Newly Generated Atractylon Derivatives in Processed Rhizomes of Atractylodes macrocephala Koidz

Thermally processed rhizomes of Atractylodes macrocephala (RAM) have a long history of use in traditional Chinese medicine (TCM) for treating various disorders, and have been an integral part of various traditional drugs and healthcare products. In TCM, herbal medicines are, in most cases, uniquely processed. Although it is thought that processing can alter the properties of herbal medicines so as to achieve desired functions, increase potency, and/or reduce side effects, the underlying chemical changes remain unclear for most thermally processed Chinese herbal medicines. In an attempt to shed some light on the scientific rationale behind the processes involved in traditional medicine, the RAM processed by stir-frying with wheat bran was investigated for the change of chemical composition. As a result, for the first time, five new chemical entities, along with ten known compounds, were isolated. Their chemical structures were determined by spectroscopic and spectrometric analyses. The possible synthetic pathway for the generation of such thermally-induced chemical entities was also proposed. Furthermore, biological activity evaluation showed that none of the compounds possessed cytotoxic effects against the tested mammalian cancer and noncancer cell lines. In addition, all compounds were ineffective at inhibiting the growth of the pathogenic microorganisms.


Introduction
The rhizome of Atractylodes macrocephala Koidz (RAM)-called Baizhu in traditional Chinese medicine (TCM)-is a well-known herbal medicine. It was documented in the earliest existing book on TCM-"Shen Nong's Materia Medica" written during the Han Dynasty (A.D. . RAM has been traditionally used for the treatment of various disorders, such as loss of appetite, diarrhea, limb weakness, gastrointestinal dysfunction, immune dysfunction, diabetes, and some chronic inflammatory diseases [1]. It was reported that RAM is used in more than 835 TCM preparations, as well as an integral part of more than 4340 classic prescriptions for treating chronic diseases [1]. Indeed, RAM is considered a functional food, a tonic, and a constituent of various health products purported for promoting digestion, alleviating fatigue, improving sleeping, enhancing immunity, and treating alimentary anemia [2]. RAM is traditionally used in its processed form which is commonly achieved by stir-frying

Structure Elucidation
A previous study using a NMR-based metabolomics method to investigate the effects of processing on the Chinese herbal medicine Flos Lonicerae revealed that the NMR approach can provide not only a holistic view on the change of chemical composition during processing, but also the information on identity of individual components [5]. In the present study, with the aid of information obtained from the NMR-based metabolomic analyses of RAM before and after processing, we could narrow the scope of isolation targets to the hexane partition fraction and its sub-fractions based on new signals of interest observed in their NMR spectra.
Compound 2 was obtained as a white amorphous powder. As determined from the [M + H] + peak at m/z = 459.3267 in HR-ACPI-MS, 2 has a molecular formula of C 32 H 42 O 2 with 12 degrees of unsaturation (DOU). Compound 2 has the same number of DOU as compound 1, but has an additional CH 2 in molecular formula. The 1 H NMR and HSQC spectra of 2 exhibit similar signal patterns as those of 1, respectively, i.e., the signals of methyl groups at δ H 1.82/1.76 (s) and 0.75/0.77 (s), exocyclic methylidene at δ H 4.68 and 4.84, aliphatic methine at δ H 2.10 (m), six methylenes in the range of δ H 1.40-2.50. A significant difference lies in that one methylene at δ H 3.82 (s) was observed for 1, but one aliphatic methine at δ H 4.17 (q) and one methyl at δ H 1.57 (d) were observed for 2 ( Table 1). The 13 C-NMR spectrum of 2 also displays a similar signal pattern as that of 1. A significant difference between 2 and 1 is the presence of one methine carbon at δ C 30.1 and one methyl carbon at δ C 18.1 in 2, corresponding, respectively, to the proton signals at δ H 4.17 and δ H 1.57. The analyses of the COSY and HSQC spectra confirmed the presence of an ethylidene unit in 2, instead of the presence of a methylene in 1. Comparing 2 to 1, another difference in their 13 C spectra is that slight splitting of carbon signals for the pair groups of C-1/1 , C-5/5 , C-9/9 , C-11/11 , C-13/13 , and C-14/14 were observed for 2 but not for 1 ( Table 1). Analyses of the HMBC and COSY spectra of 2 revealed that these pair of signals were attributed to the two atractylon moieties presented in 2. Furthermore, the trans-fused A/B ring junction in the two atractylon moieties was confirmed by the NOE observations from the NOESY spectrum. The HMBC correlations of the methyl protons (Me-17) in the ethylidene unit at δ H 1.75 with the carbons C-16 at δ C 30.1, C-11 (11 ) at δ C 112.9/113.0, and C-12 (12 ) at δ C 149.1, and the methine proton (H-16) in the ethylidene unit at δ H 4.17 with the carbons C-17 at δ C 18.1, C-11 (11 ) at δ C 112.9/113.0, and C-12 (12 ) at δ C 149.1 indicated that the two atractylon moieties were linked together through the methine in the ethylidene unit connecting at C-11 and C-11 . On the basis of the aforementioned evidence, the structure of 2 was established to be 2,2 -(ethane-1,1-diyl)-bis(3,8α-dimethyl-5-methylene-4,4α,5,6,7,8,8α,9-octahydronaphtho-[2,3-β]-furan), trivially named ethylidene-biatractylon.  (14′) in the furanoeudesmane skeleton was the same as atractylon. Observations of HMBC correlations of the methylene signal at δH 3.82 (H-16) to C-11 (11′) at δC 114.3 and C-12 (12′) at δC 145.0 in the HMBC spectrum indicated that the two symmetrical moieties were linked together through the methylene group attaching to C-11 and C-11′. Accordingly, 1 was identified to be bis(3,8αdimethyl-5-methylene-4,4α,5,6,7,8,8α,9-octahydronaphtho-[2,3-β]-furan-2-yl)methane, trivially named methylene-biatractylon.  Compound 3 was isolated as a white amorphous powder. Its molecular formula was determined as C 35 H 42 O 3 based on the HR-ACPI-MS data for the [M + H] + peak at m/z = 511.3221, accounting for 15 degrees of unsaturation. Both the 1 H and 13 C signals of 3 in the upfield NMR chemical shift range showed close similarities to those of 1 and 2, however, differences were observed in the downfield range. Four additional carbon signals at δ C 153.1, 141.4, 110.2, and 107.2 in the 13 C spectrum, and three additional proton signals at δ H 7.35 (d), 6.30 (dd), and 6.07 (d) in the 1 H spectrum were observed for 3, as compared to 1 or 2. These signals were assigned to a furan ring (3 degrees of unsaturation) by the analyses of COSY and HMBC spectra. Apart from these differences, the other signals displayed almost the same features as in 1 or 2. The analyses of the 2D NMR spectra confirmed that 3 possessed a similar biatractylon skeleton in its structure as was found in 1 and 2. Clear HMBC correlations of the methine proton (CH-16, with proton and carbon signals at δ H 5.47 (s) and δ C 36.1, respectively) to C-17 at δ C 153.1, C-18 at δ C 107.2, C-12 at δ C 144.7, and C-11 δ C 115.2 were observed, indicating that the furan ring and the two atractylon moieties were linked together through this methine. Hence, the structure of 3 was determined to be 2,2 -(furan-2-yl-methylene)-bis(3,8α-dimethyl-5-methylene-4,4α,5,6,7,8,8α,9-octahydronaphtho-[2,3-β] -furan), trivially named furan-2-methanetriyl-biatractylon.
Compound 4 was obtained as a white amorphous powder. The [M + H] + ion at m/z = 541.3299 in the HR-ACPI-MS spectrum revealed its molecular formula as C 36 H 44 O 4 , with the same 15 degrees of unsaturation but with one more carbon, two more protons, and one more oxygen when compared to 3. The 1 H-NMR spectrum of 4 showed almost the same features as that of 3, except for having one more singlet signal at δ H 4.55. The 13 C-NMR spectrum of 4 also displayed close similarity to that of 3, except for the addition of an oxygenated methylene signal at δ C 57.6 (corresponding to the above signal at δ H 4.55 in the HSQC spectrum) and the significant downfield shift of the olefinic carbon signal from δ C 114.4 to 153.0. In the HMBC spectrum of 4, the protons of this oxygenated methylene demonstrated clear correlations to the downfield-shifted carbons C-20 at δ c 153.0 and C-19 at δ c 108.7, indicating its connection at C-20. No other significant differences were observed when comparing the NOESY spectra of 4 and 3. Accordingly, the structure of 4 was established to be (5-(bis(3,8α-dimethyl-5-methylene-4,4α,5,6,7,8,8α,9-octahydronaphtho-[2,3-β]-furan-2-yl)methyl) furan-2-yl)methanol, trivially named 5-furanmethanol-2-methanetriyl-biatractylon.

Proposed Mechanism for the Formation of Compounds 1-5
To the best of our knowledge, this is the first report to delineate the possible role that processing plays in producing new chemical entities and impacting the chemical composition of traditional medicines. Five new compounds (1-5) were generated as a result of the processing of RAM by stir-frying with wheat bran. A plausible rationale for the formation of new atractylon derivatives (1-5) is outlined below. Both RAM and wheat bran contain fiber, polysaccharides, cellulose, resistant starch, inulin, lignins, and oligosaccharides [1,24]. Thermal processing of cellulose, hemicellulose, and other polysaccharides are known to produce significant amounts of diverse carbonyl compounds such as formaldehyde [25], acetaldehyde, furfural, 5-hydroxymethylfurfural, etc. [26]. Isolation of major quantities of both 5-(hydroxymethyl)furfural (6) and 5-(hydroxymethyl)-2-(dimethoxymethyl)furan (7) from the processed RAM in this study further supports the formation of carbonyl compounds. On the other hand, atractylon, the major sesquiterpene of A. macrocephala [15,27], can undergo electrophilic reactions with the pyrolytic aldehyde products. For example, as shown in the proposed plausible mechanism (Figure 2

Biological Activities of Compounds 1-5
The new compounds (1-5) were evaluated for their cytotoxicity to a panel of selected mammalian cancer and noncancer cell lines (SK-MEL, KB, BT-549, SK-OV-3, LLC-PK1, and Vero cells). None of the compounds exhibited cytotoxic effects. All compounds were ineffective at inhibiting the growth of the pathogenic microorganisms including three fungi (C. albicans, C. neoformans, and A. fumigates) and five bacteria (S. aureus, methicillin-resistant S. aureus, E. coli, P. aeruginosa, and M. intracellulare). In addition, the in vitro antileishmanial activity testing revealed that the compounds were not effective against L. donavani. Even though the preliminary biological data of these interesting class of compounds are not encouraging, further biological studies on this class of compounds concerning other biological targets, as well as SAR and medicinal properties need to be further investigated. In addition to biological activities, probing the comparative physicochemical properties of these polymeric compounds against monomeric atractylon might provide additional scientific rationale behind the traditional processes associated with polyherbal formulations.

General Experimental Procedures.
UV and IR spectra were obtained on a HP 8452A UV-Vis spectrometer (Hewlett-Packard, Palo Alto, CA, USA) and an Cary 630 FTIR spectrometer (Agilent Technologies, Santa Clara, CA, USA), respectively. High-resolution mass spectra were obtained on an Agilent TOF LC/MS spectrometer equipped with an atmospheric pressure chemical ionization (APCI) source and operated with the Analyst QS 1.1 software for data acquisition and processing. NMR spectra were acquired on an Agilent DD2-500 NMR spectrometer with a OneNMR probe at 500 MHz for 1 H and 125 MHz for 13 C

Biological Activities of Compounds 1-5
The new compounds (1-5) were evaluated for their cytotoxicity to a panel of selected mammalian cancer and noncancer cell lines (SK-MEL, KB, BT-549, SK-OV-3, LLC-PK1, and Vero cells). None of the compounds exhibited cytotoxic effects. All compounds were ineffective at inhibiting the growth of the pathogenic microorganisms including three fungi (C. albicans, C. neoformans, and A. fumigates) and five bacteria (S. aureus, methicillin-resistant S. aureus, E. coli, P. aeruginosa, and M. intracellulare). In addition, the in vitro antileishmanial activity testing revealed that the compounds were not effective against L. donavani. Even though the preliminary biological data of these interesting class of compounds are not encouraging, further biological studies on this class of compounds concerning other biological targets, as well as SAR and medicinal properties need to be further investigated. In addition to biological activities, probing the comparative physicochemical properties of these polymeric compounds against monomeric atractylon might provide additional scientific rationale behind the traditional processes associated with polyherbal formulations.

General Experimental Procedures
UV and IR spectra were obtained on a HP 8452A UV-Vis spectrometer (Hewlett-Packard, Palo Alto, CA, USA) and an Cary 630 FTIR spectrometer (Agilent Technologies, Santa Clara, CA, USA), respectively.
High-resolution mass spectra were obtained on an Agilent TOF LC/MS spectrometer equipped with an atmospheric pressure chemical ionization (APCI) source and operated with the Analyst QS 1.1 software for data acquisition and processing. NMR spectra were acquired on an Agilent DD2-500 NMR spectrometer with a OneNMR probe at 500 MHz for 1 H and 125 MHz for 13 C using the pulse programs provided by the Agilent Vnmrj 4.0 software. Silica gel (J. T. Baker, 40 µm for flash chromatography) and Sephadex LH-20 were purchased from Fisher Scientific Co. (Waltham, MA, USA), and used for column chromatographic separations. Biotage Isolera TM Prime flash chromatography system (Biotage Co, Charlotte, NC, USA) was used for further separation and purification. TLC was performed on silica gel 60 GF 254 plates (Millipore Sigma, Burlington, MA, USA) with the TLC spots being observed at 254 nm, followed by spraying with 1% vanillin-H 2 SO 4 derivatization reagent. Processing of the rhizomes of A. macrocephala (2.5 kg) was conducted in the Processing Lab of the Experimental Training Center in the Heilongjiang University of Chinese Medicine. Following the processing protocol in the Chinese Pharmacopeia, the rhizomes were cut into slices and air-dried to reach a constant weight. The wheat bran was first stir-fried in a cauldron with a temperature around 170 • C until smoke appeared. The rhizome slices were then added into the cauldron with a ratio of the rhizome to wheat bran of 1:4 (w/w), the mixture was stir-fried approximately 26 mins until the slices turned to a yellow-brown color. Next, the processed rhizome slices were separated from the wheat bran by a sifter.

Cytotoxicity Assay
Cytotoxic activity was determined against four human cancer cell lines (SK-MEL, KB, BT-549, and SKOV-3) and two noncancerous kidney cell lines (LLC-PK1 and Vero) as described earlier [28]. All cell lines were obtained from the American Type Culture Collection (ATCC, Rockville, MD, USA). Each assay was performed in 96-well tissue culture-treated microplates. Cells were seeded at a density of 25,000 cells/well and incubated for 24 h. Samples at different concentrations were added, and cells were again incubated for 48 h. At the end of incubation, the cell viability was measured using a tetrazolium dye (WST-8) which was converted to a water-soluble formazan product. The absorbance was measured at 450 nm and the percent viability of sample treated cells was calculated in comparison to the vehicle-treated cells. Doxorubicin was used as a positive control, while DMSO was used as the negative (vehicle) control.

Antimicrobial Assays
Isolates were tested against a panel of 8 pathogenic organisms including three fungi (Candida albicans ATCC 90028, Cryptococcus neoformans ATCC 90113, and Aspergillus fumigates ATCC 204305) and five bacteria (Staphylococcus aureus ATCC 29213, methicillin-resistant S. aureus ATCC 33591, Escherichia coli ATCC 35218, Pseudomonas aeruginosa ATCC 27853, and Mycobacterium intracellulare ATCC 23068). Microorganisms were obtained from the American Type Culture Collection. The assays were performed at the National Center for Natural Products Research (NCNPR), at the University of Mississippi as a part of the antimicrobial screening program following a previously reported method [29]. Drug controls, ciprofloxacin for bacteria and amphotericin B for fungi, were included in each assay.

Antileishmanial Assay
Compounds were tested in vitro for their ability to inhibit Leishmania donovani, employing the assay described by Jain et al. [30]. Amphotericin B was included as the drug control for L. donovani.

Conclusions
Pao-Zhi (frying and cooking of herbs) is an ancient pharmaceutic technique in TCM to facilitate the use of herbal medicines for specific clinic needs [31]. Traditionally, most Chinese herbal medicines undergo elaborate processing in order to become ingredients that are prescribed or utilized in the manufacturing of TCM proprietary drugs [32]. Although the practice of processing has a long history, the underlying mechanisms largely remain unclear for most Chinese herbal medicines. In the present study, through the characterization of chemical profiles coupled with NMR metabolomics approach, the chemical changes resulting from the traditional processing protocol associated with RAM preparation were investigated. For the first time, five new chemical adducts, which were formed during processing A. macrocephala with wheat bran, were isolated and their structures were identified. The findings allowed us to gain valuable insights into the chemical reactions which occur during the processing procedures. Processed RAM is widely used in various formulations of TCM drugs and health care products. Stir-frying with wheat bran is one of the most widely used traditional processing methods for RAM in TCM [12]. Although the processed RAM is listed as an item in the Chinese Pharmacopoeia, there are currently no modern standardized processing protocols or quality control standards for the processed RAM products. The findings of this study may provide useful information for developing such standards with a scientific basis. Furthermore, as the change of chemical profile will inevitably influence the associated pharmacological properties of processed RAM, further investigations of the bio-activities of the newly generated compounds are needed.