Characteristics of Airborne Bacterial Communities in Indoor and Outdoor Environments During Continuous Haze Events in Beijing I

Environment International 139 (2020) 105721

Contents lists available at ScienceDirect

Environment International

journal homepage: www.elsevier.com/locate/envint

Characteristics of airborne bacterial communities in indoor and outdoor T environments during continuous haze events in Beijing: Implications for health care Jianguo Guoa,b, Yi Xiongc, Changhua Shia,b, Ce Liud, Hongwei Lia,b, Hua Qiane, Zongke Sunf, ⁎ Chuan Qina,b, a NHC Key Laboratory of Human Disease Comparative Medicine (The Institute of Laboratory Animal Sciences, CAMS&PUMC), Beijing 100021, China b Key Laboratory of Human Diseases Animal Model, State Administration of Traditional Chinese Medicine, Beijing 100021, China c Department of Food Science and Engineering, School of Chemistry and Chemical Engineering, Harbin Institute of Technology, Harbin 150001, China d Department of Infectious Disease, Beijing Chuiyangliu Hospital, Affiliated with Tsinghua University, Beijing 100022, China e School of Energy and Environment, Southeast University, Nanjing 211189, China f National Institute of Environmental Health, China CDC, Beijing 100021, China

ARTICLE INFO ABSTRACT

Handling editor: Adrian Covaci There is solid evidence that haze pollution threatens human health owing to the abiotic pollutants it contains. Keywords: However, the characteristics of airborne bacterial communities in indoor and outdoor environments exhibiting Airborne bacterial community haze occurrence are still unknown. Thus, we examined variations in both indoor and outdoor airborne bacterial Haze events communities in Beijing from December 9–27, 2016, a period which included three haze events. The outdoor Indoor environment airborne bacterial communities were clustered into two main groups (Groups I and II), and they shifted between Outdoor environment two typical bacterial communities regardless of the haze event. The Chao1, Shannon, and phylogenetic diversity Respiratory disease indexes and abundance of dominant classes changed significantly, as did airborne bacterial community type. The Pneumonia indoor airborne bacterial community closely tracked the outdoor bacterial community type, forming two obvious groups supported by Adonis analysis, changes in dominant classes, and bacterial diversity compared to the outdoor group. Furthermore, we found that the airborne bacterial community type could affect the morbidity of respiratory diseases. Daily pneumonia cases were significantly higher in Group I(p = 0.035), whereas daily amygdalitis cases were significantly higher in Group II(p = 0.025). Interestingly, the enriched classes in the indoor environment were quite different from those in the typical airborne bacterial community environment, except for Clostridia, which had significantly higher abundance in both indoor environments. In conclusion, we found that the two indoor and outdoor airborne bacterial community types changed independently of haze events, and the special airborne bacterial community type was closely related to the incidence of pneumonia in the heavy haze season.

1. Introduction NOx, CO, VOCs, PM10, and PM2.5 emissions were reduced by 20–40% in 2020 (Guo et al. 2018). Many other measures have been implemented

The annual concentrations of PM2.5 have increased by 48% in China to reduce the level of air pollution. However, there is still a long way to and an increasing trend was observed in 87.9% of the country ac- go. The level of air pollution in China is substantially higher than that in cording to data from 1999 to 2016 (Zhao et al. 2019). High PM2.5 levels developed countries, and its threat to public health has attracted a large were mainly located in the Beijing-Tianjin-Hebei region in North China, amount of attention, particularly in relation to haze events (Guan et al.

East China, and the Henan province (Lin et al. 2014). The annual PM2.5 2016). 3 concentrations in >10% of this area have been > 35 μg/m since 2003 Haze pollution, and especially fine particulate matter (PM2.5), can (Zhaoet al. 2019). A total of 83% of the population lives in areas where increase mortality from heart disease (Pope et al. 2009; Pope et al. 3 the annual PM2.5 level exceeds 35 μg/m (Liu et al. 2016). A coal cap 2004), stroke, respiratory disease (Zhang and Cao 2015), and lung policy was implemented in the Beijing-Tianjin-Hebei region, and SO2, cancer (Lepeule et al. 2012; Pope et al. 2011), and can shorten

⁎ Corresponding author at: Pan Jia Yuan Nan Li No. 5, Chao Yang District, Beijing 100021, China. E-mail address: [email protected] (C. Qin). https://doi.org/10.1016/j.envint.2020.105721 Received 1 November 2019; Received in revised form 2 April 2020; Accepted 3 April 2020 Available online 16 April 2020 0160-4120/ © 2020 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/). J. Guo, et al. Environment International 139 (2020) 105721 )

)a( )b( 3 200 PM2.5 PM10 150 100 50

Concentration (ug/m 0 Jul-16 Oct-16 Apr-16 Jun-16 Jan-17 Mar-17 Feb-17 Nov-16 Dec-16 Aug-16 Sep-16 May-16

(c)

TC1 TP2 TC3 TP4 TC5 TP6 TC7 3 μ g/m and PM10 Concentration of PM2.5

12/9a12/9p 12/10a12/10p12/11a12/11p12/12a12/12p12/13a12/13p12/14a12/14p12/15a12/15p12/16a12/16p12/17a12/17p12/18a12/18p12/19a12/19p12/20a12/20p12/21a12/21p12/22a12/22p12/23a12/23p12/24a12/24p12/25a12/25p12/26a12/26p12/27a12/27p

Fig. 1. Sampling details. (a) Distribution of sampling sites. The room (307, 311, 315) represents the indoor environment, and balcony (Y) represents the outdoor environment. (b) Mean concentrations of PM2.5 and PM10 for each month between April 2016 and March 2017. (c) Concentrations of PM2.5 and PM10 at each sampling interval between December 9 and 27, 2016. TC: time interval < 75 μg/m3; TP: time interval > 75 μg/m3 (note: time intervals were numbered in chronological order). lifespans. Data from six low- and middle-income countries showed that outdoors (Kembel et al. 2012). Our knowledge of indoor microbial 3 each 10 μg/m increase of PM2.5 corresponded to a 0.72 increase in communities has greatly increased in recent years owing to the devel- overall disability scores based on the World Health Organization Dis- opment of high-throughput sequencing technology. This technology has ability Assessment Schedule (Lin et al. 2017). PM2.5 exposure caused shown that the indoor microbial community is affected by building 4.2 million deaths (7.6% of total global deaths) and 103.1 million factors, which govern the entry behavior of outdoor microorganisms, as disability-adjusted life-years (DALYs) (4.2% of global DALYs) in 2015 well as by occupants, who release human-associated microorganisms (Cohen et al. 2017). Premature mortalities attributed to PM2.5 in China and re-suspend microbes residing on floors and surfaces (Leung and Lee amounted to 1.37 million in 2013 (Liuet al. 2016). 2016). Therefore, both the outdoor environment and occupants act as Airborne microorganisms are crucial components of the atmo- major sources of indoor microorganisms (Adams et al. 2015; Hospodsky sphere, with bacterial cells exceeding 1 × 104 m−3 (Burrows et al. et al. 2012; Leung and Lee 2016; Stephens et al. 2015). 2009), and they can threaten human health by disseminating allergens In Beijing, severe haze pollution occurs frequently in winter, when and pathogens (Griffin 2007). Many researchers have investigated pathogenic bacteria and fungi are more abundant (Duet al. 2018). Al- structural variation in the airborne bacterial community during haze though microbial composition varies between settings, bacteria are events (Du et al. 2018; Sun et al. 2018; Yan et al. 2018; Zhen et al. generally the most abundant microbes (approximately 80%), followed 2017). However, comparisons of the airborne bacterial community on by fungi (Zhai et al. 2018). Therefore, in this study, we investigated haze and non-haze days have yielded inconsistent results. Fan et al. variations in airborne bacterial communities in indoor and outdoor screened several dominant genera in PM2.5 on heavy or severe pollution environments in Beijing from December 9–27, 2016, a period which days (Fan et al. 2019), and Sun et al. found that the dominant genera included three haze cycles. In addition, we explored the reason for the varied under different concentrations of PM including Rhizobium, Pro- shift in the bacterial community and the relationship between bacterial pionibacterium and Comamonadaceae (Abd Aziz et al. 2018; Sunet al. community structure and respiratory disease. Our results have im- 2018), while no correlation between dominant genera and pollution portant implications for health care during haze events. levels was observed in other studies (Duet al. 2018; Wei et al. 2016; Xu et al. 2017b; Yanet al. 2018). These inconsistent results may be largely 2. Material and methods due to the limited numbers of samples and lack of continuous sampling (Dong et al. 2016; Duet al. 2018), as data on unsampled days might 2.1. Exposure experiment help explain the correlation between airborne bacterial communities and air pollution. Therefore, a better understanding of the dynamic In this study, we sampled the indoor and outdoor environments on variation in airborne bacterial communities throughout the haze pro- the third floor of a three-story building at approximately 10 mabove cess is important for human health risk evaluation. the ground (116°26′35.86″E, 39°52′22.30″N). We selected four sam- Indoor airborne microorganisms are also an important considera- pling sites: one in each of three different student dormitories (307, 311, tion in human health risk evaluations during haze events, because hu- and 315), and one on a balcony (Y) (Fig. 1a). The balcony was open to mans spend most of their time indoors (up to 90% in industrialized the outdoor air, with no barriers. The samples collected from 307, 311, countries [(Hoppe and Martinac 1998; Klepeis et al. 2001)]). The in- and 315 were designated as indoor samples. The samples collected from door airborne bacterial community is influenced by ventilation type, the balcony represented the outdoor environment. There was no me- and the relative number of several operational taxonomic units (OTUs) chanical ventilation indoors. Indoor and outdoor air exchange in these (similarity to human pathogens > 97%) is higher indoors than rooms is normally conducted through natural ventilation. The windows

2 J. Guo, et al. Environment International 139 (2020) 105721 of rooms were kept closed during the experiment. There were stairs at using the Ribosomal Database Project classifier (confidence both ends of the corridor. The corridor connected the rooms and the threshold = 70%) (Wang et al. 2007). Each sample was rarefied to outdoor environment. Occupants worked and rested as usual during the 19,363 sequences, and the Chao1 richness estimator, Shannon index, experiment and forbade people from other dorms to enter. The trend of and Faith’s phylogenetic diversity index (PD index) (Faith and Baker PM2.5 and PM10 from April 2016 to March 2017 shows that the most 2007) were calculated using mothur software. We used PICRUSt to infer serious haze pollution occurred in the winter, especially in December the function of the bacterial community. (Fig. 1b). Thus, we collected samples from December 9–27, 2016. In the ambient air quality standards in China (GB 3095–2012), a 24-hour 2.4. Collection of records from a fever clinic average concentration of 75 μg/m3 is considered the threshold of the second function areas, including dwelling district. Three haze events For the collection of patient records, we selected Beijing 3 (>75 μg/m ) occurred during the sampling period (TP2, TP4, and Chuiyangliu Hospital (1.6 km from the sampling sites), where there is TP6); these were interspersed with non-haze periods (TC1, TC3, TC5, no delay in registration to see a doctor, so actual disease occurrence is and TC7) (Fig. 1c). apparent. Data were collected from the fever clinic during the experi- We used natural sedimentation to collect samples covering the ments to represent the daily disease cases in the area. The records whole time period. At each site, we placed four uncovered disposable provided patient data, including gender, age, date of visit, and diag- sterile cell culture dishes (φ90 mm) without agar medium at a height of nosis. We screened all cases related to respiratory disease and classified 1.5 m above the ground and left them for 12 h. Samples were then them as pneumonia, amygdalitis, or pharyngitis. Finally, we summar- collected by smearing the inner surface of each dish with a sterile swab ized the cases of each disease for each day and studied the relationship that had been dipped in 100 μL of sterile normal saline, and the swabs between bacterial community type and disease morbidity. were stored individually in sterile 2 mL tubes at − 80 °C until further processing. We then replaced the cell culture dishes for the next 12 h 2.5. Statistical analysis sampling period. We placed dishes and collected samples at 08:00 and 20:00 h from December 9–27, 2016, wearing masks and gloves at all We performed canonical correspondence analysis (CCA), principal times. We also recorded the environmental conditions outdoors (PM , 2.5 coordinate analysis (PCoA), and a permutational multivariate analysis PM , SO , NO , NO, O , temperature, moisture, and wind scale) and 10 2 2 3 of variance (Adonis) in the “vegan” package as well as a Mann-Whitney indoors (temperature and moisture) during the sampling period. U test (two-tailed) and Pearson correlation analysis (two-tailed) in the “state” package in the R environment. A significance level of p < 0.05 2.2. Next-generation sequencing was used for all analyses. CCA was performed to explore the relationships in the outdoor airborne bacterial community based on the OTU Genomic DNA was extracted from each sample using a PowerSoil® matrix and outdoor environmental variables (PM , PM , SO , NO , DNA Isolation Kit (MO BIO Laboratories, Inc., Carlsbad, CA, USA) ac- 2.5 10 2 2 NO, O , temperature, moisture, and wind scale). The variables that cording to the manufacturer’s recommendations. The extracted DNA 3 significantly affected the distributions of bacterial communities were was diluted to a concentration of 1 ng/μL and stored at − 20 °C until included in the CCA model. PCoA with Bray–Curtis distances was used further processing. The diluted DNA was used as a template for PCR to explore the similarities between samples. Adonis analysis with amplification of the bacterial 16S rRNA gene using barcoded primers Bray–Curtis distances was performed to partition distance matrices and HiFi HotStart ReadyMix (KAPA). The V3–V4 variable regions of the among sources of variation. We used the Mann-Whitney U test to 16S rRNA gene were amplified with the universal primers 343F compare the abundance of taxa in different bacterial community types (TACGGRAGGCAGCAG) and 798R (AGGGTATCTAATCCT). The am- and different taxa between indoor and outdoor environments. Pearson plicon quality was visualized by gel electrophoresis. Next, the ampli- correlation analysis was used to explore the relationships between cons were purified with AMPure XP beads, amplified for another round biodiversity at the sampling sites (307, 311, 315, Y) and the relation- of PCR, and then re-purified with AMPure XP beads. The final amplicon ship between bacterial taxa and daily disease cases. In order to describe concentration was quantified using a Qubit™ dsDNA Assay Kit. Equal OTU fluctuations over time, we introduced the concept of appeared and amounts of the purified amplicons were then pooled for subsequent disappeared OTUs. The appeared OTUs were those that were detected sequencing using the MiSeq Sequencing System (Illumina, Inc., San in a sample but not detected in the sample from the previous 12 h. The Diego, CA, USA). disappeared OTUs were those that were detected in the sample of a previous 12 h but not detected in the current sample. Appeared and 2.3. Sequence processing disappeared OTUs of each outdoor samples were calculated except the first one (12.9a) without data of the previous 12 h. The counts ofthe Raw sequencing data were stored in the FASTQ format. Paired-end appeared and disappeared OTUs and their abundance were used to reads were preprocessed using Trimmomatic software (Bolger et al. track their changes over time. 2014) to detect and cut off ambiguous bases (N) as well as low-quality sequences with an average quality score of < 20 using the sliding window trimming approach. Following trimming, paired-end reads 2.6. HYSPLIT trajectory model were assembled using FLASH software (Reyon et al. 2012) with the following parameters: 10 bp and 200 bp of minimum and maximum The National Oceanic and Atmospheric Administration (NOAA) overlap, respectively, and 20% maximum mismatch rate. Further de- HYSPLIT model of backward trajectories of airflow over 72 h at the noising of the sequences was conducted; this involved abandoning study sites for each day was built based on GFSG meteorological data reads with ambiguous homologous sequences and those with < 200 bp. (https://ready.arl.noaa.gov/HYSPLIT.php), which was used to de- Reads with 75% of bases above Q20 were retained, and on further termine the source and vertical motion of airflow. detection any reads with chimera were removed using QIIME software (version 1.8.0) (Caporaso et al. 2010). 2.7. Nucleotide sequence accession Clean reads were subjected to primer sequence removal and clus- tering to generate OTUs using UPARSE software with a 97% similarity All bacterial 16S rRNA gene sequences generated in the present cut off (Edgar 2013). A representative read of each OTU was then se- study were deposited in the National Center for Biotechnology lected using the QIIME package. All representative reads were anno- Information Sequence Read Archive (http://www.ncbi.nlm.nih.gov/ tated and blasted against the SILVA database version 123 (16 s rDNA) sra) under accession number PRJNA503826.

3 J. Guo, et al. Environment International 139 (2020) 105721

)a( )b(

(d) TC1 TP2 TC5 TP6 TC7 TP3 TP4

Bacteroidia Clostridia 0.6 Gammaproteobacteria Bacilli Actinobacteria Epsilonproteobacteria (c) Erysipelotrichia Alphaproteobacteria 0.4 Deltaproteobacteria Betaproteobacteria Coriobacteriia 12.15p Fusobacteriia 12.25p Gemmatimonadetes Acidimicrobiia Cytophagia 0.2 12.25a Sphingobacteriia Mollicutes Flavobacteriia I II Negativicutes Thermoleophilia 12.15a Spirochaetes 0 12 /9 12 /10 12 /11 12 /12 12 /13 12 /14 12 /15 12 /16 12 /17 12 /18 12 /19 12 /20 12 /21 12 /22 12 /23 12 /24 12 /25 12 /26 12 /27

(e) (f) (g)

Fig. 2. Analysis of the outdoor airborne bacterial community. (a) Canonical correspondence analysis (CCA) of the outdoor airborne bacterial community based on the OTU matrix and outdoor environmental variables. (b) Proportion of the total variation of the outdoor airborne bacterial community explained by significantly affected environmental variables. (c) Principal coordinate analysis (PCoA) of outdoor airborne samples collected between December 9 and 27, 2016,basedon

Bray–Curtis distances. Different colors represent different time intervals in terms of air pollution levels. Point sizes represent thePM2.5 concentration. (d) Changes in abundance of dominant classes (mean abundance > 0.001) in the outdoors over time. (e) Appeared and disappeared OTU counts over time. (f) Abundance of total appeared and disappeared OTUs over time. (g) Changes in the Chao1, Shannon, and Faith’s PD indexes over time.

3. Results 3.1. Shift in outdoor airborne bacterial communities

We collected 145 samples in total from December 9–27, 2016 (Table CCA was performed based on the outdoor airborne bacterial OTU

A.1). Data from next-generation sequencing showed that each sample matrix and environmental parameters. Variables PM2.5 and CO sig- contained > 19,363 reads and had a coverage of > 0.976, indicating nificantly affected the airborne bacterial community (Fig. 2a). How- that the samples adequately represented the actual bacterial commu- ever, PM2.5, and CO could only explain 7.5% of the total variation of the nities (Table A.1). outdoor airborne bacterial community (Fig. 2b), which indicates that the outdoor airborne bacterial community was largely shaped by

4 J. Guo, et al. Environment International 139 (2020) 105721

Fig. 3. Comparison of different outdoor bacterial community types. (a) Appeared and disappeared OTU counts and their abundance. (b) Chao1, Shannon, andFaith’s PD indexes. (c) Abundance of dominant classes (mean abundance > 0.001). * p < 0.05; ** p < 0.01; *** p < 0.001 (Mann-Whitney U test). A green star indicates that the class was significantly higher in Group I than in Group II; red stars indicate that the class was significantly higher in Group II thaninGroupI. undetected parameters. PCoA analysis showed that the outdoor samples Interestingly, the assignment of G1 and GII types could well explain the were mainly clustered into two groups (Groups I and II; GI and GII) airborne bacterial community structure at a single site and at all sites (Fig. 2c). The airborne bacterial community shifted between the two analyzed (R2: 0.252–0.35) (Table A.3). The dominant classes were si- groups (Table A.2). Transformation of GI to GII occurred in 12.15a and milar in each room and outdoors (Fig. 4b & Fig. 3c). Seventeen classes 12.15p (in TC3), and transformation from GII to GI occurred in 12.25a in the indoor environment were significantly different in GI and GII(12 and 12.25p (in TP6) (Table A.2). Thus, the outdoor samples collected higher in GI and 5 higher in GII, p < 0.05) (Fig. 4b). We selected the from December 9–14 and December 26–27 were assigned to GI, and classes that significantly differed between GI and GII in the indoor outdoor samples collected from December 16–24 were assigned to GII. environment and compared them with the classes in the outdoor en- Temporal trends of abundance of dominant classes, counts of appeared/ vironment. The total abundance of consistently significantly different disappeared OTUs, abundance of total appeared/disappeared OTUs, classes was 0.46 and 0.42 in the indoor and outdoor environment, re- and bacterial diversity all showed sharp changes in TC3 and TP6 spectively (Fig. 4b & Table A.4). In addition, we observed the changing (Fig. 2d, 2e, 2f, 2g). The appeared and disappeared OTU species were trends of each class from GI to GII in the indoor environment and se- significantly higher in GI than in GII(p < 0.001), while the abundance lected the classes with the same trend compared to the outdoor en- of appeared and disappeared OTU species was significantly lower in GI vironment. The total abundance of classes with same trend amounted to than in GII (p < 0.001) (Fig. 3a), which confirmed that the shift of 0.94 and 0.91 in the indoor and outdoor environment, respectively bacterial communities in GI and GII differed. Bacterial diversity, as (Fig. 4b & Table A.4). Moreover, comparisons of the Chao1, Shannon shown in the Chao1, Shannon, and PD indexes, was significantly higher and PD indexes between time intervals showed that all three changed in in GI than in GII (Fig. 3b). Several classes were found to significantly a similar “U”-like pattern (Fig. A.2). The values of the Chao1, Shannon, differ between GI and GII, and the two dominant classes (Bacteroidia and PD indexes in the indoor environment were significantly correlated and Clostridia) showed different trends (Fig. 3c). with those of the outdoor environment (p < 0.001) (Fig. 4c, 4d, 4e). The shift in bacterial community, Adonis analysis, changing trend of classes, and bacterial diversity all showed that the indoor airborne 3.2. Shift in indoor airborne bacterial communities bacterial community tracked the outdoor environment (Fig. 4). Based on the results of outdoor sampling, the indoor samples collected from Dec. 9–14 and Dec. 26–27 were assigned to GI, and the 3.3. Relationship between the airborne bacterial community type and samples collected from Dec. 16–24 were assigned to GII. PCoA analysis respiratory diseases of all samples from four sites showed that the samples were clustered into two obvious groups, consistent with Fig. 2c (Fig. 4a). Two major airborne bacterial community types were identified in The indoor and outdoor airborne bacterial community both changed outdoor and indoor environments (Fig. 4a). Thus, bacteria with these in a single bacterial community transformation cycle (Fig. A.1). Adonis two airborne bacterial community types entered the respiratory tract of analysis of all samples with variables showed that the sample source people. We wanted to determine whether these bacterial community significantly affected the airborne bacterial community structure types affected human health in different ways. To assess this, wecol- (R2 = 0.015, p < 0.05) (Table A.3). Adonis analysis of each indoor and lected data on respiratory diseases, including pneumonia, amygdalitis, outdoor site showed that PM2.5 concentration and haze period did not and pharyngitis, from a nearby hospital (Table A.5). We selected dis- significantly affect airborne bacterial community structure (Table A.3). ease data from Dec. 9–24 covering GI and GII and deleted the data from

5 J. Guo, et al. Environment International 139 (2020) 105721

)a( )b( 307 311 315 Indoor SC TC Bacteroidia Clostridia Gammaproteobacteria Epsilonproteobacteria Bacilli I Actinobacteria Erysipelotrichia II Alphaproteobacteria Deltaproteobacteria Betaproteobacteria Coriobacteriia Fusobacteriia Gemmatimonadetes Acidimicrobiia Mollicutes Sphingobacteriia Cytophagia Flavobacteriia Negativicutes Thermoleophilia Spirochaetes

Group IIIIIIIIIIII Indoor 0.48 0.94

Outdoor 0.42 0.91

Abundance 0.1 0.2 0.3 0.4 Total abundance

( )c )d( )e(

Fig. 4. Comparison of indoor and outdoor samples. (a) Principal coordinate analysis (PCoA) of samples collected between December 9 and 27, 2016, based on

Bray–Curtis distances. Different colors represent different time intervals in terms2.5 ofPM pollution levels. Different shapes represent the sample sites: 307, 311and 315 were indoors, Y was outdoors. (b) Abundance of dominant classes in each room indoors (307, 311, 315). Classes that significantly differed between Group Iand Group II indoors are indicated by stars. * p < 0.05; ** p < 0.01; *** p < 0.001 (Mann-Whitney U test). A green star indicates that the class was significantly higher in Group I than in Group II; a red star indicates that the class was significantly higher in Group II than in Group I. Comparing the classes between the indoorsand outdoors. SC: the classes that were significantly higher/lower in Group I than in Group II in indoor or outdoor environment were selected, and the ones consistentin indoors and outdoors were marked ; TC: the classes that were consistently higher/lower in Group I than in Group II both indoors and outdoors were marked. The total abundance of classes in SC and TC indoors and outdoor are shown separately below. (c-e) Correlations between the Chao1, Shannon, and phylogenetic diversity indexes among sampling sites, shown as correlation coefficient maps. The proportion of the pie chart that is colored represents the correlation coefficient. Adarker color represents a higher correlation coefficient. * p < 0.05; ** p < 0.01; *** p < 0.001.

Fig. 5. Daily cases of pneumonia, amygdalitis, and pharyngitis from December 9 to 24, 2016, in a nearby hospital. Number of residents suffering within<1 day (a), at least 1 day (b), and at least 2 days (c) by special bacterial community type. A Mann-Whitney U test was performed. transition days (Dec. 15, 25). We also deleted the data from Dec. 26–27 number of days. We found that the number of daily pneumonia cases to avoid combining it with the disease data from Dec. 9–14 in GI, which was significantly higher in GI than in GII(p = 0.035), whereas the would result in inaccuracy owing to the complexity of human adapta- number of daily amygdalitis cases was significantly higher in GII than tion to changes in the airborne bacterial community and the limited in GI (p = 0.025) (Fig. 5a). However, there was no difference in the

6 J. Guo, et al. Environment International 139 (2020) 105721

)a( )c(

Group I Group II

Actinobacteria Alphaproteobacteria

Clostridia ssalC Cytophagia Epsilonproteobacteria Erysipelotrichia Fusobacteriia Gammaproteobacteria Mollicutes

Alistipes Allobaculum Alloprevotella Anaeroplasma Anaerotruncus Blautia Citrobacter Haemophilus

Helicobacter suneG (b) Lachnoclostridium Lachnospiraceae_NK4A136_group Lachnospiraceae_UCG_006 Lactococcus Oscillibacter Paracocccus Prevotella_1 Prevotella_9 Prevotellaceae_UCG_001 Rikenellaceae_RC9_gut_group Roseburia Sphingomonas Streptococcus oo r Indoo r Indoo r Out doo r Outd

0 0.1 0.2 0.3 0.4

Fig. 6. Characteristics of indoor samples in different outdoor airborne bacterial community types. Principal coordinate analysis (PCoA) of samples collected inthe time range of Groups I (a) and Group II (b) based on Bray–Curtis distances. Different colors represent samples with different time scales in one day. a. 20:00–8:00; p. 8:00–20:00. Different shapes represent the sampling sites: 307, 311, and 315 were indoors, Y was outdoors. (c) The significantly different abundance ofclassesand genera in indoor and outdoor environments in Groups I and II. * p < 0.05; ** p < 0.01; *** p < 0.001 (Mann-Whitney U test). Green stars indicate that the class was significantly higher in Group I than in Group II; red stars show that the class was significantly higher in Group II than in Group I. The classes/genera thatwere consistently significantly higher/lower indoors than outdoors in both Group I and Group II are marked by aredcircle.

Table 1 morbidity of pharyngitis between groups. To consider the case of de- Adonis analysis of airborne bacterial communities in indoor and outdoor en- layed onset, we compared the daily disease cases suffered one day or vironments according to different variable types in Groups I andII. two days later to the special airborne bacterial community in G1 and Indoor Y All samples GII. The daily pneumonia cases were higher in GI than in GII after one day (p = 0.061) and two days (p = 0.050), whereas the daily amyg- 2 2 2 R p R p R p dalitis cases were higher in GII than in GI after one day (p = 0.008) and two days (p = 0.017) (Fig. 5b, 5c). This showed that the morbidity of Group I Sample source – – – – 0.073 0.001*** Room 0.116 0.001*** – – – – pneumonia and amygdalitis was significantly different in GI and GII Sampling time 0.025 0.278 0.037 0.530 0.014 0.352 (Fig. 5). PM2.5(ug/m3) 0.036 0.079 0.072 0.126 0.024 0.054 Group II Sample source – – – – 0.030 0.008** Room 0.031 0.849 – – – – 3.4. Characteristics of the indoor airborne bacterial community in GI and Sampling time 0.058 0.001*** 0.072 0.207 0.040 0.001*** GII PM2.5(ug/m3) 0.017 0.652 0.061 0.318 0.014 0.489 Although the indoor airborne bacterial community tracked the Sample source: indoor, outdoor; Room: 307, 311, 315; Sampling time: a outdoor airborne bacterial community to form two obvious groups, it (20:00–8:00), p (8:00–20:00). was valuable to explore the characteristics of bacterial community composition in the indoor and outdoor environments for the same outdoor bacterial community type. PCoA analysis of G1 and G2 showed

7 J. Guo, et al. Environment International 139 (2020) 105721 that the room and sampling time affected the indoor airborne bacterial altitude from December 16–18 (Fig. 7). Thus, a strong shift in the input community differently (Fig. 6a, 6b). Adonis analysis showed that airflow direction, distance, and vertical motion between December 16 sampling source significantly affected the airborne bacterial community and 18 might explain the marked change in the bacterial community for both types (p < 0.01) (Table 1). Room significantly affected the (Fig. 7). indoor airborne bacterial community in GI (p = 0.001), and sampling Indoor airborne bacterial community structure and bacterial di- time affected the indoor airborne bacterial community inGII versity both closely tracked those found outdoors (Fig. 4), supporting (p = 0.001) (Table 1). Comparisons of the abundance of taxa in the the findings of Meadow et al.(Meadow et al. 2014). The abundance of indoor and outdoor environments in G1 and G2 showed that the sig- Clostridia was significantly higher in the indoor environment in bothGI nificantly different classes were Actinobacteria, Bacteroidia, Clostridia, and GII, making up > 20% of the bacterial community (Fig. 6c & Table Erysipelotrichia, Fusobacteria, Gammaproteobacteria, and Mollicutes A.6). However, the taxa that significantly differed between the indoor (Fig. 6c & Table A.6); the significantly different genera were Alistipes, and outdoor environment were quite different in G1 and G2(Fig. 6c & Allobaculum, Alloprevotella, Anaeroplasma, Anaerotruncus, Blautia, Ci- Table A.6 & A.7). This may be the result of a combination of different trobacter, Clostridium_sensu_stricto_1, Haemophilus, Lachnospir- bacterial compositions entering the interior from outside and re- aceae_NK4A136_group, Lachnospiraceae_UCG_006, Lactococcus, Odor- spiratory emissions from indoor residents. Since human health is in- ibacter, Oscillibacter, Parabacteroides, Prevotella_1, Prevotella_9, fluenced by both the indoor and outdoor environments, and>60% of Prevotellaceae_UCG_001, Romboutsia, Roseburia, and Streptococcus viable bioaerosols are in the respirable size range under all weather (Fig. 6c & Table A.7). Most of the taxa showed an inconsistent trend in conditions (Li et al. 2017b), it is important to consider the relationship the indoor and outdoor environments with regard to GI and GII except between the airborne bacterial community and respiratory disease for two classes and one genus (Fig. 6c). It was showed Clostridia was during haze events. Microbial patterns in the nearby air environment with significantly higher abundance in the indoor environment, and strongly affect the nasal microbiota of animals and humans in similar Epsilonproteobacteria was with significantly higher abundance in the ways (Kraemer et al. 2018). In the present study, we found that the two outdoor environment for both GI and GII (p < 0.05) (Fig. 6c & Table different bacterial community types could result in significantly dif- A.6). Helicobacter belonging to Epsilonproteobacteria was consistent ferent numbers of daily pneumonia and amygdalitis cases (Fig. 5). Ex- with Epsilonproteobacteria (Fig. 6c & Table A.7). amination of the correlation of the classes in all samples and respiratory diseases showed that Clostridia and Deltaproteobacteria were sig- 4. Discussion nificantly negatively correlated with pneumonia, Gammaproteo- bacteria was significantly positively correlated with pneumonia, and Although the abiotic pollutants that are present during haze events Clostridia was significantly positively correlated with amygdalitis have been characterized in detail (Guanet al. 2016; Huang et al. 2014; (Table 2). At the genus level, we found that ten genera were sig- Sun et al. 2013; Wang et al. 2016), we still have limited knowledge of nificantly positively correlated with pneumonia and four genera were the airborne bacterial community, with different trends reported (Chan significantly positively correlated with amygdalitis (Table 3). These et al. 2015; Donget al. 2016; Gao et al. 2015; Gao et al. 2016; Xu et al. genera could be recognized as indicators for pneumonia or amygdalitis. 2017a). A continuous dataset from nine haze events showed that the The above selected classes and genera were mostly consistent with the airborne bacterial concentration was significantly positively correlated significantly different taxa in G1 and G2 (Table A.9 &A.10). with the concentrations of the main pollutants (PM2.5, PM10, SO2, NO2, During the course of the disease, the bacterial community compo- and CO) and the relative humidity in Beijing (Li et al. 2018). Thus, the sition often shifted toward Gammaproteobacteria, which contains many influence of haze events on airborne bacteria is not negligible. Inthe common lung-associated gram-negative pathogens such as Haemophilus present study, the bacterial community structure did not change with influenzae (Burke et al. 1991), Klebsiella (Juan et al. 2020), Pseudo- the haze cycle (Fig. 1c), which was consistent with the findings of monas aeruginosa (Garau and Gomez 2003), Legionella pneumophila previous studies (Duet al. 2018; Liet al. 2018). The outdoor airborne (Newton et al. 2010), and Escherichia coli (Okimoto et al. 2010) that bacterial communities in our samples were clearly clustered into two benefit from airway inflammation, potentially leading to a cyclical in- groups (Fig. 2c). Examination of the appearance and disappearance of flammatory mechanism (Huffnagle et al. 2017). By contrast, Clostridia OTUs in these groups showed that they had significantly different be- rarely causes pleuropulmonary infections in the absence of penetrating haviors (p < 0.001) (Fig. 3a). We found that although there were sig- chest injuries (Bayer et al. 1975; Patel and Mahler 1990), suggesting nificant differences in physiochemical parameters between GI andGII, that the pathogenicity of Clostridia is lower than that of Gammapro- including PM2.5, PM10, NO2, and CO (Table A.8), these could not ex- teobacteria for pneumonia. However, there was no clear evidence re- plain the sharp change in OTUs and the quite significant shift in bac- garding the relationship between Clostridia and amygdalitis. Pneu- terial community structure that occurred on December 15, 2016 monia is an important cause of morbidity and mortality. However, its (Fig. 2c). Beijing entered the central heating period during the experi- etiology is only identified in approximately 50% of cases of community- mental period. Major sources of PM2.5/PM10 did not show an obvious acquired pneumonia and in approximately 35% of cases of nosocomial change. Near-surface meteorological conditions with stable pressure pneumonia (Thomas et al. 2006). In our study, there was no evidence of field, high humidity, and low wind speed were the primary cause ofthe a relationship between our indicated genera and pneumonia or amyg- haze event during this period (Xu et al., 2016). Therefore, these dif- dalitis. By investigating the literature, we found that more of the ferences must have been due to a sudden change in bacterial inputs. We pneumonia-indicated taxa in our study were related to gastrointestinal tracked the source of airflow that could input bacteria from another disease, nervous system disease, and coronary artery disease, including area on each sampling date. Backward trajectories over 72 h showed Alistipes (De Angelis et al. 2013; Petrov et al. 2016; Schirmer et al. that the airflow was mainly from the northwest from December 9–14 2018), Escherichia_Shigella (de Paiva et al. 2016; Shen et al. 2017; Wang (Fig. 7a). However, from December 16 on, there was an obvious input et al. 2012), Faecalibacterium (de Paivaet al. 2016; Zhu et al. 2018), from the west, with all inputs arriving from this direction by December Pseudobutyrivibrio (de Paivaet al. 2016; Kvasnovsky et al. 2018), and 18 (Fig. 7b). The input area then returned to the north on December 22 Roseburia (Petrovet al. 2016; Raman et al. 2013). Most of the indicated (Fig. 7c). We also found that the maximum distance from which air genera were gram-negative (8/14) and/or anaerobic (10/14) sug- arrived was > 2,200 km from December 9–14, indicating that bacterial gesting that these bacteria may be of high risk to human health. inputs were received from much farther away during this period than Meteorological factors are known to affect air pollution and the from December 16–24 (Fig. 7). The frequency of vertical airflow motion bacterial community (Zhenet al. 2017). Our data showed that these at > 2,500 m was also greater from December 9–14 than from De- changes in the bacterial community could also lead to human disease. cember 16–24 (3/6 VS 2/9), and the vertical motion changed to a lower The consequences of pneumonia are much more serious than those of

8 J. Guo, et al. Environment International 139 (2020) 105721

Fig. 7. National Oceanic and Atmospheric Administration (NOAA) HYSPLIT model of the backward trajectories of airflow at the study sites using GFSG meteorological data. a. December 9 to 14, 2016. b. December 14 to 19, 2016. c. December 19 to 24, 2016. In each map, each line represents the 72-hour backward trajectories of airflow for one day and lines with same color represent the same day. The upper map shows the trajectory of the airflow in a horizontal direction.An equidistant circle centered on Beijing is marked on the map. The lower map shows the trajectory of the airflow in the vertical direction. The source was in Beijing (39.93 N, 116.28E, 200 m above ground level).

Table 2 children, the elderly, and hospital patients. Correlations between respiratory diseases and taxa at the class level.

Class Pneumonia Amygdalitis 5. Conclusions

Actinobacteria 0.36 −0.26 Two outdoor airborne bacterial community types were observed Alphaproteobacteria −0.25 0.31 Bacilli 0.19 −0.43 during the central heating period from December 9–27, 2016 in Beijing. Bacteroidia 0.3 −0.06 The formation was possibly due to a strong shift in input airflow di- Betaproteobacteria −0.33 0.04 rection, distance, and vertical motion. Indoor airborne bacterial com- Clostridia −0.54 * 0.5 * munity structure and bacterial diversity both closely tracked those Deltaproteobacteria −0.6 ** 0.42 found outdoors. Importantly, the airborne bacterial community type Epsilonproteobacteria −0.41 0.25 Erysipelotrichia −0.39 0.23 was closely related to the incidence of human respiratory diseases, Gammaproteobacteria 0.52 * −0.42 particularly pneumonia. Finally, we found that the outdoor airborne bacterial community * p < 0.05; ** p < 0.01(Pearson correlation analysis) type had a strong impact on the distribution of bacteria in both indoor and outdoor environments. Clostridia was an indicated class in the in- amygdalitis. Thus, the health risk of the bacterial community was door environment with higher abundance. higher in GI. However, the air pollution in G1 was significantly lower than that in GII (Table A.8). Therefore, the finding that the bacterial CRediT authorship contribution statement community was a lower risk to human health during the time period with higher pollution levels was unexpected. Comprehensive con- Jianguo Guo: Conceptualization, Methodology, Formal analysis, sideration of the health risk of the airborne bacterial community and air Software, Writing - original draft, Funding acquisition. Yi Xiong: pollution is important to public health. Airborne bacterial concentra- Investigation, Formal analysis. Changhua Shi: Investigation, Data tions and community composition are shaped by many factors, in- curation. Ce Liu: Investigation. Hongwei Li: Formal analysis, cluding relative humidity, carbon monoxide and ozone concentrations, Visualization. Hua Qian: Software, Resources. Zongke Sun: Software. temperature, ultraviolet (UV) radiation, and PM fractions (Bowers et al. Chuan Qin: Conceptualization. 2013; Duet al. 2018; Franzetti et al. 2011; Gaoet al. 2016; Qi et al. 2014; Qian et al. 2012). The relative humidity of the atmosphere in particular is vital for the survival of airborne microbes and can shape Acknowledgments bacterial community structure (Dannemiller et al. 2017; Liet al. 2018). In addition, changes in bioaerosols during haze events are season-de- We are grateful for the sequencing service provided by Oebiotech pendent (Liet al. 2018). Thus, in the future, we could obtain more Company in Shanghai. The authors would like to thank Enago (www.enago.cn) for the English language review. samples and build models that comprise the risks of PM2.5 and the bacterial community, both of which can be predicted by the values of the abovementioned parameters, to develop more useful guidelines for Funding policymakers dealing with air pollution. Our knowledge of the relationship between airborne bacterial community structure and disease This research was supported by the National Key R&D Program of occurrence is still in its infancy, but our results provide important im- China (grant number: 2017YFC0702800), the CAMS Innovation Fund plications for health care during haze events, particularly for those for Medical Science (CIFMS: 2018-12M-I-001, 2017-I2M-2-005, 2016- members of the population with greater susceptibility, including I2M-2-006), and the Central Public-interest Scientific Institution Basal Research Fund (grant number: 2016ZX310037).

9 J. Guo, et al. Environment International 139 (2020) 105721

Competing interests. The authors declare no competing interests.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https:// doi.org/10.1016/j.envint.2020.105721. Pneumonia Amygdalitis 0.61 ** −0.29 0.46 * −0.22 −0.49 * 0.51 * −0.38 0.5 * 0.65 ** −0.43 −0.38 0.52 * 0.6 ** −0.22 0.71 *** −0.25 0.58 ** −0.54 * 0.65 ** −0.4 0.64 ** −0.21 −0.68 ** 0.48 * 0.5 * −0.15

References

Abd Aziz, A., Lee, K., Park, B., Park, H., Park, K., Choi, I.G., Chang, I.S., 2018. Comparative study of the airborne microbial communities and their functional composition in fine particulate matter (PM2.5) under non-extreme and extreme PM2.5 conditions. Atmos. Environ. 194, 82–92. Adams, R.I., Bhangar, S., Pasut, W., Arens, E.A., Taylor, J.W., Lindow, S.E., Nazaroff, Characteristics G−, anaerobic G−, anaerobic G−, aerobic G−,anaerobic G−, facultative aerobic 0.56 * −0.51 * G+, aerobic G−,aerobic Anaerobic Anaerobic Anaerobic G−, anaerobic Anaerobic G−, anaerobic W.W., Bruns, T.D. Chamber, 2015;10... Bioaerosol Study: Outdoor Air and Human Occupants as Sources of Indoor Airborne Microbes. PLoS ONE. Bayer, A.S., Nelson, S.C., Galpin, J.E., Chow, A.W., Guze, L.B., 1975. Necrotizing pneumonia and empyema due to Clostridium perfringens. Report of a case and review of the literature. The American journal of medicine 59, 851–856. Bolger, A.M., Lohse, M., Usadel, B., 2014. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics 30, 2114–2120. Bowers, R.M., Clements, N., Emerson, J.B., Wiedinmyer, C., Hannigan, M.P., Fierer, N., 2013. Seasonal variability in bacterial and fungal diversity of the near-surface atmosphere. Environ. Sci. Technol. 47, 12097–12106. Burke, W.M., Maloney, P.J., Savdie, E., 1991. PNEUMONIA CAUSED BY HAEMOPHILUS- INFLUENZAE. Med. J. Aust. 154 219 219. Burrows, S.M., Elbert, W., Lawrence, M.G., Poschl, U., 2009. Bacteria in the global atmosphere - Part 1: Review and synthesis of literature data for different ecosystems. Atmos. Chem. Phys. 9, 9263–9280. Caporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F.D., Costello, E.K., Fierer, N., Pena, A.G., Goodrich, J.K., Gordon, J.I., Huttley, G.A., Kelley, S.T., Knights, D., Koenig, J.E., Ley, R.E., Lozupone, C.A., McDonald, D., Muegge, B.D., Pirrung, M., Reeder, J., Sevinsky, J.R., Turnbaugh, P.J., Walters, W.A., Widmann, J., Yatsunenko, T., Zaneveld, J., Knight, R., 2010. QIIME allows analysis of high- throughput community sequencing data. Nat Methods 7, 335–336. Chan, W.R., Cohn, S., Sidheswaran, M., Sullivan, D.P., Fisk, W.J., 2015. Contaminant levels, source strengths, and ventilation rates in California retail stores. Indoor Air 25, 381–392. Cohen, A.J., Brauer, M., Burnett, R., Anderson, H.R., Frostad, J., Estep, K., Balakrishnan, K., Brunekreef, B., Dandona, L., Dandona, R., Feigin, V., Freedman, G., Hubbell, B., Negatively correlated diseases GD( Ishaqet al. 2018 ), NAFLD( Jiang et2017 ), al. IBD( de 2015 ), MeijetMancabelliet DS( al. al. 2018 ) – – – – GD( Ishaqet al. 2018 ), CAD( Zhuet al. 2018 ) – – – – SLE( Heet al. 2016 ), PsA( Scheret al.) 2015 CRC( Wanget al. 2012 ), IBD( Imhannet al. 2018 ), CAD( Zhuet al. 2018 ) G+, anaerobic – – Jobling, A., Kan, H., Knibbs, L., Liu, Y., Martin, R., Morawska, L., Pope III, C.A., Shin, H., Straif, K., Shaddick, G., Thomas, M., van Dingenen, R., van Donkelaar, A., Vos, T., Murray, C.J.L., Forouzanfar, M.H., 2017. Estimates and 25-year trends of the global burden of disease attributable to ambient air pollution: an analysis of data from the Global Burden of Diseases Study 2015. Lancet 389, 1907–1918. Dannemiller, K.C., Weschler, C.J., Peccia, J., 2017. Fungal and bacterial growth in floor dust at elevated relative humidity levels. Indoor Air 27, 354–363. De Angelis, M., Piccolo, M., Vannini, L., Siragusa, S., De Giacomo, A., Serrazzanetti, D.I., Cristofori, F., Guerzoni, M.E., Gobbetti, M., Francavilla, R., 2013;8... Fecal Microbiota and Metabolome of Children with Autism and Pervasive. Developmental Disorder Not Otherwise Specified. Plos One. de Meij, T.G.J., de Groot, E.F.J., Peeters, C.F.W., de Boer, N.K.H., Kneepkens, C.M.F., Eck, A., Benninga, M.A., Savelkoul, P.H.M., van Bodegraven, A.A., Budding, A.E., 2018;13... Variability of core microbiota in newly diagnosed treatment-naive pae- diatric inflammatory bowel disease patients. PLoS ONE. de Paiva, C.S., Jones, D.B., Stern, M.E., Bian, F., Moore, Q.L., Corbiere, S., Streckfus, C.F., Hutchinson, D.S., Ajami, N.J., Petrosino, J.F., Pflugfelder, S.C., 2016. Altered Mucosal Microbiome Diversity and Disease Severity in Sjogren Syndrome. Sci Rep 6, 23561. Declercq, A.M., Haesebrouck, F., Van den Broeck, W., Bossier, P., Decostere, A., 2013. Columnaris disease in fish: a review with emphasis on bacterium-host interactions. Vet. Res. 44. Dong, L., Qi, J., Shao, C., Zhong, X., Gao, D., Cao, W., Gao, J., Bai, R., Long, G., Chu, C., 2016. Concentration and size distribution of total airborne microbes in hazy and foggy weather. Sci. Total Environ. 541, 1011–1018. Du, P., Du, R., Ren, W., Lu, Z., Fu, P., 2018. Seasonal variation characteristic of inhalable microbial communities in PM2.5 in Beijing city, China. Sci. Total Environ. 610, 308–315. Positively correlated diseases IBD( Schirmeret al. 2018 ), PD( Petrovet al.2013 ) 2016 ), PDD( De Angeliset al. – IBD( Suchodolskiet al. 2010 )(dog) ASD( Zhanget al. 2018 ) PD( Liet al. 2017a; Qianet al.CRC( Wanget 2018 ), al. NAFLD( Shenet 2012 ), al.de SS( 2017 ), Paivaet al. 2016 ) SS( de Paivaet al. 2016 ), CAD( Zhuet al. 2018 ) CD( Declercqet al. 2013 )(fish) Inflammation( Loyet al. 2017 ) (mouse) ADI( Kvasnovskyet al. 2018 ), SS( de Paivaet al. ) 2016 NDP( Petrovet al. 2016 ),Ramanet NAFLD( al. 2013 ), – – Edgar, R.C., 2013. UPARSE: highly accurate OTU sequences from microbial amplicon reads. Nat Methods 10, 996–998. Faith, D.P., Baker, A.M., 2007. Phylogenetic diversity (PD) and biodiversity conservation: some bioinformatics challenges. Evolutionary bioinformatics online 2, 121–128. Fan, X.-Y., Gao, J.-F., Pan, K.-L., Li, D.-C., Dai, H.-H., Li, X., 2019. More obvious air pollution impacts on variations in bacteria than fungi and their co-occurrences with ammonia-oxidizing microorganisms in PM2.5. Environ. Pollut. 251, 668–680. NK4A136 group – UCG-001 – Franzetti, A., Gandolfi, I., Gaspari, E., Ambrosini, R., Bestetti, G., 2011. Seasonal variability of bacteria in fine and coarse urban air particulate matter. Appl. Microbiol. Biotechnol. 90, 745–753. Gao, M., Qiu, T., Jia, R., Han, M., Song, Y., Wang, X., 2015. Concentration and size distribution of viable bioaerosols during non-haze and haze days in Beijing. Environ. Sci. Pollut. Res. Int. 22, 4359–4368. Genus Alistipes Anaeroplasma Brevundimonas Butyricimonas Escherichia_Shigella Faecalibacterium Flavobacterium Lachnospiraceae Lachnospiraceae Mucispirillum Pseudobutyrivibrio Roseburia Ruminiclostridium_9 Subdoligranulum

Table 3 Genus-level taxa that were significantly positively correlated with respiratory diseases. IBD: inflammatory bowel disease;human PD: Parkinson’s colorectal cancer; disease; SS: Sjogren’s PDD: pervasivearthritis; syndrome; developmental CAD: “-“: coronary none arterydisorder; reported. disease; Pearson CD:GD: Graves’ columnaris correlation disease; disease; analysis NAFLD: ADI: was acute non-alcoholic performed. diverticulitis; NDP: fatty neurological liver diseases except disease; PD; SLE: DS: systemic gut lupus disease; erythematosus; ASD: autism PsA: spectrum disorder; CRC: psoriatic Gao, M., Yan, X., Qiu, T., Han, M., Wang, X., 2016. Variation of correlations between

10 J. Guo, et al. Environment International 139 (2020) 105721

factors and culturable airborne bacteria and fungi. Atmos. Environ. 128, 10–19. mSystems 2 (1). https://doi.org/10.1128/mSystems.00171-16. Garau, J., Gomez, L., 2003. Pseudomonas aeruginosa pneumonia. Current Opinion in Mancabelli, L., Milani, C., Lugli, G.A., Turroni, F., Cocconi, D., van Sinderen, D., Ventura, Infectious Diseases 16, 135–143. M., 2017. Identification of universal gut microbial biomarkers of common human Griffin, D.W., 2007. Atmospheric Movement of Microorganisms in Clouds of Desert Dust intestinal diseases by meta-analysis. FEMS Microbiol. Ecol. 93. and Implications for Human Health. Clin. Microbiol. Rev. 20, 459–477. Meadow, J.F., Altrichter, A.E., Kembel, S.W., Kline, J., Mhuireach, G., Moriyama, M., Guan, W.J., Zheng, X.Y., Chung, K.F., Zhong, N.S., 2016. Impact of air pollution on the Northcutt, D., O'Connor, T.K., Womack, A.M., Brown, G.Z., Green, J.L., Bohannan, burden of chronic respiratory diseases in China: time for urgent action. Lancet 388, B.J.M., 2014. Indoor airborne bacterial communities are influenced by ventilation, 1939–1951. occupancy, and outdoor air source. Indoor Air 24, 41–48. Guo, X.R., Zhao, L.J., Chen, D.S., Jia, Y.H., Zhao, N., Liu, W.W., Cheng, S.Y., 2018. Air Newton, H.J., Ang, D.K.Y., van Driel, I.R., Hartland, E.L., 2010. Molecular Pathogenesis of quality improvement and health benefit of PM2.5 reduction from the coal cap policy Infections Caused by Legionella pneumophila. Clin. Microbiol. Rev. 23, 274–298. in the Beijing-Tianjin-Hebei (BTH) region. China. Environmental Science and Okimoto, N., Hayashi, T., Ishiga, M., Nanba, F., Kishimoto, M., Yagi, S., Kurihara, T., Pollution Research 25, 32709–32720. Asaoka, N., Tamada, S., 2010. Clinical features of Escherichia coli pneumonia. He, Z., Shao, T., Li, H., Xie, Z., Wen, C., 2016. Alterations of the gut microbiome in Journal of Infection and Chemotherapy 16, 216–218. Chinese patients with systemic lupus erythematosus. Gut pathogens 8, 64. Patel, S.B., Mahler, R., 1990. Clostridial pleuropulmonary infections: case report and Hoppe, P., Martinac, I., 1998. Indoor climate and air quality - Review of current and review of the literature. The Journal of infection 21, 81–85. future topics in the field of ISB study group 10. Int. J. Biometeorol. 42,1–7. Petrov, V.A., Alifirova, V.M., Saltykova, I.V., Zhukova, I.A., Zhukova, N.G., Dorofeeva, Hospodsky, D., Qian, J., Nazaroff, W.W., Yamamoto, N., Bibby, K., Rismani-Yazdi, H., Y.B., Tyakht, A.V., Altukhov, I.A., Kostryukova, E.S., Titova, M.A., Mironova, Y.S., Peccia, J., 2012;7... Human Occupancy as a Source of Indoor Airborne Bacteria. PLoS Izhboldina, O.P., Nikitina, M.A., Perevozchikova, T.V., Fait, E.A., Sazonov, A.E., ONE. 2016. Comparison study of gut microbiota in case of Parkinson’s disease and other Huang, R.-J., Zhang, Y., Bozzetti, C., Ho, K.-F., Cao, J.-J., Han, Y., Daellenbach, K.R., neurological disorders. Bulletin of Siberian Medicine 15, 113–125. Slowik, J.G., Platt, S.M., Canonaco, F., Zotter, P., Wolf, R., Pieber, S.M., Bruns, E.A., Pope 3rd, C.A., Burnett, R.T., Krewski, D., Jerrett, M., Shi, Y., Calle, E.E., Thun, M.J., Crippa, M., Ciarelli, G., Piazzalunga, A., Schwikowski, M., Abbaszade, G., Schnelle- 2009. Cardiovascular mortality and exposure to airborne fine particulate matter and Kreis, J., Zimmermann, R., An, Z., Szidat, S., Baltensperger, U., El Haddad, I., Prevot, cigarette smoke: shape of the exposure-response relationship. Circulation 120, A.S.H., 2014. High secondary aerosol contribution to particulate pollution during 941–948. haze events in China. Nature 514, 218–222. Pope 3rd, C.A., Burnett, R.T., Thurston, G.D., Thun, M.J., Calle, E.E., Krewski, D., Huffnagle, G.B., Dickson, R.P., Lukacs, N.W., 2017. The respiratory tract microbiome and Godleski, J.J., 2004. Cardiovascular mortality and long-term exposure to particulate lung inflammation: a two-way street. Mucosal Immunol. 10, 299–306. air pollution: epidemiological evidence of general pathophysiological pathways of Imhann, F., Vich Vila, A., Bonder, M.J., Fu, J., Gevers, D., Visschedijk, M.C., Spekhorst, disease. Circulation 109, 71–77. L.M., Alberts, R., Franke, L., van Dullemen, H.M., Ter Steege, R.W.F., Huttenhower, Pope 3rd, C.A., Burnett, R.T., Turner, M.C., Cohen, A., Krewski, D., Jerrett, M., Gapstur, C., Dijkstra, G., Xavier, R.J., Festen, E.A.M., Wijmenga, C., Zhernakova, A., Weersma, S.M., Thun, M.J., 2011. Lung cancer and cardiovascular disease mortality associated R.K., 2018. Interplay of host genetics and gut microbiota underlying the onset and with ambient air pollution and cigarette smoke: shape of the exposure-response re- clinical presentation of inflammatory bowel disease. Gut 67, 108–119. lationships. Environ. Health Perspect. 119, 1616–1621. Ishaq, H.M., Mohammad, I.S., Shahzad, M., Ma, C., Raza, M.A., Wu, X., Guo, H., Shi, P., Qi, J., Shao, Q., Xu, W., Gao, D., Jin, C., 2014. Seasonal distribution of bioaerosols in the Xu, J., 2018. Molecular Alteration Analysis of Human Gut Microbial Composition in coastal region of Qingdao. Journal of Ocean University of China 13, 57–65. Graves' disease Patients. International journal of biological sciences 14, 1558–1570. Qian, J., Hospodsky, D., Yamamoto, N., Nazaroff, W.W., Peccia, J., 2012. Size-resolved Jiang, W.W., Wu, N., Wang, X.M., Chi, Y.J., Zhang, Y.Y., Qiu, X.Y., Hu, Y., Li, J., Liu, Y.L., emission rates of airborne bacteria and fungi in an occupied classroom. Indoor Air 22, 2015;5... Dysbiosis gut microbiota associated with inflammation and impaired mu- 339–351. cosal immune function in intestine of humans with non-alcoholic fatty liver disease. Qian, Y., Yang, X., Xu, S., Wu, C., Song, Y., Qin, N., Chen, S.D., Xiao, Q., 2018. Alteration Sci. Rep. of the fecal microbiota in Chinese patients with Parkinson's disease. Brain Behav. Juan, C.H., Fang, S.Y., Chou, C.H., Tsai, T.Y., Lin, Y.T., 2020;9... Clinical characteristics of Immun. 70, 194–202. patients with pneumonia caused by Klebsiella pneumoniae in Taiwan and prevalence Raman, M., Ahmed, I., Gillevet, P.M., Probert, C.S., Ratcliffe, N.M., Smith, S., Greenwood, of antimicrobial-resistant and hypervirulent strains: a retrospective study. R., Sikaroodi, M., Lam, V., Crotty, P., Bailey, J., Myers, R.P., Rioux, K.P., 2013. Fecal Antimicrobial Resistance and Infection. Control. microbiome and volatile organic compound metabolome in obese humans with Kembel, S.W., Jones, E., Kline, J., Northcutt, D., Stenson, J., Womack, A.M., Bohannan, nonalcoholic fatty liver disease. Clinical gastroenterology and hepatology : the offi- B.J., Brown, G.Z., Green, J.L., 2012. Architectural design influences the diversity and cial clinical practice journal of the American Gastroenterological Association 11 structure of the built environment microbiome. The ISME journal 6, 1469–1479. (868–875), e861–e863. Klepeis, N.E., Nelson, W.C., Ott, W.R., Robinson, J.P., Tsang, A.M., Switzer, P., Behar, Reyon, D., Tsai, S.Q., Khayter, C., Foden, J.A., Sander, J.D., Joung, J.K., 2012. FLASH J.V., Hern, S.C., Engelmann, W.H., 2001. The National Human Activity Pattern assembly of TALENs for high-throughput genome editing. Nat Biotechnol 30, Survey (NHAPS): a resource for assessing exposure to environmental pollutants. J. 460–465. Expo. Anal. Environ. Epidemiol. 11, 231–252. Scher, J.U., Ubeda, C., Artacho, A., Attur, M., Isaac, S., Reddy, S.M., Marmon, S., Kraemer, J.G., Ramette, A., Aebi, S., Oppliger, A., 2018. Hilty M. Key Role of Airborne Neimann, A., Brusca, S., Patel, T., Manasson, J., Pamer, E.G., Littman, D.R., Microbial Communities. Applied and environmental microbiology, Influence of Pig Abramson, S.B., 2015. Decreased bacterial diversity characterizes the altered gut Farming on the Human Nasal Microbiota, pp. 84. microbiota in patients with psoriatic arthritis, resembling dysbiosis in inflammatory Kvasnovsky, C.L., Leong, L.E.X., Choo, J.M., Abell, G.C.J., Papagrigoriadis, S., Bruce, K.D., bowel disease. Arthritis & rheumatology 67, 128–139. Rogers, G.B., 2018. Clinical and symptom scores are significantly correlated with Schirmer, M., Franzosa, E.A., Lloyd-Price, J., McIver, L.J., Schwager, R., Poon, T.W., fecal microbiota features in patients with symptomatic uncomplicated diverticular Ananthakrishnan, A.N., Andrews, E., Barron, G., Lake, K., Prasad, M., Sauk, J., disease: a pilot study. Eur. J. Gastroenterol. Hepatol. 30, 107–112. Stevens, B., Wilson, R.G., Braun, J., Denson, L.A., Kugathasan, S., McGovern, D.P.B., Lepeule, J., Laden, F., Dockery, D., Schwartz, J., 2012. Chronic exposure to fine particles Vlamakis, H., Xavier, R.J., Huttenhower, C., 2018. Dynamics of metatranscription in and mortality: an extended follow-up of the Harvard Six Cities study from 1974 to the inflammatory bowel disease gut microbiome. Nat. Microbiol. 3, 337–346. 2009. Environ. Health Perspect. 120, 965–970. Shen, F., Zheng, R.D., Sun, X.Q., Ding, W.J., Wang, X.Y., Fan, J.G., 2017. Gut microbiota Leung, M.H., Lee, P.K., 2016. The roles of the outdoors and occupants in contributing to a dysbiosis in patients with non-alcoholic fatty liver disease. Hepatobiliary & pan- potential pan-microbiome of the built environment: a review. Microbiome 4, 21. creatic diseases international : HBPD INT 16, 375–381. Li, W., Wu, X.L., Hu, X., Wang, T., Liang, S., Duan, Y.F., Jin, F., Qin, B., 2017a. Structural Stephens, B., Adams, R.I., Bhangar, S., Bibby, K., Waring, M.S., 2015. From commens- changes of gut microbiota in Parkinson's disease and its correlation with clinical alism to mutualism: integrating the microbial ecology, building science, and indoor features. Science China-Life Sciences 60, 1223–1233. air communities to advance research on the indoor microbiome. Indoor Air 25, 1–3. Li, W., Yang, J., Zhang, D., Li, B., Wang, E., Yuan, H., 2018. Concentration and Suchodolski, J.S., Xenoulis, P.G., Paddock, C.G., Steiner, J.M., Jergens, A.E., 2010. Community of Airborne Bacteria in Response to Cyclical Haze Events During the Fall Molecular analysis of the bacterial microbiota in duodenal biopsies from dogs with and Midwinter in Beijing. China. Frontiers in microbiology 9, 1741. idiopathic inflammatory bowel disease. Vet. Microbiol. 142, 394–400. Li, Y., Lu, R., Li, W., Xie, Z., Song, Y., 2017b. Concentrations and size distributions of Sun, Y., Wang, Z., Fu, P., Jiang, Q., Yang, T., Li, J., Ge, X., 2013. The impact of relative viable bioaerosols under various weather conditions in a typical semi-arid city of humidity on aerosol composition and evolution processes during wintertime in Northwest China. J. Aerosol Sci. 106, 83–92. Beijing. China. Atmospheric Environment 77, 927–934. Lin, G., Fu, J.Y., Jiang, D., Hu, W.S., Dong, D.L., Huang, Y.H., Zhao, M.D., 2014. Spatio- Sun, Y., Xu, S., Zheng, D., Li, J., Tian, H., Wang, Y., 2018. Effects of haze pollution on Temporal Variation of PM2.5 Concentrations and Their Relationship with Geographic microbial community changes and correlation with chemical components in atmo- and Socioeconomic Factors in China. Int. J. Environ. Res. Public Health 11, 173–186. spheric particulate matter. The Science of the total environment 637–638, 507–516. Lin, H., Guo, Y., Zheng, Y., Zhao, X., Cao, Z., Rigdon, S.E., Xian, H., Li, X., Liu, T., Xiao, J., Thomas, V., Herrera-Rimann, K., Blanc, D.S., Greub, G., 2006. Biodiversity of amoebae 2017. Zeng W.; Weaver N.L.; Qian Z.; Ma W.; Wu F. Exposure to ambient PM2.5 and amoeba-resisting bacteria in a hospital water network. Appl. Environ. Microbiol. associated with overall and domain-specific disability among adults in six low- and 72, 2428–2438. middle-income countries. Environ. Int. 104, 69–75. Wang, G., Zhang, R., Gomez, M.E., Yang, L., Zamora, M.L., Hu, M., Lin, Y., Peng, J., Guo, Liu, J., Han, Y., Tang, X., Zhu, J., Zhu, T., 2016. Estimating adult mortality attributable to S., Meng, J., Li, J., Cheng, C., Hu, T., Ren, Y., Wang, Y., Gao, J., Cao, J., An, Z., Zhou, PM2.5 exposure in China with assimilated PM2.5 concentrations based on a ground W., Li, G., Wang, J., Tian, P., Marrero-Ortiz, W., Secrest, J., Du, Z., Zheng, J., Shang, monitoring network. Sci. Total Environ. 568, 1253–1262. D., Zeng, L., Shao, M., Wang, W., Huang, Y., Wang, Y., Zhu, Y., Li, Y., Hu, J., Pan, B., Loy, Alexander, Pfann, Carina, Steinberger, Michaela, Hanson, Buck, Herp, Simone, Cai, L., Cheng, Y., Ji, Y., Zhang, F., Rosenfeld, D., Liss, P.S., Duce, R.A., Kolb, C.E., Brugiroux, Sandrine, Gomes Neto, João Carlos, Boekschoten, Mark V., Schwab, Molina, M.J., 2016. Persistent sulfate formation from London Fog to Chinese haze. Clarissa, Urich, Tim, Ramer-Tait, Amanda E., Rattei, Thomas, Stecher, Bärbel, Berry, PNAS 113, 13630–13635. David, Lozupone, Catherine, 2017. Lifestyle and Horizontal Gene Transfer-Mediated Wang, Q., Garrity, G.M., Tiedje, J.M., Cole, J.R., 2007. Naive Bayesian classifier for rapid Evolution of Mucispirillum schaedleri , a Core Member of the Murine Gut Microbiota. assignment of rRNA sequences into the new bacterial taxonomy. Appl. Environ.

11 J. Guo, et al. Environment International 139 (2020) 105721

Microbiol. 73, 5261–5267. L.Y., 2018. Structural Variation in the Bacterial Community Associated with Airborne Wang, T., Cai, G., Qiu, Y., Fei, N., Zhang, M., Pang, X., Jia, W., Cai, S., Zhao, L., 2012. Particulate Matter in Beijing, China, during Hazy and Nonhazy Days. Appl. Environ. Structural segregation of gut microbiota between colorectal cancer patients and Microbiol. 84. healthy volunteers. The ISME journal 6, 320–329. Zhai, Y., Li, X., Wang, T., Wang, B., Li, C., Zeng, G., 2018. A review on airborne micro- Wei, K., Zou, Z., Zheng, Y., Li, J., Shen, F., Wu, C.-Y., Wu, Y., Hu, M., Yao, M., 2016. organisms in particulate matters: Composition, characteristics and influence factors. Ambient bioaerosol particle dynamics observed during haze and sunny days in Environ. Int. 113, 74–90. Beijing. Sci. Total Environ. 550, 751–759. Zhang, M.X., Ma, W., Zhang, J., He, Y., Wang, J., 2018. Analysis of gut microbiota profiles Xu, C., Wei, M., Chen, J., Wang, X., Zhu, C., Li, J., Zheng, L., Sui, G., Li, W., Wang, W., and microbe-disease associations in children with autism spectrum disorders in Zhang, Q., Mellouki, A., 2017a. Bacterial characterization in ambient submicron China. Sci. Rep. particles during severe haze episodes at Ji'nan. China. Sci Total Environ 580, Zhang, Y.-L., Cao, F., 2015. Fine particulate matter (PM2.5) in China at a city level. Sci. 188–196. Rep. Xu, C., Wei, M., Chen, J., Wang, X., Zhu, C., Li, J., Zheng, L., Sui, G., Li, W., Wang, W., Zhao, J.H., Wang, X.Z., Song, H.Q., Du, Y., Cui, W.J., Zhou, Y.C., 2019. Spatiotemporal Zhang, Q., Mellouki, A., 2017b. Bacterial characterization in ambient submicron Trend Analysis of PM2.5 Concentration in China 1999–2016. Atmosphere. particles during severe haze episodes at Ji’nan. China. Science of The Total Zhen, Q., Deng, Y., Wang, Y., Wang, X., Zhang, H., Sun, X., Ouyang, Z., 2017. Environment 580, 188–196. Meteorological factors had more impact on airborne bacterial communities than air Xu, R.; Rao, X.Q.; Zhang, H.D. Analysis on the weather characteristics of continuous pollutants. Sci. Total Environ. 601, 703–712. heavy pollution in Beijing in December 2016. 2017 science and technology annual Zhu, Q., Gao, R., Zhang, Y., Pan, D., Zhu, Y., Zhang, X., Yang, R., Jiang, R., Xu, Y., Qin, H., meeting of the Chinese society for environmental science. Xiamen, fujian, China; 2018. Dysbiosis signatures of gut microbiota in coronary artery disease. Physiol. 2017c. Genomics 50, 893–903. Yan, D., Zhang, T., Su, J., Zhao, L.L., Wang, H., Fang, X.M., Zhang, Y.Q., Liu, H.Y., Yu,