Ambient Endotoxin and Chemical Pollutant (PM10, PM2.5, and O3) Levels in South Korea

We measured the levels of airborne endotoxins in South Korea and compared them to PM10, PM2.5, and O3 levels in ambient environments; environmental factors affecting these levels were also analyzed. A total of 81 air samples were collected and analyzed using the kinetic Limulus Amebocyte Lysate (LAL) assay. The geometric mean was determined for the levels of endotoxin (0.132 EU m), PM10 (51.9 μg m), PM2.5 (22.6 μg m), and O3 (0.018 ppm). The endotoxin levels were significantly higher in fall and winter than in summer. The levels of PM10 and PM2.5 were significantly higher, and the level of O3 was by far its highest, in spring. Negative correlations were found between the endotoxin and O3 levels (r = –0.491) and between the endotoxin levels and temperature (r = –0.302). The PM10 levels were also negatively associated with the O3 levels and temperature but positively associated with the PM2.5 levels. Given the negative relationship between airborne endotoxins and O3 determined here, further studies with larger sample sizes are needed to identify the responsible mechanisms.


INTRODUCTION
A variety of air pollutants are legally required to be monitored in South Korea, including particulate matter with a diameter less than 10 and 2.5 µm (PM 10 and PM 2.5 , respectively), carbon monoxide, nitrogen dioxide, sulphur dioxide, and ozone (Ministry of Environment of Korea, 2017).However, due to a legal oversight, biological agents such as airborne endotoxins are not monitored in outdoor environments.Endotoxins such as lipopolysaccharides (LPSs) are ubiquitous in the environment and are an important structural component of the outer membranes of gramnegative bacteria (Beutler and Rietschel, 2003).Exposure to endotoxins has been found to cause and exacerbate asthma and wheezing in both children and adults (Abbing-Karahagopian et al., 2012) and has also been linked to lung function impairment (Liebers et al., 2008) and the pathogenesis of pulmonary diseases (Loh et al., 2006).In addition, a recent study found that endotoxin exposure can dramatically alter the body's white blood cell count, leading to disorders in immune function (Shang et al., 2016).
The health effects of PM 10 are predominantly respiratory and cardiovascular, with impacts ranging from functional changes (e.g., reduced lung function) and impaired activities (e.g., school absenteeism, days off work) to reduced life expectancy and ultimately death (Kuschel et al., 2012).Ambient PM 2.5 was the fifth-ranking global mortality risk factor in 2015, with exposure causing 4.2 million deaths (95% uncertainty interval, 3.7-4.8 million people) (Cohen et al., 2017).A study of 500,000 adults in the urban United States reported that overall mortality, mortality of cardiopulmonary diseases, and lung cancer increased by 4%, 6%, and 8%, respectively, for every 10 µg m -3 PM 2.5 increase, after ruling out smoking, diet, drinking, occupational, and other risk factors (Pope et al., 2002).PM 2.5 is also known to have neurotoxic effects, as these particles can enter human circulatory systems and affect various organs (Genc et al., 2012), in addition to coming into contact with the brain through the nasal olfactory mucosa (Garcia et al., 2015).Exposure to O 3 causes respiratory symptoms, increases susceptibility to pulmonary infections, and even increases the risk of mortality in those with underlying cardiorespiratory conditions (Turner et al., 2016).Moreover, endotoxin inactivation in the presence of O 3 becomes more efficient with increasing exposure time (Rezaee et al., 2008).
Past research has evaluated airborne endotoxins, PM, and O 3 in outdoor environments, such as ambient endotoxins and PM 10 in Chitwan, Nepal (Mahapatra et al., 2018); ambient concentrations of PM 10 and PM 2.5 in Palermo, Italy (Dongarra et al., 2010); spatio-temporal variations in ambient PM 10 and PM 2.5 concentrations in Beijing (Jie et al., 2016); exposure to outdoor PM 10 , PM 2.5 , and O 3 in Singapore (Gall et al., 2015); and ambient concentrations of O 3 under the influence of PM 2.5 , NO 2 , and SO 2 in Zhejiang, China (Chen et al., 2017).However, no research has evaluated the relationships between airborne endotoxins and PM 10 , PM 2.5 , and O 3 , although doing so would improve scientific understanding of these pollutants' airborne levels and distributions while collecting important background data for comparison between different countries.
Therefore, in this study we measured the ambient levels of airborne endotoxins atop two buildings in urban South Korea for one year and analyzed them with reference to PM 10 , PM 2.5 , and O 3 levels collected from Airkorea (www.airkorea.or.kr) to determine the relationship between these substances and the potential influence of environmental factors such as temperature and relative humidity.

Study Setting
We collected endotoxin samples from two buildings in Ilsan, Goyang-si, Gyeonggi-do, near Seoul (Fig. 1), in the spring, summer, autumn, and winter (from March 2016 to February 2017).Sampling Point A was located on the roof of a 12-story building, while Sampling Point B was located at the top of the highest apartment in a 19-story building.These buildings were selected to determine the differences between cities with high-traffic roads (Sampling Point A) and residential areas without high-traffic roads (Sampling Point B).Air samples were collected from 100-150 cm above floor level for about three days per month at both locations (81 total samples).

Endotoxin Sampling and Analysis
During endotoxin sampling, temperature and relative humidity (RH) were recorded at each location using a Unis digital thermometer (YTH-104 series, Unis Inc., Korea).Samples were collected onto glass fiber filters (37 mm diameter; SKC Inc., USA) preloaded in a three-piece clear plastic cassette using an air sampler (17G9 GilAir Sampler, Sensidyne, Inc., USA) at a flowrate of 2.0 L min -1 (± 5%) for an average of 6 h.One field blank was collected on each sampling day and analyzed by kinetic-turbidimetric Limulus Amebocyte Lysate (LAL) assay (Associations of Cape Cod, Inc., USA) with no contamination.Precautions were taken to avoid breathing on, touching, or otherwise exposing the sampling containers to human contamination while sampling airborne endotoxins, including the use of gloves while connecting or disconnecting the cassette and the pump.After sampling, a protective covering back (cap) was placed on the cassette's inlet and outlet, and the entire cassette was wrapped in its original packing and sealed with tape.
The samples were stored at 4 ± 2°C, sent to an analytical laboratory within a week of sampling, and analyzed immediately upon arrival.Detection and quantification of endotoxin levels were conducted by kinetic-turbidimetric LAL assay.The entire endotoxin extraction procedure was conducted at room temperature (25 ± 2°C).An extraction volume of 15 mL of pyrogen-free water was added to a test tube, which was then capped and sonicated at a minimum peak frequency of 48 kHz for 1 h.After that, samples were centrifuged at 1000 g for 15 min, and the supernatant was transferred to a pyrogen-free test tube.100 µL of each sample was distributed into a pyrogen-free 96-well microplate and incubated at 37°C for 10 min in an automated micro-plate reader (ELx808, BioTek Instruments, USA).

Fig. 1. Location of Sampling Points
A and B on building rooftops in Goyang-si Gyeonggi-do, South Korea.100 µL of LAL reagent was added to each well and analyzed in duplicate at 340 nm using Win KQCL Software (BioWhittaker, Cambrex Co., USA).The Escherichia coli O55:B5 control standard endotoxin (Lonza, USA) was utilized to draw a standard curve ranging from 0.005 to 50 endotoxin unit mL -1 .Only calibration curves greater than or equal to 0.98 were accepted for further analysis.Positive product control (PPC) recoveries within 50-200% and coefficients of variation (CV) less than 10% were considered valid.The endotoxin levels were expressed as endotoxin units per cubic meter of air (EU m -3 ).The assay limit of detection (LOD) was 0.01 EU mL -1 extract.Values below the LOD were assigned a value of LOD/√2 (Hornung and Reed, 1990).

PM 10 , PM 2.5 , and O 3 Data
The ambient PM 10 , PM 2.5 , and O 3 levels were obtained from publicly available Airkorea data (www.airkorea.or.kr), a program of the government's National Ambient air quality Monitoring Information System (NAMIS), and compared with the sampled endotoxin levels (Airkorea, 2017).The same dates used for endotoxin sampling were used for outdoor PM 10 , PM 2.5 , and O 3 data along with the same 6 h sampling time, and the outdoor concentrations were obtained at locations within 1 km of Sampling Point A and 2.6 km of Sampling Point B.

Statistical Analyses
Statistical analyses were conducted using SAS software, version 9.3 (SAS Institute, Inc., USA).A nonparametric analysis was performed since the endotoxin and PM levels were not distributed normally or log-normally according to a Shapiro-Wilk test.The relationships between the endotoxin, PM 10 , PM 2.5 , and O 3 level distributions and the recorded ambient temperature and RH were analyzed using descriptive statistics.Kruskal-Wallis tests were performed to determine the differences between the endotoxin, PM 10 , PM 2.5 , and O 3 levels and the seasons, including between Sampling Points A and B. Mann-Whitney tests with Bonferroni adjustments were also carried out to determine which seasons were significantly different.In addition, Spearman's correlation analyses were employed to examine the associations between the endotoxin, O 3 , PM 10 , and PM 2.5 levels and temperature and RH.

RESULTS
Endotoxin levels ranged from 0.007 to 1.681 EU m -3 with a geometric mean (GM) of 0.132 EU m -3 , PM 10 levels ranged from 23.0 to 166.0 µg m -3 with a GM of 51.9 µg m -3 , PM 2.5 levels ranged from 4.0 to 92.0 µg m -3 with a GM of 22.6 µg m -3 , and O 3 levels ranged from 0.003 to 0.059 ppm with a GM of 0.018 ppm (Table 1).Although endotoxin concentrations were higher at Sampling Point A (average GM of 0.147 EU m -3 ) than at Sampling Point B (average GM of 0.115 EU m -3 ), there was no significant difference (p > 0.05) between the two sampling points.
At the monthly scale, endotoxin levels were highest in October and lowest in April, PM 10 and PM 2.5 levels were highest in March and lowest in September, and O 3 levels were highest in June and lowest in January (Table 2).To evaluate seasonal variations in these pollutants, we grouped the monthly levels by season: spring (March-May), summer (June-August), fall (September-October), and winter (November-February) (Fig. 2).Endotoxins were highest in fall and winter, followed by summer and spring, with significant differences between fall and spring (p = 0.0003) and between winter and spring (p = 0.0008).PM 10 levels were highest in spring and winter and lowest in fall (p = 0.0091 between spring and fall, p = 0.0037 between winter and fall).PM 2.5 levels were highest in winter and lowest in fall (p = 0.0027 between fall and winter).O 3 levels were highest in summer and lowest in winter (p < 0.0001 between summer and winter).
Correlation analysis between endotoxin levels and PM 10 , PM 2.5 , O 3 , temperature, and RH showed a negative association between endotoxins and O 3 (r = -0.491)and between endotoxins and temperature (r = -0.302); the remaining factors were not clearly correlated with endotoxins (Table 3).PM 10 was negatively associated with O 3 and temperature.

DISCUSSION
This study analyzed the distribution of ambient airborne endotoxin levels for a year at the top of two buildings in the Ilsan area of South Korea and assessed the influence of environmental factors on endotoxin levels.Endotoxin levels ranged from 0.007 to 1.681 EU m -3 (GM of 0.132 EU m -3 ).In comparison, airborne endotoxin levels in outdoor urban areas of Stockholm, Sweden, ranged from 0.020 to 0.107 EU m -3 (GM of 0.05 EU m -3 ) (Nilsson et al., 2011), while areas with intensive livestock production in the Netherlands recorded endotoxin levels of 2.0-2.9EU m -3 and 0.46-0.66EU m -3 in residential gardens at least 500 m from the nearest farm (Schulze et al., 2006;Rooij et al., 2010).These variations in reported endotoxin levels may be due to differences in sampling and extraction methods, as well  , 2010).However, studies have shown that endotoxins affect health even at much lower concentrations (Ryan et al., 2009;Bennett et al., 2012).Rabinovitch et al. (2005) reported an increase in the severity of asthma in children exposed to endotoxin levels of 0.08 EU m -3 .Our results showed that endotoxin concentrations were higher in a high-traffic urban setting (Sampling Point A) than in a low-traffic residential area (Sampling Point B), similar to a previous study reporting that endotoxin concentrations on congested streets (median = 4.4 EU m -3 ) were higher than in residential areas (median = 0.33 EU m -3 ) (Madsen, 2006).Exposure to traffic-related particles is associated with childhood respiratory problems, and a synergistic relationship exists between co-exposure to traffic-related particles and endotoxins with regard to persistent respiratory problems during infancy through 3 years of age (Ryan et al., 2009).
The average ambient GM (GSD) PM 10 and PM 2.5 levels in this study were 51.9 (1.5) and 22.6 (1.9) µg m -3 , respectively-less than the 100 and 50 µg m -3 from Airkorea (2017) and the 50 and 25 µg m -3 from the WHO (2016).However, these average GM PM 10 levels were higher than those reported for areas with livestock farms (19.8-22.3µg m -3 ; Rooij et al., 2017), and the average PM 2.5 levels were higher than reported for urban rooftops near busy roads in Brisbane, Australia (8.0-19.0µg m -3 ; Quang et al., 2012).Other PM reports include those from atop a six-story building in Beijing (GM of PM 2.5 levels ranging from 6.4 to 463.5 µg m -3 and averaging 61.7 µg m -3 ; Guan et al., 2014), from 38 of China's largest cities (daily mean PM 10 level of 92.9 µg m -3 ; Yin et al., 2017), from a traffic site in Algeria (GMs of 105.2 µg m -3 for PM 10 and 57.8 µg m -3 for PM 2.5 ; Terrouche et al., 2016), and from two traffic sites in Lahore, Pakistan (GMs of 286 and 365 µg m -3 for PM 10 and 222 and 302 µg m -3 for PM 2.5 ; Ali et al., 2015), all of which were higher than those measured in this study.These differences between PM levels in different cities are likely caused by measurement differences, which can vary widely due to sampling procedures and equipment, even for the same pollutant in the same location (Amaral et al., 2015).
Ambient O 3 levels in this study ranged from 0.003 to 0.059 ppm (overall GM of 0.018 ppm), less than the 8 hour levels measured by Airkorea (2017).Other reports of outdoor O 3 levels include those from British Columbia (0.028 ppm) and southern Ontario (0.037 ppm) in 2014 (ECCC, 2016).A European study noted that the concentration of surface ozone had increased from an estimated preindustrial value of 0.01 ppm to 0.03-0.05ppm (Pritchard and Amthor, 2005).According to the United States Environmental Protection Agency (EPA), controlled studies of prolonged human ozone exposure at levels below 0.08 ppm showed respiratory effects, changes in lung function, and increased airway responsiveness; animal toxicology studies have provided additional evidence of such effects (NAAQSO, 2008).Previous studies have shown that airborne endotoxin levels were higher in spring and summer than in fall and winter (Carty et al., 2003;Kallawicha et al., 2015).In contrast, our results found the highest levels in fall (September-October) and the lowest levels in spring (March-May), with significant seasonal differences (Fig. 2).Differences in meteorological factors (e.g., rain, wind, sunlight hours, temperature, and humidity) likely explain this seasonal variability in endotoxin levels (Carty et al., 2003).Temperature and RH were found to be the most influential, with the highest endotoxin concentrations recorded during warm periods and moderate RH (35-75%) in ambient environments (Allen et al., 2011). Traversi (2010) observed that temperature plays a predominant role in endotoxin modulation within the environment and showed that temperature has a negative correlation with endotoxin levels.On the other hand, Mahapatra et al. (2017) reported that endotoxin levels showed a weak positive correlation with temperature (r = 0.34).This discrepancy might be due to temperature variations because Mahapatra et al. (2017) mentioned that a temperature variation of 22-28°C was not sufficient to affect endotoxin levels; Su et al. (2001) also observed no significant correlation with a small change in temperature.Although we found no significant correlation between RH and endotoxin levels, RH had a weak negative correlation with endotoxin in a similar pattern as reported by Mahapatra et al. (2017) in which a weak positive correlation (r = 0.2) between endotoxin concentration and RH from 38-60% was observed, suggesting that moderate RH aided bacterial growth but higher RH levels (60-90%) reduced endotoxin levels.
We found no clear correlations between endotoxin and PM 10 or PM 2.5 levels; this was consistent with Rooij (2017), who suggested that lower correlations related to inherent variability in endotoxin levels were due to the influence of sampling and analytical variability.However, PM 10 levels were significantly negatively correlated with O 3 levels (p < 0.05), similar to results given in Chen et al. (2017), because high particle concentrations in ambient air could make the atmosphere cooler and reflect sunlight above the ground, resisting the formation of O 3 (Moss et al., 2010).
O 3 levels varied significantly between seasons and were much higher during summer than winter (Fig. 2).This result was consistent with Chen et al. (2017) and other studies, suggesting that severe O 3 pollution can benefit from higher sunlight intensity and temperature (Stathopoulou et al., 2008).O 3 was positively correlated with temperature (p < 0.001), suggesting that higher temperature is beneficial to O 3 formation in ambient air because higher temperature can accelerate reaction among precursors and their intermediate products such as free radicals; this result was consistent with the principle reported by Coates et al. (2016).High RH indicates that the atmosphere contains more water molecules, which play a vital role in the formation of O 3 (Calvert et al., 2015), as demonstrated by the significantly positive correlation between O 3 and RH.
Our study is the first to show a correlation between endotoxin levels and chemical pollutants such as PM 10 , PM 2.5 , and O 3 in ambient environments, but some limitations should be considered.First, our outdoor measurements of endotoxins, PM, and O 3 did not necessarily accurately reflect the correlation between endotoxin levels and outcomes as they were not conducted simultaneously.Second, the sources of PM 10 and PM 2.5 were not specifically identified; further studies would need to measure PM more directly to better understand the components of atmospheric environments.Third, the short daily sampling period (6 h) may have introduced some variation among measurements, resulting in poorer representation and weaker consistency between the concentrations for the entire day.Finally, the limited sample size may not have been representative of the ambient levels of endotoxins in comparison with pollutant levels, resulting in possible biases.Despite these limitations, this study was conducted for a substantial period of time using standard air sampling methods, which increases the validity of the measurement comparisons.Moreover, identifying levels of ambient endotoxins in relation to PM 10 , PM 2.5 , and O 3 is a new step toward a better understanding of their interrelated dynamics across a large metropolitan area in South Korea.

CONCLUSION
Airborne endotoxin levels were measured from the tops of two buildings in the city of Goyang in Gyeonggi Province, South Korea, and compared with PM 10 , PM 2.5 , and O 3 data recorded in nearby locations.The endotoxin levels were significantly higher in fall and winter than in summer.The levels of PM 10 and PM 2.5 were significantly higher in spring than the other seasons, and spring also had the highest O 3 levels.The endotoxin and O 3 levels were found to be negatively correlated (r = -0.491),as were the endotoxin and temperature levels (r = -0.302).The PM 10 levels were also negatively associated with the O 3 levels and temperature.Further studies, especially with a larger sample size, are needed to identify the prevalent mechanism causing these relationships.

Table 1 .
Overall levels of chemical pollutants and climate (temperature and relative humidity) in ambient environments.

Table 2 .
Characteristics of endotoxin levels (measured in this study) and PM