Integration of a hybrid scan approach and in-house high-resolution MS² spectral database for charactering the multicomponents of Xuebijing Injection

2.1 Chemicals and reagents

A total of 56 reference compounds (Fig. 2 and Table S1), involving 21 organic acids (1–21), four phenanthrenequinones (22–25), 21 flavonoids (26–46), three terpenoids (47–49), five phthalides (50–54), and two nucleosides (55, 56), were purchased from Shanghai Standard Biotech. Co., ltd. (Shanghai, China) and Desite Biotechnology Co., ltd. (Chengdu, China). HPLC-grade acetonitrile, methanol (Fisher, Fair Lawn, NJ, USA), formic acid (Sigma-Aldrich, MO, Switzerland), and ultra-pure water in-house prepared using a Milli-Q Integral 5 water purification system (Millipore, Bedford, MA, USA), were used. The five TCMs in XBJ were purchased from Beijing Tongrentang Pharmacy (CF, CR, SMRR, ASR) and Anhui Shimao Co., ltd. (PRR). XBJ Injection (batch number: 1911291), provided by Tianjin Hongri Pharmaceutical Co., ltd., was analyzed in this work.

Close

Fig. 2

Chemical structures for the 56 reference compounds used in the current work.

Export to PPT

2.2 Preparation of the test solutions

XBJ Injection, filtered through a 0.22-µm microporous membrane, was used as the test solution. For preparing the samples of five component drugs (CF, PRR, CR, SMRR, and ASR), each powder (500 mg) was dissolved in 10 mL 20 % methanol (v/v). After being vortexed for 2 min, each sample was extracted with the aid of ultrasound (power: 400 W; frequency: 40 kHz) at 40 °C for 30 min. After centrifugation at 11,481 g (14,000 rpm) for 10 min, the supernatant was taken and further filtered through a 0.22-μm microporous membrane, yielding the final test solution (equivalent to 50 mg/mL of the drug material).

2.3 Uhplc/Im-Qtof-Ms

The ACQUITY UPLC I-Class/ Vion™ IM-QTOF system (Waters, Milford, MA, USA) was used to collect the MS data of reference compounds, component drugs, and the XBJ sample. The Zorbax Eclipse Plus C18 column (2.1 × 100 mm, 1.8 µm) maintained at 40 °C was selected for chromatographic separation. A binary mobile phase, containing 0.1 % formic acid in water (A) and acetonitrile (B), ran according to an optimal gradient program as follows: 0–3 min, 1–5 % (B); 3–4 min, 5 %–15 % (B); 4–7 min, 15 %–15 % (B); 7–12 min, 15 %–20 % (B); 12–16 min, 20 %–21 % (B); 16–17 min, 21 %–55 % (B); 17–20 min, 55 %–70 % (B); 20–22 min, 70 %–80 % (B); 22–24 min, 80 %–95 % (B). A flow rate of 0.3 mL/min was set, and the injection volume was 5 µL.

The Vion IM-QTOF mass spectrometer was used to acquire high-accuracy MS data in both the positive and negative ion modes by the hybrid HDMS^E-HDDDA scan method. Parameters for the LockSpray ion source were as follows: capillary voltage, +1.0 kV (ESI + )/–1.0 kV (ESI–); cone voltage: 60 V (ESI + )/80 V (ESI–); source offset, 80 V; source temperature, 120 °C; desolvation gas temperature (N₂), 500 °C; desolvation gas flow (N₂), 800 L/h; and cone gas flow (N₂), 50 L/h. Default parameters for the traveling wave IM separation were defined, and the calibration of CCS was conducted according to the manufacturer's guideline using a mixture of calibrants (Paglia et al., 2015). The QTOF analyzer scanned over a mass range of m/z 50–1500 at a low collision energy of 6 eV in both two ESI modes by HDMS^E at 0.3 s per scan (MS¹) and HDDDA at 0.15 s per scan (MS¹). The ramp collision energy (RCE) of 20–70 eV (ESI + )/30–70 eV (ESI–) was set in HDMS^E. MS/MS fragmentation of three most intense precursors was automatedly triggered by HDDDA when the intensity exceeded 200 detector counts under mass-dependent ramp collision energy (MDRCE) of 20–40 eV (ESI + )/20–50 eV (ESI–) in low mass ramp settings and 30–50 eV (ESI + )/30–80 eV (ESI–) in high mass ramp settings. The MS/MS acquisition stopped until time exceeded 0.5 s. MS data calibration was conducted by constantly infusing the leucine enkephalin solution (Sigma-Aldrich, St. Louis, MO, USA; 200 ng/mL) at a flow rate of 10 µL/min.

2.4 Establishment of the high-resolution MS² spectral database for XBJ

The MS² spectral library for XBJ was in-house established by two steps: 1) creation of the literature-based compounds information library for each of five component drugs (CF/PRR/CR/SMRR/ASR); 2) use of the UNIFI workflows to characterize the contained compounds, and establishment of the MS² spectral database based on the characterized ones (Fig. 1). Elaboration of the component drug databases could be illustrated using CF as a case: 1) the literature on CF phytochemistry was retrieved from Web of Science (https://www.webofknowledge.com), CNKI ( www.cnki.net), ChemSpider (https://www.chemspider.com), PubChem ( http://pubchem.ncbi.nlm.nih), Chemicalbook (https://www.chemicalbook.com), and other databases, and those reported compounds were summarized (with NMR data) and classified according to their structure types; 2) by drawing the chemical structure for each known component using ChemDraw, each structure was saved in the ″.mol″ format; 3) after importing the structural data into the UNIFI platform, the CF in-house library was established. The spectral library for five component drugs finally constituted the in-house MS² spectral library for XBJ.

The LC-MS conditions for data acquisition of five component drugs and the mixed reference standard solutions were the same as those described in 2.3. UHPLC/IM-QTOF-MS, but the MS scan method was HDMS^E. The resultant HDMS^E data (ESI + and ESI–) were imported into UNIFI for automatic data processing, by which the t_R, CCS, and MS² fragments of each compound could be obtained (major parameters were detailed in the following). The next step of characterization results checking could finally obtain the in-house MS² spectral database for five component drugs, which could be searched by UNIFI to automatedly identify the multicomponents from XBJ.

The UNIFI 1.9.3.0 (Waters) performed the data correction, peak picking, and peak annotation, to process the uncorrected HDMS^E of five component drugs. In the parameter of ″Find 4D Peaks″, high-energy intensity threshold at 100.0 counts and low-energy intensity threshold at 200.0 counts were set. For the setting of ″Target by mass″, target match tolerance at 10.0 ppm was used, and the functions of ″Screen on all isotopes in a candidate″, ″Generate predicted fragments from structure″, and ″Look for in-source fragments″, were enabled. The fragment match tolerance was 10.0 mDa. Adducts: positive adducts including + H, +Na, and + K; Lock mass: combine width, 3 scans; mass window, 0.5 m/z; reference mass, 556.2766; reference charge, +1; negative adducts including + HCOO and –H; Lock Mass: combine width, 3 scans; mass window, 0.5 m/z; reference mass, 554.2620; reference charge, –1.

2.5 Automated annotation of the HDMS^E & HDDDA data by searching the in-house MS² spectral library achieving the multicomponent characterization of XBJ

By using the UNIFI platform and automatic searching of the in-house MS² spectral library, the HDMS^E and HDDDA data (in both the ESI + and ESI– modes), obtained using the established hybrid scan strategy, were efficiently processed enabling the multicomponent characterization of XBJ. For processing the HDDDA data, the established MS² spectral library of XBJ in the positive and negative ion mode were called in the ″expected component list″. A list of the ″identified components″ was obtained involving diverse information (e.g., detected m/z, formula, mass error and adducts, expected RT (referring to the value recorded in the MS² spectral library), observed RT, observed drift, expected CCS, observed CCS, and CCS Delta, etc.). In addition to the ″identified components″ list, the compounds in the ″unknown components″ list were further characterized by comparison with the literature and MS² data analysis, aimed to identify more potential targeted compounds. Processing of the HDMS^E data of XBJ was consistent with the approach depicted for the HDDDA data. The parameters set for ″Find 4D Peaks″, ″Target by mass″, ″Adducts″, and ″Lock Mass″, were the same as those set in processing the HDMS^E data of five component drugs. The additional parameters included: Find DDA masses (in HDDDA): MS ion intensity threshold, 200.0 counts; MSMS ion intensity threshold, 100.0 counts. Target by retention time: absolute retention time identification tolerance, 0.1 min. Target by collision cross section: collision cross section tolerance, 2.0 %.

3.1 Optimization of a reversed-phase UHPLC/IM-QTOF-MS approach targeting the multiple classes of components from XBJ Injection

We were aimed to develop a reversed-phase UHPLC/IM-QTOF-MS approach to resolve more minor components from XBJ Injection with the enhanced sensitivity in detection. Given the rather complex chemical composition of XBJ (Sun et al., 2017), key parameters affecting both the UHPLC separation and Vion IM-QTOF mass spectrometer were optimized in sequence.

Selectivity of the stationary phase can determine the resolution in the UHPLC separation. Ten reversed-phase C18 column (Table S2) with different silica gel cores and bonding technologies (purchased from Agilent, Waters, and Phenomenex) were compared, by comprehensively evaluating the resolution and the number of resolved peaks (according to the positive-mode MS^E data processed by UNIFI; Fig. S1). As a result, Zorbax Eclipse Plus C18 (the number of resolvable peaks: 2601), Zorbax SB-C18 (3348), Zorbax Extend C18 (3382), Zorbax SB-Aq (2945), and BEH C18 (3136), could resolve more peaks than the other five columns from XBJ. Moreover, Zorbax Eclipse Plus C18 enabled relatively good response for the major components in the BPI chromatogram, and better peak shape than the others. What’s more, filled with high-performance particulate C18, Zorbax Eclipse Plus C18 is advantageous for the analysis of the acidic and neutral samples, and especially the basic components that show poor peak shape on the other columns (pH tolerance: 2–9). The column temperature varying between 25 and 40 °C was examined to test its influence on the resolution of XBJ components by using the selected Zorbax Eclipse Plus C18 column (Fig. S2). Comparatively, higher level of temperature could lead to better resolution in whole, and at 40 °C (8006) the largest amount of peaks were resolved. We further slightly adjusted the gradient elution program, and the satisfactory performance in chromatography was enabled, ultimately.

To enhance the sensitivity of detection and produce rich fragments beneficial to characterizing more minor components from XBJ Injection, two key ion source parameters (capillary voltage and cone voltage) together with the collision energy (RCE for HDMS^E and MDRCE for HDDDA), on the Vion^TM IM-QTOF mass spectrometer, were optimized by comparing the ion response (integrated peak area through three injections in parallel) of the reference compounds and the abundance of MS² fragments in both the positive and negative modes. Influences by the variation of capillary voltage (0.5–3.0 kV) were tested using HSYA, rutin, 9-methyl lithospermate B, oxypaeoniflorin, and benzoylpaeoniflorin in the negative ion mode, while kaempferol, quercetin, benzoyloxypaeoniflorin, salvianolic acid C, and tanshinone I in the positive mode, as the index components (Fig. S1). It could be found that the peak areas of HSYA, rutin, and oxypaeoniflorin (ESI–) gradually decreased when increasing the voltage from 0.5 to 3.0 kV, but the response of 9-methyllithospermate B and benzoylpaeoniflorin showed no significant alternation. Moreover, the reproducibility of three parallel analyses at 0.5 kV was worse than those at 1.0 kV, and accordingly, the capillary voltage of 1.0 kV in ESI– was set. In the positive mode, most of the index compounds (except salvianolic acid C) showed the downward trend with the increase of voltage value, and thus the same level of capillary voltage at 1.0 kV was selected in the positive mode. In the case of cone voltage (20–120 V), five index components showed inconsistent variation trends (the best ion response obtained at 80 V for HSYA and oxypaeoniflorin, 100 V for rutin, 20 V for 9-methyl lithospermate B, and 40 V for benzoylpaeoniflorin) in the negative mode, and we finally selected 80 V. Differently, in the positive mode, benzoyloxypaeoniflorin and tanshinone I showed the best ion response at 60 V, while kaempferol, quercetin, and salvianolic acid C, gave the strongest signals at 80 V, 100 V, and 20 V, respectively. The cone voltage at 60 V in the positive ion mode was selected. The collision energy was optimized by observing the fragmentation degree of the index compounds. For RCE in HDMS^E (Fig. S3), four different levels (including 20–60 eV, 30–70 eV, 40–80 eV, and 60–100 eV) were examined in the negative mode, taking HSYA, rutin, oxypaeoniflorin, and salvianolic acid C, as the index compounds. We could infer that the higher level of collision energy could induce more fragmentation, and the condition of 30–70 eV was considered as satisfactory, under which most of these index compounds generated the rich and diversified fragments useful for the structural elucidation. Similarly, among the four levels (including 20–70 eV, 30–90 eV, 40–80 eV, and 60–100 eV) examined in the positive mode using kaempferol, quercetin, isocryptotanshione II, and adenosine as the index compounds, 20–70 eV was a good choice. Mass-dependent RCE (MDRCE) in HDDDA can apply customized collision energy ramp on the selected precursors by separately setting the collision energy ramp at two ends (low mass and high mass) of the predefined scan range (Wang et al., 2022). For MDRCE in ESI– (Fig. S4), four different levels, involving low mass: 10–30 eV/high mass: 20–50 eV, 20–50 eV/30–80 eV, 30–50 eV/40–80 eV and 30–60 eV/40–90 eV, had been compared, while 10–30 eV/20–40 eV, 10–40 eV/20–60 eV, 20–40 eV/30–50 eV, and 30–50 eV/40–60 eV, were examined in the positive mode. Taking into account of the richness of the product ions for most index compounds, the MDRCE of 20–50 eV/30–80 eV in the negative ion mode and 20–40 eV/30–50 eV in the positive mode were selected.

The Vion IM-QTOF LC-MS platform, utilized in this work, can enable versatile high-definition MS and MS² scan methods because of the IM separation of all precursor ions (according to the charge state, size, and shape), among which HDMS^E-MS/MS represents a hybrid scan approach facilitating the alternate DIA and DDA in once injection analysis (Qian et al., 2022; Wang et al., 2022). The IM separation was found powerful in separating the co-eluted and isomeric components. For instance, the co-eluting components at 4.97, 6.61, 10.37, and 17.74 min, were separated into two independent peaks with different CCS values (Fig. S5A). Taking the co-eluting compounds at 4.97 min as the example, the two compounds showed the deprotonated precursors ([M−H]^–) at m/z 771.1970 (compound 46#, C₃₃H₄₀O₂₁) and m/z 787.1942 (compound 47#, C₃₃H₄₀O₂₂). For 46#, the product ions at m/z 609.1461 ([M−H−Rha]^–), 463.0873 ([M−H−Rha−Glc]^–), and 301.0342 ([M−H−Rha−2Glc]^–), were observed, which, by searching the in–house library, was characterized as 6-hydroxykaempferol-3-O-rutinoside-6-O-glucoside or its isomer. For 47#, with the elimination of attached sugars, the fragments at m/z 625.1407 ([M−H−Glc]^–) and 463.0874 ([M−H−2Glc]^–) were found. By the aid of reference compound comparison, this compound was identified as quercetin-3-O-β-d-glucose-7-O-β-d-gentiobioside. The isomers at m/z 313.07, 577.15, 611.16, 787.19, and 1043.26 were separated by IM, generating two remarkable peaks for each (Fig. S5B). Taking the isomers at m/z 313.07 as an example, based on the molecular formula (C₁₇H₁₄O₆) searching in the in–house library, salvianolic acid F was a possible hit. The MS² fragments for compound A1 were detected at m/z 269.0806 ([M−H−CO₂]^–) and 255.0662 ([M−H−C₂O₂H₂]^–), which could be consistent with the structure of salvianolic acid F. Compound A2 was thus deduced as an isomer of salvianolic acid F (Fig. S5C).

3.2 Establishment of an in-house high-resolution MS² spectral database and its superiority in rapid characterization of the multicomponents from XBJ Injection

Given the fact that specific MS database useful for the multicomponent characterization of XBJ was rare, in this work, we developed an in-house high-resolution MS² spectral database, and applied it to characterize the multiple classes of components from XBJ Injection based on the CID-MS² (collision-induced dissociation) data obtained using the established HDMS^E-HDDDA hybrid scan approach. The established MS² spectral library for XBJ consisted of 293 components of Carthami Flos (ESI + 96, ESI– 197), 182 ones of Paeoniae Radix Rubra (ESI + 62, ESI– 120), 216 of Chuanxiong Rhizoma (ESI + 100, ESI– 116), 257 of Salviae Miltiorrhizae Radix et Rhizoma (ESI + 111, ESI– 146), and 141 of Angelicae Sinensis Radix (ESI + 68, ESI– 73).

The in-house MS² spectral library, as a spectral library, recorded diverse structure-related information, and could greatly improve the accuracy of the identification results. This spectral library had recorded the t_R, CCS, and the MS² spectrum for each component (Fig. 3A). The identification results obtained by invoking the in-house MS² spectral library presented not only the seven major items of information inherent in UNIFI for the MS¹ library, but also the additional information like the expected RT, expected CCS, and CCS delta (ΔCCS) (Fig. 3B). Because of the absolute t_R identification tolerance (<0.1 min) and ΔCCS (<2%) filtering and MS² fragments comparison, the accuracy in the identification results was improved, compared with the application of the traditional MS¹ library. Firstly, it enabled more accurate identification, taking compound 54# (t_R 5.15 min, CCS: 238.59 Å²) as an example. Its MS² spectra gave rich product ions involving m/z 611.1611, 491.1186, 325.0705, 119.0494, and the others. By searching the in-house MS² spectral library, it was identified as HSYA, which could be confirmed by comparison with the reference standard (t_R, CCS, MS¹, and MS²). However, by applying the MS¹ library (recording 814 known compounds ever reported for the five component drugs of XBJ), compound 54# was incorrectly identified as (2S)-4′,5-dihydroxyl-6,7-di-O-β-d-glucopyranosyl flavanone. Secondly, improved characterization was enabled for the isomers. For instance, an isomer of HSYA (t_R 6.47 min) could be identified based on the characteristic DPIs similar to HSYA, even if the signal interference was remarkable (possibly dissociated from the co-eluted isomers) (Fig. 3C). Those failing to give the DPIs of HSYA were not characterized as the HSYA isomer, such as a component eluted at 7.66 min. What’s more, the in-house MS² spectral library could differentiate the isomers, such as between compounds 144# (m/z 609.1457, t_R 8.56 min, CCS 233.03 Å²) and 159# (m/z 609.1469, t_R 10.01 min, CCS 234.10 Å²). Because of the filtering by both t_R and CCS facilitated by the in-house MS² spectral library, compound 144# (ΔCCS: 1.51 %) was identified as rutin, and compound 159# (ΔCCS: 1.33 %) as an isomer of rutin. Similar cases could include compounds 216#/228# (salvianolic acid B/isomer) and 219#/226# (benzoyloxypaeoniflorin/isomer).

Close

Fig. 3

Illustration for the advantages of the in-house MS² spectral library. A-Detailed information involved in the in-house MS² spectral library of UNIFI using HSYA as the case; B-comparison of the processing results obtained between the in-house MS² spectral library and the MS¹ library; C-comparison of the identification of HSYA isomers by searching the in-house MS² spectral library (a) and the MS¹ library (b); D-a radar chart showing the differentiated performance between the in-house MS² spectral library and the MS¹ library (IR: Identification Rate; AR: Accuracy Rate; II: Identification of Isomers; IQ: Information Quantity; MD: Matching Degree).

Export to PPT

Using the characterization of the Carthami Flos (CF) components as the case, we further compared the identification results for XBJ by the in-house established MS² spectral library and the MS¹ library. The results could be interpreted from the following five aspects: Identification Rate (IR), the ratio of the number for finally confirmed identifications to a constant; Accuracy Rate (AR), the ratio of the number of finally confirmed identifications to the number of primary ″Identified Components″; Identification of Isomers (II), the ratio of the number of isomers to a constant; Information Quantity (IQ), the ratio of the information quantity (items) provided by two different libraries to a constant; Matching Degree for the known compounds of XBJ (MD), the ratio of the total finally confirmed identifications to the number of components the MS¹ library of CF contained. As shown in Fig. 3D by a radar chart, the in-house MS² spectral library showed superiority over the MS¹ library in all five aspects, especially in AR and II. Therefore, the established UNIFI workflows by searching the in-house MS² spectral library could demonstrate a potent tool for the multicomponent characterization of herbal medicines, proving the improved efficiency and accuracy, in contrast to the conventional MS¹ library.

3.3 Characterization of the multicomponents from XBJ Injection based on the streamlined workflows facilitated by UNIFI and the retrieval of the in-house MS² spectral database

Comprehensive multicomponent characterization of XBJ Injection was performed by applying the hybrid scan method of HDMS^E-HDDDA for data acquisition and interpreting the resultant MS² data based on the intelligent peak annotation workflows of UNIFI and the retrieval of the established in-house MS² spectral database (Fig. 4). The obtained HDMS^E and HDDDA data of XBJ separately in the negative and the positive modes were efficiently processed (automatic data correction, peak extraction, and peak annotation) by the UNIFI workflows (Qian et al., 2022; Wang et al., 2022; Zhang et al., 2019; Zuo et al., 2019; Zuo et al., 2020), yielding a list of identified components. The primarily characterized components suffered from re-confirmation, while those listed as unknown components were re-analyzed to enrich the identification. As a result, a total of 294 compounds were characterized from XBJ, and 28 thereof were identified by comparison with the reference compounds. Detailed information regarding these characterized components is given in Table S3. Moreover, the typical neutral loss (NL) and diagnostic product ions (DPIs) encountered in the multicomponent characterization of XBJ are summarized in Fig. 4. The known NL forms included 162.05 Da for a hexose (Glc), 146.06 Da for Rha, 30.02 Da for CH₂O, 43.99 Da for CO₂, 27.99 Da for CO, and 18.01 Da for H₂O, while the characteristic DPI, such as m/z 119.05 (C₈H₇O₅) for characterizing a quinochalcone (Si et al., 2016), m/z 271.02 (C₁₄H₇O₆) and 301.03 (C₁₅H₉O₇) representative of the flavonoid aglycones (Lu et al., 2019a). In this work, we used the typical characterization cases of flavonoids, terpenoids, organic acids, phenylpropanoids, and phthalides, to illustrate the comprehensive multicomponent characterization of XBJ Injection.

Close

Fig. 4

The overall compound characterization workflows based on the UNIFI informatics platform and the in-house MS² spectral library, showing the typical neutral loss (NL) and diagnostic product ions (DPI) useful for the rapid structural elucidation of the components from XBJ Injection.

Export to PPT

3.3.1

3.3.1 Characterization of flavonoids

Flavonoids, including the quinochalcones and flavonoid O-glycosides, are one class of main bioactive ingredients for XBJ, which mostly originated from the component drug, Carthami Flos (Li et al., 2021a). A total of 81 flavonoids were identified from XBJ Injection (accounting for 27.6 % of the total amount of the characterized components), and eleven thereof (47#/54#/63#/144#/151#/155#/165#/183#/227#/248#/251#) got confirmatively identified with the aid of reference compounds comparison. Compound 54# (t_R 5.15 min), giving the deprotonated precursor ion ([M−H]^–) at m/z 611.1611 (C₂₇H₃₂O₁₆), was unambiguously identified as HSYA by comparison with the reference standard. In its MS/MS spectrum, the fragment ion at m/z 491.1186 ([M−H−C₄H₈O₄]^–, C₂₃H₂₃O₁₂) could result from the ^0,2X cleavage of the C-glycoside moiety (Fig. 5). It could be further fragmented into the product ion of m/z 325.0705 (C₁₈H₁₃O₆) by the combined cleavage of CO + H₂O + C₄H₈O₄, and the ion of m/z 283.0596 (C₁₅H₁₁O₅) was obtained because of the additional C₂H₂O eliminated. The DPI for quinochalcone was observed at m/z 119.0494 (C₈H₇O), which corresponded to the remaining moiety of ring B (p-hydroxystyrene). These data were consistent with our previous reports (Lu et al., 2019a; Si et al., 2016). Compound 203# (t_R 13.86 min) showed the [M−H]^– ion at m/z 613.1556, suggesting the molecular formula C₃₀H₃₀O₁₄. Similarly, the combined elimination of H₂O + CO₂ occurring to the 3-(4-hydroxyphenyl) propanoic acid moiety generated the fragment at m/z 551.1553 (C₂₉H₂₇O₁₁). What’s more, some product ions at the medium intensity, such as m/z 361.1073 (C₂₂H₁₇O₅), 241.0495 (C₁₄H₉O₄), and the characteristic ion for quinochalcone at m/z 119.0494 (C₈H₇O), were observed. By searching the literature, these fragments could assist to tentatively characterize compound 203# as isosafflomin C or its isomer (Yang et al., 2016). Compounds 57# (t_R 5.21 min), 120# (t_R 7.22 min), 144# (t_R 8.56 min), and 159# (t_R = 10.01 min), were characterized as the isomeric flavonoid O-glycosides (C₂₇H₃₀O₁₆) based on their high-resolution MS data, amongst them 144# was identified as rutin by comparison with the reference standard. The [M−H]^– ions at m/z 609.15 for compounds 144# and 159# could eliminate all the attached sugars yielding the fragment ion of the free flavone moiety at m/z 301.03 ([M−H−Glc−Rha]^–; C₁₅H₉O₇), which was further fragmented into the ion at m/z 271.02 by losing CH₂O, one characteristic NL form for 3-O-flavonol (Yang et al., 2014). Accordingly, compound 159# was identified as the isomer of rutin.

Close

Fig. 5

Annotation of the negative CID-MS² spectra and proposed fragmentation pathways for the representative compounds of flavonoids, terpenoids, and organic acids from XBJ Injection.

Export to PPT

3.3.4

3.3.4 Characterization of phthalides

Phthalides are the main bioactive compounds for Chuanxiong Rhizoma, which include both the monomers and dimers (Yang et al., 2021). We could characterize 42 phthalide compounds from XBJ Injection, and most of them were in the monomer form. Notably, the hydroxyl phthalides had different numbers of hydroxyl substituents, which could have the typical protonated precursor ions ([M + H]⁺) at m/z 205.09, 207.10, 209.12, 225.11, and 279.16, while H₂O, CO, and CO₂, were the characteristic NL forms. Compound 196# (t_R 12.94 min) showed the [M + H]⁺ precursor ion at m/z 207.1017, and accordingly we could characterize its molecular formula as C₁₂H₁₄O₃. By the MS/MS fragmentation, the characteristic fragment ion at m/z 189.0908 (C₁₂H₁₃O₂) of senkyunolide F could be observed after NL of H₂O. Another product ion at m/z 161.0953 (C₁₁H₁₃O) could be generated by the combined loss of H₂O + CO (Fig. 6). By searching the literature, compound 196# was characterized as senkyunolide F or its isomer (Sun et al., 2017). Compound 277# (t_R 18.57 min) showed the [M + H]⁺ parent ion at m/z 279.1593, consistent with the molecular formula of C₁₆H₂₂O₄. The product ion at m/z 233.1556 (C₁₅H₂₁O₂) could be obtained due to the integrated cleavage of H₂O + CO. Another product ion [M + H–C₄H₈O₂]⁺ at m/z 191.1070 (C₁₂H₁₅O₂) could indicate the existence of the hydroxyl and 1-oxobutyl substituent. By comparing this information with the database, compound 277# was characterized as senkyunolide M or its isomer (Zhang et al., 2017).

Close

Fig. 6

Annotation of the positive CID-MS² spectra and proposed fragmentation pathways for the representative compounds of phthalides and phenanthrenequinones from XBJ Injection.

Export to PPT

3.4 Chemical composition analysis for XBJ Injection

According to the identification results, the chemical composition features of XBJ Injection were analyzed. A total of 294 ingredients were identified or preliminary characterized, including 81 flavonoids (27.6 %), 51 terpenoids (17.3 %), 42 phthalides (14.3 %), 40 organic acids (13.6 %), 13 phenylpropanoids (4.4 %), seven phenanthrenequinones (2.4 %), six alkaloids (2.0 %), and 54 others (18.4 %). Based on the identification results of the in-house MS² spectral library, the components characterized from XBJ could be assigned to their botanical sources. By ranking according to the percentage proportion, they followed this order: CF (31 %), PRR (22 %), CR (20 %), SMRR (18 %), and ASR (9 %), which, in general, conformed to the compatibility rule of ″Monarch, Minister, and Assistant″ in the theory of TCM. In the prescription, CF acts as the monarch drug, PRR and CR are the minister components, and SMRR and ASR serve as the adjuvant drugs (Yu et al., 2022).