Using UHPLC-MS Profiling for the Discovery of New Dihydro-β-Agarofurans from Australian Celastraceae Plant Extracts

An analytical method using UHPLC-MS was developed and applied to 16 crude CH2Cl2 extracts from Australian Celastraceae plants; the endemic plant materials were accessed from Griffith University’s NatureBank resource and included bark, fruit, leaf, root, twig and mixed samples, all of which were collected from Queensland, Australia. The generated UHPLC-MS data were analysed and dereplicated using the scientific databases Dictionary of Natural Products and SciFinder Scholar in order to potentially identify new dihydro-β-agarofurans from local Celastraceae plants. These investigations led to the large-scale extraction and isolation work on a prioritised fruit sample that belonged to the rainforest plant Denhamia celastroides. Chemical investigations resulted in the purification of four new natural products, denhaminols O–R (1–4), along with the related and known compound, denhaminol G (5). The structures of all the new compounds were determined via detailed analysis of NMR and MS data.


Introduction
Dihydro-β-agarofurans are a class of structurally unique polyoxygenated tricyclic sesquiterpenoids, which incorporate a trans-decalin and a tetrahydrofuran and are commonly found in the Celastraceae plant family [1]. This class of natural products has gained much attention due to their various and promising bioactivities, such as multidrug resistance reversal [2], antitumor-promotion [3], acetylcholinesterase inhibition [4], antifungal [5], α-glucosidase inhibition [6], antiplasmodial [7] and leucine uptake inhibition [8]. The biological effects of dihydro-β-agarofurans are related not only to several stereocentres but also to acyl groups attached to the tricyclic core scaffold. One of our group's current research interests is the isolation of new natural products (i.e., dihydro-β-agarofurans) from Australian Celastraceae native plants [8][9][10].
Within the workflow of natural product research, the time-consuming re-isolation of previously identified compounds presents a major obstacle and can significantly delay discovery efforts [11]. However, rapid and detailed dereplication methodologies can solve this problem. In the context of our continuing interest in the identification of new natural products from Celastraceae plants [8][9][10],

Results and Discussion
The CH 2 Cl 2 extracts of samples from 16 Australian Celastraceae plants were prepared using a small amount of the air-dried and ground samples. All extracts were subjected to UHPLC-MS profiling ( Figure 1 and Figure S1, Supplementary data) and the molecular masses of the major UV-active compounds in each of the extracts were determined from either the negative or positive total ion chromatogram (TIC). These data were then analysed using SciFinder Scholar and the Dictionary of Natural Products (DNP). Examination of respective UHPLC chromatograms and scientific databases provided a preliminary overview of the constituents of the extracts. The number of hits (from the database search) of the molecular weights generated from the MS data was used as a filter. Of the 16 plant extracts subjected to UHPLC-MS, three plant samples [Denhamia celastroides (F. Muell.) Jessup -fruits, Hysophila halleyana (F. Muell.) -bark and Perrottetia arborescens (F. Muell.) Loes -root] were prioritised since distinct molecular ions were detected (less than five hits reported in the databases; Table S2, Supplementary data). Since the UHPLC traces of the fruit extract of D. celastroides showed the best separation amongst all analysed plant extracts (Figure 2), this sample was chosen for additional chemical investigations.
The scientific databases search (Table 1 and Table S2, Supplementary data) indicated the possibility of new dihydro-β-agarofurans present in the fruits of a Denhamia celastroides CH 2 Cl 2 extract since it contained several distinct molecular ions in the (+)-ESI mode (m/z 615 [M + H] + ; 631 [M + H] + ; and 657 [M + H] + ), which were only found to match a few dihydro-β-agarofurans (less than five hits) reported in SciFinder Scholar and DNP search (keywords: "Celastraceae" and "agarofuran"). To confirm the presence of the new compounds and to unambiguously identify the structures, we conducted a large-scale extraction of the plant material and MS-directed purification to isolate the targeted compounds with the distinct molecular weights before NMR experiments were conducted.  The scientific databases search (Table 1 and Table S2, Supplementary data) indicated the possibility of new dihydro-β-agarofurans present in the fruits of a Denhamia celastroides CH2Cl2  The scientific databases search (Table 1 and Table S2, Supplementary data) indicated the possibility of new dihydro-β-agarofurans present in the fruits of a Denhamia celastroides CH2Cl2  The fruits of D. celastroides (10 g) were sequentially extracted with CH 2 Cl 2 . Subsequent purifications using silica gel column chromatography and RP-HPLC afforded five dihydro-βagarofurans ( Figure 3). The targeted compounds with molecular ions m/z [M + H] + of 615, 631 and 657 were confirmed to be new natural products, which were given the trivial name denhaminols O-Q (1-3), respectively. During the isolation of the three targeted compounds, another new dihydro-β-agarofuran (denhaminol R, 4) and a known congener, denhaminol G (5) [20] were also obtained. The complete structure elucidation of the new compounds is detailed below.
The positions of the ester groups were determined following HMBC data analysis. HMBC cross-peaks from two olefinic protons (δ H 6.40 and 7.65, d, J = 16.0) and H-9 (δ H 4.88) to an ester carbonyl carbon at δ H 165.9 located a trans-cinnamate group at C-9. The 1 H-NMR resonances at δ H 1.78 (3H, m), 1.79 (3H, m) and 6.81 (1H, m) were the characteristic of tigloyl moiety [20]. The tigloyl group was located at C-12 based on HMBC correlations from a pair of diastereotopic methylene protons (δ H 4.52 and 4.69), a methyl at δ H 1.79 and an olefinic proton at δ H 6.81 to a carbonyl resonance at δ C 167.6. The HMBC spectrum of 1 also exhibited correlations from a set of methylene protons at δ H 4.31 and 4.45 to two carbonyl carbons resonating at δ C 167.1 and 170.2. These data along with further HMBC correlation from H-1 (δ H 4.88) to the carbonyl carbon at 167.1 suggested the location of an acetoxyacetate functional group at C-1. Finally, two hydroxy moieties were positioned at C-4 and C-6 by considering the molecular formula of 1 and the deshielded NMR resonances of C-4 (δ C 73.3) and CH-6 (δ H 4.44 and δ C 79.4). The relative configuration of denhaminol O (1) was established by ROESY ( Figure 4) and 1 H-1 H coupling constant data analysis. The large coupling constant of H-1 (J 1,2 = 12.1 Hz) indicated the β-orientation of H-1. Similarly, the α-orientation of H-9 was assigned based on the coupling constant (J 8,9 = 7.4 Hz). ROESY cross-peaks between H 3 -14 and H-6 and between H-6 and H-9 as well as between H-9 and H 3 -15 suggested that these protons were cofacial. It is worth mentioning that ROESY correlations were also observed between H 2 -12 and H-8β as well as between H 3 -13 and H-7. Consequently, the structure of 1 was established as 1α-acetoxyacetate-8β-cinnamoyloxy-4β,6β-dihydroxy-12-tigloyloxydihydro-β-agarofuran. Denhaminol O (1) was isolated as a colourless gum with a molecular formula of C33H42O11 as assigned by HRESIMS data (m/z 637.2595). The 1 H-NMR spectrum (  (Table 3) and HSQC spectra of 1 suggested a total of 33 carbons, including 6 methyls, 5 methylenes, 12 methines and 10 non-protonated carbons. The 13 C resonances at δC 165.9, 167.1, 167.6 and 170.2 suggested the presence of four ester groups in 1. These data indicated that compound 1 was a dihydro-β-agarofuran bearing four ester groups [8][9][10], which was confirmed by COSY and HMBC experiments ( Figure 4). Denhaminol P (2, colourless gum) had a molecular formula of C 35 H 44 O 12 as assigned by (+)-HRESIMS. Comparison of 1 H and 13 C-NMR spectra of 2 and 1 showed a high degree of similarity between the two compounds. However, the 1 H resonance of H-6 was shifted downfield from δ H 4.44 in 1 to δ H 5.48 in 2, which suggested the attachment of an ester group at C-6 in 2. HMBC correlations from a methyl singlet signal at δ H 2.15 and H-6 (δ H 5.48) to a carbonyl carbon at δ C 170.6 identified the presence of an acetate group at C-6 in 2. The 2D NMR data ( Figure S28, Supplementary data) further confirmed the attachment of the remaining ester moieties in 2. Comparison of ROESY data of 1 and 2 revealed the same relative configurations for these compounds. Thus, the structure of 2 was elucidated as 6β-acetoxy-1α-acetoxyacetate-8β-cinnamoyloxy-4β-hydroxy-12-tigloyloxydihydro-β-agarofuran.      Figure S29, Supplementary data) from H-8 to C-6, C-9 and C-11 further confirmed the location of OH-8. The β-orientation of the hydroxy moiety at C-8 was assigned by ROESY correlations (Figure S29, Supplementary data) between H-8 and H 3 -15. Therefore, the structure of denhaminol Q (3) was determined as 1α-acetoxyacetate-8β-cinnamoyloxy-4β,6β,8β-trihydroxy-12-tigloyloxydihydro-β-agarofuran. During the isolation of the major compounds 1-3, denhaminol R (4) was obtained. The molecular formula of compound 4 was C 35 H 40 O 12 as indicated by (+)-HRESIMS data. Analysis of NMR and MS data showed that the structure of 4 was similar to that of 3. However, the tigloyl group at C-12 in 3 was replaced by a benzoate in 4. The presence of the benzoate group was confirmed by the characteristic 1 H NMR resonances at δ H 7.45 (2H, m), 7.57 (1H, m) and 8.03 (2H, m). The position of the benzoate at C-12 was assigned by HMBC correlations ( Figure  S30, Supplementary data) from H 2 -12 (δ H 4.88 and 5.02) and the proton at δ H 8.03 to an ester carbonyl carbon at δ C 166.4. The relative configurations of 4 were ascertained to be the same as those of 3 by ROESY experiment. Accordingly, the structure compound 4 was assigned as 1α-acetoxyacetate-12-benzoylyloxy-8β-cinnamoyloxy-4β,6β,8β-trihydroxydihydro-β-agarofuran.
Previously, we reported the isolation of dihydro-β-agarofuran sesquiterpenoids from two Australian plants belonging to the Denhamia genus, namely D. celastroides and D. pittosporoides. The chemical investigation of the leaves of D. celastroides afforded eight dihydro-β-agarofurans (denhaminol A-H) [20], while two new dihydro-β-agarofurans (denhaminols I and J) were obtained from the leaves of D. pittosporoides [9].

General Experimental Procedures
Values of specific rotations were determined with a JASCO P-1020 polarimeter and UV spectra were recorded using a JASCO V-650 UV/vis spectrophotometer (ATA Scientific, Taren Point, NSW, Australia). ECD spectra were obtained on a JASCO J-715 spectropolarimeter (Tokyo, Japan) and processed using the software SDAR v3.2 [21]. IR data were acquired using an attached universal attenuated total reflectance (UATR) two module on a PerkinElmer spectrophotometer (Waltham, MA, USA). NMR spectra were acquired from a Bruker AVANCE HDX 800 MHz NMR spectrometer (Zurich, Switzerland) equipped with a TCI cryoprobe at 25 • C. The 1 H and 13 C chemical shifts were referenced to the residual solvent signal of CDCl 3 at δ H 7.26 and δ C 77.16 ppm, respectively. HRESIMS data were acquired on a Bruker maXis II ETD ESI-qTOF (Bremen, Germany) and the mass spectrum was calibrated externally with 0.1 mg/mL of sodium trifluoroacetate. A Fritsch Universal Cutting Mill Pulverisette 19 (Idar-Oberstein, Germany) was used to grind the air-dried plant material and an Edwards Instrument Company Bio-line orbital shaker (Narangba, Australia) was used for plant extraction. Phenomenex Strata solid phase extraction (SPE) cartridges (3 cc, polypropylene, single fritted, catalogue# AH0-7806) (Torrance, California, USA) were used for the small-scale plant extractions. The UHPLC-MS was performed on an Ultimate 3000 RS UHPLC (Waltham, MA, USA) coupled to a Thermo Fisher Scientific MSQ Plus single quadruple ESI mass spectrometer (Waltham, MA, USA) using an analytical Waters ACQUITY UPLC CSH C 18 column (2.1 × 50 mm, 1.7 µm, 130 Å) (Milford, MA, USA). A Thermo Fisher Scientific Dionex Ultimate 3000 UHPLC was used for semi-preparative HPLC separations. A Phenomenex Luna C 18 (250 × 10 mm, 5 µm, 90-110 Å) column (Torrance, CA, USA) was used for semi-preparative HPLC separations. Alltech C 18 -bonded Si (35-75 µm, 150 Å) (Sydney, NSW, Australia) was used for pre-adsorption work, and the resulting material was packed into an Alltech stainless steel guard cartridge (10 × 30 mm) prior to semi-preparative HPLC separations. Merck Si gel (0.040-0.063 mm, 230-400 mesh) (Darmstadt, Germany) was used for Si gel column chromatography. All solvents (CH 2 Cl 2 and CH 3 CN) used for chromatography, specific rotation, ECD, UV and MS were RCI Labscan HPLC grade (Samutsakhorn, Thailand). H 2 O was Sartorius arium pro VF (Göttingen, Germany) filtered. All compounds were analysed for purity by 1 H NMR spectroscopy and shown to be >95%, unless otherwise stated. NMR spectra were processed using MestReNova version 11.0 (Santiago de Compostela, Spain).

Plant Materials
The 16 Celastraceae plant samples were obtained from the NatureBank biota library housed at the Griffith Institute for Drug Discovery, Griffith University, Australia [19]. All samples were collected in Queensland and taxonomically identified by the Queensland Herbarium. Voucher specimens have been deposited at the Queensland Herbarium, Australia. All plant specimens were air-dried, ground and stored at room temperature prior to extraction. The fruits of Denhamia celastroides (F. Muell.) Jessup (Voucher specimen code: AQ605014) that were used in the large-scale extraction and isolation investigations were collected on 26 November 1997 in Mt Windsor Tableland rainforest, Queensland, Australia. Details of the collection date, location and voucher specimen codes for the other Celastraceae plants used in these studies are provided in Table S31 of the supplementary data.

Preparation of Crude Plant Extracts for UHPLC-MS Analyses
Each of the air-dried and ground Celastraceae plant materials (300 mg) was packed into an SPE cartridge and extracted under gravity with 8 mL of CH 2 Cl 2 . The CH 2 Cl 2 extracts were dried, weighed and resuspended in CH 3 CN to generate a stock solution, which had a concentration of 1 mg/mL (minimum stock solution volume = 0.5 mL).

Large-scale Extraction and Isolation of the Fruits of D. celastroides
The air-dried and ground fruits of D. celastroides (10 g) were extracted with CH 2 Cl 2 (2 × 500 mL for 16 h each) to afford 245.8 mg of a crude extract. The extract was subjected to chromatography using a Si-gel column (3 × 8 cm) and a step-wise gradient system of n-hexane/EtOAc (100% n-hexane to 100% EtOAc, 10% increment, 100 mL each) to afford 11 fractions (fractions 1-11). All 11 fractions were analysed by UPLC-MS and fractions 7-9 were chosen for further purification based on UPLC-MS data analysis. Fraction 7 (22.4 mg) was pre-adsorbed to C 18 bonded Si-gel (~1 g), packed into a guard cartridge and attached to a semi-preparative C 18

Conclusions
A UHPLC-MS method and dereplication process was developed for the rapid identification of new dihydro-β-agarofurans from Celastraceae plants. This study further exemplifies how UHPLC-MS data and database mining can be used to extract molecular features related to characteristic secondary metabolites of a plant family and thus expedite new discoveries in natural products research. This approach was successfully applied to 16 Australian Celastraceae plant extracts. Consequently, four previously undescribed dihydro-β-agarofurans (denhaminols O-R, 1-4) along with a known compound denhaminol G (5) were successfully isolated and characterised from one prioritised plant sample, namely the fruits of D. celastroides. Other Celastraceae samples that were prioritised for large-scale extraction and isolation studies will be investigated in the future. The pure compounds reported in this paper will be added to the Davis Open-Access Compound Library which is housed at Compounds Australia, Griffith University [19,[22][23][24][25] and will be tested in various bioassays in the future. It is worth mentioning and well-known that the chemical profile of a plant can vary due to different geographical locations and collection seasons; however, even though this research utilised only Queensland Celastraceae plants, the UHPLC-MS method developed here should be applicable to the chemical profiling of any plant sample. Furthermore, this UHPLC-MS methodology should be adaptable to the chemical investigation of any biota material, including not only plants but also microbes, marine invertebrates and fungi.