Investigation of bacterial effects of Asian dust events through comparison with seasonal variability in outdoor airborne bacterial community

Atmospheric bacterial dispersion with aeolian dust has been reported to have a potential impact on public health and ecosystems. Asian dust is a major aeolian event that results in an estimated 4 million tons of Asian dust particles falling in Japan annually, 3,000–5,000 km away from their source regions. However, most studies have only investigated the effects of Asian dust during dust seasons. Therefore, in this study, outdoor bacterial abundance and community composition were determined by 16S rRNA quantitative PCR and amplicon sequencing, respectively, and compared on Asian and non-Asian dust days (2013–2015; 44 samples over four seasons). Seasonal variations in bacterial abundance of non-Asian dust days were not observed. Bacterial abundance of individual samples collected on non-Asian dust days changed dynamically relative to Asian dust days, with bacterial abundance occasionally reaching those of Asian dust days. The bacterial community composition on non-Asian dust days was rather stable seasonally, and did not differ from that on Asian dust days. These results indicate that bacteria in Asian dust does not immediately influence indigenous bacterial communities at the phylum/class level in distant downwind areas; accordingly, further studies of bacterial communities in downwind areas closer to the dust source are warranted.

even though the bacterial abundance and community composition of aerosols in outdoor environments are thought to be affected by seasonal and weather-related variations [15][16][17][18][19] . To assess the effects of bacteria transported with aeolian dust on public health and the environment, bacterial variations should be evaluated by long-term monitoring.
Therefore, the present study was conducted to investigate the effects of bacteria transported by Asian dust events on humans and the ecosystem based on outdoor aerosol samples collected on both Asian dust and non-Asian dust days from 2013 to 2015. We analyzed variations in bacterial abundance and bacterial community composition on non-Asian dust days to understand variations in the local airborne bacterial community. We then investigated airborne bacterial community characteristics following Asian dust events through comparison with seasonal bacterial community variability on non-Asian dust days and changes in the bacterial community on Asian dust days. Airborne bacterial abundance and community composition were determined by 16S rRNA gene-targeted quantitative PCR and amplicon sequencing, respectively.

Results
Variations in bacterial abundance on non-Asian dust days and comparison with bacterial abundance on Asian dust days. Particle size distribution of aerosols in the outdoor environments was measured ( Supplementary Fig. S1). The number of particles in the outdoor environment was changed by 10 fold. During Asian dust events, particle number was generally elevated, and comparatively large particles were dynamically elevated (P value; > 0. 3  . The results revealed a correlation between bacterial abundance and particle size distribution, with particle sizes larger than 1.0 μ m showing a greater correlation (Fig. 1).
Accordingly, we compared bacterial abundance and number of particles larger than 1.0 μ m during different seasons, rainfall events, and Asian dust occurrences (Fig. 2). The levels of particles larger than 1.0 μ m fluctuated between 3 × 10 2 and 3 × 10 3 L −1 , regardless of season. On Asian dust days, particle levels ranged from 2 × 10 3 to 7 × 10 3 L −1 . Particle numbers on Asian dust days were higher than those on non-Asian dust days, and their fluctuation was more stable on Asian dust days.
Bacterial abundance in outdoor environments varied with variations in particle number ( Fig. 2; 1 × 10 2 -1 × 10 4 cells m −3 ), and bacterial abundance was not influenced by rainfall in this study. However, bacterial abundance generally increased as the number of particles (> 1.0 μ m) increased, without response to seasonal variations or occurrence of Asian dust (correlation coefficient; r > 0.7). Bacterial abundance on Asian dust days was generally greater than 10 4 cells m −3 . The average bacterial abundance on Asian dust days ([1 ± 0.6] × 10 4 cells m −3 ) increased by approximately 5 times relative to non-Asian dust days ([2 ± 3] × 10 3 cells m −3 ). However, bacterial abundance fluctuated from 10 2 to 10 4 cells m −3 on non-Asian dust days and changed dynamically relative to Asian dust days, with bacterial abundance reaching that of Asian dust days on several occasions (20 August, 27 August,  3 September, and 14 September, 2013; 16 April and 23 April, 2015).
The ratio of bacterial abundance to number of particles (> 1.0 μ m) was comparatively higher in summer and fall (0.56% and 0.28%, respectively). However, this ratio was lower in winter than in other seasons (0.15% in spring, 0.06% in winter). On Asian dust days, the ratio of bacterial abundance to particle number was 0.30% and stable relative to non-Asian dust days.

Variations in bacterial community composition on non-Asian dust days and comparison
with bacterial community composition on Asian dust days. To investigate the bacterial effects of Asian dust, we also analyzed bacterial community composition with variations in environmental conditions on non-Asian dust days. Airborne bacterial community composition in outdoor environments has been reported to change in response to variations in environment factors [15][16][17][18][19] . In this study, the airborne bacterial community composition was determined using 16S rRNA gene targeted ion PGM sequencing in conjunction with a two-step PCR method. Two-step PCR has advantages such as increased reproducibility and recovery of higher genetic diversity during amplicon sequencing 20,21 .
In the two-step PCR method, we used the 968f-1401r primer (V6-V8) set because it produced the highest diversity in a preliminary study using PCR-DGGE to select the proper primer ( Supplementary Fig. S2). To analyze similarities in the bacterial community composition of each sample, amplicon sequencing data of bioaerosol samples were processed using the QIIME software, and the results were indicated using multidimensional scaling (MDS) (Fig. 3). The results revealed that bacterial community composition in the outdoor environment was rather stable, despite changes in season, and samples were generally not affected by variations in environmental factors. However, the bacterial community compositions of samples collected on 5 August 2013 and 11 November 2014 differed from others. In addition, bacterial community composition did not differ significantly on Asian dust days and non-Asian dust days. Comparative analysis of the bacterial community composition at the phylum and class level revealed that environmental factors such as season and rainfall had no effect on the predominant bacterial phylum and class (Fig. 4).
On 5 August, 2013, the bacterial community composition differed from that observed on other sampling dates, with Acidobacteria being the dominant member (78%). Acidobacteria are generally the dominant phyla in soil habitats 22 . Additionally, on 11 November, 2014, Gammaproteobacteria, which is known to exist in diverse environments, was dominant (50%). Bacilli can form spores and withstand severe conditions such as those found in sources of Asian dust 23 . There have been many reports of increased levels of Bacilli on Asian dust days in downwind regions far from the source regions of Asian dust 9 ; however, Bacilli accounted for more than 50% of the population on several non-Asian dust days in this study (13 June, 2013 [54%], 12 Febraury, 2015 [78%]). Bacilli did not increase in response to Asian dust events in this downwind area.

Discussion
In this study, bacterial number and community composition were calibrated based on copy number of 16S rRNA gene of each bacterial phylum. This was conducted because both high copy number bacteria (e.g., Bacilli) and low copy number bacteria (e.g., Acidobacteria) were present in the collected dust samples.
Although bacterial abundance has been reported to change in response to variations in environmental factors [15][16][17][18][19] , it was not correlated with any environmental factors (season, temperature, humidity, wind speed, wind direction, rainfall) except particle numbers in the present study. During winter, the ratio of bacterial abundance to particle number was low relative to other seasons. Atmospheric bacterial abundance would be lower in winter because of the response to low temperature 24 .
No considerable increase in bacterial abundance was observed on Asian dust days relative to fluctuations in bacterial abundance on non-Asian dust days. Bacterial abundance on aerosols of indoor environments usually ranges from 10 5 to 10 6 cells m −3 25 , and bacterial abundance in outdoor environments determined in this study was 10-100 times lower than those in indoor environments.
The results on non-Asian dust days appeared to be correlated with the environmental characteristics of the sampling location. Specifically, we monitored bioaerosols in environments in which temperatures are suitable to the growth of general bacteria (from 4 °C to 34 °C; average 21 ± 9 °C). It has been reported that bacterial abundance and community composition changed significantly in response to season in specific places (e.g., coastal sites 15 , high-elevation sites 16 ). However, none of our sampling points were located in places such as these. Variations in atmospheric bacterial community composition in outdoor environments impacted by Asian dust occurrence were more stable than those observed on non-Asian dust days. Accordingly, these findings indicate that bacterial effects on humans and ecosystems in distant downwind areas impacted by Asian dust may be lower than those of general changes in the natural environment. Information describing the viability of airborne bacteria collected on Asian dust days can help accurately estimate their influences on public health and ecosystems. Several methods, such as fluorescent vital stain 10 , can be used to estimate bacterial activity in aquatic and soil environments. However, bacterial abundance in atmospheric environments is much lower than in other natural environments; therefore, accurate evaluation of viability of airborne bacteria remains difficult. In addition, new methods are required to simultaneously obtain bacterial viability and phylogenetic information.  We demonstrated that bacteria in Asian dust that had been transported long distances did not immediately influence the bacterial community in downwind areas. Furthermore, our findings suggest that bacterial communities may be affected more by ground environments along the transfer route and local environments than by the bacterial community in the dust itself. However, more severe occurrences of Asian dust in areas closer to the dust source may result in microbes in the dust having a greater impact on the indigenous bacterial community. The amount of Asian dust fallout is estimated to be 180 g m −2 year −1 in Beijing, China 26 (500-2,500 km from the dust source region) and 0.005-0.05 g m −2 year −1 in Osaka, Japan 6 (3,000-5,000 km from the dust source region). Accordingly, future studies of the bacterial community in downwind areas closer to the source are warranted to better assess the impacts of aeolian dust on public health and ecosystems.

Materials and Methods
Sample collection. Aerosol Table 1). The occurrence of atmospheric Asian dust was confirmed using information from the Japan Meteorological Agency, LIDAR (Light Detection and Ranging) data from the Ministry of the Environment, Japan (http://www-gis5.nies.go.jp/eastasia/DustLider.php), and visibility at the sampling location ( Supplementary Fig. S3).
The geographic origins of Asian dust collected in this study were determined by back trajectory analysis (http://ready.arl.noaa.gov/HYSPLIT.php), and we confirmed that the origin of all Asian dust samples was the Gobi Desert.

Estimation of bacterial abundance.
To determine bacterial abundance, 16S rRNA gene was quantified by real-time PCR using a LightCycler (Roche Diagnostics, Mannheim, Germany). Real-time PCR was performed with eubacterial primer sets as described by Yamaguchi et al. 12 . A total of 1 × 10 1 to 1 × 10 7 copies per reaction of PCR products of E. coli W3110 were used as the DNA template to generate a standard curve for quantification of the 16S rRNA gene. The copy number of the 16S rRNA gene differed among bacterial phyla; therefore, bacterial abundance was calibrated based on the results of bacterial community composition analysis at the phylum level.
Analysis of bacterial community composition. Two-step PCR was conducted to amplify the 16S rRNA gene for pyrosequencing 28 . Using this approach, tags and adapters were added in a second round of PCR amplification. The first PCR amplification was conducted using the universal primers 968F (AACGCGAAGAACCTTAC) and 1401R (CGGTGTGTACAAGACCC) to amplify a 434-bp fragment of the 16S rRNA gene between the V6 and V8 regions 29 . PCR amplification was performed using the reagents supplied with the AmpliTaq Gold kit (Applied Biosystems, Carlsbad, CA, USA). The reaction mixture contained 2.5 U mL −1 AmpliTaq Gold, 0.5 μ of each primer, 0.2 mM of each dNTP, 1.5 mM MgCl 2 , and 12.5 μ g mL −1 8-methoxypsoralen (Sigma Aldrich, St. Louis, MO, USA; dissolved in dimethyl sulfoxide) in 49 μ L PCR buffer. A 1-μ L DNA suspension was added after irradiation of the PCR mixture with ultraviolet light 30 . The reaction cycle consisted of an initial denaturing step at 94 °C for 9 min, followed by 10 cycles of denaturing at 95 °C for 1 min, annealing at 63-53 °C (decreased by 1 °C per cycle) for 1 min, and extension at 72 °C for 3 min. This was followed by 20 cycles of denaturing at 95 °C for 1 min, annealing at 53 °C for 1 min, and extension at 72 °C for 3 min, with a final extension step at 72 °C for 10 min. Primary PCR products were then purified using a MonoFas PCR Purification Kit (GL Sciences, Tokyo, Japan) and eluted with 40 μ L of TE buffer. Next, a second round of PCR was performed as described above, except primers with an adapter and barcodes of 10 nucleotides in length were used. Furthermore, the number of PCR cycles was reduced to 20 (10 cycles of annealing at 63-53 °C and 10 cycles of annealing at 53 °C). Amplicon sequencing using Ion PGM (Thermo Fisher Scientific KK, Yokohama, Japan) was carried out at the Center for Medical Research and Education, Osaka University (Osaka, Japan).
Raw sequence data of the obtained amplicons were screened, trimmed, and filtered using the default settings of QIIME pipeline version 1.9.1 (http://qiime.org/), resulting in over 125,000 sequences across all samples (3,200 sequences per sample, on average). Total operational taxonomic units (OTUs), which were defined at the 97% nucleotide-sequence identity level using the UCLUST function of the QIIME software 31 , were identified in all sequences, with about 1,500 OTUs per sample on average being recovered. Beta diversity measures were also calculated. Differences in community composition of each sample were assessed graphically using the ordination method of non-metric MDS calculated based on the Euclidean distance.  Table 1. Sample descriptions and associated physical characteristics of the atmosphere. *Determined by information obtained from the Japan Meteorological Agency, LIDAR, and visibility at the sampling site.