Antibiotic resistance associated with air quality and transferred by airborne particulate matter

Background: Environmental spread of antibiotic resistance has become a public health problem, but the relationship between air quality and antibiotic resistance has not been fully investigated, the effect and mechanism of air pollutants on horizontal genes transfer of antibiotic resistance are rarely mentioned. Results: In the present study, significant correlation ( P < 0.05) between resistance rates of 10 clinical antibiotic-resistant pathogens and number of days when air quality index was below 100 were observed in an extensive survey from 2014 to 2017 in China. In addition, characterization of antibiotic resistance genes (ARGs), mobile genetic elements (MGEs), and bacterial communities in airborne particulate matter (PM), dust, and human airway sputum samples were profiled in a hospital. PM 2.5 and PM 10 were shown to contain ARGs and MGEs high relative abundance and diversity index. Importantly, transferable multi-resistant plasmids were identified from air, e.g., pTAir-3 with 26 MGEs and 10 ARGs, using conjugative mating assays and nanopore sequencing . Furthermore, PM 2.5 and PM 10 significantly enhanced the conjugative transfer frequencies between bacteria, via inducing increased levels of reactive oxygen species and cell membrane permeability; upregulated gene expression levels were confirmed by genome-wide RNA sequencing. Conclusions: These findings provide a new perspective on antibiotic resistance research and have profound implications for antibiotic resistance control in clinics and environments.

Horizontal gene transfer (HGT) is a major driver for disseminating ARGs within and across bacterial genera, via mobile genetic elements (MGEs), such as plasmids, integrons, transposes, and prophages [12]. In this regard, conjugation is the most dominant HGT mechanism to transfer DNA between donor and recipient bacteria [13]. Increasing antibiotic-resistant bacteria (ARB) infections and multi-environmental spreading of ARGs are associated with increasing patient hospitalization and mortality rates [14][15][16][17].
Antibiotic resistance and air pollution are included in the WHO reports "ten threats to global health in 2019", but 90% of people breathe polluted air every day [1]. Microscopic pollutants infiltrate respiratory and circulatory systems, whose antibiotic resistance potential might exacerbate morbidity and mortality. ARGs are found in a number of aerial environments including hospitals, cattle feed yards, sewage treatment plants, and urban and rural areas [18][19][20][21][22][23][24][25]. These researches demonstrated ARGs presence and abundance in air environments, however, little is known about the linkage between air quality and antibiotic resistance, and the effects and mechanisms of air pollutants on ARGs spread. To address these issues, this study tested the hypothesis that air quality correlates with antibiotic resistance, and particulate matter (PM) promotes HGT of ARGs. Antibiotic resistance and air quality data across different geographical regions in China from 2014 to 2017 were evaluated, including correlation between the number of days with good air quality (here defined as number of days when air quality index is below 100) and antibiotic resistance rates (here defined as percent of pathogens resistant to a particular antibiotic). Furthermore, Sampling sets were designed in a hospital to investigate ARGs, MGEs, and bacteria in the air, dust, and human airway sputum samples, as well as conjugation mating system to identify the effects of PM 2.5 and PM 10  to 2017. The used clinically-relevant bacterial pathogens were mainly derived from sputum specimens (averagely accounting for 41.6%), urine specimens (averagely accounting for 17.6%) and blood specimens (averagely accounting for 9.5%). In China, air quality index (AQI) is divided into six levels, ranging from first-level excellent (0-50), second-level good (51-100), third-level mild pollution (101-150), fourth-level moderate pollution (151-200), to fifth-level severe pollution (201-300), and sixth-level severe pollution (more than 300). The number of days with the AQI below 100 can more effectively represent the air quality of a year than the average concentration of 5 pollutants. To characterize ARGs, MGEs, and bacteria community distribution in clinical aerial environment, 70 samples were collected from a hospital during July 27 to August 2 in 2018, including 17 patient airway (P) sputum samples (the samples were collected from volunteers in Inpatient Respiratory Department) and 11 healthy human airway (H) sputum samples, and respiratory-related environmental samples such as 7 indoor dust (ID), 7 outdoor dust (OD), 7 indoor PM 2.5 (I-PM 2.5 ), 7 outdoor PM 2.5 (O-PM 2.5 ), 7 indoor PM 10 (I-PM 10 ), and 7 outdoor PM 10 (O-PM 10 ) samples (detailed sampling protocols, ethics approval, and DNA extraction procedures are described in Supplementary Information Text S1).
Physiochemical characterization of PM 2.5 and PM 10 samples. coli HB101 and E. coli RP4 was used, as previously reported [29]. E. coli HB101, which harbors RP4 plasmid carrying resistance to tetracycline, ampicillin, and kanamycin, was selected as the donor strain in this study; E. coli NK5449 represented the plasmid recipient. Donor and recipient strains were mixed and treated with PM 2.5 /PM 10 at 31.25, 62.5, 125, 250, and 500 µg/mL for 12 h at pH 7.0 and 37 o C [30]. PM 2.5 and PM 10 solutions were prepared as previously reported [30]. Amount of recipient bacteria and transconjugants were counted on LB-selective solid medium as colony-forming units per milliliter (CFU/mL). Transconjugants were identified using DGGE and amplified by PCR with RNA extraction, genome-wide RNA sequencing, and bioinformatics.
Conjugation mating systems were established as described above, using PM 2.5 and PM 10

Statistical analysis.
Diversity index, Mantel test, and Procrustes analysis were generated by R with 'vegan' packages. Heatmaps of China were generated by R with 'mapdata', 'maptools', 'plyr', 'mapproji', 'sp', 'maps', and 'ggplot2' packages. Linear regression analysis graphs were generated by R with 'ggplot2' package. Heatmap was generated by R with 'pheatmap' 9 packages. PCoA (Bray-Curtis distance based) was generated by CANOCO 5. Box and bar charts were generated by OriginPro 9.0 (OriginLab Corp., Northampton, MA). Network analysis was conducted in the Gephi platform and only statistically robust correlations of Spearman's correlation coefficient (ρ) > 0.9 and significance level P < 0.01 were used to form the final networks by Frucherman Reingold algorithms [26,31,32]. Phylogenetic tree was extracted from the taxa in MEGA7 program, using the maximum-likelihood tree with 500 bootstraps. SourceTracker 1.0.1 was used to estimate the relative contribution of the microbial taxa and ARGs from the source environments to the sink environments by Bayesian methods [33] (details are described in Supplementary Information Text S4).

Relationship between antibiotic resistance and air quality
Four-year average number of days when air quality index was below 100 and antibiotic resistance patterns across China from 2014 to 2017 heatmaps were generated ( Fig. 1 and   S1). Most antibiotic resistance bacteria used in this correlation are from respiratory samples, so there is rationale for the increase of antibiotic resistance with increased air pollution.
Although high air pollution increases respiratory infections, the increase in the antibiotic resistance (percentage of antibiotic resistance bacteria) is unexpected.

Abundant aerial antibiotic resistance genes
Seventy samples including air, dust, and human airway sputum samples were collected for seven consecutive days to further identify ARGs in air or in its relevant matters, observing differential number and Shannon index of detected ARGs in samples ( Fig. 2a and b).
Principal coordinate analysis (PCoA) was performed based on ARGs profiles abundance, using the Bray-Curtis dissimilarity index (Fig. 2c). PCoA showed that I-PM 2.5 , I-PM 10 , O-PM 2.5 , O-PM 10, and OD samples formed one cluster on the projection plot (Fig. 2c), whereas airway sputum and ID samples formed two separate clusters. I-PM 10 and ID samples contained the highest ARGs abundance with an average of 5.4 copies and the least ARGs abundance with an average of 0.5 copies per 16S rDNA gene copies, respectively (Fig. 2d). ARGs abundance had no significant difference between indoor and outdoor samples, whereas ARGs abundances in PM 10 and PM 2.5 were higher than those in dust and airway sputum samples (P < 0.05). ARGs distribution among 8 different samples is shown in Fig. 2e, whereas that of ARG subtypes are shown in the heatmap (Fig. S2).
Genes providing resistance to aminoglycoside and multidrugs were mostly present in dust and air samples. Tetracyclines and MLSB resistance genes contributed to 22% and 53% in human airway sputum, respectively.

Mobile genetic elements
The total abundance of transposase and integron-integrase genes in samples is shown in Figures S3 and S4. PM 10 samples contained higher transposase and integron-integrase abundance than others, with an average of 0.4 and 2.9 copies per 16S rDNA gene, respectively. Furthermore, ARGs abundance significantly (P < 0.01) correlated with that of transposases and integron-integrase genes (Fig. S5).

Microbial Taxa in air
From the 16S ribosomal gene amplicon sequencing, raw data consisted of 3237776 reads for all samples, and quality-filtered data consisted of 2767855 reads with an average value of 43247 reads per sample, corresponding to 829 operational taxonomic units, after 11 filtering and quality control. ID samples harbored more observed species and showed higher bacterial diversity than others ( Fig. S6a and b). PCoA results of bacteria communities indicated that airway sputum and other environmental samples formed two separate clusters (Fig. S6c) Co-occurrence patterns among ARGs, MGEs, and microbial taxa The resultant network consisted of 224 nodes (each node represented a subtype of ARGs, MGEs or microbial taxa) and 440 edges, with a node connectivity of 3.93 (Fig. 3a). The clustering coefficient was 0.73, and the modularity index was 0.89 (values > 0.4), suggesting that the network had a modular structure [34,35]. Based on the modularity class, the entire network was clearly separated into eight major modules. Compared with a random association, clusters of nodes in the same module contained more interactions among themselves than with other nodes. In some modules, some clinical critical microbial taxa were supposed to carry some specific ARGs. For example, Campylobacter spp. positively correlated with ARGs, conferring resistance to fluoroquinolone (vatC-02, mexD), β-lactams (fox5), tetracycline (tet36-01, tetA-01, tetPB-02), and vancomycin (vanSB, vanSE, vanTC-02, vanTE, vanYB) in module I, whereas Staphylococcus spp.

SourceTracker analysis of bacteria and antibiotic resistance genes
SourceTracker was used to analyze the relative contribution of environmental bacteria and ARGs to human airway sputum (Fig. S7). SourceTracker analysis has been performed to identify the extent of contribution of each source to the sink, as previously reported [36][37][38]. It can be used for estimating the relative contributions of taxa and ARGs from exposed environments (sources) in the human airway sputum (sinks). In healthy human airway sputum, I-PM 10 was the highest relative contribution of bacteria from six environments, accounting for 2.5%. In patient airways sputum, O-PM 2.5 and I-PM 2.5 contributed to a higher proportion of bacteria than others, accounting for 3.5% and 3.6%, respectively. Similar trends were observed when SourceTracker analyses were run on ARG data, whereas the relative contribution of ARGs was significantly (P < 0.05) lower than bacterial taxa (Fig. S7b). In the patient airways sputum, O-PM 2.5 contributed to a higher proportion of ARGs than others, accounting for 1.8%.
Conjugative multi-resistant plasmids in air 13 Three conjugative multi-resistant plasmids were demonstrated from air. The plasmid pTAir-3 was 146,145 nucleotides in length, had a GC content of 51% and carried 26 horizontal transfer genes and 10 antibiotic resistance genes, as well as type IV secretion systems and fimbrial genes (Fig. 4). Nine metal resistance genes (including cusB/F copper, silC/E/R silver, and merA/C/D/E/P cadmium resistance genes) and four antibiotic resistance genes (bleO bleomycin, EamA family, and tetG and tetR tetracycline resistance genes) were present in the genomic island regions spanning nucleotide positions 38656-68101; four horizontal transfer genes and three antibiotic resistance genes (qacE disinfectants, aadA21 aminoglycoside, and drfA1 trimethoprim resistance genes) were present in another genomic island regions from 72848-80820. Other conjugative multi-resistance plasmids' genetic maps are shown in Figures S8 and S9.
Effects of PM 2 . 5 and PM 1 0 on conjugative transfer Physiochemical characterization of PM 2.5 and PM 10 samples were described in Table S2, in the supporting information. PM 2.5 and PM 10 size distributions were 832 to 1096 nm and 2884 to 11482 nm, respectively. Significant (P < 0.05) increases in transfer frequency fold changes occurred in the presence of PM 2.5 and PM 10 Fig. 5). For example, at 125 µg/mL, PM 2.5 and PM 10 increased in the conjugative transfer rate by 1.3-and 2.1-fold, respectively, compared with that of control. Effects of PM 2.5 and PM 10 on ROS generation and cell membrane permeability were analyzed. ROS is considered to enhance the frequency of bacterial conjugative transfer [12]. Results of the present study showed that PM 2.5 and PM 10 induced a significant (P < 0.05) increase in donor intracellular ROS (up to 3.6-and 4.6-fold increases for PM 2.5 and PM 10 , respectively, compared with that of control). TEM graphs showed the structure of bacteria during conjugation (Fig. S10).
14 Genes expression during PM 1 0 and PM 2 . 5 exposure Exposure to all levels of PM 10 and PM 2.5 evidenced increased expression of conjugative processes relevant genes (Fig. 6). In this concern, ROS (ytfE, uspF, and napF), conjugation (ihfA and ompA) and pilus generations (htrE, sfmC, ybgP, yehB, and yraJ) genes increased expression during PM exposure. Increased changes in SOS (yafO, recQ, umuC, sulA, and yebG) gene expression were also detected. Genes relevant to drug transmembrane transport (mdtM and ompF) were upregulated. PM 10 and PM 2.5 further increased antibiotic response genes expression.

Discussion
This research provides a new perspective between air quality and human health. By integrating four-year data on resistance rates of ten clinic antibiotic-resistant pathogens and air quality, we compared the relationship between air quality and clinical antibiotic resistance across China. Significant relationship (P < 0.05) between air quality and antibiotic resistance was further demonstrated. We generated robust correlation tests based on four-year data sets, which are universal in most antibiotic-resistant bacteria. The robust correlation can be a powerful driver to an in-depth study of ARGs behavior in air environments.
ARGs in the clinical air environment were comprehensively characterized. ARGs from air may be mobilized and transferred to human commensals and abundant ARGs in PM 10 and PM 2.5 may infiltrate the respiratory system, causing an increased health risk to humans. In addition, air environments may be suitable ARGs carriers for long-distance propagation. A probable reason to explain the high abundance of ARGs and MGEs in air samples is that air environments contain low nutrition, high UV exposures, and multiple environmental pollution particles, which agree with previous reports demonstrating that such factors may increase abundance and transfer of ARGs in microbes [29,39,40]. In microbial taxa, Proteobacteria were mostly found in dust and air samples, whereas Firmicutes prevail in airway sputum samples, as reported by others [25]. Furthermore, daylight exposure can affect dust communities [41], particularly, Actinobacteria are associated with darkness and light exposure, which may be the reason for a higher abundance of Actinobacteria in ID samples than in other environments. Human beings are surrounded by dust and air, whose ARGs and bacteria content are relevant to public health. Moreover, the significant correlation between bacteria community and ARGs indicated bacterial communities may shape ARGs distribution and abundance in air.
Network analysis indicated co-spread of ARGs, MGEs, and bacteria among air, dust, and human sputum. Co-occurrence between fluoroquinolone resistance gene and Campylobacter spp., methicillin/ vancomycin resistance genes, and Staphylococcus spp.
were observed. Campylobacter spp. and Staphylococcus spp. are zoonotic disease pathogens, exhibiting high levels of acquired antibiotic resistance, which has a worldwide public health impact [42][43][44]. In particular, fluoroquinolone-resistant Campylobacter spp. and methicillin/vancomycin-resistant Staphylococcus aureus are "high priority pathogens", according to the WHO; Campylobacter spp. causes serious diarrhea and Staphylococcus aureus is associated with skin infections, pneumonia, and bloodstream infections [45].
Methicillin-resistant Staphylococcus aureus has previously been found in hospital and poultry farming air environments [46,47]. Prevotella which is known to be resistant to tetracycline [48], shows strong associations with tetracycline resistance genes such as tetA-01, tet(36)-01, and tetPB-02; Veillonella, which is resistant to erythromycin and gentamicin, correlated to MLSB resistance genes (ermK-02) in the present study.
Furthermore, Prevotella and Veillonella are commonly detected in airway sputum in high abundance; these genera can potentially influence airway microbiota composition by their metabolic products or by forming intergeneric coaggregates with other microbes [49][50][51].
Moreover, co-occurrence between these genera and MGEs (IS613, intI2) were observed; MGEs may contribute to ARGs spread in the microbial community via horizontal gene transfer (HGT). In this regard, Campylobacter spp., Prevotella, multiple ARGs, and MGEs co-occurred in module I. This module may represent a typical community revealing the potential co-spread pattern of bacteria and ARGs in the environments, in which common bacteria may maintain the structure of community with pathogens and MGEs are more actively transferring resistance genes among microbial communities to resist environmental exposures.
Based on SourceTracker analysis, the relative contributions of bacteria and ARGs from PM 2.5 samples are higher than that of other environmental samples, probably due to the increased infiltration of small particles to the airways. The relative contributions of environmental microbial to human airway sputum are higher than those of ARGs. It is reasonable that bacteria can be inhaled into human airways, whereas competitions from growth and reproduction may lead to the loss of antibiotic-resistant bacteria or loss of ARGs, and some extracellular ARGs degradation in human airways. Researchers have also noted that antibiotic resistance results in reduced fitness, compared with that of antibiotic-sensitive bacteria, which is known as the fitness cost [52,53]. Furthermore, most bacterial community and ARGs contributions in airway sputum are predicted to be of unknown origin, which may result from the large variety of individuals and other factor contributions, such as age, foodborne origin, home environment, personal care products, and different ecological conditions [54].
Importantly, conjugative multi-resistance plasmids were identified from the air. It seems possible that the plasmid can be transported by air, and transfer ARGs into human commensals at suitable conditions. pTAir-3 carried gene resistance to aminoglycoside, bleomycin, quinolone, tetracycline, trimethoprim, copper, and mercury. Plasmids carrying antibiotic and metal resistance genes allow bacteria to survive under difficult air conditions. Furthermore, pTAir-3 carried intI1, transposons, type IV secretion systems and fimbrial genes, which related to conjugation transfer processes. This research provides a feasible and effective way to identify the conjugative multi-resistance plasmids in the air. We also investigated potential mechanisms to explain the role of PM 2.5 and PM 10 in the enhancement of conjugative transfer of RP4 plasmid. Multiple investigative approaches were used to reveal the underlying mechanisms, including flow cytometry and TEM to measure ROS response level and membrane permeability, and whole-genome RNA sequencing to evaluate gene expression. We verified the increased ROS level and membrane permeability with exposure to PM. ROS was activated in the presence of PM 10 and PM 2.5 ; increased ROS levels induces SOS, which damage cell membranes. This can weaken the barriers of cell membranes and promote horizontal ARGs dissemination [38].
RNA genomics revealed comprehensive gene expression of bacteria during PM exposure, 18 which provides solid data to uncover the physiological changes of bacteria. Genes of pilus generation were upregulated on exposure to PM, which could facilitate physical cell-to-cell contact between donor and recipients, and enhance plasmid transfer. In addition, PM could increase the expression of genes response to antibiotic and drug transmembrane transport, augmenting bacteria antibiotic tolerance. Thus, PM 2.5 and PM 10 were confirmed to promote ARGs transfer.

Conclusion
In conclusion, this study demonstrates a significant relationship between resistance rates of ten clinic antibiotic-resistant pathogens and air quality. Abundant ARGs were found in clinical air, which may accelerate ARGs transfer between distant places. Importantly, conjugative multi-resistance plasmids and other MGEs in the air may frequently transfer between environments and within bacteria, exacerbating ARGs dissemination. In addition,

Ethics approval and consent to participate
Sputum samples were obtained from volunteers in the Inpatient Respiratory Department with approval of the Zhejiang University School of Medicine ethics committee, and all subjects were provided with written informed consent before participation.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no conflict of interest.