MS-grade methanol and acetonitrile were purchased from Merck (Germany). MS-grade formic acid was bought from Thermo Fisher Technology Co., Ltd. (United States). Mullein isoflavone glucoside, naringin, hesperidin, ononin, epimedoside A, icariin, nobiletin, and tangeretin (purities > 98%) were all purchased from Sichuan Vicky Biotechnology Co., Ltd. (Chengdu, China). Betaine, adenosine, magnoflorine, peimine, peiminine, vanillin, hesperetin, wogonin, honokiol, magnolol, and linoleic acid (purities > 98%) were all acquired from Chengdu Pufei De Biotech Co., Ltd. (Chengdu, China). Lipopolysaccharide (LPS) was offered by Shanghai Sulaibao Biotechnology Co., Ltd. (Shanghai, China). Interleukin-8 (IL-8), interleukin-6 (IL-6), tumor necrosis factor α (TNF-α), prostaglandin E2 (PGE2), matrix metalloproteinase-9 (MMP-9), and nitric oxide (NO) testing kits were purchased from Shanghai Langton Biotechnology Co., Ltd. (Shanghai, China).

BJG was provided by the First Affiliated Hospital of Henan University of Traditional Chinese Medicine and produced by Jiangyin Tianjiang Pharmaceutical Co., Ltd. (batch no. 1905301, China). The preparation process of BJG is an industrial production process amplified proportionally according to the following single feeding amount: Astragalus mongholicus Bunge [Fabaceae; Astragalus mongholicus radix] 12 g, Codonopsis pilosula (Franch.) Nannf [Campanulaceae; Codonopsis pilosula radix] 6 g, Polygonatum kingianum Collett & Hemsl [Asparagaceae; Polygonatum kingianum rhizoma] 12 g, Atractylodes macrocephala Koidz [Asteraceae; Atractylodis macrocephalae rhizoma] 9 g, Poria cocos (Schw.) Wolf [Poromycelidae; Poria] 9 g, Fritillaria thunbergii Miq [Liliaceae; Fritillariae thunbergii bulbus] 6 g, Pheretima aspergillum (E. Perrier) [Lumbricidae; Pheretima] 9 g, Magnolia officinalis Rehder & E. H. Wilson [Magnoliaceae; Magnolia officinalis cortex] 6 g, Citrus reticulata Blanco [Rutaceae; Citri reticulatae pericarpium] 9 g, Aster tataricus L. f [Asteraceae; Asteris radix et rhizoma] 6 g, Epimedium brevicornu Maxim [Berberidaceae; Epimedii folium] 6 g, and Ardisia japonica (Thunb.) Blume [Primulaceae; Ardisia japonicae herba] 15 g.

After being crushed, Atractylodis macrocephalae rhizoma was combined with Citri reticulatae pericarpium, six times the volume of water was added, and the sample was hydrodistilled for 5 h. The volatile oil was collected. Then, the sample was filtered to get decoction (I), and the residue was collected for later use. Extract (I) was obtained as follows: Magnolia officinalis cortex, Asteris radix et rhizoma, and Fritillariae thunbergii bulbus are taken, six times the volume of 70% ethanol is added, the extract is refluxed for 1.5 h, and the process is repeated three times. The filtrates of three repetitions are combined, the sample is decompressed to recover ethanol, and it is concentrated to a relative density of 1.15 at 50°C to obtain extract (I). The residue was collected for later use. Extract (II) was obtained as follows: we combine the aforementioned drug residues, add eight times the volume of water, and decoct the sample three times for 1 h. Then, we filter and mix the filtrates with decoction (I), decompress the sample, and concentrate it to a relative density of 1.15 at 50°C to obtain extract (II). Dry powder (I) was obtained as follows: we combine extract (I) with extract (II) to obtain the total extract, 20% dextrin is added, and after spray-drying, dry powder (I) is obtained. Dry powder (II) was obtained as follows: the volatile oil is mixed with 10 times its amount of β-cyclodextrin by the grinding method to obtain the inclusion complex; after vacuum-drying and crushing, dry powder (II) is obtained. We mix dry powder (I), dry powder (II), and 10% honey of the total powder weight uniformly and make granules; after drying, the granules are formed (Li and Li, 2011).

The quality of BJG was verified by HPLC detection (Figure 1). Mullein isoflavone glucoside, naringin, hesperidin, ononin, epimedoside A, icariin, nobiletin, tangeretin, and honokiol were quantified for controlling the quality of the granule, and their contents in the batch 1905301 sample were 0.071 mg/g, 0.172 mg/g, 0.028 mg/g, 0.017 mg/g, 0.021 mg/g, 0.343 mg/g, 0.047 mg/g, 0.018 mg/g, and 0.042 mg/g, respectively. The chemical composition of BJG was described in our previous study by the HPLC-QTOF-MS technology (Cui et al., 2020). Based on this, the authenticity of these nine quality control indicators was confirmed by comparing the retention time, accurate mass number, isotope peak, and other information of these chemical components (please refer to the Supplementary Material for details of HPLC and HPLC-QTOF-MS).

Male Sprague–Dawley (SD) rats (200 ± 20 g) were purchased from Liaoning Changsheng Biotechnology Co., Ltd. [number: SCXK (Liao) 2020-0001]. All rats were acclimated for 1 week after arrival and housed in a room under controlled temperature (22 ± 2°C) and humidity with a 12-h light/dark cycle and free access to water and food. Apart from that, all experiments were carried out following the approved animal protocols and guidelines established by the Medicine Ethics Review Committee for Animal Experiments of Liaoning University of Traditional Chinese Medicine with approval number 2020YS013(KT)-013-01.

A total of 72 SD rats were randomly divided into six groups, namely, the blank group, COPD model group, AMINOPH (aminophylline) (0.0054 g/kg) positive control group, BJG high-dose (2.36 g/kg) group, medium-dose (1.18 g/kg) group, and low-dose (0.59 g/kg) group, with 12 rats in each group. The medium dose of BJG is determined according to the conversion of animal dose to human equivalent dose. For the high dose, we used two times the medium dose, and for low dose, we used two times below the medium dose, to investigate the dose–effect relationship. Except for the blank group, for rats in the other groups, the COPD model was established by administering LPS combined with smoking according to the method with minor modification (Shin et al., 2017; Mao et al., 2022; Pelgrim et al., 2022). The administration intervention was started in the 6th week, twice a day. The blank group and the model group were given distilled water of equal volume, and the whole process lasted for 6 weeks. During this period, the activity and behavior changes of the rats such as hair, mental state, activity ability, weight, death, and other physiological state indicators were observed and detected. After the drug intervention, the Buxco noninvasive animal airway detection system was used to measure the lung function of rats (Luo et al., 2021). The rats to be tested were placed into the noninvasive animal airway detection cavity. After the rats’ breathing was stable, the minute ventilation volume (MV), special airway resistance (sRaw), airway resistance (Raw), functional residual volume (FRC), peak inspiratory flow rate (PIF), peak expiratory flow rate (PEF), and expiratory flow rate at 50% ventilation (EF50) were measured continuously. In addition, the contents of IL-6, IL-8, TNF-α, PGE2, MMP-9, and NO in the serum and BALF were determined using ELISA kits.

A total of 24 SD rats were randomly classified into the blank group, blank administration group, model group, and model administration group. In the model groups, the rat model of COPD was established by administering LPS combined with the smoking method (Shin et al., 2017; Mao et al., 2022; Pelgrim et al., 2022), while in the administration groups, rats were gavaged with 2.36 g/kg solution of BJG, twice a day for 7 consecutive days. The blank group and the model group were given the same amount of normal saline. Then, 12 h before the last administration, fasting without water was performed, followed by 60 min after the last administration, blood being taken from the abdominal aorta into centrifuge tubes with heparin sodium, and 30 min later, it being centrifuged at 3,000 rpm for 15 min to separate the plasma. Afterward, the supernatant was taken and kept in a refrigerator at -80°C for later use.

The plasma samples were thawed at room temperature, and 400 µl of plasma samples was accurately extracted, followed by protein removal with 1,200 µl of methanol. After that, by vortexing for 3 min and centrifugation at 13,000 rpm (4°C) for 10 min, the supernatant was taken, cryogenic freeze-dried, redissolved in 50 µl of methanol, and vortexed for 3 min. Furthermore, the supernatant was centrifuged again at 13,000 rpm (4°C) for 10 min and was directly detected by LC-MS (Li et al., 2022; Wang et al., 2022).

The positive ion mode is adjusted as follows: Agilent Poroshell 120 SB-C18 (100 mm × 4.6 mm, 2.7 μm) is used; the mobile phase consists of 0.1% formic acid water (A)–acetonitrile (B); gradient elution conditions are 0–25 min, 5–40% B; 25–35 min, 40–75% B; 35–40 min, and 75–100% B; the flow rate is 0.8 ml min⁻¹; the column temperature is 30°C; the injection volume is 1 μl; the electrospray ion source (ESI) is detected in the positive ion mode; the drying gas flow is 13 L/min; the drying gas temperature is 350°C; the capillary voltage (Vcap) is 4,000 V; the neutralizer pressure is 45 psig; the fragmentor voltage is 125 V; the skimmer voltage is 65 V; the mass scanning range is 50–1,000 m/z; and the secondary MS collision voltage is 40 eV.

The negative ion mode is set as follows: Agilent Poroshell 120 SB-C18 (100 mm × 4.6 mm, 2.7 μm) is used; the mobile phase consists of water (A)–acetonitrile (B); gradient elution conditions are 0–30 min, 5–100% B; the flow rate is 0.8 ml min⁻¹; the column temperature is 30°C; the injection volume is 5 μl; the drying gas flow is 11 L/min; the drying gas temperature is 250°C; the Vcap is 3,500 V; the neutralizer pressure is 45 psig; the fragmentor voltage is 125 V; the skimmer voltage is 65 V; the mass scanning range is 50–1,000 m/z; and the secondary MS collision voltage is 40 eV (Wang et al., 2021).

SPSS 19.0 software was adopted for statistical analysis. The data were represented as mean ± standard deviation (mean ± SD). Statistical comparisons were analyzed by one-way analysis of variance (ANOVA). A value of p < 0.05 suggests a difference, and p < 0.01 represents a significant difference (Chen et al., 2019; Liu et al., 2021).

MS data processing was performed with MassHunter software and the PCDL database, which contained extracted ion chromatograms and calculations of elemental compositions with mass errors within 10 ppm. The chemical structures of all analytes were explained and verified based on their elemental compositions, accurately measured mass values, the elution order on a C18 column, retention time, fragmentation behavior, and the comparison with authentic standards and literature as far as possible (Lu et al., 2021).

The rats in the blank group featured normal hair color and luster, good mental state, flexible and frequent activities, stable exhalation, and normal diet and water intake, while those in the model group had dull and yellow hair, no luster, cough, phlegm in the throat, mental fatigue, slow movement, reduced diet, and weight loss. In the BJG high-dose group and the AMINOPH group, cough was significantly relieved and phlegm and sound in the larynx disappeared, with daily activities of rats being more frequent. There was no significant difference in drinking water compared with the blank group. Furthermore, compared to the model group, the cough condition of rats in the BJG medium- and low-dose groups was reduced and their mental state was improved, while their water intake was reduced compared with the blank group.

Before the experiment, there existed no significant difference in the weight of rats in each group. After the experiment, a significant difference in the body weight existed between the blank group and the model group (p < 0.01). In comparison with the model group, the body weight in the BJG groups and the AMINOPH group was obviously higher (p < 0.01) (Figure 2A). Before and after the experiment, the weight of rats in each group increased linearly. The weight growth rate of the blank group was faster, and that of the model group was the slowest (Figure 2B). Furthermore, during the whole experiment, there was no death in the blank control group and BJG high-dose, medium-dose, and AMINOPH groups, while deaths occurred in the model and BJG low-dose groups, with mortality rates of 16.7% and 8.3%, respectively.

In comparison with the blank group, MV, PIF, PEF, and EF50 in the model group were dramatically decreased, while sRaw, Raw, and FRC were significantly increased (p < 0.01). Compared with the model group, MV, PIF, PEF, and EF50 in the BJG groups and the AMINOPH group were observably increased, while sRaw, Raw, and FRC were obviously decreased (p < 0.01) (Figure 2C). According to the results, compared with the blank group, the model group had increased airway resistance, decreased lung compliance, restricted airflow, and airway obstruction. However, compared with the model group, the indexes of each treatment group, especially the BJG high-dose and AMINOPH groups, had improved, indicating that BJG can ameliorate airway obstruction, enhance pulmonary ventilation, and exert a protective effect on the pulmonary function of COPD rats. In addition, a certain dose–effect relationship was observed.

The lung tissues of the blank control group were smooth, shiny, soft, light red in color, and normal in size. In the model group, the surface of lung tissue was rough, the whole lung was white, luster was lost, and lung swelling was obvious. Compared with the model group, the lung volume of rats in each administration group was significantly reduced, shiny, and soft. The color of the BJG high-dose group and the AMINOPH group was close to that of the normal group, and the volume was reduced compared with the model group to a great extent. Moreover, the color of the BJG medium- and low-dose groups was lighter, and the volume was larger than that of the blank group (Figure 3A).

Under the light microscope, the lung tissue structure of the blank group was complete and the bronchial tube wall was not thickened; other than that, there was no exudate in the lumen, the cilia were arranged neatly, and no inflammatory cell infiltration occurred. The size of the alveoli was normal, and the structure of the alveolar septum was clear. Compared with the blank group, the bronchial smooth muscle in the model group was significantly thickened and the lumen was filled with a large number of neutrophils. The alveolar septum was severely broken, alveolar cells were fibrotic, neutrophils were denatured and necrotic, cellulose was dissolved, alveolar walls were thinner, alveolar cavities were significantly expanded, and some alveoli fused to form pulmonary bullae. The pathological changes of the abovementioned trachea and lung tissues were consistent with those of COPD patients. Compared with the model group, the lung tissue damage in each drug intervention group was less than that in the COPD group and the lung tissue damage in the AMINOPH and BJG high-dose groups was the least (Figure 3B).

The contents of IL-6, IL-8, TNF-α, PGE2, MMP-9, and NO in the serum and BALF of rats in each group are shown in Figure 4. Compared to the blank group, the expression levels of IL-6, IL-8, and TNF-α were increased in the model group and all treatment groups, and compared to the model group, they were lowered dramatically in all treatment groups (p < 0.01), especially in the BJG high-dose group and the AMINOPH group. Compared with the blank group, the expression levels of NO and MMP-9 factors in the model group and each treatment group were increased, while compared with the model group, they were decreased to a great extent in all treatment groups (p < 0.01), especially in the BJG high-dose group and the AMINOPH group. Moreover, compared with the blank group, the expression levels of PGE2 in the model group and each treatment group were increased, whereas compared with the model group, they were noticeably decreased in all treatment groups (p < 0.01), especially in the BJG high-dose group and the AMINOPH group. These findings suggest that all treatment groups could improve the inflammatory response in COPD rats.

By comparing the profiles of positive and negative blank plasma and drug-containing plasma, 13 different compounds, including 10 prototype components and three metabolites, were found in the pathological plasma in the positive ion mode. In physiological plasma, it contains all pathological components, but there is one prototypical component and one metabolite more than that in the pathological state. In the negative ion mode, eight different compounds, involving five prototypes and three metabolites, were discovered in the pathological plasma. In the physiological state, it contains all pathological components, but there is one more prototypical blood-entering component than that in the pathological state. The total ion-flow chromatogram (TIC) of each group is shown in Figure 5, while the retention time and molecular weight of each component are displayed in Table 1.

In the positive ion mode, for compound P1 (1.829 min, m/z 118.0861) (C₅H₁₁NO₂, M + H)⁺, the secondary fragment ions were m/z 84.9595 (M + H-2OH)⁺ and m/z 58.0650 (M-C₂H₃O₂)⁺. After comparison with the betaine reference substance, the fragment ion information was basically consistent with that of betaine and it was determined to be betaine. For compound P2 (3.022 min, m/z 268.1040) (C₁₀H₁₃N₅O₄, M + H)⁺, the secondary fragment ions were m/z 136.0616 (M + H-C₅H₈O₄)⁺, m/z 119.0349 (M + H-C₅H₈O₄-NH₃)⁺, and m/z 94.0400 (M-C₅H₈O₄-NH₃-CN)⁺. After comparison with the adenosine control substance, the fragment ion information was in line with that of adenosine and, finally, it was identified as adenosine. For compound P3 (5.427 min, m/z 254.1387) (C₁₃H₁₉NO₄+H)⁺, the secondary fragment ion was m/z 161.0580 (M + H-H₂O-C₃H₉NO)⁺. In addition, after comparison with the standard database information and references, the fragment ion information is roughly consistent with that of codonopsinol B. It was speculated to be codonopsinol B. For compound P4 (11.245 min, m/z 342.1695) (C₂₀H₂₄NO₄+H)⁺, the secondary fragment ions were m/z 297.1130 (C₂₀H₂₄NO₄-CH₂O-CH₃)⁺, m/z 282.0884 (C₂₀H₂₄NO₄-CH₂O-2CH₃)⁺, m/z 222.0670 (C₂₀H₂₄NO₄-CH₂O-2CH₃-C₂H₄O₂)⁺, and m/z 191.0851 (C₂₀H₂₄NO₄-CH₂O-2CH₃-C₂H₄O₂-CH₃O)⁺. The fragment ion information was in accordance with that of magnoflorine, by comparison with that of magnoflorine, and it was detected as magnoflorine. For compound P5 (15.924 min, m/z 428.3139) (C₂₇H₄₁NO₃+H)⁺, the secondary fragment ion was m/z 412.3205 (C₂₇H₄₁NO₃+H-O)⁺. After comparison with the standard database information, the fragment ion information was roughly consistent with that of peimisine, so it was speculated that it might be peimisine. For compound P6 (16.551 min, m/z 432.3460) (C₂₇H₄₅NO₃)⁺, the main secondary fragment ions were m/z 414.3358 (C₂₇H₄₅NO₃-H₂O)⁺, m/z 398.3054 (C₂₇H₄₅NO₃-H₂O-O)⁺, and m/z 299.2351 (C₂₇H₄₅NO₃-H₂O-O-C₆H₁₁O)⁺. After comparison with the control substance, the fragment ion information was basically in line with that of the control substance and the fragment ion information was determined to be peimine. In terms of compound P7 (16.934 min, m/z 430.3308) (C₂₇H₄₃NO₃+H)⁺, the secondary fragment ion was m/z 412.3199 (C₂₇H₄₃NO₃+H-H₂O)⁺. After comparison with standard database information, fragment ion information was roughly consistent with that of zhebeinone, so it was speculated to be zhebeinone. As for compound P8 (17.553 min, m/z 430.3301) (C₂₇H₄₃NO₃+H)⁺, the secondary fragment ions were m/z 412.3199 (C₂₇H₄₃NO₃+H-H₂O)⁺, m/z 396.2898 (C₂₇H₄₃NO₃+H-H₂O-O)⁺, and m/z 175.1474 (C₂₇H₄₃NO₃+H-H₂O-O-C₁₅H₂₇N)⁺. After comparison with the control substance, it was confirmed that it was peiminine. For compound P9 (19.934 min, m/z 432.3461) (C₂₇H₄₅NO₃+H)⁺, the main secondary fragment ion is m/z 414.3359 (C₂₇H₄₅NO₃+H-H₂O)⁺. After comparison of the fragment ion information with the standard database, it was speculated to be zhebeinine. For compounds P10 (21.570 min, m/z 414.3359) and P11 (22.379 min, m/z 414.3357), (C₂₇H₄₃NO₂+H)⁺ was compared with standard database information and fragment ion information, and it was speculated to be ebeiedinone or zhebeirine. The specific association needs to be further confirmed by a reference substance. Compound PM1 with molecular ion peak m/z 169.0354 (C₅H₄N₄O₃+H)⁺, which is 32 Da more than the hypoxanthine excimer ion peak 137 (C₅H₄N₄O + H)⁺, prompted that O is added continuously and two oxygenation reactions occurred. Combined with the fragments m/z 152.0319 (C₅H₄N₄O₃+H-NH₃)⁺, m/z 141.0669 (C₅H₄N₄O₃+H-CO)⁺, and m/z 125.9855 (C₅H₄N₄O₃+H-CONH)⁺, it is speculated to be hypoxanthine oxidation products. Compound PM2 m/z 349.1822 (C₁₆H₁₂O₇S + H)⁺, which was found to be 80 Da more than the formononetin excimer ion peak 269 (C₁₆H₁₂O₄+H)⁺, suggested that the compound is a sulfate metabolite. Combined with fragment m/z 269.1024 (C₁₆H₁₂O₄+H)⁺, which is the prototype component of formononetin, it was speculated that the combination is formononetin sulfate. Furthermore, for compound PM3 with the molecular ion peak of m/z 565.2861, the main secondary fragment ion is m/z 403.1357 (C₂₁H₂₂O₈+H)⁺, which is the excimer ion peak of nobiletin. The molecular weight of this substance in primary mass spectrometry is 162 Da more than that of nobiletin. It may be the product of glucuronidation and demethylation of nobiletin. Compound PM4 with m/z 461.1066, which is 176 Da more than that of calycosin excimer ion peak 285, demonstrated that the compound is a metabolite of glucoaldehyde acidification and the corresponding molecular formula is C₂₂H₂₀O₁₁. Combined with fragments m/z 285.0753 (C₁₆H₁₂O₅+H)⁺, m/z 270.0509 (C₁₆H₁₂O₅+H-CH₃)⁺, and m/z 125.9863 (C₁₆H₁₂O₅+H-C₉H₄O₃)⁺, m/z 285.0753 is the prototype component of the compound calycosin, so that the compound is speculated to be glucuronized calycosin.

In the negative ion mode, for compound N1 (11.495 min, m/z 593.1872) (C₂₈H₃₄O₁₄-H)^-, the secondary fragment ions were m/z 510.2635 (C₂₈H₃₄O₁₄-C₄H₄O₂)^-, m/z 285.0783 (C₂₈H₃₄O₁₄-C₁₂H₂₁O₇)^-, and m/z 96.9587 (C₂₈H₃₄O₁₄-C₂₀H₂₅O₁₂)^-. After comparison with vanillin, the fragment ion information was basically consistent and it was identified as vanillin. For compound N2 (14.287 min, m/z 301.0724) (C₂₈H₃₄O₁₄-H)^-, the main secondary fragment ions were m/z 136.0171 (C₂₈H₃₄O₁₄-C₂₀H₂₆O₁₂)^-, m/z 108.0223 (C₂₈H₃₄O₁₄-C₂₁H₂₆O₁₃)^-, and m/z 65.0039 (C₂₈H₃₄O₁₄-C₂₃H₂₉O₁₄)^-. After comparison with hesperetin, the fragment ion information was consistent with that of hesperetin. Therefore, it was identified as hesperetin. For compound N3 (17.067 min, m/z 283.0619) (C₁₆H₁₂O₅-H)^-, the secondary fragment ions were m/z 239.0344 (C₁₆H₁₂O₅-C₂H₅O)^-, m/z 163.0039 (C₁₆H₁₂O₅-C₇H₅O₂)^-, m/z 135.0099 (C₁₆H₁₂O₅-C₇H₅O₂-CO)^-, and m/z 110.0007 (C₁₆H₁₂O₅-C₇H₅O₂-CO-C₂H₅)^-. After comparison with the control substance, the fragment ion information was in line with that of wogonin. It was finally determined to be wogonin. For compound N4 (20.129 min, m/z 265.1250) (C₁₈H₁₈O₂-H)^-, the secondary fragment ions were m/z 249.0936 (C₁₈H₁₈O₂-OH)^-, m/z 223.0771 (C₁₈H₁₈O₂-C₃H₇O)^-, and m/z 197.0617 (C₁₈H₁₈O₂-C₄H₅O₂)^-. After comparison with the honokiol control substance, fragment ion information was basically consistent with that of honokiol and it was identified as honokiol. For compound N5 (21.576 min, m/z 265.1248) (C₁₈H₁₈O₂-H)^-, the main secondary fragment ions were m/z 223.0767 (C₁₈H₁₈O₂-H-C₃H₅)^-, m/z 204.0577 (C₁₈H₁₈O₂-H-C₃H₅-OH)^-, and m/z 119.0501 (C₁₈H₁₈O₂-C₉H₇O₂)^-. After comparison with the reference substance, it was identified as magnolol. For compound N6 (29.296 min, m/z 279.2328) (C₁₈H₃₂O₂-H)^-, no fragment ions were produced at 40 V. After comparison with the linoleic acid reference substance, it was determined to be linoleic acid. The molecular ion peak of compound NM1 was m/z 167.0215, which is the methyl group, 14 Da more than protocatechuic acid. When combined with the fragment m/z 122.4525 (C₇H₆O₄-2O)^- and according to the literature (Guo et al., 2019) as well as database comparison, it was speculated as a methylated product of protocatechuic acid. For compound NM2 with m/z 273.0065, which increased by 94 Da (M + CH₂+SO₃) compared with caffeic acid (C₉H₈O₄), according to the comparison of literature studies and databases, it was speculated that the compound was 3-methoxy-caffeine-4-O-sulfate. Moreover, the basic ion peak of compound NM3 was m/z 259.1026, 80 Da more than that of caffeic acid 179, suggesting that M + SO₃ was increased. Beyond that, the corresponding molecular formula was C₉H₈O₇S, and in comparison with the characteristic ion of caffeic acid with m/z 179.0346, it was speculated to be caffeic acid-3-O-sulfate.

According to Table 2 and Figure 6, betaine, codonopsinol B, magnoflorine, peimisine, peimine, zhebeinone, peiminine, zhebeinine, ebeiedinone, zhebeirine, vanillin, hesperetin, honokiol, magnolol, and linoleic acid were all absorbed into the blood as the prototype under physiological and pathological conditions. Taking the variation of abundance greater than 1.5 times as the statistical standard, nine of the prototype components increased in the blood, five decreased, and one was not detected under the pathological state, while honokiol and linoleic acid did not change.

It can be seen from Table 3 and Figure 7 that there are seven metabolic components that change under the pathological state compared with the physiological state. Among them, five components increase in the blood concentration under the pathological state, while one metabolite decreases in the blood concentration and hypoxanthine oxidation products are not detected under the pathological state, which may not affect the absorption of the aforementioned components under the pathological state.