Maslinic acid, corosolic acid, oleanolic acid, and ursolic acid were obtained from Phytolab (Vestenbergsgreuth, Germany), lithocholic acid (LCA), chenodeoxycholic acid (CDCA), oleic acid, and palmitic acid from Sigma Aldrich (St. Louis, MO, United States), stearic acid and 4-methoxycinnamic acid from Fluka (Buchs, Switzerland), chlorogenic acid from Carl Roth (Karlsruhe, Germany), and linoleic acid from Acros (Geel, Belgium). Eugenol was isolated from clove and the identity and purity tested by ATR-IR spectroscopy. All reference and standard compounds were obtained in a purity of ≥90%. They were dissolved to 1.0 mg/mL in MeOH. The triterpene acids (TTAs) used for the quantification were mixed and diluted to give a standard solution containing 100 μg/mL of each corosolic acid, maslinic acid, oleanolic acid, and ursolic acid followed by a serial dilution down to 2 μg/mL.

Dulbecco’s modified Eagle’s medium (DMEM), containing 4.5 g/L glucose, L-glutamine, benzylpenicillin, and streptomycin were obtained from Lonza (Basel, Switzerland). Fetal bovine serum (FBS) and trypsin was supplied by Gibco via Invitrogen (Lofer, Austria). The TGR5 expression plasmid, GPBAR1, transcript variant 3, NM_170699.1 (OG-SC123315) was obtained from Origene via Biomedica (Vienna, Austria). The CRE-Luc plasmid, pGL4.29[luc2P/CRE/Hygro], was obtained from Promega (Mannheim, Germany) and the plasmid encoding enhanced green fluorescence protein (pEGFP-N1) was obtained from Clontech (Mountain View, CA, United States).

Dried and powdered herbal material was kindly provided by PADMA AG (Wetzikon, Switzerland). Information regarding the used organ, plant species, batch number, and acronym of the prepared extracts are given in Table 1. Voucher specimens are deposited at the Department of Pharmacognosy, University of Vienna, Austria.

Small scale extraction was adapted from Camp et al. (2012) as previously described (Kratz et al., 2016). In brief, 300 mg plant material was defatted with n-hexane before subsequent extraction with CH₂Cl₂ (7 mL) and MeOH (13 mL) for 15 min each by ultrasonication. The combined and dried extracts were resuspended in 4 mL MeOH and loaded onto a SPE polyamide gel cartridge for tannin depletion. The flow-through and eluate from washing with 2 × 4 mL MeOH was combined and dried under reduced pressure. Dried extracts were dissolved in dimethyl sulfoxide (DMSO) prior to bioactivity measurements. For bioassays, the final DMSO concentration was 0.1%.

To produce sufficient amounts of extract for further fractionation, the whole extraction and tannin-depletion procedure (see above) was scaled up by a factor of 10. The obtained extract (KgalE, 70 mg) was then fractionated by flash chromatography (PuriFlash 4250 system, Interchim, Montlucon Cedex, France) on a PuriFlash column (PF-RPC18HQ, 15 μm, 6g, Interchim) using water and acetonitrile as mobile phases A and B, respectively. The sample was applied as dry load onto the PuriFlash column and eluted at 25°C with a flow rate of 5 mL/min and a linear gradient of 5–95% B in 45 min. Eluting compounds were detected with both, a UV-detector (200–400 nm) and an evaporative light-scattering detector (ELSD). Six fractions (KgalE-1 to 6) were collected based on the ELSD signal followed by TLC monitoring (stationary phase: Merck silica gel 60 PF₂₅₄; mobile phase: toluol/ethyl acetate/formic acid/methanol/acetone, 4:2:1:0.5:0.25:0.25; detection with staining reagents vanillin/H₂SO₄ at vis, UV₂₅₄, UV₃₆₆), with the following elution volumes: KgalE-1, 90–138 mL; KgalE-2, 138–204 mL; KgalE-3, 204–237 mL; KgalE-4, 237–261 mL; KgalE-5, 267–327 mL; KgalE-6, 327–351 mL.

Essential oil of K. galanga was obtained by small-scale steam distillation of 2.0 g of the dried and pulverized drug with 20 mL of water for 1.5 h. The main constituents of the volatile oil were identified by GC-MS (mass spectrometry) on a QP2010 System (Shimadzu Austria, Korneuburg, Austria) equipped with a Zebron ZB-5 capillary column (60 m length, 0.25 μm film thickness; Phenomenex, Aschaffenburg, Germany). One microliter of the diluted essential oil (1:100 in CH₂Cl₂) was injected and separated with He (5.0) as carrier gas applying a temperature gradient from 50 to 270°C in 45 min. EI mass spectra were recorded in the range of m/z 40–500. The chemical composition in terms of normalized % peak area was determined on an AutoSystem XL equipped with a SE-54-CB capillary column (10 m length, 0.25 μm film thickness) and an FID detector. In this case, N₂ (5.0) was used as carrier gas and the temperature was increased from 50 to 155°C in 35 min.

Simulated gastric fluid (SGF) was prepared according to the European Pharmacopoeia (Ph. Eur. 8th): 2.0 g NaCl were dissolved in water. Eighty milliliters of 1 M HCl were added and the solution was diluted with water to achieve 1000 mL solution (pH = 1.2). Neutralization to a pH of ∼7 after incubation was achieved by adding diluted NaOH (Ph. Eur. 8; 8.5 g NaOH dissolved in water to achieve a 100 mL solution). Phosphate-buffered saline (PBS) was prepared with 137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, and 1.8 mM KH₂PO₄ to achieve a pH of 7.4.

Fecal slurry was prepared with fecal pellets obtained from C57BL/6 wild type mice (kindly provided by the Max F. Perutz Laboratories Animal Facility, Vienna). All procedures involving animal handling and sampling were performed according to protocols approved by the Austrian law (BMWF 68.205/0032-WF/II/3b/2014). Handling of the animals for sample collection was carried out by FELASA B degree holding personnel. Freshly evacuated fecal pellets were collected from mice that were housed together in the same cage, and immediately stored into an Oxoid^TM Anaerobic 3.5 L jar, where oxygen is removed by the presence of a AnaeroGen sachet (Thermo Scientific AN0035). The jar containing the fecal pellets was then introduced in a Vinyl Anaerobic Chamber (Coy Laboratory Products). Fecal pellets (∼12) were resuspended in 18 mL of PBS, and used instantly after preparation. For control samples (labeled with superscript 0), suspended fecal slurry was autoclaved before incubation.

Dried tannin-depleted extracts (4 mg) or CDCA (100 nmol) were incubated with 1 mL SGF for 90 min at 37°C followed by neutralization with diluted NaOH. Then, 1 mL of either autoclaved or non-autoclaved fecal slurry was added followed by incubation for 24 h at 37°C under anaerobic conditions. The incubation was terminated by adding ethanol to a final concentration of 48 vol% ethanol followed by storage at -20°C for about 18 h. Centrifugation and sterile filtration were applied before subjecting the samples to analytical and pharmacological investigations.

To identify the TGR5-activating compounds, samples were analyzed by high-performance liquid chromatography (HPLC) with charged aerosol detection (CAD) and HPLC-MS. These analyses were performed on an UltiMate 3000 RSLC-series system (Dionex/Thermo Fisher Scientific, Germering, Germany) coupled in parallel to a Corona ultra RS charged aerosol detector (CAD, Dionex/Thermo Fisher Scientific) and an HCT 3D quadrupole ion trap mass spectrometer equipped with an orthogonal ESI source (Bruker Daltonics, Bremen, Germany). Separation was carried out on an Acclaim 120 C18, 2.1 mm × 150 mm, 3 μm HPLC column (Dionex/Thermo Fisher Scientific) using 0.1% aqueous formic acid and acetonitrile as mobile phase A and B, respectively. The gradient was slightly modified for each extract (Syzygium aromaticum: 5–25% B in 10 min, 25–65% B in 8 min, 65–89% B in 12 min, and 89–95% B in 1 min; Pimenta dioica: 5–26% B in 14 min, 26–75% B in 14 min, and 75–95% B in 10 min; K. galanga: 5–95% B in 45 min). Each gradient was followed by a washing (10 min at 95% B) and re-equilibration step (10 min at 5% B). The flow rate was 0.5 mL/min and the column oven temperature was set to 25°C. After passing the diode-array detector (DAD), the eluate flow was split 4:1 between the CAD and the MS, respectively. The CAD nebulizer temperature was 35°C and the ESI ion source was operated as follows: capillary voltage: +3.5/-3.7 kV, nebulizer: 26 psi (N₂), dry gas flow: 9 L/min (N₂), and dry temperature: 340°C. Positive and negative ion mode multistage mass spectra up to MS³ were obtained in automated data-dependent acquisition (DDA) mode using helium as collision gas, an isolation window of Δm/z = 4, and a fragmentation amplitude of 1.0 V.

To confirm the tentative identifications achieved with the above system, high-resolution mass spectra were recorded on a maXis HD ESI-Qq-TOF mass spectrometer (Bruker Daltonics) that was also connected to an UltiMate 3000 RSLC-series system. The separation was performed with the above-described HPLC methods. The eluate flow was split approximately 1:8 and the following ESI ion source settings were applied: capillary voltage: ±4.5 kV, nebulizer: 0.8 bar (N₂), dry gas flow: 7.0 L/min (N₂), and dry temperature: 200°C. The sum formulas of the detected ions were determined using Bruker Compass DataAnalysis 4.2 based on the mass accuracy (Δm/z ≤ 10 ppm) and isotopic pattern matching (SmartFormula algorithm).

The main TTAs in SaroE and PdioE were quantified with the above-described HPLC-DAD-CAD system using external calibration. While the same stationary and mobile phase was applied as for the identification, other parameters had to be adapted to achieve better separation of the analytes: the flow rate was set to 0.3 mL/min, the column oven temperature to 10°C; the gradient consisted of a linear increase from 80 to 91% B in 22 min (followed by a column cleaning and a re-equilibration step). The extracts, quantification standards (100, 40, 20, 10, 5, and 2 μg/mL each of corosolic acid, maslinic acid, oleanolic acid, and ursolic acid), a quality control sample (12 μg/mL each of corosolic acid, maslinic acid, oleanolic acid, and ursolic acid) and a blank sample (pure MeOH) were analyzed in quadruplicates. For each analyte, the peak area was obtained from the CAD chromatograms and a linear calibration function was determined from the quantification standards, which was then used to calculate the concentration of the four TTAs in the active extracts.

For the TGR5 luciferase reporter gene assay HEK-293 cells (ATCC, Manassas, VA, United States) were maintained in DMEM containing 4.5 g/L glucose supplemented with 2 mM glutamine, 10% FBS, 100 U/mL benzylpenicillin and 100 μg/mL streptomycin. Cells were grown in 15 cm dishes at a density of 6 × 10⁶ cells per dish for 19 h and then transfected via the calcium phosphate precipitation method with 5 μg TGR5 expression plasmid, 5 μg CRE-Luc plasmid, and 3 μg of a plasmid encoding EGFP as internal control. After 6 h the cells were reseeded to 96-well plates (5 × 10⁴ cells/well) in DMEM containing 4.5 g/L glucose supplemented with 2 mM glutamine, 100 U/mL benzylpenicillin, 100 μg/mL streptomycin, and 5% charcoal-stripped FBS. Cells were treated with indicated concentrations of test extracts, positive control (10 μM LCA) or solvent vehicle (0.1% DMSO) and incubated for 18 h. Cells were then lysed and the luminescence of the firefly luciferase and the fluorescence of EGFP were quantified with a GeniosPro plate reader (Tecan, Grödig, Austria). The luciferase-derived luminescence was normalized to the EGFP-derived fluorescence from each well to account for differences in transfection efficiency or cell number.

Statistical analyses were performed with the GraphPad prism software version 4.03. Non-linear regression (sigmoidal dose response) was used to calculate the EC₅₀ values and maximal fold activation. The data are presented as the mean ± SD. Statistical evaluation was performed by one-way analysis of variance (ANOVA) and Student’s t-test. Differences between groups with a p-value <0.05 were considered statistically significant.

Nineteen herbal drugs of a Tibetan formula used against different symptoms associated with the metabolic syndrome were selected and tested for their potential to activate TGR5. The formula contains the pulverized material of the following plant materials: costus root, Iceland moss, neem fruit, cardamom fruit, myrobalan fruit, red sandalwood, allspice, bael tree fruit, columbine, liquorice root, ribwort plantain, knotgrass, potentilla golden herb, clove, aromatic ginger, heart-leaved sida, valerian root, lettuce leaf, and marigold.

In accordance to the oral application of the pulverized material, extracts were prepared as previously described (Kratz et al., 2016) by a protocol adapted from Camp et al. (2012). Herein, defatted plant material is extracted successively with dichloromethane and methanol. The two resulting extracts were combined in order to cover drug-like properties in a wide range of polarity. Further, tannin depletion via polyamide gel was carried out in order to remove common assay-interfering compounds.

The obtained extracts were screened in HEK-293 cells transfected with the TGR5 receptor and a luciferase gene containing the respective response element in its promoter (CRE-Luc). Three plant extracts were able to significantly activate TGR5 at concentrations of 100 and 30 μg/mL, namely those of the dried flower buds of S. aromaticum (SaroE), commonly known as cloves, the unripe fruits of P. dioica (PdioE), commonly known as allspice, and the rhizomes of K. galanga (KgalE), known as aromatic ginger. These three medicinal plants are used as herbal remedies for the treatment of digestive problems, diabetes, as antioxidants and analgesics in traditional medicines, especially of Middle Eastern and Asian countries as well as the Caribbean (Achuthan and Padikkala, 1997; Zhang and Lokeshwar, 2012; Kumar et al., 2013; Cortes-Rojas et al., 2014; Opara and Chohan, 2014; Sanae et al., 2014; Tu et al., 2014). Interestingly, all three herbal drugs are also widely used as spices.

Concentration–response studies were performed with the three TGR5-activating extracts PdioE, SaroE, and KgalE to determine their EC₅₀ and E_max values. As reference compound, the well-established TGR5 agonist LCA (Figure 1) was used at a concentration of 10 μM, which was set as 100% TGR5 activation. The assay was validated with CDCA (Figure 1), which showed an EC₅₀ of 33.2 ± 4.5 μM with 34.6 ± 8.6% maximum relative TGR5 activation in our assay. All three plant extracts were able to increase TGR5 activity in a concentration-dependent manner; KgalE showed the lowest EC₅₀ with 12.9 ± 24.1 μg/mL and an E_max of 33.2 ± 4.5% compared to LCA (Figure 2A and Table 2). PdioE elicited an E_max of 29.5 ± 8.3% compared to LCA and the highest EC₅₀ of all three plant extracts with 60.2 ± 13.7 μg/mL (Figure 2B). Intriguingly, SaroE revealed the highest efficiency with an E_max of 112.8 ± 12.1% compared to LCA, overreaching by far the efficiency of our positive control CDCA. The EC₅₀ of SaroE was determined to be 19.8 ± 16.6 μg/mL (Figure 2C and Table 2).

To simulate the passage of orally ingested compounds through the gastrointestinal system, the positive control CDCA and the prepared extracts were first suspended in SGF and incubated for 90 min to account for the physiological retention time in the stomach. After neutralization of this mixture with sodium hydroxide, either (i) deactivated fecal microbiota (i.e., non-transformed, labeled with superscript 0), or (ii) active fecal microbiota (i.e., transformed, superscript FS) from mice were added to facilitate the biotransformation of the constituents.

Incubation of the positive control CDCA with mouse fecal microbiota led to a strong biotransformation, thus proving the suitability of the experimental workflow. HPLC-DAD-CAD and LC-MS analyses showed that about half of the CDCA was consumed after 24 h while at least four different metabolites were produced (Supplementary Figure S1), presumably by 7-dehydroxylation, 3- or 7-dehydrogenation, and 3- or 7-epimerization as previously described (Gerard, 2013). After 24 h incubation, the most abundant metabolite was identified as LCA by comparison with a reference standard. LCA is the most potent endogenous agonist for TGR5 with CDCA being significantly less active (Kawamata et al., 2003). To evaluate our assay system, CDCA was incubated with active and deactivated fecal slurry for 24 h (CDCA^FS and CDCA⁰, respectively). In accordance with the observed transformation, CDCA^FS elicited a much higher response than CDCA⁰ in the TGR5 luciferase reporter gene assay (Figure 3).

Instrumental analysis of the transformed (KgalE^FS, SaroE^FS, and PdioE^FS) and non-transformed extracts (KgalE⁰, SaroE⁰, and PdioE⁰) revealed that only PdioE and KgalE showed differences in the metabolite profile after transformation. In both cases, an increase in the abundance of compounds that were tentatively identified as free fatty acids and hydroxylated or otherwise oxidized fatty acids was detected, which can be explained by the action of lipolytic enzymes on remaining polar lipids in the extracts. For PdioE, it was additionally observed that a compound tentatively identified as ericifolin (Shamaladevi et al., 2013) by high-resolution mass spectrometry (HRMS) and fragment ion spectra was completely degraded after 24 h of incubation with mouse fecal microbiota. The comparison of KgalE⁰ and KgalE^FS showed a marked increase in the abundance of 4-methoxycinnamic acid, presumably by hydrolysis of larger phenylpropanoid structures, and higher concentrations of some not unambiguously identified small polar metabolites.

To elucidate a possible impact caused by incubation with fecal slurry, KgalE and PdioE were retested for their TGR5-activating potential, once incubated with active fecal slurry and once prepared with deactivated fecal slurry. However, other than in the case of CDCA, no significant difference in activity using the TGR5 reporter gene assay could be observed between KgalE⁰ and KgalE^FS, and PdioE⁰ and PdioE^FS, suggesting that the analytically observed biotransformation has no influence on the TGR5 activity. Therefore, we further focused on the identification of plant metabolites not targeted and modified by the fecal microbiota.

To identify the compounds responsible for TGR5 activity, all three active extracts were analyzed by HPLC and LC-MS. The accurate masses of the individual compounds were determined by HPLC-HRMS on an ESI-QqTOF system. Sum formulas were predicted based on mass accuracy (Δm/z ≤ 10 ppm) and isotopic pattern matching. Finally, structural information was obtained by HPLC-multistage MS (MSⁿ) on an ESI-ion trap system in DDA mode. Since all three plant species represent prominent spices and herbal drugs, they are well studied in terms of their chemical constituents. Thus, the main compounds could be tentatively identified as known metabolites of the investigated plant parts (see Tables 3–5 and Supplementary Figures S2–S4 in the Supporting Information). The assignment of several of these structures was confirmed by comparison of the retention time and spectral data with reference standards.

All three extracts contained common fatty acids (e.g., palmitic, stearic, oleic, and linolenic acids) as well as cyclic oligomers of 𝜀-caprolactam—the latter being non-TGR5-influencing artifacts from the extract preparation.

The HPLC-DAD-CAD and LC-MS analyses of KgalE indicated the presence of essential oil ingredients (Supplementary Figure S2) with the main component ethyl p-methoxycinnamate and the less abundant ethyl cinnamate (Wong et al., 1992; Raina and Abraham, 2016). Related derivatives, such as the corresponding propyl-esters and p-methoxycinnamic acid as well as fatty acid derivatives and minor unknown compounds were additionally detected (Table 3). Since none of these constituents have been described as TGR5 activators or are structurally similar to known TGR5 binders, KgalE was fractionated by RP-18 column chromatography to further assign the observed activity to a specific compound class or constituent. GC-MS analysis of the essential oil, which was obtained by steam distillation, revealed 92.2% p-methoxycinnamate and 1.3% ethyl cinnamate. The obtained six fractions as well as the steam distilled essential oil of K. galanga rhizomes were retested in the TGR5 luciferase reporter gene assay. Interestingly, the TGR5 activity was found to be distributed mainly between the essential oil and fractions 4 and 5 of KgalE (Figure 4). Thereby, the previously observed TGR5-activating potential of the extract could not be acuminated in one of these fractions. Moreover, the bioactive fractions negatively influenced cell viability, therefore we refrained from further separation and isolation of their constituents.

The extracts of S. aromaticum (SaroE) and P. dioica (PdioE) shared some similarities in their phytochemical profile (Tables 3, 4). Both samples contained some volatile and non-volatile phenolic compounds, such as isobiflorin, biflorin, eugenol, and acetyleugenol in SaroE, several dihydroxyeugenol derivatives and ericifolin in PdioE, and gallic acid in both. More importantly, TTAs were identified as major constituents of the extracts. The common monohydroxy TTAs betulinic, oleanolic, and ursolic acid were found to activate TGR5 with EC₅₀ values in the range of 1–3 μM in previous studies (Sato et al., 2007; Genet et al., 2010). Indeed, comparison with reference compounds revealed oleanolic acid as the base peak in the CAD chromatogram of SaroE and both, oleanolic and ursolic acids as major constituents of PdioE. In addition, maslinic acid (in SaroE and PdioE) and corosolic acid (in PdioE), which are the 2α-hydroxy-derivatives of oleanolic acid and ursolic acid, respectively, were identified as prominent ingredients of these two extracts (Figure 1). Whether and how the additional hydroxy-group in 2α-position affects the activity on TGR5 has not been assessed so far.

To verify the assumption that the measured TGR5 activity of SaroE and PdioE can be related to the identified pentacyclic TTAs, we developed a quantitative HPLC-CAD method with external standardization to determine their concentration in the tested extracts. Our method showed good linearity in the range of 2–100 μg/mL for each of the four compounds. SaroE was found to contain 6.56 ± 0.20% oleanolic acid and 2.70 ± 0.04% maslinic acid. In correlation with the lower activity, the concentration of the TTAs was less in PdioE, namely 0.290 ± 0.019% maslinic acid, 0.256 ± 0.012% ursolic acid, 0.198 ± 0.010% corosolic acid, and 0.148 ± 0.007% oleanolic acid. Based on these quantification results, reconstituted mixtures of the identified TTAs were prepared in the same concentration and composition as in the final test concentration (100 μg/mL) of SaroE, i.e., 14.4 μM oleanolic acid and 5.72 μM maslinic acid, and PdioE, i.e., 0.32 μM oleanolic acid, 0.61 μM maslinic acid, 0.42 μM corosolic acid, and 0.56 μM ursolic acid. These reconstituted mixtures, together with the individual pure compounds, were tested in the TGR5 luciferase reporter gene assay. In both cases, the reconstituted mixtures of TTAs were equally potent compared to the respective extracts (Figure 5). The EC₅₀ and E_max values of the extracts, the reconstituted mixtures and the single compounds are listed in Table 6.