Shimadzu LC-2030C system (Shimadzu Corporation, Japan) and LabSolutions software program were used for chromatographic separation of components and analyzing HPLC data. Preparative-HPLC (Gelaipu Corporation, Chengdu, China), equipping with a GL6000-cp1000 high-pressure infusion pump, GL6000-UV3292 ultraviolet detector, and GelaiInst software program. The packing used in the pre-HPLC column (200 × 100 mm, 10 μm) was purchased from Acchrom Co., Ltd. (Wenling, China). BL-420F Organism Functional Experimental System (Chengdu Techman Co., Ltd. Chengdu, China) was used to record the lead II electrocardiograms (ECGs). CPA2250 electronic analytical balance (Sartorius, Germany) was used to correctly weigh the lab samples. The process of sand frying was simulated by oil bath heating using an HH-SJ heat-collecting magnetic stirrer (Jintan Chengdong Xinrui Instrument Factory, Changzhou, China). Optical rotations were obtained on a Perkin-Elmer 341 polarimeter (PerkinElmer, United States). Melting points were measured on an X-4 micro-melting point apparatus and uncorrected (Shanghai Precision Scientific Instrument Corporation, Shanghai, China). NMR spectra were measured on a Bruker Avance 600 spectrometer (Bruker, Germany) in CDCl₃ with tetramethylsilane (TMS) as internal standard. Mass spectra were carried out on a micrO-TOF-Q-II mass spectrometer (Bruker, Germany). RE-3000B rotary evaporator (Shanghai Yarong Biochemical Instrument Factory, Shanghai, China) was used to evaporate the solvent. Silica gel G (200–300 mesh) for column chromatography and TLC plates (silica gel G) were obtained from Qingdao Sea Chemical Factory (Qingdao, China).

Crassicauline A (purity ≥98%) was purchased from Chengdu Desite Bio-Technology Co., Ltd. (Chengdu, China), aconitine (purity = 99%) from Shaanxi Herbchem Bio-Technology Co., Ltd. (Xi’an, China). Propafenone hydrochloride (purity = 99.8%) and lidocaine hydrochloride (purity = 93.4%) were purchased from National Institutes for Food and Drug Control (Beijing, China), urethane from Chengdu Kelong Chemical Reagent Factory (Chengdu, China). Compounds 1–4 (purity ≥95%) were separated and identified from processed crassicauline A, the structures were verified by ¹H NMR, ¹³C NMR, and HR-ESI-MS.

HPLC-grade acetonitrile was supplied by Fisher Chemical Co. (New Jersey, United States). Ultrapure water was directly obtained from a ULUP-I-10T water purification system (Chengdu, China). The remaining reagents were all of analytical grade. Spots on TLC plates were visualized with Dragendorff’s reagent or iodine vapor.

SPF grade Sprague Dawley rats (male and female) weighing between 180 and 220 g [certificate No. SCXK (CHUAN) 2015-030 and 2020-030] were supplied from Chengdu Dossy experimental animals Co., Ltd. (Chengdu, China) and maintained under a controlled light/dark cycle and temperature (20 ± 2°C), with free access to food and water. They were left for 2 days for acclimatization to animal room conditions. The animal study was reviewed and approved by the Animal Experimentation Ethics Committee of Chengdu University of Traditional Chinese Medicine (permission No. 2020-15).

100.60 mg crassicauline A standard was weighed accurately into a 250 ml volumetric flask and dissolved in dichloromethane to make a concentration of 402.4 μg/ml reaction stock solution. 4 ml of the reaction stock solution was added into a 10 ml volumetric flask, with dichloromethane evaporating, 0.1% (v/v) HCl-methanol was added for dissolution to prepare a final concentration of 160.96 μg/ml. The standard solution was filtered with 0.45 μm syringe filter before injection into the HPLC.

Twenty-four 100 ml round-bottomed flasks were divided into four temperature groups (120°C, 140°C, 160°C, and 180°C), six reaction time points (1, 5, 10, 20, 30, and 40 min) were set at each temperature. Accurately adding 4 ml of the reaction stock solution into each flask, and the solvent was evaporated. The flasks were subsequently immersed in an oil bath and processing according to set parameters, cooling to room temperature after reaction. The residue was diluted with 0.1% (v/v) HCl-methanol in a 10 ml volumetric flask and subsequently filtered with 0.45 μm syringe filter before injection into the HPLC.

The rats were anesthetized by intraperitoneal (i.p.) injection of 20% urethane (1.2 g/kg), with their back fixed and four limbs in subcutaneous penetration of needle-electrodes. Lead II ECGs were recorded after the administration of urethane.

In vivo studies were carried out to assess the cardiotoxicity of crassicauline A and its converted products. 40 rats were equally divided into four groups, five female and five male rats for each group were intravenously injected with the same dose of compounds. Lead II ECGs were recorded for 20 min prior to drug administration, recording the ECG changes within 30 min after administration (Figure 1).

Through the preliminary experiment, it was found that 0.10 mg/kg crassicauline A could cause ventricular premature beat (VPB), ventricular tachycardia (VT), and ventricular fibrillation (VF) in rats. Comparing under the same dose, whether the converted products caused arrhythmias can directly reflect the cardiotoxicity pre- and post-processing.

To further investigate the antiarrhythmic activity of the three converted alkaloids, the testing was performed in vivo in 233 rats. The rats were randomly divided into the following 15 groups, a minimum of ten animals were used in each group: the blank solvent group (Take 4 ml of 1% HCl-ethanol solution, dilute it with saline, add 5% NaOH to adjust pH 7, and fixed volume with saline to 100 ml), the control group (Aconitine, 0.03 mg/kg body weight), the positive control groups (Lidocaine, 5.0 mg/kg (Hong et al., 2019) body weight; propafenone, 3.2 mg/kg (Li et al., 2006) body weight), the compound 1 groups (0.05, 0.20, 0.40, and 0.60 mg/kg body weight), the compound 2 groups (0.05, 0.20, 0.30, and 0.40 mg/kg body weight), and the compound 4 groups (0.05, 0.20, and 0.40 mg/kg body weight). The preparation methods of aconitine, positive drugs, and the experimental compounds were the same as the blank solvent. To confirm whether the blank solvent had an influence on rats, the rats of solvent group were only administered the same volume of blank solvent to observe ECG changes during the recording time.

The rats were anesthetized using 20% urethane (1.2 g/kg, i.p.) (Bai et al., 2012; Mohamed et al., 2016). Recording the lead II ECGs for 20 min prior to administration, the experimental compounds, positive drugs, and equal volume of saline were subsequently administered via the exposed vena femoral, respectively. After stabilization for 10 min (Zhang et al., 2006), aconitine was administered into the uncovered vena femoral at a dosage of 0.03 mg/kg (Bartosová et al., 2005; Klekot, 2006; Bartosova et al., 2007) to establish arrhythmia (Figure 1). The onset time of VPB (Qiu et al., 2016) was recorded within 30 min (Sun et al., 2006; Zhang and Xiong, 2015; Qiu et al., 2016) after aconitine. Meanwhile, the VT (Meng et al., 1992; Ma et al., 2018) or arrhythmia (Wang et al., 1997a; Wang et al., 1997b), if any, was recorded for each group at the end of the observation period.

The data obtained were analysed using statistical package for social sciences (SPSS) version 20 software. Experimental data were expressed as mean ± SD or proportion. Descriptive statistics were examined individually. When the data conformed to the normal distribution, the one-way ANOVA would be used for those with homogeneous variance. If the variance was not homogeneous, Tamhane’s T2 test would be used for comparison between groups. Comparisons of proportions were made with the Pearson’s chi-square (χ ²) test. Statistical significance was set at p < 0.05.

Analysis of the peak area of the standard sample resulted from HPLC as well as linear regression (Y = 21239X-3038.8, r = 0.9996), the obtained results showed that crassicauline A exhibited satisfactory linearity within the range of 2–200 μg/ml. The LOD and LOQ of crassicauline A under the chromatographic conditions were 0.18 μg/ml and 0.60 μg/ml, respectively.

Precision, stability, and repeatability of the method were additionally verified. During the determination, the RSD values of crassicauline A for the precision, stability, and repeatability were 1.29, 2.33, and 2.69%, respectively, indicating that the established method was precise, stable, and accurate enough for determination.

To screen out the temperature and time range for the structural transformation of crassicauline A in the sand frying process, the sample solutions, acquired under different processing temperatures and time, along with the standard solutions, were injected into the HPLC, respectively. The chromatograms were obtained after determination, which could visually reflect the dynamic change of crassicauline A. The chromatograms are shown in Figures 2A–D.

In light of the above experimental findings (Figures 2A–E), it could be found that the content of crassicauline A before processing was 160.96 μg/ml. When crassicauline A was processed at 140–180°C, with the elevation of temperature and prolongation of processing time, the content of crassicauline A decreased. For instance, the content of crassicauline A was 69.07 μg/ml after being processed for 10 min at 160°C and continuously dropped to 24.42 μg/ml for 20 min. When processed at 180°C for 5 min, the content decreased to 55.47 μg/ml, even declined to 0 μg/ml for another 15 min. Meanwhile, some unknown chromatographic peaks emerged in chromatograms. As a result, the quantity and content of the above transformed products were increased obviously when processed at 160°C for 30 min. Therefore, in this paper, the processed products were prepared based on above processing parameters.

Compounds 1-4 were subsequently isolated from processed crassicauline A by procedures described in the experimental section, and the purity of converted products was determined by re-injected into the HPLC (Figure 2F).

16-demethoxy-∆¹⁵⁽¹⁶⁾-8-deacetylcrassicauline A (1) was isolated as an amorphous powder, [ α ] D 20 + 47.1°(c = 0.92, CH₃OH). It showed a positive reaction with Dragendorff’s reagent. Its molecular formula was deduced to be C₃₂H₄₃NO₈ from a protonated molecule ion at m/z 570.3052 (M + H)⁺ (calcd. 570.3061) in its HR-ESI-MS. The NMR spectra (Table 1) of compound 1 showed the presence of an N-ethyl group (δ _H 1.06, 3H, t, J = 7.32 Hz; δ _H 2.51, 2.59, each 1H, m; δ _C 13.3 q, 49.3 t), three methoxyl groups (δ _H 3.24, 3.30, 3.32, each 3H, s; δ _C 56.2 q, 57.6 q, 59.2 q), one p-methoxyl benzoyloxy group (δ _H 6.92, 7.90, each 2H, AA’BB’ system, J = 8.82 Hz; 3.85, 3H, s; δ _C 167.1 s, 121.6 s, 131.7 d (2C), 114.0 d (2C), 163.9 s, 55.5 q), as well as four quaternary carbons (δ _C 39.4, 50.2, 73.9, 76.9). The aforementioned NMR features suggested an aconitine-type alkaloid for compound 1 (Pelletier et al., 1984). The ¹H-doublet signal at δ _H 5.17 (J = 5.1 Hz) was assigned to H-14β (Pelletier and Joshi, 1991; Gao et al., 2006), resulting in location of the p-methoxyl benzoyloxy group to C-14. Three methoxyl groups were attributed to C-1, C-6, and C-18 based on the cross-peaks between 1-OCH₃ (δ _H 3.24) and C-1, 6-OCH₃ (δ _H 3.30) and C-6, 18-OCH₃ (δ _H 3.32) and C-18 in its HMBC spectrum (Figure 3). Two hydroxyl groups were assigned to C-8 and C-13 based on the correlations between the C-8 and H-6, H-10, H-14, H-16, H-17, as well as C-13 and H-9, H-15 in the HMBC of 1. The ¹H NMR signals at δ 5.61 (1H, d, J = 9.54 Hz, H-15), δ 5.91 (1H, d, J = 9.54 Hz, H-16), and the ¹³C NMR signals at δ 130.1 and 134.8 indicated the existence of a disubstituted double bond in the molecule (Yue et al., 1994a; Yue et al., 1994b; Desai et al., 1998). Also, long-range correlations between C-13 and H-15, C-8 and H-16 confirmed the presence of a double bond between C-15 and C-16. In addition, the coupling constant between H-15 and H-16 was 9.54 Hz, suggesting that the two hydrogen atoms were CIS structures. The key NOE correlations (Figure 4) between H-15 and H-17, H-16 and H-17 also confirmed the configuration of H-15 and H-16 in 1 was α-orientation. The NMR spectra of compound 1 lacked a methoxyl group at C-16, when compared with crassicauline A (Chen et al., 2002), and the ¹³C NMR spectra of 1 clearly showed changes in the chemical shifts of C-8, C-15, and C-16 due to the absence of an acetoxyl group and the presence of a double bond. Except for these points, the ¹³C NMR spectra of the two alkaloids were very similar. From these deductions, the structure of compound 1 was assigned as 16-demethoxy-∆¹⁵⁽¹⁶⁾-8-deacetylcrassicauline A, a new diterpenoid alkaloid (Supplementary Figures S1–S13).

16-demethoxy-∆¹⁵⁽¹⁶⁾-8-O-methylcrassicauline A (2) was isolated as a white amorphous powder, [ α ] D 20 + 61.5°(c = 0.78, CH₃OH). Its molecular formula C₃₃H₄₅NO₈ was derived from the HR-ESI-MS spectral data: m/z 584.3203 (M + H)⁺ (cacld. 584.3218). ¹H- and ¹³C-NMR spectra of 2 (Table 2) showed the distinct NMR features of an aconitine-type C₁₉-diterpenoid alkaloid skeleton (Pelletier et al., 1984), bearing an N-ethyl group (δ _H 1.05, 3H, t, J = 7.38 Hz; δ _H 2.49, 2.54, each 1H, m; δ _C 13.3 q, 49.2 t), four methoxyl groups (δ _H 3.12, 3.23, 3.26, 3.29, each 3H, s; δ _C 49.1 q, 56.2 q, 57.8 q, 59.2 q), a disubstituted double bond (δ _H 5.80, 1H, d, J = 9.9 Hz; δ _H 6.03, 1H, d, J = 9.9 Hz), and a p-methoxyl benzoyloxy group (δ _H 6.89, 7.98, each 2H, AA’BB’ system, J = 8.82 Hz; 3.84, 3H, s; δ _C 167.3 s, 122.7 s, 131.9 d (2C), 113.5 d (2C), 163.4 s, 55.4 q). The ¹H-doublet signal at δ _H 4.95 (J = 4.08 Hz) was assigned to H-14β (Pelletier and Joshi, 1991; Gao et al., 2006), resulting in location of the p-methoxyl benzoyloxy group to C-14. Four methoxyl groups were assigned to C-1, C-6, C-8, C-18 due to the long-range correlations between 1-OCH₃ (δ _H 3.23) and C-1, 6-OCH₃ (δ _H 3.26) and C-6, 8-OCH₃ (δ _H 3.12) and C-8, 18-OCH₃ (δ _H 3.29) and C-18 in the HMBC spectrum of 2 (Figure 3). A remaining hydroxyl group was assigned to C-13 based on the correlations between the C-13 and H-9, H-15 in the HMBC of 2. Also, the presence of a double bond between C-15 and C-16 could be corroborated by the HMBC correlations (Figure 3) from H-15 to C-13 and from H-16 to C-8 (Yue et al., 1994a; Yue et al., 1994b; Desai et al., 1998). The NOE correlations (Figure 4) could be observed between H-15 and H-17, H-16 and H-17 in compound 2. As a result, the configuration of H-15 and H-16 in compound 2 was unambiguously established to have an α-orientation.

Comparison of the NMR spectra of 2 with that of crassicauline A (Chen et al., 2002), it was showed that compound 2 bore a methoxyl group at C-8 instead of an acetoxyl group, except for the absence of a methoxyl group at C-16 and the induction of a double bond at C-15 and C-16. Compound 2 exhibited nearly identical ¹H and ¹³C NMR resonances to those of 1, and the distinction between the two sets of spectra was demonstrated by the presence of an additional methoxyl group at C-8 in 2 instead of a hydroxyl group in 1. All of the above arguments determined the structure of 2 as 16-demethoxy-∆¹⁵⁽¹⁶⁾-8-O-methylcrassicauline A, a new diterpenoid alkaloid (Supplementary Figures S14–S26).

16-demethoxy-∆¹⁵⁽¹⁶⁾-crassicauline A (3) was obtained as colorless needles, mp 155–157°C, [ α ] D 20 + 25.4°(c = 0.80, CH₃OH). A protonated molecular ion at m/z 612.3163 (M + H)⁺ (cacld. 612.3121) in the HR-ESI-MS spectrum suggested a molecular formula of C₃₄H₄₆NO₉. Compound 3 exhibited characteristic NMR features (Table 3) of an aconitine-type alkaloid (Pelletier et al., 1984), bearing an N-ethyl group (δ _H 1.05, brs, 3H; δ _H 2.50, 2.53, each 1H, m; δ _C 13.3 q, 49.3 t), three methoxyl groups (δ _H 3.15, 3.24, 3.27, each 3H, s; δ _C 57.4 q, 56.1 q, 59.2 q), an acetoxyl group (δ _H 1.39, 3H, s; δ _C 169.6 s, 21.8 q) (Joshi et al., 1987), as well as a p-methoxyl benzoyloxy group (δ _H 6.88, 7.94, each 2H, AA’BB’ system, J = 6.96 Hz; 3.85, 3H, s; δ _C 166.6 s, 122.4 s, 131.7 d (2C), 113.7 d (2C), 163.6 s, 55.4 q). The one-proton double signal (J = 4.8 Hz) at δ _H 4.92 was attributed to H-14β (Pelletier and Joshi, 1991; Gao et al., 2006), implying the appearance of p-methoxyl benzoyloxy group at C-14 position. The location of three methoxyl groups was based on the long-range correlations between 1-OCH₃ (δ _H 3.24) and C-1, 6-OCH₃ (δ _H 3.15) and C-6, 18-OCH₃ (δ _H 3.27) and C-18 in HMBC spectrum (Figure 3). Furthermore, a hydroxyl group was assigned to C-13 based on the correlations between the C-13 and H-9, H-15 in the HMBC of 3. The ¹³C-NMR signals at δ 125.4 and 137.3 showed that a disubstituted double bond was located at C-15 and C-16. The ¹H-NMR signal at δ 6.53 (1H, d, J = 9.9 Hz, H-15) and δ 6.05 (1H, d, J = 9.9 Hz, H-16) also confirmed this deduction (Yue et al., 1994a; Yue et al., 1994b; Desai et al., 1998).

Comparison of the NMR data of 3 with crassicauline A (Chen et al., 2002) showed that except a disubstituted double bond, compound 3 lacked a methoxyl group at C-16. The NMR spectra of compound 3 were also similar to those of 1 and 2 except for the substituents at C-8. The C-8 substituents of 1 and 2 were hydroxyl group and methoxyl group, respectively, while 3 was acetoxyl group. The key NOE correlations (Figure 4) could be observed between 8-OAc and H-2′, 6′ in compound 3. Therefore, the configuration of 8-OAc in 3 was unambiguously established to have a β-orientation. Thus, the structure of 3 was elucidated 16-demethoxy-∆¹⁵⁽¹⁶⁾-crassicauline A, a new diterpenoid alkaloid (Supplementary Figures S27–S39).

Pyrocrassicauline A (4) was obtained as colorless needles, mp 129–131°C, [ α ] D 20 + 182.4°(c = 0.81, CH₃OH). Its molecular formula, determined to be C₃₃H₄₅NO₈, was derived from HR-ESI-MS: m/z 584.3222 (M + H)⁺ (calcd. 584.3218). ¹H and ¹³C NMR spectra of 4 (Supplementary Table S1) showed the distinct NMR features of an aconitine-type C₁₉-diterpenoid alkaloid skeleton (Pelletier et al., 1984), bearing an N-ethyl group (δ _H 1.07, 3H, t, J = 7.2 Hz; δ _H 2.54, 2.66, each 1H, m; δ _C 13.6 q, 49.9 t), four methoxyl groups (δ _H 3.24, 3.28, 3.30, 3.38, each 3H, s; δ _C 56.4 q, 58.2 q, 59.3 q, 57.2 q), and a p-methoxyl benzoyloxy group (δ _H 6.88, 7.99, each 2H, AA’BB’ system, J = 8.4 Hz; 3.84, 3H, s; δ _C 167.9 s, 122.9 s, 132.0 d (2C), 113.5 d (2C), 163.3 s, 55.4 q). The one-proton double signal (J = 3.0 Hz) at δ _H 4.93 was attributed to H-14β (Pelletier and Joshi, 1991; Gao et al., 2006), implying the appearance of a p-methoxyl benzoyloxy group at C-14. Four methoxyl groups were assigned to C-1, C-6, C-16, C-18 due to the long-range correlations between 1-OCH₃ (δ _H 3.24) and C-1, 6-OCH₃ (δ _H 3.28) and C-6, 16-OCH₃ (δ _H 3.38) and C-16, 18-OCH₃ (δ _H 3.30) and C-18 in the HMBC spectrum of 4 (Figure 3). Along with the above-mentioned signals, its ¹³C NMR displayed eight oxygenated carbon signals, suggesting that compound 4 possessed an additional hydroxyl group, which could be located at C-13 due to the HMBC correlations between the C-13 and H-15 of 4. Furthermore, the ¹³C NMR signals at δ 146.9 and 116.0 showed that a trisubstituted double bond was located at C-8 and C-15. The ¹H NMR signal at δ 5.53 (1H, d, J = 6.0 Hz, H-15) also confirmed this deduction (Wang et al., 2009). The configuration of 16-OCH₃ in 4 was also determined to have a β-orientation according to the key NOE correlations between 16-OCH₃ and H-2′, 6′ (Figure 4). The ¹³C NMR spectra of 4 and crassicauline A were very similar except for the chemical shifts of C-8, C-15 signal appeared at low field, caused by the absence of an acetoxyl group and the presence of a double bond (Yue et al., 1994a; Yue et al., 1994b; Desai et al., 1998; Chen et al., 2002). All of the above arguments determined the structure of 4 as pyrocrassicauline A (Supplementary Figures S40–S53).

Intravenous injection of 0.10 mg/kg crassicauline A caused arrhythmias in normal rats, such as VPB, VT, and VF, accompanied by regular chest twitching and convulsions. The incubation period of crassicauline A induced arrhythmia was (182.8 ± 84.58) s. In contrast, the three converted products did not exhibit arrhythmias and behavioral manifestations under the same dose. Through this comparative experiment, it could be concluded that the cardiotoxicity was reduced after processing (Table 4; Supplementary Figure S54).

VPB is the initial manifestation of aconitine-induced arrhythmia model rats, and the ECG of VPB is characterized by premature and bizarrely shaped QRS complexes that appear wide on the ECG, these complexes are not preceded by a P wave, and a T wave is usually oriented in a direction opposite the major deflection of the QRS (Casas et al., 2018; Bae and Kwon, 2019). VPB incubation period refers to the time after the administration of aconitine to the first occurrence of VPB (Qiu et al., 2016). The length of VPB incubation period can reflect the antiarrhythmic effect of the experimental compounds, the longer the incubation period is, the better the antiarrhythmic effect is.

In the control group, the typical characteristic ECGs of VPB and VT occurred successively following the injection of aconitine, even VF occurred in parts of the rats, and lasted for more than 30 min, suggesting that the arrhythmia model was successfully replicated. The VPB incubation period of the control group was (116.5 ± 36.4) s, lidocaine and propafenone groups were (280.3 ± 128.7) s and (193.3 ± 39.9) s, respectively. Two positive drugs had a significant difference in comparison with the control group (p < 0.05).

The effects of compounds 1, 2, and 4 on VPB incubation period in rats are shown in Table 5; Figure 5. Compared with the control group, different dose groups of compound 1 (0.20, 0.40, and 0.60 mg/kg) significantly delayed the emerge of VPB (p < 0.05, Figure 5A). The VPB incubation periods in 0.40 and 0.60 mg/kg groups were (493.7 ± 148.7) s and (547.3 ± 241.8) s, respectively. They were significantly different from propafenone (p < 0.05). Similarly, there was a significant difference between 0.40 mg/kg group and lidocaine group (p < 0.05), indicating that a 0.40 mg/kg or more intravenous dose of compound 1 exhibited a marked activity relative to positive drugs.

As mentioned above, the differentiation between compounds 1 and 2 is the substituents at C-8 position. To explore the effects of these structural distinctions, the antiarrhythmic effect of compound 2 was further evaluated. In comparison with the control group, different dose groups of compound 2 (0.20, 0.30, and 0.40 mg/kg) had a significant delay on the VPB incubation period (p < 0.05, Figure 5B). The VPB incubation periods in 0.20, 0.30, and 0.40 mg/kg groups were (441.4 ± 202.3) s, (800.5 ± 353.2) s, and (920.5 ± 358.1) s, respectively, which were significantly different from propafenone (p < 0.05). Compared with lidocaine, 0.30 and 0.40 mg/kg groups had significant difference (p < 0.05). It was demonstrated that a 0.30 mg/kg or more dose of compound 2 exhibited superior antiarrhythmic activity relative to compound 1 and the positive drugs.

On the contrary, compound 3, bearing two ester bonds at C-8 and C-14, belonging to DDAs, may exert relatively strong toxicity (Gong, 2005). Consequently, its antiarrhythmic activity was not investigated. Different from compounds 1 and 2, the double bond of compound 4 is located at C-8/C-15 rather than C-15/C-16. To exemplify different positions of double bonds on the strength of the antiarrhythmic efficacy, we further investigated the antiarrhythmic effect of compound 4. Different dose groups of compound 4 (0.20 and 0.40 mg/kg) significantly delayed the onset time of VPB compared with the control group (p < 0.05, Figure 5C). The incubation period of VPB was (382.9 ± 146.5) s in the 0.40 mg/kg group, and it was significantly different from propafenone (p < 0.05), but showed an equivalent effect relative to lidocaine (p > 0.05).

In summary, within their own dose ranges, compounds 1, 2, and 4 could delay the onset time of VPB in a dose-dependent manner. Additionally, at the dose of 0.40 mg/kg, the VPB incubation periods were compound 2 >compound 1 >compound 4, indicating that compound 2 had the best antiarrhythmic effect among these compounds.

VT is the result of further progression of VPB, as well as an established risk factor for sudden cardiac death (Batiste et al., 2019). The ECGs of VT are characterized by three or more consecutive beats of apparent ventricular origin, the QRS complexes are wide and deformed, without constant P wave (Swerdlow et al., 1983; Baldzizhar et al., 2016; Brady et al., 2017). The incidence of VT can be used to evaluate whether the experimental compound can effectively prevent the progress of VPB, the lower the occurrence is, the better the efficacy is.

In accordance with the data requirements of chi-square test, the adjacent dose groups with similar VT incidence should be combined to carry out chi-square test. The subgroups of compound 1 were 0.05–0.20, 0.40, and 0.60 mg/kg. As vividly displayed in Table 6; Figure 6A, chi-square test showed a significant difference in the incidence of VT among the three groups (χ ² = 14.704, p = 0.001). The pairwise comparison revealed a highly significant drop in the incidence of VT from 85.7% in the 0.05–0.20 mg/kg group to only 18.2% in the 0.60 mg/kg group (χ ² = 11.313, p = 0.001).

Following the identical research approach as VPB incubation period, we further investigated the incidence of VT in compounds 2 and 4. Compound 2 was divided into the 0.05, 0.20–0.30, and 0.40 mg/kg groups. As described in Table 6; Figure 6B, chi-square test showed that there was a marked difference in the incidence of VT among the three groups (χ ² = 13.000, p = 0.002). In comparison with the 0.05 mg/kg (81.8%) and 0.20–0.30 mg/kg (61.9%) groups, the incidence of VT was significantly dropped to 10% in 0.40 mg/kg group (χ ² = 8.144, p = 0.004; χ ² = 5.422, p = 0.020).

The subgroups of compound 4 were 0.05–0.20 and 0.40 mg/kg groups. As presented in Table 6; Figure 6C, the chi-square test showed a highly difference in the incidence of VT between the two subgroups (χ ² = 10.165, p = 0.001). Furthermore, there was a greatly significant reduction in the incidence of the VT from 85.7% in 0.05–0.20 mg/kg group to 36.8% in 0.40 mg/kg group (χ ² = 10.165, p = 0.001).

On the basic of aforementioned results, it can be concluded that compounds 1, 2, and 4 could reduce the incidence of VT in a dose-dependent manner, suggesting that they could effectively prevent the further progression of VPB.

The incidence of arrhythmia (Wang et al., 1997a; Wang et al., 1997b) is defined as the proportion of arrhythmia that occurs within 30 min after pre-intravenous injections of test drugs, followed by an arrhythmia model established with aconitine. The occurrence of any kinds of ECGs such as VPB, VT, or VF should be judged as arrhythmia. Arrhythmia inhibition rate (%) = 100%-arrhythmia incidence rate (%), that is, the proportion that no arrhythmia occurs within 30 min, which can be used to evaluate whether the experimental compound could completely inhibit the proarrhythmic effect of aconitine. Arrhythmia inhibition rate is the most intuitive index reflecting the strength of the drug efficacy, the higher it gets, the better the efficacy is.

Considering the number of rats without arrhythmia in dose groups of 0.05 and 0.20 mg/kg was small, the original grouping could not meet the requirements of the chi-square test. Therefore, adopting the similar data analysis method as VT incidence, chi-square analysis was carried out in this paper by combining 0.05 and 0.20 mg/kg groups into 0.05–0.20 mg/kg group in each experimental compound.

In this paper, the effects of different doses of compound 1 (0.05–0.20, 0.40, and 0.60 mg/kg) and lidocaine on arrhythmia inhibition rate in rats were compared. Chi-square test showed that the arrhythmia inhibition rates in the four groups were significantly different (χ ² = 15.457, p = 0.001), as shown in Table 7. Compared with 0.05–0.20 mg/kg group, the arrhythmia inhibition rates in 0.60 mg/kg and lidocaine groups significantly increased from 4.5% to 56% and 42.9%, respectively (χ ² = 14.258, p = 0.0002; χ ² = 6.820, p = 0.009). Notably, the arrhythmia inhibition rate in 0.60 mg/kg group was higher than lidocaine, exhibiting a better effect, as shown in Figure 6D.

To ascertain the influence of different substituents at C-8 and the location of double bond on drug efficacy, we further analysed the arrhythmia inhibition rate of compounds 2 and 4, the result of chi-square test showed that the arrhythmia inhibition rates in compound 2 subgroups (0.05–0.20, 0.30, and 0.40 mg/kg) and lidocaine were statistically different (χ ² = 18.123, p = 0.0004), as shown in Table 7. In comparison with 0.05–0.20 mg/kg group, the arrhythmia inhibition rate in 0.30 mg/kg, 0.40 mg/kg, and lidocaine groups significantly rose from 8.3% to 54.5%, 64.3%, and 42.9%, respectively (χ ² = 11.578, p = 0.001; χ ² = 17.093, p = 0.00004; χ ² = 7.228, p = 0.007, respectively.), indicating that the antiarrhythmic effect of compound 2 was stronger with the increasing dose. Furthermore, a 0.30 mg/kg or more dosage of compound 2 exhibited a superior antiarrhythmic effect than lidocaine, as illustrated in Figure 6E.

Chi-square test showed that the arrhythmia inhibition rates in different dose groups (0.05–0.20 and 0.40 mg/kg) of compound 4 and lidocaine were significantly different (χ ² = 6.908, p = 0.032), as shown in Table 7. The pairwise comparison result showed that the arrhythmia inhibition rate of lidocaine was significantly higher than 0.05–0.20 mg/kg group (χ ² = 6.832, p = 0.009). However, a 0.40 mg/kg intravenously dose of compound 4 exhibited an equal antiarrhythmic activity relative to lidocaine (p > 0.05), as displayed in Figure 6F.

In summary, within their respective dose ranges, compounds 1, 2, and 4 could increase the arrhythmia inhibition rate in a dose-dependent manner. Among them, compound 1 in 0.60 mg/kg group (56%), compound 2 in 0.30 and 0.40 mg/kg groups (54.5%; 64.3%) had a higher arrhythmia inhibition rate than lidocaine (42.9%), manifesting that these two compounds possessed excellent antiarrhythmic effects.