Chemical Profiling of Re-Du-Ning Injection by Ultra-Performance Liquid Chromatography Coupled with Electrospray Ionization Tandem Quadrupole Time-of-Flight Mass Spectrometry through the Screening of Diagnostic Ions in MSE Mode

The broad applications and mechanism explorations of traditional Chinese medicine prescriptions (TCMPs) require a clear understanding of TCMP chemical constituents. In the present study, we describe an efficient and universally applicable analytical approach based on ultra-performance liquid chromatography coupled to electrospray ionization tandem quadrupole time-of-flight mass spectrometry (UPLC-ESI-Q/TOF-MS) with the MSE (E denotes collision energy) data acquisition mode, which allowed the rapid separation and reliable determination of TCMP chemical constituents. By monitoring diagnostic ions in the high energy function of MSE, target peaks of analogous compounds in TCMPs could be rapidly screened and identified. “Re-Du-Ning” injection (RDN), a eutherapeutic traditional Chinese medicine injection (TCMI) that has been widely used to reduce fever caused by viral infections in clinical practice, was studied as an example. In total, 90 compounds, including five new iridoids and one new sesquiterpene, were identified or tentatively characterized by accurate mass measurements within 5 ppm error. This analysis was accompanied by MS fragmentation and reference standard comparison analyses. Furthermore, the herbal sources of these compounds were unambiguously confirmed by comparing the extracted ion chromatograms (EICs) of RDN and ingredient herbal extracts. Our work provides a certain foundation for further studies of RDN. Moreover, the analytical approach developed herein has proven to be generally applicable for profiling the chemical constituents in TCMPs and other complicated mixtures.


Introduction
Traditional Chinese medicine prescriptions (TCMPs), which are combinations of several medicinal herbs, have been widely employed for thousands of years in China and other Asian countries. In clinical practice, TCMPs often exhibit significant advantages of low therapeutic risk and remarkable effect for some chronic, multifactorial and systemic diseases [1][2][3][4]. However, due to the extreme complexities of multiple TCMP components, revealing their pharmacological material basis and mechanism of action remains challenging. Consequently, an effective and reliable analytical approach for the rapid screening and identification of the multiple components contained in TCMPs is in high demand.
Currently, due to its significant advantages in analytical speed and detection sensitivity, ultra-performance liquid chromatography coupled with electrospray ionization tandem quadrupole time-of-flight mass spectrometry (UPLC-ESI-Q/TOF-MS) has become an irreplaceable technique for the on-line structural elucidation of multiple components in mixtures, especially for complex TCMs/TCMPs, biological samples and pesticide residues [5][6][7][8][9]. UPLC coupled with MS E ( E represents collision energy) technology provides an automated strategy to decrease analysis time and maximize duty cycles by using parallel alternating scans at low collision energy in the collision cell to obtain precursor ion information or at high collision energy to obtain accurate full-scan mass fragment, precursor ion and neutral loss information. Therefore, both precursor and fragmentation data in exact mass mode were collected in a single run; this method has provided excellent chromatographic and MS efficiencies for the rapid structural elucidation of multiple constituents in complex mixtures [10][11][12]. In the present study, based on the point of view that a certain type of chemical compounds could produce identical or similar characteristic fragment ions under a suitable collision energy in their tandem mass spectra, a well-designed analytical approach that enabled rapid screening and characterization of multiple TCMP constituents was developed. By virtue of UPLC-ESI-Q/TOF-MS and optimized MS E method, diagnostic fragment ions can be used as invaluable evidence for the detection of both expected and unexpected chemical constituents within TCMPs.
Re-Du-Ning injection (RDN), a traditional Chinese medicine injection (TCMI), was manufactured by Jiangsu Kanion Pharmaceutical Co. Ltd. (Lianyungang, China) and consists of three common herbs: Lonicera japonica Thunb. (L. japonica Thunb.; Jin-yin-hua), Gardenia jasminoides Ellis (G. jasminoides Ellis; Zhi-zi) and Artemisia annua L. (A. annua L.; Qing-hao). In China, RDN is widely used for the treatment of viral infection, such as hand-foot-mouth disease [13][14], influenza [15] and herpes angina efficacy [16]. Although RDN has proven to be clinically effective, the knowledge of its chemical constituents is still limited. The elucidation of the various components contained in RDN is urgently necessary and of great importance to RDN quality control and to understanding its mechanism of action.
In this paper, a robust Waters UPLC-ESI-Q/TOF-MS system and optimized MS E method was utilized, employing RDN as an example for illustration. To our knowledge, this work is the first study on the chemical components contained in RDN using the methodology developed herein. As a result, a total of 90 compounds, including 45 iridoids, 21 organic acids, nine flavonoids, seven lignans, four sesquiterpenes, three coumarins and one monoterpene were identified or tentatively characterized in RDN. In addition, the source plants of these compounds were confirmed by comparing the extracted ion chromatograms (EICs) of RDN to the corresponding ingredient herbs. This work provides a certain foundation for further studies of RDN. More importantly, this novel approach is expected to be widely applied for analyzing other TCMPs and complex mixtures.

Chemicals and materials
G. jasminoides Ellis, L. japonica Thunb. and A. annua L. were purchased from the Ji'an Medical Material Market (Jiangxi, China). All herbal medicines were identified by Professor Zhou Wu (Jiangsu Kanion Pharmaceutical Co. Ltd.). A voucher specimen was deposited in Jiangsu Kanion Pharmaceuticals (Lianyungang, China). The Re-Du-Ning injection (Batch number: 100906) was manufactured and supplied by Jiangsu Kanion Pharmaceutical Co. Ltd. (Lianyungang, China).
All reference standards were isolated from the RDN injection by various column chromatography techniques and were unambiguously identified by nuclear magnetic resonance (NMR) and MS methods in our laboratory.
Liquid chromatography (LC)-MS-grade acetonitrile and water were purchased from Fisher Scientific (Fair Lawn, New Jersey, USA). LC-MS-grade formic acid was obtained from Sigma-Aldrich (St. Louis, USA). The water, methanol and ethanol used for sample extraction were all of analytical grade.

Sample preparation
The RDN samples were directly evaporated with a rotary evaporator and then diluted to 10 mg/mL. Next, 2 mL of these solutions were transferred into separate clean tubes and dried under nitrogen gas at room temperature. The residues were reconstituted in 2 mL of water and then centrifuged at 10000 rpm for 10 min. Solid-phase extraction (SPE) cartridges, (Vac 3cc, 200 mg, Phenomenex strata C18-E, Torrance, CA) were preconditioned with 3 mL of methanol, followed by 3 mL of water before use. The supernatants were loaded onto the SPE cartridges and washed with 2 mL of water. The SPE cartridges were then eluted with 4 mL of methanol, and the eluents were centrifuged at 10000 rpm for 10 min. Supernatant aliquots of 2 μL were injected into the UPLC/Q-TOF-MS system for analyses.
Ingredient herbal medicine samples (G. jasminoides, 2 g; L. japonica, 2 g; A. annua, 2 g) were immersed in 20 mL of deionized water for 1 h. The solutions were then decocted by boiling 3 times (1 h each time). The extracts were diluted to generate 20 mg/mL solutions. All samples were filtered through a 0.22 μm filter membrane before UPLC-MS analyses.

UPLC-Q/TOF-MS analyses
UPLC analyses were performed using an ACQUITY UPLC system equipped with a binary solvent system, an automatic sample manger and photodiode array (PDA) detector. The chromatographic separation was performed on an Acquity UPLC BEH C18 Column (3.0 mm × 150 mm, 1.7 μm, waters, Ireland) at a temperature of 40°C. The mobile phases consisted of eluent A (0.1% formic acid in water, v/v) and eluent B (0.1% formic acid in acetonitrile, v/v). These eluents were delivered at a flow rate of 0.4 mL/min with a linear gradient program as follows: 2-5% B from 0 to 5.0 min, 5-12% B from 5.0 to 10.0 min, 12-30% B from 10.0 to 15.0 min, 30-55% B from 15.0 to 19.0 min and 55-100% B from 19.0 to 20.0 min. After maintaining 100% B for 3 min, the column was returned to its initial condition.
The UPLC system was coupled to a hybrid quadrupole, orthogonal time-of-flight (Q-TOF) tandem mass spectrometer (SYNAPT G2 HDMS, Waters, Manchester, U.K.) equipped with ESI. The operating parameters were as follows: capillary voltage of 3 kV (ESI+) or -2.5 kV (ESI-), sample cone voltage of 35 V, extraction cone voltage of 4 V, source temperature of 100°C, desolvation temperature of 300°C, cone gas flow of 50 L/h and desolvation gas flow of 800 L/h. In MS E mode, the trap collision energy for the low-energy function was set at 5 eV, while the ramp trap collision energy for the high-energy function was set at 20-50 eV. Argon was used as the collision gas for collision-induced dissociation (CID) in MS E and MS 2 modes. To ensure mass accuracy and reproducibility, the mass spectrometer was calibrated over a range of 50-1500 Da using a solution of sodium formate. Leucine-enkephalin (m/z 556.2771 in positive ion mode; m/z 554.2615 in negative ion mode) was used as an external reference for the LockSpray and was infused at a constant flow of 5 μL/min. The data were centroided during acquisition.

Optimization of UPLC and mass spectrometry conditions
The MS E acquisition mode required well-resolved peaks to ensure that the predominant fragments were collected from a single precursor ion. Obtaining a desirable chromatographic profile with satisfactory separation and peak shapes without excessive peak tailing was necessary. Different UPLC conditions that included both mobile phase systems (methanol-aqueous and acetonitrile-aqueous) were tested. When the mobile phase was acetonitrile-aqueous, the separation resolution was greatly improved compared to methanol-aqueous. Addition of 0.1% formic acid to the mobile phase reduced peak tailing and enhanced the resolution. Thus, an acetonitrile-aqueous solution with 0.1% formic acid was selected as the mobile phase. In addition, four analytical columns, including the Acquity BEH C18 column (3.0 mm × 150 mm, 1.7 μm), Acquity BEH C18 column (2.1 mm × 50 mm, 1.7 μm), Acquity HSS T3 column (2.1 mm × 50 mm, 1.8 μm) and Acquity Shield PR18 column (2.1 mm × 50 mm, 1.7 μm) were compared to achieve better separation performance. Unfortunately, the results of three different 50 mm columns were not satisfactory as shown in (S1 Fig). Then, we found that the column length could influence the separation efficiency significantly. Thus, the Acquity BEH C18 column (3.0 mm × 150 mm, 1.7 μm) was chosen for analysis in current condition. For mass spectrometry, both positive and negative ion modes were tested, and each target compound type was analyzed in a suitable ESI mode.

Establishment of the supporting database
A systematic investigation of the chemical constituents in RDN was conducted. A self-built database of compounds that were isolated from three medicinal herb ingredients of RDN was established by retrieving on-line databases or Internet search engines, such as Chemical Abstracts Service (CAS) database, Massbank, Web of Science and ChemSpider. The emphasis was placed on analyzing structural characteristics and MS fragmentation behaviors, especially for diagnostic ions (characteristic fragments). As a result, 259 constituents, including iridoids, organic acids, flavonoids, sesquiterpenes, lignans and coumarins were collected. Five items, compound name, molecular formula, accurate mass, diagnostic fragment ions or neutral losses and UV absorption, were recorded.

Diagnostic ion screening using the optimized MS E method
All chemical constituents in herbal medicine can be categorized into different families based on structural types. Thus, a certain family of compounds with identical carbon skeletons could produce similar characteristic fragment ions under CID conditions in mass spectrometry.
Accordingly, the core idea of our approach is to use the diagnostic ions as markers for target compound detection. To simultaneously generate both precursor and fragment ions using the MS E method, low-and high-energy scan functions were switched rapidly and continuously for data acquisition. The high-energy scan function that is used to collect information on fragment ions is generally equivalent to a non-selective MS/MS scan. With such a function, specific diagnostic ions of diverse compounds contained in TCMPs and their precursor ions and neutral losses were simultaneously collected, providing large quantities of valuable information regarding the structural identification of chemical constituents.
In this study, the screening process of caffeoylquinic acids was considered to describe the approach in detail. Based on the aforementioned self-built database, fragment ions at m/z 191.0556 and m/z 179.0340, which can be produced from caffeoylquinic acids as common substructures, were selected as diagnostic ions for detecting other analogues. As shown in (Fig 1), the peaks that appeared in the EIC of the high-energy function of the MS E mode were considered as target compounds and further characterized by accurate mass measurements, MS fragmentation analyses and reference standards. Interestingly, an unexpected compound (labeled as 35) that possessed a novel structure with a rare caffeoylquinic ester acylated at the C-10 position of geniposide was similarly screened out. By comparison, only a few peaks could be detected in the EIC mode of the low-energy MS E function. A wide range of ramped CE (20-50 eV in the present study) in the high-energy MS E function will help reveal the MS fragmentation behaviors of different compounds simultaneously. Similarly, other types of analogues were rapidly screened out by our proposed approach, such as iridoids (S2

Identification of chemical constituents in RDN
A total of 90 compounds, including 45 iridoids, 21 organic acids, nine flavonoids, seven lignans, four sesquiterpenes, three coumarins and one monoterpene were identified or tentatively characterized in RDN (Table 1; S4 Fig). The herb sources of these compounds were confirmed by comparing the base peak chromatograms of RDN to a single herbal extract. The main active constituents of RDN (i.e., caffeoylquinic acids and iridoids) were rapidly screened out by UPLC-ESI/Q-TOF mass spectrometry through diagnostic ion screening with MS E . The  remaining compounds were identified according to their accurate mass measurements within 5 ppm error, tandem MS behaviors, database-matching and reference standards. Both negative and positive ion modes were examined, and the base peak intensity (BPI) profiles of RDN and three ingredient herbs are shown in (Fig 2, S5 and S6 Figs). Recently, ultra performance liquid chromatography coupled with quadrupole time-of-flight mass spectrometry (UPLC-Q-TOF-MS) has been widely used to characterize chemical profiling of herbal medicines and TCMPs. This method has become one of the most frequently applied approaches in the area of fast chromatographic separations [17][18][19]. High-resolution tandem mass spectrometry can provide a more specific and accurate mass as long as the co-eluting compounds possess different m/z values [20]. And the isotopic abundances and the elemental composition of fragment ions are greatly conducive to the structural elucidation of unknown compounds. However, it should be pointed out that identification of chemical components from complex  TCMPs relying solely on mass spectrometry-based approaches was insufficient. As the spectral differences for some isomers are very small and they cannot be differentiated and unambiguously identified. Therefore, in present study, some reference standards isolated from the entitled injection were used to validate the elucidation of those isomers. Thus, it provides the enhanced accuracy and reliability of MS quantitative results.

Iridoids
Iridoids are the main constituents of RDN. This category of compounds was primarily derived from L. japonica and G. jasminoides. In this study, 45 iridoids were identified in positive ion mode. Four of the iridoids were new compounds, which were previously isolated and identified using NMR [21]. The diagnostic fragment ions of these compounds were previously reported [22][23][24]. .1643 (C 21 H 28 O 12 Na), respectively. All of these peaks were produced by the loss of a C 11 H 12 O 4 fragment (Fig 3). This fragmentation pathway is different from that of iridoid glycosides in which the C6-C3 unit is not substituted onto the C-6 position of the glucose unit. We presumed that the C6-C3 unit, an electron-donating group, might have led to this phenomenon. This mechanism should be further investigated.
Compound 44 showed a [M+H] + ion at m/z 506.2032 with an elemental composition of C 25 H 32 NO 10   identified as L-phenylalaninosecologanin B (Fig 4), which was further confirmed by comparison to a reference standard.
Compounds 37 and 41 gave the same molecular formula of C 34 H 46 O 19 from their precursor [M+Na] + ions at m/z 781.2515 and 781.2533, respectively. In (Fig 5a and 5b) Fig 6). Thus, compound 35 was identified as jasmigeniposide A, which was a new compound isolated from RDN. This result was further confirmed through reference standard comparison.
In addition to the above compounds, 37 iridoid glycosides (compounds 1-34, 36, 38 and 43) were identified or tentatively characterized from RDN (Table 1) based on their molecular weights and the tandem fragmentation patterns.

Organic acids
According to previous research, caffeoylquinic acids as the main bioactive components in RDN were found in L. japonica Thunb., G. jasminoides Ellis and A. annua L. The structures of these typical constituents generally consist of a quinic acid moiety and mono-or dicaffeic acids that are linked to the 3-OH and/or 4-OH and/or 5-OH [22]. These compounds exhibit common proposed fragmentation pathways and diagnostic fragmentation ions, such as m/z 353, 191, 179, 173, 135, etc. The differences in the diagnostic fragmentation ion intensity could be used to identify their structures. As shown in (Fig 1), 15 peaks, including 14 caffeoylquinic acids and one caffeoylquinic substituted new iridoid glycoside, appeared in EIC mode and were considered as target compounds by extracting diagnostic ions 191.0556 and 179.0340 in the high-energy MS E function. Six peaks were easily located in the chromatogram of RDN by extracting m/z 353.0873. Similarly, three parent ions at m/z 515.1190 were located. By comparison with accurate retention times, the first three ions were assigned as monocaffeic acids, while the latter three were identified as dicaffeic acids (Fig 7). According to the literature [25][26][27], the linkage position of the caffeoyl groups on quinic acid could be determined according to its MS 2 fragmentation behavior. Briefly, when the caffeoyl group was linked to 3-OH or 5-OH, the [quinic acid-H]ion at m/z 191 was the base peak, and the [caffeic acid-H]ion at m/z 179 was more significant for 3-O-caffeoylquinic acids. The [quinic acid-H 2 O-H]ion at m/z 173 was the prominent peak when the caffeoyl group was linked to 4-OH. In our experiment, this fragmentation behavior was also observed in the negative mode MS E spectra. Thus, compounds 47, 49 and 50 were identified as 5-O-caffeoylquinic acid, 3-O-caffeoylquinic acid and 4-O-caffeoylquinic acid, respectively. Similarly, compounds 51, 53 and 55 were identified as 5-O-caffeoylquinic methyl ester, 3-O-caffeoylquinic methyl ester and 4-O-caffeoylquinic methyl ester, respectively.
Compound 61 had a base peak ion at m/z 191.0561 and a secondary peak at m/z 179.0349. As reviewed above, 61 could be identified as a 3-substituted quinic acid. Therefore, peak 18 was identified as 3,5-di-O-caffeoylquinic acid, which was further confirmed by comparison to a reference standard. Compounds 60 and 62 both produced a base peak at m/z 173, indicating that they were both 4-substituted quinic acids. According to literature [28], the retention time of 3,4-di-O-caffeoylquinic acid is shorter than that of 4,5-di-O-caffeoylquinic acid, and thus, the compounds were identified as 3,4-di-O-caffeoylquinic acid and 4,5-di-O-caffeoylquinic acid, respectively. These retention times were consistent with those of the separate compounds. In addition, compounds 63, 64 and 66 were identified as 3,4-di-O-caffeoylquinic methyl ester, 3,5-di-O-caffeoylquinic methyl ester and 4,5-di-O-caffeoylquinic methyl ester, respectively.
Compound 46 was identified as quinic acid by comparison with a standard compound. Compounds 54, 57 and 59 were tentatively assigned as 3-hydroxy-4-methoxy styrene acrylic acid, 3-O-caffeoylshikimic acid and syringaldehyde, respectively, by matching their accurate molecular weights with those in a chemical database. These assignments were further corroborated by comparison with standard substances. Compounds 48, 52, 56, 58 and 66 were identified by comparison with isolated compounds from RDN (as listed in Table 1).

Flavonoids
The MS/MS behaviors of flavonoids and their glycosides have been extensively described [22,[29][30][31]. Briefly, the primary MS/MS behavior of aglycones was described by the RDA fragmentation pathway. Successive loss of CO from the ketone group, C-fragmentation and loss of radicals, such as CH 3 and CHO, have been described. For flavonoid glycosides, the glycosidic bond is easily cleaved in positive ion mode, and the neutral loss of 162 Da is the characteristic fragment ion of flavonoid O-glycosides. The fragment ion at [M+H-308] + corresponds to the loss of a rutinose unit.
As shown in Table 1, a total of nine flavonoids were screened from RDN, four of which were unambiguously identified as rutin (67)

Conclusion
In this work, an approach involving UPLC-ESI-Q/TOF-MS coupled with MS E data acquisition was developed to profile multiple chemical constituents in RDN. Diagnostic ions were used as invaluable markers for the screening of target compounds. A total of 53 compounds, including two new iridoids, were identified or tentatively characterized using this method. Due to the structural complexity of the chemical constituent types in TCMPs, the present analytical approach still has a limitation in the detection of low-abundance components. The remaining 37 compounds were identified according to their accurate mass measurements within 5 ppm error, tandem MS behaviors, database-matching and reference standards. The RDN herbal sources were unambiguously confirmed by comparing the extracted ion chromatograms (EICs) of RDN and ingredient herbal extracts. The results of our study not only provide a certain foundation for further studies of RDN but also demonstrate chemical profile analyses of TCMPs via UPLC-ESI-Q/TOF-MS and diagnostic ion screening using MS E .