Chemical Profiles of Cultivated Agarwood Induced by Different Techniques

Agarwood is the resinous wood produced in some Aquilaria species and is highly valued for wide usages in medicine, incense, and perfume. To protect the threatened Aquilaria species, the cultivation of Aquilaria sinensis and artificial agarwood induction techniques have been effectively established in China. To evaluate the quality of agarwood induced by different techniques, patterns of chemical constituents in artificial agarwood by four methods (wounding using an axe, burning-chisel-drilling, chemical inducer, and biological inoculation) were analyzed and compared by UPLC-ESI-MS/MS and GC-EI-MS in this study. Results of GC-MS gave a panorama of chemical constituents in agarwood, including aromatic compounds, steroids, fatty acids, sesquiterpenoids, and 2-(2-phenlyethyl)-chromones (PECs). Sesquiterpenoids were dominant in agarwood induced by wounding using an axe. PEC comprised over 60% of components in agarwood produced by biological inoculation and chemical inducers. PECs were identified by UPLC-ESI-MS/MS in all artificial agarwood and the relative contents varied in different groups. Tetrahydro-2-(2-phenylethyl)-chromones (THPECs) in wounding by axes induced agarwood were lower while 2-(2-phenylethyl)-chromones (FPECs) were higher than other groups. The results showed that methods used for inducing agarwood formation in Aquilaria sinensis affect the chemical constituents of agarwood.


Introduction
The process of agarwood formation is initiated in the parenchyma cells in some species belonging to Thymelaeaceae family [1]. The cells activate protective machinery to synthesize secondary metabolites which defend against the stress [2]. The chemicals are produced, stored, and distributed in the wounded areas, filling the anatomical compartments [3]. Based on the process, various artificial agarwood induction methods have been invented and classified into two groups: conventional and nonconventional methods [4]. The conventional techniques include physical intrusions such as burning-chisel-drilling and wounding using an axe, which have been established for nearly 1000 years in ancient China [5]. Non-conventional methods have been newly developed in recent years [4]. Solvents containing ions or some microbes are injected into trunks of health trees to stimulate agarwood formation [6,7]. These methods are intensively practiced in plantations of China and other agarwood producing countries [5]. Agarwood produced by artificial induction techniques will inevitably be the majority in the commercial and medical market in the near future. Therefore, the quality of cultivated agarwood is of great concern. Some recent reports suggested that agarwood induced by nonconventional methods were comparable with wild agarwood while others reported the obvious differences in chemical constituents between wild and cultivated agarwood [8][9][10]. Studies Although morphology sometimes affect the commercial prices of agarwood used in sculptures or artifacts, resin contents which is scaled as ethanol extract contents are the crucial criteria in agarwood qualification system in most agarwood producing and consuming regions [26]. Resin contents partly depend on the extent of whitewood removal, especially for the artificial agarwood whose morphology is regular. Chemical constituents including chromones and sesquiterpenes in ethanol extract are contributors to the pleasant odor and medical effects of agarwood. Thus, the chemical constituents in ethanol extracts should be carefully analyzed to assess their qualities.

GC-MS Analysis
As agarwood is valued for its unique and pleasant odor, GC-MS were widely used in determining the components and access the quality. To fully acquire the chemical profiles of agarwoods, the ethanol extracts were eluted and analyzed by GC-MS. The typical TIC chromatographs of agarwood by four inducing methods are shown in Figure 2. Chemicals that showed up before retention time (RT) of 57 min were mainly sesquiterpenes and aromatic compounds. PECs were detected between 57 and 87 min with some long chain fatty acid. Steroids showed up in the last 15 min. Chemicals, except for some PECs, were identified by searching NIST14 with MS data and retention index. Tentatively identified chemicals with similarity over 80% and retention index within ±20 are listed in Table 1. Some PECs identity was done by referring to previous reports. Proportion of sesquiterpenoid was highest in groupA. PECs made up most proportions of Although morphology sometimes affect the commercial prices of agarwood used in sculptures or artifacts, resin contents which is scaled as ethanol extract contents are the crucial criteria in agarwood qualification system in most agarwood producing and consuming regions [26]. Resin contents partly depend on the extent of whitewood removal, especially for the artificial agarwood whose morphology is regular. Chemical constituents including chromones and sesquiterpenes in ethanol extract are contributors to the pleasant odor and medical effects of agarwood. Thus, the chemical constituents in ethanol extracts should be carefully analyzed to assess their qualities.

GC-MS Analysis
As agarwood is valued for its unique and pleasant odor, GC-MS were widely used in determining the components and access the quality. To fully acquire the chemical profiles of agarwoods, the ethanol extracts were eluted and analyzed by GC-MS. The typical TIC chromatographs of agarwood by four inducing methods are shown in Figure 2. Chemicals that showed up before retention time (RT) of 57 min were mainly sesquiterpenes and aromatic compounds. PECs were detected between 57 and 87 min with some long chain fatty acid. Steroids showed up in the last 15 min. Chemicals, except for some PECs, were identified by searching NIST14 with MS data and retention index. Tentatively identified chemicals with similarity over 80% and retention index within ±20 are listed in Table 1. Some PECs identity was done by referring to previous reports. Proportion of sesquiterpenoid was highest in groupA. PECs made up most proportions of chemicals detected in burning-chisel and non-conventional method induced agarwood ( Figure 2). However, THPECs overlapped with DPECs and EPECs in GC-MS analysis ( Figure 2B).    However, GC-MS is only suitable for volatile constituents and identification via GC-MS can only be achieved within limited library searching. Therefore, given the fact that PECs comprised a high percentage of the identified compounds, and while some nonvolatile PECs perhaps failed to be detected by GC-MS, UPLC-MS analysis was carried out to study the dynamic changes of PECs during the process of agarwood formation.

LC-MS Analysis
2-(2-phenylethyl) chromones are widely reported as the characteristic constituents in agarwood in regardless of their origins and induction methods. The chromone patterns of each groups were acquired by LC-MS/MS. Total ion chromatograph (TIC) obtained by scan mode showed that all artificial agarwood were qualitatively similar ( Figure 3A). Basal peaks were chosen and further identified by LC-MS/MS by production ion mode. All four types of 2-(2-phenylethyl) chromones were detected in artificial agarwood and their backbone structures are shown in Figure 3B.
Agarotetrol, 2-[2-(4-methoxphenyl)ethyl] chromone and 2-(2-phenylethly) chromone were identified by comparing their retention times and MS spectra with reference compounds. Tentative identification of chromones according to MS/MS fragments and previous reports are listed in Table 1. THPEC with 4 OHs at ring A were eluted firstly before 7 min. THPECs with less than 4 OHs, EPECs and DEPECs appear between 7 and 13 min. Chemicals detected after 13 min were molecular with least polarity such as FTPECs and sesquiterpenoids ( Figure 3 and Table 2). Proportions of each type of chromone in artificial agarwood samples and the most abundant chemical was shown in Figure 4.
FTPECs comprise most part of chromones in artificial agarwood ( Figure 4). THPECs is lower while EPECs and DEPECS were more abundant in group A than in other groups ( Figure 4). The above results strongly suggested chromones in artificial agarwood differ between groups.
detected by GC-MS, UPLC-MS analysis was carried out to study the dynamic changes of PECs during the process of agarwood formation.

LC-MS Analysis
2-(2-phenylethyl) chromones are widely reported as the characteristic constituents in agarwood in regardless of their origins and induction methods. The chromone patterns of each groups were acquired by LC-MS/MS. Total ion chromatograph (TIC) obtained by scan mode showed that all artificial agarwood were qualitatively similar ( Figure 3A). Basal peaks were chosen and further identified by LC-MS/MS by production ion mode. All four types of 2-(2-phenylethyl) chromones were detected in artificial agarwood and their backbone structures are shown in Figure 3B.

Multivariate Analysis
As the above results show, chemical constituents in artificial agarwood are complex. To further investigate the differences within groups, all MS data were further processed for multivariate analysis. PCA respectively based on GC-MS and LC-MS showed similar grouping tendency. Most samples of group A were allocated together while samples for other groups mixed up ( Figure 5A,B). Samples from group B (blue dots) scattered which suggested great variety both in sesquiterpenes and PECs. To further identify the molecules contributing to the classification of artificial agarwood, LC-MS data were used for features selection for each group. 26 chemicals were confirmed important for artificial agarwood grouping. Random Forest based on the intensity of above 26 chemicals showed that 9 with VIP > 1.5 ( Figure 6B). Those molecules were further identified by LC-MS/MS and the possible formula were listed in Table 3

Multivariate Analysis
As the above results show, chemical constituents in artificial agarwood are complex. To further investigate the differences within groups, all MS data were further processed for multivariate analysis. PCA respectively based on GC-MS and LC-MS showed similar grouping tendency. Most samples of group A were allocated together while samples for other groups mixed up ( Figure 5A,B). Samples from group B (blue dots) scattered which suggested great variety both in sesquiterpenes and PECs.

Multivariate Analysis
As the above results show, chemical constituents in artificial agarwood are complex. To further investigate the differences within groups, all MS data were further processed for multivariate analysis. PCA respectively based on GC-MS and LC-MS showed similar grouping tendency. Most samples of group A were allocated together while samples for other groups mixed up ( Figure 5A,B). Samples from group B (blue dots) scattered which suggested great variety both in sesquiterpenes and PECs. To further identify the molecules contributing to the classification of artificial agarwood, LC-MS data were used for features selection for each group. 26 chemicals were confirmed important for artificial agarwood grouping. Random Forest based on the intensity of above 26 chemicals showed that 9 with VIP > 1.5 ( Figure 6B). Those molecules were further identified by LC-MS/MS and the possible formula were listed in Table 3  To further identify the molecules contributing to the classification of artificial agarwood, LC-MS data were used for features selection for each group. 26 chemicals were confirmed important for artificial agarwood grouping. Random Forest based on the intensity of above 26 chemicals showed that 9 with VIP > 1.5 ( Figure 6B). Those molecules were further identified by LC-MS/MS and the possible formula were listed in Table 3  Nearly all chemicals contributing to artificial agarwood classification came out at RT 14-19 min, which are most tentatively identified as FTPECs or sesquiterpenes with less polarity (Table 3).

Discussion
In this study, morphology and chemical profile of artificial agarwood by four induction methods were investigated. Artificial agarwood from different induction methods vary and results acquired from multivariate analysis strongly suggested that inducing methods affect the chemical constituents of artificial agarwood. There are over 200 chemicals found in agarwood and the number is increasingly growing. In this study, GC-MS tentatively identified 71 and LC-MS/MS tentatively identified 43 chemicals in artificial agarwood (Tables 1 and 2). All samples had similar chemicals in the ethanol extracts while the relative contents differed (Figures 2 and 4).
Unlike PECs in agarwood, although numerous sesquiterpenoids were tentatively identified in GC-MS (Table 1), no characteristic chemicals were found even in group A. One reason for relative low proportion of sesquiterpenoids in all samples is that the variety of sesquiterpenoid skeletons made them difficult to be identified without standard chemicals. Samples from group A were obviously different from other samples for its higher contents of sesquiterpenoids and lower contents of THEPCs (Figures 2 and 4). One explanation for the relative lower proportions of sesquiterpenoids might be the shorter formation time in group B-D compared with group A. Sesquiterpenoids were reported be produced later than PECs [15]. THPECs are reported appear early during agarwood formation and evolved into FPECs as time goes on [11]. The formation time of samples in group A is usually over 18 months. Many samples in group B-D are about 9-16 months if known. Time length  Nearly all chemicals contributing to artificial agarwood classification came out at RT 14-19 min, which are most tentatively identified as FTPECs or sesquiterpenes with less polarity (Table 3).

Discussion
In this study, morphology and chemical profile of artificial agarwood by four induction methods were investigated. Artificial agarwood from different induction methods vary and results acquired from multivariate analysis strongly suggested that inducing methods affect the chemical constituents of artificial agarwood. There are over 200 chemicals found in agarwood and the number is increasingly growing. In this study, GC-MS tentatively identified 71 and LC-MS/MS tentatively identified 43 chemicals in artificial agarwood (Tables 1 and 2). All samples had similar chemicals in the ethanol extracts while the relative contents differed (Figures 2 and 4).
Unlike PECs in agarwood, although numerous sesquiterpenoids were tentatively identified in GC-MS (Table 1), no characteristic chemicals were found even in group A. One reason for relative low proportion of sesquiterpenoids in all samples is that the variety of sesquiterpenoid skeletons made them difficult to be identified without standard chemicals. Samples from group A were obviously different from other samples for its higher contents of sesquiterpenoids and lower contents of THEPCs (Figures 2  and 4). One explanation for the relative lower proportions of sesquiterpenoids might be the shorter formation time in group B-D compared with group A. Sesquiterpenoids were reported be produced later than PECs [15]. THPECs are reported appear early during agarwood formation and evolved into FPECs as time goes on [11]. The formation time of samples in group A is usually over 18 months. Many samples in group B-D are about 9-16 months if known. Time length of agarwood formation might be a crucial contributor for the differences between group A and other groups. However, other factors might also contribute to the chemical composition differences. As reports on mechanisms of agarwood formation suggest differences reside between the wounding by an axe and chemical inducer method through transcriptome and microtome analysis [2,27,28]. Whether those divergences accounts for the chemical variety of artificial agarwood should be an interesting question to be resolved.
Previous studies focused on agarwood qualifications proposed certain sesquiterpenes and PECs were accumulated in high-quality agarwood. In this study, agarwood in group A contained more sesquiterpenes and PECs than other groups (Figures 2 and 4), which suggested that agarwood in group A might be superior than other groups. Although many of the chemicals identified in this study can also been found in other plants, the complexity and variety of chemicals in agarwood confer its irreplaceable odor and pharmaceutical effects. Results from random forest point out several chemicals strongly correlated with samples groupings (Figure 6). It is still hard to conclude that those molecules are quality determinants. Further studies combining both chemicals analysis and medical effects or fragrant assays might help to identify the qualification markers.
Agarwood formation is believed to be the co-production of microbe and parenchyma cell at the injured sites as reported in recent research. Micro-environment of agarwood formation in conventional methods induced agarwood and the non-conventional ones differs, which can be observed by the morphology of sample in Figure 1A-D. Samples from group A were exposed to air and sunlight in whole inducing procedure while samples in group C-D were in dark and anaerobic niche until harvest. Agarwood in group B was produced in both niches; as shown in Figure 1B, holes were opened to the outside and the part near the hole were covered inside trunks. Group allocations by PCA on chemical profiles also reflect the tendency ( Figure 5). Group B is closer to group A than groups C and D (Figures 5 and 6). Samples in group C and group D were similar in morphology and chemical profiles (Figures 1 and 6). The results strongly suggested that inducing methods affect the chemicals in cultivated agarwood. Studies focused on the microtomic and biological activities in agarwood formation under different niches might shed more light on the complex mechanisms of agarwood formation.
Agarwood is precious due to the rare formation under natural environment. Various artificial inducing techniques have been developed. In this study, cultivated agarwood induced by four methods were analyzed and compared. Types of components in the ethanol extracts are similar while differing in proportions, which strongly suggest that inducing method affect chemicals in resins of agarwood. The results are helpful for study of mechanism behind agarwood formation. Agarwood induced by wounding of axes contains more sesquiterpenes and FPECs compared with other groups. The results provide a comprehensive assessment of agarwood induced by four popular artificial methods which will help to evaluate artificial agarwood quality.

Agarwood Materials and Reagents
All agarwood were from plantations in Guangdong, Guangxi, Yunnan, and Hainan provinces in South China. In total, 48 artificial agarwood induced by wounding using an axe (14), burning-chiseldrilling (9), chemical inducers (17) and biological inoculation (8) were investigated. Detailed sample information was listed in Table 4 Before the study, all samples were analyzed according the instructions of monography "agarwood" in Chinese Pharmacopoeia (2015 edition) to make sure the fake or adulterated samples were excluded from the following study [29].

Data Processing and Analysis
Raw LC-MS data were converted tomzXML format using MSconvert tools (Version 3,64 bit, proteowizard, Palo Alto, CA, USA). Raw GC-MS data were converted to mzXML format using GC-MS Postrun analysis (Shimadzu, Kyoto, Japan). Preprocessing of MS data including peak picking, peak grouping, retention time (RT) correction, and integration was performed using the XCMS implemented with R software (Version 3.5, University of Auckland, New Zealand). Each ion was identified by the RT and m/z data. Intensities of each peak were recorded and a three-dimensional matrix containing arbitrarily assigned peak indices and ion intensity information was generated. The intensities of each ions identified were normalized and the quantitative data were analyzed by several unsupervised methods and supervised methods in R. PCA (principle components analysis) was used for multivariate exploration of clusters and trends among the observation. Feature selection were performed by Boruta package in R.