A total of 60 kg of Prismatomeris connata Y. Z. Ruan root was subjected to extraction through reflux with 95% ethanol, with each cycle lasting 3 hours. After concentration, a yield of 1200 g of extract was achieved. This resulting extract was completely dissolved to create an aqueous solution, which was subsequently subjected to sequential extractions with petroleum ether, ethyl acetate, and n-butanol. The ethyl acetate extraction portion was subjected to decompression and concentration, thus ultimately yielding 112 g of ethyl acetate extract, denoted HG2.

We used the Bioconductor/R CIBERSORT package with the CIBERSORT algorithm to conduct an in-depth analysis of immune cell infiltration in lung tissue. This procedure allowed us to evaluate the abundance of 22 distinct types of immune cells in the lung tissue samples (38). Furthermore, the Bioconductor/R GSVA package and ssGSEA were used to investigate the immune infiltration specifically pertaining to AMs (39). We tailored the gene sets representative of M1 and M2 macrophages, aligning them with established biomarkers associated with the polarization of human macrophages. The M1 macrophage signature included the expression of CD68, CD80, IL1R1, TLR2, TLR4, IL12A, IL12B, IL23A, IL27, CXCL9, CXCL10, CXCL11, CXCL16, CCL5, ARG2, NOS2, TNF, IL1A, IL1B, IL6, CD86, HLA-DMA, HLA-DMB, HLA-DOA, HLA-DOB, HLA-DPA1, HLA-DPB1, HLA-DQA1, HLA-DQA2, HLA-DQB1, HLA-DRA, HLA-DQB2, HLA-DRB1, HLA-DRB3, HLA-DRB4, HLA-DRB5, CLEC7A, STAT1, IFNG, FCGR3B, FCGR2C, FCGR1A, MERTK, and IRF5. In contrast, the M2 macrophage signature encompassed ARG1, CCL1, CCL16, CCL17, CCL18, CCL22, CCL24, CD14, CD163, CD36, CHI3L1, CHI3L2, CLEC10A, CXCL13, EBI3, F13A1, FCGR3A, IGF1, IL1R2, IL1RN, IL4R, ITGAX, MRC2, PLA2R1, RETNLB, SLAMF1, SPHK1, STAB1, TGFB1, TLR1, TLR8, VEGF, VTCN1, STAT6, IDO1, CSF1R, MSR1, CD209, CLEC4M, MS4A2, FCER1G, FCER1A, VSIG4, and IRF4 (40–43).

The least absolute shrinkage and selection operator (LASSO) regression prediction model was constructed with the Bioconductor/R glmnet package. Parameters were set to alpha = 1 and nlambda = 1000, with lambda.min selected as the optimal lambda. Subsequently, a logistic regression model was developed with the Bioconductor/R rms package, and the calibrate function was used to fine-tune the regression model, with parameters B = 1000 and sls = 0.05.

For random forest analysis, the Bioconductor/R pacman package was used, with the randomForest function. The mtry node value was determined on the basis of the minimum error, whereas the ntree value was selected according to image stability trends.

To identify the top ten critical differential immune cells, we used the mean decrease accuracy and mean decrease Gini metrics. Further refinement of the key differential immune cells was achieved through a comparative assessment across the three algorithms.

We performed GO and KEGG enrichment analyses with the Bioconductor/R clusterProfiler, enrichment, and ggplot2 packages. Additionally, for CTpathway enrichment analysis, we used the online analysis platform available at http://www.jianglab.cn/CTpathway/ (44).

The gene set enrichment analysis (GSEA) analysis of patients with IPF was conducted with the gseGO and gseKEGG functions in the Bioconductor/R clusterProfiler package. Results meeting the criteria of an |normalized enrichment score| > 1, adjusted p value < 0.05, and false discovery rate (FDR) < 0.25 were considered statistically significant.

We used the Bioconductor/R weighted gene co-expression network analysis (WGCNA) package to construct co-expression networks between IPF lung tissue and normal samples, as well as between IPF AMs and normal samples (45). The sample’s phenotype was categorized as either IPF group or normal control. For lung tissue, the soft threshold was set to β = 4, thus yielding a scale-free R² of 0.85. For AMs, the soft threshold was adjusted to β = 9, and a scale-free R² of 0.85 was maintained. Subsequently, the co-expression matrix was generated with the expression matrix, and the respective β values were determined. Genes exhibiting similar expression patterns were grouped into corresponding gene modules, thus establishing co-expression modules ( Supplementary Figures S2, S3 ). Characteristic gene functions were applied to assess disparities in modular characteristic genes (Mes) and the correlation between modules and phenotypes. Pearson coefficients were computed to determine the correlation between genes and phenotypes within modules, thereby identifying central hub genes.

Specific pathogen-free (SPF) male C57BL/6N mice, weighing between 18 and 22 g were procured from Beijing Vital River Laboratory Animal Technology Co., Ltd. The company holds a production license (SCXK- (Beijing) 2016-0011) issued by the Beijing Science and Technology Commission. The animals were housed in the facilities of Beijing Union-Genius Pharmaceutical Technology Development Co., Ltd. This company is further accredited by the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC International) under number 143.

Each cage housed five mice, and a controlled environment was provided with a room temperature of (25 ± 2)°C and 45% relative humidity. A 12-hour light-dark cycle was maintained, and the mice were given unrestricted access to food and water throughout the feeding period. The experimental protocol was approved by the company’s Experimental Animal Welfare Ethics Committee. The experimental animal license number is SCXK (Beijing) 2020-0021, and the experimental animal certificate number is 2018-002 (research).

A total of 65 SPF male C57BL/6N mice, 6–8 weeks of age, weighing 18–22 g, were selected. Among them, ten mice served as the control group, whereas the remaining 55 mice were subjected to BLM treatment for inducing pulmonary fibrosis. After 1 week of acclimation, each mouse underwent a precise intratracheal injection of either BLM (1.25 mg/kg, dissolved in sterile PBS buffer, provided by Zhejiang HISUN Pharmaceutical Co., LTD., China) or sterile PBS buffer (50 μL) with an HRH-MAG4 quantitative pulmonary fluid atomizer from Beijing HuiRongHe Technology Co., Ltd. (Beijing, China).

On the 15th day after BLM treatment, HG2 (at doses of 100 mg/kg, 200 mg/kg, or 400 mg/kg), nintedanib (50 mg/kg, sourced from the Institute of Materia Medica, Chinese Academy of Medical Sciences, China), or vehicle (0.5% CMC-Na) was administered via gastric gavage once daily for 14 consecutive days. Within 24 hours after the last administration, three mice were randomly selected from each group and humanely euthanized, and the residual blood in the lung tissue was cleared through cardiac perfusion. The lungs were then extracted; the left lung was subjected to histopathological examination, and the right lung was promptly frozen at −80°C.

The remaining mice were anesthetized via intraperitoneal injection of pentobarbital sodium (90 mg/kg), then underwent tracheotomy and tracheal intubation, and were connected to an Flexivent small animal pulmonary function instrument from SCIREQ (city, France). Various pulmonary function parameters, including respiratory resistance (Rrs), central airway resistance (Crs), peak expiratory flow (PEF), FVC, and forced expiratory volume in the first 0.1 second (FEV0.1), were measured.

Bronchoalveolar lavage was performed with pre-chilled PBS buffer at 4°C (1 ml/each, performed three times), and the bronchoalveolar lavage fluid (BALF) was collected and stored at 4°C for subsequent cytokine analysis.

To further investigate the mechanism of HG2, we selected 30 SPF male C57BL/6N mice, 6–8 weeks old, with an average body weight of 18–22 g. Among these, 25 mice were subjected to induced pulmonary fibrosis with the previously described modeling approach, whereas five mice were designated as the control group.

On the 15th day after administration of BLM, HG2 was administered via gastric gavage at varying doses (50 mg/kg, 100 mg/kg, and 200 mg/kg), alongside nintedanib (at a dose of 50 mg/kg), or a vehicle solution (0.5% CMC-Na), once per day for 14 days. Subsequently, all mice were euthanized within 24 hours of the final administration. Residual blood in the lung tissue was removed via cardiac perfusion. The lungs were then extracted, promptly frozen at −80°C, and prepared for subsequent western blot analysis.

The left lung samples from the mice were preserved in 4% paraformaldehyde for more than 72 hours. Subsequently, the samples underwent dehydration and paraffin embedding, then were sectioned into slices with 4 μm thickness. These slices were then subjected to staining with hematoxylin and eosin, as well as Masson’s trichrome, both from Beijing SoonBio Technology Co., Ltd., China. Subsequently, the samples were imaged with a NIKONCI-S microscope (Nikon, Japan) in conjunction with a DS-F12 imaging system (Nikon).

RAW264.7 cells, generously provided by the Stem Cell Bank at the Chinese Academy of Sciences, were cultivated in DMEM:F-12 1:1 medium (C3132-0500, Viva Cell, Israel) supplemented with 10% fetal bovine serum (C04001-500, Viva Cell, Israel), 100 U/mL penicillin, 0.1 g/mL streptomycin, and 12.5 U/mL mycostatin (C3422-0100, Viva Cell, Israel) under 5% CO₂ at 37°C until they reached 70% confluence.

At 24 hours after RAW264.7 cell seeding in 96-well cell culture plates, the medium was aspirated, and 200 μL of either vehicle (0.5% DMSO) or various concentrations of HG2 solution (673.72, 336.86, 168.43, 84.2, 42.10, 21.05, and 10.50 μg/mL) was added to each well. The plates were then incubated for 12 hours at 37°C under 5% CO₂. Subsequently, the medium was removed, and the cells were treated with Cell Counting Kit-8 (CCK-8) reagent (10 μL, CK001-500T, Lablead, China) for 1.5 hours at 37°C. The optical density (OD) at 450 nm was measured with a TriStar2 SLB942 fluorescence enzyme labeling instrument (Berthold Technologies, Germany). Cell viability was calculated with the following formula: cell viability (%) = [OD (treatment) − OD (blank)]/[OD (control) − OD (blank)].

After determination of HG2’s median lethal concentration (LC50) in RAW264.7 cells, the cells were subjected to the following treatments: serum-free medium, serum-free medium with 1 μg/mL lipopolysaccharide (LPS), or 1 μg/mL LPS combined with 3 ng/mL interferon (IFN)-γ or 20 ng/mL IL-4. Additionally, cells were treated with serum-free medium containing 1 μg/mL LPS or 1 μg/mL LPS combined with 3 ng/mL IFN-γ or 20 ng/mL IL-4, along with varying concentrations of HG2 (84.2, 42.1, 21.05, and 10.5 μg/mL). Subsequently, cells and supernatants were collected for nitric oxide (NO) and cytokine determination, western blotting, real-time quantitative PCR (RT-qPCR), flow cytometry, and immunocytochemistry.

RAW264.7 cells were harvested, and 200 μL of RIPA lysis buffer was added to each cell sample. The cells were thoroughly lysed through a series of freeze-thaw cycles. Subsequently, the lysates were centrifuged for 20 minutes at 12,000 rpm and 4°C and, and the resulting supernatant was collected to obtain intracellular proteins. For extracellular protein extraction, culture supernatants were collected and mixed with pre-cooled acetone at a 1:4 volume ratio. The mixture was then frozen at −20°C for a minimum of 12 hours to ensure complete protein precipitation. Afterward, the sample was centrifuged for 20 minutes at 12,000 rpm and 4°C, and the resulting pellet was collected. This pellet contained the extracellular proteins, which were reconstituted in the previously obtained intracellular protein solution to yield the total cellular protein.

Total protein was extracted from lung tissues with an animal tissue total protein extraction kit (BC3790, Solarbio, China). A BCA protein quantification kit (B5000–500T, Lablead, China) was used to determine protein concentrations. Cell samples (50 μg) and tissue samples (70 μg) were separated on 4%–12% SDS-PAGE gels and then transferred to a PVDF membrane. The membrane was blocked with 5% skim milk at room temperature for 1 hour, then incubated overnight with primary antibodies (1:1500) at 4°C. After thorough washing, the membrane was incubated with HRP-conjugated secondary antibodies (1:3000) for 1–2 hours at room temperature. Details regarding the primary and secondary antibodies are provided in the Supplementary Materials ( Supplementary Table S1 ). After additional washes, protein signals were detected with an ECL kit (E1060, Lablead, China). Quantitative analysis was conducted with ImageJ (National Institute of Mental Health, USA).

The aforementioned tissue sections underwent immunohistochemical staining for collagen I (1:400, ab279711, Abcam, UK) after high-pressure heat-induced antigen retrieval, with a SABC-AP immunohistochemical staining kit (SA1052, Boster, China). Images were captured with an AxioVert. A1 Microscopic Imaging System (CARL ZEISS, Germany) and subjected to quantitative analysis in ImageJ.

Multiple fluorescence immunohistochemical staining was performed with a PANO 4-plex IHC kit (10001100100, Panovue, China). The aforementioned tissue slices were subjected to high-pressure heat-induced antigen retrieval and sealed with a specialized immunohistochemistry sealing solution. Subsequently, the corresponding primary antibody (1:400) and HRP-labeled secondary antibody were sequentially applied. After TSA signal amplification, the slices underwent high-pressure heat treatment after each TSA application to decrease background noise. After the labeling of all antigens, nuclei were stained with DAPI, and images were captured with an AxioVert. A1 microscopic imaging system. ZEN3.1 (CARL ZEISS, Germany) was used for image processing and Plot Profile analysis, whereas ImageJ was used for quantitative analysis.

RAW264.7 cells were seeded onto cell tablets (BS-24-RC, Bio-sharp, China). After cell adhesion, the cells were fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton, and blocked with 10% goat serum (50012-8615, SiJiQing, China). Subsequently, the cells were incubated with the corresponding primary antibody (1:500) and fluorescently labeled secondary antibodies (1:500). Nuclei were counterstained with DAPI (SIGMA-ALDRICH, Germany), and images were visualized with an AxioVert. A1 microimaging system. The images were further processed with ZEN3.1.

The supernatants from cell cultures and mouse BALF were carefully collected, subjected to centrifugation for 10 minutes at 1000 rpm and 4°C, and the supernatant subsequently aspirated. The NO content within the cell culture supernatant was quantified with an NO one-step assay kit (A013-2-1, Nanjing Jiancheng Bioengineering Institute, China). Additionally, the expression levels of key cytokines, including IL-6, IL-10, IL-12, tumor necrosis factor (TNF)-α, and TGF-β1 in the cell culture supernatant, were determined according to the protocols outlined by the respective commercial enzyme-linked immunosorbent assay (ELISA) kits (Boster, China).

Furthermore, the levels of cyclooxygenase (COX)-2, plasminogen activator inhibitor (PAI)-1, MMP7, and IL-17B in BALF were evaluated with respective commercial ELISA kits (all from Cloud-clone, China). Finally, the levels of TGF-β1, IL-10, E-cadherin, IL-17A, and MMP3 in BALF were determined with respective commercial ELISA kits (all from Boster) according to the manufacturer’s instructions for accurate determination.

Total RNA was extracted from both RAW264.7 cells and lung tissue homogenate with an RNeasy™ Plus animal RNA extraction kit (R0032, Beyotime, China). Subsequently, a PrimeScript™ RT reagent Kit with gDNA Eraser (Perfect Real Time) (RR047A, Takara, Japan) was used for cDNA reverse transcription. The quantitative PCR reactions were conducted in a 20 μL reaction mixture system using TB Green Premix Ex TaqTM II (Tli RNaseH Plus) (RR820A, Takara, Japan) and specific primers. Amplification was performed with an ABI 7900HT 96 real-time PCR system (Applied Biosystems, USA). The mRNA levels were normalized to those of GAPDH, and the fold change in mRNA was calculated with the ΔΔCt method. Primer sequences are provided in the Supplementary Material ( Supplementary Table S2 ).

RAW264.7 cultured cells were harvested by centrifugation at 3000 rpm at 4°C for 5 minutes and subsequently labeled with anti-NOS2-PE (696805, Biolegend, USA), anti-MHC-II-FITC (107606, Biolegend, USA), and anti-F4/80-APC (123116, Biolegend, USA). Flow cytometry was performed with a FACS Verse™ Cell Analyzer (BD Biosciences, USA), and the data were analyzed in FlowJo software (Tree Star, USA) to determine the percentage macrophage polarization (iNOS⁺F4/80⁺ or MHC-II⁺F4/80⁺).

The data are presented as mean ± standard deviation (SD), and each experiment was conducted independently and repeated at least three times. Initially, the homogeneity of variances was assessed. In cases of homogeneous variances, one-way ANOVA was used to determine the statistical significance of differences between various samples (P < 0.05). In the event of heterogeneous variances, an independent samples Student’s t-test was used to assess whether the differences between samples were statistically significant (P < 0.05). Data analysis was performed in IBM SPSS Statistics 21.0 (IBM, USA), and GraphPad Prism 8.0.2 (GraphPad Software, USA) was used for graphical representation.

To investigate immune cell infiltration patterns in the lungs of patients with IPF versus healthy people, we analyzed the expression matrix (comprising seven datasets) with Cibersort. Our findings indicated a notable increase in the infiltration levels of plasma cells, T follicular helper cells, M2 macrophages, and activated dendritic cells in patients with IPF compared with healthy controls. In contrast, the infiltration levels of naive B cells, naive CD4 T cells, resting CD4 memory T cells, resting mast cells, and neutrophils exhibited a significant decrease. Additionally, a slight increase was observed in the infiltration levels of M1 macrophages and regulatory T cells (Tregs) ( Figures 1A, B ).

Further exploration of the correlation between IPF and immune cell infiltration in the lungs revealed a strong association between M2 macrophages and IPF. These M2 macrophages exhibited a positive correlation with activated dendritic cells and a negative correlation with naive B cells ( Supplementary Figure S1A ). With LASSO regression, we identified ten immune cell types closely associated with IPF ( Supplementary Figure S1B ). Subsequently, using multivariate logistic regression to control for potential confounding factors, we determined that M2 macrophages, plasma cells, T follicular helper cells, and monocytes were characteristic immune cell types in IPF ( Supplementary Figure S1C ). In the random forest algorithm, immune cell types were ranked on the basis of the feature weights of the mean decrease accuracy and mean decrease Gini, and the top ten cell types were selected as key contributors ( Figure 1C ). By combining the results from the three algorithms, we conclusively established M2 macrophages as the most critical immune cell type in IPF. Additionally, T follicular helper cells and plasma cells also played substantial roles in the condition. To corroborate our findings, we assessed the gene expression of markers associated with macrophage polarization in the lung tissue in patients with IPF compared with healthy controls, and observed the anticipated expression patterns ( Supplementary Figure S1D ).

We further assessed differences in AMs between patients with IPF and healthy individuals, which exhibited elevated expression of genes associated with the inflammatory response and macrophage polarization, including CCL2, CXCL3, CD36, and TLR2 ( Supplementary Figure S2A ). CTpathway-based enrichment analysis of gene expression revealed pathways associated predominantly with macrophage polarization and collagen, such as the IL-4, IL-13, IL-10, and IFN-γ pathways; MHC II antigen presentation; and collagen degradation ( Figure 2A , Supplementary Figure S2B ). GSEA analysis further demonstrated the activation of immune-associated KEGG pathways (e.g., Toll-like receptor signaling pathway, TNF signaling pathway, and NF-κB signaling pathway) in the AMs of patients with IPF ( Figure 2B ). Notably, inflammatory, immune, and cytokine-associated GO pathways, including immune response-activated cell surface receptor signaling pathway, viral response, and lymphocyte activation regulation, exhibited substantial alterations ( Supplementary Figure S2C ). Through ssGSEA analysis of the AM dataset, we observed significantly greater abundance of M2 AM infiltration in patients with IPF than healthy controls ( Figure 2C ).

Traditional Chinese medicine is distinguished by its multi-component and multi-target synergy. Through an extensive literature review, we identified 51 compounds with known structures in the root of Prismatomeris connata Y. Z. Ruan ( Supplementary Figure S3A ). SwissADME (http://www.swissadme.ch/) was used to screen for active ingredients, thus yielding 36 compounds (compounds 1–36). Furthermore, SwissTargetPrediction (http://www.swisstargetprediction.ch/), SuperPred (https://prediction.charite.de/), and SEA (https://sea.bkslab.org/) were used to predict potential drug targets for these active compounds, thus yielding a comprehensive list of 670 target genes.

For a deeper exploration of the targets of active compounds in IPF, we curated therapeutic target information for IPF from various datasets in the GEO database, including lung tissue samples from GSE110147, GSE15197, GSE24988, GSE31934, GSE32537, GSE35145, and GSE53845, as well as AM samples from GSE49072, GSE70864, and GSE90010. Differentially expressed genes were identified with the Bioconductor/R limma package. Disease-associated gene modules were examined via WGCNA, and hub genes were extracted through CTpathway analysis.

Simultaneously, therapeutic target information for IPF was obtained from three online databases: GeneCards (https://www.genecards.org/), DisGeNET (https://www.disgenet.org/), and OMIM (https://omim.org/), thus resulting in 2980 targets ( Figure 3A ). With the CytoNCA plug-in in Cytoscape software, we filtered core targets with criteria including closeness ≥ 0.001899, betweenness ≥ 278.766, and degree ≥ 37.4789, thus resulting in the identification of 52 core targets ( Figure 3A ).

The outcomes of GO enrichment and KEGG enrichment analyses indicated that the active ingredients within the root of Prismatomeris connata Y. Z. Ruan regulate pathways associated primarily with immunity, such as the TNF signaling pathway, IL-17 signaling pathway, pathways associated with the coronavirus disease COVID-19, and pathways associated with cell differentiation, such as Th17 cell differentiation and sites of polarized growth in IPF therapy ( Figure 3B ).

A data independent model was used for proteomic analysis across all sample groups. By setting the FDR of the peptides to 1% and ensuring that each protein was represented by at least one characteristic peptide, we identified 6035 proteins in all four sample groups. PLS-DA revealed clear distinctions between each group, most prominently between the model group and control group. Furthermore, after administration of HG2 or nintedanib, we observed a discernible trend in reversion of protein expression levels ( Figure 4A ).

Further GSEA demonstrated significant activation of immune-associated pathways, such as the complement and coagulation cascade, NF-κB signaling pathway, Toll-like receptor signaling pathway, and pathways associated with COVID-19, as well as collagen-associated pathways including ECM receptor interaction, in the model group compared with the control group ( Figure 4C ). These findings were consistent with the results obtained from the bioinformatics analysis. Additionally, pathways associated with immunoglobulin production, innate immune response, inflammatory response, and viral response were notably activated in the model group compared with the control group ( Supplementary Figure S4A ).

HG2 and nintedanib exhibited distinct action profiles. Whereas both compounds down-regulated expression of the members of complement and coagulation cascade, HG2 demonstrated more potent inhibition of pathways associated with immunoglobulin production, including molecular regulator production, immunoglobulin production, humoral immune response, and complement activation, in the immune response ( Figure 4B , Supplementary Figure S4A ). In comparison to the HG2 group, the nintedanib group displayed significant up-regulation of the expression of Coronavirus disease-COVID19, oxidative phosphorylation, and PPAR signaling pathway members, whereas the expression of carbon metabolism and B-cell receptor signaling pathway members was notably down-regulated. These findings indicated that HG2 exerts a more focused role in immune regulation than nintedanib ( Figure 4B , Supplementary Figure S4A ).

We further examined HG2’s effects on cytokine release and macrophage polarization by using an in vitro macrophage model. Initially, we assessed the effect of HG2 stimulation at varying concentrations on RAW264.7 cell proliferation over 12 hours. The calculated 12-hour LC50 of HG2 on RAW264.7 cells was 128.1 ± 1.4 μg/mL ( Figure 5A ). On the basis of this result, we established four concentrations (84.2, 42.1, 21.05, and 10.5 μg/mL) as distinct HG2 administration groups.

Next, RAW264.7 cells were stimulated with LPS (1 μg/mL) for 12 hours and subsequently treated with different concentrations of HG2. We determined cytokine and NO levels in the supernatant with ELISA and NO one-step assay kits. HG2 significantly decreased the levels of IL-6, TGF-β1, TNF-α, IL-10, and NO in the supernatant of RAW264.7 cells induced by LPS. Notably, the decrease in TGF-β1 and TNF-α exhibited a dose-dependent pattern. We also observed a similar trend when HG2 was applied to RAW264.7 cells co-stimulated with LPS (1 μg/mL) and IFN-γ (3 ng/mL), thus substantially decreasing NO and IL-10 release ( Figures 5B–D ).

To understand HG2’s influence on macrophage polarization, we subjected RAW264.7 cells to stimulation with LPS (1 μg/mL) separately, LPS (1 μg/mL) combined with IFN-γ (3 ng/mL) and IL-4 (20 ng/mL) separately for 12 hours, with subsequent HG2 treatment. We monitored changes in mRNA ( Figure 6A , Supplementary Figure S5A ) and protein ( Figures 6B, C ) expression levels of polarization-associated markers in RAW264.7 cells. HG2 notably suppressed the elevation in Cd206, Nos2, and Il6 mRNA expression, and enhanced Mgl2 mRNA expression induced by LPS. Furthermore, HG2 significantly inhibited the increases in Cd80, Cd86, and Nos2 mRNA expression triggered by LPS combined with IFN-γ, as well as the increases in Cd206, Arg1, Chi3l3, and Mgl2 mRNA expression stimulated by IL-4. Western blotting findings supported HG2’s inhibitory effect on iNOS expression in RAW264.7 cells induced by LPS or LPS combined with IFN-γ, and the expression of CD206 and Arg-1 in RAW264.7 cells induced by IL-4. Flow cytometry ( Figures 5E–G ) and immunocytochemistry ( Supplementary Figure S5C ) corroborated the effects of HG2 on the regulation of macrophage polarization.

Interestingly, western blotting demonstrated that HG2 dose-dependently increased the expression of total COX-2 and intracellular CD80 in RAW264.7 cells. Flow cytometry revealed that HG2 dose-dependently increased MHC-II expression in RAW264.7 cells, and the effect was not influenced by LPS or LPS combined with IFN-γ or IL-4 ( Figures 6B , 5G ).

Therefore, these findings suggested that HG2 has nuanced effects on macrophage polarization, not only influencing M1 and M2 polarization, but also promoting the transition of macrophages to a state intermediate between M1 and M2. Additionally, HG2 enhances the expression of total COX-2, intracellular CD80, and MHC-II, while inhibiting the release of inflammatory factors induced by LPS or LPS combined with IFN-γ. Thus, HG2 may modulate immune responses by affecting the antigen presentation function of macrophages.

To investigate the therapeutic potential of HG2 in pulmonary fibrosis, we used a mouse model induced by BLM and examined lung pathology and collagen deposition. Pulmonary fibrosis is characterized by restrictive ventilation disorder. Before dissection, we assessed pulmonary function with Flexivent, measurements of Rrs, Crs, PEF, FEV0.1, and FVC, and generated flow-volume (F-V) and pressure-volume curves. Compared with the control group, the model group exhibited significantly higher Rrs and Crs, and lower PEF, FVC, and FEV0.1. The F-V curve indicated restrictive and obstructive lesions, thus suggesting severe ventilation restriction in the model group. Different concentrations of HG2 and nintedanib treatment led to improved outcomes ( Figure 7C ).

We assessed the expression of factors in BALF. Compared with that in the control group, PAI-1 expression in the model group’s BALF was significantly lower. Both the nintedanib and HG2 treatment groups demonstrated clear improvement, although no significant difference was observed in the expression of other factors, possibly because of the measurement timing (28th day post-modeling) ( Figure 7B ). Furthermore, we tracked survival rates and body weight changes in the mice ( Figure 7A ).

After 28 days of modeling, we prepared paraffin sections of lung tissue for HE staining. Pathological examination revealed alveolar and alveolar duct occlusion, interstitial thickening with fibrous tissue hyperplasia, and pulmonary bronchiolar epithelial cell exfoliation in the model group compared with the control group. Additionally, localized alveolar and alveolar duct complete occlusion, as well as interstitial fibrous tissue hyperplasia, occurred without a significant increase in dense fibers. These findings were characteristic of pulmonary fibrosis development. Lung injury was alleviated in both the HG2 and nintedanib treatment groups, and HG2 demonstrated milder effects, thus indicating superior effects against BLM-induced pulmonary fibrosis in mice ( Figure 8A ).

Massive release and deposition of ECM proteins, including collagen, laminin, and fibronectin, are characteristic of pulmonary fibrosis. Masson staining and type I collagen immunohistochemistry were used to observe collagen deposition in lung tissue sections. In contrast to the control group, the model group exhibited strong positive staining, whereas both the HG2 and nintedanib treatment groups demonstrated a notable reduction ( Figure 8B ). Immunohistochemical results also indicated specific type I collagen deposition around tracheal and vascular areas of the lungs ( Figure 8C ). ImageJ was used to semi-quantitatively detect areas in lung tissue with positive Masson and type I collagen staining. HG2 treatment at various doses, along with nintedanib treatment, showed superior anti-pulmonary fibrosis effects, and the 400 mg/kg HG2 group showed the best performance ( Figures 8D, E ).

To further validate HG2’s therapeutic effect, we assessed the expression of collagen I, collagen III, and ECM deposition-associated mRNA and proteins in the mouse lungs. The model group showed greater expression of type I collagen, type III collagen, and fibronectin mRNA than the control group. Notably, HG2 exhibited a more pronounced inhibitory effect on collagen and fibronectin mRNA expression. However, nintedanib demonstrated only anti-collagen expression effects ( Figure 7D ). These findings were corroborated by western blotting. Both HG2 and nintedanib effectively inhibited type I and type III collagen protein expression, whereas HG2 had a pronounced inhibitory effect on ECM deposition-associated proteins, such as vimentin (400 mg/kg), fibronectin (200 mg/kg), and Snail Family Transcriptional Repressor 2 (SNAI2) (200 mg/kg). In contrast, no statistically significant difference was observed in the nintedanib group ( Figures 7E, F ).

Additionally, after confirmation of HG2’s anti-BLM-induced pulmonary fibrosis effect, we conducted an acute-subacute toxicity test to preliminarily assess safety. The results revealed no acute-subacute toxicity in ICR mice ( Supplementary Tables S3–S6 ).

In our investigation of the polarization state of pulmonary macrophages in mice with BLM-induced pulmonary fibrosis, we assessed the mRNA and protein expression levels of macrophage polarization-associated biomarkers. Notably, compared with the control group, the model group exhibited substantial upregulation of the M2 macrophage markers CD206 and Arg-1, as well as the macrophage marker F4/80. Intriguingly, we also observed the presence of M1 macrophage markers, including dimeric iNOS, CD80, and Cox-2. This observation was further corroborated by RT-qPCR, which revealed elevated expression levels of M2 macrophage-associated mRNA, such as Cd206, Mgl2, Retnla, and Chi3l3. Additionally, Nos2 and H2ab1 (M1 biomarkers) were also highly expressed. The presence of both M1 and M2 type macrophage markers in the model group underscores the importance of the balance between inflammation and repair and highlights the dynamic shifts in macrophage phenotype throughout idiopathic pulmonary fibrosis progression. Of note, the treatment with nintedanib and HG2 at various doses had distinct inhibitory effects on the expression of macrophage markers. The 400 mg/kg HG2 treatment demonstrated superior efficacy to nintedanib treatment ( Figures 9A–C ).

We sought to visually depict the intricate relationship between pulmonary fibrosis and lung macrophages, particularly the M2 subtype, by using a multi-fluorescence immunohistochemical staining approach targeting CD206, F4/80, and collagen I. Our findings revealed substantial accumulation of collagen I surrounding the trachea and interstitial spaces in the model group, accompanied by macrophage infiltration and elevated CD206 expression. Encouragingly, both the HG2 treatment group and the nintedanib treatment group demonstrated significant decreases in collagen deposition, macrophage infiltration, and CD206 expression. This observation was further corroborated by fluorescence intensity analysis, which revealed markedly higher levels of CD206, F4/80, and collagen I in the model group than the control group. Notably, the 400 mg/kg HG2 group demonstrated a substantial decrease in fluorescence intensity. Additionally, qualitative analysis with Plot Profile indicated clear co-localization of collagen I, CD206, and F4/80 ( Figures 10A–C ). This comprehensive visualization provided crucial insights into the dynamic interplay among lung macrophages, collagen deposition, and the development of pulmonary fibrosis.

Given the absence of significant disparities in factors within the BALF, we sought to gain deeper insights into the effects of macrophage polarization on the expression of pulmonary inflammatory factors in mice with pulmonary fibrosis. To this end, we used western blotting to assess the levels of TGF-β, TNF-α, IL-6, IL-10, and IL-1β in lung tissue ( Figures 9D, E ). In comparison to the control group, the model group exhibited markedly higher levels of TGF-β in lung tissue, alongside significantly diminished levels of IL-10. Notably, various doses of HG2 and nintedanib demonstrated restorative effects: the 400 mg/kg and 200 mg/kg HG2 groups, as well as the nintedanib treatment group, exhibited a substantial decrease in TGF-β expression. In contrast, the 400 mg/kg and 100 mg/kg HG2 groups displayed a notable increase in IL-10 expression. These compelling results strongly indicated that HG2 modulates the expression of pulmonary inflammatory factors in mice with pulmonary fibrosis.

TGF-β has a critical role in both macrophage polarization and the development of pulmonary fibrosis: it is synthesized and secreted by M2 macrophages and induces further M2 polarization, thereby performing diverse functions including anti-inflammatory, pro-repair, and pro-fibrotic activities. We conducted RT-qPCR analyses to assess the TGF-β/Smad pathway gene expression levels in mice with BLM-induced pulmonary fibrosis. Compared with the control group, the BLM-induced pulmonary fibrosis group exhibited significantly higher expression of Fmod and Tgfbr2 mRNA; in contrast, the mRNA expression levels of Bambi, Skp1, Zfyve9, Zfyve16, Ppp2ca, Ppp2r1a, Ppp2r1b, Rps6kb1, and Rps6kb2 were markedly lower. Interestingly, HG2 exhibited a clear restorative effect ( Figure 11A ). Ppp2Ca, Ppp2r1b, and Rps6kb2 demonstrated the most pronounced response.

Building on the qPCR findings, we examined protein expression profiles through western blot analysis ( Figures 11B, C , Supplementary Figure S5B ). In contrast to the control mice, mice with pulmonary fibrosis exhibited notably higher levels of lung proteins such as TGF-βR2, SKP-1, CHI3L1, OGG1, and Galectin-3, alongside lower expression of p-4Ebp1, SARA, p-P70S6K, and PP2A. Both nintedanib and the various doses of HG2 displayed clear restorative effects, and the high-dose HG2 group exhibited the most pronounced response.

Our results thus indicated that HG2 significantly modulates pulmonary macrophage polarization and the expression of inflammatory factors in mice with BLM-induced pulmonary fibrosis. This effect appeared to be more potent than that of nintedanib. The co-localization of CD206, F4/80, and collagen I fluorescence signals underscored the importance of assessing pulmonary macrophage polarization in pulmonary fibrosis, particularly the M2 subtype. Notably, despite the elevated expression of M2 macrophage markers in the lungs of mice with pulmonary fibrosis, the elevated levels of dimeric iNOS, CD80, and COX-2 implied that chronic pulmonary inflammation might persist in the pathogenesis of pulmonary fibrosis. HG2 and nintedanib exhibited distinct anti-inflammatory effects from those previously reported, and HG2 was found to demonstrate a more pronounced anti-inflammatory effect ( Figures 7B , 8A , 9A–E ). Furthermore, HG2 exerted an anti-fibrotic role by influencing the expression levels of proteins and genes in the TGF-β/Smad pathway in mice with pulmonary fibrosis. The western blotting results also prompted us to focus on apoptosis and oxidative stress in the lungs of mice with pulmonary fibrosis, where HG2 actively intervenes. Therefore, HG2, an extract from traditional Chinese medicine, targets multiple regulatory facets across various networks.