Ultrasound-assisted extraction of withanolides from Tubocapsicum anomalum: Process optimization, isolation and identification, and antiproliferative activity

Graphical abstract

Abstract

Tubocapsicum anomalum, a Chinese medicinal plant rich in anti-tumor withanolides, requires efficient extraction methods. In this paper, an HPLC method was first established for the detection of withanolides, and gradient elution was carried out using a methanol–water solvent system. It was found that the content of withanolides was the highest in the leaves of T. anomalum, followed by the stems and fruits, and almost none in the roots. During the actual picking process, the quantity of leaves collected was relatively small, while the number of stems was the highest. Therefore, the Box-Behnken response surface method was used to optimize the ultrasonic-assisted extraction process of withanolides from the stems of T. anomalum. The optimal extraction conditions were determined as follows: the liquid–solid ratio was 20:1, the extraction solvent was 70 % ethanol, the ultrasonic power was 250 W, the ultrasonic time was 40 min, and the ultrasonic temperature was 50 °C. Under these conditions, the average yields of tubocapsenolide A (Te-A) and tubocapsanolide A (Ta-A) can reach 2.87 ± 0.12 mg/g and 1.18 ± 0.05 mg/g, respectively. We further compared extraction rates of two withanolides from different parts of T. anomalum using ultrasonic and traditional extraction methods. Ultrasonic extraction significantly increased rates, with the highest yields from leaves, followed by stems and fruits. The results show that ultrasonic optimization can improve extraction rate, reduce time, lower costs, enhance quality, and increase yield. Therefore, the optimized ultrasonic-assisted extraction process was adopted to extract the aerial parts of T. anomalum and separate the components. After optimization, the extract underwent several chromatographic separations to isolate eight previously undescribed withanolides (1–8) and two artificial withanolides (9–10), in addition to fifteen known compounds (11–25). Their structures were established through extensive spectroscopic data analysis. The compounds were evaluated for their antiproliferative effects against multiple cancer cell lines, including human hepatocellular carcinoma cells (HepG2, Hep3B, and MHCC97-H), human lung cancer cells (A549), human fibro-sarcoma cancer cells (HT1080), human chronic myeloid leukemia cells (K562), and human breast cancer cells (MDA-MB-231 and MCF7). Compounds 1–3, 5, 7, 11, 13, 15–16, and 22 displayed significant activity with IC₅₀ values of 5.14–19.87 μM. The above results indicate that ultrasonic-assisted extraction technology can be used to obtain new withanolides more efficiently from T. anomalum, thereby enhancing the utilization rate of T. anomalum resources.

1. Introduction

Withanolides are a group of natural compounds consisting of C₂₈ steroidal δ- or γ- lactones that are derived from an ergostane skeleton. They are primarily found in 25 genera of the Solanaceae family, including Jaborosa, Physalis, Vassobia, Withania, Tubocapsicum, etc [1]. In recent years, with the continuous discovery of new structural types of withanolides and the in-depth study of modern pharmacology, some withanolides stand out because of their significant biological activities, which also makes this class of compounds have greater development potential to become drug candidates [2], [3]. Therefore, the extraction technology of withanolides has become a research hotspot.

In order to ensure the effectiveness of the research on the extraction technology of withanolides, it is necessary to establish a comprehensive set of chemical component analysis and detection methods [4], [5]. At present, the primary methods used for analyzing withanolides in Solanaceae involve thin-layer chromatography (TLC) and high-performance liquid chromatography/high-performance liquid chromatography-mass spectrometry (HPLC/HPLC-MS) [6], [7], [8], [9]. However, TLC is less commonly used in practice due to its lower precision and repeatability compared to HPLC. At present, research on the extraction of withanolides is still relatively limited, with heating reflux extraction being the primary method. However, the heated reflux extraction method may adversely affect components that are easily destroyed during the heating process [10]. A large number of studies have pointed out that the A-ring α, β-unsaturated ketone and 4β-hydroxyl, and the 5β,6β-epoxide rings of the B ring in the structure of withanolides are the key to the anti-tumor activity of these compounds [11], [12]. However, during the extraction process, the compounds containing α,β-unsaturated carbonyl structural units are prone to change under the influence of methanol when heated, resulting in a significant decrease in their biological activity [13]. Therefore, methanol and other polar nucleophiles should be avoided as much as possible in the extraction and separation process, to reduce the formation of potential artifacts. In addition to traditional solvent extraction methods, scholars at home and abroad have developed various efficient extraction technologies, including microwave-assisted, ultrasonic-assisted, and supercritical fluid extractions [14], [15], [16], [17]. However, most of the extraction processes described in the literature are single-factor investigations, and the processes have not been further optimized. Therefore, it is necessary to enhance the research on the extraction process of withanolides in Solanaceae.

Tubocapsicum anomalum (Solanaceae), a medicinal plant found in China, Korea, Japan, and the Philippines, has been documented as a potential source of withanolides [18], [19]. Our previous research indicated that the aerial parts of T. anomalum was comprehensively separated, revealing a high content of withanolides and demonstrating potential anti-tumor effects [20], [21]. However, the conventional extraction and enrichment methods for these components are time-consuming and inefficient, which not only limits the development and utilization of T. anomalum resources, but also leads to the increase of production costs. Ultrasound-assisted extraction (UAE), as a modern green extraction method, makes use of the characteristics of ultrasonic propagation in the medium for efficient extraction, which has the advantages of high efficiency, simple operation, less solvent use and high active components in the extraction solution [22], [23]. Scientists typically conduct one-factor-at-a-time (OFAT) experiments, in which only one factor is changed at a time while the other factors are held fixed. In contrast, response surface methodology (RSM) allows the study of the interactive effects of multiple factors to provide a more complete understanding on how the factors work together, and RSM can create a mathematical model that predicts the response under new factor settings, which can be valuable for process control and design [24]. Compared to OFAT, RSM is more efficient in terms of the number of experiments required, while response surfaces can be used to optimize the process to find the best combination of factor levels that maximize or minimize the response, which is impossible with the OFAT method [25]. To enhance the extraction rate of withanolides from T. anomalum, ultrasonic-assisted technology was employed for the extraction from its stems. The effects of solvent type, volume fraction, liquid–solid ratio, ultrasonic power, extraction temperature and time on the extraction rate were studied by single factor experiment. The RSM was used to optimize the extraction process, in order to develop a high extraction rate, simple operation and efficient extraction and enrichment method of T. anomalum. And then, we conducted a chemical investigation on the optimized extract of T. anomalum. This investigation has resulted in the discovery of eight new withanolides (1–8), two artificial withanolides (9–10), and fifteen previously identified analogues (11–26). And all the isolated compounds were tested for their antiproliferative effects against various cancer cell lines. The results of this paper show that using UAE technique can obtain new withanolides from T. anomalum faster and more efficiently, thus improving the utilization of T. anomalum.

2. Materials and methods

2.1. Materials and instruments

The plants of T. anomalum were collected from Fuding City, Ningde City, Fujian Province, China. A voucher specimen (No. 20180118) has been deposited in the Laboratory of Structural Pharmacology & TCM Chemical Biology at Shenyang Pharmaceutical University. The fruits, leaves, roots, and stems were dried, crushed separately using a high-speed grinder, sieved through an 80-mesh screen, and then set aside. UAE was performed on an ultrasound cleaning bath (KQ-600KDE, Kunshan Ultrasonic Instrument Co., Ltd., Kunshan, China). High Performance of Liquid Chromatography (HPLC) analysis was carried out using an SPD-M20A (LC-30A) HPLC system. The methanol was of high-performance liquid chromatography (HPLC) grade. All other chemicals and solvents utilized were of analytical grade.

2.2. Ultrasound-assisted extraction (UAE) experiment

2 g of dried powder from different parts of T. anomalum (fruits, leaves, roots, and stems) were weighed, crushed, and mixed with the extraction solvent. Then, parameters were set according to the experimental scheme, including ultrasonic power, ultrasonic temperature, and ultrasonic time. Finally, the collected products were centrifuged at 1500 rpm for 10 min. The supernatant in the centrifuge tube was transferred to a rotary evaporator for drying. It was then adjusted with methanol and filtered using a 0.45 μm filter membrane. The initial filtrate was discarded, and the remaining filtrate was used as the sample pretreatment liquid for HPLC detection and extraction rate calculation (mg/g) [26].

2.3. Determination of the content of withanolides

2.3.1. High-performance liquid chromatography analysis of extracts

The ultrasonic extracts of various parts of T. anomalum were analyzed using a C18 reversed-phase column with HPLC [27]. The HPLC gradient elution procedures were set as follows:

Column: YSC-TriART C18 (5 μm, 4.6 × 250 mm), column temperature: 30 ℃, sample chamber temperature: 4 ℃, sample size: 20 μL, flow rate: 0.80 mL/min. The liquid phase elution procedure was set as follows: 10 % MeOH-H₂O, 0.5 min; 15 ∼ 45 % MeOH-H₂O, 0.5 ∼ 30 min; 45 ∼ 65 % MeOH-H₂O, 30 ∼ 50 min; 65–80 % MeOH-H₂O, 50 ∼ 60 min; 80 %MeOH-H₂O, 60–70 min; 90 ∼ 100 % MeOH-H₂O, for 70 ∼ 75 min.

2.3.2. Drawing of standard curves

2 mg of tubocapsenolide A (Te-A) and tubocapsanolide A (Ta-A) were accurately weighed, and dissolved in 2 mL of methanol. Then, the reference solutions with concentrations of 0.04, 0.08, 0.12, 0.16, 0.20, and 0.24 mg/mL were diluted with methanol, respectively. The peak area of the standard product with various concentrations was determined using HPLC at 210 nm [7]. The concentration of the standard product (mg/mL) was plotted on the horizontal axis, while the peak area was plotted on the vertical axis. A standard curve was drawn to obtain a linear regression equation for the withanolides.

2.3.3. Precision detection of regression equation

Two standard solutions with a concentration of 0.12 mg/mL were accurately dispensed and repeated 6 times with 20 μL per injection to measure the peak area and calculate the relative standard deviation (RSD) using the following formula [8].

RSD=SD/Cm × 150 %.

RSD stands for relative standard deviation, SD stands for standard deviation, and Cm stands for average test results.

2.4. Single-factor experiment on T. Anomalum extraction

Due to the potential impact of various factors on the extraction rate of withanolides from T. anomalum, a single-factor test was designed and conducted [14]. The range of different factors used in this chapter is as follows: Different extraction solvents (ethanol, methanol, acetone, ethyl acetate), ultrasonic temperature (20, 30, 40, 50, 60 ℃), ethanol concentration (40 %, 50 %, 60 %, 70 %, 80 %, 95 %), ultrasonic power (50, 150, 200 W), ultrasonic time (10, 20, 30, 40, 50, 60 min), and liquid-to-solid ratio (5:1, 10:1, 15:1, 20:1, 30:1 mL/g).

2.5. Process optimization experiment of response surface method

In this study, ultrasonic power, ultrasonic time, and ultrasonic temperature were set as independent variables, denoted by X₁, X₂, and X₃, respectively. Three different levels of each variable were represented by 1, 0, and −1. Taking the extraction rates (Y) of two withanolides (Te-A and Ta-A) from T. anomalum as the dependent variable, a Box-Behnken design (BBD) was used to optimize the extraction rates of these two compounds with three factors and three levels (Table 1) [28]. Subsequently, the response surface model (RSM) is constructed based on BBD test results. Three independent experiments are conducted to compare experimental and predicted values. This aims to assess the superiority of the predicted model after optimizing the extraction procedure of withanolides from T. anomalum.

Table 1.

Factors and levels of Box-Behnken design.

	Independent variables
Level	X1-Power (W)	X2-Time (min)	X3-Temperature (℃)
−1	150	20	40
0	200	35	50
1	250	50	60

2.6. Heating reflux extraction experiment

The heating reflux method was used to extract 2 g of dried powder from different parts of T. anomalum (fruits, leaves, roots, and stems) in 70 % ethanol at 50 °C for 30 min. After filtering to remove suspended particles, the filtrate was taken and centrifuged for 20 min until the extract was clarified. Supernatant was filtered with a 0.45 μm membrane, discarding the initial filtrate [26]. The remaining filtrate was analyzed by HPLC. The experiment was repeated thrice.

2.7. Extraction and isolation

The total dry weight of the aerial parts of T. anomalum was about 20 kg, and the above ultrasonic extraction process was used to extract them using a 70 % ethanol–water solution. The total crude extract of 2815 g was obtained by concentrating the combined extract under reduced pressure. The crude extract was dispersed in 5 L of water and extracted twice with the same volume of petroleum ether and ethyl acetate, respectively, to obtain the petroleum ether extract (183 g), ethyl acetate extract (296 g), and water layer (2336 g). Specific details on sample separation, information on the instruments and compound information are described in SI Appendix.

2.8. Cell culture

All cell lines (HepG2, Hep3B, MHCC97-H, A549, HT1080, K562, MDA-MB-231, and MCF7) were purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA). Cells were routinely cultured in high Glucose Dulbecco’s modified Eagle’s medium (DMEM), which was supplemented with 10 % fetal bovine serum (FBS) (Sigma) and 100 U/mL penicillin/streptomycin (Invitrogen). The cells were incubated at 37 °C in 5 % CO₂ (v/v) [29].

2.9. Cell viability assay

The exponentially growing cells were seeded in 96-well plates with 6000 cells per well and incubated overnight at 37 °C with 5 % CO₂. The original medium was discarded 100 μL medium containing 10 % CCK8 (Cell Counting Kit-8) was added to each well and incubated in an incubator at 37 °C for 1 h. The absorbance was measured at 450 nm using a microplate reader (Synergy neo HTS multimode microplate reader, BioTek) [21]. All experiments were repeated for three times.

2.10. Statistical analysis

The experimental data of response surface optimization were analyzed using Design-Expert software (Ver 13.0, Stat. Ease Inc. USA) data processing system. Origin Pro 2019 software was used for plotting and curve fitting.

3. Results and discussion

3.1. HPLC analysis of the different parts of T. anomalum

Taking T. anomalum medicinal materials produced in Fujian as raw materials, withanolides were extracted from the roots, stems, leaves, and fruits of T. anomalum using ultrasonic assistance. HPLC was employed to analyze and detect withanolides. Upon comparison with the control product of withanolides previously isolated by our research group [20], peak 1 was identified as anomanolide C, peak 2 as Te-A, peak 3 as acnistin A, peak 4 as tubocapsenolide F, and peak 5 as Ta-A in the HPLC chromatograms of the leaf extract (Fig. 1A). In the stems of T. anomalum (Fig. 1B), peak 6 was anomanolide A. The compounds Te-A and Ta-A can be clearly identified from the extract of T. anomalum fruits (Fig. 1C). Fig. 1D shows the HPLC chromatograms of the extract of T. anomalum roots, without identifying withanolides.

Fig. 1.

High-performance liquid chromatogram (HPLC) of T. anomalum: (A) leaves, (B) stems, (C) fruits, (D) roots, (E) compound structures.

By comparing the content of withanolides in different parts of the plant, we found that the highest concentration of withanolides was in the leaves, followed by the stems and fruits, with almost none detected in the roots. The compounds Te-A and Ta-A can be clearly identified in the leaves, fruits, and stems of T. anomalum. However, in the actual picking process, the amounts of leaves collected was relatively small, while the number of stems was the largest. Therefore, to enhance the efficiency of extracting withanolides in T. anomalum, we chose the dried stems of the T. anomalum plant as the research material. We selected two representative compounds (Te-A and Ta-A) with high bioactivity in T. anomalum as reference substances for HPLC quantitative analysis to determine the extraction rate of withanolides in T. anomalum stems.

3.2. Drawing of standard curves and precision test results

A series of standard solutions were prepared with the concentration of standard products as the horizontal axis and the peak area of each component as the vertical axis. Fig. 2 shows the standard curve and linear range. The standard curves for compounds were as follows: Y_Te-A=7115.8x + 649.67, R² = 0.9991, Y_Ta-A=4808.6x + 701, R² = 0.9991. The results of the relative standard deviation (RSD) analysis are presented in Table S1. The RSD of the HPLC peak area for the two withanolides was less than 2 %. The experimental results proved that the HPLC instrument used demonstrated good precision.

Fig. 2.

Standard curves of Te-A and Ta-A.

3.3. Single factor test results

The choice of extraction solvent is crucial for ultrasonic extraction of withanolides. The impact of solvents on extraction rate was evaluated (Fig. 3A). Methanol and ethanol both showed good extraction effects, but methanol may alter the structures of withanolides. Hence, ethanol was chosen as the extraction solvent for further study. The effect of ethanol concentration on the extraction rate of withanolides was investigated. The concentration of ethanol in the solvent is one of the main factors that affect the extraction of withanolides by ultrasonic extraction. As shown in Fig. 3B, the extraction rate of withanolides first increased and then decreased with the increase in ethanol concentration. It is speculated that the optimal concentration of ethanol can penetrate the cell to dissolve withanolides. High concentrations, however, may lead to protein denaturation, thus hindering the dissolution of withanolides and resulting in a lower extraction rate. Therefore, 70 % ethanol was selected for further study in this experiment. As shown in Fig. 4A, with the increase in the liquid–solid ratio, the extraction of withanolides first increased and then decreased, reaching the maximum value when the liquid–solid ratio was 20:1 mL/g. When the liquid–solid ratio increased from 20:1 mL/g to 30:1 mL/g, the yield decreased slightly. The reason is that in the case of a large volume of solution, excessive solvent will reduce the efficiency and strength of ultrasonic cavitation [22]. Additionally, it can lead to a certain degree of waste, making the operation difficult to be carried out. Considering that the experiment should adhere to the principles of efficiency and cost-effectiveness, the optimal experimental parameter for the liquid–solid ratio should be selected as 20:1 mL/g. It can be observed from Fig. 4B that with the increase in ultrasonic power, the extraction rate of withanolides first increased and then showed a decreasing trend. The maximum value was achieved when the ultrasonic power is set at 150 W. Nevertheless, when the ultrasonic power was raised to 150–250 W, the extraction rate of Te-A started to decline gradually, whereas the extraction rate of Ta-A showed a slight initial increase before also decreasing. This may be because the high ultrasonic power produces varying degrees of damage to different components. Therefore, in the response surface optimization test, the extended optimization range for ultrasonic power was 150–250 W.

Fig. 3.

Effect of extraction solvent and ethanol concentration on extraction amount of withanolides.

Fig. 4.

Single factor test results: the effects of (A) liquid-to-solid ratio, (B) ultrasonic power, (C) ultrasonic time, (D) ultrasonic temperature.

Ultrasonic time is crucial for enhancing extraction rate. As shown in Fig. 4C, increasing ultrasonic time from 10 to 40 min gradually raised the extraction rate of withanolides. However, extending the time to 60 min leads to a gradual decrease in extraction rate. This could be attributed to the excessive duration of ultrasonic treatment, which might have disrupted the structure of withanolides. Therefore, the response surface for ultrasonic time selection was further optimized, with a range of 20–50 min. As shown in Fig. 4D, in the ultrasonic extraction temperature range of 20 to 50 ℃, the extraction rate exhibited an increasing trend with the rise in temperature. When the ultrasonic extraction temperature reached 50 to 60 ℃, the extraction rate decreased as the extraction time extended. In order to thoroughly comprehend the impact of temperature on the extraction rate, an ultrasonic temperature range of 40 to 60 ℃ was chosen to optimize the experiment.

3.4. Response surface experiment results

Based on the BBD principle, this study employed the response surface method to refine screening conditions for exploring factor interactions. The study focused on optimizing ultrasonic power, time, and temperature. The specific experimental design scheme and data results are presented in Table 2.

Table 2.

Box-Behnken design response surface methodology design scheme and test results.

Run	X1 (Power) W	X2 (Time) min	X3 (Temperature) ℃	Y1 (Te-A) mg/g	Y2 (Ta-A) mg/g
1	150	40	50	2.74	1.12
2	150	20	50	2.64	1.11
3	150	30	40	2.42	0.97
4	150	30	60	2.28	0.98
5	200	30	50	2.81	1.10
6	200	30	50	2.35	0.99
7	200	20	60	2.65	0.93
8	200	30	50	2.15	0.83
9	200	40	40	2.50	1.12
10	200	20	40	2.31	0.92
11	200	30	50	2.70	1.08
12	200	30	50	2.68	1.16
13	200	40	60	2.84	1.15
14	250	20	50	2.39	1.01
15	250	30	60	2.42	1.18
16	250	30	40	2.88	1.19
17	250	40	50	2.43	1.12

The response surface data analysis software Design Expert 13 was used to analyze the experimental data in Table 3 [30]. The quadratic polynomial regression equation describing the relationship between the extraction rate (Y_Te-A)/(Y_Te-A) of two withanolides in T. anomalum and the ultrasonic power (X₁), ultrasonic time (X₂) and ultrasonic temperature (X₃) was derived.

Table 3.

Analysis of variance (ANOVA) of the Box-Behnken design.

Source	Sum of squares		df	Mean Square		F value		Prob＞F
	YTe-A	YTa-A		YTe-A	YTa-A	YTe-A	YTa-A	YTe-A	YTa-A
Model	0.172	0.7231	9	0.0191	0.0803	35.63	15.71	< 0.0001	0.0007
X1	0.099	0.0903	1	0.099	0.0903	184.58	17.66	< 0.0001	0.004
X2	0.0008	0.0001	1	0.0008	0.0001	1.49	0.0220	0.2615	0.8863
X3	0.0010	0.0000	1	0.001	0.0000	1.89	0.0098	0.2119	0.924
X1 X2	0.0000	0.0462	1	0.0000	0.0462	0.0466	9.04	0.8352	0.0198
X1 X3	0.0169	0.0081	1	0.0169	0.0081	31.5	1.58	0.0008	0.2485
X2 X3	0.0182	0.0225	1	0.0182	0.0225	33.97	4.4	0.0006	0.0741
X12	0.0144	0.0275	1	0.0144	0.0275	26.86	5.37	0.0013	0.0536
X22	0.0003	0.0001	1	0.0003	0.0001	0.6358	0.0272	0.4514	0.8736
X32	0.0198	0.5106	1	0.0198	0.5106	36.83	99.86	0.0005	< 0.0001
Lack of fit	0.0011	0.0107	3	0.0004	0.0036	0.5348	0.5666	0.6828	0.6657
Credibility analysis of the regression equations		Index mark	Standard deviation	Mean	CV%	Press	R2	Predicted R2	Adequacy precision
		YTe-A	0.0232	1.06	2.19	0.02	0.9786	0.8783	20.33
		YTa-A	0.0715	2.54	2.82	0.21	0.9528	0.7232	12.79

*p < 0.05, significant；**p < 0.01, highly significant；***p < 0.001, extremely significant. The Design Expert 13.0 software was applied to obtain the results.

Y_(Te-A) = -2.2 + 1.9 × 10^-2X₁-3.8 × 10^-2X₂ + 7.4 × 10^-2X₃ + 2.0 × 10^-7X₁X₂ + 6.9 × 10^-4X₂X₃-1.4 × 10^-4X₁X₃-2.4 × 10^-5X₁² + 7.7 × 10^-5X₂²-6.9 × 10^-4X₃².

Y_(Ta-A) = -6.1 + 1.3 × 10^-2X₁-3.4 × 10^-2X₂ + 3.5 × 10^-2X₃ + 6.1 × 10^-7X₁X₂ + 4.3 × 10^-4X₂X₃-1.8 × 10^-4X₁X₃-3.1 × 10^-5X₁² + 3.2 × 10^-5X₂²-9.1 × 10^-4X₃².

Based on the above results, an RSM model was constructed, and the model's accuracy was evaluated using analysis of variance (ANOVA). According to the results of ANOVA (Table 3), the significance of the impact of the regression equation on the response value is determined by the F-value and probability. The F-values of Te-A and Ta-A in this model were 35.63 and 15.71, respectively, with p < 0.0001, indicating that the model is highly significant. Among the factors of Te-A, X₁, X₁X₃, X₂X₃, and X₃² were extremely significant, while X₁² was highly significant. In Ta-A, X₁ and X₃² were highly significant items, while the rest were not significant. The lack of fit (LOF) term shows the variation in the data around the fitted model [31], [32]. The LOF terms were 0.6828 and 0.6657, both greater than 0.05, suggesting they were not significant. The regression model was able to accurately fit the true response value. The coefficient of determination (R²) for Te-A was 0.9786, and for Ta-A was 0.9528, suggesting that the fitting equations closely match the data [33]. The predicted R² values for Te-A and Ta-A were 0.8783 and 0.7232, respectively, indicating the reliability and effectiveness of the predictions [34]. The signal-to-noise ratios were 20.33 and 12.79, respectively, both of which were greater than 4. This indicates that the signal was accurate. The coefficients of variation (CV) were 2.19 % and 2.82 %, respectively, indicating a good fit for the model. This suggests that the equation was suitable for analyzing the process of ultrasonic-assisted extraction of withanolides.

RSM incorporates statistical methods to produce response surface plots that can be evaluated for model validity and predictive certainty, thus providing more reliable results [35], [36]. Optimal parameter conditions and interactions can be observed from this graph, as shown in Fig. 5. It can be seen from Fig. 5A-C that the influence of three factors on the extraction rate of Te-A was as follows: X₁ (ultrasonic power) > X₂ (ultrasonic time) > X₃ (ultrasonic temperature). And the influence of three factors on the yield of Ta-A was as follows (Fig. 5D-F): X₁ (ultrasonic power) > X₃ (ultrasonic temperature) > X₂ (ultrasonic time). After optimizing and analyzing the response surface, the optimal process conditions for Te-A and Ta-A were determined as follows: ultrasonic power of 250 W, ultrasonic time of 40 min, and ultrasonic temperature of 50 °C. Under these conditions, the theoretical extraction rates of Te-A and Ta-A after optimizing the response surface were 2.88 mg/g and 1.19 mg/g, respectively. Under optimal prediction conditions, the analysis results of the response surface method were verified. The average yield of Te-A and Ta-A reached 2.87 ± 0.12 mg/g and 1.18 ± 0.05 mg/g through five repeated experiments. These values were very close to the predicted extraction rate, demonstrating that the model was suitable for optimizing parameters in the extraction process of withanolides.

Fig. 5.

The 3D surface plots showing interaction diagram of any two variables on Te-A (A-C) and Ta-A (D-F) yield.

3.5. Effect of two extraction methods on the extraction rate of withanolides

In order to evaluate the actual effect of the optimized ultrasonic extraction process, different parts of T. anomalum (fruits, leaves, roots, and stems) were chosen as the subjects of the study. They were extracted using the heating reflux extraction method and ultrasonic extraction method. The results obtained were presented in Table 4. The ultrasound-assisted extraction technology developed by us is not only suitable for extracting withanolides from different parts of T. anomalum, but also its extraction rate significantly surpasses that of the heating reflux extraction method within the same timeframe. At the same time, we found that the stems, leaves, and fruits contained withanolides, but the roots did not. In order to obtain enough withanolides for the pharmacological activity research, we then adopted an optimized ultrasonic-assisted extraction process to extract the aerial parts of T. anomalum and separate the components.

Table 4.

Comparison of different extraction methods.

	Heating reflux extraction		Ultrasound-assisted extraction
Sample	Te-A	Ta-A	Te-A	Ta-A
Leaves	2.12 ± 0.11	1.04 ± 0.03	3.16 ± 0.11	1.74 ± 0.04
Stems	2.07 ± 0.12	0.75 ± 0.04	2.87 ± 0.12	1.18 ± 0.05
Fruits	0.20 ± 0.05	0.45 ± 0.05	0.25 ± 0.017	0.55 ± 0.08
Roots	−	−	−	−

3.6. Identification and isolation of compounds

The investigation of the constituents from the aerial parts of T. anomalum led to the isolation of ten new withanolides (1–10) and fifteen known analogues (11–25) (Fig. 6). The known ones were identified as 20-hydroxytubocapsanolide A (11) [18], 23-hydroxytubocapsanolide A (12) [18], tubocapsanolide D (13) [19], 4β-hydroxyacnistin I (14) [19], withanolide D (15) [29], 17R-hydroxywithanolide D (16) [29], tubocapsenolide G (17) [37], tubonolide A (18) [29], withajardin J (19) [29], 17-epi-withajardin J (20) [29], anomanolide D (21) [18], 17-epiacnistin A (22) [29], anomanolide B (23) [38], 4β-hydroxyanomanolide E (24) [19], and 3β-ethoxy-2,3-dihydrotubocapsanolide A (25) [29] based on a comparison of their ¹H and ¹³C NMR data with those reported in the literatures.

Fig. 6.

Structures of compounds 1–25 from T. anomalum.

Compound 1 was obtained as white powder. The molecular formula was determined to be C₂₉H₃₈O₆ by its HRESIMS at m/z 483.2745 [M+H]^- (calcd. for C₂₉H₃₉O₆, 483.2741), indicating eleven indices of hydrogen deficiency. The ¹H NMR spectrum (Table 5) of compound 1 showed two olefinic protons at δ_H 6.86 (1H, dd, J=10.2, 2.6 Hz) and 6.22 (1H, dd, J=10.2, 1.7 Hz), five methyl groups at δ_H 1.85, 1.70, 1.51, 1.32 (each 3H, s), and 1.23 (3H, d, J=7.0 Hz). The ¹³C NMR spectrum (Table 6) involved 29 carbon signals, including one ketone carbonyl signal at δ_C 209.9, one ester carbonyl signal at δ_C 166.6, three groups of double bond signals at δ_C 147.6/126.6, 162.5/122.7 and 149.9/122.1, and two carbon resonances of a three-membered epoxy ring at δ_C 64.8 and 59.3. The above NMR data revealed a typical withanolide of compound 1. Comparison of the NMR data of 1 with those of withangulatin A [39] showed the main difference between them in the significant upfield shifts of C-14 (δ_C 84.0) to δ_C 44.9, and an extra methoxyl [δ_H 3.30 (3H, s); δ_C 57.2] instead of an acetoxy group, indicating the absence of a hydroxy group at C-14 and the presence of a methoxyl substituent at C-15 in 1. The HMBC cross-peaks from H-12b to C-14/C-13, H-16 to C-14/C-15, CH₃-19 to C-14, and H-15 to OCH₃-15 confirmed the above assumption. Taking biosynthetic grounds into consideration [40], the orientation of CH₃-18 and CH₃-19 was β. The NOESY cross-peaks of H-15 with CH₃-18 showed that the α-orientation of OCH₃-15. The ECD spectrum indicated a (22R)-configuration via the positive Cotton effect at 250 nm [41]. Therefore, the structure of 1 was established as 15α-methoxy-5β,6β-epoxy-4β-hydroxy-21,24-cycloergost-2-en-1-one and named 15α-methoxywithangulatin A.

Table 5.

¹H NMR Data of Compounds 1–10 (Pyridine‑d₅, 600 MHz, δ in ppm, J in Hz).

pos.	1	2	3	4	5	6	7	8	9	10
2	6.22 dd (10.2, 1.7)	6.28 dd (10.3, 2.4)	6.15 dd (10.1, 2.8)	6.30 dd (1.9, 10.3)	6.43 d (9.9)	6.09 dd (10.3, 1.9)	6.40 d (9.8)	3.30 dd (15.2, 4.4) 3.14 dd (15.2, 8.1)	3.14 dd (15.4, 6.4) 2.94 d (15.4)	3.10 dd (15.3, 7.4) 2.89 dd (15.3, 3.4)
3	6.86 dd (10.2, 2.6)	6.94 dd (10.3, 2.0)	6.66 m	6.95 dd (2.6, 10.3)	7.22 dd (9.9, 6.3)	6.65 dd (10.3, 2.6)	7.20 dd (9.8, 6.3)	3.96 br s	3.91 br s	3.91 dd (7.4, 3.2)
4	3.94 m	5.61 br s	3.79 d (19.5) 2.45 overlapped	5.29 br s	4.01 d (6.3)	5.09 br s	3.99 d (6.3)	3.92 br s	3.87 br s	3.89 br s
6	3.51 br s	4.47 dd (11.6, 5.2)	4.20 s	4.67 dd (12.6, 4.6)	3.23 br s	4.65 dd (12.8, 4.6)	3.21 br s	3.46 br s	3.35 br s	3.36 br s
7	2.98 m 2.12 m	2.26–2.30 m 1.71 m	2.25 m 1.96 m	2.20 d (14.9) 1.35 m	2.08 m 1.24 m	2.27 dt (13.3, 4.4) 1.77 d (12.8)	1.78 m 1.24 m	2.34 1.82 m	2.20 d (14.9) 1.35 m	2.21 d (14.3) 1.36 m
8	2.28 m	1.78–1.84 m	2.38 m	1.56 m	1.53o	1.59 m	1.48 m	2.30 br s	1.56 m	1.60 m
9	2.26 m	1.52–1.54 m	2.40 overlapped	1.33o	1.01 m	1.49 m	0.97 m	1.87 m	1.33o	1.39o
11	1.69 m 1.48 m	1.14–1.17 m 1.59 dd (12.8, 3.3)	2.77–2.81 m 1.71 m	1.52 m	2.00 dd (14.3, 3.4) 1.53o	1.38 m 1.21 m	1.94 dd (14.3, 3.4) 0.90o	2.03 m 1.86o	1.52 m	1.65 m 1.57 m
12	1.75 m 1.58 m	1.01 m 2.1 m	2.41 m 1.66 m	1.74 m 1.39 m	1.69 m 0.92 m	1.87 td (12.7, 4.1) 1.46 m	1.66 m 0.90 m	1.94 m 1.21 m	1.74 m 1.39 m	1,98 m 1.57o
14	0.77 m	0.93 m	1.14 m	1.79 m	0.83 m	2.01 m	0.81 m	0.81 m	1.79 m	1.93 m
15	5.89 d (2.5)	2.37 overlapped	1.43–1.48 m 2.31 m	1.68 m 1.08 m	1.48 m 0.94 m	1.60o 1.11 m	1.46 m 0.90 m	2.55 m 1.87o	1.68 m 1.08 m	1.71 m 1.12 m
16	6.12 d (2.5)	4.85 br s	4.74 br s	1.98 m	1.67 m 1.23o	1.98 m	1.64 m 1.20o	5.26 dt (13.1, 2.9)	1.95 t (12.8) 1.71 m	1.60 m
17		2.16 m
18	1.32 s	1.45 s	1.57 s	0.64 s	0.50 s	0.57 s	0.41 s	1.69 s	0.64 s	0.69 s
19	1.70 s	1.62 s	1.73 s	1.71 s	1.82 s	1.54 s	1.77 s	2.78 dd (7.2, 2.8)	1.71 s	1.75 s
20	2.62 m			2.51 m	1.02o	2.64 t (8.0)	1.01o	1.15 s	2.51 m	2.56 t (8.5)
21	1.23 d (7.0)	1.74 s	1.77 s	2.53o 1.57o	2.76 d (12.9) 1.40 m	2.92 d (10.8) 1.83 m	2.99 dd (13.4, 8.5) 1.56 dd (13.4, 3.2)	4.46 m	2.53o 1.57o	2.07 m 2.01 m
22	4.43 dt (13.0, 3.7)	4.59 br s	4.56 s	4.76 br s	4.53 d (2.2)	4.89 br s	4.47 br s	3.01 d (17.5) 2.60 m	4.76 br s	4.98 d (2.9)
23	2.42 m 2.01 m	2.42 m 2.01 m	2.42 m 2.01 m	2.76 d (14.9) 2.07 d (14.9)	2.11 m 1.31 dd (13.7, 5.8)	2.10 m	2.12 d (13.4) 1.22o	2.11 d (13.4) 1.21o	2.76 d (14.9) 2.07 d (14.9)	2.70 d (12.0) 1.96 m
27	1.85 s	1.20 s	1.18 s	1.42 s	1.73 s	1.66 s	1.63 s	1.65 s	1.42 s	1.80 s
28	1.51 s	1.21 s	1.19 s	1.32 s	4.53 d (14.5) 4.61 d (14.5)	1.26 s	1.30 s	1.28 s	1.32 s	1.38 s
3-OCH3									3.27 s	3.29 s
15-OCH3	3.30 s
OH-4		7.89 s			7.91 s
OH-16							6.19 br s
OH-17							5.29 br s

Table 6.

13C NMR Data of Compounds 1–10 (Pyridine‑d₅, 150 MHz, δ in ppm).

pos.	1	2	3	4	5	6	7	8	9	10
1	209.9	201.7	203.6	209.3	202.4	201.3	202.4	209.5	209.3	209.3
2	126.6	127.8	129.5	129.5	132.0	126.0	132.0	40.9	40.7	40.8
3	147.6	147.2	142.4	142.4	145.0	146.6	144.9	79.1	78.6	78.6
4	74.5	72.1	37.1	74.5	70.1	64.4	69.6	74.8	74.5	74.5
5	64.8	63.4	65.4	64.8	64.2	75.7	79.7	65.0	64.8	64.8
6	59.3	57.1	58.9	57.9	59.9	65.7	32.7	58.3	57.9	57.9
7	25.6	38.4	34.6	31.9	31.4	39.7	28.9	30.0	31.9	31.8
8	36.6	33.6	30.3	30.2	30.0	35.3	30.0	30.8	30.2	30.1
9	39.0	45.7	42.7	42.9	44.4	46.3	44.4	41.7	42.9	42.9
10	51.3	57.1	53.0	50.7	48.3	57.9	48.2	50.7	50.7	50.4
11	21.4	23.6	24.1	21.1	21.4	22.9	21.4	24.5	21.1	21.1
12	37.8	40.5	38.6	31.5	39.1	31.8	39.2	22.9	31.5	31.6
13	43.5	44.5	44.6	48.0	42.2	47.7	42.2	139.6	48.0	47.3
14	44.9	54.5	54.9	49.8	55.7	49.5	55.7	134.9	49.8	50.7
15	84.8	38.3	41.8	23.6	24.3	23.5	24.3	41.7	23.6	24.1
16	122.7	73.0	73.1	35.3	27.6	61.6	27.7	79.3	35.3	38.0
17	162.5	83.5	83.6	83.3	49.1	71.2	49.1	52.7	83.3	83.3
18	16.6	15.8	16.1	15.3	12.5	14.7	12.4	15.4	15.5	15.1
19	15.5	11.1	16.5	15.3	17.0	10.2	17.0	14.5	15.3	15.3
20	35.6	80.2	80.1	54.3	54.6	54.1	54.6	42.1	41.2	54.1
21	18.2	23.1	23.1	27.2	37.4	34.9	39.0	26.1	27.2	35.5
22	79.1	74.3	74.2	77.1	84.8	85.0	84.9	81.3	77.1	84.8
23	32.7	79.7	79.6	42.0	39.1	41.0	39.0	42.5	42.0	39.7
24	149.9	148.7	147.8	70.2	47.3	47.6	42.5	71.2	70.2	47.4
25	122.1	120.9	123.2	48.2	76.7	76.9	41.0	47.2	48.2	76.7
26	166.6	167.4	166.7	177.7	175.0	178.4	179.8	178.7	177.7	175.9
27	12.9	12.3	13.9	15.0	19.9	25.3	24.9	11.8	15.0	20.3
28	20.5	20.0	20.1	27.9	60.8	20.0	20.5	60.8	27.9	19.7
OCH3-3									56.4	56.4
OCH3-15	57.2

Compounds 2–3 both had NMR (Table 5, Table 6) signal characteristics of an α,β-unsaturated ketone and an α,β-unsaturated δ-lactone, and the molecular formulas for 2 and 3 were both defined as C₂₈H₃₈O₇ (2: [M+H]⁺ m/z 487.2693, 3: [M+H]⁺ m/z 487.2692, calcd. for C₂₈H₃₉O₇, 487.2690). The NMR signals at δ_C 36.7 (C-4) and 33.6 (C-16) in withanolide E [42] were shifted downfield to δ_C 72.1 (C-4) in 2 and δ_C 73.0 (C-16) in 3, respectively. Thus, together with HSQC and HMBC analysis, OH groups were placed at C-16 (2) and C-4 (3), respectively. The structures and relative stereochemistry were established by NOESY data as 5β,6β-epoxy-4β,17α,20β-trihydroxy-21,24-cycloergost-2-en-1-one (2) and 5β,6β-epoxy-16α,17α,20β-trihydroxy-21,24-cycloergost-2-en-1-one (3). Hence, 2 and 3 were hydroxy derivatives of withanolide E, and they were named 4-hydroxywithanolide E (2) and 16-hydroxywithanolide E (3).

Compound 4 was obtained as a white amorphous powder. Its molecular formula C₂₈H₃₉ClO₇ with nine degrees of unsaturation was established by HRESIMS (m/z 545.2259 [M+Na]⁺ (calcd. for C₂₈H₃₉ClNaO₇, 545.2277) and NMR data. The NMR signals (Table 5, Table 6) of compound 4 showed similarities to those observed in tubonolide A (18) [29]. These signals were characteristic of a 2-oxabicyclo[2.2.2]octan-3-one moiety, which is typical of withajardin-type withanolides. By conducting a thorough comparison of the data for these two compounds, we found the absence of a hydroxy group at position C-16 compound 4. It was confirmed by the HMBC correlations (Fig. 7) from OH-17 [δ_H 5.29 (1H, br s)] to C-17 (δ_C 83.3), C-20 (δ_C 54.3), C-13 (δ_C 48.0), and C-16 (δ_C 35.3), and H-16 [δ_H 1.98 (2H, m)] to C-17 (δ_C 83.3) and C-15 (δ_C 23.6). The NOESY correlation (Fig. 8) of OH-17 with H-14 indicated that an α-orientation for OH-17. Thus, the structure of 4 was assigned as 6α-chloro-4β,5β,17α,25α-tetrahydroxy-21,25-cycloergost-2-en-1-one, and named tubonolide F.

Fig. 7.

Key HMBC correlations of compounds 1–8.

Fig. 8.

Key NOESY correlations of compounds 1–8.

Unlike the aforementioned withanolides, compounds 5–10 instead exhibited a 4-hydroxy-4,5-dimethyl-2-oxa-bicyclo[3.2.1]octan-3-one ring on C-17, similar to acnistin-type withanolides. Characteristic signals in the ¹H and ¹³C NMR (Table 5, Table 6) spectra of compound 5 were assigned to an epoxy group at C-5/C-6 and OH groups at C-4 and C-17. The ¹³C NMR spectra of 5 and acnistin E [29] were similar. The differences between them were the absence of a singlet signal corresponding to CH₃-28 at the high-field region in the ¹H NMR spectrum of 5. Additionally, the presence of two doublets at δ_H 4.53 (1H, d, J=14.5 Hz) and 4.61 (1H, d, J=14.5 Hz) suggested the existence of a C-28 hydroxymethylene group in 5. This group was confirmed by the HMBC correlations of CH₃-28 with C-23/C-24 and C-25. In the NOESY spectrum (Fig. 3), H-4 was correlated with H-9, indicating the β-orientation of OH-4. Thus, compound 5 is 28-hydroxy derivative of acnistin E and has been designated as 28-hydroxyacnistin E (5β,6β-epoxy-4β,17α,25α,28-tetrahydroxy-21,24-cycloergost-2-en-1-one).

The molecular formula of 6 was established by HRESIMS and NMR data as C₂₈H₃₇ClO₇. Comparison of its NMR data (Table 5, Table 6) with those of the known tubocapsanolide D (13) [18], showed the carbon signal of C-17 at δ_C 85.2 was shifted towards the higher field to δ_C 71.2, while lower-field shift of C-16 (δ_C 35.2) to δ_C 61.6, suggesting that C-16 and C-17 formed an epoxy ring in 6. Combined with its HRESIMS data, we supposed the presence of a chlorine atom at C-6 in compound 6. The above supposition was confirmed by the detailed interpretation of its HMBC correlations (Fig. 7) from H-16 to C-17, C-14 and C-15, and H-6 to C-5, C-4 and C-7. The 16α,17α-epoxy group was determined based on the NOE correlations of H-8 with CH₃-18, and H-16 with CH₃-18, as well as the NOE correlation between H-6 and CH₃-19, indicating the β-orientation of H-6. Therefore, the structure of 6 was established as 6α-chloro-16α,17α-epoxy-4β,25α-dihydroxy-21,24-cycloergost-2-en-1-one, and it was named anomanolide K.

The molecular formula of 7 was established as C₂₈H₃₈O₇. The 1D NMR signals were similar to those of 6, including a 16α,17α-epoxy ring and the 4β,5β-diol. Compound 7 exhibited NMR signals at δ_H 1.24 (1H, m, H-6a), 1.78 (1H, m, H-6b)/δ_C 32.7 (C-6), and the lower-field shift of C-16 (δ_C 35.7) to δ_C 26.9 when compared with the data of compound 6, indicating the absence of a chlorine group at C-6 in 7. These deductions were confirmed by the HMBC correlations (Fig. 7) from H-6b to C-5, C-7 and C-8, and H-6a to C-4, C-5 and C-7. In the NOESY spectrum (Fig. 8), the observed correlations between H-4 and H-9/H-14 indicated an α-orientation of H-4. Thus, the structure of anomanolide O (7) was defined as 16α,17α-epoxy-4β,5β,25α-trihydroxy-21,24-cycloergost-2-en-1-one.

Compound 8 was assigned a molecular formula of C₂₈H₄₀O₉ based on HRESIMS and ¹³C NMR data. Comparison of the NMR data (Table 5, Table 6) of 8 with anomanolide C (22) [29] indicated that signals of a methylene [δ_H 3.30 (1H, dd, J=15.2, 4.4 Hz, H-2a) and 3.14 (1H, dd, J=15.2, 8.1 Hz, H-2b); δ_C 40.9 (C-2)], and an oxygenated methine [δ_H 3.96 (1H, br s, H- 3); δ_C 79.1 (C-3)] in 8 replaced the signals of an olefinic moiety in anomanolide C. These features were reminiscent of the hydration of the Δ^2,3 double bond. The HMBC cross-peaks from H-2a to C-1/C-3, H-2b to C-1/C-3/C-4, H-4a to C-3, and H-4b to C-3/C-5/C-10, confirmed the above assumption. The NOESY correlations of H-3 with H-4/H-9, and H-16 with CH₃-18 showed that the β-orientation of OH-3, OH-4, and H-16. Therefore, the structure of anomanolide L (8) was defined as 5β,6β-epoxy-3β,4β,16α,17α,25α-pentahydroxy-21,24-cycloergost-2-en-1-one.

Compounds 9 and 10, which are artificial withanolides, were also isolated during the extraction and isolation procedures. Comparison of the NMR spectral data (Table 5, Table 6) of 9 and 10, with those of the reported withanolides anomanolide D [29] and 25-hydroxytubocapsanolide A [18], demonstrates the presence of an additional methoxy signal at C-3 in 9 and 10, instead of a Δ^2,3 double bond. The key HMBC correlations from H-2 to C-1/C-3/C-4, H-3 to C-1/C-2/C-4, and from OCH₃-3 to C-3 suggest the presence of a methoxy signal at C-3. In addition, the relative configuration of OCH₃-3 was assigned to be β-oriented by NOESY interaction of H-4 and H-3/H-6, H-6 and H-9 (Fig. S54-S74). Consequently, the structures of 9 and 10 were determined as 3β-methoxy-2,3-dihydro-anomanolide D and 3β-methoxy-2,3-dihydro-25-hydroxytubocapsanolide A, respectively.

3.7. Preliminary screening of antitumor activity of the isolated compounds

To investigate the preliminary anti-tumor activity of 25 withanolides isolated from T. anomalum, the cell viability was determined on eight human cancer cell lines. These include hepatocellular carcinoma, lung cancer, fibro-sarcoma, chronic myeloid leukemia, and breast cancer cells. Table 7 showed that most compounds exhibited cytotoxic activity. Compounds 1–3, 5, 7, 11, 13, 15–16, and 22 showed significant cytotoxic activity against the above eight cancer cell lines with IC₅₀ values of 5.14–19.87 μM. Compounds 4, 6, 8–10, 12, 17–21 and 23–25, at the concentration up to 20 μM, exhibited no cytotoxic effects. This lack of cytotoxicity can be attributed to the absence of an enone group in ring A, which is crucial for inducing cytotoxicity. Moreover, converting the 5β,6β-epoxide into the corresponding 5β,6α-diol could potentially reduce its cytotoxic effects (5 vs 6). However, the introduction of β-hydroxylation at C-4 would likely enhance its cytotoxicity (2 vs 3). These findings align with the earlier analysis of the structure–activity relationship (SAR), confirming the consistency of the results[12], [43], [44]. It was notable that compound 1 exhibited significant anti-proliferative activity against eight different cell lines.

Table 7.

IC₅₀ values of compounds 1–25 against eight human cancer cell lines.

	Cell lines (IC50, µM)
Com.	MDA-MB-231	HT1080	A549	HepG2	Hep3B	MCF7	K562	MHCC97-H
1	13.35 ± 1.05	8.56 ± 1.09	>20	7.90 ± 1.21	5.14 ± 1.10	13.87 ± 1.05	17.56 ± 1.84	13.35 ± 1.05
2	>20	>20	>20	>20	>20	>20	19.22 ± 0.12	>20
3	>20	8.17 ± 1.09	>20	>20	9.70 ± 1.15	>20	19.87 ± 1.21	>20
5	>20	8.54 ± 1.09	>20	8.17 ± 1.09	>20	>20	14.25 ± 1.04	>20
7	14.88 ± 1.05	>20	15.13 ± 1.15	>20	17.38 ± 2.02	13.43 ± 1.03	6.88 ± 1.06	9.88 ± 1.05
11	8.17 ± 1.09	>20	>20	>20	14.21 ± 2.01	>20	>20	>20
13	16.28 ± 1.09	7.02 ± 1.07	>20	6.10 ± 1.14	7.34 ± 1.18	16.26 ± 1.03	7.36 ± 1.08	16.54 ± 1.09
15	>20	>20	>20	>20	>20	>20	19.22 ± 0.12	>20
16	>20	8.17 ± 1.09	>20	>20	9.70 ± 1.15	>20	19.87 ± 1.21	>20
22	>20	>20	>20	>20	17.38 ± 2.02	>20	>20	>20

^a Inhibition of cell proliferation measured by the CCK8 assay. Values are represented as the means ± S.D. based on three independent experiments. Compounds 4, 6, 8–10, 12, 17–21 and 23–25 were inactive (IC₅₀ > 20 μM).

4. Conclusions

Withanolides are active antitumor components in T. anomalum. In order to establish an efficient extraction and enrichment method for withanolides, this paper first developed a high-performance liquid chromatography method for the analysis and detection of withanolides in T. anomalum. The HPLC analysis revealed that the highest content of withanolides was found in the leaves of T. anomalum, followed by the stems and fruits, with almost none detected in the roots. Therefore, we utilized the ultrasonic extraction method to optimize the efficient extraction process of withanolides from dried stems of T. anomalum. We determined the optimal extraction conditions through a Box-Behnken response surface experiment to obtain withanolides with a high yield. HPLC was utilized to determine and compare the effects of ultrasonic extraction and heated reflux extraction on the extraction rates of two types of withanolides from different parts (fruits, leaves, roots, and stems) of T. anomalum. It was found that the extraction rate was highest for leaves, followed by stems and fruits. Additionally, the ultrasonically assisted extraction rate was significantly higher than that of traditional heated reflux extraction.

In order to obtain more withanolides and establish a chemical composition foundation for pharmacological activity research, we adopted an optimized ultrasonic-assisted extraction process to extract the aerial parts of T. anomalum and separate the components. In the end, we isolated ten new withanolides (1–10) along with fifteen known analogues (11–25) from the aerial parts of T. anomalum. The structures of new compounds were identified by using NMR, HRESIMS, and ECD data. Cytotoxicity was evaluated against eight tumor cell lines for the isolated compounds. Among them, compounds 1–3, 5, 7, 11, 13, 15–16, and 22 showed significant cytotoxic activity with IC₅₀ values of 5.14–19.87 μM. Additionally, compound 1 demonstrated potent inhibition in all tested cell lines. In this paper, we enriched the structural diversity of natural withanolides, and their biological activities may need further studies. The above results indicate that the ultrasonic assisted extraction technology can be used to obtain new withanolides more quickly and efficiently from T. anomalum, thus improving the utilization rate of T. anomalum resources.

CRediT authorship contribution statement

Ke Xiang: Writing – original draft, Methodology, Investigation. Rui Zhu: Methodology, Data curation. Yueying Yang: Visualization, Validation, Methodology. Yang Xu: Visualization, Validation, Investigation. Kuiru Sa: Investigation, Data curation. Hua Li: Writing – review & editing, Project administration, Funding acquisition. Lixia Chen: Project administration, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article: