Integrated Metabolomic and Gut Microbiome Analyses Reveal the Therapeutic Effects of Xuanfei Heji in Rats With Chronic Obstructive Pulmonary Disease

ABSTRACT

Background

Chronic obstructive pulmonary disease (COPD) is a leading cause of death, underscoring the need for improved therapies. Xuanfei Heji (XFHJ), a hospital‐prepared herbal formula, has been used clinically in the treatment of COPD. However, its mechanisms remain unclear.

Methods

XFHJ constituents were profiled using UHPLC–HRMS. COPD was induced in rats by intratracheal lipopolysaccharide instillation and cigarette smoke exposure. Treatment effects were assessed using pulmonary function, lung histopathology, and proinflammatory cytokines. Untargeted serum metabolomics and fecal 16S rRNA gene sequencing were performed; associations among differential metabolites, microbial taxa, and inflammatory markers were evaluated using Spearman's rank correlation analysis.

Results

Chemical profiling tentatively identified 374 constituents. XFHJ improved pulmonary function and attenuated lung histopathological injury and inflammation. Tryptophan and glycerophospholipid metabolism were the principal treatment‐associated pathways. XFHJ also altered gut microbial diversity and composition, with enrichment of potentially beneficial taxa such as Bifidobacterium, Roseburia, and several Clostridia‐related taxa. Treatment‐responsive taxa correlated positively with indole‐related metabolites, which correlated inversely with pulmonary inflammatory markers.

Conclusions

XFHJ exhibited significant therapeutic effects on COPD rats, and its mechanism may be correlated with regulating the intestinal microbiota structure and metabolic profiles of COPD rats, thereby attenuating lung histopathological injury and pulmonary inflammation.

1. Introduction

Chronic obstructive pulmonary disease (COPD) is a highly prevalent chronic respiratory disorder defined by persistent respiratory symptoms secondary to irreversible airflow limitation [1]. Inhalation of toxic particulates, including tobacco smoke and ambient air pollutants, constitutes a primary risk factor for COPD pathogenesis, and advanced age is another critical factor that contributes to the disease. Elderly individuals demonstrate elevated susceptibility to COPD, which is ascribed to prolonged smoking histories, cumulative exposure to other risk factors, and age‐associated pulmonary function deficits [2]. Given the global demographic shift toward aging populations and the ongoing rise in smoking rates across developing countries, the incidence and prevalence of COPD are expected to increase steadily. The Global Burden of Disease Study reported that COPD caused 3.3 million deaths worldwide in 2019, accounting for 5% of all deaths and ranking as the third leading cause of death globally [3]. An independent study confirmed that COPD‐associated deaths accounted for approximately 85.25% of all deaths attributable to chronic respiratory diseases in 2021. The public health burden of COPD is particularly pronounced in China, other Asian regions, southern sub‐Saharan Africa, Oceania, and other developing economies [4].

Current management of COPD combines risk‐factor reduction, nonpharmacological interventions, and individualized pharmacotherapy. Smoking cessation is the principal measure for limiting ongoing exposure‐related lung injury, whereas vaccination, regular physical activity, patient education, and pulmonary rehabilitation are recommended to prevent respiratory infections, reduce dyspnea, and improve exercise capacity and health‐related quality of life [5]. However, the effectiveness of these interventions is limited by inadequate access, poor adherence, and difficulty sustaining long‐term behavioral change. Pharmacotherapy centers on inhaled bronchodilators, including long‐acting muscarinic antagonists and long‐acting β ₂‐agonists, which improve airflow, relieve symptoms, and reduce exacerbation risk; dual bronchodilator therapy generally provides greater clinical benefit than monotherapy in patients with persistent symptoms [6]. Treatment responses vary according to inhaler technique, adherence, disease phenotype, and comorbidities. Inhaled corticosteroids are primarily used in patients with recurrent exacerbations and evidence of corticosteroid‐responsive inflammation, particularly elevated blood eosinophil counts. Triple therapy combining an inhaled corticosteroid, a long‐acting muscarinic antagonist, and a long‐acting β ₂‐agonist can further reduce moderate or severe exacerbations in selected patients, but broader use is constrained by increased risks of pneumonia and other corticosteroid‐related adverse effects [7]. Additional treatments, such as phosphodiesterase‐4 inhibitors, long‐term macrolides, and oxygen therapy, are reserved for selected clinical phenotypes because of their restricted indications and potential adverse effects; notably, long‐term oxygen therapy benefits patients with severe chronic resting hypoxemia but does not provide sustained clinical benefit in those with only moderate desaturation [8]. Thus, current therapies control symptoms and reduce exacerbations but do not reverse established airflow limitation, alveolar destruction, or airway remodeling, and persistent inflammation, oxidative stress, and systemic metabolic abnormalities may remain inadequately controlled. These limitations support investigating adjunctive therapies that target multiple COPD‐related pathological processes.

COPD is assigned to the syndromic categories of Fei Zhang (pulmonary distension) and Chuan Bing (asthmatic disease), with its pathogenesis predominantly ascribed to the synergistic interplay of Tan Zhuo (phlegm turbidity), Shui Yin (fluid retention), and Xue Yu (blood stasis) in Traditional Chinese Medicine (TCM). In TCM theory, the lung is highly vulnerable to invasion by external pathogenic factors. During the progressive course of COPD, symptoms such as a worsening cough and wheezing, yellow and sticky phlegm, and fever may appear. The most common approach to treating COPD involves clearing heat and resolving phlegm. Tan Re (phlegm‐heat) is considered a persistent pathological state throughout the course of COPD and is regarded as central to disease pathogenesis, making the therapeutic principle of clearing heat and resolving phlegm the most commonly used clinical strategy for COPD management.

Xuanfei Heji (XFHJ) is a hospital‐prepared herbal formula developed by the Second Affiliated Hospital of Nanjing University of Chinese Medicine, consisting of nine medicinal components: Ephedrae Herba, Gypsum Fibrosum, Mori Cortex, Pinelliae Rhizoma, Armeniacae Semen Amarum, Scutellariae Radix, Fritillariae Cirrhosae Bulbus, Gardeniae Fructus, and Glycyrrhizae Radix et Rhizoma. XFHJ is an experience‐based modified formula derived from the classic TCM formula Ma‐Xing‐Shi‐Gan Decoction [9, 10] and is used as an institutional preparation at our hospital. Accumulated clinical experience and preliminary internal observations suggested its potential benefit in COPD, providing the clinical rationale for the present investigation. However, its efficacy has not been evaluated in published controlled clinical studies. The present study therefore provides an initial preclinical assessment of its effects and potential mechanisms in COPD.

By comprehensively detecting and identifying metabolites in complex biological matrices, untargeted metabolomics offers distinct advantages for discovering novel biomarkers and elucidating the underlying pathophysiological pathways [11]. This holistic analytical strategy has been increasingly integrated into TCM research, where it facilitates a systems‐level understanding of how herbal formulas prevent and treat diseases.

The interplay between pulmonary pathogenesis and gastrointestinal microbial communities, bridged by the gut–lung axis, has emerged as a critical focal point in COPD research [12, 13]. Cigarette smoking, the primary risk factor for COPD, exerts detrimental effects on both the respiratory and gastrointestinal tracts. Furthermore, smoke‐induced gas‐exchange impairment in patients with COPD may increase intestinal permeability, which in turn triggers gut dysbiosis and immune dysfunction. Inflammatory bowel disease (IBD) and irritable bowel syndrome are markedly more prevalent among patients with COPD [14]. Recent evidence suggests that TCM may exert multitarget therapeutic effects by reshaping gut microbial composition and regulating microbiota‐derived metabolites [15].

In this study, ultrahigh‐performance liquid chromatography coupled with Q Exactive HF‐X mass spectrometry (UHPLC‐Q Exactive HF‐X) was used to characterize the chemical constituents of XFHJ. A rat model of COPD was established via intratracheal instillation of lipopolysaccharide (LPS) combined with cigarette smoke (CS) exposure, and the effects of XFHJ on pulmonary function, lung histopathology, and inflammatory markers were assessed in COPD rats. Additionally, untargeted metabolomics and 16S rRNA gene sequencing were used to identify differential metabolites and microbial taxa and to elucidate potential associations among XFHJ‐responsive metabolites, the gut microbiota, and inflammatory markers.

2. Materials and Methods

2.1. Preparation of XFHJ

The herbal ingredients of XFHJ were supplied by the TCM Pharmacy of the Second Affiliated Hospital of Nanjing University of Chinese Medicine and were authenticated by Chief Pharmacist Haiyan Qian. The herbal materials were extracted twice using water as the extraction solvent. For the first extraction, a 12‐fold volume of water was added, followed by soaking for 0.5 h and then decoction for 1.5 h. For the second extraction, the residue was decocted with a 10‐fold volume of water for 1 h. The resulting extracts were filtered, pooled, and subsequently concentrated under reduced pressure to three final concentrations of 0.5‐, 1‐, or 2‐g crude‐herb equivalent/mL.

2.2. Analysis of the Chemical Constituents of XFHJ

The chemical constituents of the XFHJ extract were comprehensively profiled using a Vanquish UHPLC system coupled with a Q Exactive HF‐X mass spectrometer. The analytical conditions, including the chromatographic separation conditions and mass‐spectrometer parameters, are detailed in Table 1. The nonlinear mobile‐phase gradient program is provided in Table 2. Following data acquisition, raw MS files were processed using Progenesis QI software (Waters Corporation) for baseline filtering and peak alignment. Detected constituents were tentatively identified by systematically matching the experimental MS/MS spectra to an integrated mass‐spectral database, which was compiled by combining data from public repositories (MassBank, PubMed, Web of Science, and CNKI) and an in‐house mass spectral library.

TABLE 1.

UHPLC–MS instrumental parameters for the chemical constituents of XFHJ.

Parameter	Setting/value
UHPLC system	Vanquish UHPLC (Thermo Fisher Scientific, USA)
Mass spectrometer	Q Exactive HF‐X (Thermo Fisher Scientific, USA)
Chromatographic column	ACQUITY UPLC HSS T3 (100 × 2.1 mm, 1.8 μm; Waters)
Column temperature	40°C
Flow rate	0.3 mL/min
Injection volume	2 μL
Mobile Phase A	0.1% formic acid in water
Mobile Phase B	0.1% formic acid in acetonitrile
ESI spray voltage	+3000 V (positive‐ion mode)/−2800 V (negative‐ion mode)
Ion source temperature	350°C
Ion transfer tube temperature	320°C
Sheath gas flow rate	40 arb
Auxiliary gas flow rate	10 arb
Full‐scan mass range	m/z 70–1050
MS resolution	70 000 (Primary)/17500 (MS/MS)

TABLE 2.

Mobile‐phase gradient elution program for the chemical constituents of XFHJ.

Time (min)	Mobile Phase A (%)	Mobile Phase B (%)
0–1.0	100	0
1.0–12.0	100–5	0–95
12.0–13.0	5	95
13.0–13.1	5–100	95–0
13.1–17.0	100	0

2.3. Animals and Treatment

2.3.1. Animal Care and Experimental Groups

Healthy male Sprague–Dawley rats (specific pathogen‐free, aged 6 to 8 weeks) were supplied by Suzhou Sibeifu Biotechnology Co. Ltd. (License: SCXK [Su] 2022‐0006; Certification: A202508060739). All in vivo protocols received formal ethical clearance from the Institutional Animal Care and Use Committee at Nanjing University of Chinese Medicine (Approval No. A250301). The animals were housed under controlled laboratory conditions at 24°C ± 2°C with a 12‐h light/dark cycle and unrestricted access to standard rodent chow and water. All invasive procedures were performed under isoflurane anesthesia to minimize suffering. For intratracheal LPS administration, rats were anesthetized with isoflurane. At the designated endpoints, the rats were deeply anesthetized with isoflurane; biological samples were then collected, and the animals were euthanized by exsanguination.

Male rats were randomly assigned to six groups (n = 8) using a computer‐generated randomization sequence: the control group (CG), model group (MG), aminophylline group (AP), Xuanfei Heji low‐dose group (XFHJ‐L), Xuanfei Heji medium‐dose group (XFHJ‐M), and Xuanfei Heji high‐dose group (XFHJ‐H).

2.3.2. Animal Treatment

Before the study, the rats were acclimatized for 1 week. A COPD model was then established via intratracheal lipopolysaccharide (LPS, Merck KGaA, L6529) instillation combined with passive cigarette smoke (CS) exposure. Rats in all groups except the CG were exposed to passive cigarette smoke once daily in a custom‐made chamber (120 × 80 × 80 cm). Each 1‐h session used 30 cigarettes (Hongjinlong brand; tar content, 11‐mg/cigarette; nicotine content, 1.0‐mg/cigarette; carbon monoxide content, 11‐mg/cigarette). The total duration of CS exposure was 28 days. On Day 1 and Day 14 of the experiment, rats were anesthetized with isoflurane, immobilized, and slowly administered 100 μL of LPS solution (1.0 mg/mL) via intratracheal instillation. After instillation, the rats were held upright and rotated for 20 s to ensure uniform distribution of the LPS solution in the lungs. No CS exposure was performed on the days of LPS instillation. Rats in the CG received an equal volume of normal saline via intratracheal instillation on Day 1 and Day 14, without CS exposure throughout the experiment. On Day 27, all rats subjected to CS exposure and intratracheal LPS instillation showed weight loss and wheezing. Examination of six additional model‐validation rats showed decreased pulmonary function, disrupted lung‐tissue architecture, and inflammatory‐cell infiltration in H&E‐stained lung sections, as well as elevated levels of IL‐1β, IL‐6, and TNF‐α in serum, confirming the successful establishment of the COPD model.

Following the successful establishment of the COPD model, rats in the AP group received aminophylline (Shandong Xinhua Pharmaceutical, H37020351) by oral gavage at 0.054 g/kg/day. The adult clinical dose of XFHJ is 97 g of crude herbs/day. For a 60‐kg adult, this corresponds to 1.62 g crude‐herb equivalent/kg/day. Body‐surface‐area normalization used a human‐to‐rat conversion factor of 6.2 [16]. The calculated rat‐equivalent dose (1.62 × 6.2 = 10.044 g crude‐herb equivalent/kg/day) was rounded to 10 g/kg/day and designated as the medium dose. Accordingly, doses of 5‐, 10‐, and 20‐g crude‐herb equivalent/kg/day were selected as 0.5‐, 1‐, and 2‐fold the estimated rat‐equivalent clinical dose, respectively. The dosing solutions contained 0.5‐, 1.0‐, or 2.0‐g crude‐herb equivalent/mL and were administered at 10 mL/kg to deliver 5, 10, or 20 g/kg/day, respectively. Animals in the CG and MG received matched volumes of physiological saline. Treatments were administered by oral gavage each morning for 2 weeks.

2.4. Pulmonary Function Testing

To evaluate respiratory function, each animal was placed in the plethysmograph chamber of an EMKA noninvasive testing system. The rats were allowed to acclimatize until a stable breathing pattern was achieved. Once breathing had stabilized, respiratory data were recorded continuously using EMKA IOX software. The following functional indices were measured: tidal volume (TV), minute ventilation (MV), expiratory time (Te), peak inspiratory flow (PIF), peak expiratory flow (PEF), and expiratory flow at 50% tidal volume (EF50).

2.5. Collection of Samples

Sampling commenced exactly 12 h following the final drug administration. Fresh fecal samples were collected directly into sterile tubes, immediately snap‐frozen in liquid nitrogen, and stored at −80°C until 16S rRNA gene sequencing. Subsequently, the animals were placed under isoflurane anesthesia to facilitate blood collection from the abdominal aorta. The whole blood was left undisturbed at room temperature for 40 min to ensure complete coagulation. After centrifugation at 4000 rpm for 10 min at 4°C, the serum was aliquoted and stored at −80°C. Following blood collection, the intact lungs were immediately isolated. The left lobes were processed into tissue homogenates, whereas the right lobes were fixed in 4% paraformaldehyde for subsequent histological examination.

2.6. Pulmonary Histopathology

Following a 24‐h fixation period in 4% paraformaldehyde, the right lung samples were dehydrated through a graded ethanol series, embedded in paraffin, and serially sectioned. To assess general tissue architecture, the sections underwent routine hematoxylin and eosin (H&E) staining. To quantify structural damage, the mean linear intercept (MLI) was calculated using Image‐Pro Plus software from three randomly selected microscopic fields per tissue section. A central crosshair was digitally overlaid on each field, and the MLI was calculated as L/NS, where L is the total line length and NS is the number of alveolar septal intersections. In parallel, adjacent sections were processed with Masson's trichrome stain to visualize airway collagen accumulation. The same image‐analysis software was used to quantify the proportion of the area occupied by deposited collagen fibers.

2.7. Measurement of Interleukin‐1β (IL‐1β), Interleukin‐6 (IL‐6), and Tumor Necrosis Factor‐α (TNF‐α) in Lung Tissue

To quantify the local inflammatory response, commercial enzyme‐linked immunosorbent assay (ELISA) kits were used to measure IL‐1β, IL‐6, and TNF‐α concentrations in lung homogenates. Before cytokine quantification, total protein concentrations in each lung‐tissue sample were measured using a bicinchoninic acid (BCA) assay to normalize the cytokine data.

2.8. Metabolomic Analysis

2.8.1. Sample Preparation

An aliquot of 400 μL of extraction solvent (methanol/water, 4:1, v/v) containing the internal standard 2‐chloro‐L‐phenylalanine (0.02 mg/mL) was added to the serum sample. After brief vortex mixing, the mixture was sonicated in an ice bath for three cycles of 10 min each. The sample was then incubated at −20°C for 30 min and centrifuged (13 000 ×g, 4°C, 15 min). The resulting clear supernatant was collected for instrumental analysis. A quality‐control (QC) sample was prepared by pooling 20 μL of supernatant from each sample.

2.8.2. LC–MS Analysis Conditions

Serum metabolic profiles were acquired using a Vanquish UHPLC‐Orbitrap Exploris 240 system. Instrument settings, internal standard, and 8‐min gradient programs are provided in Table 3 and Table 4. For system conditioning and stability monitoring, a pooled quality‐control (QC) sample was injected after every eight study samples throughout the analytical run.

TABLE 3.

UHPLC–MS instrumental parameters for serum metabolomics analysis.

Parameter	Setting/value
UHPLC system	Vanquish UHPLC (Thermo Fisher Scientific, USA)
Mass spectrometer	Orbitrap Exploris 240 (Thermo Fisher Scientific, USA)
Chromatographic column	ACQUITY UPLC HSS T3 (100 × 2.1 mm, 1.8 μm; waters)
Column temperature	40°C
Flow rate	0.40 mL/min
Injection volume	3 μL
Mobile Phase A	0.1% formic acid in 95% water and 5% acetonitrile
Mobile Phase B	0.1% formic acid in 47.5% acetonitrile, 47.5% isopropanol, and 5% water
Internal standard	2‐Chloro‐L‐phenylalanine
Lock mass (positive mode)	m/z 200.0472
Lock mass (negative mode)	m/z 198.0323

TABLE 4.

Mobile‐phase gradient elution programs for serum metabolomics analysis.

(a) Positive‐ion mode
Time (min)	Mobile Phase A (%)	Mobile Phase B (%)
0–3.0	100–80	0–20
3.0–4.5	80–65	20–35
4.5–5.0	65–0	35–100
5.0–6.3	0	100
6.3–6.4	0–100	100–0
6.4–8.0	100	0

(b) Negative‐ion mode
Time (min)	Mobile Phase A (%)	Mobile Phase B (%)
0–1.5	100–95	0–5
1.5–2.0	95–90	5–10
2.0–4.5	90–70	10–30
4.5–5.0	70–0	30–100
5.0–6.3	0	100
6.3–6.4	0–100	100–0
6.4–8.0	100	0

2.8.3. Data Processing and Quality Control

Raw‐data conversion, baseline filtering, and peak integration were performed using Progenesis QI v3.0 (Waters Corporation). Metabolites were identified by matching accurate masses (mass‐error tolerance, 10 ppm) and MS/MS fragmentation patterns against the METLIN and HMDB databases and a custom spectral repository. During the preprocessing phase, variables with a missing‐value rate exceeding 20% in any group were discarded, and half‐minimum values were used to impute the remaining missing values. Peak intensities were subsequently normalized to the internal standard (2‐chloro‐L‐phenylalanine). To ensure analytical robustness, features with a relative standard deviation (RSD) > 30% in the pooled QC runs were removed. The remaining data were then log10‐transformed to approximate a Gaussian distribution for downstream multivariate modeling.

2.9. Gut Microbiota Analysis

2.9.1. DNA Extraction, PCR Amplification, and Library Preparation

Genomic DNA was extracted from fecal pellets, and the V3‐V4 hypervariable region of the 16S rRNA gene was amplified in triplicate using uniquely barcoded 338F/806R primers (sequences are provided in Table S2). Triplicate PCR products per sample were pooled, run on a 2% agarose gel, and recovered using a Tris–HCl elution kit. Following fluorometric quantification, the purified DNA was pooled in equimolar amounts. Sequencing libraries were prepared according to the TruSeq kit instructions (Illumina), including adapter ligation, secondary purification, and NaOH denaturation. Library size and concentration were assessed using an Agilent 2100 Bioanalyzer and qPCR, respectively, before 300‐bp paired‐end sequencing on an Illumina NextSeq 2000 platform; the required library concentration was ≥ 2 nM.

2.9.2. Data Processing

Raw reads were processed using the QIIME pipeline (v1.9.1). fastp (v0.23.4) was used for adapter trimming and low‐quality‐read removal (Phred < 20), whereas FLASH (v1.2.11) merged the clean pairs into 3 830 347 tags (mean length 413 bp). The UPARSE algorithm (v11) was used to cluster sequences into operational taxonomic units (OTUs) at a ≥ 97% identity threshold, and UCHIME was used to detect and remove chimeric sequences. For taxonomic assignment, representative OTU sequences were classified against the SILVA v138 database using the RDP classifier at a confidence threshold of 0.7. All analytical workflows were performed on the Majorbio Cloud Platform.

2.10. Statistical Analysis

Data visualization and hypothesis testing were performed using GraphPad Prism 9.0. Results are presented as the mean ± SD. Prior to significance testing, data were evaluated for normality and homogeneity of variance using the Shapiro–Wilk and Brown–Forsythe tests, respectively. Subsequently, between‐group differences were assessed using Student's t‐test for two‐group comparisons and one‐way analysis of variance followed by Tukey's post hoc test for multiple‐group comparisons.

2.10.1. Statistical Analysis of Metabolomic Data

Metabolomic data were Pareto‐scaled before the construction of PLS‐DA and OPLS‐DA models, each of which was validated using 200 permutation tests. Metabolites with a variable importance in projection (VIP) score > 1 and a p < 0.05 were defined as biomarkers and subsequently mapped to KEGG pathways using the MetaboAnalyst web tool.

2.10.2. Statistical Analysis of Gut Microbiota Data

For microbiome datasets, differences in α‐diversity were assessed using the Wilcoxon rank‐sum test. Between‐sample differences in community structure (β‐diversity) were visualized using principal coordinate analysis (PCoA) based on Euclidean distances, and overall differences in community structure were tested using the Kruskal–Wallis test. Linear discriminant analysis effect size (LEfSe) was then used to identify taxonomic biomarkers (from phylum to genus) distinguishing the experimental groups, applying an LDA score threshold > 2 and p < 0.05.

3. Results

3.1. Chemical Constituents of XFHJ

The components were tentatively identified by searching relevant databases and the literature and by interpreting MS and MS/MS fragmentation patterns. Representative total ion chromatograms of XFHJ acquired in both positive‐ and negative‐ion modes are shown in Figure 1A,B, and the detailed information is provided in Table S1. A total of 374 chemical constituents were tentatively identified, of which 180 were detected in negative‐ion mode and 194 in positive‐ion mode. The identified constituents were classified as follows [17]: 44 prenol lipids (11.76%), 78 flavonoids (20.86%), 42 organooxygen compounds (11.23%), 32 fatty acyls (8.56%), 14 coumarins and derivatives (3.74%), 12 benzene and substituted derivatives (3.21%), 21 carboxylic acids and derivatives (5.61%), 16 isoflavonoids (4.28%), 13 cinnamic acids and derivatives (3.48%), and 102 other compounds (27.27%) (Figure 1C).

FIGURE 1.

(A,B) Representative UHPLC–HRMS chromatograms of XFHJ acquired in positive‐ and negative‐ion modes, respectively. (C) Distribution of the 374 tentatively identified constituents among the major chemical classes. (D) Pulmonary function parameters. (E) Representative hematoxylin and eosin–stained lung sections. Scale bars, 50 μm. Data are presented as mean ± SD (n = 8). ^# p < 0.05, ^## p < 0.01, ^### p < 0.001, and ^#### p < 0.0001 versus the Control group; *p < 0.05 and **p < 0.01 versus the Model group.

3.2. Effects of XFHJ on Pulmonary Function and Lung Histopathology in COPD Model Rats

Significant differences were detected among the groups for PIF (F = 5.175, p < 0.01), PEF (F = 6.769, p < 0.001), MV (F = 6.356, p < 0.001), TV (F = 4.742, p < 0.01), Te (F = 6.553, p < 0.001), and EF50 (F = 6.643, p < 0.001). Compared with the CG, the MG exhibited significantly reduced PIF (p < 0.01), PEF (p < 0.001), MV (p < 0.0001), TV (p < 0.05), and EF50 (p < 0.0001), whereas Te was significantly prolonged (p < 0.001), indicating comprehensive pulmonary function impairment. Compared with the MG, the AP group showed significant increases in PIF, PEF, TV, and EF50 (all p < 0.05), a numerical increase in MV, and a significant shortening of Te (p < 0.01). In the XFHJ‐M group, PIF (p < 0.01), PEF (p < 0.01), MV (p < 0.01), TV (p < 0.05), and EF50 (p < 0.05) were significantly increased, and Te was significantly shortened (p < 0.05), with most parameters approaching control levels. In the XFHJ‐H group, PIF (p < 0.05), PEF (p < 0.05), and EF50 (p < 0.05) were significantly increased, Te was significantly shortened (p < 0.05), and the remaining parameters improved to varying degrees. No significant improvements were observed in the XFHJ‐L group (Figure 1D). These results suggest that both XFHJ‐M and XFHJ‐H confer a comprehensive protective effect against pulmonary function impairment in COPD rats.

Hematoxylin and eosin (H&E) staining (Figure 1E) revealed that, compared with the CG, the MG had thickened alveolar walls with evident macrophage and neutrophil infiltration in the alveolar lumens, compensatory emphysema adjacent to some alveolar lumens, lymphocyte infiltration in lung tissues, constricted small bronchi with intraluminal mucus accumulation, and a significant elevation in MLI (F = 6.192, p < 0.001; MG vs. CG, p < 0.0001). Compared with the MG, MLI was reduced to varying extents in rats receiving the AP, XFHJ‐L, XFHJ‐M, and XFHJ‐H. Among the treatment groups, the decrease in MLI was statistically significant in the AP (p < 0.05), XFHJ‐M (p < 0.05), and XFHJ‐H (p < 0.05) groups but not in the XFHJ‐L group (Figure 2B).

FIGURE 2.

(A) Representative Masson's trichrome–stained lung sections showing collagen fibers in blue. Scale bars, 50 μm. (B) Quantification of the MLI in hematoxylin and eosin–stained lung sections. (C) Quantification of the collagen‐positive area fraction in Masson's trichrome–stained sections. (D) Pulmonary concentrations of IL‐1β, IL‐6, and TNF‐α. Data are presented as mean ± SD (n = 8). ^#### p < 0.0001 versus the Control group; *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 versus the Model group.

Masson's trichrome staining (Figure 2A) revealed marked thickening of the bronchiolar airway walls in the MG compared with the CG, accompanied by extensive peribronchiolar and perivascular blue collagen fiber deposition. A significant overall group effect was observed (F = 10.13, p < 0.0001). Relative to the MG, collagen‐fiber deposition was reduced in the AP, XFHJ‐M, and XFHJ‐H groups, whereas no obvious change was observed in the XFHJ‐L group. Quantitative analysis using Image‐Pro Plus confirmed that the percentage of the area occupied by collagen fibers was significantly increased in the MG (p < 0.0001) compared with the CG. Relative to the MG, this percentage was reduced to varying extents in rats treated with AP, XFHJ‐L, XFHJ‐M, and XFHJ‐H (Figure 2C). However, only the reduction in the XFHJ‐M group reached statistical significance (p < 0.05).

3.3. Anti‐Inflammatory Effects of XFHJ in COPD Rats

Significant differences in IL‐1β (F = 13.08, p < 0.0001), IL‐6 (F = 14.67, p < 0.0001), and TNF‐α (F = 48.56, p < 0.0001) were detected among the groups (Figure 2D). Compared with the CG, the MG exhibited significantly elevated levels of IL‐1β, IL‐6, and TNF‐α in lung tissue (p < 0.0001), indicating a notable hyperinflammatory state in the lungs of COPD rats. Relative to the MG, the AP group showed significantly reduced levels of IL‐1β (p < 0.001), IL‐6 (p < 0.01), and TNF‐α (p < 0.0001). In the XFHJ‐L group, IL‐1β and IL‐6 levels were decreased numerically, whereas TNF‐α was significantly reduced (p < 0.0001). The XFHJ‐M group exhibited significant reductions in IL‐1β (p < 0.0001), IL‐6 (p < 0.001), and TNF‐α (p < 0.0001), and the XFHJ‐H group showed similar reductions (IL‐1 β , p < 0.0001; IL‐6, p < 0.001; TNF‐α, p < 0.0001). Collectively, AP, XFHJ‐M, and XFHJ‐H groups significantly attenuated pulmonary inflammation in COPD rats.

3.4. Effects of XFHJ on Metabolic Profiles of COPD Rats

In the PCA score plot, the QC samples clustered tightly together (Figure S1), indicating that the analytical method was stable and reliable, and therefore suitable for subsequent analysis. PLS‐DA showed distinct separation among the CG, MG, XFHJ‐M, and XFHJ‐H groups, suggesting that COPD induction and subsequent interventions were associated with marked alterations in the serum metabolome. To further characterize treatment‐related metabolic perturbations, OPLS‐DA models were constructed for pairwise comparisons, including MG versus CG, XFHJ‐M versus MG, and XFHJ‐H versus MG. The corresponding score plots and 200‐permutation validation results are shown in Figure S2A–D. The Q ² regression intercepts and permutation test results confirmed that these models were robust and not overfitted, supporting their use for subsequent differential metabolite screening.

Differential metabolites were screened using the criteria of p < 0.05 and VIP score > 1. Volcano plots revealed substantial serum metabolic alterations between the CG and MG, as well as marked metabolic responses to XFHJ‐M and XFHJ‐H treatment (Figure 3A–C). Venn analysis further identified 17 differential metabolites shared across all three pairwise comparisons. Of these, 13 metabolites, including LPE(16:0), indolelactic acid, quinoline‐4,8‐diol, 2,4‐quinolinediol, DL‐4‐hydroxyphenyllactic acid, and O‐sulfotyrosine, were decreased in the MG relative to the CG and increased after both XFHJ‐M and XFHJ‐H treatment. The remaining four metabolites showed the opposite pattern; for example, corchoionol C 9‐glucoside and DG(6‐keto‐PGF1α/i‐19:0/0:0) were elevated in the MG and decreased following both XFHJ interventions. In addition, 13 metabolites were shared between the MG‐versus‐CG and XFHJ‐M‐versus‐MG comparisons, 15 were shared between the MG‐versus‐CG and XFHJ‐H‐versus‐MG comparisons, and 53 were shared between the two XFHJ treatment comparisons. XFHJ‐M and XFHJ‐H produced both overlapping and distinct metabolic responses (Figure 3D). The heatmap likewise showed distinct groupwise abundance patterns among the differential metabolites (Figure 3E).

FIGURE 3.

(A–C) Volcano plots of differential serum metabolites for the Model versus Control, XFHJ‐M versus Model, and XFHJ‐H versus Model comparisons, respectively. (D) Venn diagram showing unique and overlapping differential metabolites among the three comparisons. (E) Hierarchical clustering heatmap of representative differential metabolites across the Control, Model, Aminophylline, XFHJ‐M, and XFHJ‐H groups. (F–H) KEGG pathway analyses of differential metabolites from the Model versus Control, XFHJ‐M versus Model, and XFHJ‐H versus Model comparisons, respectively.

KEGG pathway enrichment and topology analyses were used to contextualize these metabolite changes. Metabolic disturbances in the MG primarily involved histidine metabolism, glycerophospholipid metabolism, the citrate cycle (TCA cycle), and tryptophan metabolism (Figure 3F). XFHJ‐M predominantly affected tryptophan and glycerophospholipid metabolism, and energy‐ and redox‐related pathways, including pyruvate and glutathione metabolism (Figure 3G). XFHJ‐H also affected tryptophan and glycerophospholipid metabolism, and additionally involved sphingolipid metabolism and phenylalanine, tyrosine, and tryptophan biosynthesis (Figure 3H). These analyses identified tryptophan and glycerophospholipid metabolism as the principal pathways shared by the two XFHJ doses and indicated distinct pathway‐level responses to XFHJ‐M and XFHJ‐H.

In accordance with these pathway‐level changes, several representative metabolites exhibited distinct response patterns (Figure 4). Compared with the CG, the MG had a significantly lower level of indolelactic acid (p < 0.05), as well as lower levels of quinoline‐4,8‐diol and 2,4‐quinolinediol (both p < 0.01). In contrast, GPCho(20:0/18:2) and sphinganine were significantly elevated in the MG (p < 0.05 and p < 0.01, respectively). No significant CG‐MG differences were detected for 3‐indoleacrylic acid, indole‐3‐propionic acid, or LPC(18:3) (p > 0.05). Relative to the MG, XFHJ‐M significantly increased indolelactic acid, 3‐indoleacrylic acid, indole‐3‐propionic acid, LPC(18:3), and quinoline‐4,8‐diol (all p < 0.05), and 2,4‐quinolinediol (p < 0.001). XFHJ‐M also significantly reduced the elevated levels of GPCho(20:0/18:2) and sphinganine (both p < 0.05). XFHJ‐H similarly increased 3‐indoleacrylic acid (p < 0.05), indole‐3‐propionic acid, LPC(18:3), and quinoline‐4,8‐diol (all p < 0.01) and significantly increased indolelactic acid and 2,4‐quinolinediol (both p < 0.001). However, XFHJ‐H did not significantly alter GPCho(20:0/18:2) or sphinganine relative to the MG.

FIGURE 4.

Levels of key metabolites in the Control, Model, XFHJ‐M, and XFHJ‐H groups. For each named comparison, the left panel shows the mean metabolite abundance in the two groups, whereas the right panel shows the between‐group mean difference with its 95% confidence interval. Each group comprised 7 animals. Asterisks denote the significance of the corresponding pairwise comparison: *p < 0.05, **p < 0.01, and ***p < 0.001.

Collectively, XFHJ altered the serum metabolic profile of COPD rats through shared and dose‐associated effects, particularly in tryptophan‐ and glycerophospholipid‐related metabolites.

3.5. Effects of XFHJ on the Gut Microbiota of Rats With Model

Rarefaction curves approached plateaus, indicating that the sequencing depth was sufficient to capture the majority of microbial diversity in the samples (Figure S3A–D). The Shannon, Simpson, Chao1, and ACE indices were calculated to evaluate the α‐diversity of the gut microbiota (Figure 5A–D). No significant differences in these indices were observed between the control group (CG) and model group (MG). Compared with the MG, XFHJ‐H significantly increased the Shannon index (p < 0.01), Chao1 index (p < 0.05), and ACE index (p < 0.01), while significantly decreasing the Simpson index (p < 0.01). Aminophylline treatment also significantly decreased the Simpson index (p < 0.05). XFHJ‐M showed similar trends in these indices, although the differences did not reach statistical significance. Thus, high‐dose XFHJ affected microbial richness and diversity more strongly than the medium dose. Gut microbial β‐diversity was evaluated using three‐dimensional principal component analysis (3D‐PCA) and principal coordinate analysis (PCoA) at the OTU level (Figure 5E,F). In the 3D‐PCA plot, PC1, PC2, and PC3 explained 8.41%, 7.52%, and 5.64% of the total variation, respectively. Significant differences in microbial community structure were detected among the five groups (R = 0.3182, p = 0.001). Similarly, the first two coordinates of the PCoA explained 23.49% and 18.12% of the total variation, respectively, and the overall between‐group difference was significant (R ² = 0.28619, p = 0.001). The MG exhibited a microbial community structure distinct from that of the CG, whereas the XFHJ‐M and XFHJ‐H groups partially separated from the MG in ordination space. The two XFHJ dose groups also displayed different clustering patterns, suggesting that XFHJ‐induced dose‐associated restructuring of the gut microbial community rather than a uniform shift toward the control profile.

FIGURE 5.

(A–D) Shannon, Simpson, Chao1, and ACE indices, respectively, at the OTU level. Data are presented as the mean ± SD (n = 7). *p < 0.05 and **p < 0.01 versus the Model group. (E) Three‐dimensional PCA score plot based on OTU‐level microbial profiles. (F) PCoA based on OTU‐level microbial profiles.

The relative‐abundance profiles of the dominant microbial taxa were visualized at the genus and family levels (Figure 6A,B). At the genus level, the microbial communities were mainly composed of unclassified_f_Lachnospiraceae, Lactobacillus, Romboutsia, Ligilactobacillus, norank_f_Muribaculaceae, norank_o_Clostridia_UCG‐014, and Blautia. At the family level, Lachnospiraceae, Lactobacillaceae, Peptostreptococcaceae, Erysipelotrichaceae, and Muribaculaceae were the dominant families. The relative abundances of these taxa differed among the experimental groups, indicating that COPD induction and XFHJ treatment altered gut microbial composition. The Kruskal–Wallis H test identified significant overall differences among the CG, MG, XFHJ‐M, and XFHJ‐H groups in the relative abundances of Limosilactobacillus (p = 0.044), norank_o_Clostridia_UCG‐014 (p = 0.035), Ligilactobacillus (p = 0.003), Christensenellaceae_R‐7_group (p = 0.026), Roseburia (p = 0.028), Coprococcus (p = 0.009), NK4A214_group (p = 0.011), norank_o_Clostridia_vadinBB60_group (p = 0.040), Bifidobacterium (p = 0.030), Nosocomiicoccus (p = 0.010), norank_c_Clostridia (p = 0.036), and Anaerotruncus (p = 0.002; Figure 6C). Groupwise abundance patterns indicated that the model‐associated increases in Ligilactobacillus and Nosocomiicoccus were attenuated after XFHJ‐M and XFHJ‐H treatment. Both doses were also associated with increased abundances of Roseburia, Coprococcus, and several Clostridia‐related taxa. Notably, XFHJ‐M was associated with marked enrichment of Bifidobacterium, whereas XFHJ‐H was associated with greater increases in norank_o_Clostridia_UCG‐014, NK4A214_group, norank_o_Clostridia_vadinBB60_group, norank_c_Clostridia, and Anaerotruncus.

FIGURE 6.

(A) Relative‐abundance profiles of the predominant bacterial taxa at the genus level. (B) Relative‐abundance profiles of the predominant bacterial taxa at the family level. (C) Mean relative abundances of bacterial taxa that differed among the Control, Model, XFHJ‐M, and XFHJ‐H groups. Overall group differences were evaluated using the Kruskal–Wallis H test, and the corresponding p‐values are shown. Each group comprised seven animals.

LEfSe analysis identified the taxa that most strongly distinguished the CG, MG, XFHJ‐M, and XFHJ‐H groups, using an LDA score > 2 and p < 0.05 as thresholds (Figure 7A,B). The CG was characterized by the enrichment of Limosilactobacillus, Mediterraneibacter, Coprococcus, Holdemania, Frisingicoccus, and CAG‐196, together with members of Cyanobacteriota and Gastranaerophilaceae. In contrast, Lactobacillaceae, Ligilactobacillus, the [Eubacterium]_ruminantium_group, and the Prevotellaceae_NK3B31_group were enriched in the MG. XFHJ‐M was characterized by enrichment of Bacillota, Bifidobacterium, Bifidobacteriaceae, Mammaliicoccus, and Jeotgalicoccus. XFHJ‐H was predominantly associated with Bacteroidia, Bacteroidales, norank_f_Muribaculaceae, norank_o_Clostridia_UCG‐014, Ruminococcus, Roseburia, NK4A214_group, the [Eubacterium]_xylanophilum_group, Dubosiella, and several other Clostridia‐related taxa.

FIGURE 7.

(A) Taxonomic cladogram showing bacterial taxa that distinguished the Control, Model, XFHJ‐M, and XFHJ‐H groups. Circles from the center outward represent taxa from the phylum to genus levels, and circle size reflects relative abundance. Colored nodes indicate taxa enriched in the corresponding group, whereas yellow nodes indicate taxa without significant groupwise differences. (B) LDA scores of differentially abundant taxa identified by LEfSe analysis. Taxonomic prefixes indicate phylum (p), class (c), order (o), family (f), and genus (g).

Overall, XFHJ modulated gut microbial diversity and composition in COPD rats in a dose‐associated manner. XFHJ‐H produced broader changes in α‐diversity and preferentially enriched Bacteroidales‐ and Clostridia‐related taxa, whereas XFHJ‐M was distinguished by the marked enrichment of Bifidobacterium. These dose‐specific microbial responses warrant further investigation of associations between XFHJ‐responsive gut bacteria and circulating metabolite changes.

3.6. Spearman's Rank Correlation Analysis Between Metabolites and the Gut Microbiota

To explore the potential associations among XFHJ‐responsive microbial alterations, serum metabolites, and pulmonary inflammation, Spearman's rank correlation analysis was performed using the selected differential bacterial taxa, representative metabolites, and pulmonary inflammatory cytokines. As shown in Figure 8A, several treatment‐associated bacterial taxa were positively correlated with metabolites that were restored or elevated after XFHJ administration. In particular, Bifidobacterium was positively correlated with indole‐3‐propionic acid (r = 0.4662), 3‐indoleacrylic acid (r = 0.4983), and indolelactic acid (r = 0.4273). The norank_o_Clostridia_UCG‐014 taxon was positively associated with indolelactic acid (r = 0.4218), 3‐indoleacrylic acid (r = 0.3730), and 2,4‐quinolinediol (r = 0.4488). Moreover, Roseburia was positively correlated with indolelactic acid, 2‐hydroxyquinoline, indole‐3‐carbinol, quinoline‐4,8‐diol, and O‐sulfotyrosine. Quinoline‐4,8‐diol was positively correlated with several Clostridia‐related taxa, Bifidobacterium, and Roseburia; the strongest correlation was with norank_o_Clostridia_UCG‐014 (r = 0.6194). By contrast, Nosocomiicoccus was negatively correlated with indolelactic acid (r = −0.4732), quinoline‐4,8‐diol (r = −0.4410), O‐sulfotyrosine (r = −0.4352), 5‐methyldeoxycytidine (r = −0.5387), and several other treatment‐responsive metabolites, but positively correlated with corchoionol C 9‐glucoside (r = 0.5850).

FIGURE 8.

(A) Correlations between XFHJ‐responsive bacterial taxa and representative serum metabolites. (B) Correlations between representative serum metabolites and pulmonary IL‐6, IL‐1β, and TNF‐α levels. Colors represent Spearman's correlation coefficients, with red indicating positive correlations and blue indicating negative correlations. Asterisks denote the significance of the corresponding Spearman correlation: *p < 0.05, **p < 0.01, and ***p < 0.001.

Additional analyses showed associations between the altered serum metabolites and pulmonary inflammatory cytokines (Figure 8B). Most metabolites that decreased in the COPD model and increased after XFHJ treatment were inversely correlated with IL‐6, IL‐1β, and/or TNF‐α. Indolelactic acid was negatively correlated with IL‐6, IL‐1β, and TNF‐α (r = −0.5152, −0.5735, and −0.5486, respectively). LPE(16:0) was also negatively correlated with all three cytokines (r = −0.5733 to −0.6439), whereas LPC(18:3) was negatively correlated with IL‐6 (r = −0.3887), IL‐1β (r = −0.6339), and TNF‐α (r = −0.5020). Strong inverse correlations were observed between 2,4‐quinolinediol and IL‐1β (r = −0.7253) and TNF‐α (r = −0.6670), as well as between indole‐3‐carbinol and IL‐1β (r = −0.7066) and TNF‐α (r = −0.6782). Indole‐3‐propionic acid and 3‐indoleacrylic acid were negatively correlated with IL‐1β and TNF‐α. In contrast, sphinganine was positively correlated with IL‐6, IL‐1β, and TNF‐α, whereas DG(6‐keto‐PGF1α/I‐19:0/0:0) was positively correlated with IL‐1β and TNF‐α.

These results indicate that XFHJ‐associated changes in Bifidobacterium, Clostridia‐related taxa, and Roseburia covaried with increased levels of indole‐related metabolites and selected lysophospholipids. Higher metabolite levels were also associated with lower pulmonary inflammatory cytokine levels. These correlations are consistent with a potential microbiota–metabolite–pulmonary inflammation network underlying the effects of XFHJ, although they do not establish direct microbial production of the metabolites or causal mediation of the anti‐inflammatory effects.

4. Discussion

COPD is a heterogeneous respiratory disorder characterized by persistent airflow limitation, chronic inflammation, and progressive structural injury [18]. In the present study, an experimental COPD model was established using intratracheal LPS instillation combined with cigarette smoke exposure. The resulting pulmonary function impairment, inflammatory infiltration, alveolar enlargement, airway collagen deposition, and elevation of pulmonary cytokines confirmed successful model establishment. XFHJ treatment attenuated these abnormalities, supporting further investigation of its chemical, metabolic, and microbiota‐associated effects.

UHPLC–HRMS profiling tentatively identified 374 constituents in XFHJ, including oroxylin A, baicalin, and wogonin from Scutellaria baicalensis; liquiritin, licochalcone B, and glycyrrhetinic acid from Glycyrrhiza uralensis; and the Fritillaria alkaloid peiminine. These constituents may collectively contribute to XFHJ activity. Oroxylin A alleviates cigarette‐smoke–induced oxidative stress and inflammation through Nrf2 signaling, whereas baicalin suppresses airway inflammation through HDAC2/NF‐κB‐related regulation [19, 20]. Both constituents have also been associated with modulation of the gut microbiota and maintenance of mucosal barrier homeostasis, potentially contributing to gut–lung communication [21, 22]. Liquiritin and licochalcone B exert antioxidant, anti‐inflammatory, and antifibrotic effects, partly through inhibition of HCK signaling [23], whereas glycyrrhetinic acid attenuates inflammation, oxidative stress, and mitochondrial dysfunction in COPD‐related models [24]. In a nonpulmonary model, glycyrrhetinic acid was also associated with enrichment of Bifidobacterium and regulation of tryptophan and sphingolipid metabolism, although its relevance to COPD remains to be established [25]. Peiminine may also inhibit EGFR‐related signaling and pulmonary fibrosis [26]. Collectively, these activities may contribute to the improvements in pulmonary inflammation, tissue injury, and airway remodeling observed after XFHJ treatment.

COPD pathogenesis involves interacting oxidative stress, chronic inflammation, protease–antiprotease imbalance, epithelial dysfunction, mucus hypersecretion, small‐airway remodeling, and emphysematous destruction. Cigarette smoke and other noxious exposures activate epithelial and immune cells, sustaining the release of reactive oxygen species, proteases, chemokines, and cytokines, thereby promoting alveolar injury, extracellular‐matrix remodeling, airflow limitation, and impaired pulmonary function [27, 28, 29]. Beyond pulmonary injury, COPD is accompanied by systemic metabolic reprogramming [30]. Glycerophospholipid and tryptophan metabolism are particularly relevant and may contribute through distinct mechanisms. Disrupted glycerophospholipid metabolism can compromise epithelial membranes and surfactant homeostasis while generating lipid mediators that amplify inflammation [30, 31]. In contrast, altered tryptophan metabolism affects immune and mucosal homeostasis; inflammatory activation may shift tryptophan flux among the kynurenine, serotonin, and microbial indole pathways, reducing protective indole signaling and potentially linking gut dysbiosis and barrier dysfunction to pulmonary inflammation [32, 33]. Thus, host glycerophospholipid dysregulation and host–microbial tryptophan disturbance may represent complementary pathways associated with the effects of XFHJ on COPD‐related inflammation and structural injury.

In this context, the glycerophospholipid alterations observed in the present study may reflect impaired membrane and surfactant homeostasis in COPD. Persistent oxidative stress can induce phospholipid peroxidation and accelerate membrane lipid degradation, thereby impairing epithelial integrity and surfactant function [34, 35]. Phospholipid hydrolysis also generates bioactive lysophospholipids that regulate leukocyte migration, epithelial responses, and inflammatory signaling through Toll‐like receptors and G‐protein‐coupled receptors [36]. However, the biological activity of individual lysophospholipid species depends on their fatty‐acyl composition and degree of unsaturation; consequently, different LPC and LPE species may have distinct inflammatory effects [37]. In the present study, XFHJ partially normalized glycerophospholipid‐related metabolites, including LPE(16:0), LPC(18:3), and GPCho(20:0/18:2). XFHJ also reduced the COPD‐associated elevation of sphinganine, suggesting an additional effect on sphingolipid metabolism. Moreover, LPE(16:0) and LPC(18:3) were inversely associated with pulmonary inflammatory cytokines, whereas sphinganine was positively associated with IL‐1β, IL‐6, and TNF‐α. Thus, XFHJ‐associated pulmonary improvements were accompanied by partial restoration of lipid homeostasis. The normalization of glycerophospholipid metabolism may preserve epithelial membrane stability and pulmonary surfactant homeostasis and reduce the generation of damage‐associated inflammatory signals [34, 35].

In contrast to glycerophospholipid metabolism, tryptophan metabolism provides a direct interface between host immune regulation and the intestinal microbiota. Gut microorganisms convert a portion of dietary and endogenous tryptophan into indole derivatives that participate in epithelial and immune homeostasis [38, 39]. In the COPD model, several tryptophan‐related metabolites, including indolelactic acid, indole‐3‐propionic acid, 3‐indoleacrylic acid, and quinoline‐4,8‐diol, were altered. XFHJ‐M and XFHJ‐H partially normalized these metabolites, suggesting that modulation of tryptophan metabolism may contribute to the therapeutic effects of XFHJ. The distinct microbial profiles induced by XFHJ‐M and XFHJ‐H indicate that the two effective doses produced distinct intestinal responses. XFHJ‐M was characterized primarily by Bifidobacterium enrichment, whereas XFHJ‐H induced broader changes in α‐diversity and preferentially enriched several Clostridia‐ and Bacteroidales‐related taxa. These findings support separate interpretation of the two dose groups and suggest that partially distinct upstream microbial responses may converge on similar improvements in pulmonary inflammation.

The enrichment of several obligate anaerobic taxa associated with intestinal homeostasis suggests partial restructuring of the intestinal microbial ecosystem after XFHJ treatment. Bifidobacterium participates in carbohydrate fermentation and produces acetate and lactate, which can support epithelial defense and serve as substrates in cross‐feeding interactions with other intestinal microorganisms [40, 41]. Certain Bifidobacterium species can also convert aromatic amino acids into aromatic lactic acids, including indolelactic acid, providing a potential link between microbial community changes and tryptophan‐related metabolism. Roseburia is an anaerobic butyrate producer, whereas selected Anaerotruncus strains produce acetate and butyrate [42, 43]. Butyrate is preferentially used as an energy source by the colonic mucosa and supports epithelial barrier and immune homeostasis [43, 44].

Among these microbial changes, the relationship between Bifidobacterium and indolelactic acid merits particular attention. XFHJ‐M markedly enriched Bifidobacterium, and the abundance of Bifidobacterium was positively correlated with serum indolelactic acid. Selected Bifidobacterium species, particularly Bifidobacterium longum , B. breve , and Bifidobacterium bifidum , can convert tryptophan into indolelactic acid through an aromatic lactate dehydrogenase pathway [45]. Culture‐based studies showed that indolelactic acid production varies among bifidobacterial species and strains [46]. Genetic experiments identified Aldh as a key gene in bifidobacterial indolelactic acid synthesis; its disruption markedly reduced indolelactic acid production [47]. These findings provide biological support for the positive association between Bifidobacterium and indolelactic acid observed in the present study, although they do not establish that the enriched Bifidobacterium directly produced the circulating indolelactic acid detected in the COPD rats. Indolelactic acid is both a microbial metabolite and a bioactive signaling molecule. In intestinal epithelial and macrophage models, indolelactic acid attenuated LPS‐ or TNF‐α‐induced inflammatory responses and was associated with activation of AhR‐ and Nrf2‐related pathways and inhibition of NF‐κB signaling [48]. Indolelactic acid secreted by B. longum subsp. infantis suppressed IL‐1β‐induced inflammatory responses in immature intestinal epithelial cells via an AhR‐dependent mechanism [49]. In an experimental colitis model, bacterial indolelactic acid reduced epithelial CCL2 and CCL7 production by regulating AhR‐dependent NF‐κB, HIF, and glycolytic pathways, thereby limiting the recruitment of inflammatory macrophages [50]. In the present study, serum indolelactic acid was negatively correlated with pulmonary IL‐1β, IL‐6, and TNF‐α. Together with the enrichment of Bifidobacterium and its positive correlation with indolelactic acid, these results indicate an association among Bifidobacterium, indolelactic acid, and pulmonary inflammatory status after XFHJ treatment. Nevertheless, the genus‐level taxonomic resolution, the absence of microbial functional gene measurements, and lack of intestinal indolelactic acid quantification preclude direct attribution of circulating indolelactic acid to the enriched Bifidobacterium. Reductive tryptophan metabolism also occurs in selected Clostridia strains, including Clostridium sporogenes [51]. Following XFHJ‐H treatment, changes in several Clostridia‐related taxa coincided with partial recovery of indole‐3‐propionic acid and 3‐indoleacrylic acid, metabolites with reported barrier‐protective and anti‐inflammatory activities [52, 53]. However, the available data did not establish direct microbial production of these metabolites or their causal involvement in gut–lung regulation.

XFHJ alleviated pulmonary dysfunction, inflammatory injury, and airway remodeling in COPD rats, accompanied by changes in host lipid metabolism, gut microbial composition, and tryptophan‐related metabolites. The partial normalization of glycerophospholipid metabolism may support membrane and surfactant homeostasis and attenuate lipid‐associated inflammation. In parallel, XFHJ‐induced microbial alterations were associated with increased levels of tryptophan‐related metabolites. Notably, Bifidobacterium abundance was positively correlated with serum indolelactic acid, which was inversely correlated with pulmonary IL‐1β, IL‐6, and TNF‐α. These findings suggest that host lipid‐metabolic regulation and a potential microbiota–tryptophan–inflammation pathway may jointly contribute to the protective effects of XFHJ. However, causality and underlying mechanisms require experimental validation.

This study has several limitations. First, the noninvasive pulmonary function assessment reduced procedure‐related injury but measured only spontaneous‐breathing parameters and did not provide core forced expiratory indices such as FEV0.1, FEV0.3, and FVC. Moreover, aminophylline is not a preferred first‐line maintenance therapy for COPD and the clinical use of methylxanthines is limited by their relatively modest therapeutic benefit, narrow therapeutic window, potential adverse effects, and drug interactions; therefore, the study does not establish the comparative efficacy of XFHJ against current LABA/LAMA‐based regimens. Second, untargeted serum metabolomics has a high false‐positive rate and lacks definitive MSI level 1 structural identification. Third, 16S rRNA sequencing had limited species‐, strain‐, and functional‐level resolution. Correlations between microbial taxa and circulating metabolites did not establish direct microbial production of these metabolites. Finally, intestinal barrier function, microbial functional genes, receptor activation, and downstream signaling pathways were not examined; therefore, the proposed chemical constituent–microbiota–metabolite–lung relationships require experimental validation. Future studies should integrate targeted quantitative metabolomics, tissue‐specific lipid analysis, shotgun metagenomics or metatranscriptomics, stable‐isotope tracing, microbiota‐transfer experiments, and direct assessment of barrier function and inflammatory signaling. Forced‐expiratory lung function measurements, clinically relevant positive controls, and validation in human samples are also needed to establish the translational relevance of these findings.

5. Conclusion

In summary, this study integrated pharmacodynamic assessment, untargeted serum metabolomics, 16S rRNA gene sequencing, and cross‐dataset correlation analysis to evaluate the effects of XFHJ in a rat model of COPD induced by LPS and cigarette smoke. XFHJ improved pulmonary function and attenuated alveolar structural injury, inflammatory‐cell infiltration, and pulmonary inflammation. Serum metabolomics implicated tryptophan and glycerophospholipid metabolism as the principal treatment‐associated pathways. Gut microbiota analysis further showed that XFHJ altered microbial diversity and community composition, including enrichment of potentially beneficial taxa such as Bifidobacterium, Roseburia, and several Clostridia‐related taxa. Treatment‐responsive taxa were positively associated with indole‐related metabolites, which were inversely associated with pulmonary inflammatory markers. Collectively, these findings indicate that the pulmonary effects of XFHJ were accompanied by changes in glycerophospholipid metabolism and microbiota‐related tryptophan metabolism. This study provides a preclinical basis for further investigation of the biological and causal relevance of these changes and for evaluating XFHJ as a potential adjunctive intervention for COPD.

Author Contributions

Jianping Liu: conceptualization, data curation, formal analysis, investigation, resources, methodology, visualization, writing – original draft, writing – review and editing. Ningxing Duan: data curation, Formal analysis, investigation, visualization, writing – original draft. Tian Xin: data curation, formal analysis, investigation. Yang Cao: resources. Haiyan Qian: conceptualization, funding acquisition, writing – review and editing.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: (A) PCA score plot acquired in positive‐ion mode. (B) PCA score plot acquired in negative‐ion mode. Control group (CG), model group (MG), aminophylline group (AP), XFHJ‐M group, and XFHJ‐H group are shown separately. Quality‐control (QC) samples, prepared by pooling equal aliquots of all study samples, clustered tightly in both ionization modes (red circles), indicating good analytical stability, reproducibility, and reliability of the LC–MS platform throughout the metabolomic analysis.

Figure S2: Supervised multivariate analysis of serum metabolomic profiles and model validation. (A) PLS‐DA score plot and corresponding 200‐permutation test among the CG, MG, XFHJ‐M, and XFHJ‐H groups. The model parameters were R ² Y = 0.845 and Q ² = 0.548, with a Q ² regression intercept of −0.6018. (B) OPLS‐DA score plot and corresponding 200‐permutation test for MG versus CG. The model parameters were R ² Y = 0.989 and Q ² = 0.744, with a Q ² regression intercept of −0.007. (C) OPLS‐DA score plot and corresponding 200‐permutation test for XFHJ‐M versus MG. The model parameters were R ² Y = 0.998 and Q ² = 0.629, with a Q ² regression intercept of 0.0288. (D) OPLS‐DA score plot and corresponding 200‐permutation test for XFHJ‐H versus MG. The model parameters were R ² Y = 0.993 and Q ² = 0.645, with a Q ² regression intercept of −0.0144. The Q ² intercepts and permutation validation results indicate that the models were reliable and not overfitted.

Figure S3: (A) Shannon curves of α‐diversity analysis. (B) Simpson curves of α‐diversity analysis. (C) Ace curves of α‐diversity analysis. (D) Chao1 curves of α‐diversity analysis.

Table S2: Primer sequences utilized for bacterial 16S rRNA gene amplification.

Table S1: Chemical constituents of XFHJ.

Data Availability Statement

The raw UHPLC‐Orbitrap Exploris 240 mass spectrometry data have been deposited in the MetaboLights database under accession number MTBLS14337 (https://www.ebi.ac.uk/metabolights/reviewer919c65bf‐2879‐4961‐9cd0‐642e80b57c7c). The raw 16S rRNA gene sequencing data have been deposited in the NCBI SRA database under BioProject accession number PRJNA1456130 (https://dataview.ncbi.nlm.nih.gov/object/PRJNA1456130?reviewer=2f3d1hqhh1irot1bujsm9gggom).