted using a hemocytometer (INCYTO, Republic of Korea). One ml of the spore suspension was inoculated into 5 ml of potato dextrose broth (PDB) in glass tube and incubated for 24 h at 28°C with shaking at 150 rpm. Subsequently, the grown mycelia were inoculated into a 500 ml-baffled Erlenmeyer flask containing 100 ml of the fermentation medium and incubated for 144 h at 28°C with shaking at 150 rpm. The fermentation medium before the optimization of inducer and carbon and nitrogen sources contained 20 g/l citrus pectin, 10 g/l corn steep solid, 5 g/l KH 2 PO 4 , 5 g/l Na 2 HPO 4 , 0.3 g/l MgSO 4 ·7H 2 O, 0.3 g/l CaCl 2 , 5 mg/l FeSO 4 ·7H 2 O, and 1.3 mg/l MnSO 4 ·H 2 O. The pH of fermentation media was adjusted to 5.0. Unless otherwise stated, fermentation was conducted in a shaking incubator at 28°C and 150 rpm for 144 h, and 1 g/l AGE containing 0.36 mM PPD-type ginsenosides as the substrate for C-K production was added at 72 h.

Optimization of the Inducer and Carbon and Nitrogen Sources
The effect of inducer on C-K production during fermentation of A. tubingensis was investigated using 2 g/l of arabinogalactan, arabic gum, grapefruit pectin, apple pectin, rice straw, sugar beet sludge, and sugar beet fiber. The effect of carbon source on C-K production was tested using 20 g/l of citrus pectin, cellulose, fructose, sucrose, lactose, and wheat bran, supplemented with 2 g/l rice straw. The effect of nitrogen source on C-K production was evaluated using 10 g/l of corn steep solid, soybean powder, soy protein concentrate, yeast extract, and urea supplemented with 2 g/l rice straw.

Optimization of Feeding Methods for AGE
To determine the optimal feeding timing of AGE as a substrate for C-K production, 1 g/l AGE was fed one time in the optimized fermentation medium at 24-84 h during fermentation. The optimal feeding concentration of AGE at 60 h was determined using varying the AGE concentrations from 1 to 16 g/l). To reduce the toxicity towards the fungus and the inhibition of C-K production caused by the high concentration of AGE, feeding was performed two or three times at 48, 60, or/and 72 h during fermentation.

Production of C-K via Fermentation under Optimized Conditions
C-K production via fermentation by A. tubingensis was performed under the optimized feeding conditions of feeding two times with 8 g/l AGE at 48 and 60 h, with a total of 16 g/l AGE containing 5.76 mM or 6.16 g/l PPDtype ginsenosides to the optimized fermentation medium containing 2 g/l rice straw, 10 g/l sucrose, 10 g/l soy protein concentrate, 5 g/l KH 2 PO 4 , 5 g/l Na 2 HPO 4 , 0.3 g/l MgSO 4 ·7H 2 O, 0.3 g/l CaCl 2 , 5 mg/l FeSO 4 ·7H 2 O, and 1.3 mg/l MnSO 4 ·H 2 O with shaking at 150 rpm for 156 h at 28°C.

Analytical Methods
After fermentation, the fermentation broth was extracted by adding an equal volume of n-butanol mixed with 1.0 mg/ml digoxin as an internal standard. The n-butanol layer was separated and collected via centrifugation at 13,000 ×g for 10 min. The collected layer was evaporated until it was completely dried, and an equal volume of methanol was added to the dried residue. Ginsenosides dissolved in methanol were quantitatively analyzed using the Agilent 1100 Infinity HPLC system with a UV detector at 203 nm and an octadecylsilica column (4.6 × 150 mm, 5 μm particle size; YMC, Japan). The column was eluted at 40°C with a gradient of acetonitrile/water (v/ v) from 30:70 to 60:40 for 20 min, 60:40 to 90:10 for 10 min, 90:10 to 30:70 for 5 min, and 30:70 for 10 min, at a flow rate of 1 ml/min. All ginsenosides, including reagent ginsenoside standards, ginsenosides in AGE, and biotransformed ginsenosides in AGE, were quantified by the calibration curves using the ginsenoside standards. The fermentation broth was filtered with Whatman No. 1 filter paper and washed with water, and the collected cells were dried in a dry oven at 105°C for 12 h. After drying, the dry cell weight was measured.

Statistical Analysis
The means and standard errors for all experiments, including evaluation of fermentation conditions, and HPLC analysis, were calculated from duplicate. One-way analysis of variance (ANOVA) was carried out using Tukey's method with a significance level of p < 0.05 using SigmaPlot 10.0 (Systat Software, USA).

Effect of Inducer on C-K Production via Fermentation by A. tubingensis
The extracellular enzyme from A. tubingensis KCTC 14166 with D-glucosidase, L-arabinofuranosidase, and Larabinopyranosidase activities showed the highest concentration and productivity of C-K among the enzymatic transformation reported so far [20]. Polysaccharides used in fungal cultivation are known to induce the synthesis of hydrolytic enzymes [22] and increase their activity [23]. Therefore, in the present study, C-K production via fermentation by A. tubingensis KCTC 14166 was evaluated by adding 1 g/l AGE at 72 h with 2 g/l polysaccharide as an inducer in the medium containing 20 g/l citrus pectin and 10 g/l corn steep solid, as carbon and nitrogen sources, respectively. Among the polysaccharides, rice straw produced C-K with the highest concentration (0.087 g/l) and molar conversion of PPD-type ginsenosides in AGE to C-K (36.2%), which were 123% higher than those (0.071 g/l and 29.4%, respectively) without polysaccharides (Fig. 1). Therefore, 2 g/l rice straw was supplemented in the fermentation medium for effective C-K production.

Effects of Carbon and Nitrogen Sources on C-K Production via Fermentation by A. tubingensis
Carbon sources have been used to increase the synthesis of hydrolytic enzymes during fungal cultivation [24]. To select the optimal carbon source, 20 g/l of various carbon sources were added into the medium containing 10 g/l corn steep solid and 2 g/l rice straw ( Fig. 2A). Sucrose was identified as the optimal carbon source because it led to the highest production of C-K.
The activity of hydrolytic enzymes increases by optimizing the nitrogen source during fermentation [25]. Therefore, C-K production was investigated using 10 g/l of various nitrogen sources in the medium containing 20 g/l sucrose and 2 g/l rice straw. C-K production was the highest using soy protein concentrate (Fig. 2B).
The optimization of the concentrations of inducer, sucrose, and soy protein concentrate for C-K production during fermentation by A. tubingensis was performed using RSM. BBD of RSM for the three independent variables, rice straw concentration (X 1 ), sucrose concentration (X 2 ), and soy protein concentrate concentration (X 3 ) for C-K production was made with 17 experimental runs ( Table 1). The final polynomial equations derived from regression equations using analysis of variance (ANOVA) in terms of coded and actual factors, which were represented by the following equations : The ANOVA for the model and factor is indicated in Table 2. The probability value (P value) indicated the significance of the interaction between each independent variable. The regression model was statistically significant (p < 0.05), meaning that it could accurately predict the results of ANOVA for each variable (rice straw, sucrose, and soy protein concentrate concentration). According to the ANOVA, the regression coefficient (R 2 ) of the model was calculated to be 99.6%, indicating that the model had adequately represented the real relationships between the parameters chosen.
The interactive relationships between the concentrations of rice straw, sucrose, and soy protein concentrate for C-K production are represented using three-dimensional contour plots in Fig. 3. The optimal values of the three variables in coded units were X 1 = 0, X 2 = 0, and X 3 = 0. Therefore the optimal concentrations of rice straw, sucrose, and soy protein concentrate for C-K production during fermentation by A. tubingensis were 2 g/l rice straw, 10 g/l sucrose, and 10 g/l soy protein concentrate, respectively.
C-K production (0.22 g/l) and molar conversion (91.0%) using 2 g/l rice straw, 10 g/l sucrose and 10 g/l soy protein concentrate as the optimal inducer and carbon and nitrogen sources, respectively, were increased by 2.5-

Fig. 2. Effects of carbon and nitrogen source on C-K production by fermentation by A. tubingensis. (A) Effect
of carbon source. (B) Effect of nitrogen source. The substrate AGE at 1 g/l was added at 72 h. Data were examined in duplicate, and the error bar represents the standard deviation.

Run
Factors a C-K (g/l)  2.1-fold compared to the C-K production (0.087 g/l) and molar conversion (43.5%) using 20 g/l citrus pectin and 10 g/l corn steep solid before the optimization. These results demonstrate that the inducer and the types and concentrations of the carbon and nitrogen sources are important for increasing C-K production by A. tubingensis. Moreover, corn steep solid is limited as a food source, whereas soy protein concentrate is not limited.

Effects of Feeding Timing and Concentration of AGE on C-K Production via Fermentation by A. tubingensis
To optimize the feeding timing of AGE as a substrate for C-K production, 1 g/l AGE was added to fermentation broth at different timepoints from 24 to 84 h (Fig. 4A). C-K production was the highest by feeding at 60 h, with a molar conversion of 97.2%. Subsequently, the effect of AGE concentration on C-K production at 60 h of feeding timing was investigated by varying the concentration from 1 to 16 g/l (Fig. 4B). An increase in AGE concentration resulted in a decrease in conversion, whereas an increase in AGE concentration within 9 g/l led to a proportional increase in C-K production. However, above 9 g/l AGE, the C-K production decreased with increasing AGE concentration. These results indicate that the optimal AGE concentration at the optimal feeding timing is 9 g/l.

Effect of Feeding Frequency of AGE on C-K Production via Fermentation by A. tubingensis
When a high concentration of substrate has a negative effect, it can be overcome via intermittent feeding [26]. Intermittent feeding was performed by feeding AGE two to three times to increase C-K concentration and productivity. Table 3 shows the C-K production during intermittent feeding and AGE concentration during fermentation. The optimal feeding timings were selected to be 48, 60, and 72 h because the molar conversion was   (Fig. 4A). The molar conversion, concentration, and productivity of C-K obtained from feeding two times with 4.5 g/l AGE at 48 and 60 h, with a total of 9 g/l AGE (Run 8 of Table 3), were 110% higher than those fed one time with 9 g/l AGE at 60 h (Fig. 4B). When 8 g/l AGE was fed two times at 48 and 60 h, with a total concentration of 16 g/l (Run 14), C-K concentration (2.47 g/l) and productivity (17.1 mg/ l /h) were the highest among those obtained by feeding of AGE. This feeding type resulted in a 4.9-fold increase in the concentration and productivity compared to feeding one time with 16 g/l AGE. The concentration and productivity of C-K after optimizing the feeding increased 11-fold compared to those before optimization (0.22 g/l and 1.52 mg/ l / h, respectively). C-K production by feeding three times with AGE at 48, 60, and 72 h was also examined. When a total of more than 16 g/l AGE was fed three times, C-K production was reduced compared to feeding two times. For instance, when a total of 16 g/l AGE was added three times (6 g/l AGE at 48 h, 6 g/l AGE at 60 h, and 4 g/l AGE at 72 h), the concentration and productivity of C-K were reduced to 94.7% compared to those obtained by feeding two times with 8 g/l AGE at 48 and 60 h. Meanwhile, the concentration and productivity of C-K fed with a total of 24 g/l AGE three times (8 g/l AGE at 48, 60, and 72 h) were reduced to 42% compared to those obtained by feeding two times with 12.5 g/l AGE at 48 and 60 h. These results indicate that feeding two times is more effective for C-K production than feeding one time and three times. Moreover, feeding with 8 g/l AGE at 48 and 60 h is optimal for C-K production.

C-K Production from PPD-Type Ginsenosides in AGE via Fermentation by A. tubingensis under Optimized Conditions
AGE contained 38.5% (w/w) PPD-type ginsenosides, which consisted of Rb1 (204 mg/g), Rb2 (12 mg/g), Rc (74 mg/g), and Rd (95 mg/g). Fermentation for the biotransformation of PPD-type ginsenosides in AGE to C-K was performed for 156 h in the optimized fermentation medium with shaking at 150 rpm at 28°C, with feeding two times with 8 g/l AGE at 48 h and 60 h. After 144 h of fermentation, C-K production reached a plateau (Fig. 5A). A. tubingensis produced 2.47 g/l C-K at 144 h, with a molar conversion of 64.3%, productivity of 17.1 mg/l/h, and dry cell weight of 8.78 g/l. The extracellular enzymes of A. tubingensis showed glycoside-hydrolyzing pathways as follows: Rb1 → Rd → F2 → C-K, Rb2 → C-O or Rd → C-Y or F2 → C-K, and Rc → Rd or C-Mc1 → F2 or C-Mc → C-K [20]. Rb1, Rb2, and Rc in AGE supplied at 48 and 60 h were converted into Rd, which accumulated to approximately 5 mM at 96 h (Fig. 5B) and the accumulated Rd was converted into C-K via F2. Rb2 and Rc in AGE were converted into C-K via C-O and C-Y; and C-Mc1 and C-Mc using alternative pathways, respectively. During the fermentation by A. tubingensis, further hydrolysis of the formed C-K was not found. After 144 h, the residual concentrations of byproduct ginsenosides followed the order Rd The production of C-K from PPD-type ginsenosides in the ginseng extracts via fermentation is shown in Table 4. Paecilomyces bainier sp. 229 produced 1.25 g/l C-K via fermentation for 144 h, with a productivity of 8.6 mg/l/h and a molar conversion of 82.6%, all of which were the previously highest recorded values for C-K [15]. The concentration (2.47 g/l) and productivity (17.1 mg/l/h) of C-K obtained via fermentation by A. tubingensis under the optimized conditions were approximately 2-fold higher than those of P. bainier sp. 229. A. tubingensis also produced 1.87 g/l C-K at 144 h, with a productivity of 12.9 mg/l/h and a molar conversion of 86.4%. These results showed a higher concentration, productivity, and conversion than those of P. bainier sp. 229. Among the GRAS fungi, Aspergillus niger FMBS 494 produced the highest concentration (0.04 g/l) and productivity