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

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.


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

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).

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 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).

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 CH 2 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 CH 2 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 CH 2 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.

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 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.

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.

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, CO 2 , 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.  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 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, CH 3 , 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.

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], antitumor [31], antioxidant [32], and neuroprotective activities [33]. Therefore, it is crucial to characterize the saponin components in the SFI.
In 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.  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.

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. characterized in the SFI. Eleven of the 71 saponins were confirmed by comparison with the reference substances.

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.    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.  Compounds 183, 192, 194-196, 198, and 200 in the MS spectra exhibited one more C 6 H 10 O 5 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.

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 Figure 4D.

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

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 Figure 4D.

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. 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.

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, C 22  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).

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.

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 MS 2 would trigger MS 3 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/MS 2 and CID/MS 3 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.

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. 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 CH 2 groups [40]. The KMD formula is as follows: Kendrick mass (KM) = m/z × (Nominal mass of CH 2 (14))/(Exact mass of CH 2 (14.01565)) (1) 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).

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.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/molecules27154734/s1, Figure S1. Chemical structures of nine reference standards; Figure S2. Fingerprint chromatogram of UHPLC-CAD. (A: Codonopsis pilosula, B: Astragalus memeranaceus, C: SFI); Figure S3. Definition of PMDF windows of flavonoids through several excel formulas; Figure S4. Definition of PMDF windows of saponins through several excel formulas; Figure S5. Definition of PMDF windows of alkaloids through several excel formulas; Figure S6. Definition of PMDF windows of polyacetylenes, and polyenes through several excel formulas; Figure S7.  Table S1. The compound information used in establishing the PMDF window for flavonoids in SFI; Table S2. The compound information used in establishing the PMDF window for saponins in SFI; Table S3. The compound information used in establishing the PMDF window for alkaloids in SFI; Table S4. The compound information used in establishing the PMDF window for polyene and polytylene components in SFI; Table S5. Detailed information about the characterized compounds in SFI. Wang only participated in the collection of data and were not involved in any other processes of the study. Authors L.F. Liu and W.H. Huang served as part of the funders in the current study and these funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.
Sample Availability: Samples of the compounds are available from the authors.