Authentication of Shenqi Fuzheng Injection via UPLC-Coupled Ion Mobility—Mass Spectrometry and Chemometrics with Kendrick Mass Defect Filter Data Mining

Abstract

Nearly 5% of the Shenqi Fuzheng Injection’s dry weight comes from the secondary metabolites of Radix codonopsis and Radix astragali. However, the chemical composition of these metabolites is still vague, which hinders the authentication of Shenqi Fuzheng Injection (SFI). Ultra-high performance liquid chromatography with a charged aerosol detector was used to achieve the profiling of these secondary metabolites in SFI in a single chromatogram. The chemical information in the chromatographic profile was characterized by ion mobility and high-resolution mass spectrometry. Polygonal mass defect filtering (PMDF) combined with Kendrick mass defect filtering (KMDF) was performed to screen potential secondary metabolites. A total of 223 secondary metabolites were characterized from the SFI fingerprints, including 58 flavonoids, 71 saponins, 50 alkaloids, 30 polyene and polycynes, and 14 other compounds. Among them, 106 components, mainly flavonoids and saponins, are contributed by Radix astragali, while 54 components, mainly alkaloids and polyene and polycynes, are contributed by Radix codonopsis, with 33 components coming from both herbs. There were 64 components characterized using the KMDF method, which increased the number of characterized components in SFI by 28.70%. This study provides a solid foundation for the authentification of SFIs and the analysis of its chemical composition.

1. Introduction

The unambiguous identification of chemical composition is important for the authentification of natural product drugs, especially herbal injections. The Shenqi Fuzheng injection (SFI) is used to improve the quality of life of patients with lung and stomach cancer, especially with respect to symptoms like fatigue, laziness, and spontaneous perspiration. SFI was approved by the national medical products administration of the People’s Republic of China in 1999 (Drug Approval Number: Z19990065) [1]. SFI is comprised of an equal ratio of two herbal water extracts, Radix codonopsis (Dangshen, Co) and Radix astragali (Huangqi, As) [2]. It has wide clinical application in China and has been on China’s National Reimbursement Drug List (NRDL) since 2004.

Until now, the chimeric composition of SFI has been investigated using several methods. Eighty-one major constituents have been tentatively characterized by ultra-fast liquid chromatography (UFLC) coupled with electrospray ionization quadrupole time of flight mass spectrometry (ESI-QTOF). These constituents include organic acids, amino acids, oligosaccharides, alkaloids, nucleosides, phenylpropanoids, polyacetylenes, flavonoids, isoflavonoids, and saponins [2]. Of these constituents, three saccharides, fructose, glucose, and sucrose, were demonstrated to account for nearly 90% of the SFI’s dry weight (except for the NaCl added), as assayed using the HPLC-ESLD method [3]. Nine of the fifteen analytes, including major polyacetylenes, flavonoids, isoflavonoids, and saponins, were determined [3,4], but only 0.18–0.21% of the SFI could be quantitatively analyzed. There was still nearly 5% of the key chimerical composition of the SFI that could be characterized, mainly attributed to the secondary metabolites of Radix codonopsis (Dangshen, Co) and Radix astragali (Huangqi, As), which seriously hinders the authentification of the SFI.

The chromatographic profile (also called the chromatographic fingerprint) technique has been widely used in the quality control and authentication of herbal medicines [5] and food [6]. For SFI, two chromatographic fingerprints have been used for quality control purposes. One is used to control nucleosides, phenylpropanoids, polyacetylenes, and flavonoids at 245 nm, and the other is used to assay saponins using an evaporative light-scattering detector (ELSD). However, it was difficult and laborious to characterize SFI using two independent methods. The charged aerosol detector (CAD), as a universal and high sensitivity detector, has advantages for fingerprint analysis compared with an ultraviolet detector (UV) and ELSD [7]. The CAD can simultaneously detect analytes with or without ultraviolet absorption. Additionally, due to its universal response for most analytes, the chromatographic profile obtained with CAD can readily reflect the actual contents of all analytes in herbal medicines.

High-resolution mass spectrometry (HRMS) with ion mobility (IM) is a high dimensionality analysis tool that can provide abundant chemical composition information when coupled with liquid chromatography [8,9,10]. It has been successfully used for the deep chemical analysis of the root of Panax notoginseng [11], the stem and hooks of Uncaria rhynchophylla (Miq.) Miq.ex Havil [12], the tuber of Gastrodia elata Bl [9], and the bulbs of Lilium brownii var. viridulum Baker [13]. The collision cross section (CCS) value obtained by ion mobility provides separation with the potential to distinguish isomers for HRMS [9].

To obtain a high content of chemical structural information with HRMS, many data acquisition modes, such as data-dependent acquisition (DDA) and data-independent acquisition (DIA), including data post-processing methods, should be performed [14]. The fragmentation ions obtained by DDA and DIA modes are crucial for the elucidation of structures. This is especially true for DIA mode (MSE mode, all-ion fragmentation, or SWATH), which can provide lossless fragmentation information for all potential analytes in herbal medicines. With online or in-house library matching using software (such as UNIFI and Progenesis QI [15]), the structural information in DIA or DDA data can be quickly screened [16]. In addition, data post-processing methods such as mass defect filtering (MDF) [12] and Kendrick mass defect filtering (KMD) [17] can also be used to screen potential analytes with similar structures.

In this study, a single chromatographic fingerprint was developed to cover all secondary metabolites in the SFI and its intermediates by adopting the UHPLC-CAD approach. Then, combined with IM-QTOF or LTQ-Orbitrap-MS, all peaks in the chromatographic fingerprint were dissected in-depth with mass defect filtering (MDF) and Kendrick mass defect filtering (KMD). Additionally, the CCS values of all tentatively identified ions were also demonstrated for the first time. The resulting in-depth chemical information regarding the SFI may provide useful information for its authentication.

2. Results and Discussion

2.1. Development of Chromatographic Fingerprint of SFI by UHPLC-CAD

Nearly 5% of the dry weight in SFI is still unknown, which hinders its authentication. Most of these components come from the secondary metabolites of Radix codonopsis and Radix astragali. Thus, a solid-phase extraction method with non-polar divinyl benzenebased neutral polymeric sorbent was optimized and developed to enrich those secondary metabolites. The discarded part of SPE consistent of primary saccharides (fructose, glucose, and sucrose, representing approximately 88% of the dry weight in the SFI), amino acids (approximately 2% of the dry weight in the SFI), and nucleosides (unpublished data). The sample preparation method, including the SPE type, loading sample volume, eluent volume of water and methanol, and the flow rate of the eluent and solvent of the sample, were comprehensively optimized (unpublished data).

The enriched part of the SPE column was mainly composed of polyacetylenes, flavonoids, and saponins. It was challenging to profile these components sensitively and simultaneously in a single chromatogram. Due to the prominent advantage of the sensitivity of the CAD as a universal detector, it was adopted to construct the chromatographic fingerprint of secondary metabolites in the SFI (Figure 1A). The chromatographic system was optimized comprehensively, including different columns, mobile system, mobile gradient, flow rate, column temperature, and injection volume. The CAD parameters were also compared one by one. The method validation was evaluated based on precision obtained on different columns, instruments, and labs (data not shown).

To dissect the chemical component information from the chromatographic fingerprint of SFI, the developed UHPLC method was coupled to IM-QTOF-MS and LTQ-Orbitrap Velos to obtain the high-resolution m/z value, CCS value, and MS/MS fragmentation information. The typical base peak chromatograms (BPC) in the positive and negative modes are shown in Figure 1B,C. The corresponding chromatograms of Co_E and As_E are shown in the Supplementary Materials (Figure S2).

2.2. Combining PMDF and KMDF for MS Data Mining of SFI

The chemical composition of natural products is complex and diverse, and comprehensively characterizing the chemical composition of natural products is still a considerable challenge. Due to the high sensitivity, high resolution, and high accuracy of high-resolution mass spectrometry (HRMS), the data collected with HRMS contain massive compound information, including a large amount of background interfering information. In addition, when using HRMS to collect chemical information on natural products, it is more challenging to analyze the structure of chemical components due to the in-source dissociation of compounds and the additive effects on compounds when solvents are added. Mass defect filtering (MDF) is a method of quickly and efficiently screening target components in complex systems based on the law between the precise mass range and the mass loss range of compounds of the same structural type. PMDF, derived from MDF, establishes a polygonal m/z window for potential target components by using the reported compound information to eliminate matrix interference. Using PMDF, target components could be screened more precisely in complex systems. KMDF is based on the theory raised by Edward Kendrick. He proposes that compounds, when sharing the same skeleton type but differing in one or more CH₂ groups, will result in the same KMD value. This theory can be applied to quickly screen homologs in natural products and is currently primarily used to analyze lipids.

Previous literature indicates that the SFI contains many types of compound skeletons, including flavonoids, saponins, polyene and polyacetylene, and alkaloids. The literature also indicates that there are one or more CH₂ group differences within each type of compound’s skeletons. Therefore, by combining these two data filtering methods, we can quickly characterize the same type of components and their homologs in the SFI. A total of 198 components were collected by retrieving the previous literature and sorted into flavonoids (50), saponins (79), alkaloids (30), and polyene and polyacetylene (39) based on their structures, and various PMDF windows of different components were created by the programmed formula using Excel (Figures S3–S6, Tables S1–S4).

The QI data were filtered according to the established PMDF window, and the results were screened for potential flavonoids (366), saponins (389), alkaloids (391), and alkenes/alkyne (396) components. The potential flavonoids, saponins, alkaloids, and alkene components were further filtered using the KMDF method, where the bias was set to ±5 ppm to characterize as many constituents as possible in the SFI. The results are shown in Figure 2A. Taking compounds 34 and 44 as examples, they have the same KMD value, indicating that compounds 34 and 44 might be the same type of compound. Moreover, the NKM values differed by 14, indicating one CH₂ unit difference between compounds 34 and 44. Similarly, more types of components including flavonoids, saponins, alkaloids, and alkenes can be analyzed in the SFI using this strategy. The numbers of homologs screened using the KMDF method were 30, 16, 16, and 2, for flavonoids, saponins, alkaloids, and alkenes/alkynes, respectively. Unfortunately, the KMDF method could not precisely characterize the absolute configurations of isomers.

When comparing the CCS values acquired by UPLC-IMS-Q-TOF analysis with the CCS values retrieved from the literature [18], the CCS values of the same components characterized in the current study were consistent with previous literature within 1% variation. This indicated that the CCS values obtained in the current study are accurate and can be applied to future analysis. The KMD method and the CCS values were integrated to analyze the characterized components and the results are shown in Figure 2B. Figure 2B shows that the higher the molecular weight of the identified compound, the larger the CCS value. Additionally, the CCS values of the saponins were larger than those of other types of components. These data contribute to establishing a CCS database of characterized compounds which can provide a reference for identifying chemical components in the SFI.

2.3. Characterization of Compounds in the SFI

The SFI is composed of Radix codonopsis and Radix astragali. Radix astragali mainly contains flavonoids and saponins, while Radix codonopsis mainly contains alkenes/alkynes and alkaloids [19,20,21,22]. Studies have shown that the flavonoids and saponins in Radix astragali have anti-inflammatory, fatigue-alleviating, anti-tumor, and immunoregulatory activity [23,24], and the alkenes/alkynes and alkaloids in Radix codonopsis have antioxidant and anti-tumor activity [25,26]. To characterize more flavonoids, saponins, alkaloids, and alkene in SFIs, the components in the SFI were identified by comparing the components with in-house databases, the mass spectrometry data from the literature, and the reference substances’ chromatography, spectra, and retention time data. Then, diagnostic fragment ions were identified for different types of components by summarizing the fragment ion peaks of the identified components. Additionally, more homologs were explored using identified components with the KMDF method.

2.3.1. Characterization of Flavonoids

Both Radix codonopsis and Radix astragali contain flavonoids. The predominant flavonoids in Radix codonopsis are flavonol, flavanones, and their glycosides [20], while Radix astragali has mainly isoflavones, flavonoids, isoflavones, pterocarpans, chalcones, and their glycosides [19]. By summarizing the pattern of flavonoids’ break bonds in MS, it was noted that flavonoids mainly underwent glycoside bond breakage, dehydration, CO₂, CO, C-ring fracture, and retro-Diels–Alder (RDA) reactions. In this study, 58 flavonoids (compound 1–58) were identified by in-house databases, MS data from the literature, and retention time and MS data obtained using reference substances.

Compounds 1, 2, and 12 were presumed to be flavonols according to their MS/MS spectra. In compound 1, the [M − H]⁻ ion at m/z 623.16 Da and the [M + HCOO]⁻ ion at m/z 669.17 Da were detected in the MS spectra. Between the two ions, m/z 623.16 Da produced m/z 461.11, m/z 299.06, and m/z 151.04 fragmented ions, which were derived from successive losses of two Glc and C-ring fractures, respectively, followed by dicarbonyl group. Compound 1 was identified as a complanatuside or its isomer after comparison with MS data from the literature [27]. In MS, compound 2 gave an [M − H]⁻ ion at m/z 609.15 Da in negative ion mode and generated a molecular formula of C₂₇H₃₀O₁₆ with an error of −1.72 ppm. In MS/MS spectra, compound 2 showed three principal fragment ions at m/z 447.09, m/z 285.04, and m/z 193.05 Da, which were from successive losses of two Glc (162 Da) and a C-ring fracture (Figure S7B), respectively. Through comparison of the mass spectrometric data from the literature, compound 2 was tentatively identified as kaempferol-3-O-sophoroside [28]. In the negative ion mode, the pseudo-molecular ion peak of compound 12 was m/z 447.09 Da. It has one less Glc (162 Da) than compound 2. In MS/MS spectra, three ion fragments were detected at m/z 285.04, 255.03, and 193.05 Da, which came from the consecutive loss of one Glc (162 Da) and HCOH, and the C-ring fracture of the flavonoids. Based on the information of the compound parent ion and its fragment ions, compound 12 was identified as astragalin.

Comparison with the reference standards’ retention time and the MS data information, compounds 6 and 44 were identified as pratensein-7-O-β-d-glucopyranoside and calycosin, respectively, and both belonged in the isoflavone group. Calycosin is an isoflavone glycoside, and the precursor ion [M − H]⁻ was m/z 283.06 in the negative ion mode. The main fragment ions in MS/MS were m/z 268.05, 239.04, 211.04, 195.05, and 183.05 Da, and these ions were derived from calycosin losing CH₃ (15.02 Da), CHO (29.00 Da), O (15.97 Da), and 2CO (55.99 Da). The MS fragmentation pattern and secondary fragmentation plot of calycosin are shown in Figure S7C. In negative ion mode, the precursor ions [M-H]⁻ of pratensein-7-O-β-d-glucopyranoside (6) was m/z 461.11 Da, and thus the molecular formula was C₂₂H₂₂O₁₁. The MS/MS fragments derived from m/z 461.11 Da were m/z 299.06, m/z 284.03, and m/z 266.02 Da, created by successive losses of Glc, CH₃, and H₂O from pratensein-7-O-β-d-glucopyranoside.

Compound 47 was identified as isomucronulatol 7-O-glucoside through comparison with the literature’s MS data information, which belongs to the isoflavane group [29]. In the negative ion mode, the pseudo-molecular ion [M-H]^- was m/z 463.16 Da, and the fragment ions were m/z 301.11, 286.09, 271.06, and 147.05 Da in the MS/MS spectra (Figure S7D). These fragment ions were derived from isomucronulatol 7-O-glucoside by consecutive losses of Glc, CH₃, and CO, as well as C-ring fracture in the basic skeleton of isoflavane.

Three of the 58 flavonoids were identified using reference substances, and the results showed that the analyzed structures contained isoflavones, flavonols, and isoflavone components. Unfortunately, the various classifications of flavonoid components and the presence of many isomers among different types make it impossible to characterize the flavonoid components one by one.

2.3.2. Characterization of Saponins

The saponins in the SFI are mainly derived from Radix astragali and Radix codonopsis herbs and most of them are tetracyclic triterpenes and pentacyclic triterpenes [20,27]. Pharmacological studies have shown that saponins have immunomodulatory [30], anti-tumor [31], antioxidant [32], and neuroprotective activities [33]. Therefore, it is crucial to characterize the saponin components in the SFI.

In negative ion mode, the molecular ion peaks [M − H]⁻ and additive ion peaks [M + HCOO]⁻ of compounds 98, 100, and 102 were at m/z 783.45 Da and m/z 829.46 Da, respectively, and the three were also isomers. Compounds 98, 100, and 102 were identified as astragaloside IV, astragaloside III, and isoastragaloside IV through comparison of the retention time and MS information using reference substances, and the three belonged in the tetracyclic triterpenoid group. In the MS/MS spectra, m/z 621.40 Da and m/z 489.36 Da peaks were detected in compounds 98, 100, and 102 fragments, and the two were derived by successive loss of one Xyl and one Glc from the precursor ions, respectively. Thus, the fragment ions m/z 621.40 Da and m/z 489.36 Da can be applied for the identification of tetracyclic triterpenoids in the SFI when their glycoside belongs to cycloastragenol.

The MS/MS spectra and possible MS fragment patterns of astragaloside IV are shown in Figure S8B. In negative ion mode, compounds 71, 74, and 92, and their precursor ions ([M − H]⁻ m/z 945.51 Da and [M + HCOO]⁻ m/z 991.52 Da) were detected, and they contained one more Glc than astragaloside IV. Further comparisons were made with the retention time and MS data of the reference substances, and compounds 71, 74, and 92 were identified as astragaloside VII, astragaloside VI, and astragaloside V, respectively. Compared to compound 71, the molecular ion peaks of compound 91 had two less H atoms and MS/MS fragments than compound 71. In negative ion mode, it was shown that compound 91 is a tetracyclic triterpene saponin with a skeleton type containing two less H atoms than cycloastragenol.

Furthermore, compounds 105, 114, and 117 were identified as rajanoside, isoastragaloside II, and cyclocephaloside II through comparison with reference substances. The fragment ions of compound 79 were [M − H]⁻ m/z 785.47 Da and [M + HCOO]⁻ m/z 831.48 Da, and its molecular formula was tentatively identified as C₄₁H₇₀O₁₄ (−0.52 ppm). In the MS/MS spectra, fragment ions at m/z 623.42, 491.37, 473.33, and 415.32 Da were detected, which can be derived by successive losses of Xyl, Glc, H₂O, and CO groups from the m/z 785.47 Da ion. After review of the literature, compound 79 was identified as cyclocanthoside E [33]. The MS/MS spectra and possible MS fragmentation patterns of cyclocanthoside E are shown in Figure S8C.

Compound 106 precursor ions [M − H]⁻ m/z 941.51 Da and [M + HCOO]⁻ m/z 987.51 Da were identified as soasaponin I through comparison of the retention time and MS data with the reference substance. Soyasaponin I is a pentacyclic triterpenoid saponin, and its fragment ions were detected in the MS/MS spectra as m/z 795.46, m/z 615.40, and m/z 457.37. These fragment ions were derived from the successive losses of Gal, Glc, and C₆H₆O₅ from soyasaponin I. The MS/MS spectra and possible MS fragmentation patterns of soyasaponin I are shown in Figure S8D. Additiionally, the m/z 457.37 Da fragment ion can be used as a diagnostic ion to characterize pentacyclic triterpenoids in the SFI. In negative ion mode, compound 119’s pseudo-molecular ion [M − H]⁻ and MS/MS fragment had two H less than soyasaponin I, indicating that compound 119 may be a compound containing a double bond in the soyasaponin I skeleton. Based on the in-house database, PMDF combined with the KMDF approach was further adopted, and 71 saponins were characterized in the SFI. Eleven of the 71 saponins were confirmed by comparison with the reference substances.

2.3.3. Characterization of Alkaloids

Many studies have demonstrated that alkaloids in natural products have good biological activity [34,35]. For example, most analgesic drugs are solely made of alkaloid components. Though a small number of alkaloids have been reported in Radix codonopsis, few studies have been performed focusing on alkaloid components in the SFI. Therefore, it is necessary to characterize the alkaloid components in the SFI systematically.

Previous studies have shown that alkaloids are prone to obtain a proton in the positive ion mode and generate a [M + H]⁺ molecular ion peak with a better response [36]. Therefore, the characterization of alkaloids in the SFI was performed in positive ion mode. The ion peaks [M + H]⁺ at m/z 416.19 Da and [M + Na]⁺ at m/z 438.17 Da of compounds 127, 132, and 134 were detected in positive ion mode. Through comparison with the MS data in the literature, it was indicated that these ion peaks are codonopiloside A and its isomers [2]. The main fragment ions produced by codonopiloside A in MS/MS spectra were at m/z 398.18, m/z 254.14, m/z 236.13, m/z 218.12, m/z 205.09, and m/z 161.06 Da, and these ions were derived from codonopiloside A after successively losing one molecule of Glc, H₂O, NH₂CH₃, and C₂H₄O. The possible fragment patterns of codonopiloside A in MS spectra are shown in Figure 3B. The m/z 398.18, m/z 254.14, m/z 236.13, m/z 218.12, m/z 205.09, and m/z 161.06 Da fragment ions can be applied as components of the diagnostic ion in Radix codonopsis, whose hydrogen in the N atom of pyrolidine alkaloids was substituted by a single CH₃ group.

Compounds 153 and 158 are a pair of isomers and m/z 350.20 Da [M + H]⁺ ions were observed in positive ion mode. Through comparison with MS data from the literature, compounds 153 and 158 can be identified as codonopyrrolidium A and its isomers. Fragment ions at m/z 268.15, m/z 250.14, m/z 220.13, m/z 205.09, and m/z 161.06 Da were observed in the MS/MS spectra, and these fragment ions were derived by successive losses of C₅H₇O, H₂O, NH(CH₃)₂, and C₂H₄O from the pseudo-molecular ion, respectively. According to the codonopyrrolidium A MS/MS fragment ions, their possible MS fragmentation patterns are shown in Figure 3D. The fragment ions at m/z 268.15, m/z 250.14, m/z 220.13, m/z 205.09, and m/z 161.06 Da can be used as diagnostic ions in the Radix codonopsis, whose hydrogens in the N atom of pyrolidine alkaloids are substituted by two CH₃ groups. The precursor ions [M + H]⁺ of compounds 133 and 141 were detected at m/z 430.21 Da. In the MS/MS spectra, it can be observed that there are more CH₂ fragments than those of codonopiloside A, indicating that compounds 133 and 141 might come from the triple CH₃ replacement of the hydrogen on the N atom of codonopiloside A. The KMDF analysis showed that compound 136 had one more CH₂ group than compound 130 and its pseudo-molecular ion peak was at [M + H]⁺ m/z 430.21 Da. Fragment ions at m/z 412.19, m/z 268.15, m/z 250.14, m/z 232.13, and m/z 205.09 Da were detected in the MS/MS spectra, which indicated that compound 136 was derived from the replacement of the H atom with two CH₃ groups on the N atom of codonopyrrolidium A. The possible MS fragment pattern of compound 136 is shown in Figure 3C and no information regarding compound 136 can be retrieved in the SciFindern database, indicating that it is potentially a new compound.

In this study, 50 alkaloid components were characterized in the SFI using the PMDF combined with KMDF strategy, Most of the alkaloid components are pyrrolidine alkaloids from Radix codonopsis. However, few studies have been reported on alkaloids in Radix codonopsis and Radix astragali. The structure of the characterized alkaloids cannot be determined.

2.3.4. Characterization of Polyacetylenes, Polyenes, and Their Glycosides

Polynephthalenes, polyenes, and their glycoside components are widely distributed among Campanulaceae plants and are the characteristic components of Radix codonopsis. Multiple unsaturated bonds make this type of component more active and prone to addition and oxidation reactions. Therefore, a large number of prototypes of such components can be characterized by the MS method.

Compounds 181 and 182 are a pair of isomers, with [M − H]⁻ and [M + HCOO]⁻ ions at m/z 423.19 and m/z 469.19 Da, respectively. In negative ion mode, fragment ions at m/z 261.13, m/z 221.07, m/z 179.06, and m/z 161.05 were detected in the MS/MS spectra, and they were derived from m/z 423.19 Da by losing one Glu, hexenyl fracture, [Glu-H]⁻, and [Glu-H₂O-H]⁻. Through comparison of the MS data and the polarity of the compounds with the literature, compounds 181 and 182 can be inferred to be (E)-2-Hexenyl-β-sophorodide and (Z)-3-Hexenyl-β-sophorodide, respectively [37].

Compounds 201, 202, 204, and 205 have the same pseudo-molecular ions [M − H]⁻ and addition ions in primary mass spectrometry [M + HCOO]⁻ m/z 395.17 and 441.18 Da, indicating that they are isomers of each other, and are inferred to lobetyolin through comparison with the literature mass spectrometry data and that of its isomer [2]. Fragment ions at m/z 305.12, m/z 233.12, m/z 215.11, m/z 179.05, and m/z 143.07 Da were detected in the MS/MS spectra, and they were derived from lobetyolin’s lipid chain breakage and losses of Glu, H₂O, and CH₂O, respectively. The MS/MS spectra and their possible fragment patterns of lobetyolin are shown in Figure 4B, indicating that m/z 305.12, 233.12, 215.11, 179.05, and 143.07 Da can be used as diagnostic ions for this class of components. Compounds 183, 192, 194–196, 198, and 200 in the MS spectra exhibited one more C₆H₁₀O₅ group (m/z 162.05 Da) than lobetyolin, and they were presumably identified as lobetyolinin and its isomers. Furthermore, compound 203 had two more hydrogen atoms than lobetyolin, while the RDB value was 6.5 (lobetyolin, RDB = 7.5). This indicated that compound 203 might be codonopiloenynenoside A. Using the diagnostic ion method, a total of 30 compounds, including polyacetylenes, polyenes and their glycosides, were characterized from the SFI.

2.3.5. Characterization of Other Components

In addition to flavonoids, saponins, alkaloids, polyacetylenes, and polyenes, there are also a small amount of phenylpropanin, lignans, and unsaturated fatty acids in SFI [2]. Other classes of ingredients in the SFI were characterized by comparing the MS data of chemical constituents in Radix codonopsis and Radix astragali plants to previous literature. The molecular ion peak [M − H]⁻ of compound 217 was m/z 327.22 Da, detected in negative ion mode, which was inferred to belong to 9,12,13-Trihydroxy-10,15-octadecadienoic acid using MS data from the literature [38]. This compound belongs to the unsaturated fatty acid group. Fragment ions at m/z 309.16, m/z 291.18, m/z 229.12, m/z 211.11, and m/z 171.10 Da were detected in the MS/MS spectra and they are derived from the continuous loss of H₂O molecules and the breakage of fat chains from 9,12,13-trihydroxy-10,15-octadecadienoic acid. Their MS/MS fragment patterns are shown in Figure 4C. Similarly, compounds 217–224 were identified as unsaturated fatty acids. Compound 215 was identified as syringaresinol 4’-O-glucopyranoside through comparison of the retention time and MS data with those of the reference substance. Fragment ions at m/z 417.16, m/z 402.13, m/z 387.10, and m/z 371.15 Da were detected in the MS/MS spectra, derived by the losses of CH₃ and HCOOH groups from the precursor ions, respectively. The MS/MS spectra of lobetyolin and their possible fragment patterns are shown in Figure 4D.

2.4. Analysis of Identified Compounds in SFI

To further analyze the source of the constituents in the SFI, the identified 233 compounds (Table S5) were uploaded into the EHBIO (http://www.ehbio.com/test/venn/#/ (accessed on 1 June 2022) data analysis platform for analysis [39]. As shown in Figure 5A, 163 components were identified from Radix astragali, 111 components from Radix codonopsis, and 194 compounds were identified in the SFI using IMS-Q-TOF. Both Radix astragali and Radix codonopsis had five exclusive compounds in them and one exclusive component was identified in the SFI. Figure 5B shows that the identified components include 58 flavonoids, 71 saponins, 50 alkaloids, 30 polyenes, and polynephthyls, and 14 other types of compounds (such as unsaturated fatty acids, phenylpropanins, and lignans). Figure 5C showed that alkaloids, polyenes, and polynephthenes detected in SFI are mainly contributed by Radix codonopsis, while flavonoids and saponins are mainly contributed by Radix astragali. Figure 5D showed that the flavonoids (33) and saponins in the identified components mainly originated from Radix astragali (3), the alkaloids are from Radix astragali (25) and Radix codonopsis (25), and polyene and polynephthalene (25) components are mainly derived from Radix codonopsis herbs. These results lead to a detailed understanding of the origin and types of chemical components in the SFI, providing a reference to select quality control markers for SFI.

3. Materials and Methods

3.1. Reagents and Samples

A total of 18 standards were used for fragmentation behavior studies and confirmation of identification, including pratensein 7-O-glucopyranoside (CAS 36191-03-4, C₂₂H₂₂O₁₁, MW. 462.12 Da; 06), calycosin 7-O-β-d-glucoside (CAS 20633-67-4, C₂₂H₂₂O₁₀, MW. 446.12 Da; 14), calycosin (CAS 20575-57-9, C₁₆H₁₂O₅, MW. 284.07 Da; 44), 9-O-Methylnissolin 3-O-glucoside (CAS 94367-42-7, C₂₃H₂₆O₁₀, MW. 462.15 Da; 40), ononin (CAS 486-62-4, C₂₂H₂₂O₉, MW. 430.12 Da; 32), isomucronulatol 7-O-glucoside (CAS 94367-43-8, C₂₃H₂₈O₁₀, MW. 464.17 Da; 47), syringin (CAS 118-34-3, C₁₇H₂₄O₉, MW. 371.14 Da), lobetyolin (CAS 136085-37-5, C₂₀H₂₈O₈, MW. 396.18 Da; 200), lobetyolinin (CAS 142451-48-7, C₂₆H₃₈O₁₃, MW. 558.23 Da; 193), astragaloside IV (CAS 84687-43-4, C₄₁H₆₈O₁₄, MW. 784.46 Da; 98), astragaloside III (CAS 84687-42-3, C₄₁H₆₈O₁₄, MW. 784.46 Da; 100), astragaloside II (CAS 84676-89-1, C₄₃H₇₀O₁₅, MW. 826.47 Da; 105), isoastragaloside II (CAS 86764-11-6, C₄₃H₇₀O₁₅, MW. 826.47 Da; 114), astragaloside I (CAS 84680-75-1, C₄₅H₇₂O₁₆, MW. 868.48 Da; 127), astragaloside VI (CAS 84687-45-6, C₄₇H₇₈O₁₉, MW. 946.51 Da; 74), cyclocephaloside II (CAS 215776-78-6, C₄₃H₇₀O₁₅, MW. 826.47 Da; 117), isoastragaloside I (CAS 84676-88-0, C₄₅H₇₂O₁₆, MW. 868.48 Da; 127), soyasaponin I (CAS 51330-27-9, C₄₈H₇₈O₁₈, MW. 942.52 Da; 106), isoastragaloside IV (CAS 136033-55-1, C₄₁H₆₈O₁₄, MW. 784.46 Da; 101). All standards were obtained from Shanghai Standard Technology Co. Ltd. The purity of all compounds met the requirement (>98%).

HPLC-Grade acetonitrile, methanol (Merck, Darmstadt, Germany), formic acid (FA), and deionized water (18.2 MΩ at 25 °C), prepared by the Millipore Alpha-Q water purification system (Millipore, Bedford, MA, USA), were used in the mobile phase for chromatographic separation and the extraction solvent. Leucine-enkephalin was purchased from Sigma-Aldrich (St. Louis, MO, USA).

The SFI (NO:190135) and the corresponding samples during the manufacture, including the single herb extract Co_E (NO: 190135-Co) and As_E (NO: 190135-As) were provided by Livzon Pharmaceutical Group Inc (Zhuhai, China).

3.2. Sample Preparation

The SFI sample was enriched with SPE (Bond Elut Plexa, 6 mL/200 mg, Agilent). After washing with water, the methanol eluent was collected and evaporated to dry. The residue was dissolved in 5.0 mL water with a final concentration 0.6 g/mL Co or As.

The other samples, including Co_E, and As_E, were prepared with the same procedure, and the concentration was adjusted to a final concentration 0.6 g/mL Co or As.

3.3. Chromatogram Fingerprint Obtained with UHPLC-CAD

The SFI was separated on an Agilent Poroshell 120 EC C18 column (150 mm × 4.6 mm, 2.7 μm) on an Ultimate ™ 3000 Standard Dual System (Thermo Fisher Scientific, San Jose, CA, USA) equipped with dual-gradient pumps, an autosampler, and a charged aerosol detector (CAD, Corona Veo). A binary mobile phase consisting of acetonitrile (A) and 0.1% formic acid in water (B) at a flow rate of 0.8 mL/min was adopted with a gradient elution program: 0–1 min, 12% (A); 1–13 min, 12–18% (A); 13–34 min, 18–25% (A); 34–48 min, 25–35% (A); 48–59 min, 35–70% (A); 59–59.1 min, 70–95% (A); 59.1–64 min; 95–95% (A); 64–65 min, 95–12% (A); 65–75 min, 12–12% (A). The column temperature was set at 25 °C and the CAD detector was optimized at a sampling frequency of 10 kHz, filter 3.6 s, and a low evaporator temperature (35 °C).

3.4. CCS Value Obtained from UHPLC -IM-QTOF

The CCS value and DIA analysis (MS^E) were performed on a traveling wave ion-mobility hybrid quadrupole time-of-flight mass spectrometer with an ESI ion source (SYNAPT G2-Si, Waters, Wilmslow, UK) in resolution mode (~30,000 FWHM at m/z 500). The MS data were acquired from 100 to 1500 Da in positive and negative ion modes. The ESI source parameters were set as follows: capillary voltages of 0.5 kV (ESI+) or 2.5 kV (ESI−), a cone voltage of 40 V, a source offset of 60 V, a source temperature of 120 °C, desolvation gas flow of 800 L/h, desolvation gas temperature of 500 °C, and a cone gas (N₂) flow of 20 L/h. The traveling wave ion-mobility mass separation parameters were set as follows: helium cell 95 mL/min; triWave, IMS wave velocity 650 m/s, wave height 20 V, transfer wave velocity 176 m/s; IMS wave velocity ramp starting at 80 m/s and ending at 1000 m/s. The MS and MS^E data scanning time was set at 0.5 s. Leucine enkephalin (LE; Sigma-Aldrich, St. Louis, MO, USA; 200 ng/mL) and sodium formate solvent (0.5 mmol/L) were used as the calibration standards. The CCS accuracy was calibrated with poly-dl-alanine solution (10 mg/L). MassLynx V4.1 software (Waters, Milford, MA, USA) was employed for data acquisition and processing.

3.5. Fragmentation Obtained from UHPLC-LTQ-Orbitrap

The MS/MS fragmentation information was obtained on the LTQ-Orbitrap Velos Pro-hybrid mass spectrometer (Thermo Fisher Scientific, San Jose, CA, USA) equipped with a heated electrospray interface (HESI) source in positive and negative ion mode. The parameters were set as follows: ion spray voltage, 3.8 kV (ESI+) or 2.7 kV (ESI−); collision energy, 30 eV; capillary temperature, 350 °C; source heater temperature, 300 °C. Nitrogen was used as the sheath gas and the auxiliary gas, and the gas flows were at 15 and 8 arbitrary units, respectively. The MS scan range of m/z 300–2000 was at the resolution of 30,000 (FWHM at m/z 400) as Event I. MS spectrum of top one intense ion from the parent list (PIL) was collected as Event Ⅱ. The most intense ion of MS² would trigger MS³ fragmentation and was recorded as Event Ⅲ. The following parameters were used for dynamic exclusion: repeat count, 3; repeat duration, 2 s; exclusion list size, 50; exclusion duration, 30 s. Both CID/MS² and CID/MS³ in normalized collision energies (NCE) were 35%. The resolution for Event II and III was set as 7,500 (FWHM at m/z 400). Xcalibur 4.1 software (Thermo Fisher Scientific, San Jose, CA, USA) was used to assist the data acquisition and processing.

3.6. Data Processing

According to previous literature, the organic small molecule compounds in Shenqi Fuzheng Injection are mainly composed of flavonoids, saponins, polyacetylenes and polyenes, alkaloids, and other components. To systematically characterize the different types of components in SFI, we imported the original MS file (QTOF MSE data) into QI, and the parameters of data processing were set as follows: the peak width was at 0.2 min; the minimum MS response was at 2000; the positive adduct ions were set as [M + H]⁺, [M + Na]⁺, [M + K]⁺, [2M + H]⁺, and [2M + Na]⁺; the negative adduct ions were set as [M − H]⁻, [M + COOH]⁻, [2M − H]⁻, and [2M + COOH]⁻, respectively. After retention time alignment, noise signal removal, and peak extraction, an .msp file was generated to integrate m/z, retention time, peak intensity, and CCS values in .scv format with data matrix and MS/MS data. The reported compounds in Radix codonopsis and Radix astragali were retrieved from literature, and an in-house database containing 198 identified compounds was established and imported into the UNIFI database for the following compound identification in the SFI. The compounds were collected, classified, and compiled with multiple formulas using Excel. PMDF windows were established for multiple components in the SFI to screen the potential flavonoids, saponins, polyacetylenes and polyenes, alkaloids, and other components in the obtained MS spectra for the SFI. To identify more components in the SFI, components were first identified with the PMDF method, then the KMDF method was applied to identify and recognize compounds with one or more CH₂ groups [40]. The KMD formula is as follows:

Kendrick mass (KM) = m/z × (Nominal mass of CH₂ (14))/(Exact mass of CH₂ (14.01565))

(1)

KMD = (KM − NKM) × 1000

(2)

In the above equations, m/z is the measured mass-to-charge ratio, and NKM is the nearest integer of KM.

In order to obtain more abundant and accurate information on the MS fragments from the identified components, the information of the fragments was further improved using LTQ-Orbitrap (Table S5).

4. Conclusions

In the current study, a fingerprint of the secondary metabolites in the SFI was established by UHPLC-CAD and the fingerprints in the SFI were systematically characterized by chemical composition in combination with the IMS-Q-TOF and LTQ-Orbitrap techniques. Based on the characterization of the components in the SFI using an in-house database and reference substances, PMDF combined with the KMDF data processing method was used to mine the chemical composition information of the SFI. A total of 223 secondary metabolites were characterized from the SFI fingerprints, including 58 flavonoids, 71 saponins, 50 alkaloids, 30 polyene and polycynes, and 14 other types of compounds. Among them, the flavonoids and saponins in SFI are mainly contributed by Radix astragali and the enotylene components are mainly contributed by Radix codonopsis. Specifically, 106 components come from Radix astragali, including 33 flavonoids, 53 saponins, and 19 alkaloids. There were 64 components characterized using the KMDF method, which increased the characterization of the components in the SFI by 28.70%. This study provides a solid foundation for authentification analysis of the chemical compositions the SFI.