Chronic exposure to ambient particulate matter induces gut microbial dysbiosis in a rat COPD model

Figure 1a depicts the design of the study. Rats were exposed to PM as described previously [6]. Briefly, all animals were exposed in whole-body inhalation chambers for 4 h/day, 5 days/week either 4, 12, or 24 weeks. PM mass concentrations, particle size distributions, and gas concentrations were monitored each day of exposure. DustTrak II aerosol monitors (model 8530, TSI, Shoreview, MN, USA) were used to monitored PM mass concentrations and particle size distributions. Testo 340 portable flue gas analyzers (Testo, Lenzkirch, Germany) were used to monitored gas concentrations (O₂, carbon monoxide, nitrogen oxides, and sulfur dioxide) in the exposure rooms. The control group was exposed to clean air.

Spirometry data were obtained as previously described using a Forced Pulmonary Maneuver System (Buxco Research Systems, Wilmington, NC, USA) [6]. Rat were sedated with 3% pentobarbital (1 mL/Kg), tracheostomized and intubated, then placed supine in the body chamber and connected to the system. According to the procedures, the FRC (functional residual capacity), FEV20 (forced expiratory volume in 20 s), FEV100 (forced expiratory volume in 100 s) and PEF (peak expiratory flow) were measure. At least three acceptable maneuvers for each test of every mice were conducted to obtain a reliable mean spirometry data.

Rats were sacrificed by CO₂ after 4, 12, and 24 weeks of exposure (on days 29, 85, and 169, respectively). Blood samples were collected from the heart and centrifuged at 1700×g for 15 min at 4 °C. Serum was stored at − 80 °C. Proximal colon contents were harvested using sterile instruments for each individual animal and site. Fresh proximal colon contents samples were snap-frozen in liquid nitrogen then stored at − 80 °C for microbial and SCFA analysis.

Bronchoalveolar lavage fluid (BALF) was collected as previously reported [14]. Cells were isolated by centrifugation at 300×g for 10 min at 4 °C and stained with Diff-Quik stain (Baso Diagnostics, Zhuhai, China). Differential cell counts were assessed from 400 cells counted on each slide.

As described previously [6], lung tissues were fixed with 4% paraformaldehyde solution and embedded in paraffin using standard methods. Sectioning and staining were performed by the Pathology Center of the First Affiliated Hospital of Guangzhou Medical University (Guangzhou, China). All slides were scanned and analyzed using an image analyzer platform (Leica, Wetzlar, Germany). Alveolar enlargement and destruction, and the bronchial wall thickness was calculated as describe previous [15].

Serum levels of LPS were measured using a commercial chromogenic end-point TAL kit (Xiamen Bioendo Technology Co., Ltd., Xiamen, China). All procedures were performed according to the manufacturer’s instructions. The total protein in the BALF determined by Bicinchoninic Acid (BCA) method using a commercial BCA Protein Assay Kit (Thermo Fisher Scientific, Waltham, USA). The concentration of endotoxin from DMSO-extract of particulate matter were performed at the Kingmed Diagnostics Center (Guangzhou, China). All procedures were performed according to the Pharmacopoeia of the China (2015 edition) volume IV.

DNA was isolated from colon contents and BALF samples using a Qiagen QIAamp® DNA extraction kit (Qiagen, Hilden, Germany) according to the manufacturer’s recommendations. 16S rRNA gene amplification, in vitro transcription, and labeling and hybridization were performed following the Illumina 16S Metagenomic Sequencing Library Preparation guide [16]. We used a MiSeq rRNA amplicon sequencing protocol to PCR-amplify the V3–V4 variable regions (amplicon size expected: approximately 460 bp). 16S amplicon PCR forward primer was 5′-(TCG TCG GCA GCG TCA GAT GTG TAT AAG AGA CAG CCT ACG GGN GGC WGC AG)-3′ and 16S amplicon PCR reverse primer was 5′-(GTC TCG TGG GCT CGG AGA TGT GTA TAA GAG ACA GGA CTA CHV GGG TAT CTA ATC C)-3′ [17]. All samples were paired-end sequenced on an Illumina MiSeq PE250 platform (San Diego, CA, USA) by the RiboBio Genome Center (Guangzhou, China). 16S rRNA gene sequence analysis, including raw sequence filtering and taxonomic classification, was performed as described previously [18]. The bar diagrams of alpha diversity indices and relative abundance were drawn using GraphPad Prism 8 software (La Jolla, CA, USA).

Seven SCFAs (acetic, propionic, butyric, isobutyric, valeric, isovaleric, and caproic acids) in the proximal colon were measured by high-performance gas chromatography-mass spectrometry (Agilent 6890 N; Agilent Technologies, Santa Clara, CA, USA) according to the manufacturer’s recommendations. Briefly, a total of 100 mg of the colon contents were homogenized by ultrasonication in 600-μL reactions containing 100 μL of phosphoric acid (15%), 100 μL of 4-methylvaleric acid (250 μg/mL), and 400 μL of diethyl ether. The mixtures were vortexed and centrifuged until a clear supernatant was obtained. The sample was subjected to a high-performance liquid chromatography column (Agilent Technologies) for analysis of SCFAs.

Data are reported as mean ± standard deviation (SD). Comparisons were performed using ANOVA and p values were corrected for multiple testing with the Bonferroni method. Statistical analysis was performed using SPSS version 24 (IBM SPSS, Armonk, NY, USA). Correlations between serum LPS levels and the pulmonary mean linear intercept (MLI) were assessed using Spearman’s rank correlation. p < 0.05 was considered significant.

Particle size distributions and gas concentrations were measured during PM exposure. Figure 1b and supplementary Table 1 show the concentrations of PM with a diameter ≤ 10, 2.5, and 1 μm (PM₁₀, PM_2.5, and PM₁) during BMF and MVE exposures. Average concentrations of PM₁₀, PM_2.5, and PM₁ in the BMF exposure were 25.87 ± 2.99 mg/m³, 21.91 ± 1.84 mg/m³, and 19.60 ± 1.76 mg/m³, respectively. In the MVE exposure, average concentrations were 1.86 ± 0.20 mg/m³, 1.85 ± 0.24 mg/m³, and 1.82 ± 0.27 mg/m³, respectively. BMF exposure caused notably higher particulate emissions. Health impact assessments of PM considering not only mass concentration but also the composition of atmospheric aerosol. Supplementary Table 1 lists the mean ± SD concentrations of O₂, carbon monoxide, nitrogen oxides, and sulfur dioxide during exposure. Supplementary Tables 2, 3 and 4 list the elemental composition of the particulates. The exposure concentrations of BMF and MVE reported here are based on our previous research on indoor air pollution in rural areas of China [4, 5, 19], and with those online reporting of the heavy haze conditions [20]. In contrast with our previous study [6], we improved the ventilation system and burning process to reduce the concentration of nitrogen oxides during the exposure, which is more in line with the real-world exposure.

In order to evaluate the concentration of LPS attached to particulate matter, we tested the concentration of endotoxin from DMSO-extract of particulate matter, which collected from BMF exposure room, MVE exposure room and the clean air. The endotoxin levels from BMF exposure room, MVE exposure room and the clean air were 12.0 EU/ml, 10.0 EU/ml and 12.5 EU/ml, respectively (Supplementary Table 5). Our data suggest that there is no significant difference in endotoxin levels from the three group during the exposure.

Exposure to PM affected body weight and induced pulmonary inflammation. Rats continued to gain weight during PM exposure, with a total weight gain of 162.3% (controls), 135.2% (BMF group), and 130.4% (MVE group). The BMF and MVE groups maintained a steady rate of weight gain throughout the exposure period, but the rate was slower than the rate of weight gain in the controls after 4 weeks (p < 0.01; Fig. 2a). The rats exposed to BMF or MVE had more sediment particulate matter in BALF as early as 4 weeks into the exposure period (Fig. 2b). The levels of total protein in BALF were both higher in the MVE group or BMF group than in controls since a 4-week exposure (p < 0.05 or p < 0.01, Fig. 2c). Moreover, a significant increase in total leukocyte counts were observed both in the BMF group and MVE group (p < 0.05 or p < 0.01, Fig. 2d). The total leukocyte counts started to increase greatly since a 4-week exposure and remained elevated to the 24 weeks of exposure. Differential cell counts showing that increase in total leukocyte counts were due mainly to increases in alveolar macrophages and neutrophils counts (p < 0.05 or p < 0.01, Fig. 2e-f), and the alveolar macrophages have the highest percentage in total leukocyte counts.

Exposure to PM also induced COPD-like changes in rat lung. Histological analysis showed that significant increases in the mean linear intercept after 24 weeks of exposure in both the BMF (p < 0.01) and MVE (p < 0.01) groups compared to controls (Fig. 3a). Long-term PM exposure damaged the lung parenchyma. After 24 weeks BMF and MVE particles exposure, the thickness of bronchial walls in the lungs of rats was increase greatly to that of controls (p < 0.01, Fig. 3b), suggesting that the PM exposures induced airway remodeling. Moreover, after 24-weeks of exposure, the pulmonary function test results of FRC were significantly increased, but the PEF, FEV₂₀/FVC and FEV₁₀₀ were decreased significantly both in the BMF group and MVE group than from controls (p < 0.01, Fig. 3c). Taken together, these findings suggest that chronic exposure to BMF and MVE induced COPD-like changes in the rat lung.

To determine whether chronic exposure to ambient PM induces intestinal and lung microbial shifts, we performed 16S rRNA gene sequencing of proximal colon contents and BALF samples from rats exposed to BMF or MVE after 4, 12, and 24 weeks of exposure.

For gut microbial analysis, after filtering for low-quality reads, 4,479,175 sequence reads from 54 samples were used for analysis and resulted in 22,000 operational taxonomic units (OTU). Comparisons between the three groups showed that intra-individual diversity, as measured by the number of OTUs, decreased significantly in the MVE group after 24 weeks of exposure (p = 0.062, vs CON group; p = 0.122, vs BMF group, Fig. 4a). Other indices (Chao 1and PD_whole_tree) were calculated to estimate the within-sample (alpha) diversity. The Chao 1 (p = 0.090, vs CON group; Fig. 4b) and PD_whole_tree indices (p = 0.216, vs CON group, Fig. 4c) were lower in the MVE group after 24 weeks exposure. The reduced richness of OTUs and alpha diversity in the gut microbiota suggest a possible deficiency in healthy microflora in the BMF and MVE groups. To investigate the difference between gut microbial communities in the three groups, we analyzed the taxonomical community structure of the microbiome in colon contents. At the phylum level, all samples at all measurement time points from the three groups contained four major bacterial phyla: Bacteroidetes, Firmicutes, Proteobacteria, and Actinobacteria. The first three phyla accounted for over 97% of the total sequences in all three groups (Fig. 5). After a 4-week exposure, the MVE group had a lower relative abundance of Firmicutes (p = 0.003 vs. the control group; p = 0.002 vs. the BMF group) but a higher relative abundance of Proteobacteria (p = 0.099 vs. the control group; p = 0.055 vs. the BMF group, Fig. 5a). After a 12-week exposure, the MVE group exhibited relative abundances of Firmicutes (p = 0.019 vs. the control group; p = 0.072 vs. the BMF group, Fig. 5b) that were consistent with the trend seen after 4 weeks of exposure. However, after 24 weeks of exposure, the three groups exhibited a similar relative abundance of four major bacterial phyla (p > 0.05, Fig. 5c). These findings indicate gut microbiota dysbiosis after PM exposure.

For lung microbial analysis, high-quality sequence reads were used for subsequent analyses and resulted in 28,336 OTUs. As shown in Fig. 6a, there were no significant difference of number of OTUs between the three groups in 4, 12, and 24 weeks time-point (p > 0.05). Chao 1 and the Shannon index indicated that there was no significant difference between the three groups in 4, 12, and 24 weeks time-point (p > 0.05, Fig. 6b-c). At the phylum level, all samples from the control group, BMF group and MVE group also contained the four major bacterial phyla: Bacteroidetes, Firmicutes, Proteobacteria, and Actinobacteria. Despite the control group, BMF group and MVE group showed different microbial composition in 4 and 12 weeks, there was no statistically significant difference with its limited sample size (p > 0.05, Fig. 6d-f).

We analyzed the levels of SCFAs in colon contents following 4- and 24-week exposures using gas chromatography. Acetic, propionic, and caproic acids were detected in all colon contents, but butyric, isobutyric, valeric, and isovaleric acids were not detected in some samples. Levels of total SCFAs were significantly lower in the MVE group than in controls (p = 0.125) after 4 weeks and 24-week exposure (p = 0.041, Fig. 7a). Moreover, a significant decrease in acetic acid was observed in colon contents from the MVE group relative to contents in controls (p = 0.036, Fig. 7b). Propionic and caproic acids exhibited similar levels after PM exposure (Fig. 7c, d).

Serum LPS levels were higher in the MVE group than in controls after a 4-week exposure (p = 0.054). After a 24-week exposure, serum LPS levels were significantly higher in both the BMF (p = 0.007) and MVE (p = 0.007) groups than in controls (Fig. 7e), and exhibited a weak but significant positive correlation with the pulmonary mean linear intercept (r² = 0.4552, p < 0.001; Fig. 7f).