The BFHX capsule consists of Astragalus exscapus L., Paeonia lactiflora Pall, Psoralea aphylla L., Paeonia lactiflora Pall, and Psoralea aphylla L. Astragalus exscapus L. is dry root of Astragalus membranaceus (Fisch.) Bge. var. mongholicus (Bge.) Hsiao and Paeonia lactiflora Pall is dry foot of Paeonia lactiflora Pall. In addition, Psoralea aphylla L. is dry and contains mature seeds of Psoralea corylifolia L.

Approximately 720 g of Astragalus exscapus L., 540 g of Paeonia lactiflora Pall, and 360 g of Psoralea aphylla L. were boiled with 8 times the amount of water for the first time and 6 times the amount of water for the second time, for 1 h, respectively. The combined decoction was filtered and concentrated to a relative density of 1.05–1.15 (80°C); next, ethanol was added to the decoction to achieve a 60% alcohol content for 24 h, ethyl alcohol was recycled, and the supernatant was taken to concentrate to a relative density of 1.35–1.40 (80°C). A measure of 180 g of Paeonia Lactiflora Pall was ground to fine powder, and then the powder was mixed well with the supernatant and ground to fine powder. The precipitation was collected and washed with 90% ethanol in turn, and then it was dried routinely to obtain the powder.

BFHX preparation and quality control were conducted as described previously (Li et al., 2023). The C-18 column (150 mm × 4.6 mm, Agilent, Italy) was equipped with a guard column for BFHX capsules and standard solutions. Three mobile phases were conducted as follows: (1) an aqueous solution containing water/acetonitrile (v/v, 68:32) (Astragalus exscapus L.), (2) an aqueous solution containing 0.05 mol/L potassium dihydrogen phosphate/methanol (v/v, 60:40) (Paeonia lactiflora Pall), and (3) an aqueous solution containing water/methanol (v/v, 52:48) (Psoralea aphylla L.).

Six- to eight-week-old male C57 mice were purchased from Charles River (Beijing, China) and kept under SPF conditions (12 h light/dark cycle) at a temperature of 24°C ± 2°C and 50%–60% humidity. Mice were provided water and food freely. The animal experiment was approved by the Animal Health Research Institute of Guangdong Academy of Agricultural Sciences (ethics no. SPF2020034). After a 1-week quarantine period, mice were randomly divided into the following groups: 1) control group (CTL), 2) COPD model group (MOD), 3) COPD + roflumilast (POS group, 0.5 mg/kg/d), 4) COPD + BFHX low-dose group (low group, 1.9 g/kg/d, based on the clinical dose), and 5) COPD + BFHX high-dose group (high group, 3.8 g/kg/d), n = 9 per group. All groups, except the CTL group, received cigarette smoke (CS) exposure and lipopolysaccharide (LPS) instillation, as described previously (Guan et al., 2020). Mouse airways were instilled with PBS or LPS (75 μg in 50 μL) on experimental days 1 and 14; all mice, except the CTL group, received CS exposure for 4 h with 4 h intervals. The animals were orally administered on the 61st day of the experiment for 30 days, the weights of the animals were recorded, and their states, including their hair status and movement, were observed. After administration, the animal was harvested for further analysis (n = 5–6 mice per group were analyzed).

The small animal pulmonary function instruction-forced pulmonary maneuver system (DSI, CA, United States) was used to evaluate the pulmonary function of mice. Mice were anesthetized with 1.2% avertin (0.2 mL/10 g) by intraperitoneal injection. The trachea of each mouse was cannulated and placed in the body chamber, before the pulmonary function of mice was recorded. Lung function readings were excluded from the analysis after quality control.

BALF was collected by injection and aspiration of 0.8 mL of sterile saline solution repeated six times. The BALF was centrifuged at 800 rpm for 10 min at 4°C, and the supernatant was collected and stored at −80°C for cytokine assessment. The number of inflammatory cells in the BALF was counted, and the cell morphology was evaluated with hematoxylin–eosin (HE) staining. At least 200 cells on the slide were classified in a blinded manner.

The left lungs of the mice were inflated with paraformaldehyde fixative and excised at a 20-cm H₂O pressure. Subsequently, the lung tissues were embedded in paraffin, cut, dehydrated, and subjected to H&E staining. The mean linear intercept (MLI) of the lung tissue was measured. Periodic acid-Schiff (PAS) staining was performed to access the level of goblet cell hyperplasia, Masson’s staining was performed to evaluate collagen deposition, and α smooth muscle actin alpha (αSMA) IF staining was conducted to evaluate the expression level of the lung airway.

The concentrations of monocyte chemoattractant protein-1 (MCP1) and KC (functional IL-8 in mice) in the BALF supernatant were detected using ELISA kits, according to the respective manufacturer’s instructions. Briefly, the BALF supernatant was added to the ELISA plates pre-coated with capture antibodies and incubated overnight at 4°C. The ELISA plates were then washed with phosphate-buffered saline (PBS) and incubated with anti-detection antibodies at room temperature. Streptavidin-conjugated HRP was added to each well, followed by the enzyme-substrate TMB, and incubated for 10 min. The enzyme reaction was stopped by adding the 2N H₂SO₄ solution. The absorbance was read at 450 and 570 nm using a microplate reader (Gene, Hongkong, China). The concentrations of the investigated cytokines in the BALF were determined using standard curves.

Frozen lungs were lysed in lysis buffer containing the protease inhibitor PMSF (100:1). Lysates were cleared by centrifugation at 4°C for 30 min, and the total protein concentration of tissues was determined using a BCA kit (Beyotime Biotechnology, Wuhan, China). Next, 10% SDS PAGE and Western blotting were performed using a standard method, and the proteins were transferred to a polyvinylidene difluoride (PVDF) membrane (Millipore, Billerica, MA) by Western blotting (Bio-Rad Laboratories, Hercules, CA). Membranes were blocked with 5% milk and incubated with primary antibodies against β-actin, HO1, NQO1 (all from Abcam), and NOX4 (Servicebio, Wuhan, China) at 4°C overnight. The membranes were then washed in TBST followed by incubation with goat anti-rabbit/mouse and horseradish peroxidase (HRP)-linked secondary antibodies, and the bands were detected using an enhanced chemiluminescence substrate (Tanon, Shanghai, China). The intensity of each band was determined using IPP software and was corrected to the intensity of β-actin.

Immunofluorescence analysis was performed with the following steps: after citrate antigen was repaired, the lung sections were blocked with 10% bovine serum albumin (BSA), subjected to incubation at 4°C overnight with primary antibodies. On the second day, the slides were incubated with the fluorescent secondary antibody at 37°C for 1 h, and then the sections were restrained with 4′,6′-diamino-2-phenylindole (DAPI) for nuclear location. The slides were observed under an upright fluorescence microscope (Leica, Germany).

Total fecal DNA was extracted, and 1% agarose gel electrophoresis was performed to check the extracted genomic DNA. The 16S rRNA gene V3–V4 region was amplified with the following primers (forward primer, 5′-ACT​CCT​ACG​GGA​GGC​AGC​AG-3′; reverse primer, 5′-GGACTACHVGGGTWTCTAAT-3′) by polymerase chain reaction (PCR) (ABI 3 GeneAmp^® 9700, ABI, United States) before using QuantiFluor™-ST (Promega, United States) to quantify the PCR products, which were paired-end sequenced on an Illumina MiSeq instrument (Illumina, United States).

Operational taxonomic units (OTUs) were divided according to different similarity levels, and bio-information from OTUs with a similarity level >97% was analyzed using the RDP classifier (V2.13) and compared with the SILVA database (V138). In the diversity analysis of samples, alpha diversity reflects the richness and diversity of the microbial community, and mothur (V1.30.2) was used to analyze the species richness and diversity of the intestinal fora. R software was employed to create curves and analyze PCA graphics. LEfSe analysis was used to analyze the correlation between OTU and gate-level annotation.

Feces were collected from the rectums of mice into tubes with sterile scissors and tweezers, stored at −80°C, and then analyzed using a liquid chromatography–tandem mass spectrometry (LC–MS) system according to machine orders. Information from LC–MS was matched with metabolic public databases HMDB (http://www.hmdb.ca/) and KEGG (http://www.genome.jp/kegg/compound/) to obtain the classification of the metabolites. The annotated metabolites were then mapped to the KEGG pathway database (http://www.kegg.jp/kegg/pathway.html), and groups of differential metabolites were analyzed. Differential metabolite screening was performed in both groups with VIP (VIP >1) and p-value (p < 0.05).

SPSS v.25.0 statistical software (IBM Corporation, USA) was used for statistical analyses. All data are represented as mean ± SEM. All assays were performed in at least three independent experiments. The statistical significance was calculated by a one-way ANOVA, which was conducted to analyze quantitative data. Dunnett’s post hoc test was used after one-way ANOVA analysis, for which p < 0.05 was considered significant.

During the experiment, the status of mice was observed, and the weight of mice was recorded. CTL group mice had normal movement, stable breathing, and shiny hair, whereas CS exposure mice had decreased activity, rapid respiration, and ruffled hair. Compared with the CTL group, CS exposure mice had slowed weight gain, and after the administration, there was no significant increase in weight gain (Supplementary Figure S1). To determine whether BFHX is harmful to other systems, the organ index (liver, spleen, and kidney) was tested. The results revealed no significant difference between the MOD and BFHX groups (Supplementary Figure S2). Airway inflammation and emphysema are the main pathological characteristics of COPD. Mice exposed to CS exposure displayed inflammation accumulation, destroyed alveoli, and increased MLI in lung tissue compared with the MOD group. The BFHX low and high groups showed reduced MLIs, indicating that BFHX reduced CS-induced emphysema (Figure 1).

To determine the effect of BFHX on COPD mice, we performed pulmonary function testing and pulmonary function test after administration. As shown in Figures 2A–D, CS exposure significantly increased the functional residual capacity (FRC), percentage of forced vital capacity in 100 milliseconds (FEV100%), total lung capacity (TLC), and inspiratory capacity (IC) in the CS group. These results indicate that BFHX significantly improved the pulmonary function of COPD mice.

CS induced accumulation inflammation in the cells of the lung, especially macrophages, which are the major cells in lung tissues. Acute inflammation induces acute exacerbation and promotes the progression of COPD. Consistent with the report (Arora et al., 2018), in our study, CS exposure induced inflammation accumulation in BALF (Figure 3A), predominantly macrophages (Figure 3B), whereas BFHX decreased the number of total inflammatory cells and cytokines (Figures 3E, F), indicating that BFHX suppressed CS-induced inflammation.

As mucus promotes COPD progression and airway goblet cells produce mucus with CS exposure, we next evaluated the airway goblet cell hyperplasia of COPD mice. As shown in Figure 4, PAS staining of lung tissue revealed increased and decreased goblet cells in the airways of the MOD and BFHX groups, respectively.

Collagen deposition induces airway obstruction, which is a pathological characteristic of COPD. Therefore, we next conducted Masson’s staining of lung tissue to access collagen deposition. As shown in Figure 5, BFHX decreased collagen deposition in the airways of COPD mice. We also investigated the level of αSMA in the airways of COPD mice, as shown in Figure 5B. BFHX reduced the level of αSMA induced by CS.

CS exposure can cause oxidative stress accumulation and induce chronic lung inflammation. Excessive reactive oxygen species (ROS) from CS-induced oxidative stress accelerates lung inflammation and injury in COPD, both of which produce endogenous ROS from cellular sources such as dihydronicotinamide-adenine dinucleotide phosphate (NADPH) oxidase (NOX). As shown in Figure 6, BFHX decreased the level of NOX4 induced by CS exposure and LPS. In addition, the levels of antioxidant enzymes and phase 2 detoxifying enzymes, such as NADP(H): quinone oxidoreductase (NQO1) and the stress response protein heme oxygenase (HO-1), were increased in COPD mice and were found to be downregulated by BFHX.

Next, the diversity of mice was detected following the administration of BFHX. The obtained dilution curve gradually fattened, indicating a sufficient amount of sequencing data and demonstrating the rationality of the sequencing data and, indirectly, the species richness (Supplementary Figure S3). The good coverage of each group was >99.0%, indicating that the sequencing depth of the microbiome analysis meets the requirements (Supplementary Figure S4). The α diversity of the intestinal flora includes the richness and diversity of microbial communities; the Sobs index was used to evaluate the richness, and the Simpson and Shannon indices were used to evaluate the diversity of the intestinal flora of COPD mice. As shown in Figure 7A, the richness of flora in the MOD group was lower, but it was increased by the BFHX administration. Figures 7B, C shows that the diversity of the bacterial population (Simpson and Shannon indices) in the MOD group was significantly lower than that in the CTL group, whereas BFHX reversed both the diversity and richness.

The beta diversity of each group was calculated based on cluster tree analysis, principal component analysis (PCA), and non-metric multidimensional analysis (NMDS). The PCA and hierarchical clustering tree (Figure 7D) were used to explore the similarities and differences in intestinal flora composition in each group. The results showed that the bacterial structure of the CTL group was separated from that of the MOD group and the BFHX administration group was similar to that of the MOD group. Based on the first two PCA scores, the total variations accounted for 15.39% and 9.84%, suggesting that COPD may change the percentages of bacteria in intestinal flora, and BFHX could improve bacterial composition (Figure 7E). The NMDS of the bacterial community also showed that BFHX could improve bacterial percentage (Figure 7F).

Next, a heatmap was constructed based on the top 30 dominant genera using the SILVA database. The results shown in the heatmap revealed that all 30 genera belong to six main phyla: Firmicutes, Bacteroidetes, Actinobacteria, Verrucomicrobiota, Desulfobacterota, and Proteobacteria. At the phylum level, most of the dominant genera in the CTL group belonged to Bacteroidetes, and most of the dominant genera in the MOD group belonged to Firmicutes (Figure 8A). In addition, the proportion of Verrucomicrobiota and Desulfobacterota increased while that of Actinobacteria decreased in the MOD group compared with the CTL group. The results revealed that BFHX reversed the proportion of bacteria at the phylum level. At the genus level, the proportions of Murbaculaceae, Lactobacillus, Desulfuribrio, Clostridiumthe, and Akkermansia in the CTL group decreased sharply compared to those in the MOD group, whereas the proportions of Lactobacillus, Desulfuribrio, and Bacteroidetes increased substantially, indicating that BFHX reversed the proportions (Figures 8B–D).

To confirm the species differences at multiple levels, including the phylum, class, order, family, genus, and species levels, LEfSe was used to identify specialized communities in flora with LDA scores ≥2, as shown in Figure 9. From the phylum to the genus level, the number of dominant bacteria in each group were as follows: 33 in the CTL group, 14 in the MOD group, 26 in the low group, and 15 in the high group. In the CTL group, Bacteroidota dominated, whereas in the MOD, low, and high groups, Firmicutes dominated.

Next, non-targeted detection of mice fecal metabolites was performed based on LC–MS. As a result, 1784 fecal metabolites were identified in the CTL group, 1791 metabolites in the CS group, 1790 metabolites in the mid group, and 1794 metabolites in the high group. In addition, there were 17 metabolites in the CTL, mid, and high groups, and 20 metabolites in the MOD, mid, and high groups. According to different biological roles, all metabolites were classified as amino acids, vitamins and cofactors, nucleic acids, hormones and transmitters, carboxylic acids, antibiotics, steroids, organic acids, lipids, and carbohydrates. As shown in Figure 10B, PLS-DA analysis showed that the CTL, mid, and high groups’ data points were clearly separated from those of the MOD group. According to the PLS-DA analysis, the fecal metabolites of mice from the high group were much closer to those of the CTL group, and lysine degradation and phenylalanine metabolism were significantly altered. The differential metabolites between each group were screened by p-value and VIP bar graph (Figures 10D–F).