The PM samples were obtained each day by MD8 Air Scan sampling device (Air Scan Sartorius AG, Gottingen, Germany) with the average flow rate of 40 L/hour, through the sterile gelatin filters (diameter of 80 mm, pore size of 3 µm, 17528-80-ACD, Sartorius) during November 30 to December 21 of 2016 in different functional regions of Wuqing District, Tianjin (Table 1), including: the control groups sampled in clean days (Cle1 from hospitals, and Cle2 from traffic hubs), Group 1 sampled from hospitals (Hosp1, Hosp2, and Hosp3), Group 2 sampled from traffic hubs(Trans1, Trans2, and Trans3), Group 3 sampled from schools (Sch1, Sch2, and Sch3), and traffic hubs (Tran1, Tran2, and Tran3). Initially, we also sampled the materials in clean days from the school as the third control (Cle3). Unfortunately, the sequencing of this sample was failed. Therefore, we did not include this control sample to the manuscript. All the samplers were placed 1.5 m above the ground. The variations of PM2.5 and PM10 concentrations during the sampling periods were made available online (China national environmental monitoring center). The gelatin filters were stored at −20 °C prior to extraction of total DNA.
Groups Functional areas Abbreviations of sampling sites Control Clean days of hospitals and traffic hubs Cle1 (hospital areas), Cle2 (traffic hubs) Group1 Hospitals Hosp1, Hosp2, Hosp3 Group2 Traffic hubs Trans1, Trans2, Trans3 Group3 Schools Sch1, Sch2, Sch3
Table 1. Sampling groups of the different functional areas
The morphology of PM samples on the gelatin membranes was characterized by a Phenom scanning electron microscope (SEM) supported with image software (SEM, ProX, Holland).
The gelatin filters were dissolved in 5 mL sterile water (preheated to 37 °C). The pathogenic microorganisms captured by gelatin filter were lysed by the alkaline lysing liquid (NaOH 50 mmol/L, SDS 1%, Protease K 10 mg/L, RNase 20 mg/L), the total DNA was enriched using magnetic nanoparticles modified by polyquaternary amino salt polymers, and the total DNA was eluted by elution (Jinping Biotech, China) according to the instruction. The extracted DNA was stored at −80 °C for further use.
The 16S rDNA V3+V4 region was amplified using the forward primer 338F (5′-ACTCCTACGGGAGGCAGCA-3′) and the reverse primer 806R (5′-GGACTACHVGGGTWTCTAAT-3′), the 18S rDNA ITS region was amplified using primers ITS1 (5′-CTTGGTCATTTAGAGGAAGTAA-3′) and ITS2 (5′-GCTGCGTTCTTCATCGATGC-3′). The primers were synthesized by Sangon Ltd., Beijing. The PCR reaction procedure is shown in Supplementary Table 1, available online.
The obtained total 16S rDNA and 18S rDNA were sequenced by double end sequencing (Paired-End) method through FLASH v1.2.7 using the Illumina HiSeq 2500 sequencing platform (BioMarker Technologies Co., Ltd., Beijing, China). Bioinformatics analysis of sequences was conducted using the QIIME (V1.7.0) software package. Sequences with similarities greater than or equal to 97% were grouped into operational taxonomic units (OTUs). The Shannon index was used to estimate the biodiversity of bacteria in a single sample. The principal component analysis (PCA) and the cluster heatmap analysis were also performed to assess the bacterial composition of the samples.
Fig. 1 shows the variations of PM2.5 and PM10 concentrations during the sampling periods. The concentrations of PM2.5 were higher than 200 μg/m3 in 8 days, and the concentrations of PM10 were higher than 200 μg/m3 in 10 days, with the highest PM10 concentration being almost 350 μg/m3, which far exceeded the PRC National Standard PM standards: 50 μg/m3 for PM10, and 35 μg/m3 for PM2.5, respectively (PRC National Standard, 2012). As shown in Fig. 2, the statistical mean diameters of the sampled particles were in the ranges of 100–900 nm and 1.0 –2.5 μm, confirming that the particles were in the scope of fine particulate matter.
Figure 1. Daily average PM2.5 and PM10 concentrations estimated from the samples collected from November 30 to December 21, 2016.
Rarefaction curves derived from the observed OTU number and shannon index were flattened (Fig. 3), showing that our sequencing depth was sufficient to cover the vast majority of bacteria and fungi in the samples. The statistics of OTU species in different ranks of the bacteria and the fungi are shown in Table 2 and 3, and a total of 584 OTUs of bacteria and 370 OTUs of fungi at the genus level were identified during hazy days.
Sample Kindom Phylum Class Order Family Genus Species Cle1 1 20 52 83 142 205 143 Cle2 2 20 50 79 135 220 158 Hosp1 2 31 74 120 226 524 362 Hosp2 1 26 55 100 203 507 316 Hosp3 1 31 79 125 243 487 328 Sch1 1 23 52 88 157 261 175 Sch2 1 27 70 114 187 340 240 Sch3 1 39 105 170 318 584 420 Tran1 1 22 50 80 134 221 151 Tran2 1 22 58 112 221 521 336 Tran3 1 26 71 128 253 582 396 OUTs: operational taxonomic units.
Table 2. Statistics of OTU species of bacteria in different ranks
Sample Kindom Phylum Class Order Family Genus Species Cle1 4 18 39 87 209 348 335 Cle2 4 18 41 88 210 339 361 Hosp1 3 8 22 44 93 131 152 Hosp2 3 9 127 63 134 192 254 Hosp3 4 16 41 83 197 334 319 Sch1 4 17 39 81 198 330 313 Sch2 4 16 39 86 197 331 325 Sch3 4 16 39 79 197 345 377 Tran1 4 16 40 87 213 370 373 Tran2 1 6 19 34 58 85 81 Tran3 3 10 28 64 146 223 259 OUTs: operational taxonomic units.
Table 3. Statistics of OTU species of fungi in different ranks
As shown in Fig. 4, in all samples, Proteobacteria was the most abundant phylum, and four other dominant phyla were Firmicutes, Bacteroidetes, Actinobacteria, and Cyanobacteria (Fig. 4A). Compared to the samples from the clear days, the samples from hospitals and traffic hubs during hazy days contain much higher levels of human health-related bacteria, such as Acinetobacter, Staphylococcus, Corynebacterium, Lactobacillus, and Duganella (Fig. 4B). Especially, Staphylococcus and Corynebacterium were the predominant pathogens around the high-density-population functional areas, such as hospital and transportation areas. For example, Staphylococcus on PM collected around the hospital areas was the dominant bacteria (18% in Hosp2 samples), which may be derived from the high density of patients and health care personnel. Nevertheless, the average abundance of Staphylococcus in Sch2 was only 1.2% because of the regular population. The dominant bacteria were Corynebacterium (34.3%) in Tran2 and Lachnospiraceae_Nk4A136 (10.5%) in Tran3. In Sch3 areas, the dominant bacteria were Lactobacillus (13.5%) and Klebsiella (11.8%). Interestingly, the genus Sphingomonas is detected on both the clear days and hazy days, which are the environment-associated bacteria widely found in surface water, the rhizosphere, sediments, and even soils. Fig. 4C shows that the dominant fungi were Malassezia (58.0%) in Hosp1 and (54.9%) in Hosp2, Alternaria (55.5%) and Cladosporium (9.17%) in Tran3. The genus Malassezia has been associated with a number of diseases affecting the human skin, such as pityriasis versicolor, Malassezia (Pityrosporum) folliculitis, seborrheic dermatitis and dandruff, atopic dermatitis, and psoriasis. Malassezia yeasts are a part of the normal micro-flora, but under certain conditions they can cause superficial skin infection. Cladosporium, Aspergillus and Alternaria could enter the deep lung and cause respiratory diseases.
The variation of bacteria in collected samples were analyzed by two methods: the PCA analysis and the clustering heatmap analysis. PCA analysis showed that the high-population-density functional areas (Hosp1, Hosp2, Tran2, and Tran3 but not Tran1) had a similar bacterial diversity (Fig. 4D). However, the samples from Trans1 and Sch1 displayed little similarity in the bacterial composition. The clustering heat map analysis also demonstrated the similarity of the bacterial composition between the hospital areas and the transportation areas (Fig. 5). One cluster was composed of the hospital areas and the transportation areas; another cluster was composed of the samples from control and the school areas (samples Cle1, Cle2, Sch1, and Sch3). The bacterial diversity around school areas (Sch1 and Sch2) is similar to the control group, but not always the case: the sample Sch3 showed partially similar bacterial diversity to the samples from the hubs and hospitals, which may be attributed to transportation of PM from hubs and hospitals.
Amplification of 16S and 18S rDNA gene sequences
High-throughput sequencing of 16S rDNA
Concentration and diameter of fine particles during the haze events
Microbial species richness and diversity