Ultrasound-Assisted Extraction Optimization of α-Glucosidase Inhibitors from Ceratophyllum demersum L. and Identification of Phytochemical Profiling by HPLC-QTOF-MS/MS

Ceratophyllum demersum L. (CDL) is a traditional Chinese herb to treat many diseases, but research on its anti-diabetic activity is not available. In this research, the α-glucosidase inhibitory ability and phytochemical constituents of CDL extract were firstly studied. Optimal ultrasound-assisted extraction conditions for α-glucosidase inhibitors (AGIs) were optimized by single factor experiment and response surface methodology (RSM), which was confirmed as 70% methanol, liquid-to-solid ratio of 43 (mL/g), extraction time of 54 min, ultrasonic power of 350 W, and extraction temperature of 40 °C. The lowest IC50 value for α-glucosidase inhibition was 0.15 mg dried material/mL (mg DM/mL), which was much lower than that of acarbose (IC50 value of 0.64 mg DM/mL). In total, 80 compounds including 8 organic acids, 11 phenolic acids, 25 flavonoids, 21 fatty acids, and 15 others were identified or tentatively identified from CDL extract by HPLC-QTOF-MS/MS analysis. The results suggested that CDL could be a potential source of α-glucosidase inhibitors. It can also provide useful phytochemical information for research into other bioactivities.


Introduction
α-Glucosidase is a vital carbohydrate hydrolase situated in the brush border surface membrane of the small intestine, which is involved in the last step of carbohydrate digestion by hydrolyzing the α-(1,4) glycosidic bond to release glucose at the non-reducing end [1]. α-Glucosidase inhibitors (AGIs) can effectively alleviate the release of glucose from dietary carbohydrates and delay the absorption of glucose by inhibiting the action of α-glucosidase, resulting in delayed postprandial blood glucose level [2]. Currently, acarbose, miglitol, and voglibose are the commonly used AGIs to treat diabetes and its complications, but these drugs exhibit toxic side effects, such as flatus, diarrhea, abdominal colic, and so on [3]. At present, numerous studies have proved that many plant extracts possess the potential to be excellent sources of AGIs, with the advantages of being natural, highly-efficient, inexpensive, and with low toxicity. Moreover, many highly active AGIs have been isolated and identified, such as flavones, phenolic acids, alkaloids, terpenes, anthocyanins, glycosides, and so on [4,5]. Zhang et al. [6] evaluated the α-glucosidase inhibitory activity of four Acer species leaves, and the IC 50 values were 167-433 fold of that of acarbose; hydrolysable tannins were the major The recovery of bioactive compounds varied greatly with the changes of solvent polarity. Therefore, the influence of different concentrations of methanol on the extraction of AGIs from CDL was evaluated separately; the result is given in Figure 1. All extracts had considerable α-glucosidase inhibition in the sample concentration range of 0.17-2.5 mg DM/mL and exhibited an obvious doseeffect relationship. The 70% methanol extract possessed the best α-glucosidase inhibition with the lowest IC50 value of 0.17 mg DM/mL. The inhibition was 3.7 times higher than acarbose (0.76 mg/mL), a clinical diabetes treatment drug, indicating the hypoglycemic potential of CDL extracts (Figure 1a). Analysis of total phenolic content (TPC) and total flavonoid content (TFC) indicated that 30% methanol extract possessed the highest TPC, with the value of 3.76 mg GAE/g DM. The highest TFC was found in 70% methanol extract (27.88 mg quercetin equivalents per gram of dried material (mg QuE/g DM,)). The 10% methanol extract possessed the lowest TPC and TFC, which were only 3.11 mg gallic acid equivalents per gram of dried material (mg GAE/g DM) and 0.23 mg QuE/g DM, respectively (Figure 1b). These indicated that the medium polar solvent is more suitable for extracting phenols in CDL, and the weak polar solvent is suitable for extracting flavonoids. Correlation coefficient analysis (Table S1) revealed that the flavonoids in CDL correlated well (r = −0.648) with the α-glucosidase inhibition, so flavonoids could be the major contributor to the α-glucosidase inhibition of CDL. Thus, 70% methanol was selected for further extraction of AGIs from CDL.

Preliminary Screening of Each Single Factor Analysis
Extraction temperature, time, ultrasound power and liquid-to-solid ratio also played an important role in the recovery of bioactive constituents. Generally, the more solvents, the higher mass transfer efficiency and extraction rate, but too many solvents cause solvent waste and increase the extraction cost [25]. As shown in Figure 2a, the IC50 value of extracts decreased with increasing liquidto-solid ratio with the minimum value detected at 40 mL/g, but a slight increment was observed when the ratio was set at 50 mL/g. In Figure 2b, increased ultrasonic power (250-350 W) resulted in increased α-glucosidase inhibition of extracts. Further increasing ultrasonic power resulted in reduced α-glucosidase inhibition. Therefore, 350 W was considered to be the optimal ultrasound power due to the highest α-glucosidase inhibition and relatively low energy consumption. Reasonable extraction time can facilitate the contact between solvent and raw material, which is beneficial to the release of target compounds, and increase the extraction rate [26], but continuous heating is not conducive for retention of activity. As shown in Figure 2c, the sample extracted for 60 min gave the strongest α-glucosidase inhibition.

Preliminary Screening of Each Single Factor Analysis
Extraction temperature, time, ultrasound power and liquid-to-solid ratio also played an important role in the recovery of bioactive constituents. Generally, the more solvents, the higher mass transfer efficiency and extraction rate, but too many solvents cause solvent waste and increase the extraction cost [25]. As shown in Figure 2a, the IC 50 value of extracts decreased with increasing liquid-to-solid ratio with the minimum value detected at 40 mL/g, but a slight increment was observed when the ratio was set at 50 mL/g. In Figure 2b, increased ultrasonic power (250-350 W) resulted in increased α-glucosidase inhibition of extracts. Further increasing ultrasonic power resulted in reduced α-glucosidase inhibition. Therefore, 350 W was considered to be the optimal ultrasound power due to the highest α-glucosidase inhibition and relatively low energy consumption. Reasonable extraction time can facilitate the contact between solvent and raw material, which is beneficial to the release of target compounds, and increase the extraction rate [26], but continuous heating is not conducive for retention of activity. As shown in Figure 2c, the sample extracted for 60 min gave the strongest α-glucosidase inhibition.
With the increase of extraction temperature from 40 • C to 70 • C, a significant increase in IC 50 value was observed; the minimum α-glucosidase inhibition was detected at 70 • C (Figure 2d). Usually, a higher extraction temperature can destroy cell structure more effectively, leading to increased extraction yield [27]. However, low temperature (40 • C) is more conducive to the extraction of α-glucosidase inhibitors from CDL, therefore, 40 • C was selected as the suitable extraction temperature.
With the increase of extraction temperature from 40 °C to 70 °C, a significant increase in IC50 value was observed; the minimum α-glucosidase inhibition was detected at 70 °C ( Figure 2d). Usually, a higher extraction temperature can destroy cell structure more effectively, leading to increased extraction yield [27]. However, low temperature (40 °C) is more conducive to the extraction of α-glucosidase inhibitors from CDL, therefore, 40 °C was selected as the suitable extraction temperature.

Response Surface Analysis
Based on the results of single factorial experiments, liquid-to-solid ratio, ultrasonic power, and extraction time were chosen for further RSM analysis. The experiments were performed according to Box-Behnken design (BBD), and results are presented in Table 1  Effects of liquid-to-solid ratio (a), ultrasonic power (b), extraction time (c), extraction temperature (d), on the α-glucosidase inhibitory ability (IC 50 ) of CDL extracts.

Response Surface Analysis
Based on the results of single factorial experiments, liquid-to-solid ratio, ultrasonic power, and extraction time were chosen for further RSM analysis. The experiments were performed according to Box-Behnken design (BBD), and results are presented in Table 1. The results indicate the effect of process variables on the α-glucosidase inhibition of CDL extracts. Estimated regression coefficients for the response (IC 50 value) in the second order polynomial equations (Equation (1)) are as follows: ANOVA statistics ( Table 2) were generated to assess the goodness of fit, the significance of the model, coefficient of determination, and related probability values (p-value) [10]. The overall quadratic model, individual and interaction effects of liquid-to-solid ratio (mL/g), ultrasonic power (W), extraction time (min) are indicated by F and p-values. The p-value (<0.0001) showed that the model was statistically significant. At the same time, the values of R 2 and Adj-R 2 were 0.9798 and 0.9538, respectively, implying a strong correlation between the predicted results and actual results. Moreover, the linear effect of liquid-to-solid ratio, extraction time, and square effect of liquid-to-solid ratio, extraction power, and extraction time, were found to be significant for α-glucosidase inhibitory activity. The interaction terms of liquid-solid ratio and time have a significant effect on α-glucosidase inhibitory activity. The interaction effects of individual process variables on dependent variable (IC 50 value) were clearly studied through the pictorial representation in the form of 3D plot and 2D contour map ( Figure 3). Figure 3a illustrates that there was no significant interaction between ultrasonic power and liquid-to-solid ratio. At any liquid-to-solid ratio, the α-glucosidase inhibitory activity increased with improved ultrasonic power. As revealed by Figure 3b, when the ultrasonic power was set at 350 W, the IC 50 value decreased by simultaneous increase of liquid-to-solid ratio and extraction time. A higher α-glucosidase inhibition was obtained when the extraction time and liquid-solid ratio reached 53 min and 43 mL/g, respectively, which implied a significant interaction between the two parameters. In Figure 3c, within the scope of 40-54 min and 300-341 W, the inhibition ability of α-glucosidase increased with the sonication time and power increase, then decreased when beyond this range. According to the significance of regression coefficients, it was evident that extraction time was the most significant factor affecting the inhibitory activity, followed by liquid-to-solid ratio and ultrasonic power. Molecules 2020, 25, x FOR PEER REVIEW 7 of 23

Optimal Extraction Conditions Analysis
To obtain the maximized response of α-glucosidase inhibition, a response optimizer tool was used to determine the optimal level of the chosen variables. The lowest IC50 value of 143.88 µg DM/mL was predicted at the optimal conditions of liquid-to-solid ratio of 43 mL/g, extraction time of 54 min, and power of 340 W. Validation experiments for the predicted optimum conditions were carried out to verify the model accuracy. However, due to the limitations of actual operating conditions, the actual parameter of each variable was adjusted to 43 (mL/g), 54 min, 350 W. The experimental IC50 value was observed to be 146.23 µg DM/mL, which fitted well (98.37%) with the predicted IC50 value. This demonstrates that the developed RSM model is practicable and can be used to describe the relationship between extraction factors and α-glucosidase suppression of CDL extracts.

Optimal Extraction Conditions Analysis
To obtain the maximized response of α-glucosidase inhibition, a response optimizer tool was used to determine the optimal level of the chosen variables. The lowest IC 50 value of 143.88 µg DM/mL was predicted at the optimal conditions of liquid-to-solid ratio of 43 mL/g, extraction time of 54 min, and power of 340 W. Validation experiments for the predicted optimum conditions were carried out to verify the model accuracy. However, due to the limitations of actual operating conditions, the actual parameter of each variable was adjusted to 43 (mL/g), 54 min, 350 W. The experimental IC 50 value was observed to be 146.23 µg DM/mL, which fitted well (98.37%) with the predicted IC 50 value. This demonstrates that the developed RSM model is practicable and can be used to describe the relationship between extraction factors and α-glucosidase suppression of CDL extracts.

Analysis of Phytochemical Constituents
To investigate the major chemical components of the CDL extract giving the strongest α-glucosidase inhibition, HPLC-QTOF-MS/MS analysis was carried out. The base peak chromatogram (BPC) is shown in Figure 4. Identified or tentatively identified compounds are listed in Table 3; identities were confirmed by analyzing the fragmentation pattern of each deprotonated molecule, and by matching the data with that recorded in available references and databases. In total, 80 compounds were identified or tentatively identified, including 8 organic acids, 11 phenolic acids, 25 flavonoids, 21 fatty acids, and 15 othercompounds.
Molecules 2020, 25, x FOR PEER REVIEW 8 of 23 To investigate the major chemical components of the CDL extract giving the strongest αglucosidase inhibition, HPLC-QTOF-MS/MS analysis was carried out. The base peak chromatogram (BPC) is shown in Figure 4. Identified or tentatively identified compounds are listed in Table 3; identities were confirmed by analyzing the fragmentation pattern of each deprotonated molecule, and by matching the data with that recorded in available references and databases. In total, 80 compounds were identified or tentatively identified, including 8 organic acids, 11 phenolic acids, 25 flavonoids, 21 fatty acids, and 15 other compounds. .

Phenols and Derivatives
A total of 11 phenolic acids were characterized, which can be further classified into hydroxybenzoic acids and their derivatives.
Three hydroxybenzoic acid derivatives were identified.  [32]. The fragment ions of peak 14 at 208.0322, 193.0161, and 149.0253 resulting from the loss of CH 3 , 2 CH 3, and 2 CH 3 + COOH respectively, indicating the presence of two methyl groups and one propenoic acid moiety. So it was identified as sinapinic acid [34], and the detected fragmentation pattern is given in Figure 5b. Peaks

Phenols and Derivatives
A total of 11 phenolic acids were characterized, which can be further classified into hydroxybenzoic acids and their derivatives.
Three hydroxybenzoic acid derivatives were identified.         [6]. The detected fragmentation pattern of peak 22 was shown in Figure 5c [28]. Taking peak 65 as an example, the diagnostic fragment ion at 197.12 suggested the presence of a hydroperoxide at C11, so it was identified as 11-hydroperoxy-octadecatrienoic acid; the fragmentation pattern is shown in Figure 5f. Peaks 60, 64, and 66 with product ions at 171.10 resulted from the loss of C 9 H 14 O, indicating the hydroperoxide at C9, but this could not reveal the position of the double bonds. Peaks 67, 69, and 71 were identified as 9-hydroperoxy -octadecadienoic acid due to the MS/MS at 171.10.
Peak 72 with molecular ion at m/z 291.1980 was identified as 12-oxo-phytodienoic acid, and the fragmentation pattern is shown in Figure 5g. Its MS/MS ions at 273.1857 and 247.2078 result from the loss of a water molecule and a carboxylic residue, respectively [43]. Peak 73 (559.3142, C 28 H 48 O 11 ) was tentatively assigned as dirhamnosyl linolenic acid, fragment ion at 277.2186 resulted from the loss of a dirhamnosyl (C 10 H 18 O 9 , 282 Da) [36].

Preparation of Extracts
Fresh CDL was bought in Shuyang County, Jiangshu Province, in April 2019. The CDL was dried, pulverized into powder with a high-speed disintegrator (Hangzhou, China), and sieved through a 50 mesh screen. The plant material s moisture content was 8.2% (w/w), which was determined by measuring the weight before and after drying at 105 • C in a bake oven to a constant weight. The CDL powder was stored in a refrigerator at −20 • C until used.
Selecting a suitable solvent is very important for extracting the target product. In this research, a methanol solution was selected as the best extraction solvent after pre-experiment. The CDL powder (1 g) was suspended in 10%, 30%, 50%, 70%, and 90% methanol aqueous solution at a liquid-to-solid ratio of 20 mL/g, respectively, and then sonicated for 120 min at 50 • C, 200 W. The mixtures were centrifuged at 5000 rpm/min for 10 min, and the supernatants were collected for further analysis.

Determination of Total Phenolic and Flavonoid Content
The total flavonoid content (TFC) and total phenolic content (TPC) of different crude extracts were measured with the AlCl 3 colorimetric method and the Folin-Ciocalteu method [48] with some modifications, respectively. In the experiment of measuring TFC, 0.5 mL of properly diluted sample was mixed with 100 µL of 5% NaNO 2 for 6 min, followed by adding 100 µL 10% AlCl 3 for 6 min, then adding 1 mL 4% NaOH and 1 mL distilled water. The mixtures were incubated at room temperature for 15 min, and 200 µL of mixtures were pipetted into a 96-well plate. The absorbance was measured at 510 nm using a microplate reader (SpectraMax M2, Molecular Devices Corp., Sunnyvale, CA, USA). In the experiment of measuring TPC, 200 µL of properly diluted sample was incubated with 100 µL of Folin-Ciocalteu reagent for 5 min, followed by adding 300 µL 20% Na 2 CO 3 and 1 mL distilled water. The mixtures were incubated at room temperature for 30 min in the dark. After 2 min of centrifugation at 7000 rpm, 200 µL of supernatants were pipetted into a 96-well plate, and absorbance at 765 nm was read with a micro-plate reader. The TFC was expressed as mg quercetin equivalents per gram of dried material (mg QuE/g DM). The TPC was expressed as mg of gallic acid equivalents per gram of dried material (mg GAE/g DM.). All experiments were done in triplicate.

Single Factor Experiments
The liquid-to-solid ratio, ultrasonic power, extraction time, and extraction temperature were the major factors affecting the recovery of bioactive compounds from plant materials. The experiments were performed by changing the level of one factor and maintaining the other factors at a constant level of 70% methanol aqueous solvent, liquid-to-solid ratio at 40 mL/g, extraction time at 60 min, ultrasonic power at 300 W, and extraction temperature at 50 • C. Briefly, CDL was extracted with 70% methanol aqueous solvent in different liquid-to-solid ratios (from 10 to 50 mL/g) at different extraction times (from 40 to 120 min), ultrasonic powers (from 250 to 450 W), and temperatures (from 40 to 80 • C) controlled by a digitally-controlled ultrasonic bath (KQ-500DE, Kunshan ultrasonic instrument CO., LTD, Kunshan, China).

α-Glucosidase Inhibition Assay
The α-glucosidase inhibition was assessed using the method reported by reference [6]. All α-glucosidase and pNPG solutions were prepared with 0.1 M, pH 6.9 phosphate buffer. Different concentrations of samples (50 µL) and 50 µL of 0.1 U/mL α-glucosidase solution were incubated in 96-well plates at 25 • C for 10 min. Then, 50 µL of 5 mM pNPG solution was added and incubated for 15 min at 37 • C. Finally, the reaction was terminated with 100 µL of 0.2 M Na 2 CO 3 , and absorbance at 405 nm was recorded with a micro-plate reader. Acarbose was used as positive control. All experiments were done in triplicate. The concentration required to inhibit 50% activity of α-glucosidase (IC 50 value) was expressed as mg dried material/mL (mg DM/mL).

Statistical Optimization of UAE
RSM with BBD was used to optimize the extraction of AGIs in CDL. As shown in Table 4, three extraction variables (ratio of material to liquid: 1:30, 1:40, and 1:50 g/mL; extraction time: 40, 60, and 80 min; ultrasonic power: 300, 350, and 400 W) were chosen to evaluate the effect on response value (α-glucosidase inhibitory ability). The response variables were fitted to the following, a second order polynomial model equation: where Y is the predicted response value (α-glucosidase inhibitory ability); X i and X j are independent variables; α 0 , α i , α ii , and α ij are the constant coefficient, linear coefficient, quadratic coefficient, cross-product coefficient, respectively.

HPLC-QTOF-MS/MS Analysis
For compound separation, an Agilent 1260 HPLC infinity system (Agilent, Palo Alto, CA, USA) equipped with a DAD detector, a binary pump, and a Sun Fire C 18 column (250 × 4.60 mm, 5 µm, Waters, Milford, MA, USA) was applied. The mobile phase consisted of 0.1% formic acid in de-ionized water (A) and acetonitrile (B). The sample was eluted with a gradient from 10% B to 100% B in 35 min at a flow rate of 0.8 mL/min. The detection wavelength, column temperature, and injection volume were set at 280 nm, 35 • C, and 5 µL, respectively.
To obtain the MS and MS/MS information of detected compounds, the elutes were directly interfaced to a Hybrid Quadrupole-TOF 6600 system (AB Sciex) equipped with an electrospray ionization source (ESI). The full scan mass spectrum was detected at a mass range of m/z 100-1500 under negative ion mode. Other parameters were spray gas pressure of 50 psi, capillary voltage of 3.5 kV, ion source temperature of 550 • C, flow rate of 0.8 mL/min, and ion spray voltage floating of − 4500 V. Nitrogen and helium were used as auxiliary and collision gases, respectively. The MS data was processed by MassHunter. A molecular formula calculator was used to calculate the elemental composition of each parent and product ion. The compounds were characterized or tentatively characterized by comparing the parent ion and MS 2 fragments with those in references and database.

Statistical Analysis
Statistical analyses were carried out on SPSS 17.0 (IBM, Armonk, NY, USA) and Origin 8.0 (OriginLab, Northampton, MA, USA), all data were expressed as mean ± SD (standard deviation). The statistical analysis of the proposed regression model was analyzed by Design Expert 8.0.6 (Stat-ease INC., Minneapolis, MN, USA). Significant difference among data was performed by Tukey's-b, One-way analysis of variance (ANOVA), p < 0.05 was considered significant. The correlation between the bioactivity and content of constituents was evaluated by Pearson's correlation analysis.

Conclusions
This is the first research to optimize the extraction conditions of AGIs from CDL, and to analyze the major phytochemical constituents. The optimal extraction parameters were confirmed as extraction solvent of 70% methanol, liquid-to-solid ratio of 43 (mL/g), extraction time of 54 min, ultrasonic power of 350 W, and extraction temperature of 40 • C, under which, the strongest α-glucosidase inhibition ability was achieved (IC 50 , 146.23 µg DM/mL). In addition, 30% and 70% methanol aqueous solutions are suitable for recovering the phenolics and flavonoids in CDL, respectively. HPLC-QTOF-MS/MS analyses permitted the identification of 80 compounds, including flavonoids, phenolic acids, fatty acids, and others. The major active compounds in CDL extract are caffeic acid derivatives, ferulic acid and its derivatives, apigenin, quercetin, kaempferol, naringenin, luteolin, and catechin and their derivatives, many of which have been reported to be promising AGIs. In addition, fatty acids with 18 carbons were also identified as the main components. This study can provide a theoretical basis for the study of CDL as a natural anti-diabetic drug, and the structure and inhibition mechanism of AGIs from CDL need further study.
Supplementary Materials: The following are available online, Table S1: Correlation analysis between bioactive compounds and α-glucosidase inhibitory activity in methanol extracts of CDL.
Author Contributions: Z.L. was in charge of literature search, figures, study design, data collection, data analysis, and writing. Z.T. and H.W. were in charge of financial support and study design. L.Z. was in charge of study design, data analysis, revision of manuscripts and financial support. All authors have read and agreed to the published version of the manuscript.