Increase water solubility of Centellaasiatica extract by indigenous bioenhancers could improve oral bioavailability and disposition kinetics of triterpenoid glycosides in beagle dogs

A newly standardised extract of Centella asiatica (Centell-S) with better water solubility than the previous standardised extract of C. asiatica (ECa 233) was developed, and pharmacokinetic profiles of bioactive triterpenoids were investigated in beagle dogs. The test substances were administered via intravenous or oral administration with single and multiple doses for 7 days. The concentrations of major bioactive triterpenoids, including madecassoside, asiaticoside, madecassic acid, and asiatic acid, in biological samples were measured by liquid chromatography–tandem mass spectrometry. The dogs in this study showed good tolerability to all test substances, based on the physical appearance and blood chemistry 24 h after dosing. The major bioactive triterpenoids found in systemic blood circulation were madecassoside, asiaticoside, and asiatic acid; the concentration of these components ranged from 1 to 10,000 µg/L after intravenous administration of 1.0 mg/kg Centell-S. Oral administration of 10 and 20 mg/kg Centell-S generated approximately twofold higher plasma levels of both madecassoside and asiaticoside compared with equivalent doses of ECa 233. In addition, there was an accumulation of triterpenoid glycosides after multiple oral administrations of Centell-S for 7 days, while triterpenic acids showed little tendency for accumulation. Beagles had good tolerability to both standardised extracts of C. asiatica, and showed a similar pattern of bioactive triterpenoids to humans. Centell-S increased oral bioavailability of major triterpenoid glycosides and can be further developed into a phytopharmaceutical product.

Animals. Eight healthy 8-month-old male beagles were obtained from the Chulalongkorn University Laboratory Animal Centre (CULAC), Chulalongkorn University. The dogs were randomly divided into two groups and housed in a controlled environment with temperature of 22 ± 2 °C, relative humidity 50% ± 20%, and a 12 h light/dark cycle. Food was provided twice a day, and water was available ad libitum. The dogs were fasted 12 h before the experiment, with free access to water. After the administration of the test substances, the dogs were placed in metabolic cages until 48 h post-administration. The animal experiments were approved by the Institutional Animal Care and Use Committee of the CULAC, Chulalongkorn University, Bangkok, Thailand (Protocol number 2073002, Approval date: February 21, 2020). All animal procedures were complied with the ethical principles and guidelines for the use of animals (National Research Council of Thailand, 2015). All animal experiments followed the ARRIVE reporting guidelines for documenting animal experiments.
Pharmacokinetic study. For intravenous administration, there were four groups: vehicle, 0.46 mg/kg madecassoside, 0.41 mg/kg asiaticoside, and 1.00 mg/kg Centell-S. The test substance solutions were freshly prepared by using 20% v/v dimethyl sulfoxide/sterile water for injection and filtered through 0. 22 Sample preparation and method validation. All biological samples (plasma, urine, and faeces) were subjected to protein precipitation to extract the major triterpenoids from the samples. Two hundred microliters of methanol with two internal standards (glycyrrhizin and glycyrrhetinic acid) was added to 50 µL of the biological sample and mixed for 10 min. Then, the sample was centrifuged at 12,000g for 10 min at 4 °C, and 10 µL of supernatant was injected into the LC-MS/MS system. Faeces samples were homogenised by adding methanol, then the mixture was centrifuged at 1500g for 10 min at 4 °C to collect the supernatant. In the case of the concentration of the biological sample exceeded the linear calibration curve; blank matrices were used to dilute before protein precipitation. For method validation, all major components in the biological sample of dog were determined and validated according to the previous reports with little modification 15 . The method showed  www.nature.com/scientificreports/ good specificity and linearity and no interference from other substances at the retention times of the targeted analytes and internal standards.
Instrumentation. The LC-MS/MS system was set up according to a previous published method 15 . In brief, an LCMS-8060 coupled with a triple quadrupole mass spectrometer was controlled by LabSolution software version 5.86 (Shimadzu Corp.). A Synergi Fusion-RP C18 column (Phenomenex Inc.) maintained at 40 °C was used as the stationary phase. The analysis was run using absolute methanol and 0.2% formic acid in water as the mobile phase. The mobile phase used for gradient elution was 10% methanol at 0.0-0.5 min, increased to 90% methanol at 1.5 min and maintained until 3.0 min, then decreased to 10% methanol until 5.0 min at a flow rate of 0.5 mL/min. Multiple reaction monitoring with negative mode ionisation was performed to obtain the peak of madecassoside ( , with a range from 1 to 10,000 µg/L for parent compounds (madecassoside and asiaticoside) and from 1 to 1000 µg/L for active metabolites (madecassic acid and asiatic acid). The lower limits of detection for both parent compounds and active metabolites were 1.00 µg/L. For identification of minor compositions of Centell-S, an Agilent 6540 Q-TOF-MS spectrometer (Agilent Technologies Inc.) coupled with an Agilent 1260 Infinity Series High performance liquid chromatography system (Agilent Technologies Inc.). The separation was performed with a Luna C18 column, size 4.6 mm × 150 mm, 5 µm (Phenomenex Inc.) at a flow rate of 500 µL/min and the control temperature at 35 °C. The mobile phase A was water type I and B was acetonitrile. Both phases contained 0.1% (v/v) formic acid. The gradient elution mode started with 5% solvent B to 95% solvent B linear gradient within 30 min, and maintained at this ratio for 10 min after that post-run for 5 min. The injection volume was 10 µL. The operating parameters for MS detection were as follows: drying gas (N 2 ) flow rate 10 Data analysis. To evaluate significant differences between the pre-and post-dose of the test substances to the health status of dogs, the biochemical data were compared by using Student's t-test. PK solution software version 2.0 (Summit Research Services) was used to analyse pharmacokinetic parameters by non-compartmental analysis. C max and T max were directly determined from plasma concentration-time curves. AUC 0-24 was calculated using the trapezoidal rule and extrapolated to time infinity by the equation AUC 0-inf = AUC 0-24 + (C last /k el ), where C last is the last observed plasma concentration after dosing and k el is the elimination rate constant. MRT was calculated as AUMC 0-inf /AUC 0-inf , where AUMC 0-inf is the area under the first moment concentration-time curve. Vd was equal to dose/C 0 , where C 0 is the plasma concentration at time zero and CL was calculated as dose/AUC 0-inf . The terminal elimination half-life was determined by dividing 0.693 by k el . Absolute bioavailability was calculated as (AUC p.o. /DOSE p.o. )/(AUC i.v. /DOSE i.v. ) × 100. Relative bioavailability was calculated from (AUC CTS /DOSE CTS )/(AUC ECa /DOSE ECa ) × 100. Before performing the statistical analysis in SPSS version 22.0 (IBM Corp.), the Shapiro-Wilk test was used to determine whether the data were normally distributed. A paired Student's t test or Wilcoxon signed-rank test was used to determine significant differences of pharmacokinetic parameters between the groups. The percent recovery of triterpenoid glycosides and triterpenic acids were calculated using the total amount of the test substance found in urine or faeces divided by the given dose. All data are expressed as the mean ± standard deviation except for Tmax and half-life, which were expressed as median (IQR). Differences were considered significant at p < 0.05.

Results
Plasma concentration-time profiles. All of the dogs used in the study tolerated the test substances.
Blood samples were collected from each dog pre-and post-administration of madecassoside, asiaticoside, ECa 233, and Centell-S. Then, the plasma was sent for clinical laboratory testing to determine haematology, kidney, and liver functions. There were no significant differences in the measured biochemical parameters between predose and post-dose in any group (Supplementary Table S1).
The plasma concentration-time profile of triterpenoid glycosides after single and multiple administrations of intravenous and oral doses of madecassoside, asiaticoside, ECa 233, and Centell-S are shown in Fig. 2. For intravenous administration of 0.46 mg/kg madecassoside, the plasma concentration of this substance reached a maximum of approximately 10,000 µg/L, then declined to 10 µg/L 24 h after dosing (Fig. 2a). For intravenous administration of 0.41 mg/kg asiaticoside, the plasma concentration reached a maximum of approximately 3000 µg/L, then declined to 10 µg/L 8 h after dosing (Fig. 2b). For intravenous administration of 1 mg/kg Centell-S, the plasma concentration of madecassoside and asiaticoside showed similar pharmacokinetic patterns to the pure substances. Remarkably, intravenous administration of pure madecassoside generated a significant amount of asiaticoside and vice versa; this finding may indicate bidirectional in vivo interconversion between the two triterpenoid glycosides. A single oral administration of 20 mg/kg ECa 233 showed maximal madecassoside and asiaticoside plasma concentrations that were approximately twofold higher than oral administration of 10 mg/ kg ECa 233 (Fig. 2c,d). After 7 days of multiple oral administration of 20 mg/kg ECa 233, the maximum plasma concentration of madecassoside was approximately 1.5 fold greater than the first day of administration. The plasma concentrations of madecassoside and asiaticoside showed similar pharmacokinetic patterns between www.nature.com/scientificreports/ oral administration of ECa 233 and Centell-S. However, oral administration of Centell-S was associated with maximal plasma concentrations that were approximately twofold higher compared with equivalent doses of ECa 233. The plasma concentration-time profile of the active metabolites, madecassic acid and asiatic acid, after single and multiple oral administration of ECa 233 and Centell-S are shown in Fig. 3. These active metabolites were detected in the plasma from 30 min to 8 h after administration. The asiatic acid concentration was approximately tenfold higher than madecassic acid. There was no statistically significant difference in the triterpenic acid level between oral administration of ECa 233 and Centell-S.  (Table 1). For oral administration of Centell-S, the time to reach the maximum plasma concentration (T max ) of madecassoside and asiaticoside was observed 0.5-2 h after oral administration. The absolute bioavailability of triterpenoid glycosides after single oral dosing was approximately 0.6-1.2% for Centell-S and 0.3-0.6% for ECa 233. Noticeably, the maximum plasma concentration (C max ) of madecassoside, asiaticoside, and asiatic acid after oral administration of 20 mg/kg Centell-S was approximately twofold higher than the group that received oral administration      www.nature.com/scientificreports/ of 10 mg/kg Centell-S. Furthermore, the accumulation of asiaticoside, as indicated by the C max , showed a statistically significant difference (p < 0.05) between single and multiple doses of Centell-S (Table 2). Oral administration of ECa 233 generated a lower C max and a delayed T max for both triterpenoid glycosides compared with Centell-S. The relative oral bioavailability of both triterpenoid glycosides in Centell-S group was approximately 200% of ECa 233 group. The Vd and CL of madecassoside and asiaticoside after single and multiple oral dosing showed higher values than intravenous administration of Centell-S. Repeated oral administration of ECa 233 and Centell-S prolonged the mean residence time (MRT) of both madecassoside and asiaticoside. The major active metabolite found in beagles was asiatic acid, and repeated oral administration increased the level of asiatic acid. After intravenous administration of madecassoside, asiaticoside, and Centell-S, most of the parent triterpenoid glycosides were excreted via urine within 24 h after dosing. Madecassoside and asiaticoside from oral ECa 233 and Centell-S were mainly excreted via faeces. Negligible amounts of the active metabolites, madecassic and asiatic acids, were found in the excreta of beagles after intravenous or oral dosing of all test compounds (Supplementary Table S2). The volume and weight of excreta which was used to calculate percent recovery of bioactive triterpenoids after receiving the test compounds were shown in Supplementary Table S3.

LC ESI-QTOF-MS/MS analysis.
Regarding the composition of Centell-S assessed by LC ESI-QTOF-MS/ MS analysis, two major and two minor compositions were found (Fig. 4). The first major peak was detected at m/z 1009.4766 [M+Cl] − with a retention time of 14.749 min and was deduced to be a chloride adduct of madecassoside. The second major peak was identified at m/z 993.4824 [M+Cl] − with a retention time of 15.441 min and was deduced to be a chloride adduct of asiaticoside. The standards of madecassoside and asiaticoside were used to confirm the major peaks that showed the identical molecular weight and retention time. The first minor peak was detected at a retention time of 15.058 min with a molecular weight 957.5096 g/mol which was pro-  www.nature.com/scientificreports/ posed as a hydrogen abduct [M−H] − of centellasaponin A (exact molecular weight: 958.51 g/mol). The second minor peak was identified at a retention time of 18.172 min with a molecular weight 977.5004 g/mol which was considered as a chloride adduct [M+Cl] − of asiaticoside F (exact molecular weight: 942.52 g/mol). However, the confirmation of these two minor peaks remained undetermined because of the absence of commercial standards. QTOF-MS/MS was used to confirm the identity of the saponin compound in Centell-S. The pseudo-molecular ion can be seen in chloride adduct form in negative mode and ammonium adduct in positive mode. Madecassoside and asiaticoside were identified and confirmed the mass fragmentation pattern and retention time with authentic compounds. Centellasaponin A and asiaticoside had the same molecular weight and almost the same MS/MS fragmentation pattern. When considering Xlog P, centellasaponin A (XlogP = 0.59) was more hydrophilic than asiaticoside (XlogP = 0.73), so the retention time of centellasaponin A was ahead of asiaticoside. The next compound with m/z 977.5004 [M+Cl] − had 16 Da less than centellasaponin A and was proposed as asiaticoside F.

Discussion
Drug absorption can be altered by many factors, the most common of which is water solubility. A test compound with limited water solubility usually has low absorption, resulting in low oral bioavailability 20 . In general, triterpenoid glycosides, especially madecassoside and asiaticoside, have large molecular sizes and contain a sugar moiety that reduces absorption through the intestinal membrane 13 . Moreover, efflux transporters, for example, P-glycoprotein could efflux the absorbed madecassoside return to the gastrointestinal tract 21 . The standardised extract of ECa 233 has low oral bioavailability due to its limited water solubility 14,15 . To improve the solubility of ECa 233, Centell-S was developed. This test compound is freely soluble in water and, similar to ECa 233, its major components are madecassoside and asiaticoside. The absolute oral bioavailability of both triterpenoid glycosides was determined, and Centell-S had higher value by approximately twofold. Two minor components, centellasaponin A and asiaticoside F were detected in Centell-S that might act as bioenhancers. Previously, our group has reported the minor components of ECa 233, and found that one centellasaponin in ECa 233 could act as a bioenhancer 22 . The improved bioavailability of Centell-S might be due to the improvement in the water solubility of Centell-S.
In this study, we aimed to determine the pharmacokinetic profile of bioactive triterpenoids in Centell-S compared with ECa 233. All experimental animals showed good tolerability to the test substances after oral and intravenous administration. The results implied that the test substances, especially Centell-S, are safe in animals, a finding that is in line with our previous reports in standardised extract of C. asiatica 14,15,19,23 . After intravenous administration of Centell-S, the plasma concentrations of madecassoside and asiaticoside were slightly lower than after pure compound administration. The potential mechanism could involve a minor component found in Centell-S that might act as a bioenhancer and increase the tissue distribution of triterpenoid glycosides. This was correlated with a study by Yurdakok-Dikmen et al. reporting that C. asiatica contains some bioenhancers that can facilitate tissue distribution 24 . Therefore, a higher Vd could be observed for madecassoside and asiaticoside of Centell-S. Improved tissue distribution could result in a lower plasma level of the triterpenoid glycosides, since the blood volume in the vascular compartment is accounts for only 5-10% of the total body volume. Moreover, asiaticoside showed a higher Vd than madecassoside, which also implies that asiaticoside may distribute to tissues better than madecassoside. This phenomenon relates to the partition coefficient of asiaticoside (XlogP = 0.1), which is higher than that of madecassoside (XlogP = − 1.2). Therefore, asiaticoside has greater lipophilic properties than madecassoside prefers residing in tissue compartments. This result is well correlated with a study on ECa 233 in rats, which showed the distribution of asiaticoside to many tissues, including the lungs, heart, brain, skin, and liver, after intravenous administration of ECa 233 14 . We accidentally detected a small peak of asiaticoside generated by intravenous administration of pure madecassoside or vice versa in experimented animals. In addition, there was no peak of asiaticoside found in the fresh preparation and blank plasma of madecassoside group or vice versa. This result was consistent with our report from 2018, which was a comparative pharmacokinetic study between the pure compound and the standardised extract of C. asiatica in rats by Hengjumrut et al. 19 . These results suggested that bidirectional interconversion between madecassoside and asiaticoside could be detected in both rats and dogs. This bidirectional interconversion could prolong systemic exposure of the two glycosides and might increase pharmacodynamic activities of these triterpenoid glycosides in vivo. To add hydroxyl group into asiaticoside structure, it might be needed some oxidative enzymes, e.g., CYP isoforms. In contrary, conversion of madecassoside to asiaticoside might require reduction processes in order to remove hydroxyl group from madecassoside structure. Further study to determine responsible enzymes and appropriate conditions of this bidirectional interconversion is required.
Increasing the Centell-S or ECa 233 oral dose from 10 to 20 mg/kg in beagles increased the C max and AUC of madecassoside and asiaticoside by approximately twofold. After oral administration of Centell-S or ECa 233, the parent triterpenoid glycosides were metabolised to triterpenic acids by anaerobic bacteria in the intestine, a phenomenon that has been reported in many studies 13,15,25 . Asiatic acid might be converted from asiaticoside by hydrolytic degradation of the sugar moiety 26 . In this study, asiatic acid was the main active metabolite detected in the plasma of dogs from 0.5 to 8.0 h after oral administration of Centell-S or ECa 233. This finding correlates well with results from humans, namely that asiatic acid is a major active metabolite in plasma after oral administration of 250-500 mg capsules of ECa 233 15 . The similar pharmacokinetic profiles of bioactive triterpenoids in humans and dogs might imply that these species have a similar gut microbiota, which can biotransform triterpenoid glycosides into triterpenic acids 27,28 . This result was different from the pharmacokinetic study of ECa 233 in rats, which did not detect any active metabolites in the plasma and internal organs of rats 14 . After increasing the concentration of a single oral dose of Centell-S from 10 to 20 mg/kg, the C max of asiatic acid increased approximately twofold. Furthermore, 7 days of repeated Centell-S administration significantly changed www.nature.com/scientificreports/ the C max of asiatic acid compared with the first day. Asiatic acid seems to be the major active metabolite in both ECa 233 and Centell-S in dogs, and many studies have determined that asiatic acid has therapeutic effects for the treatment of various diseases [29][30][31] . In our study, madecassoside and asiaticoside were mainly excreted via the urine after intravenous administration of all test substances. However, after oral administration of ECa 233 and Centell-S, the main route for madecassoside and asiaticoside excretion was via the faeces. A certain amount of asiatic acid was found in the faeces within 24 h after oral administration of Centell-S and ECa 233, suggesting that the gut microbiota had the ability to hydrolyse the sugar moiety of the parent triterpenoid glycosides of C. asiatica to active triterpenic acid. This result agrees with a study by Songvut et al. who reported that active triterpenic acids are eliminated via the faecal route 32 .

Conclusion
Beagles had good tolerability to the standardised extract of C. asiatica, and showed similar pharmacokinetic profiles of bioactive triterpenoids to humans. Improved water solubility of Centell-S led to higher oral bioavailability of madecassoside and asiaticoside than ECa 233 in beagle dogs. The pharmacokinetic results could benefit future phytopharmaceutical product development from C. asiatica.