Microwave- and Ultrasound-Assisted Extraction of Cucurbitane-Type Triterpenoids from Momordica charantia L. Cultivars and Their Antiproliferative Effect on SAS Human Oral Cancer Cells

Cucurbitane-type triterpenoids are a major class of bioactive compounds present in bitter melon. In the present study, six different cultivars of bitter melon were extracted by using microwave- or ultrasound-assisted techniques to identify the prominent method that can extract the majority of cucurbitane-type triterpenoids. A UHPLC–MS/MS (ultra-high-performance liquid chromatography tandem mass spectrometry) system was used for the identification and quantification of ten cucurbitane-type triterpenoids. The results suggest that the use of microwave-assisted extraction on cultivars 4 and 5 produced higher amounts of the selected cucurbitane-type triterpenoids. The interpretation of principal component analysis also identified that cultivar 4 is significantly different from the others in which the compounds 3β,7β,25-trihydroxycucurbita-5,23(E)-dien-19-al and momordicine I were found in higher quantities. Upon further evaluation, it was also identified that these two triterpenoids can act as antiproliferative agents due to their effects on SAS human oral cancer cell lines.


Introduction
Even with the significant improvements and advancements in therapies in recent years, cancer remains the second leading cause of death. According to the latest global statistics, it is estimated that more than 22 million new cancer cases will develop by 2030. Oral squamous cell carcinoma (OSCC) is one of the common cancers in Taiwan and worldwide [1,2]. The death rate associated with oral cancer is significantly higher than that of other cancers, not due to the fact that it is difficult to diagnose or discover but due to the cancer being routinely discovered late in its development [2,3]. Even though potential target-based therapies are available in proper dosing or as combination treatment protocols, prepared by a Direct-Q ® 3 UV water purification system (Merck, Darmstadt, Germany). Triterpenoid standards (Figure 1), momordicoside L, momordicoside K, momordicine I, momordicoside F 1 , momordicoside F 2 , 3β,7β,25 trihydroxycucurbita-5,23(E)-dien-19-al, and momordicoside I were purchased from Starsci Biotech Co., Ltd. (Taipei, Taiwan). Momordicoside A was purchased from ALB Technology (Mongkok Kowloon, Hong Kong). Momordicoside G and momordicoside I aglycone were purchased from Grand Chemical (Bangkok, Thailand).

Plant Materials
The fruits of wild bitter melon (Momordica charantia Linn. var. abbreviata Ser.) Hualien No. 1 to 6 were provided by the Hualien District Agricultural Research and Extension Station, Hualien, Taiwan. The material was cleaned to remove any dirt particles with water and was lyophilized and ground to form a fine powder. The powder of bitter melon Hualien No. 4 was directly purchased from Aquavan technology Co., Ltd., Taipei, Taiwan. (i) momordicoside F 1 ; (j) momordicoside G. Structures were drawn using ChemDraw 20.

Plant Materials
The fruits of wild bitter melon (Momordica charantia Linn. var. abbreviata Ser.) Hualien No. 1 to 6 were provided by the Hualien District Agricultural Research and Extension Station, Hualien, Taiwan. The material was cleaned to remove any dirt particles with water and was lyophilized and ground to form a fine powder. The powder of bitter melon Hualien No. 4 was directly purchased from Aquavan technology Co., Ltd., Taipei, Taiwan.

Ultrasound-Assisted Extraction
An amount of 0.5 g of bitter melon powder was weighed and mixed with 40 mL of methanol. This mixture was sonicated (Branson 5800, Brookfield, CT, USA) for 30 min at 25 • C and was then centrifuged (Beckman Allegra X-15R, Brea, CA, USA) at 4000 rpm for 15 min, according to a method established by Ma et al. (2012) [19]. After centrifugation, the supernatant was drained and collected, and the procedure was repeated five times. The collected mother liquor was concentrated until the volume was less than 40 mL, and the volume was adjusted to 40 mL with methanol.

Microwave-Assisted Extraction
Microwave-assisted extraction was performed using a Microwave Digestion System (MARS 6, CEM, Matthews, NC, USA). An amount of 0.5 g of bitter melon powder was mixed with 50 mL of methanol in 100 mL MARSXpress plus vessels (CEM, Matthews, NC, USA). The temperature was ramped up in 5 min and was held at 80 • C for 5 min, where the oven power was set at 600 W. After this process, the samples were transferred to centrifuge tubes and centrifuged at 4000 rpm for 10 min. The final volume of the collected mother liquor was made up of methanol to 50 mL.

Solid Phase Extraction Process Test
The obtained extracts were then further cleaned up by reversed-phase C-18 solid-phase extraction (SPE) cartridges (500 mg/6 mL, Agela Technologies, Torrance, CA, USA). SPE cartridges were conditioned with 6 mL methanol and 6 mL ultrapure water. An amount of 0.5 mL of 100 ppb triterpenoid standards was loaded onto a preactivated SPE cartridge and then eluted out at 10% intervals for 6 mL methanol to evaluate the effect of SPE cleanup. Each fraction was collected, pooled, and concentrated. On other hand, 0.5 mL of the obtained extracts was loaded individually onto a preactivated SPE column and washed with 6 mL 30% methanol. The elution was carried out with 6 mL 100% methanol, and the fraction was collected and concentrated. The concentrates were dissolved in 1 mL LC-MS grade methanol and filtered through a 0.22 µm nylon membrane filter. Both of the concentrated fractions were injected into the UHPLC-MS/MS system to analyze the cucurbitane-type triterpenoids present in them.

UHPLC-MS/MS Analysis
UHPLC-MS/MS analysis was performed by using a Waters Acquity UPLC system coupled to a Xevo TQ-S tandem quadrupole MS (Waters Corporation, Milford, MA, USA). An Acquity UPLC BEH C18 column (100 × 2.1 mm, 1.7 µm, Waters Corporation, Milford, MA, USA) was used. The column temperature was maintained at 35 • C, and the injection volume was 2 µL at a flow rate of 0.3 mL/min. The mobile phase consisted of 0.01% formic acid in 5% methanol (A) and 0.01% formic acid in methanol (B). The elution was performed by using gradient programming as follows: 72% B at 0-1 min, 72-74% B at 1-2 min, 74% B at 2-6 min, 74-85% B at 6-7 min, 85% B at 7-8 min, 85-99% B at 8-9.5 min, 99% B at 9.5-10.5 min, and 99-72% B at 10.5-11 min. The instrument was operated using electrospray source ionization (ESI) in positive mode. The MS capillary voltage was set at 3.0 kV. Desolvation and cone gas flows were set at 900 and 150 L/h, and the desolvation temperature was maintained at 450 • C. Detection was performed in the multiple-reaction-monitoring mode; the transitions, cone voltage, and collision energy used are listed in Table 1. Instrument control, data acquisition, and evaluation were performed with MassLynx 4.1 software (Waters, Milford, MA, USA). The contents of 10 cucurbitane-type triterpenoids were expressed in the dried plant material (µg/g).

Method Validation Assay
An amount of 1 mg of each triterpenoid standard was accurately weighed and dissolved in LC-MS grade methanol to make a concentration of 1.0 mg/mL and was stored at −30 • C until required for further analysis. The serial dilution method was used to prepare different concentrations of the standards ranging from 2 µg/mL to 200 µg/mL. We further diluted the standard solutions to provide a series of concentrations until signal-to-noise (S/N) > 3 to determine the limits of detection (LOD) and S/N > 10 to determine the limits of quantification (LOQ). For evaluating intraday accuracy and precision, five samples were spiked standards with similar concentrations as that of the 10 cucurbitane-type triter- penoids in extract of Hualien No. 6 on the same day. Interday accuracy and precision were evaluated by spiking three samples over a period of three consecutive days. The final concentration levels of cucurbitane-type triterpenoids and momordicoside I aglycone were identified as 50 ppb and 10 ppb, which covered the entire specified linear range. The precision was calculated by analyzing the coefficient of variation (CV%). The accuracy was determined by analyzing percent recovery. All samples were filtered through a 0.22 µm nylon membrane filter before injecting into the UHPLC-MS/MS system.

Cell Culture
SAS, a high-grade tumorigenic human tongue squamous cell carcinoma, was used for preliminary experiments, which was obtained from the Japanese Collection of Research Bioresources (Tokyo, Japan) [20]. The cells were cultured in DMEM (10% FBS, 1.2 g/L NaHCO 3 , 0.5 mM C 3 H 3 NaO 3 , and 2.5 mM L-glutamine). The primary human skin fibroblasts cells (HFB cells) were kindly provided by Dr. Pan-Chyr Yang of National Taiwan University (Taipei, Taiwan). The cells were routinely maintained in DMEM supplemented with 10% FBS. All cells were grown in a 5% CO 2 humidified incubator at 37 • C.

Cell Proliferation Assay
The SAS cells were seeded into a 96-well culture plate at 1.3 × 10 6 cells/well, and HFB cells were seeded at 1.26 × 10 6 cells/well. After incubation in DMEM at 37 • C under a humidified 5% CO 2 for 24 h, the cells were treated with 20 and 40 µM 3β,7β,25trihydroxycucurbita-5,23(E)-dien-19-al or momordicine I dissolved in DMSO, while the control group was treated with DMSO. Real-time monitoring of the proliferation of cells was performed with the xCELLigence system (ACEA Biosciences, Inc., San Diego, CA, USA), and the cell number was expressed as the cell index.

Statistical Analysis
The data are presented as the means ± standard deviation (SD). A t-test or one-way ANOVA with Duncan's post hoc test was employed. A p value of <0.05 was considered statistically significant. Principal component analysis (PCA) was used to interpret the differences between the samples according to the cucurbitane-type triterpenoids contents of bitter melon. Statistical analysis was performed using the SPSS 21 software.

Solid Phase Extraction Process
Elution of 10 cucurbitane-type triterpenoids was observed with different concentrations of methanol (10-100%), as shown in Figure S1. Overall, 10 cucurbitane-type triterpenoids were not detected with elution of 10, 20, and 30% methanol. A small amount of momordicoside A was eluted in 40% methanol, while most of momordicoside A could be eluted in 50% methanol. However, momordicoside G, momordicoside F 1 , and 3β,7β,25trihydroxycucurbita-5,23 were completely eluted in 90% methanol, while the other six cucurbitane-type triterpenoids could be eluted in 60-80% methanol. Therefore, the solidphase extraction column was washed with 30% methanol, and 100% methanol was used as the elution buffer in subsequent sample clean-up experiments. The elution concentration of cucurbitane-type triterpenoids from bitter melon was similar to those of previous studies [21,22]. The retention time sequence was cucurbitane-type triterpenoid with five > four > three glucoside [23]. Polar compounds such as Momordica A eluted faster through reversed-phase SPE columns.

Method Validation
We analyzed the 10 cucurbitane-type triterpenoids from bitter melon using UHPLC-MS/MS with the multiple-reaction-monitoring (MRM) mode. A BEH C18 column with a mobile phase system, consisting of 0.01% aqueous formic acid in methanol and 100% methanol, successfully separated 10 analytes in 15 min, with a good resolution. The UHPLC-MS/MS chromatogram is shown in Supplementary Figure S2. All cucurbitanetype triterpenoids were analyzed with sodium adduct [M + Na] + as the parent ion in ESI positive mode. The identification of all analytes was achieved from the retention time, molecular ion, and fragmentation pattern for the quantitative determination. The analytical method showed that the results had good linearity (r > 0.9917-0.9991), ranging from 2 ng/mL to 200 ng/mL, as shown in Table 2. The results reveal that the 10 cucurbitane-type triterpenoids could be reliably determined; LOD and LOQ values were 0.125-10 ng/mL and 0.25-15 ng/mL, respectively. The intraday recovery and precision of the 10 cucurbitanetype triterpenoids ranged from 91.9% to 107.8% and from 3.68% to 12.59%, respectively (data not shown). The interday precision was 7.77-14.69% ( Table 2). The interday recovery of the 10 cucurbitane-type triterpenoids ranged from 85.5% to 115.3%, which indicated that the matrix effects of these analytes were negligible. All calculated intra-and interday precision rates were <15%. The results indicated good linearity and accuracy over the selected range. These LOD and LOQ results were similar to those of other LC/MS/MS detection methods in the literature [19,24], and the LOD and LOQ values were much lower than HPLC/ELSD [25]. This is the first report of the qualification of 10 terpenoids in bitter melon by UHPLC-MS/MS.

Comparison of Ultrasound-Assisted Extraction and Microwave-Assisted Extraction
Since ultrasound-assisted extraction is simple, inexpensive, safe, and more efficient than conventional extraction techniques, the extraction of triterpenoids carried out by ultrasound was common [26]. However, when compared with microwave-assisted extraction, ultrasound-assisted extraction is less robust, and the particle size is also a critical factor [27]. Zheng et al. (2020) [28] showed that the triterpenoid yield increased as the extraction temperature changed from 40 • C to 80 • C and then decreased above 80 • C by microwave. The bitter melon extracts were heated at various temperatures, including 30, 60, and 100 • C, and the samples were collected at 5 min intervals for saponin. The levels of momordicoside F 1 , momordicoside F 2 , momordicoside I, momordicoside K, momordicoside L, 3β,7β,25-trihydroxycucurbita-5, 23(E)-dien-19-al, and momordicine I remained stable during the 5 min of treatment at 30, 60, and 100°C temperatures. However, heating at 100 • C might change the levels of momordicoside F 2 , compared with those at lower temperatures. The levels of 3β,7β,25-trihydroxycucurbita-5,23(E)-dien-19-al were significantly reduced after 10 min at 100 • C [22]. Therefore, the microwave-assisted ex-Foods 2022, 11, 729 7 of 13 traction condition was 80 • C in this study. Our preliminary experimental results showed that the microwave extraction time for 2, 5, and 10 min did not affect the content of total cucurbitane-type triterpenoids from bitter melon (data not shown). To compare the effects of ultrasonic extraction and microwave-assisted extraction on the amount of triterpenoid in bitter melon extract, the amounts of the content of 10 bitter melon triterpenoids were added up as total triterpenoid content, as shown in Figure 2. The triterpenoid content of bitter melon extracted by microwave-assisted extraction was significantly higher than that of bitter melon extracted by ultrasonic extraction. Among the 10 targeted triterpenoids, the contents of momordicoside L, momordicoside K, momordicine I, momordicoside F 1 , and momordicoside G were significantly higher in microwave-assisted extracted bitter melon than in ultrasonic-extracted bitter melon, while the concentrations of momordicoside A, 3β,7β,25-trihydroxycucurbita-5,23(E)-dien-19-al, momordicoside I, and momordicoside F 2 showed no significant difference between the ultrasonic-extracted and microwave-assisted extracted group (Figure 2). This result suggested that microwave-assisted extraction had a higher efficiency in bitter melon triterpenoid extraction than ultrasound-assisted extraction. The extraction efficiency of microwave-assisted extraction is related to the mechanical effects of the rotation of molecules, which are due to a microwave-induced dipole, causing the migration of the ions, which can improve the penetration of the solvent into the matrix. This disrupts the cell wall, thereby releasing the intracellular products, permitting the dissolution of components that are to be extracted [28,29]. Microwave-assisted extraction has been used for the extraction of Ginseng and triterpene triterpenoids from the defatted residue of yellow horn (Xanthoceras sorbifolia Bunge) [30,31]. Chen et al. (2007) [31] found that, in comparison with shaking extraction, supercritical fluid carbon dioxide extraction, heat reflux extraction, and ultrasound-assisted extraction, microwave-assisted extraction had the highest efficiency in extracting total triterpenoid triterpenoids from Ganoderma atrum [30]. However, a previous study suggests that cucurbitane-type triterpenoids, such as 3β,7β,25-trihydroxycucurbita-5,23 (E)-dien-19-al and momordicine I, were extremely sensitive at 100 • C [22]; hence, temperatures below 100 • C are recommended for such compounds through the microwave-assisted extraction technique. critical fluid carbon dioxide extraction, heat reflux extraction, and ultrasound-assisted ex-traction, microwave-assisted extraction had the highest efficiency in extracting total triterpenoid triterpenoids from Ganoderma atrum [30]. However, a previous study suggests that cucurbitane-type triterpenoids, such as 3β,7β,25-trihydroxycucurbita-5,23 (E)-dien-19-al and momordicine I, were extremely sensitive at 100 °C [22]; hence, temperatures below 100 °C are recommended for such compounds through the microwave-assisted extraction technique. Figure 2. The microwave-assisted extraction efficiency of cucurbitane-type triterpenoids (based on ultrasonic extraction). The concentration of the nine triterpenoids in bitter melon extracted by microwave was divided by the concentration of the nine triterpenoids in Hualien No. 3 bitter melon extracted by ultrasound individually and presented as a percentage. All data are reported as the mean (± standard deviation) of three separate experiments. Statistical analysis was performed using a t-test, with significant differences determined at the level of * p < 0.05.

Application of Microwave-Assisted Extraction on Six Different Wild Bitter Melons
The bitter melon fruit powders of Hualien No. 1 to 6 were extracted with methanol using microwave and analyzed by the UHPLC-MS/MS system. Upon investigation, Hualien No. 4 and 5 were found to contain maximum triterpenoid concentration, at 1509.71 ± 55.97 µg/g and 1150.31 ± 52.06 µg/g, while Hualien No. 6 was observed to have the lowest content, at 90.58 ± 7.57 µg/g (Table 3). Momordicoside A was found to be the most abundant triterpenoid in Hualien No. 1, 2, 3, and 5, with concentrations of 358.59 ± 18.12 µg/g, 339.61 ± 31.92 µg/g, 195.55 ± 6.88 µg/g, and 1261.6 ± 51.54 µg/g, while momordicine I was the abundant bioactive in Hualien No. 4 (470.01 ± 25.03 µg/g) and 6 (43.93 ± 3.60 µg/g). The composition of triterpenoids was similar to that of Hualien No. 1 and 2, except the Figure 2. The microwave-assisted extraction efficiency of cucurbitane-type triterpenoids (based on ultrasonic extraction). The concentration of the nine triterpenoids in bitter melon extracted by microwave was divided by the concentration of the nine triterpenoids in Hualien No. 3 bitter melon extracted by ultrasound individually and presented as a percentage. All data are reported as the mean (± standard deviation) of three separate experiments. Statistical analysis was performed using a t-test, with significant differences determined at the level of * p < 0.05.

Application of Microwave-Assisted Extraction on Six Different Wild Bitter Melons
The bitter melon fruit powders of Hualien No. 1 to 6 were extracted with methanol using microwave and analyzed by the UHPLC-MS/MS system. Upon investigation, Hualien No. 4 and 5 were found to contain maximum triterpenoid concentration, at 1509.71 ± 55.97 µg/g and 1150.31 ± 52.06 µg/g, while Hualien No. 6 was observed to have the lowest content, at 90.58 ± 7.57 µg/g (Table 3). Momordicoside A was found to be the most abundant triterpenoid in Hualien No. 1, 2, 3, and 5, with concentrations of 358.59 ± 18.12 µg/g, 339.61 ± 31.92 µg/g, 195.55 ± 6.88 µg/g, and 1261.6 ± 51.54 µg/g, while momordicine I was the abundant bioactive in Hualien No. 4 (470.01 ± 25.03 µg/g) and 6 (43.93 ± 3.60 µg/g). The composition of triterpenoids was similar to that of Hualien No. 1 and 2, except the contents of momordicine I and 3β,7β,25-trihydroxycucurbita-5,23(E)dien-19-al which were found to be higher in No. 2. In Hualien No. 3, the concentration of momordicoside L was significantly higher than that of the other five bitter melon Hualien cultivars. This is the first report on the analysis of the triterpenoid contents in different varieties of Momordica charantia. These data demonstrated that the compositions of triterpenoids in different bitter melon cultivars were also different.
We further analyzed the content of cucurbitane-type triterpenoids (excluding momordicoside I aglycone) of bitter melon through multivariate principal component analysis. The results show that there were three principal components with eigenvalues greater than 1. When the three principal components replaced the original variables, 93.3% of the total variation from the original data could be explained. Among the three principal components, PC1 and PC2 had the largest contribution to the total variation. Their ability to explain the total variation from the original data was 48.6% and 26.7%, respectively.

3β,7β,25-Trihydroxycucurbita-5,23(E)-dien-19-al and Momordicine I Suppressed the Proliferation of SAS Cells
The effects of 3β,7β,25-trihydroxycucurbita-5,23(E)-dien-19-al and momordicine I on the cell growth of SAS cells (oral squamous cell carcinoma) were evaluated. The results indicate that 20 µM of 3β,7β,25-trihydroxycucurbita-5,23(E)-dien-19-al or momordicine I was not significantly different from the control group within 24 h (Figure 4a). Therefore, the dosage was increased to 40 µM 3β,7β,25-trihydroxycucurbita-5,23(E)-dien-19-al or momordicine I, but both were significantly lower than the cell index of the control group after 18 h. Additionally, there was no suppression of cell growth observed after the treatment of 3β,7β,25-trihydroxycucurbita-5,23(E)-dien-19-al and momordicine I in HFB cells (Figure 4b). These results suggested that 3β,7β,25-trihydroxycucurbita-5,23(E)-dien-19-al and momordicine I suppressed the proliferation of SAS cells and had no adverse effect on the cell growth of HFB cells at 40 µM. In addition, 3β,7β,25-trihydroxycucurbita-5,23(E)-dien-19-al was found to have the ability to suppress the proliferation of MCF-7 and MDA-MB-231 breast cancer [35]. Further reports also suggest that momordicine I can inhibit HNC cell (JHU022, JHU029, and Cal27) proliferation involving c-Met and downstream signaling. Studies also report that momordicine I exhibited similar activities to prevent HNC tumor growth in mice, with no apparent side effects [36]. In addition, Sur and Ray (2021) [37] found that bitter melon extract could induce cell death, inhibit cell proliferation and metabolism, and enhance the immune defense system in the prevention of OSCC in vitro and in vivo. The effect of 3β,7β,25-trihydroxycucurbita-5,23(E)-dien-19-al or momordicine I as therapeutics requires further experiments to confirm.

3β,7β,25-Trihydroxycucurbita-5,23(E)-Dien-19-Al and Momordicine I Suppressed the Proliferation of SAS Cells
The effects of 3β,7β,25-trihydroxycucurbita-5,23(E)-dien-19-al and momordicine I on the cell growth of SAS cells (oral squamous cell carcinoma) were evaluated. The results indicate that 20 µM of 3β,7β,25-trihydroxycucurbita-5,23(E)-dien-19-al or momordicine I was not significantly different from the control group within 24 h (Figure 4a). Therefore, the dosage was increased to 40 µM 3β,7β,25-trihydroxycucurbita-5,23(E)-dien-19-al or momordicine I, but both were significantly lower than the cell index of the control group after 18 h. Additionally, there was no suppression of cell growth observed after the treat- HNC tumor growth in mice, with no apparent side effects [36]. In addition, Sur and Ray (2021) [37] found that bitter melon extract could induce cell death, inhibit cell proliferation and metabolism, and enhance the immune defense system in the prevention of OSCC in vitro and in vivo. The effect of 3β,7β,25-trihydroxycucurbita-5,23(E)-dien-19-al or momordicine I as therapeutics requires further experiments to confirm. .

Conclusions
In summary, microwave-assisted extraction increases the efficiency of triterpenoid extraction in bitter melon in comparison with ultrasound-assisted extraction. Via the use of microwave-assisted extraction, bitter melon Hualien No. 4 had the highest amount of 3β,7β,25-trihydroxycucurbita-5,23(E)-dien-19-al, momordicine I, and momordicoside F 2 among the other six bitter melon cultivars. Furthermore, 3β,7β,25-trihydroxycucurbita-5,23(E)-dien-19-al and momordicine I were found to suppress the proliferation of SAS cells but did not suppress the proliferation of HFB cells. However, further investigation is required to fully understand the effects of 3β,7β,25-trihydroxycucurbita-5,23(E)-dien-19-al and momordicine I on SAS and HFB cell lines.