Discrimination and Evaluation of Wild Paris Using UHPLC-QTOF-MS and FT-IR Spectroscopy in Combination with Multivariable Analysis

Genus Paris has numerous bioactive constituents such as steroid saponins, flavonoids, and polysaccharose which are responsible for antitumor, hemostatic, and anthelmintic, etc. In this study, ultrahigh performance liquid chromatography coupled to time-of-flight mass spectrometer (UHPLC-QTOF-MS) and Fourier transform infrared (FT-IR) spectroscopy in combination with multivariable analysis were employed to discriminate the different species of Paris including P. polyphylla var. yunnanensis (PPY), P. polyphylla var. alba, P. mairei (PM), P. vietnamensis, and P. polyphylla var. stenophylla. Partial least square discriminate analysis based on UHPLC, FT-IR, and midlevel data fusion was used to distinguish 43 batches of Paris. Chemical constituents of different species Paris were determined by UHPLC-QTOF-MS. The result indicated that midlevel data fusion had a good performance in the classification compared to a single analytical technology. A total of 47 compounds were identified in different species Paris. The similar results indicated that PM could be treated as a proposal substitute of PPY.


Introduction
Genus Paris (Lilaceae family) including 24 species, is mainly distributed in the southwest of China except for Paris birmanica and P. japonica [1]. As an ethnic medicine, it is recorded in Shennong Materia Medica which is one of the longest medical books in China for treatment of hyperspasmia and bite wound in the folk for a long history [2]. In addition, P. polyphylla var. yunnanensis (PPY) and P. polyphylla var. chinensis have been documented in Chinese Pharmacopoeia (ChP) named as Rhizome Paridis (RP) for treating in furunculosis, throat-swelling, traumatic injury, and so on [3]. Modern investigation suggests Paris has bioactive constituents such as steroid saponins, favonoids, and polysaccharose which are responsible for antitumor, hemostatic, and anthelmintic, etc [4,5]. Te extraction of RP is material to make the Chinese patent drugs contain "Gongxuening capsule" and "Jidesheng Sheyao Tablet." Numerous sources are gradually applied to the industrial production in decade years. Following price of commercial produce including PPY and PPC is higher than the past in the market. Up to now, P. polyphylla var. stenophylla (PPS), P. mairei (PM), and P. vietnamensis (PV) are barely investigated by isolation and identifcation of the major bioactive steroidal saponins [6][7][8]. At the same time, few research studies could not notice the P. polyphylla var. alba (PPA). Te relationship between diferent species is hardly illuminated in previous study either.
In previous study, the research indicated an amount of active compounds was found in Paris, while only four bioactive markers were determined by ChP for the quality control of RP. It could not respond to multiconstituents and multitargets for traditional Chinese medicine. It was therefore necessary to develop an analytical method which could be responsible for the comprehensive chemical compounds. Fortunately, fngerprints analysis with the advantage of systematic and efective characteristics is used to evaluate the sample of diferent geographical origin [9,10], parts [11], and species [12,13]. As far as we know, chromatographic and spectrographic fngerprints were investigated by previous study. Te former could provide the unambiguous and specifc information of compounds based on the peak signal. At the same time, the later with the overall and convenient characteristics are used to illustrate the chemical profle. For evaluation of herbal medicine, ultrahigh performance liquid chromatography (UHPLC) as a fast and efective method could provide the targeted compounds of samples [14]. Meanwhile, Fourier transform infrared spectroscopy (FT-IR) with nondestructive and feasible character is a characteristic of integrated chemical information [15]. However, few research studies have been focused on the strategy which combined UHPLC and FT-IR with chemometric for evaluation of Genus Paris.
Compared to a single analytical technique, data fusion of diferent instrumental (spectroscopic and chromatographic) techniques together with multivariate chemometrics can increasethe model classifcation [16]. Tree date fusion strategies (low-, mid-, and high level) can be carried out basically. Especially, mid-level fusion can be able to flter block noise and enables interpretation of the results, frstly extracts some relevant features from each data of instrument, and then merges them into a single array that is used for classifcation [17]. Te aim of this investigation is to validate and develop an analytical strategy which can evaluate and discriminate different species of Paris with regard to PPY, PPA, PPS, PM, and PA. Firstly, chemical profling of Paris was identifed by UHPLC-QTOF-MS. Ten, diferent species of Paris were discriminated through UHPLC, FT-IR, and data fusion coupled to chemometrics. It is expect that the study can provide a fundamental of genus Paris sources.  Figure 1.

Apparatus.
A UHPLC-UV system (Shimadzu, Kyoto, Japan) equipped with a degasser, binary gradient pumps, UV detector, a column oven, and an auto-sampler was utilized to obtain UHPLC fngerprints. System control and data analysis were conducted by LabSolution software (Shimadzu). Te chromatographic separation was operated on Shim-pack XR-ODS III column (150 × 2.0 mm, 2. Other chromatographic parameters were as follows: Te column temperature was maintained at 45°C. Te fow rate was 0.45 mL/min. All samples were injected with 1 μL. Mass spectrometry was performed on a Triple-TOFTM 5600 + system mass spectrometer (AB SCIEX, Foster City, CA, USA). Data acquisition was conducted in the negative electrospray (ESI) ionization mode. Te ESI-MS parameters were as follows: mass scan range for both TOF-MS and TOF-MS/MS: m/z 50-1200, ion spray voltage: 4500 V, atomizer temperature (TEM): 500°C; declustering potential (DP): 80 V; curtain air pressure (CUR): 40 psi; nebulize gas (Gas 1) and auxiliary gas (Gas 2) pressure: 50 psi. Te collision energy was set at 50 eV. Instrument control, data acquisition, and analysis of data were Analyst TF 1.6 and PeakView 1.2.
For FT-IR, table press (Shanghai, China) was employed to press the samples into thin sample. FT-IR (Perkin-Elmer, Norwalk, CT, USA) was equipped with a deuterated triglycine sulfate detector. IR spectra were recorded from the accumulation of 16 scans in 4000-400 cm −1 range with a resolution of 5 cm −1 .

Sample Preparation.
For UHPLC, 0.1 g of sample powder after passing through a 60 mesh sieve was weighed accurately, transferring into 10 mL glass stopper tube and adding 2 mL of 80% methanol with ultrasonically extracted for 40 min at room temperature. Te fltrates were collected by flter paper after sample solution cooling to the room temperature. All solution was stored at 4°C and passed through a 0.22 μm membrane flter before injection into the UHPLC system. Te standard stock solution for qualitative analysis was dissolved all reference standards with methanol to fnal concentration of 1 mg/mL. For FT-IR, 1.5 mg of dried sample and 100 mg KBr was weighted accurately and mixed completely by the agate mortar and then a homogeneous tablet was pressed . Te KBr pellet background of CO 2 and H 2 O was deducted before experiment. Te FT-IR spectra of sample were scanned 2 times and recorded after preheating 60 min under relative humidity 65% at the temperature 25°C.

Data Analysis. UHPLC chromatographic fngerprints
were recorded by the sampling points from 2.5 to 29 min with the time interval of 0.0833 min. Owing to the disturbance of methanol signal noise, the sampling points were rejected before 2.5 min. Date of UHPLC was entered into Similarity Evaluation System for Chromatographic Fingerprint of Traditional Chinese Medicine (Version 2004A, Chinese Pharmacopoeia Committee) for similarity analysis. For FT-IR, the data of spectra were entered into the software of OMNIC (Version 8.2, Termo Fisher Scientifc Inc, USA), which were used by noise reduction, baseline correction, and so on.
As a supervised method, PLS-DA is used to concentrate the multidimension data into two dimension, which can illustrate a sample whether it belong to a special class [18]. To evaluate and discriminate Paris of diferent species, PLS-DA based on UHPLC and FT-IR spectra was used to make the further study. All of data were imported into SIMCA-P + 10.0 (Umetrics AB, Umea, Sweden). To remove the baseline shifts and overlap peaks, the data were subjected to second derivative before analysis [19].

Method Validation.
For UHPLC, steroid saponins including PSI, PSII, PSVI, and PSVII were used to develop and validate method for UHPLC. Precision was performed by the inter-and intraday variations to make the analytical method accurate. Te intraday was accumulated by retention time and peak areas using six repetitive injection of mixed standards in a day and three consecutive days for the interday variation. Te results implied that relative standard deviations (RSD) of intra-and interday were no more than 2.98% and 2.83%, respectively (Table 2).
For FT-IR, the precision was performed using fve consecutive scan with a sample, and the results indicated that the correlation coefcient was ranged 0.9996-0.9999 with RSD � 0.05%. Stability Stability was calculated by  International Journal of Analytical Chemistry scanning a sample every ten minutes and fve consecutive times. Te results suggested that correlation coefcient was between 0.9995 and 0.9999, and RSD � 0.03%. Repeatability was carried out by fve tablets with a sample for scanning, and the results implied that correlation coefcient was more than 0.9996, and RSD � 0.02%.

Similarity Analysis for UHPLC.
UHPLC chromatographic fngerprint of 43 batches fve species of Paris is shown in Figure 2. Peaks 1, 2, 3, and 4 in chromatographic pattern were assigned the bioactive compounds of PSVII, PSVI, PSII, and PSI based on the reference standards. As we can see, PSII and PSI changed signifcantly in each sample with even not being detected in PPA, whereas PSVI was found in the most of samples. It indicated that chemical constituents varied markedly for the Paris of diferent species. Te result was similar to the previous research that the contents of chemical components have diference among collections of Paris of diferent geographical origins and species [20]. For similarity evaluation, the average chromatogram of 43 batches of samples was treated as the standard characteristic fngerprint. Similarity values were calculated by compared with each chromatogram of the Paris samples to the average chromatogram. As shown in Figure 3, the similarities by all of samples were no more than 0.66 (PM3). For the most of samples, the values were ranged from 0.55 to 0.65. Te changed chemical constituents of sample may be discovered in PM9 (0.49) and PPS2 (0.38) with the value less than 0.5. However, it is difcult to clarify the relationship between diferent species of Paris. Figure 4, a two-dimension score scatter plot with 95% confdence ellipses of PLS-DA was used to evaluate and discriminate the diferent species of Paris. It is signifcant among various sample PV from Simao City of Yunnan Province with four outliers PV1, PV2, PV3, and PV4. Te collection sample with multiple soil, moisture, and surrounding environment may lead to vary in chemical components of each sample [9]. Te phenomenon is necessary to work in the further study. It is obvious that the chemistry in each sample changed markedly for the similar species in the rhizome of PV. Moreover, group 1 including PPA, PPY, and PM was distributed in the positive principle compound (PC) 1; on the contrary, other species of Paris in the negative PC1 were assigned group 2. Te previous morphological research reported that PPA, PPY, and PPS belong to the varieties of P. polyphylla Smith [1]. Te similar profle was found according to the theory of macroscopical phenotype. Fortunately, it is associated with the result that is a close relationship between PPA and PPY compared to other species, especially PPA2, PPA6, PPA7, and PPA10 which are close to all of PPY species except for PPY2. However, the species of PPS had tight distance to PV without macroscopical theory. In group 1, P. marirei was closely related to P. polyphylla Smith, particularly in the species of PPY. As mentioned above, it is indicated that the chemical profle was usually related to the macroscopical characteristic with the diferent Paris. Te similar report in the previous research shows that chemotaxonomic studies of nine gentianaceae indicate some secondary metabolites could be treated as potential chemotaxonomic markers to diferentiate gentianaceae species. Moreover, the plots were weakly loose among the same species in the scatter plot which indicated that the chemical profle varied notably among the most of samples. As an ofcial medical resource, PPY has been widely applied to industrial production and clinic. Due to long growth cycle and excessive excavation, the wild population of PPY is gradually decreased. Obviously, since they had the similar chemical profle, PPA and PM could be treated as a proposal substitute of PPY. UHPLC was used to evaluate and discriminate diferent species of Paris with the characteristic of rapid and accurate compared with other technology. However, the chromatographic pattern may not be responsible for the entire bioactive chemistry of sample. A convenient and nondestructive method by FT-IR is necessary for distinguishing Paris.

Discrimination of Paris by FT-IR.
FT-IR is a technique based on the absorbance of light from 4000 to 400 cm −1 , which has been widely used for evaluation of herbal medicine and authentication of food products [21]. In this study, the samples of diferent species of Paris were analyzed by the FT-IR spectra shown in Figure 5. Te similar position and shape of peak were found in the most of sample. Te peaks were assigned as following: 3390, 2930, 1740, 1650, 1400, 1370, 1250, 1150, 1049, 930, 860, 765, and 700 cm −1 , which are characteristic common peaks of spectra. Te wavelength at 3390 cm −1 , generally the most prominent peak, was due to the stretching vibration peak of O-H in sugar moiety. Te peak at 2930 cm −1 was assigned to the stretch of methylene group. Absorption band at around 1740 and 1650 cm −1 corresponding to bending C�O and C�C stretching, respectively, was potentially related to steroid saponins, favonoids, and fatty acids. Te peaks at 1400 and 1370 cm −1 were due to plane deviational vibrations of methylene or methyl group. A bond of 1250 cm −1 was responded to the vibration peak of C-O in alcohol hydroxyl group. In addition, according to the previous research that peaks at 930, 860, 1150, and 1049 cm −1 , these were due to steroidal saponins skeletal vibration [22]. As mentioned above, the similar reported study shows that the main constituents were steroids saponins in Pairs, others including favonoids, fatty acids, and so on [5]. "t" the retention time of common peak of standard; "P" the peak areas of common peak of standard.

International Journal of Analytical Chemistry
As shown in Figure 6, the two-dimension score plots were carried out by FT-IR in conjunction with PLS-DA for evaluation of diferent species of Paris. Te result agreed with the analysis of UHPLC that PPY, PPA, and PM belonged to group 1 in positive PC2, and PV was attributed group 2 in negative PC2. In addition, the same phenomenon found that PPY was more closely related with PM than with PPA. However, the various results show that PPS was located in group 1 for this investigation. As the varieties of P. polyphylla Smith, PPS had the similar chemical constituents while the various contents were compared with UHPLC. It could draw a conclusion that the varieties of P. polyphylla including PPY and PPA were closely related with P. marirei no matter what analytical technology UHPLC or FT-IR held.

Discrimination of Paris by Midlevel Data Fusion.
Midlevel data fusion for PLS-DA was used to discriminate Paris of diferent species. Te combination of relevant features based on UHPLC and FT-IR was applied to establish a model which could be used for synthetic evaluation of Paris. As show in Figure 7, two dimension score PLS-DA of midlevel data fusion is displayed. Five species of Paris had a good performance for classifcation in the data fusion compared to a single instrumental data. Interestingly, it was similar to a single data model that PPY and PM had a close relation. It demonstrated that PM as a substitute of PPY with the further study was used to solve the resource shortage. In addition, all of PPA and PPS were located in the positive PC1 and PC2 except for PPS2. Te chemical profle was related to macroscopical characteristic which was proved in this study.

Chemical Analysis of Paris Using UHPLC-QTOF-MS.
Te mix sample of diferent species Paris was determined by UHPLC-Q-TOF MS in the negative electrospray modes. Te references PSI, PSII, PSVI, and PSVII were classifed, and the cleavage patterns were summarized. In addition, the characteristic ionic fragments of compounds were summarized based on cleavage fragments and structures of Paris. Finally, the compounds were identifed by molecular weight and secondary fragmentation. A total of 47 compounds were identifed in diferent species Paris. As shown in Table 3, it included twenty-three isosproterenol saponins, fve furostanol saponins, four cholestanol saponins, four favonoids, three pentacyclic triterpenoids, two C21 steroids, two phytosterol saponins, and four others.

Conclusions
To the best of our knowledge, the wild Paris was gradually decreasing due to the industrial production. Te aim of this study was to distinguish Genus Paris including PPA, PPY, PM, PV, and PPS by using UHPLC, FT-IR and midlevel data fusion in combination with multivariate analysis. Data fusion could classify the sample better than a single data array.  Chemical constituents of diferent species Paris were determined by UHPLC-Q-TOF MS. Te similar results showed that chemical constituents of PM were similar to PYY. It is expected that PM as a substitute of PPY with the further study was used to solve the resource shortage.

Data Availability
Te data used to support the fndings of this study are included within the article.

Conflicts of Interest
Te authors declare that they have no conficts of interest.