Comparative Research of Chemical Profiling in Different Parts of Fissistigma oldhamii by Ultra-High-Performance Liquid Chromatography Coupled with Hybrid Quadrupole-Orbitrap Mass Spectrometry

The roots of Fissistigma oldhamii (FO) are widely used as medicine with the effect of dispelling wind and dampness, promoting blood circulation and relieving pains, and its fruits are considered delicious. However, Hakka people always utilize its above-ground parts as a famous folk medicine, Xiangteng, with significant differences from literatures. Studies of chemical composition showed there were multiple aristolactams that possessed high nephrotoxicity, pending evaluation research about their distribution in FO. In this study, a sensitive, selective, rapid and reliable method was established to comparatively perform qualitative and semi-quantitative analysis of the constituents in roots, stems, leaves, fruits and insect galls, using an Ultra-High-Performance Liquid Chromatography coupled with Hybrid Quadrupole Orbitrap Mass Spectrometry (UPLC-Q-Exactive Orbitrap MS, or Q-Exactive for short). To make more accurate identification and comparison of FO chemicals, all MS data were aligned and screened by XCMS, then their structures were elucidated according to MSn ion fragments between the detected and standards, published ones or these generated by MS fragmenter. A total of 79 compounds were identified, including 33 alkaloids, 29 flavonoids, 11 phenylpropanoids, etc. There were 54 common components in all five parts, while another 25 components were just detected in some parts. Six toxic aristolactams were detected in this experiment, including aristolactam AII, AIIIa, BII, BIII, FI and FII, of which the relative contents in above-ground stems were much higher than roots. Meanwhile, multivariate statistical analysis was performed and showed significant differences both in type and content of the ingredients within all FO parts. The results implied that above-ground FO parts should be carefully valued for oral administration and eating fruits. This study demonstrated that the high-resolution mass spectrometry coupled with multivariate statistical methods was a powerful tool in compound analysis of complicated herbal extracts, and the results provide the basis for its further application, scientific development of quality standard and utilization.


Different Components among FO Parts Screened by XCMS
All the data were processed by XCMS with sample name, accurate mass-to-charge ratio, retention time, P value, Q value and intensity of each fragment. The zero-intensity fragments were manually searched and obtained, which meant the components represented by these ions were not detected or distributed in the related samples. Moreover,  The standards used in this experiment were marked in bolds; 43*: CAS number of 8,9-dimethoxy-7-met-hyl-10,11-dihydrodib-enzo[b,f]oxepine-1,6-diol. The standards used in this experiment were marked in bolds; "" implied the compounds i The standards used in this experiment were marked in bolds; "" implied the comp The standards used in this experiment were marked in bolds; "" implied the compounds in the re The standards used in this experiment were marked in bolds; "" implied the compounds i The standards used in this experiment were marked in bolds; "" implied the comp The standards used in this experiment were marked in bolds; "" implied the compounds in the re The standards used in this experiment were marked in bolds; "" implied the compounds i The standards used in this experiment were marked in bolds; "" implied the comp The standards used in this experiment were marked in bolds; "" implied the compounds in the re The standards used in this experiment were marked in bolds; "" implied the compounds i The standards used in this experiment were marked in bolds; "" implied the comp The standards used in this experiment were marked in bolds; "" implied th The standards used in this experiment were marked in bolds; "" implied the compounds in the related p The standards used in this experiment were marked in bolds; "" implied the compounds in the re The standards used in this experiment were marked in bolds; "" implied the comp The standards used in this experiment were marked in bolds; "" implied the compounds in the re The standards used in this experiment were marked in bolds; "" implied the compounds i The standards used in this experiment were marked in bolds; "" implied the comp The standards used in this experiment were marked in bolds; "" implied the compounds in the related p The standards used in this experiment were marked in bolds; "" implied the compounds in the re The standards used in this experiment were marked in bolds; "" implied the compounds i The standards used in this experiment were marked in bolds; "" implied the compounds i The standards used in this experiment were marked in bolds; "" implied the compounds in the related p -- The standards used in this experiment were marked in bolds; "" implied the compounds in the related p The standards used in this experiment were marked in bolds; "" implied the compounds i The standards used in this experiment were marked in bolds; "" implied the compounds in the related p The standards used in this experiment were marked in bolds; "" implied the compounds in the re The standards used in this experiment were marked in bolds; "" implied the compounds i The standards used in this experiment were marked in bolds; "" implied the compounds in the related p The standards used in this experiment were marked in bolds; "" implied the compounds i The standards used in this experiment were marked in bolds; "" implied the compounds i The standards used in this experiment were marked in bolds; "" implied the compounds in the related p The standards used in this experiment were marked in bolds; "" implied the compounds in the re The standards used in this experiment were marked in bolds; "" implied the comp The standards used in this experiment were marked in bolds; "" implied th The standards used in this experiment were marked in bolds; "" implied the compounds in the related p The standards used in this experiment were marked in bolds; "" implied the compounds in the re The standards used in this experiment were marked in bolds; "" implied th The standards used in this experiment were marked in bolds; "" implied the compounds in the related p The standards used in this experiment were marked in bolds; "" implied the compounds in the related p The standards used in this experiment were marked in bolds; "" implied the compounds i The standards used in this experiment were marked in bolds; "" implied th The standards used in this experiment were marked in bolds; "" implied the compounds in the related p The standards used in this experiment were marked in bolds; "" implied th The standards used in this experiment were marked in bolds; "" implied the compounds in the related p The standards used in this experiment were marked in bolds; "" implied the compounds in the re The standards used in this experiment were marked in bolds; "" implied th The standards used in this experiment were marked in bolds; "" implied the compounds i The standards used in this experiment were marked in bolds; "" implied the compounds in the related p The standards used in this experiment were marked in bolds; "" implied the compounds in the related p The standards used in this experiment were marked in bolds; "" implied the compounds in the related p ---- The standards used in this experiment were marked in bolds; " The standards used in this experiment were marked in bolds; "" implied the compounds in the related part, while "-" meat no distribution.
" implied the compounds in the related part, while "-" meat no distribution.

Different Components among FO Parts Screened by XCMS
All the data were processed by XCMS with sample name, accurate mass-to-charge ratio, retention time, p value, Q value and intensity of each fragment. The zero-intensity fragments were manually searched and obtained, which meant the components represented by these ions were not detected or distributed in the related samples. Moreover, only the ones that showed zero value in all three-parallel samples of each part were considered as the none-distributed components. In this experiment, XCMS-aligned data contained 12,238 fragments in positive-ion-mode data and 19,861 in negative-ion-mode data. After rearranging the retention times in Excel, the zero-intensity fragments at the same or veryclose (<0.05) retention time were used for qualitative analysis. A total of 25 differential components of FO parts were identified and obtained, including 6 alkaloids, 13 flavonoids, 3 phenylpropanoids, 3 triterpenes ( Table 2).
The result indicated an interesting phenomenon about the chemical absence in plant part(s). For instance, three components were only detected in roots, such as norfissilandione, norcepharadione B and gedunin. This may function specially for roots, generated by specific root tissues (e.g., root tip, root hairs, etc.) or the different environments (lightless in soil) from other parts [34,35], pending further research, like laser-capturing to detect the tissue or cell chemicals [36,37]. In contrast, there were five components distributed in all the parts except root, including artabotryside A, apiin, nicotiflorin, eupatolin and quercetin-3-O-rhamn-oside, which may be generated by the above-ground tissues. Meanwhile, the stem and insect gall showed same chemical distribution, and two compounds, claussequinone and methyl 3-(2-oxo-2-prop-2-enoxyethyl)-1-benzofuran-2-carboxylate, were only absent in these two parts. Leaves and fruits had the similar results that cnidimol B, claussequinone, morusin and fissohamione distributed only here, while aristolactam FI was absented. Aristolactam FII, Kwangsienin A and Isopedicin were only absent in leaves, while isoquercitrin was just detected in leaves. The components of these different distributions were determined by the specific metabolic functions of these tissues, or the different external environment leading to plant defense [38] or other actions, which benefited the metabolic mechanism research [39]. In other aspects, the identification of Chinese traditional patent medicine is more difficult than that of Chinese herbal medicine [40], due to the usage of their powders or extracts. Although the identification techniques (liquid or gas chromatography, etc.) have been developed significantly already, it is still hard to find a specific way to tell the real resource. Hence, these differences of FO compounds can be used as identification markers of which FO tissues used in proprietary Chinese medicine, increasing the safety and efficacy of medication.

Multivariate Statistical Analysis and Comparison
In recent years, high-resolution LC/MS technology combined with multivariate statistical methods were applied to accurately perform for the omics study of traditional Chinese medicine [41][42][43]. In this study, we also performed this method to compare the component analysis of multiple samples. After sieving by XCMS, all the data without zero were considered as the common fragments of each FO part. Multivariate statistical analysis was performed to compare and analyze their differences, including principal component analysis (PCA), partial least-squares discrimination analysis (PLS-DA) and orthogonal partial least-squares discrimination analysis (OPLS-DA). The purpose was to establish a relationship model between component expression and samples to realize the prediction and judgment analysis of the data category. The steps included importing the excel data into the SIMCA-P 14.0 for modeling and automatically fitting to screen the different models [44][45][46][47][48]. The experimental data was optimally modeled under OPLS-DA, and the UV model was built with the following parameters: positive ion mode: R 2 X = 0.975, R 2 Y = 1, Q 2 = 0.959; negative ion mode: R2X = 0.999, R2Y = 1, Q2 = 0.886. The closer R 2 and Q 2 values were to 1, the better the model was [47,48]. From Figure 2a,b, the OPLS-DA model of positive and negative ion data has been successfully established, and all the FO samples were clearly clustered into five categories, indicating that the expressions of common components in FO parts were significantly different, and the fruit and insect galls showed much similarity under this model.  The extent of component differences was evaluated by VIP values (variable importance for the projection importance) [44,45] to compare the strength and explanatory power of each component expression on their classification and discrimination. The ions with VIP value > 1.0 were consider as the fragments of markers, and their expression intensities were used to compare relative content in different FO components. The fragments farther from the clusters had higher VIP values, and all the ions with a VIP value greater than 1 were screened out in Excel format. According to the retention time and mass-to-charge ratio of these ions, the MS data were targeted for searching, and the corresponding common components were sorted out. Afterwards, the average intensity of each compounds in three parallel data for each FO part was used as the comparison benchmark for relative content (%) [30,47,48] (Figure 3). A total of 54 components (markers) in different FO parts were obtained and identified, including 27 alkaloids, 16 flavonoids, 8 phenylpropanoids, 1 sesquiterpene, 1 quinone and 1 phenolic acid component (see Table 1). Figure 3 showed the relative percentage of 54 common components, in which the content of most compounds in roots, stems and leaves were generally higher than others. Their intensities were used to express the relative percentage content in Figure  3a,b. Comparing the contents in FO parts, most components were significantly more in roots, stems and leaves than other parts. The constituents mainly distributed in FO roots were glaucine, aloenin, duguevanine, licocoumarin A, clausamine F, thunberginol C, oxocrebanine, anolobine, etc., of which, notably, isoquercitrin, norfissilandione, norcepharadione B and gedunin were only detected in roots. The constituents dominating in stems were found to be glucosyringic acid, dehydrodeguelin, lactucin, etc., which, being polar species, can be transported in stem xylems [49]. The main components detected in leaves were astilbin, corytuberine, demethylmoracin I, calycinine, noraristolodione, ammidin, 5, 6, 7, 8, -Tetramethoxyflavone, Aristolactam BIII, etc., while fruits had the biggest rele-vant contents of fissistigmatin A, xylopine and fissilandione. However, only crebanine was mainly distributed in insect galls, and with good antibacterial activity [50,51]; crebanine should be a way to protect the plant from insect injured. ratio of these ions, the MS data were targeted for searching, and the corresponding common components were sorted out. Afterwards, the average intensity of each compounds in three parallel data for each FO part was used as the comparison benchmark for relative content (%) [30,47,48] (Figure 3). A total of 54 components (markers) in different FO parts were obtained and identified, including 27 alkaloids, 16 flavonoids, 8 phenylpropanoids, 1 sesquiterpene, 1 quinone and 1 phenolic acid component (see Table 1).  Table 1. Figure 3 showed the relative percentage of 54 common components, in which the content of most compounds in roots, stems and leaves were generally higher than others. Their intensities were used to express the relative percentage content in Figure 3a,b. Comparing the contents in FO parts, most components were significantly more in roots, stems and leaves than other parts. The constituents mainly distributed in FO roots were glaucine, aloenin, duguevanine, licocoumarin A, clausamine F, thunberginol C, oxocrebanine, anolobine, etc., of which, notably, isoquercitrin, norfissilandione, norcepharadione B and gedunin were only detected in roots. The constituents dominating in stems were found to be glucosyringic acid, dehydrodeguelin, lactucin, etc., which, being polar species, can be transported in stem xylems [49]. The main components detected in leaves were astilbin, corytuberine, demethylmoracin I, calycinine, noraristolodione, ammidin, 5, 6, 7, 8, -Tetramethoxyflavone, Aristolactam BIII, etc., while fruits had the biggest relevant contents of fissistigmatin A, xylopine and fissilandione. However, only crebanine was mainly distributed in insect galls, and with good antibacterial activity [50,51]; crebanine should be a way to protect the plant from insect injured.

Aristolactams Differences
Since the first-reported over 100 cases of aristolochic acid nephropathy (AAN) in Belgium [52], much attention has been paid to the distribution and toxicity of aristolochic acid (AA) and aristolactam (AL) in traditional herbs, especially for the traditional Chinese medicine (TCM) [53]. Epidemiological studies showed that AA and AL exposure is associated with a high risk of nephrotoxicity and upper urinary tract carcinoma (UUC), indicating that these toxic compounds' distribution was of wide concern [54][55][56][57][58]. According to the previous studies, 14 aristolactams have been isolated or detected from FO, including aristolactam A II, A III a, B II, B III, F I, F II, GI, GII, oldhamactam, stigmalactam, piperolactam C, enterocarpam I, velutinam and goniothalactam, most of which were derived from FO stems [9,[59][60][61].  Table 1.

Aristolactams Differences
Since the first-reported over 100 cases of aristolochic acid nephropathy (AAN) in Belgium [52], much attention has been paid to the distribution and toxicity of aristolochic acid (AA) and aristolactam (AL) in traditional herbs, especially for the traditional Chinese medicine (TCM) [53]. Epidemiological studies showed that AA and AL exposure is associated with a high risk of nephrotoxicity and upper urinary tract carcinoma (UUC), indicating that these toxic compounds' distribution was of wide concern [54][55][56][57][58]. According to the previous studies, 14 aristolactams have been isolated or detected from FO, including aristolactam A II, A III a, B II, B III, F I, F II, GI, GII, oldhamactam, stigmalactam, piperolactam C, enterocarpam I, velutinam and goniothalactam, most of which were derived from FO stems [9,[59][60][61].
In this study, a total of six aristolactams were identified in different parts of FO, including Aristolactam AII, AIIIa, BII, BIII, FI and FII. After the XCMS screening, three differently distributed aristolactams were found, in that aristolactam B II and F I were not found both in leaves and fruits and aristolactam F II only not in leaves (Figure 3b, Table 2). For all six aristolactams, they were mainly distributed in stems, insect galls and roots. Notably, all the relative content of aristolactams in above-ground stems were more than 50%, quite higher than other parts. For example, the percentages of aristolactam BII and BIII in stems were even 88% and 84%, respectively. FO leaves and fruits had less aristolactams both in number and content. However, eating fruits was considered as a risk due to these toxic components, and both underground roots (Guangxiangteng) and above-ground stems and leaves (Xiangteng) should also be carefully used as medicine, especially for oral administration.

Identification and Component Pyrolysis Based on MS Fragmentater
For component identification, the published spectra under same MS condition were quite useful to confirm the structures with the comparison of the experimental spectra, especially for the MS 2 spectra. However, most FO compounds did not have these sources in public databases. Hence, several standard components were used to assist the identification, such as (−)-epicatechin and aristolactam AIIIa shown in Figure 4. Meanwhile, to interpret other components without standards or published spectra, their MS fragmentations were predicted by MS Fragmentater [62,63] for the comparison with the detected fragments. Tak-ing the cleavage of Fissistigine C [64] as an example, it was detected in positive ion mode with a strong molecular ion peak, m/z 342. 1699 ] + . The experimental fragmentation fit very well with the fragments predicted above; therefore, it was identified as Fissistigine C. The fragment ions were generated via several processes, including protonation, π bond dissociation, inductive cleavage, rearrangements, etc. (Figure 5c).

Method Validation
Based on the above optimized condition, the mixed stock solution was diluted with methanol to different concentrations for constructing calibration curves. The ratio of peak areas (y) of each compound was plotted against the concentrations (x, ng mL −1 ). All the curves exhibited good linearity with determination coefficients (R 2 ) from 0.9988 to 0.9999. The LOQ and LOD were evaluated by the signal-to-noise (S/N) of 10 and 3. The values of LOD and LOQ in this experiment were 0.12-6.41 and 0.21-5.45 ng mL −1 , respectively.
Meanwhile, the intra-and inter-day precision were calculated by analyzing the stock solution under the above optimized conditions, with respective RSD values less than 4.8% and 4.9%. In addition, the acceptable RSD values of repeatability (n = 6) and stability in 8 h (n = 6) were 2.3-4.5% and 2.8-4.8%.     The recovery was validated by adding a known amount of stock standard at different concentration levels (high, middle and low, n = 3) into a selected sample. The mixtures were analyzed in triplicate with the optimized method. The recovery of this method varied from 99.5% to 103.5% with the RSD value between 0.5% and 3.58%. The detailed data are listed in Table 3. 50-1000. The optimal MS parameters were as follows: spray voltage −2.8 Kv/+3.5 Kv; sheath gas flow rate, 35 arbitrary units; auxiliary gas flow rate, 10 arbitrary units; capillary temperature, 320 • C; auxiliary gas heater temperature, 350 • C. The resolution of full scan and dd-ms 2 were 70,000 and 35,000 FWHM (full width at half maximum), while their AGC target were set as 3 × 10 6 and 1 × 10 5 , with their maximum IT (the maximum injection time allowed to obtain the set AGC target) 100 and 50 ms, respectively. The stepped NCE (normalized collision energy) was set to 35 V for MS/MS acquisition.

Data Processing and Analysis
All the data operation, acquisition and analysis were done by Xcalibur 4.2 (Thermo Fisher Scientific Inc., Waltham, MA, USA), ACD/MS Workbook Suite 2020 (Advanced Chemistry Development, Inc. Toronto, ON, Canada), combined with MS Fragmentater 2020 (ACD/labs), XCMS-online (Scripps Research Institute, La Jolla, CA, USA) and SIMCA 14.0 (Umetrics, Umeå, Sweden). The original UHPLC-MS data of each samples were exported, and their background were subtracted from mass spectrum data of the blank solvent using Xcalibur for both positive and negative ion data. Then, the processed data were imported to XCMS to extract peaks and align chromatograms to make the comparison more accurate than manual work. The processing parameters were as follows: mass range: 100-1000 Da; mass tolerance: 10 ppm; RT tolerance [min]: 0.05; S/N threshold: 5. Then, all the aligned MS data of FO samples without interfering peaks were obtained, including the retention time and precise molecular mass which were necessary for chemical identification.
Before detection, the published compounds of FO were comprehensively collected by checking Dictionary of Natural Products, Sci Finder, Web of Science, CNKI and other databases. Then, a manual compound library was created, including the chemical name, molecular formula, exact molecular weight, and structure information. Under Xcalibur 4.2, the mass spectrums and their secondary spectrum (top 5 ions of MS 2 fragmentation) of the molecular ion peaks were screened and checked by comparing the detected formula and MS fragments with the component library we built, and most compounds were targeted and identified. Then, their structures were elucidated by comparing with their detected MS n ion fragments and the fragmentation patterns of each compound were summarized and further verified according to MS Fragmenter.
Meanwhile, the aligned Excel data converted by XCMS were manually screened for the fragments with zero intensity, represented the absence of the corresponding compounds in the related FO part. By this method, the different compounds among FO parts were obtained. Secondly, the data without zero-intensity fragments were carried out with multivariate statistical analysis by SIMCA to identify the difference of component contents in each FO part. PLS-DA and OPLS-DA analysis of the processed data were then performed to identify the compounds with VIP > 1, which implied the markers as the main components with significant content differences among FO parts. Meanwhile, their intensity was considered as the relative content and were used for the content comparison.

Conclusions
In this paper, a rapid and reliable method using UHPLC-Q -Exactive-MS was established to perform qualitative and semi-quantitative analysis of the constituents in all Fissistigma oldhamii (FO) parts (roots, stems, leaves, fruits and insect galls). The analysis revealed a high content of virous compositions in FO, in which alkaloids and flavonoids were the main compounds. Via comparing both MS 1 and MS 2 spectra with standards and fragmentations, 79 components were identified, and they were compared by XCMS screening and multivariate statistical analysis. This showed that there were 25 components differentially distributed in FO parts, and 54 common ones (markers) were recognized with obviously different relative content, most of which exhibited quite higher contents in the roots and leaves. The results showed that the chemical components of the roots, stems, leaves, fruits and insect galls had obvious differences in types and relative contents.
Notably, due to the distribution of highly nephrotoxic aristolactams, all FO parts should be carefully valued for oral administration, especially for eating fruits. The method used in our experiment can quickly recognize the component differences between multiple samples, while the MS n comparison based on MS fragmenter can provide much more information for the identification of "unknown" compounds. However, the high-density data and charged software could be a barrier for the method application. In summary, the high-resolution LC/MS technology combined with multivariate statistical methods can accurately perform component analysis of different FO samples, and should be used for studies of other traditional Chinese medicine. Clear distributions of effective or toxic components in plants can provide a basis for the medicinal and edible usage of various plant tissues and for further studies of the metabolic process and mechanism of plant components.