Environmental Pollution 233 (2018) 483e493

Contents lists available at ScienceDirect

Environmental Pollution
j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m/ l o c a t e / e n v p o l

Effect of air pollution on the total bacteria and pathogenic bacteria in different
sizes of particulate matter*
Huan Liu a, Xu Zhang a, Hao Zhang a, Xiangwu Yao a, Meng Zhou a, Jiaqi Wang a, Zhanfei He a,
Huihui Zhang a, Liping Lou a, Weihua Mao c, Ping Zheng a, Baolan Hu a, b, *
a Department of Environmental Engineering, Zhejiang University, Hangzhou, 310058, China
b
Research Center for Air Pollution and Health, Zhejiang University, Hangzhou, 310058, China
c
The Center of Analysis and Measurement, Zhejiang University, Hangzhou, 310058, China

article info abstract

Article history: In recent years, air pollution events have occurred frequently in China during the winter. Most studies have focused on the
Received 5 March 2017 physical and chemical composition of polluted air. Some studies have examined the bacterial bioaerosols both indoors and
Received in revised form outdoors. But few studies have focused on the relationship be-tween air pollution and bacteria, especially pathogenic bacteria.
16 October 2017
Airborne PM samples with different diameters and different air quality index values were collected in Hangzhou, China from
Accepted 17 October 2017
December 2014 to January 2015. High-throughput sequencing of 16S rRNA was used to categorize the airborne bacteria.
Based on the NCBI database, the “Human Pathogen Database” was established, which is related to human health. Among all
the PM samples, the diversity and concentration of total bacteria were lowest in the moderately or heavily polluted air.
Keywords:
However, in the PM2.5 and PM10 samples, the relative abundances of pathogenic bacteria were highest in the heavily and
Pathogenic bacteria
Total bacteria moderately polluted air respectively. Considering the PM samples with different particle sizes, the diversities of total bacteria
Airborne particulate matter Air and the proportion of pathogenic bacteria in the PM10 samples were different from those in the PM2.5 and TSP samples. The
quality index composition of PM samples with different sizes range may be responsible for the variances. The relative humidity, carbon
monoxide and ozone concentrations were the main factors, which affected the diversity of total bacteria and the proportion of
pathogenic bacteria. Among the different environmental samples, the compositions of the total bacteria were very similar in all
the airborne PM samples, but different from those in the water, surface soil, and ground dust samples. Which may be
attributed to that the long-distance transport of the airflow may influence the composition of the airborne bacteria. This study
of the pathogenic bacteria in airborne PM samples can provide a reference for environmental and public health researchers.

© 2017 Published by Elsevier Ltd.

1. Introduction
Outdoor air pollution causes 3.3 million premature deaths per year
worldwide, foremost in Asia (Lelieveld et al., 2015). Among the various
factors of polluted air, particulate matter (PM) plays an important role in
posing a risk to human health (Chang et al., 2009). Total suspended
particulate (TSP) attached on the skin and clog pores. PM10 (particulate
matter with aerodynamic diameter
*
This paper has been recommended for acceptance by Eddy Y. Zeng.
* Corresponding author. Department of Environmental Engineering, Zhejiang University,
Hangzhou, 310058, China.
E-mail address: blhu@zju.edu.cn (B. Hu).
10 mm) inhaled via the upper respiratory tract and induce cyto-toxicity and
inflammation (Happo et al., 2014). PM2.5 (particulate matter with
aerodynamic diameter 2.5 mm) can easily penetrate to the lungs and
bloodstream tissues, causing respiratory and vascular diseases (Lelieveld et
al., 2015).
In addition to the physical and chemical components of airborne PM, the
bacteriological component also exist in the PM (Darwin, 1846). Some
researchers have reported that bacteria can influence public health and
ecological systems. The effects of the long-distance airborne bacterial
transmission can extend to entire eco-systems; bacteria in dust from the
Sahara Desert affected the Spanish alpine lake (Barberan et al., 2014) and
Caribbean coral reefs (Prospero et al., 2005). Pathogenic bacteria in air can
induce several
human diseases (Balloy and Chignard, 2009; Cerdeno~-Tarragaet al.,

https://doi.org/10.1016/j.envpol.2017.10.070 0269-
7491/© 2017 Published by Elsevier Ltd.
484 H. Liu et al. / Environmental Pollution 233 (2018) 483e493
2003). Legionella pneumophilacan cause severe pneumonia espe-cially
Legionnaires’ disease. Legionella can be transmitted and inhaled with the
contaminated aerosols (Allegra et al., 2016; Daumas et al., 2012; Yamaguchi
et al., 2017). Mycobacterium tuberculosisis responsible for tuberculosis,
which can easily spread in the air (Eames et al., 2009). Staphylococcus aureus
causes ab-scesses (Masalha et al., 2001) and may attach to skin in an aerosol
form (Eames et al., 2009). Clostridium difficile is responsible for
inflammation and is thought to be transmitted in the air (Roberts et al., 2008).
For the outdoor airborne bacteria, some researchers focused on their
concentration, diversity, sources and environmental impact factors (Alghamdi
et al., 2014; Barberan et al., 2015; Gonzalez et al., 2000; Hara and Zhang,
2012). On dusty days the concentrations of total bacterial cell were higher
than those on non-dusty days (Hara and Zhang, 2012). On hazy days, the
concentrations of detected bioaerosol were more than 6-fold higher than those
on sunny days (Wei et al., 2016). Considering PM of different diameters, the
di-versity of total bacteria was higher in PM10 sample than in PM2.5 sample
(Alghamdi et al., 2014). Various potential sources, such as water, soil, plants,
animals, and human activities may affect the distribution of airborne bacteria
(Gonzalez et al., 2000; Gangamma et al., 2011). Overall, the bacteria in
outdoor air samples are pre-dominantly derived from soil and plants
(Barberan et al., 2015; Gandolfi et al., 2015). Factors such as the soil pH,
mean annual precipitation, net primary productivity, and mean annual temper-
ature are responsible for the variability observed in bacterial composition
(Barberan et al., 2015).
Although several allergens and pathogens have been observed in previous
reports (Cao et al., 2014; Woo et al., 2013), it is not clear about the properties
of the pathogenic bacteria in the air. In this study, we have collected airborne
PM samples with various particle sizes and different AQI in Hangzhou (Table
S1) from December 2014 to January 2015. In addition, a pathogen library
named the “Human Pathogen Database” was established. We hypothesized
that com-positions of total bacteria and pathogenic bacteria were influenced
by the polluted air, PM sizes and environmental factors. Based on this
hypothesis, the diversity of the total airborne bacteria and the relative
abundance of pathogenic bacteria were studied in different PM sizes and at
different air quality levels. In addition, we also hypothesized that the potential
source environment would affect the structure of the total bacteria
community. To test the hypoth-esis, the adjacent surface soil, ground dust,
and water samples were collected as the potential source environmental
samples.
2. Materials and methods
2.1. Sample collection
Airborne PM samples were collected from the rooftop of a building in
Hangzhou, China (30 160 N, 120 070 E, ~10 m above the ground). The
sampling site is located approximately 10 m away from the nearby rivulet,
and is approximately 15 m in distance from Zijinhua Road. PM was collected
using a high-volume air sampler (Laoshan Application, Qingdao, China) with
3 1
a flow of 1.05 m min . The sampling filter membrane (Whatman, UK) was
an ordinary rectangular glass fiber membrane with an effective size of 230
180 mm, which underwent high-temperature sterilization at 100 C in a muffle
furnace for 12 h. From December 2014 to January 2015, the PM samples
were collected on different days by the same sampler. According to the air
quality levels and particulate sizes, 11 airborne PM samples were selected.
Every air sample was collected for 12 h every day from 8 a.m. to 8 p.m. After
sampling the mem-branes were stored at 20 C.
Hourly air pollutants data were collected from local air quality
monitoring station (Table S1), including PM2.5, PM10, carbon monoxide
(CO), nitrogen dioxide (NO2), ozone (O3) and sulfur di-oxide (SO2)
concentrations. In addition, the corresponding hourly meteorological data
were also collected from the local meteoro-logical monitoring station (Table
S1), including air temperature (T), relative humidity (RH), and wind scale
(WS). According to the Technical Regulation on Ambient Air Quality Index
(HJ 663e2012, China), the air quality were divided into six indexes, which
were described as good, moderate, lightly polluted, moderately polluted,
heavily polluted and severely polluted air. Except severely polluted air
quality, the samples were collected with five different air quality indexes. The
number of sample's description represented the cor-responding air quality
index. The PM2.5, PM10 and TSP were described as A, B and C, respectively
(Table S1).
Samples with different PM sizes and various air quality levels were
categorized to analyze the total and pathogenic bacteria community
compositions (Table S1). Four PM2.5 samples (A2, A3, A4, and A5) were
categorized from moderate to heavily polluted air quality levels. Four PM10
samples (B1, B2, B3, and B4) were collected from good to moderately
polluted air. Three TSP samples (C2, C3, and C4) were selected, ranging
from moderate to moder-ately polluted air quality levels. At the same level of
air quality, different PM sizes may affect the total bacterial and pathogenic
bacteria community structures. Furthermore, two water samples (E1 and E2),
two surface soil samples (E3 and E4) and two ground dust samples (E5 and
E6) were collected as surrounding environ-mental samples within 10 m of the
airborne PM sampling site.
2.2. DNA extraction and PCR amplification
DNA was extracted from the airborne PM samples using a Power Soil
DNA extraction kit (MO BIO Laboratories, Carlsbad, CA, USA). Firstly, the
filter was divided into 8 equal parts. Every part con-tained both the center and
the edges of a filter. Secondly, 1/8th of a filter with PM2.5, PM10 or TSP
were cut into pieces and loaded into the bead tubes. Then, the tubes were
heated to 65 C for 10 min, followed by shaking for 20 min. The remaining
steps were per-formed according to the manufacturer's instructions. The soil
and dust samples were extracted directly using this kit. The water samples
were extracted by Power Water DNA extraction kit (MO BIO Laboratories,
Carlsbad, CA, USA). The nucleic acid concentration sensor NanoDrop 2000
(Thermo Fisher Scientific, Merinton, USA) was used to check DNA quality.
Additionally, the DNA concentration was detected by Qubit Fluorometer
(Thermo Fisher Scientific, Merinton, USA). To quantify the total bacteria,
quantitative PCR (qPCR) based on SYBR Green was used. Forward primer
515F and reverse primer 806R of total bacteria were used. The plasmid DNA
was 10-fold diluted to construct the standard curve. Then, ac-cording to the
threshold cycle values of the standard curve, the concentrations of bacterial
cells were calculated. The detailed qPCR protocol was described in
previously reported work (Ren et al., 2015). For sequence analyses, 16S
rRNA genes were amplified us-ing the primer pair 515-F (50-GT GCC AGC
MGC CGC GG-30) and 907-R (50-AGA CAT GGT GCC AGC MGC CGC
GG-30). A AxyPrep DNA Gel Extraction Kit (Axygen, Silicon Valley, USA)
was used to purify the PCR amplification products. All purified amplified
DNA samples were stored at 20 C for downstream analysis (Hu et al., 2011).
High-throughput sequencing of 16S rRNA was conducted using the Ion
Torrent sequencing platform. Blank filters and DNase/RNase-Free water were
analyzed with the same procedure as the nega-tive controls. After extraction
and amplification, the DNA concen-trations of the two kinds of negative
controls were below the detection limit.
 H. Liu et al. / Environmental Pollution 233 (2018) 483e493 485

2.3. Sequence processing

QIIME package was used to handle the original sequence. The detailed
steps were based on the published methods (Caporaso et al., 2010; Liu et al.,
2013). After obtaining clean reads, Usearch software was used to remove the
chimeras and cluster data. Operational taxonomic units (OTUs) were obtained
through stan-dard clusters using 97% similarity. Next reads were randomly
extracted and distributed among the corresponding OTUs in each sample.
Pumping parameters could be reasonably obtained by the QIIME software
according to the alpha diversity indices dilution curve. A representative
sequence was chosen for each phylotype using a BLAST analysis against the
SILVA database. According to the annotation of taxa, the relative abundance
of total bacteria in each sample was categorized to different classification
levels (Kingdom, Phylum, Class, Order, Family, and Genera). The similarity
and vari-ability of total bacteria and pathogenic bacteria were compared
among different samples (Quast et al., 2013; Wang et al., 2007).

2.4. Statistical analyses

The relationship between air quality index (AQI) and the di-versity of
total bacteria or the proportion of pathogenic bacteria was assessed by Origin
(V8.0). The linear regression analyses were conducted between
environmental factors and bacterial composi-tion. To perform the redundancy
analysis (RDA) of environmental factors and the structure of total bacteria or Fig. 1. The top five abundant phyla of total bacteria in all the airborne PM samples.
pathogenic bacteria, CANOCO (V4.5) was used. Among different particle-
size PM sam-ples, the composition of total bacteria and pathogenic bacteria
were carried out in the R (V3.3.2) using gplots package. The corre-sponding most common classes, with average relative abundance of 17.2%, 15.6%,
significant differences were tested by ANOVA using SPSS (V20). The 9.7%, 9.2% and 8.3%, respectively. At the genus level, Thio-bacillus,
coordinate analyses (PCoA) of bacteria community similarity in different Methylobacterium, Rubellimicrobium and Paracoccus were the most
environments were mapped using QIIME package. abundant genera, with average relative abundances of 2.6%, 2.5%, 1.8% and
1.7%, respectively. All of these genera belong to the phylum Proteobacteria.

2.5. Establishment of the pathogen library and correlation analysis
To establish a database of pathogens related to human health, 74 pathogen
species were selected from the NCBI database (Bibby et al., 2010; Guo and
Zhang, 2012; Quast et al., 2013; Ye and Zhang, 2011). The sequences
corresponding to these pathogens were extracted from NCBI and imported
into the new local database named the “Human Pathogen Database”. Based on
the BLAST (2.2.26 version) software, the available sequences ofpathogenic
bacteria in the air PM samples were analyzed against the “Human Pathogen
Database”. A value of 98% similarity and 2% mismatch degrees were
allowed. A matching length of 300 bp was set as a screening parameter to
output the detected pathogenic bacteria sequence. The number of obtained
reads of pathogenic sequences was assigned to each sample.
The sequencing data of the pathogenic bacteria were screened against the
“Human Pathogen Database”. Overall, 23 pathogenic bacteria species
belonging to 16 genera were detected in all the airborne PM samples (Table
1). Staphylococcus, Bacillus, Clostridium, Enterobacter and Klebsiella were
the dominant pathogenic bacteria genera in all the airborne PM samples
(Table S4). Staphylococcus saprophyticus, Bacillus anthracis, Clostridium
perfrigens, Enterobacter aerogenes and Staphylococcus haemoluticus held the
highest pro-portion compared with the other pathogenic species (Table S5).
Staphylococcus saprophyticus with the average relative abundance of 1.98%
usually causes uncomplicated urinary tract infection (Kim et al., 2009).
Bacillus anthracis, a major cause of anthrax (Spencer,
Table 1
Genus and species of pathogenic bacteria detected in all the airborne PM samples.

Genus Pathogenic species

3. Results Bacillus anthracis
Burkholderia cepacia
3.1. The relative abundance of total bacteria and pathogenic bacteria Clostridium botulinum, baratii, perfringens, tetani
in the airborne PM samples Corynebacterium diphtheriae
Enterobacter aerogenes
4 Enterococcus faecium, faecalis
Total bacterial abundances ranged from 7.710 to Haemophilus influenzae
1.6 107 cells m 3 air in PM samples. The average airborne PM Klebsiella pneumoniae, oxytoca
samples harbored 31 phyla and 379 genera, and the total airborne PM samples Legionella longbeachae
Mycobacterium scrofulaceum
harbored 42 phyla and 569 genera (Table S2). Proteo-bacteria, Cyanobacteria, Nocardia brasiliensis
Actinobacteria, Firmicutes and Bacteroidetes were the dominant bacteria Pseudomonas aeruginosa
phyla in all the airborne PM samples, with the average relative abundances of Staphylococcus haemolyticus, saprophyticus
32.2%, 18.0%,16.5%, 15.5% and 11.6%, respectively (Fig. 1). Stenotrophomonas maltophilia
Alphaproteobacteria, Actinobacteria, Betaproteobacteria, Streptococcus pneumoniae, pyogenes
Vibrio cholerae
Oscillatoriophycideae and Clostridia were the
486 H. Liu et al. / Environmental Pollution 233 (2018) 483e493

2003) had a relative abundance of 0.72%. Clostridium constituted the largest

number of detected pathogenic species. Clostridium perfringens (average
relative abundance was 0.54%) is responsible for food poisoning (Warrell et
al., 2003). Clostridium botulinum (average relative abundance was 0.0063%)
can produce the neurotoxin botulinum, which causes flaccid paralytic disease
(Lindstrom€ and Korkeala, 2006). Clostridium tetani (average relative
abundance was 0.0032%) causes tetanus (Baron, 1996). Strepto-coccus
contained some common pathogenic species. Streptococcus pneumoniae
(average relative abundance was 0.091%) causes pneumonia in the upper
respiratory tract (Ryan and Ray, 2004) and Streptococcus pyogenes (average
relative abundance was 0.0016%) typically infects the throat or skin (Pericone
et al., 2000).

3.2. The relationship of AQI and the community compositions of the total
bacteria and pathogenic bacteria

Among all the airborne PM samples, the lowest concentrations were 7.7
4 5 3
10 and 1.5 10 cells m air, which were detected in good (AQI ¼ 37,
PM10) and heavily polluted air (AQI ¼ 286, PM2.5) respectively (Table S3).
Considering the diversity of bacteria com-munity, the numbers of OTUs were
analyzed (Fig. 2). In the PM2.5 samples, the number of OTUs peaked in Fig. 3. The relative abundance of pathogenic bacteria in all the airborne PM samples.
moderately polluted air (AQI ¼ 185), while in the PM10 and TSP samples
the maximum number of OTUs appeared under the conditions of moderate
(AQI ¼ 92) and lightly polluted (AQI ¼ 107) air, respectively. This result is that in the good (AQI ¼ 37) air samples with the relative abundance of 2.3%.
different from the hypothesis that the number of OTUs would increase with In the TSP samples, the relative abundance of pathogenic bacteria was
the increase of AQI. Besides, AQI had a strong relationship with the structure changed not obviously among different air quality levels. Additionally, AQI
of total bacteria community (Fig. S1). was related with the structure of patho-genic bacteria community, which is
presented in Fig. S1.

With the increasing of AQI, the proportion of pathogenic bac-teria may

increase in the PM2.5 and PM10 samples (Fig. 3). In the PM2.5 samples, the
relative abundance of pathogenic bacteria was highest of 7.0% in the heavily
polluted air (AQI ¼ 286), which was an increase of 1.8 times than that in the
moderate (AQI ¼ 88) air samples with the relative abundance of 2.5%. In the
PM10 samples, the relative abundance of pathogenic bacteria increased with
an increase of AQI and peaked in moderately polluted (AQI ¼ 195) air
samples. The relative abundance of pathogens in the moderately polluted air
samples was 9.9%, which was 3.3 times greater than
3.3. The relationship of particle size and the distribution of the total bacteria
and pathogenic bacteria
The community compositions of total bacteria in the PM10 samples were
different from that in the PM2.5 and TSP samples. Some dominant phyla may
have contributed to this difference (Fig. 4). The average relative abundance of
Firmicutes was 28.65 ± 0.44% in the PM10 samples, which was 2.15 times
higher than that in the PM2.5 samples (9.09 ± 0.04%, P < 0.005) and 1.71
times higher than that in the TSP samples (10.56 ± 0.10%, P < 0.005).
Cyanobacteria exhibited the lowest relative abundance of 10.25 ± 0.10% in
the PM10 samples, whose relative abundance was 0.58 times lower than that
in the PM2.5 samples (24.32 ± 0.31%, P < 0.005) and 0.52 times lower than
that in the TSP samples (21.41 ± 0.36%, P < 0.05). In addition, the proportion
of Actino-bacteria increased with an increase of particle sizes. The proportion
of Proteobacteria was highest in the PM2.5 samples.

The composition of pathogenic bacteria was related to the par-ticle size

distribution. In the PM10 samples, the relative abundance of pathogenic
bacteria genus was always lower than its in the PM2.5 and TSP samples (Fig.
5). The average relative abundance of Enterobacter was 0.22% ± 0.00006% in
the PM10 samples, a 0.62 times decrease from the value of 0.58 ± 0.0007%
(P < 0.05) in the PM2.5 samples and a 0.53 times decrease from the value of
0.47% ± 0.0002% (P < 0.05) in the TSP samples. Klebsiella, Entero-coccus,
Stenotrophomonas, Burkholderia, and Pseudomonas also presented low
proportions in the PM10 samples. By contrast, Clostridium and
Staphylococcus presented the highest proportion in the PM10 samples. The
average relative abundance of Clostridium was 1.20% ± 0.0003% in the
PM10 samples, an increase of 2.36 times than the value of 0.36% ± 0.0006%
(P < 0.01) in the PM2.5 samples and an increase of 2.31 times than the value
of 0.36% ± 0.0008% (P < 0.05) in the TSP samples.

Fig. 2. The number of total bacterial OTUs in all the airborne PM samples.
 H. Liu et al. / Environmental Pollution 233 (2018) 483e493 487

Fig. 4. The relative abundance of 20 most abundant total bacteria phyla in the airborne PM samples with different particle sizes (n ¼ 3).

Fig. 5. The relative abundance of pathogenic bacteria genera in the airborne PM samples with different particle sizes (n ¼ 3).
488 H. Liu et al. / Environmental Pollution 233 (2018) 483e493
3.4. The relationship of environmental factors and the community structure
of total bacteria and pathogenic bacteria
To investigate the variable composition of total bacteria and pathogenic
bacteria at different levels of air quality, some factors were taken into
account. The correlations between the numbers of OTUs in the total airborne
bacteria and the environmental factors were analyzed (Fig. 6). Among the
meteorological factors, the best predictor of total bacteria community
2 C, D and E was showed in Table S1.
diversity was relative hu-midity (R ¼ 0.11). Both the relative humidity and
2
wind scale (R ¼ 0.09) were negatively correlated with the total bacteria
2
community diversity. However, the air temperature (R ¼ 0.03) was not
closely correlated with the diversity of the total bacteria com-munities.
Among the chemical factors, the CO and O3 concentra-tions were the most
important chemical factors. The total number of bacterial OTUs was
2 were the dominant phyla among the total bacteria, which is in agreement with
positively correlated with the O3 concen-tration (R ¼ 0.15), whereas it was
2 2
negatively correlated with the CO concentration (R ¼ 0.10). The NO2 (R
2
¼ 0.03) and SO2 (R ¼ 0.01) concentrations had no obvious correlation with
the total bacteria community diversity. Additionally, the relationship be-
tween the total bacteria community structure and the environ-mental factors
were analyzed (Fig. S1). Air temperature was the least relevant environmental
factors. The O3 concentration had negative relationship with the other
environmental factors except air temperature.
Environmental factors include Meteorological factors: (A) Relative
humidity, (B) Wind scale and (C) Air temperature, and Chemical factors: (D)
CO concentration, (E) O3 concentration, (F) NO2 concentration and (G) SO2
concentration.
The correlations between the relative abundance of pathogenic bacteria
and environmental factors were analyzed (Fig. 7). Among all the
2
meteorological factors, relative humidity (R ¼ 0.36) was the most correlated
factor with the proportion of pathogenic bacteria. The proportion of
pathogenic bacteria was positively correlated with the relative humidity,
2
whereas it was negatively correlated with the wind scale (R ¼ 0.30) and the
2
air temperature (R ¼ 0.10). Among the chemical factors, the CO
2 In PM2.5 and PM10 samples, the proportions of pathogenic bacteria were
concentration was most closely correlated (R ¼ 0.42) with the proportion of
pathogenic bacteria. The relative abundance of pathogenic bacteria was
2 samples, the proportion of Streptococcus pneumoniae was also higher in
positively correlated with the concentrations of CO and SO2 (R ¼ 0.36) but
2
was negatively correlated with the O3 (R ¼ 0.10) concentration. The NO2
2 air toxins than other bacteria. When air pollution becomes severe, patho-genic
(R ¼ 0.01) concentration was not obviously correlated with the relative
abundance of pathogenic bacteria. Considering the whole structure of
pathogenic bacteria communities, the RDA analysis showed that the less
relevant environmental factors were air temperature and NO2 concentration.
Whereas the more relevant environmental factors were wind scale, relative
humidity, O3, SO2 and CO concentration. Additionally, the wind scale and
O3 con-centration were negatively related with the relative humidity, SO2 and
CO concentration (Fig. S1).
Meteorological factors: (A) Relative humidity, (B) Wind scale, (C) Air
temperature; Chemical factors: (D) CO concentration, (E) O3 concentration,
(F) NO2 concentration, (G) SO2 concentration.
3.5. Comparison of the total bacteria communities and potential source
environment
To investigate whether the total bacteria communities in the airborne
particles were derived from the surrounding environ-ments, two water
samples (E1, E2), two surface soil samples (E3, E4) and two ground dust
samples (E5, E6) were analyzed. The results showed that the total bacteria
community compositions were very similar among the airborne PM samples
but were very different from those in the surrounding environmental samples
(Fig. 8). Furthermore, the bacterial communities in surrounding
environmental samples from the same habitats were obviously clustered,
especially in the water samples (E1, E2). It is indicated that some bacteria
may be transmitted via airflow over a long distance.
Airborne PM samples: red for PM2.5, blue for PM10, orange for TSP;
surrounding environmental samples: green, E1 and E2 for water, E3 and E4
for surface soil, E5 and E6 for ground dust. The detailed information of A, B,
4. Discussion
The compositions of total bacteria and pathogenic bacteria communities
were affected by both the level of air quality and the size of the PM.
Proteobacteria, Cyanobacteria, Firmicutes, Actino-bacteria, and Bacteroidetes
other reported studies (Barberan et al., 2015; Cao et al., 2014). Cyanobacteria,
phototrophic bacteria, were active in atmosphere (Al-Bader et al., 2012). In
the previous study, Cyanobacteria were dominant at the altitudes of 1000 m
and 10 m over the Noto Peninsula (Maki et al., 2015). In addition, 23
pathogenic bacteria species belonging to 16 genera were observed in the
“Human Pathogen Database”. A previous study has also detected several
pathogens in the polluted air (Cao et al., 2014), but more pathogenic bacteria
were observed and discussed in our study.
With the increasing of AQI value, the numbers of total bacterial OTUs
were rose first and then descended. The concentrations of total bacteria were
lower in good and heavily polluted air samples than those in the air with other
quality levels. On sunny days, Wei et al. (2016) reported that the
concentrations of bioaerosol and microbe were low. On hazy days, Gao et al.
(2015) concluded that the concentrations of airborne bacteria were low. Some
bacteria prefer to live under the condition of low PM concentrations and
high light (Sanchez-baracaldo, 2015). The compositions of heavily polluted
air might cause damage to the total bacteria (Raisi et al., 2010).
almost increased with the increasing of air quality levels. In Beijing's smog
heavily polluted samples than it in less polluted airborne samples (Cao et al.,
2014). It is speculated that pathogenic bacteria have the ability to endure more
bacteria could still survive, whereas the other bacteria may die. And the
decreased proportions of other bacteria could lead to an increasing proportion
of pathogenic bacteria. Some studies have showed that the air pollution had
adverse effects on lungs (Baccarelli et al., 2014; Smith et al., 2016) and
cardiovascular (Hemmingsen et al., 2015; Su et al., 2016). The newborn baby
(Hao et al. 2016; Raz et al., 2015) and the senior people (Chen et al., 2015)
should be paid more attention. In heavily polluted air, the patho-genic bacteria
potentially posed a concern hazard to human health, although there was no
quantitative correlation between pathogenic bacteria and human disease.
Considering the influence of particle size on total bacteria and pathogenic
bacteria, the PM10 samples were distinct from the PM2.5 and TSP samples.
Similarly, other research has also concluded that spore-forming bacteria such
as Firmicutes was more dominant in PM10 samples than in PM2.5 samples
(Polymenakou et al., 2008). With the increasing of PM particle size, the
relative abundance of pathogenic bacteria genera neither increased nor
decreased. The sizes of the total bacteria and pathogenic bacteria may be
responsible for the bacterial proportions in the PM samples with different
particle sizes. A previous study has shown that there was a larger range of
bacterial sizes in PM2.5-10 samples than that
 H. Liu et al. / Environmental Pollution 233 (2018) 483e493 489

Fig. 6. Relationships between environmental factors and total bacteria community diversity.
490 H. Liu et al. / Environmental Pollution 233 (2018) 483e493

Fig. 7. Relationships between environmental factors and the proportion of pathogenic bacteria.
 H. Liu et al. / Environmental Pollution 233 (2018) 483e493
Fig. 8. The distribution of total bacteria communities in the airborne PM samples and surrounding environmental samples.
in the PM2.5 samples (Cao et al., 2014). Compared with the PM2.5 and
PM10, the TSP had a broader range of particle sizes. Although PM2.5 and
PM10 were parts of the TSP, the TSP samples still con-tained some larger
sizes (Brook et al., 1997). These large particles of TSP may interfere with the
composition of total bacteria and pathogenic bacteria.
The studies showed that some meteorological factors had re-lationships
with the structure of total bacterial and pathogenic bacteria community. The
relative humidity was significantly related with the airborne bacterial
concentrations (Hwang et al., 2017). It is known that high relative humidity
was important for microor-ganism growth (Tsai and Liu. 2009). In this paper,
the relative hu-midity was the most crucial meteorological factor, which was
positively related with the relative abundance of pathogenic bac-teria. In
polluted air, a high relative humidity may contribute to the survival of
bacteria (Cao et al., 2014; Stanier et al., 2004). Previous study showed that
the wind speed was positively correlated with the microorganisms in PM10
and PM2.5 (Alghamdi et al., 2014). But our research indicated that the wind
scale was negatively related with bacteria. As we know, wind can dilute the
air pollution. In winter, good air usually comes after the windy days. Both
total bacteria and pathogenic bacteria had a low proportion in good air. So the
correlation was negative between the wind scale and the bacterial diversity.
Air temperature had no obvious correlations with the total airborne bacteria
and pathogenic bacteria, although air temperature could influence the
concentration of the PM10 (Alghamdi et al., 2014; Hwang et al., 2017).
The structure of the total bacterial and pathogenic bacteria community
was correlated with the AQI, CO, SO2, NO2 and O3 concentrations. The AQI
value was the most important factor, which was significantly correlated with
the airborne microbes concentration (Dong et al., 2016). The concentrations
of CO, SO2 and NO2 had a positive relationship with the total bacteria. It is
also reported that the concentrations of CO, SO2 and NO2 had significant
491
relationships with fungal concentrations (Ho et al., 2005). Thees factors may
act as the energy which promoted the presence of total bacteria. The O 3
concentration was negatively correlated with the relative abundance of
pathogenic bacteria, which was in line with previous researches (Cox, 1995;
Dong et al., 2016; Ho et al., 2005). Pathogenic bacteria may be influenced by
O3 for its well-known phototoxic oxidant (Tiedemann and Firsching, 2000).
In the above analyses, both meteorological and chemical factors affected the
total bacterial diversity and the proportion.
Compared with the surrounding environmental samples, the composition
of total bacteria community was manifestly different from that in the airborne
PM samples. It is speculated that the bacteria in airborne particles were
transmitted with long-distance via airflow. In the process of transmission, the
structure of the airborne bacteria community may change. A similar finding
was concluded from the study of the East China Sea, where the long-distance
migration of marine eukaryotic bacteria in the atmo-sphere was thought to be
possible (Cho and Hwang, 2011). How-ever, Bowers et al. (2011) suggested
that most of the bacteria in the air originated from local soil and dust. So, the
speculation of long-distance transport of the airborne bacteria should be
confirmed with more PM samples and more statistics in the future work.
5. Conclusions
The compositions of total bacteria and pathogenic bacteria were observed
at different air quality levels. Observed pathogens may be harmful to human
health, but the influence mechanism of patho-genic on human health remains
unknown. The factors that affect the structure of the total bacteria and
pathogenic bacteria com-munities were also studied. However, there is still an
opportunity to analyze additional influential factors. In the future work,
models can be built to predict air quality according to the changes of bac-teria
communities.
492 H. Liu et al. / Environmental Pollution 233 (2018) 483e493
Competing financial interest declaration
The authors declare that they have no competing financial interests.
Acknowledgments
This work was supported by the Research Center for Air Pollu-tion and
Health in Zhejiang University and the Zhejiang Provincial Analysis Test
Foundation of China (2016C37003). We also acknowledge the support of the
IBM high performance computing cluster of Bio-micromolecules Analysis
Lab, Analysis Center of Aerobiology and Environmental Sciences, Zhejiang
University. Sincerely thanks to the help of Dr. Muhammad Aamir.
Appendix A. Supplementary data
Supplementary data related to this article can be found at
https://doi.org/10.1016/j.envpol.2017.10.070.
References
Al-Bader, D., Eliyas, M., Rayan, R., Radwan, S., 2012. Airedust-borne associations of
phototrophic and hydrocarbon-utilizing microorganisms: promising consortia in volatile
hydrocarbon bioremediation. Environ. Sci. Pollut. R. 19, 3997e4005.
Alghamdi, M.A., Shamy, M., Redal, M.A., Khoder, M., Awad, A.H., Elserougy, Safaa, 2014.
Microorganisms associated particulate matter: a preliminary study. Sci. Total Environ.
479e480, 109e116.
Allegra, S., Leclerc, L., Massard, P.A., Girardot, F., Riffard, S., Pourchez, J., 2016.
Characterization of aerosols containing Legionella generated upon nebulization.
Sci. Rep-UK 6, 1e8.
Baccarelli, A.A., Zheng, Y.N., Zhang, X., Chang, D., Liu, L., Wolf, K.R., et al., 2014. Air
pollution exposure and lung function in highly exposed subjects in Beijing, China: a
repeated-measure study. Part Fibre Toxicol. 11, 51.
Balloy, V., Chignard, M., 2009. The innate immune response to Aspergillusfumiga-tus.
Microbes Infect. 11, 919e927.
Barberan, A., Henley, J., Fierer, N., Casamayor, E.O., 2014. Structure, inter-annual recurrence,
and global-scale connectivity of airborne microbial communities. Sci. Total Environ. 487,
187e195.
Barberan, A., Ladau, J., Leff, J.W., Pollard, K.S., Menninger, H.L., Dunn, R.R., et al., 2015.
Continental-scale distributions of dust-associated bacteria and fungi. Proc. Natl. Acad. Sci.
U. S. A. 112, 5756e5761.
Baron, S., 1996. Medical Microbiology, fourth ed. Univ of Texas Medical Branch. Bibby, K.,
Viau, E., Peccia, J., 2010. Pyrosequencing of the 16S rRNA gene to reveal
bacterial pathogen diversity in biosolids. Water Res. 44, 4252e4260.
Bowers, R.M., Mcletchie, S., Knight, R., Fierer, N., 2011. Spatial variability in airborne
bacterial communities across land-use types and their relationship to the bacterial
communities of potential source environments. Isme J. 5, 601e612.
Brook, J.R., Dann, T.F., Burnett, R.T., 1997. The relationship among TSP, PM10, PM2.5, and
inorganic constituents of atmospheric participate matter at multiple ca-nadian locations. J.
Air Waste Manage 47, 2e19.
Caporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F.D., Costello1, E.K., et
al., 2010. QIIME allows analysis of high-throughput commu-nity sequencing data. Nat.
Methods 7, 335e336.
Cao, C., Jiang, W.J., Wang, B.Y., Fang, J.H., Lang, J.D., 2014. Inhalable microorganisms in
Beijing's PM2.5 and PM10 Pollutants during a Severe smog event. Environ. Sci. Technol.
48, 1499e1507.
Cerdeno~-Tarraga, A.M., Efstratiou, A., Dover, L.G., Holden, M.T., Pallen, M., Bentley, S.D.,
et al., 2003. The complete genome sequence and analysis of Corynebacteriumdiphtheriae
NCTC13129. Nucleic Acids Res. 31, 6516e6523.
Chang, D., Song, Y., Liu, B., 2009. Visibility trends in six megacities in China 1973-2007.
Atmos. Res. 94, 161e167.
Chen, S.Y., Wu, C.F., Lee, J.H., Hoffmann, B., Peters, A., Brunekreef, B., et al., 2015.
Pressure in elderly residents of Taipei city: a cross-sectional study. Environ.
Health Persp. 123, 779e784.
Cho, B.C., Hwang, C.Y., 2011. Prokaryotic abundance and 16S rRNA gene sequences detected
in marine aerosols on the East Sea (Korea). FEMS Microbiol. Ecol. 76, 327e341.
Cox, C.S., 1995. Stability of airborne microbes and allergens. In: Cox, C.S., Wathes, C.M.
(Eds.), Bioaerosols Handbook. Lewis Publishers, New York, pp. 77e99.
Darwin, C., 1846. An account of the fine dust which often falls on vessels in the Atlantic ocean.
Q. J. Geol. Soc. 2, 26e30.
Daumas, A., El-Mekaouiet, F., Batailleal, S., Daniel, L., Caporossi, J.M., Fournier, P.E., et al.,
2012. Acute tubulointerstitial nephritis complicating Legionnaires' dis-ease: a case report.
J. Med. Case Rep. 6, 100.
Dong, L.J., Qi, J.H., Shao, C.C., Zhong, X., Gao, D.M., Cao, W.W., et al., 2016. Concen-tration
and size distribution of total airborne microbes in hazy. Sci. Total
Environ. 541, 1011e1018.
Eames, I., Tang, J.W., Li, Y., Wilson, P., 2009. Airborne transmission of disease in hospitals. J.
R. Soc. Interface 6, 697e702.
Gandolfi, I., Bertolini, V., Bestetti, G., Ambrosini, R., Innocente, E., Rampazzo, G., et al., 2015.
Spatio-temporal variability of airborne bacterial communities and their correlation with
particulate matter chemical composition across two ur-ban areas. Appl. Microbiol. Biot. 99,
4867e4877.
Gao, M., Jia, R.Z., Qiu, T.L., Han, M.L., Song, Y., Wang, X.M., 2015. Seasonal size dis-
tribution of airborne culturable bacteria and fungi and preliminary estimation of their
deposition in human lungs during non-haze and haze days. Atmos. Environ. 118, 203e210.
Gonzalez, A.J., Landeras, E., Mendoza, M.C., 2000. Pathovars of Pseudomonas syringae
causing bacterial brown spot and halo blight in Phaseolus vulgaris L.Are distinguishable by
ribotyping. Appl. Environ. Microb. 66, 850e854.
Gangamma, S., Patil, R.S., Mukherji, S., 2011. Characterization and proinflammatory response
of airborne biological particles from wastewater treatment plants. Environ. Sci. Technol.
45, 3282e3287.
Guo, F., Zhang, T., 2012. Profiling bulking and foaming bacteria in activated sludge by high
throughput sequencing. Water Res. 46, 2772e2782.
Hao, H., Chang, H.H., Holmes, H.A., Mulholland, J.A., Klein, M., Darrow, L.A., et al.,
2016. Air pollution and preterm birth in the U.S. State of Georgia (2002e2006):
associations with concentrations of 11 ambient air pollutants estimated by combining
community multiscale air quality model (CMAQ) simulations with stationary monitor
measurements. Environ. Health Persp. 124, 875e880.
Happo, M.S., Sippula, O., Jalava, P.I., Rintala, H., Leskinen, A., Komppula, M., et al., 2014.
Role of microbial and chemical composition in toxicological properties of indoor and
outdoor airparticulate matter. Part Fibre Toxicol. 11, 60.
Hara, K., Zhang, D.Z., 2012. Bacterial abundance and viability in long-range trans-ported dust.
Atmos. Environ. 47, 20e25.
Hemmingsen, J.G., Rissler, J., Lykkesfeldt, J., Sallsten, G., Kristiansen, J., Peter, M.P., et al.,
2015. Controlled exposure to particulate matter from urban street air is associated with
decreased vasodilation and heart rate variability in overweight and older adults. Part Fibre
Toxicol. 12, 6.
Ho, H.M., Rao, C.Y., Hsu, H.H., Chiu, Y.H., Liu, C.M., Chao, H.J., 2005. Characteristics and
determinants of ambient fungal spores in Hualien, Taiwan. Atmos. Environ. 39,
5839e5850.
Hu, B.L., Darci, R., Erwin, V.D.B., Zheng, P., Mark, V.M., 2011. New anaerobic, ammonium-
oxidizing communities enriched from peat soil. Appl. Environ. Microb. 77, 966e971.
Hwang, S.H., Kim, I.S., Park, W.M., 2017. Concentrations of PM10 and airborne bacteria in
daycare centers in Seoul relative to indoor environmental factors and daycare center
characteristics. Air Qual. Atmos. Health 10, 139e145.
Kim, S.J., Kwon, O., Uh, Y., Hwang, G.Y., Jang, I.H., Yoon, K.J., et al., 2009. Frequency and
clinical characteristics of urinary tract infections caused by Staphylococcus saprophyticus.
Korean. J. Clin. Microbiol. 12, 62e66.
Lelieveld, J., Evans, J.S., Fnais, M., Giannadaki, D., Pozzer, A., 2015. The contribution of
outdoor air pollution sources to premature mortality on a global scale. Na-ture 525,
367e371.
Lindstrom,€ M., Korkeala, H., 2006. Laboratory diagnostics of botulism. Clin. Micro-biol. Rev.
19, 298e314.
Liu, S., Lou, L.P., Tian, G.M., Zheng, P., Hu, B.L., 2013. Spatial distribution and factors
shaping the niche segregation of ammonia-oxidizing microorganisms in the Qiantang
River, China. Appl. Environ. Microb. 79, 4065e4071.
Maki, T., Hara, K., Kobayashi, F., Kurosaki, Y., Kakikawa, M., Matsuki, A., et al., 2015.
Vertical distribution of airborne bacterial communities in an Asiandust down-wind area,
Noto Peninsula. Atmos. Environ. 119, 282e293.
Masalha, M., Borovok, I., Schreiber, R., Aharonowitz, Y., Cohen, G., 2001. Analysis of
transcription of the Staphylococcus Aureus aerobic Class Ib and anaerobic Class
III ribonucleotide reductase genes in response to oxygen. J. Bacterio 183, 7260e7272.
Pericone, C.D., Overweg, K., Hermans, P.W., Weiser, J.N., 2000. Inhibitory and bactericidal
effects of hydrogen peroxide production by Streptococcus pneu-moniae on other
inhabitants of the upper respiratory tract. Infect. Immun. 68, 3990e3997.
Polymenakou, P.N., Mandalakis, M., Stephanou, E.G., Tselepides, A., 2008. Particle size
distribution of airborne microorganisms and pathogens during an intense African dust event
in the Eastern Mediterranean. Environ. Health Persp. 116, 292e296.
Prospero, J.M., Blades, E., Mathison, G., Naidu, R., 2005. Interhemispheric transport of viable
fungi and bacteria from Africa to the Caribbean with soil dust. Aero-biologia 21, 1e19.
Quast, C., Pruesse, E., Yilmaz, P., Gerken, J., Schweer, T., 2013. The SILVA ribosomal RNA
gene database project: improved data processing and web-based tools. Nucleic Acids Res.
41, 590e596.
Raisi, L., Lazaridis, M., Katsivela, E., 2010. Relationship between airborne microbial and
particulate matter concentrations in the ambient air at a Mediterranean site. Glob. Nest J.
12, 84e91.
Raz, R., Roberts, A.L., Lyall, K., Hart, J.E., Just, A.C., Laden, F., et al., 2015. Autism spectrum
disorder and particulate matter air pollution before,during, and after pregnancy: a nested
caseecontrol analysis within the nurses' health study II cohort. Environ. Health Persp. 123,
264e270.
Ren, H.X., Wang, W., Liu, Y., Liu, S., Lou, L.P., Cheng, D.Q., et al., 2015. Pyrosequencing
analysis of bacterial communities in biofilms from different pipe materials in a city
drinking water distribution system of east China. Appl. Microbiol. Biot. 99, 10713e10724.
 H. Liu et al. / Environmental Pollution 233 (2018) 483e493
Roberts, K., Smith, C.F., Snelling, A.M., Kerr, K.G., Nafield, K.R., Sleigh, P.A., et al., 2008.
Aerial dissemination of Clostridium difficle spores. BMC Infect. Dis. 8, 7.
Ryan, K.J., Ray, C.G., 2004. Sherris Medical Microbiology:An Introduction to Infec-tious
Disease, fourth ed. McGraw-Hill, New York.
Sanchez-baracaldo, P., 2015. Origin of marine planktonic cyanobacteria. Sci. Rep-UK 5, 17418.
Smith, G.S., Van Den Eeden, S.K., Garcia, C., Shan, J., Baxter, R., Herring, A.H., et al., 2016.
Air pollution and pulmonary tuberculosis: a nested caseecontrol study among members of a
northern California health plan. Environ. Health Persp. 124, 761e768.
Spencer, R.C., 2003. Bacillus anthracis. J. Clin. Pathol. 56, 182e187.
Stanier, C.O., Khlystov, A.Y., Chan, W.R., Mandiro, M., Pandis, S.N., 2004. A method for the
in situ measurement of fine aerosol water content of ambient aerosols: the dry-ambient
aerosol size spectrometer (DAASS). Aerosol Sci. Tech. 38, 215e228.
Su, T.C., Hwang, J.J., Shen, Y.C., Chan, C.C., 2016. Carotid intima-media thickness and long-
term exposure to traffic-related air pollution in middle-aged residents of Taiwan: a cross-
sectional study. Environ. Health Persp. 123, 773e778.
Tiedemann, A.V., Firsching, K.H., 2000. Interactive effect of elevated ozone and carbon dioxide
on growth and yield of leaf rust infected versus non infected
493
wheat. Environ. Pollut. 108, 357e363.
Tsai, M.Y., Liu, H.M., 2009. Exposure to culturable airborne bioaerosols during noodle
manufacturing in central Taiwan. Sci. Total Environ. 407, 1536e1546.
Wang, Q., Garrity, G.M., Tiedje, J.M., Cole, J.R., 2007. Naive bayesian classifier for rapid
assignment of rRNA sequences into the new bacterial taxonomy. Appl. Environ. Microb.
73, 5261e5267.
Warrell, D.A., Cox, T.M., Firth, J.D., Benz Jr., E.J., 2003. Oxford Textbook of Medicine, fourth
ed. Oxford University Press, London.
Wei, K., Zou, Z.L., Zheng, Y.H., Li, J., Shen, F.X., Wu, C.Y., et al., 2016. Ambient bio-aerosol
particle dynamics observed during haze and sunny days in Beijing. Sci. Total Environ. 550,
751e759.
Woo, A.C., Brar, M.S., Chan, Y.K., Lau, M.C.Y., Leung, F.C.C., James, A., et al., 2013.
Temporal variation in airborne microbial populations and microbially-derived allergens in a
tropical urban landscape. Atmos. Environ. 74, 291e300.
Yamaguchi, N., Tokunaga, Y., Goto, S., Fujii, Y., Banno, F., Edagawa, A., 2017. Rapid on-site
monitoring of Legionella pneumophila in cooling tower water using a portable microfluidic
system. Sci. Rep-UK 7, 1e8.
Ye, L., Zhang, T., 2011. Pathogenic bacteria in sewage treatment plants as revealed by 454
pyrosequencing. Environ. Sci. Technol. 45, 7173e7179.