Comparative Toxicity of Size-Fractionated Airborne Particulate Matter Collected at Different Distances from an Urban Highway

Background Epidemiologic studies have reported an association between proximity to highway traffic and increased cardiopulmonary illnesses. Objectives We investigated the effect of size-fractionated particulate matter (PM), obtained at different distances from a highway, on acute cardiopulmonary toxicity in mice. Methods We collected PM for 2 weeks in July–August 2006 using a three-stage (ultrafine, < 0.1 μm; fine, 0.1–2.5 μm; coarse, 2.5–10 μm) high-volume impactor at distances of 20 m [near road (NR)] and 275 m [far road (FR)] from an interstate highway in Raleigh, North Carolina. Samples were extracted in methanol, dried, diluted in saline, and then analyzed for chemical constituents. Female CD-1 mice received either 25 or 100 μg of each size fraction via oropharyngeal aspiration. At 4 and 18 hr postexposure, mice were assessed for pulmonary responsiveness to inhaled methacholine, biomarkers of lung injury and inflammation; ex vivo cardiac pathophysiology was assessed at 18 hr only. Results Overall chemical composition between NR and FR PM was similar, although NR samples comprised larger amounts of PM, endotoxin, and certain metals than did the FR samples. Each PM size fraction showed differences in ratios of major chemical classes. Both NR and FR coarse PM produced significant pulmonary inflammation irrespective of distance, whereas both NR and FR ultrafine PM induced cardiac ischemia–reperfusion injury. Conclusions On a comparative mass basis, the coarse and ultrafine PM affected the lung and heart, respectively. We observed no significant differences in the overall toxicity end points and chemical makeup between the NR and FR PM. The results suggest that PM of different size-specific chemistry might be associated with different toxicologic mechanisms in cardiac and pulmonary tissues.


Research
During the last several decades, industrializa tion and urbanization have resulted in dra matic increases in vehicleassociated emissions. More than 50% of the total particulate matter (PM) emissions in urban areas are related to road traffic (Briggs et al. 1997). A number of studies have reported significant associa tions between traffic density or proximity to roads and various respiratory symptoms in children (Brunekreef et al. 1997;Edwards et al. 1994;Venn et al. 2001). A recent study from London reported that asthmatic indi viduals showed increased symptoms (lung dysfunction and inflammation) while walk ing around heavily trafficked Oxford Street compared with performing similar activities in Hyde Park (McCreanor et al. 2007). In addi tion, increased cardiovascular effects have also been reported with environmental exposure to vehiculartraffic-derived pollutants (Dockery et al. 2005;Peters et al. 2004).
Roadside aerosols are a complex mixture of different size particles that segregate into three fractions. Coarse PM [2.5-10 µm in aero dynamic diameter (PM 2.5-10 )] is produced by abrasion of automobile brakes and tires and dispersion of road dust, whereas fine (PM 2.5 ) and ultrafine (PM 0.1 ) materials are emitted directly from the tail pipe or form as second ary aerosols through atmospheric reactions. Vehicle emission factors have been estimated from dynamometer emission monitoring, as well as from locational studies of tunnels and nearroadside environments (Kuhn et al. 2005;Phuleria et al. 2006;Riddle et al. 2008). These studies have reported larger PM emis sions from diesel trucks compared with gaso line vehicles, with higher levels of direct vehicle emissions related to proximity to the roadside. Because the physicochemical characteristics of roadside airborne particles are also influ enced by factors such as location, climatic conditions, traffic volumes, type of fuel and engine, and operating conditions, the toxic potential of these mixtures can be quite diverse Gilmour et al. 2007;Kleinman et al. 2007;Seagrave et al. 2002;Singh et al. 2004). Direct exposure to com bustion emissions has resulted in a wide vari ety of cardio pulmonary effects in animals and human volunteers (reviewed by Hesterberg et al. 2009;Lewtas 2007;Sydbom et al. 2001). Furthermore, many of these pathophysiologic changes have also been noted after inhalation exposure to ambient PM (Brook et al. 2002;Gong et al. 2003;Gurgueira et al. 2002;Samet et al. 2009;Sun et al. 2005), sometimes in association with trafficrelated pollution. We designed the present study to characterize the chemical composition of sizefractionated airborne PM collected at either 20 or 275 m from an interstate highway and to determine the cardio pulmonary toxicity of the particles after oro pharyngeal aspiration in mice.

Materials and Methods
Sample collection. We collected two consecu tive weekly samples of airborne PM in three different size ranges (coarse, 2.5-10 µm; fine, 0.1-2.5 µm; ultrafine, < 0.1 µm) using ChemVol highvolume cascade impactors (model 2400, Rupprecht & Patashnick Co. Inc., Albany, NY) during the weeks of 26-31 July and 3-10 August 2006. Sampling sites were located approximately 20 m and 275 m from the nearest travel lane of a high way in Raleigh, North Carolina, and are referred to as the near road (NR) and far road (FR) sites, respectively. The sites were cho sen based on prevalent wind direction and distance from the highway and represented potentially high and low exposure to pollu tion emitted by traffic on the adjacent high way (Kuhn et al. 2005;Riddle et al. 2008). The vehicle fleet mix at the highway consisted mostly of lightduty, gasolinepowered motor vehicles. Detailed description of sampling site charac teristics have been published previously (Baldauf et al. 2008). The coarse and fine fractions of PM and the ultrafine PM were collected onto precleaned, preweighed poly urethane foam (PUF) and polypropylene fiber filters, respectively. The amount of material collected was determined gravimetrically. The PM was extracted for chemical analyses and toxicity testing.
PM extraction and recovery. For the coarse and fine PM fractions, PUF strips were transferred to sterile 50cc polypropylene tubes and submerged in 35 mL of analytical grade methanol. The tubes were placed in an ultrasonic water bath at < 25°C for 1 hr at 100% power, 25 kHz frequency, and nor mal ultrasonic mode (model TIH15, Elma Hans Schmidbauer GmbH & Co. KG, Singen, Germany). After sonication, a sterile Teflon rod was pressed against the PUF to help liberate material in the pores. Extracted material was decanted into preweighed sterile polypropylene 50cc tubes. For the ultrafine PM fractions, a poly propylene fiber filter was placed in a sterile 4L glass beaker with the PMcollected side downward and submerged in 150 mL methanol. The beaker was covered with Parafilm and sonicated for 1 hr under the same conditions as the PUF filters. The extracted material was decanted into a sterile 250mL polypropylene bottle. The beaker was rinsed with additional volumes of methanol and then combined with extracted material. Aliquots of all size fractions for each collec tion week were evaporated to dryness under nitrogen, and the tubes were reweighed. PM recovery efficiency was calculated as the dif ference between the pre and postweight of the tubes, divided by the PM mass collected on the weekly filters.
After determination of PM mass, the dried materials of each sample were dissolved in 100% methanol, followed by addition of 0.9% sterile saline, producing a PM suspension containing 0.5% methanol and 10 mg/mL PM. A control saline solution contained 0.5% methanol. The reconstituted PM suspensions and saline were sonicated, vortexed for 1 min, and analyzed for chemical constituents; 1mL aliquots of the material were stored at -80°C until toxicity testing. For the toxicologic study, we combined two weekly PM suspensions of the same size fraction and location to produce the same mass ratio of collected PM samples from the individual weeks.
Chemical analyses of PM. We analyzed NR and FR PM samples of each week for inorganic elements, ions, and carbon con tent. For elemental analysis, 2mg aliquots of PM suspension in saline used for the animal exposures were digested in 10% nitric acid to solubilize the metals (Wichers et al. 2006). Supernatants separated by centrifugation were diluted to a final concentration of 1.5% HNO 3 and then assayed for 28 inorganic ele ments by inductively coupled plasmaoptical emission spectrometry (ICPOES) using U.S. Environmental Protection Agency (EPA) Method 200.7, rev4.4 (U.S. EPA 1994). For ionic components, 250µg aliquots of the saline PM suspensions were diluted with 6 mL ultrapure water and analyzed for nitrate, sulfate (SO 4 2-), chloride, sodium, ammo nium (NH 4 + ), potassium, magnesium, and calcium by ion chromatography using EPA Compendium Method IO4.2 (U.S. EPA 1999) with calibration and quality control standards obtained from Dionex Corporation (Sunnyvale, CA) and Alltech Associates, Inc. (Deerfield, IL). Excessive amounts of sodium and chloride were present due to the saline in the PM suspension and were not reported in the result (Table 1). For carbon analy sis, 1mg aliquots of PM in methanol were pipetted onto prefired 1.45cm 2 quartz filters, dried, and analyzed by thermooptical meth ods based on sequential pyrolytic vaporization and detection by transmittance using a car bon analyzer (model 107A; Sunset Laboratory Inc., Tigard, OR) (Birch and Cary 1996).
Endotoxin measurements. Particle sus pensions in saline were sonicated, vortexed for 15 min, and centrifuged for 15 min at 2,100 rpm. Supernatants were collected and diluted in endotoxinfree water at PMequivalent concentrations of 1, 0.1, and 0.01 mg/mL. Endotoxin measurements were performed using the Limulus amoebo cyte lysate assay (QCL1000; Cambrex BioScience Inc., Walkersville, MD).
In vivo toxicity of PM. Experimental animals. Female CD1 mice were obtained from Charles River Laboratory (Raleigh, NC) and housed five per cage in a barrierisolated ani mal facility approved by the Association for Assessment and Accreditation of Laboratory Animal Care. Food and water were provided ad libitum. We used two different groups of mice: one for pulmonary assessment and blood and tissue collection [n = 7/group (total, n = 196); 7-8 weeks of age; body weight, 29.7 ± 2.3 g], and another for cardiac evalua tion [n = 5/group (total, n = 40); 16 weeks of age; 33.8 ± 0.8 g]. All animals were treated humanely and with regard for alleviation of suffering in accordance with National Institutes of Health guidelines (Institute of Laboratory Animal Resources 1996).
Pharyngeal aspiration. Surgical anesthesia was induced using a commercially available system (VetEquip Inc., Pleasanton, CA) to generate a 5% isoflurane/oxygen mixture. PM suspensions (25 or 100 µg) or the saline vehicle were administered by pharyngeal aspiration (Gilmour et al. 2007). Similarly, a separate group of mice received 2 µg lipopoly saccharide (LPS) (Escherichia coli endotoxin; 011:B4 con taining 10 6 unit/mg material; Sigma, St. Louis, MO) as a positive control to demonstrate maximal inflammatory response to this well characterized inflammatory agent.
Pulmonary responsiveness testing. Two hours before euthanasia and necropsy, a sub group of mice was challenged with increas ing concentrations of methacholine (MCh) aerosol in order to determine pulmonary responsiveness. We used a 12chamber whole body plethysmograph system using Buxco BioSystem XA software (Buxco Electronics, Wilmington, NC) to measure ventilatory parameters and the enhanced pause (Penh), an index of airflow limitation and a surrogate for bronchoconstriction (Hamelmann et al. 1997). After measuring baseline parameters for 10 min, an aerosol of saline or 6.25, 12.5, or 25 mg MCh/mL saline was pulled through each plethysmograph at a constant and identi cal flow rate (2 mL/sec). Response to saline and MCh was measured for 10 min, both dur ing the 2min aerosolization period and then for 8 min afterward. A MCh dose-response curve was then constructed for each animal. In addition, by interpolating the dose-response curve, the concentration of MCh required to produce a doubling of the Penh from the saline (EC 200 ) was calculated for each animal. Biological sample collection and hematology. At 4 or 18 hr postaspiration, mice from each treatment group were euthanized and plasma was obtained by cardiac puncture using a 1mL syringe containing 50 µL EDTA acid as anti coagulant. The trachea was cannulated and the left main stem bronchus occluded while the right lung lobes were lavaged once with 3 vol of 0.6 mL (35 mL/kg) Hank's balanced salt solution (HBSS) maintained at 37°C. The recovered broncho alveolar lavage fluid (BALF) was centrifuged at 130 × g for 10 min at 4°C with an aliquot stored at either 4°C for bio chemical measurement or -80°C for cytokine measure ment. Cell pellets were resuspended in 1 mL HBSS. Total BALF cell count for each mouse was measured with a Coulter Counter (Beckman, CA). Samples (200 µL) were pre pared in duplicate onto slides using a Cytospin (Shandon, PA) and stained with HEMA 3 stain set (Fisher Diagnostics, Middletown, VA) for cell differentiation determination using light microscopy. Hematologic data including total white blood cells, total red blood cells, hemoglobin, hematocrit, mean corpuscular volume, mean corpuscular hemoglobin con centration, platelets, and lymphocytes were measured using a Coulter AcT 10 Hematology Analyzer (Beckman Coulter Inc., Miami, FL). Residual blood samples were centrifuged and processed to plasma for subsequent clinical chemistry analyses.
Cytokine measurements. Concentrations of the pro inflammatory cytokines macrophage inhibitory protein2 (MIP2), tumor necrosis factorα (TNFα), and interleukin6 (IL6) in BALF were measured by enzymelinked immunosorbent assay using mouse Quantikine kits (R&D Systems, Minneapolis, MN). The lowest value in the standard curve for each cytokine was 15.625, 19.531, and 15.625 pg/ mL for MIP2, TNFα, and IL6, respectively, and all values below these lowest values were replaced with a fixed value of onehalf of the lowest value.
Cardiac perfusion. We used a separate group of mice for assessment of cardiac toxic ity. Mice aspirated 100 µg each of coarse, fine, or ultrafine PM collected from both NR and FR collection sites. At 18 hr post aspiration of PM, ex vivo Langendorff cardiac perfusion and subsequent ischemia reperfusion were per formed as described previously (Tong et al. 2009). The Langendorff perfusion model has been used to investigate cardiac responses after PM exposure in experimental animal studies (Bagate et al. 2006;Wold et al. 2006). We used the PowerLab system (AD Instruments, Milford, MA) to collect and process the left ventricular developed pressure (LVDP), heart rate, maximum of the first derivative of left ventricular pressure rise (+dP/dt max ), and mini mum of the first derivative of left ventricular pressure fall (-dP/dt min ) data. One heart did not achieve a developed pressure of at least 90 cm H 2 O and was excluded from data analy sis. Coronary flow rates were measured before and after induced ischemia reperfusion. After ischemia reperfusion, recovery of cardiac func tion and infarct size were measured at 1 and 2 hr of reperfusion, respectively. A detailed procedure of necrosis evaluation has been described by Tong et al. (2009).
Statistical analysis. We analyzed data in three increments. Initially, we performed a two way analysis of variance (ANOVA) to evaluate BALF and plasma data, and we used a three way ANOVA to evaluate the Penh data. These models examined overall effects of particle size (coarse, fine, and ultrafine) and sampling location (NR and FR sites) in the twoway ANOVA, or the influence of particle size, sampling location, and MCh concentration [levels: -1 (baseline), 0 (saline), 6.25 (6.25 mg MCh/mL saline), 12.5 (12.5 mg MCh/mL saline), and 25 (25 mg MCh/mL saline)] in the threeway ANOVA. Control exposure groups (saline and LPS) were not included in this analysis. The second step employed a oneway ANOVA model using an independent variable formed by combining location and size into a single treatment group parameter, and the saline control was included in the model. In the third step we used Dunnett's test to compare the effect between each PMexposed group and saline control group, adjusting significance level for multiple comparison. All the data were analyzed separately by each post exposure time period and PM dose. Significance was assigned at p < 0.05.

Results
Particle characteristics. Table 1 shows the col lected particle masses and recovered fractions after extraction from two weekly samples from each sampling location, as well as the chemical analyses. Approximately 20% more total PM mass was collected at the NR site than at the FR site. Coarse and fine NR PM masses were 27% and 20% greater, respectively, than cor responding fractions of FR PM; we found no difference in ultrafine fractions. The percent recovery ranges of sizefractionated PM after methanol extraction and evaporation were 54-80%, 41-70%, and 68-101% for the coarse, fine, and ultrafine PM samples, respec tively. The percent recoveries were relatively higher for FR PM and showed greater consis tency between weekly samples compared with NR PM. Losses likely occurred from particles remaining in the filter substrates as well as during the extraction procedure.
Calcium (Ca 2+ ), potassium (K + ), and nitrate (NO 3 -) were much enriched in the coarse fractions, whereas SO 4 2and NH 4 + were more abundant in the fine and ultra fine fractions. The ion chromatography anal yses of calcium and SO 4 2contents showed good agreement with elemental analysis using ICPOES. Calcium levels were 1.6 times higher in the ultrafine fraction of NR PM compared with the FR sample. The total car bon consisted of 19-22% of total mass of coarse and fine PM and 24-26% of ultrafine PM without apparent difference by location. Organic carbon amount ranged between 19% and 26%, with similar levels at both locations. Elemental carbon content was therefore much lower than the organic carbon and was mostly enriched in the fine fractions. The sum of ele mental carbon in all fractions of NR PM was 2.4 times higher than that of FR PM. With regard to biogenic components, the coarse PM contained approximately 100 times higher levels of watersoluble endotoxin than did the fine and ultrafine PM, and the NR coarse PM had 2 times more meas ureable endotoxin than did the FR coarse PM. Table 2 shows the inorganic elemental constituents of each size fraction from both locations. The concentrations of elements contained in NR PM [aluminum (Al), barium (Ba), calcium (Ca), cadmium (Cd), chromium (Cr), copper (Cu), iron (Fe), manganese (Mn), lead (Pb), antimony (Sb), silicon (Si), strontium (Sr), titanium (Ti), vanadium (V), zinc (Zn)] were at least 30% higher per gram of PM mass in any of three size fractions than in the FR PM. Lead (Pb) and tin (Sn) were present in relatively higher concentrations in the FR fine and ultrafine samples than in the NR samples. Figure 1 shows the mass balance of major classes of chemical components from each salinesuspended PM sample used for pharyn geal aspiration in mice. Overall, the chemi cal profile of the coarse fraction had a pattern different than that of the fine and ultrafine fractions, whose relative proportions of the major chemical classes were quite similar. Furthermore, none of the ratios of these major classes was appreciably altered by location. Unidentified moieties increased with particle size: 12-13% for ultrafine PM, 20-25% for fine PM, and 54-55% for coarse PM. These were presumably mineral components such as refractory alumino silicates and associated alkali and oxygen components, which, unlike the anthropogenic metals, did not fully dissolve in nitric acid.
Inflammatory biomarkers in BALF. With the exception of cytokines, which peaked at the 4hr time point, all other markers of injury and inflammation were highest at 18 hr post aspiration. Saline aspiration produced no apparent increase in markers compared with historic untreated control data, whereas LPS treated animals had high values of inflamma tion, cytokines, and markers of lung injury. The overall statistical analysis of numbers of poly morpho nuclear neutrophils (PMNs) by twoway ANOVA showed a significant particle size effect but no sampling location effect. At 18 hr post exposure, for the 100µg dose, the PMN numbers were significantly higher in the coarse PMexposed groups com pared with the saline group regardless of sam pling location, comprising 34-39% of the total cells (Figure 2), whereas the 25µg PM dose did not induce a significant effect. We observed no significant increases in PMNs for any of the fine or ultrafine PM at either concentration or location. Lymphocyte num bers were also significantly increased (< 1% of the total cells) at 18 hr postexposure to the FR coarse PM of 100 µg but not by the NR coarse PM (data not shown). We found no significant differences in numbers of alveolar macrophages between the various treatment groups, although we did detect particles in the alveolar macrophages from PMexposed mice.
Overall, the production of cytokines increased in a dosedependent manner. The analysis of inflammatory cytokines by two way ANOVA showed a particle size effect for MIP2, TNFα, and IL6. At 4 hr (Figure 3), both the 100µg and 25µg doses of NR and FR coarse PM significantly increased the production of the MIP2 and TNFα. The responses to both NR and FR coarse PM were still significant with the 100µg dose at 18 hr and to a lesser extent with the 25 µg (TNFα only; data not shown). We found no significant differences for the other size frac tions of any doses at any time points. At 4 hr (Figure 3), IL6 was significantly increased with the highdose coarse samples from both locations, and this effect persisted at 18 hr for the FRsample (data not shown). LPS responses were much higher than PM at the 4 hr, indicating that the mice were capable of responding to an inflammatory agent.
The overall statistical analysis of lung injury end points showed no significant effect by either particle size or sampling location. Analysis of individual treatment groups indi cated that only total protein in BALF was increased at a small but significant degree with the 100µg NR coarse PM at the 18 hr post exposure (Figure 4). LPS treatment showed a clear increase in BALF protein concentrations over saline controls at 18 hr. For complete data for pulmonary markers and responsiveness to MCh see Supplemental Material, Table S1 (doi:10.1289/ehp.0900730.S1 via http:// dx.doi.org/).

Hematologic markers in whole blood and plasma.
We monitored a variety of cellu lar and humoral markers in whole blood and plasma. The overall analysis by twoway ANOVA showed no significant effect by either particle size or location. In subsequent analyses, we found that the NR coarse PM     sample caused a significant increase in creatine kinase in the plasma at 4 hr post exposure (data not shown), indicating systemic effects after aspiration with this material. We found no significant changes in cell numbers, plate lets, or hematocrit markers or in plasma con centrations of A1AT, ALT, AST, CRP, FIB, HPT, or LDH.
Pulmonary responsiveness testing. The overall analysis of MCh responsiveness by threeway ANOVA indicated a significant effect by particle size and MCh dose. In the subsequent analyses, animals exposed to 100 µg of either NR or FR coarse PM showed significantly enhanced pulmonary responsive ness to the 12.5 mg/mL MCh dose at 18 hr post exposure compared with the saline group ( Figure 5A). The 25mg/mL MCh dose also enhanced pulmonary responsiveness but not to a significant level. When the data were transformed to reflect the EC 200 ( Figure 5B), we detected no statistical significance for the location and size factors. Nevertheless, mice exposed to 100 µg coarse PM required a non significant but notably lower dose of MCh to achieve a 100% increase in Penh compared with the saline group.
Cardiac ischemia-reperfusion injury. To examine the cardiac effects of the various PM samples, we isolated hearts at 18 hr after aspiration of 100 µg PM, and perfused and evaluated them for baseline hemo dynamics before under going ischemia reperfusion. The baseline hemo dynamics are shown in Table 3. The coronary flow rates in the NR fine and ultrafine groups were significantly lower than that of the saline group. We also found a notable but non significant reduction in the coronary flow rate in the FR ultrafine group. We observed increases in baseline LVDP and +dP/dt max , and decreases in -dP/dt min in the coarse and ultrafine groups of both NR and FR PM, but these were not significant.
After ischemia reperfusion, the FR ultra fine group showed significantly lower recov ery of postischemic LVDP ( Figure 6A) and increased infarct size ( Figure 6B) compared with the saline controls. For representative photomicrographs, see Supplemental Material, Figure S1 (doi:10.1289/ehp.0900730.S1). The NR ultrafine group also showed similar responses at a non significant level. The coarse PM groups showed the smallest changes fol lowed by fine PM groups, indicating these effects were dependent on particle size frac tion. Overall, the results showed that ultrafine PM exposure increased ischemia-reperfusion injury in the mouse heart.

Discussion
Many studies have compared the physico chemical characteristics and relative toxicity of sizefractionated PM from different sources and locations. In general, coarse PM obtained from ambient air sampling contains larger amounts of bacterial endotoxin and insoluble crustal components than smaller size fractions of PM and, on a mass basis, have more pro inflammatory potential than do fine or ultra fine fractions (Becker et al. 2005;Dick et al. 2003;Gilmour et al. 2007;Happo et al. 2007). Combustion aerosols in the fine and ultra fine modes can also generate significant lung inflammation Seagrave et al. 2005;Singh et al. 2004), whereas simple chemicals such as carbon black and titanium dioxide tend to cause stronger effects as sur face area increases and size decreases (reviewed by Donaldson et al. 1998). These results sug gest that inflammatory potential is dependent on both chemistry and particle size, and that effects may vary depending on the method of collection, freshness of the sample, and coconstituents. Near roadway environments are of particular interest because they com prise an array of freshly generated and aged particles that coexist in each of the three broad size ranges. Although it has been known for some time that the size, number concentration, and chemistry of PM change with distance from highways (Kuhn et al. 2005), toxicity data for PM samples from different proximities to roadways have only recently been examined. Kleinman et al. (2007) recently reported that NR (50 m) emissions in the fine and ultra fine range significantly enhanced allergic responses in sensitized animals compared with expo sures that occurred at FR locations (150 m).
However, toxicity data simultaneously compar ing cardiac and pulmonary effects after expo sure to each of the size fractions from NR and FR locations require further investigation.
Trafficassociated airborne particles feature a tri modal size distribution typically found in other ambient samples and display a distinct chemical profile related to the vehicle fleet mix, operating conditions, meteorology, dis tance from highway, and presence of other air pollutants. In this study, NR PM was more enriched with inorganic elemental compounds than FR PM. Specifically, the coarse NR PM fraction contained more Ba, Cu, Sb, Zn, Fe, Pb, and Cd, which have been attributed to brake and tire wear (Querol et al. 2007;Wåhlin et al. 2006), and Al, Mn, and Si, asso ciated with road dust, asphalt, and concrete. The fine and ultrafine fractions of NR PM contained larger amounts of Ba, Ca, Fe, Si, Sb, Ti, and V compared with FR PM. Along with some overlap with coarse PM, these elements are also components of lubricating and fuel oils (Dyke et al. 2007) and are emitted in vehicle exhaust (McDonald et al. 2004). The enrich ment of these elements in NR PM indicated a significant contribution of trafficrelated PM, whereas the high SO 4 2concentrations in the fine and ultrafine fractions at both locations likely reflected locally generated traffic emis sions as well as the regional pollutant profile consistent with the chemical makeup of fine PM in the eastern United States (U.S. EPA 2008). Relatively small amounts of elemental carbon in NR PM reflected most of the vehicle fleet at the highway being gasolinepowered vehicles (Baldauf et al. 2008).
The toxicity results showed that inflam matory responses in mice were increased in a dosedependent manner after aspiration of coarse PM samples, whereas treatment with equal masses of the smaller particles had much less effect. These findings are in agreement with previous reports from different cities across the United States and Europe that employed this sampling methodology and type of bioassay (Dick et al. 2003;Gilmour et al. 2007;Happo et al. 2007;Hetland et al. 2005). It should be noted that more coarse and fine PM collected at the NR location over the same sampling period would have resulted in a greater cumu lative exposure compared with FR location, whereas the experimental exposure regimen described here was designed to compare toxic ity on an equal mass basis. Furthermore, traf ficassociated gases and volatile compounds that were not collected by the PM samplers are steadily diluted with distance from the roadway (Baldauf et al. 2008) and may be major con tributors to reported health effects associated with proximity to roadways.
Much of the proinflammatory effect we observed from coarse PM was likely caused by the presence of LPS, whereas the smaller sized PM samples that produced no inflamma tion contained very little LPS. The NR coarse sample had twice as much LPS as did the FR course sample per gram of material but did not have substantially increased toxicity, sug gesting that the effects were not solely driven by this substance. In addition to LPS, the coarse PM comprised many other soluble and insoluble components capable of activating lung cells and inducing chemo tactic mediators with the subsequent recruitment of inflam matory cells (Hetland et al. 2004). Chemical and source apportionment analysis from the European study indicated that other compo nents, including sea salt, soilrelated elements, indicators of traffic, and nitrates, were also related to the inflammatory responses of coarse PM (Happo et al. 2007(Happo et al. , 2008. In agreement with the reports mentioned above, in the present study aspiration of fine or ultrafine PM caused little in the way of pulmonary inflammation despite the dose being as high as 100 µg per mouse. Because equivalent dosing of sourcederived material, including fine and ultrafine coal fly ash, die sel exhaust particles, and oil fly ash (Gavett et al. 1999;Gilmour et al. 2004;Singh et al. 2004), produce varying degrees of pulmo nary inflammation, this would suggest either that the fine and ultrafine ambient PM sam ples were less toxic than these sourcespecific PM samples, or that the atmospheric aging, collection and extraction procedures, or the presence of coconstituents reduced the activity of the sample.
In contrast to the pulmonary responses, the fine and ultrafine PM caused a decrease in coronary flow rate; after ischemia reperfusion, recovery of cardiac function was decreased; and the infarct size was increased by ultrafine PM. Previous studies have also reported decreases in vascular function and post ischemic recovery with exposure to urban ambient ultrafine PM (Cozzi et al. 2006) and trafficrelated ultrafine PM (Hwang et al. 2008;Peretz et al. 2008) in human volun teers and animals. Overall, the results suggest that ambient ultrafine PM has a higher potential to cause vascular dysfunction and cardiac damage than other size fractions and provides experimental support for recent epidemiologic and clinical reports (reviewed by Brook et al. 2004;Mills et al. 2009). There was also some suggestion that the NR ultrafine PM had greater effects on vascular function, but the FR ultrafine PM caused slightly more cardiac tissue damage after ischemia. Possibly these differences could be explained by the higher concentrations of elemental carbon at the NR location with increased amounts of hetero geneous organic aerosols (presumably second ary organic aerosols) at the distant location as has been previously reported (Kuhn et al. 2005;Riddle et al. 2008).
Although we did not observe cardiac effects for every particle size and location, the results illustrate that extrapulmonary responses can occur after fine or ultrafine PM exposure in the     (Gong et al. 2003;Samet et al. 2009). In one specific experimen tal example, inhalation of zinc sulfate did not cause lung injury in rats, but the soluble metal translocated to the cardiac tissue and inhibited phosphatases and kinase activation with subse quent effects on cardiac function (Wallenborn et al. 2008(Wallenborn et al. , 2009). It has also been demon strated that insoluble ultrafine PM can trans locate from the lungs to the circulation and be detected in other organs, including heart and liver (Furuyama et al. 2009;Takenaka et al. 2004). PM or its soluble constituents may be able to induce vaso constriction by increasing endothelin release (Brook et al. 2002;Cozzi et al. 2006;Peretz et al. 2008). Reactive oxy gen species (ROS), contained in the PM or endogenously generated in the lungs and heart, could also result in tissue damage (Gurgueira et al. 2002), which would decrease an ability to recover cardiac function and increase myocar dial infarction after ischemia. Also ROS could affect the nitric oxide pathway and further con tribute to the progression of atherosclerosis or other ROS (Sun et al. 2005).
On an equivalent mass basis, significant pulmonary and cardiac effects were observed with the coarse and ultrafine PM samples, respectively; however, in reality, unequal mass distributions for the different size frac tions would likely affect the extent of toxicity. Particle penetration and deposition efficiency after aspirations exposure would also not reflect sitespecific tissue doses after inhala tion of hetero geneous aerosols (Foster et al. 2001). These challenges can be overcome in toxicologic and clinical studies by develop ing advanced exposure techniques to separate, characterize, and deliver differentsized par ticles and accompanying gases from complex air pollution atmospheres in order to improve understanding of the physico chemical basis of air pollution health effects.

Conclusions
On a comparative mass basis, mice exposed to coarse PM from sites proximal to a high way showed significant increases in neutrophils, IL6, MIP2, TNFα, and protein in BALF and altered pulmonary function in healthy mice with no noticeable effects on the heart. Meanwhile, the same dose of ultrafine PM affected the cardiac system in the absence of any pulmonary responses. Chemical analyses showed clear compositional difference between coarse PM and smaller PM but less overall changes in the mass balance between NR and FR locations and little in the way of altered biological effects. However, the higher levels of coarse and fine PM (as well as gases and vapors) next to the highway would presumably have resulted in increased exposure and potential for more severe health effects. Further studies are needed to examine realtime toxicity of NR and FR atmospheres to identify the levels and types of particle and gas phase components that cause health effects in multiple organ systems. This type of work will ultimately expand knowledge on the chemical composition and health effects of air pollution mixtures between NR and FR sites and could be used to improve human health risk assessment and provide guidance for planning of new roads and communities.