Toxic evaluation of organic extracts from airborne particulate matter in Puerto Rico.

In recent years, several hypotheses have emerged to explain the toxicologic activity of particulate matter. Organic compounds, ultrafine particles, biologic components, and transition metals are some of the constituents that reportedly exert some type of adverse effect on human health. A considerable fraction of the urban particulate matter consists of carbon compounds, which originate mostly from anthropogenic sources. The toxicity of organic fractions from particulate matter have been mainly evaluated by considering their mutagenic activity. This research expands on the toxicologic profile of organic compounds adsorbed to particulate matter, specifically in Puerto Rico, by using the cytotoxic neutral red bioassay (NRB). The NRB uses normal human epidermal keratinocytes or other types of cells to measure the effect on cell viability when exposed to organic compounds associated to the particles in the air. We validated the NRB for particulate matter by using a standard reference material (SRM 1649). We used the NRB to determine toxicologic differences of extracts between an urban industrialized site with anthropogenic activity versus a coastal region with less human activity. The cytotoxicity associated with organic compounds in particulate matter collected at the urban industrialized site was detected in both the particulate matter (3/4) 10 microm in aerodynamic diameter (PM(10)) and particulate matter (3/4) 100 microm in aerodynamic diameter (PM(100)). Greater toxic effects were observed in PM(10) extracts than in PM(100) extracts, but PM(10) toxic effects were not significantly different from those in PM(100). The extracts from the industrialized site were more cytotoxic than the extracts from coastal reference site, although in the summer, extracts from both sites were significantly cytotoxic to normal human epidermal keratinocytes. In addition, the nonpolar extracts of both PM(10) and PM(100) exerted the greatest cytotoxicity, followed by the polar, and, finally, the moderately polar extract. This study demonstrates that extracts from the Guaynabo industrialized site were more toxic than similar extracts obtained from a reference coastal site in Fajardo, Puerto Rico. ImagesFigure 1Figure 2Figure 3Figure 4

have been mainly evaluated by considering their mutagenic activity. This research expands on the toxicologic profile of organic compounds adsorbed to particulate matter, specifically in Puerto Rico, by using the cytotoxic neutral red bioassay (NRB). The NRB uses normal hun epidermal keratinocytes or other ypes ofcels to measure the effect on cell viability when eTosed to organic compounds associated to the partides in the air. We validated the NRB for particulate matter by using a standard reference material (SRM 1649). We used the NRB to detere toxicologic differences of extracts between an urban industrialized site with anthropogenic activity versus a coastal region with less human activity. The cytotoxicity associated with organic compounds in particulate matter collected at the urban industrialized site was detected in both the partculate matter c 10 pm in aerodynamic diameter (PM10) and particulate matter < 100 pm in aerodynamic diameter (PMl,,). Greater toxic effect were observed in PM10 extacts than in PMIOO extracts, but PMIO toxic effects were not significantly different from those in PM 1 . The extracts from the industrialized site were more cytotoxic than the extacts from coastal reference site, although in the summer, extrcts from both sites were significantly ctotoxic to normal human epidermal keratinocytes. In addition, the nonpolr extracts of both PM10 and PM1i exerted the greatest cytotoxicity, followed by the polar, and, finally, the moderately polar extract. This study demonstrates that cxtracts from the Guaynabo indalized site were more toxic than similar etracts obtained firom a reference coastal site in Fajardo, Puerto Rico. Key words air pollution, bioassay, cytotoxicity, hunan cells, organic extmacts, particulate matter, PM1O, PMIWO, SRM 1649, toxicity.
Environ Health Peect 108: 635-640 (2000). [Online 1 June 2000] hb:/Xehpnetl.niehs.ni ogovldoci2000/1 08p635-6Orqyes/abstract.htmI Air pollution in Latin American and Caribbean countries has been a major environmental health issue for people living in metropolitan areas. These health issues arise as a consequence of urban development, population growth, and industrialization. Epidemiologic studies suggesting airborne particulate matter as one of the principal causes of respiratory illnesses (1,X, cardiovascular disease mortality (3,4), acute bronchitis (5,6), and asthma attacks (7)(8)(9) have been controversial (10). The differences center on the meta-analysis performed on data from U.S. cities, London, and Athens, Greece. However, other meta-analytical studies recently showed that a relationship between air pollutants and health does exist after adjustment for numerous variables (11)(12)(13).
Several hypotheses from these epidemiologic data attribute health effects to active toxicologic components of particulate matter. These hypotheses have focused on identifying components of particulate matter responsible for the observed health problems. The factors suggested to cause these health effects are the associated organic components (14-17); ultrafine particle constituents (18); the presence of biologic material (19); and the transition metal content of particulate matter (20).
The organic fraction of urban airborne particulate matter has been evaluated with regard to the mutagenic and genotoxic activity of the compounds found in the various extracts. Results from this research indicate that semipolar and polar fractions exhibit the greatest toxicologic activity using the Ames test (15)(16)(17), sister chromatid exchange assay (17), and a human cell mutagenic assay (14). Organic compounds such as polyaromatic hydrocarbons (PAHs), substituted aromatic hydrocarbons, and heterocyclic aromatic compounds have been targeted for these assays. A more recent study showed evidence that biologically effective amounts of PAH compounds associated to particulate matter were transferred to the intracellular environment and elicited the activation of various genes (21). Previous studies have demonstrated with great detail the activation of the aryl hydrocarbon receptor by tetrachlorodibenzo-p-dioxin (22). The activation of specific genes by organic pollutants is an area of great interest and future development.
Ultrafine particles < 0.10 pm in diameter could cause an acute inflammatory reaction as measured by lung lavage parameters in rats (18). Researchers have demonstrated an increase in pulmonary toxicity with exposure to ultrafine particles and a correlation with the surface area of the retained particles rather than their mass, volume, or numbers.
Another possible source of toxicity by particulate matter is biologic components such as aeroallergens, which have been associated with asthma-related mortality (19). An epidemiologic study showed that during [1985][1986][1987][1988][1989] in Chicago, the asthma-related deaths were associated with high mold spore counts. The death rate was 2.16 times higher than the death rates related to asthma at levels < 1,000 spores/mi3. Finally, the metal portion of particulate matter, specifically the elements iron, nickel, and vanadium, causes pulmonary toxicity (20). Lung injury interactions were observed among these metals when tested as mixtures. The toxicity of metals can also be potentiated and enhanced by other factors such as organic material that can induce inflammation in the lungs of rodents (23). All of the aforementioned studies lead to the unified hypothesis that the toxicity of airborne particulate matter is multifactorial and is influenced by the nature and origin ofthe partides.
Epidemiologic data gathered from an urban area in Puerto Rico (Guaynabo) revealed that approximately 30% of children 13-14 years of age and > 40% of children 6-7 years of age suffer from asthma (24). Furthermore, the U.S. Centers for Disease Control and the Puerto Rico Health Department reported that almost 50% of the children from this area have learning disabilities (25). In addition to this finding, they report that 41% suffer from one or several medical conditions (as compared to 23% in the rest of the island), and that there are more types and cases of cancer in this community than in other regions of Puerto Rico. The incidence of respiratory diseases reported in this area and the relatively high concentration of airborne particulate matter are higher than other regions of Puerto Rico, suggesting a direct link between these health conditions and airborne particulate matter at Guaynabo. The problem is exacerbated even further by susceptibility factors; Puerto Rican children are more affected by asthma than are other Latino children (26). This difference is attributed to a combination of unique genetic and sociocultural factors, such as smaller airway size, more severe inflammatory reactions, and lower birth weight than that of Cuban-American or Mexican-American children.
Most of the toxicity tests performed on urban airborne particulate matter have been based mainly on traditional mutagenic tests using the Salmonellalmicrosome bioassay (27)(28)(29)(30)(31)(32)(33)(34). Although this is a useful and informative test, it only measures the mutagenic potential of compounds tested and does not consider direct toxicologic effects to human cells or systems. We used the neutral red uptake bioassay (NRUB) with primary cell cultures to determine cellular toxicity (lysosomal activity) caused by the exposure of organic components of airborne particulate matter from an urban polluted area and a coastal reference environment (Fajardo). Primary cultures of cells obtained from normal human tissues maintain many of the complex functions and biochemical systems that are active in vivo. The in vitro NRUB has been used to determine the cytotoxic effect on a myriad of chemical moieties such as organic (33)(34)(35)(36)(37)(38) and inorganic chemicals (39), metals (3639-42), pesticides (34), polycyclic aromatic hydrocarbons (32'), food additives (45), chemotherapeutic agents (44,45), drug activity (45,46), and structure-activity correlation of toxins (427). Results from this assay have been compared to those of other known toxicologic assays, and the assay has been used with many cell lines including human hepatoma and melanoma tumor cells (45,47), normal human melanocytes (45,4?), normal human epithermal keratinocytes (35,43,45,47), human fibroblast cells (43,45,47), human kidney cells, human endothelial cells, and human renal carcinoma cells (4?), among others.
The results presented here encompass the toxicity of extracts obtained from two sites (Guaynabo and Fajardo) that contain low anthropogenic inputs of airborne particulate matter. This work is the first step to evaluate the toxicity of organic constituents from airborne particulate matter at two specific sites, and represents the first study of this nature in Puerto Rico. It approaches the characterization of toxicity of organic extract constituents based on differences in polarity using a cytotoxicity assay. The cytotoxicity of the extracts was seasonally dependent; summer components were the most toxic. Through this approach we conclude that extracts from the Guaynabo site are more toxic to human cells than those obtained from the Fajardo site.

Materials and Methods
Sampling procedure. Airborne particulate matter samples were collected at two sites, an urban industrialized area (Guaynabo), which is approximately 2 miles south of the entrance to the San Juan Harbor, and a reference site (Fajardo) on the east coast of Puerto Rico ( Figure 1). Fajardo is on the coast and receives trade winds from the Atlantic Ocean; the only expected input is from boating activities, which occur mainly during the weekends. We simultaneously collected total suspended particulate matter [particulate matter < 100 pm in aerodynamic diameter (PM100)] and inhalant particulate matter < 10 pm in aerodynamic diameter (PM10)] at these sites during weekdays between May 1996 and September 1997. Sampling was performed for 48 hr on glass fiber filters using high-volume samplers and size-selective inlets; the samplers were placed 10 ft above the ground. All of the air filters were handled with latex gloves, immediately transported on dry ice, and stored in the laboratory at -200C until processed. Sampling rates were on the average of 1.48 m3/min for PM0OO and 1.05 m3/min for PM1O. We determined the amount of particulate matter gravimetrically by weighing the filters before and after sampling. Filters obtained from both stations and an urban dust standard reference material [SRM 1649; National Institute of Standards and Technology (NIST), Gaithersburg, MD] was extracted using a Soxhlet apparatus with solvents of different polarity (HPLC grade; Fisher Scientific, Fair Lawn, NJ). One-fourth of each filter was sequentially extracted for 24 hr with 175 mL hexane (nonpolar), dichloromethane (DCM) (moderately polar), and acetone (polar). Each extract was reduced to approximately 10 mL using a rotatory evaporator under vacuum and then dried completely under a gentle stream of nitrogen. The extracts obtained after this process were weighed and dissolved in 100 pL DMSO (ACS reagent; Sigma, St. Louis, MO). Several dilutions of the dissolved DMSO extracts were prepared with cell culture medium for NRUB cytotoxicity screening. The DMSO in the cell culture medium was always kept below 1% to avoid cytotoxicity of the carrier solvent. Blank samples prepared by extracting unused filters showed no NRUB toxicity.
NRUB. The selection of human primary epidermal keratinocytes for this study was based on two facts. First, these cells possess xenobiotic detoxification systems that would also contribute to the toxicity of metabolites transformed by biologic systems (48); hence, the use of these cells integrates the effects of both the parent xenobiotics and their metabolic products. Second, the amount of information that exists on a great number of compounds using keratinocytes is extensive and allows for comparisons with particulate matter extracts. We obtained human epidermal keratinocytes and all solutions used in the NRUB from Clonetics Corporation (San Diego, CA). Cells were treated as follows: keratinocyte growth medium (KGM) was supplemented with epidermal growth factor, insulin, hydrocortisone, antimicrobial agents, and bovine pituitary extract. Cells were subcultured in KGM at 37°C and 5% CO2. We added fresh growth media on alternate days. At culture growth between 50 and 80% of confluence, cells were trypsinized, counted, and seeded onto 96well tissue culture microtiter plates at approximately 3,000 cells/well. After 3 days, the medium was replaced with supplemented KGM media containing different concentrations of extracts. Cells were exposed to compounds in the extracts for periods of 36-48 hr. Each batch was treated with Triton X (a positive control), which is a known toxicant [concentration that kills 50% of the cell population (NR50) between 3 and 32 pg/mL]. We also examined negative controls (untreated cells) on each plate. After exposure time, the extracts from VOLUME 108 1 NUMBER 7 July 2000 * Environmental Health Perspectives particulate matter were decanted and a neutral red dye solution of 100 pg/mL was added to all wells, with the exception of the blanks. Cells were incubated with the dye solution for 3 hr at 37°C and 5% CO2. Viable cells incorporate the dye, whereas damaged cells do not. After the incubation period, we removed the neutral red dye solution and fixed the cells. A solvent solution containing acetic acid and ethanol was added to the cells to allow the release of the neutral red dye. We determined the absorbance of the neutral red dye in each well spectrophotometrically at 540 nm, using a microplate reader. The amount of viable cells in each well was proportional to the absorbance readings. The absorbance of untreated cells was used to determine the percent of viability at each dose analyzed. The percent of viability of cells was obtained by dividing the average absorbance ofwells at each extract dose by the average absorbance of the negative controls. The NR50 concentration value was obtained from a percent viability curve. Quantitation was determined as previously described (47). We used four replicate wells per dose of extract for determining the NR50 concentration.
Statistics. We used unpaired t-test or Mann-Whitney and analysis of variance tests to evaluate statistical differences between concentrations among sites. We evaluated differences between seasons and sampling sites. The statistical analyses were performed using the StatView computer software package (version 5.0; SAS Institute, Inc., Cary, NC). We established the probability value ofp < 0.05 as the level for statistical significance.

Results and Discussion
Particle characterization per site. The monthly average concentration of PM1O (in micrograms per cubic meter) collected at Guaynabo and Fajardo during the years 1996-1997 is shown in Figure 2. By comparing the differences between the average monthly concentrations throughout the year for both sites, we determined that the amount of PM1O collected at Guaynabo is 34% higher than that obtained at Fajardo. The PM1O concentrations obtained for Guaynabo and Fajardo using our sampling procedures were not significantly different from those obtained by the Puerto Rico Environmental Quality Board (EQB) (49). Therefore, the amount of PMIO that was used in our assays (collected during 48 hr) was within the expected range of material normally reported by the local environmental agency.
Because PM1OO regulations were no longer required by the state regulatory agency after 1993, we combined the available EQB Guaynabo data for 1993-1995 to generate an average PM1OO profile with which to compare our results. All of the material obtained during our sampling period (fall, winter, spring, and summer) fluctuated between the average monthly concentration obtained from the 1993-1995 PMOO-generated profile at the Guaynabo site. Statistical analyses showed no significant differences among the PM1OO concentration at Guaynabo during the years of 1993-1995 and our PM1OO concentration at the same site (1 996-1997).
A comparison of seasonal variations clearly indicates that the highest amount of airborne particulate matter (PM1O and PM100) is evident during the summer season ( Figure 2). The average concentration of particulate matter obtained during the summer was higher than that obtained during other seasons. This finding could be partially explained by the increases due to normal chemical and photochemical oxidation processes occurring during the warm season (50). Although our data identify the summer as containing the highest levels of particulate matter, the PMIO concentrations at both sites did not exceed the maximum daily permissible limit established by the EQB (50 pg/m3) (51). However, the levels of PM1OO for Guaynabo exceeded the maximum concentration of 150 pg/m3 (secondary standard established to protect public welfare). The second highest concentration of particulate matter among the seasons was observed during the fall, followed by winter and spring. Thus, the trend observed in this study for levels of particulate matter for both PM1O and PM1OO in Puerto Rico followed the order: summer > fall > winter = spring ( Figure 3).
When these two sites in Puerto Rico are compared between themselves, we observe similar PM1O values between the urban and coastal site. Furthermore, PM1OO concentrations from Fajardo were always lower than those from Guaynabo. Hence, the major profile of particles at the urban site is between 10 and 100 pm (Figure 3). We observed characteristic patterns, in terms of the PMIO/PMIOO ratio, at both sites. Fajardo Environmental Health Perspectives * VOLUME 108 1 NUMBER 7 July 2000 different sources. The particulate matter at the reference site (predominantly PM10) is believed to contain organic compounds of marine origin, some anthropogenic compounds, and, occasionally, Saharan dust (52,53) and tridecane. Their concentrations range between 1 and 100 ng/m3. Toxicologic characterization. Urban dust standard reference material (SRM 1649) obtained from the NIST contains a series of organic compounds including PAHs and inorganic elements (heavy metals). The mutagenic activities (Ames test) of the moderately polar (DCM) and polar extracts (acetone) from this material are relatively similar (20-23 revertants/jg) (28). Furthermore, the mutagenic activity of the reference material was 3-4 times higher than that of airborne partides collected from a New Jersey site (27). These data provide evidence that moderately polar and polar fractions from urban dust (SRM 1649) contain mutagenic organic compounds. However, the cytotoxicity of these fractions has not been reported for human cells. Thus, it is important to establish the cytotoxicity of this material using human cells to establish a reference point to compare its relative toxicity, because other deleterious and toxic effects can develop through mechanisms not associated with mutagenicity. Cytotoxicity results of the urban dust SRM 1649 using the NRUB are illustrated in Figure 4.
The SRM 1649 extracts showed the greatest effect in the nonpolar fraction (hexane; NR50 = 5.4 ± 0.2 pg/mL), followed by an equal toxicity in the moderately polar (DCM; NR50 = 40 ± 2 pg/mL) and polar (acetone; NR50 = 43 ± 10 pg/mL) extracts. The similarity in toxicity between DCM and acetone extracts follows the same pattern observed with the mutagenic tests previously reported for this material (28). The increase in toxicity of the nonpolar extract can be explained by the formation of toxic metabolites of organic compounds inherent in this reference material. This is supported by the fact that human cells (primary keratinocytes) contain detoxification enzymes such as cytochrome P450s that are responsible for the metabolism of many foreign compounds (55). A previous study performed by Clonetics (56) using 52 petrochemical agents of different nature (lube oil additives, gasoline additives, polybutylene, solventrefined paraffinic petroleum oil, and various metal-working fluids) suggested a direct relationship between the Draize score (in vivo eye irritation assay) and the NR50 using the NRUB. The Clonetics study showed that NR50 values > 150 pg/mL were nontoxic. Using this threshold, all three SRM 1649 extracts can be considered highly cytotoxic. site were more cytotoxic than extracts collected at Fajardo during the same season. This is supported by the fact that many extracts collected from the Fajardo coastal site in the fall were nontoxic, with the exception of three fractions (   Fajardo and Guaynabo were cytotoxic (Table 1).
To further evaluate the relative toxicity of extract fractions by site and by season, we determined statistical differences between extract fractions per site and season. This analysis yielded an order of toxicity for each extract fraction (Table 2). Two major findings arise from this data. First, differences in the toxicity of extract fractions with respect to particle size were noticeable. An increase in toxicity was associated with increased polarity of extract constituents for PM1OO Conversely, extracts obtained from PMIO were more toxic as polarity decreased (hexane constituents are more toxic). This difference in toxicity could be explained by the type of chemical constituents partitioning into each solvent. For example, metals and polar organic compounds are expected to partition into more polar fractions, whereas other organics are associated to nonpolar fractions. The next major differences are those between sites. Extract fractions obtained from Guaynabo were toxic throughout the year; extracts from Fajardo were mainly toxic during the summer. Because the major toxicity of extracts was observed during the summer, we compared extract toxicity between sites during this season. Guaynabo had significantly higher toxicity in the nonpolar (hexane) and semipolar (DCM) fractions of PM10, whereas Fajardo exhibited significantly higher toxicity in the PM1o Acetone polar extract. The most significant toxic effect of PMIOO extracts between sites during the summer was observed in Guaynabo. This effect was observed with the polar extract fraction. Overall, comparable extract fractions between sites were more toxic at the Guaynabo site. The average NR50 for summer in the industrialized (urban) site was 31 ± 7 pg/mL; the overall average NR50 for Fajardo was 37 ± 11 pg/mL. This Guaynabo NR50 value is comparable to the NR50 reported for Triton X (3-32 pg/mL), a wellestablished toxic detergent used as a positive control with this assay. This NR50 toxicity value is also comparable to that obtained for extracts from our standard reference material (urban dust SRM 1649), which ranged from 5.4 to 43 pg/mL (Figure 4).
In summary, although the NRUB is an in vitro assay that uses primary human cells, only considers one end point, and was not evaluated simultaneously with other toxicologic assays, it provided significant data to establish marked differences in seasonal samples and urban versus rural airborne particulate extracts. The cytotoxicity of organic compounds extracted from particulate matter is higher in the nonpolar extract component. This was shown for both SRM 1649 and the particulate matter extract from the industrialized site, particularly in the PMIO.
Finally, we determined that the cytotoxicity was seasonal dependent; summer was the most toxic season at both sites studied. This work supports the hypothesis put forth by others that organic constituents adsorbed to particulate matter (particularly those of lower particle size) can contribute to the toxicity of airborne particulate matter. This work is the first report linking a regional and seasonal variation in an urban site as compared to a less-polluted site in Puerto Rico. We will attempt to isolate and identify strongly cytotoxic compounds from summer samples collected at Guaynabo-specifically from nonpolar and polar extracts. We also plan to evaluate heavy metal components in these extracts at Fajardo.