采用超声辅助纤维素酶法制备NHE。该过程包括用纤维素酶酶酶水解，然后超声提取类黄酮和多酚。采用单因素实验和RSM对制备过程进行优化。为了探索不同条件的影响，我们在单因素实验中开发了四个因素和五个实验程序的水平。选择纤维素酶用量、液固比、酶水解时间和乙醇浓度等因素，确定其对黄酮含量的影响。如图1a所示，类黄酮含量首先随着液固比从20：1（mL/g）增加到40：1（mL/g），然后下降到40：1（mL/g）到60：1（mL/g）的范围内。NHE类黄酮含量的条件达到最大值，液固比为40：1（mL/g）。溶剂与固体接触面积的增加有利于活性物质的扩散[27]，因此类黄酮含量随着液固比的增加而增加。当溶剂过量时，纤维素酶的有效浓度和非目标活性物质溶解的底物浓度增加，减少纤维素酶和底物的减少，导致NHE的类黄酮含量下降[22，42]。图1b显示，类黄酮含量随着纤维素酶量从2%增加到4%而增加，但当纤维素酶的添加量高于4%时，类黄酮含量开始下降。纤维素酶可以将细胞壁分解成更小的分子，有利于活性物质的释放。过量的纤维素酶可能会产生粘稠的酶溶液，不利于破坏细胞井和酶促反应过程[43]。如图1c所示，在酶解时间60min处，类黄酮含量达到最大值。然而，类黄酮含量在60 min后没有变化，可能是由于类黄酮的扩散平衡。乙醇的浓度对活性组分的提取速率有影响。根据类似混溶性的原理，乙醇的浓度对活性组分的提取速率有影响[43]。如图1d所示，当乙醇浓度为50%时，类黄酮含量达到最大值。

Based on the results of the single factor experiments, the liquid-to-solid ratio (A), cellulase amount (B) and ethanol concentration (C) were selected as the independent variables for optimizing the extraction process using the flavonoids content as an index. The results of the Box–Behnken experiment were shown in Table 1. According to different experiments, the flavonoids content of NHE was 20.67–27.06 mg/g. As in Table 2, multiple regression analysis was used to analyze the experimental data. Through the analysis of the multivariate regression fitting, the relationship between the response variables (flavonoids content of NHE, Y) and the quadratic regression equation is shown in the formula as follows:

According to the result of regression analysis, the response surface 3D map and the contour plot maps were drawn in Figure 2. The 3D diagram can intuitively reflect the influence of the interaction of each factor on the response value, and through it, we can find the best parameters and the interactions among the parameters. According to the p-value, we can determine two interaction coefficients (AB, BC). Figure 2a shows the interaction between the liquid-to-solid ratio and the cellulase amount (AB) displayed significantly (p < 0.05). The flavonoid content increases with the increasing liquid-to-solid ratio and cellulase amount, while at a higher liquid-to-solid ratio (40:1–60:1 mL/g) and cellulase amount (4–6%), the interaction effect was not obvious. When the amount of cellulase reaches a certain point, the flavonoid content may decrease with the increase in cellulase amount and enlarge the liquid-to-solid ratio. The reason for this may be due to the appropriate cellulase amount and an enlargement of the liquid-to-solid ratio will improve the solubility and increase the mass transfer rate, but the excessive cellulase amount with too high a liquid-to-solid ratio will affect the reaction efficiency and mass transfer efficiency, resulting in a flavonoid content decrease with the liquid-to-solid ratio. A significant interaction effect (p < 0.05) between cellulase amount and ethanol concentration (BC) was also observed in Figure 2c.

According to the response surface experiment, the optimal process conditions for extraction were as follows: the liquid-to-solid ratio was 40.45:1 (mL/g), cellulase amount was 5.23% (w/w), the ethanol concentration was 52% (v/v). Under such conditions, the predicted flavonoid content of NHE was 26.92 mg/g. To verify these optimal conditions, the experiment was repeated five times and the average actual flavonoid content of NHE was 27.26 ± 0.64 mg/g (n = 5), which was close to the predicted value and proved the validity of the models.