Composition, Toxicity, and Mutagenicity of Particulate and Semivolatile Emissions from Heavy-Duty Compressed Natural Gas-Powered Vehicles

JeanClare Seagrave*,1, Andrew Gigliotti*, Jacob D. McDonald*, Steven K. Seilkop{dagger}, Kevin A. Whitney{ddagger}, Barbara Zielinska§ and Joe L. Mauderly*

* Lovelace Respiratory Research Institute, 2425 Ridgecrest Dr. SE, Albuquerque, New Mexico 87108; {dagger} SKS Consulting Services, 3942 Rives Chapel Rd., Siler City, North Carolina; {ddagger} Southwest Research Institute, 6220 Culebra Rd., San Antonio, Texas 78238-5166; and § Desert Research Institute, Atmospheric Sciences Division, 2251 Ragio Parkway, Reno, Nevada 98512-1095

1 To whom correspondence should be addressed. Fax: (505) 348-8567. E-mail: jseagrav{at}LRRI.org.

Received April 1, 2005; accepted June 9, 2005


    ABSTRACT
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 SUPPLEMENTARY DATA
 REFERENCES
 
Particulate matter (PM) and vapor-phase semivolatile organic compounds (SVOC) were collected from three buses fueled by compressed natural gas. The bus engines included a well-functioning, conventional engine; a "high emitter" engine; and a new technology engine with an oxidation catalyst. Chemical analysis of the emissions showed differences among these samples, with the high emitter sample containing markers of engine oil constituents. PM + SVOC samples were also collected for mutagenicity and toxicity testing. Extraction efficiencies from the collection media were lower than for similarly collected samples from gasoline or diesel vehicles. Responses to the recovered samples were compared on the basis of exhaust volume, to incorporate the emission rates into the potency factors. Mutagenicity was assessed by Salmonella reverse mutation assay. Mutagenicity was greatest for the high emitter sample and lowest for the new technology sample. Metabolic activation reduced mutagenicity in strain TA100, but not TA98. Toxicity, including inflammation, cytotoxicity, and parenchymal changes, was assessed 24 h after intratracheal instillation into rat lungs. Lung responses were generally mild, with little difference between the responses to equivalent volumes of emissions from the normal emitter and the new technology, but greater responses for the high emitter. These emission sample potencies are further compared on the basis of recovered mass with previously reported samples from normal and high-emitter gasoline and diesel vehicles. While mutagenic potencies for the CNG emission samples were similar to the range observed in the gasoline and diesel emission samples, lung toxicity potency factors were generally lower than those for the gasoline and diesel samples.

Key Words: engine emissions; compressed natural gas; comparative toxicology; intratracheal instillation; bacterial mutagenicity.


    INTRODUCTION
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 SUPPLEMENTARY DATA
 REFERENCES
 
Concerns about potential health effects from inhalation of vehicle emissions have increased as the result of epidemiological studies showing increased morbidity and mortality in people exposed to air pollutants (recently reviewed in Englert, 2004Go). Some studies have shown that health effects are inversely correlated with distance from major highways (Brauer et al., 2002Go; Brunekreef et al., 1997Go; Finkelstein, 2003Go). In addition, controlled human exposure studies of diesel exhaust or diesel exhaust particulate matter (PM) have shown inflammatory responses or alterations in immune responses (Nightingale et al., 2000Go; Nordenhall et al., 2000Go; Rudell et al., 1999Go; Salvi et al., 1999Go; Stenfors et al., 2004Go). As a result of these concerns, compressed natural gas (CNG) has been promoted as a cleaner-burning fuel. In fact, CNG-fueled heavy duty vehicles appear to produce less PM, but greater total hydrocarbons compared to diesel-fueled vehicles (McCormick et al., 2000Go). Characterization of the emissions from such vehicles has demonstrated that the emissions contain potentially toxic materials (Lapin et al., 2002Go). However, to date, few of the biological effects of these emissions have been examined. Despite the lack of toxicological data, portions of some heavy-duty vehicle fleets have been converted to CNG.

Previously, we compared the toxicity of collected PM and vapor-phase semivolatile organic compounds (SVOC) from a series of gasoline and diesel vehicles using bacterial mutagenicity testing and intratracheal instillation of these emission samples into rats (Seagrave et al., 2002Go). The current study was conducted to extend this comparison to emissions from heavy-duty CNG vehicles. The study tests the hypotheses that CNG emissions are less toxic than gasoline or diesel emissions, and that high emitter CNG vehicle emissions are more toxic than those from low emitters.


    MATERIALS AND METHODS
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 SUPPLEMENTARY DATA
 REFERENCES
 
The experimental design for this study was similar to the previously published comparison of diesel and gasoline emissions (Seagrave et al., 2002Go; Zielinska et al., 2004Go). Because the methods for collecting and processing the emissions samples were identical to that study, they are only described briefly here. The vehicles tested in this study were all heavy-duty passenger buses obtained from Dallas, Fort Worth, and Houston, TX, transit systems. The new technology (NT) vehicle was a recently purchased (216 miles) 2002 model Nova bus with a Detroit Diesel Series 50G engine and an exhaust oxidation catalyst. The normal emitter (NE) vehicle was a properly functioning, in-use (134,259 miles) 1997 model New Flyer bus with a Detroit Diesel Series 50G engine and no exhaust after-treatment. The high-emitter (HE) vehicle was a retired (over 250,000 miles, odometer not working) 1992 Flexible bus with a Cummins L10G engine and no exhaust after-treatment. The lube oil was used as received. The buses were fueled with natural gas supplied by the San Antonio, TX, City Public Service, connected directly to the intake system (bypassing the vehicles' on-board fuel tanks). Fuel composition was very similar for the three buses, ranging from 96.0 to 96.8% methane, 1.6 to 1.9% ethane, 0.9 to 1.1% carbon dioxide, 0.6 to 0.8% nitrogen, and traces of other gases. The sulfur content of the fuel was not measured.

Emissions Collections and Sample Processing
Vehicles were operated at an ambient temperature of ~25°C on a chassis dynamometer. Samples were collected over several days, with each day beginning with a cold-start U.S. Environmental Protection Agency heavy-duty urban dynamometer driving schedule (HD-UDDS) followed by three central business district cycles (Federal Transit Administration of the U.S. Department of Transportation). This series was repeated seven times each day with hot-start HD-UDDS (after the first). Emissions passed directly into a constant volume dilution tunnel to a final dilution of approximately 1:35. The dilution ratios were within 3% of each other for the three buses. Approximately 25% (23.9 to 24.8%) of the total emission stream was directed through a 20- x 20-in. Teflon-impregnated glass fiber filter (Pallflex Product Corp., Putnam, CT) that was followed by a canister containing a 4-in. plug of polyurethane foam (PUF), 20 g cross-linked divinyl benzene (XAD-4), and a second PUF plug. The temperature at the filter collection point was ~35 to 40°C. The filters were extracted into acetone by sonication and gentle brushing, and the PUF/XAD-4/PUF fractions were extracted into acetone by Soxhlet extraction. Both fractions were concentrated by evaporation under nitrogen, and the two fractions were recombined. An equivalent volume of air from the tunnel in the absence of an emission source was collected on to the media and extracted similarly as the Tunnel Background (TB) sample.

Chemical Characterization of Emission Samples
PM mass collected on the 20- x 20-in. filters was measured on an analytical balance following subtraction of the precollection filter weight. PM emission rates were calculated based on the PM mass, volumetric flow rate through the filters, and calculated vehicle miles traveled over the course of the collections. Masses of the recovered PM and SVOC samples were determined by weighing the residual mass following evaporation of the solvent at room temperature. Emission rates for recovered SVOC were determined identically to the PM.

The chemical composition of the PM and SVOC fractions of each of the samples was analyzed at the Desert Research Institute (Reno, NV), as described in detail previously for diesel- and gasoline-powered motor vehicle samples (Zielinska et al., 2004Go). Separate samples for chemical analysis were collected on media optimized for the different chemical analyses in parallel to samples collected and processed for the toxicity testing. Specifically, PM collected on 47-mm-quartz fiber filters (Pall Corp., East Hills, NY) was analyzed for organic and elemental carbon by thermal/optical reflectance (Chow et al., 2001Go). The filters were then extracted into purified water for determination of inorganic ions (sulfate, nitrate) by ion chromatography. Teflon membrane filters were used to collect PM for analysis of elements (metals and associated analytes) using x-ray fluorescence (XRF). To analyze organic compounds, we collected PM on Teflon-coated glass fiber filters (Pall) and SVOC on PUF/XAD-4/PUF cartridges downstream from the filters. The filters were sequentially extracted into dichloromethane and acetone, and the cartridges were extracted into acetone. The samples were concentrated under nitrogen, transferred to acetonitrile, and analyzed by gas chromatography/mass spectrometry. The organic analysis focused on chemicals used in previous studies to illustrate differences among motor vehicle and other types of emissions. These included polycyclic aromatic hydrocarbons ([PAH], 75 species), nitrogenated PAH (25 species), and hopanes and steranes (37 species). Speciated organics included vapor-phase SVOC, compounds that distribute between both vapor/particle phase, and compounds that are exclusively in the particle phase. To conserve space, individual data for all of the organic and inorganic species are not presented. Instead, chemical classes are summarized as fractions of either particle mass or fractions of specific chemical classes. The speciation data can be obtained by request from the authors.

Measurement of Bacterial Mutagenicity
Suspensions of the recovered PM were combined with the recovered SVOC extracts and transferred to dimethylsulfoxide and shipped overnight at 0°C to BioReliance Corp. (Rockville, MD) for mutagenicity testing using the Ames Salmonella plate incorporation assay (Ames et al., 1975Go). Mutagenicity was assessed in Salmonella strains TA98 and TA100, each with and without metabolic activation by the microsomal fraction of Aroclor-induced rat liver homogenates (S9) (Maron and Ames, 1983Go). Each sample was tested in duplicate over a 200-fold range of concentrations in parallel with negative (solvent) and positive controls. The positive controls were 2-nitrofluorene for TA98-S9, sodium azide for TA100-S9, and 2-aminoanthracene for TA98+S9 and TA100+S9. Colonies on plates with sufficient precipitate to interfere with electronic counting were counted by hand; other plates were counted using an automated colony counter. All samples met the response criteria of dose-related increases in mean revertants per plate, rising to at least twice the negative control value. However, high concentrations were toxic to the bacteria and suppressed the observed revertant values. Responses were, therefore, quantified as the slope of the initial linear portion of the dose-response curve fit to the linear model as described under statistical analysis of data below.

Measurement of In Vivo Toxicity
Animals.
Male F344/Crl BR rats were purchased from Charles River Laboratories (Wilmington, MA). The rats, 8 ± 1 weeks old at receipt, were quarantined for three weeks before dosing and were 11 ± 1 weeks old at the time of instillation. Serological testing confirmed the absence of common rodent pathogens. The rats were housed two per shoebox cage with hardwood chip bedding and filter caps. The light/dark cycle was 12 h light/12 h dark with lights on at 0600 h. Food (Harlan Teklad Lab Blox, Madison, WI) and water were provided ad libitum. The room temperature was maintained at 20–22°C, with a relative humidity of 20–50%. All procedures involving animals were approved by the Institutional Animal Care and Use Committee.

Reagents and supplies.
General chemicals and reagents were purchased from Sigma Chemical Company (St. Louis, MO), unless otherwise specified.

Sample preparation.
The PM/SVOC mixtures for toxicological testing were prepared by dilution (at least 100-fold from the stock in acetone) and sonication in sterile 0.9% NaCl containing 0.01% Tween-80 as previously described (Seagrave et al., 2002Go). The final acetone content of these solutions was adjusted to 1%. Dilutions were made in 0.9% NaCl with 1% acetone and 0.01% Tween-80. Three dilutions of the TB control were prepared similarly from the concentrated acetone extract of filters and PUF/XAD-4/PUF canisters exposed to equivalent volumes of the tunnel dilution air. As a positive control, National Institutes of Standards (NIST) Standard Research Material 2975 (fork lift diesel soot: [DS]) was prepared in the same manner.

Intratracheal instillation.
Rats were anesthetized using 5% halothane in oxygen with nitrous oxide. Once anesthetized, each rat was intubated orally with an intratracheal cannula and instilled with an emission sample or control material suspended in a total volume of 0.5 ml saline. The rats were allowed to recover and returned to their cages.

Groups of five rats for each dose were instilled with the combined sample, with the highest dose determined by a preliminary range-finding experiment to cause responses without lethality. The first experiment tested the responses to material collected from 0.3, 3, or 30 m3 of the TB and 3, 6, or 12 m3 of NE emissions. The second experiment tested responses to material collected from 6, 12, or 24 m3 of NT; 0.75, 1.5, 3 m3 of HE; and 24 m3 of NE. Each experiment included a positive control of a 1 mg/rat of DS and a blank instillation consisting of 0.01% Tween-80 and 1% acetone in 0.9% saline.

Euthanasia and processing.
Previous studies indicated that the strongest inflammatory and cytotoxic effects were observed 24 h after the instillations (Seagrave et al., 2002Go). Rats were therefore killed by pentobarbital overdose 24 h after instillation, and body weights were recorded. The lungs were removed and weighed. The right cranial lobe was tied off and the left bronchus was temporarily clamped. The right middle, caudal, and accessory lobes were then lavaged with two, 3-ml aliquots of Dulbecco's phosphate buffered saline. The recovered bronchoalveolar lavage fluid (BALF) was combined, the volume was recorded, and the BALF was kept on ice until processed. After lavage, the right cranial lobe was removed and snap frozen in liquid nitrogen. The left bronchus was then unclamped and the lungs were gently instilled with 10% neutral buffered formalin to approximate normal volume, the trachea was ligated, and the lungs were immersed in the same fixative for at least 48 h.

Lung lavage cells and fluid were collected for analyses. An aliquot of BALF was diluted in red blood cell lysing buffer (Roo's solution: 9 g/l NH4Cl2 + 1 g/l K2CO3 + 0.0372 g/l EDTA) and 0.02% (final) Trypan blue to determine total cell numbers and cell differentials. The total number of leukocytes was determined using a hemocytometer. Cytocentrifuge preparations were used for evaluation of differential cell counts. The slides were stained with Wright-Giemsa using a Hema-Tek Slide Stainer, and the percentages of macrophages, polymorphonuclear leukocytes (PMNs), and lymphocytes in 300 cells per sample were counted using a 20X objective.

The remaining BALF was centrifuged (10 min at 1000 rpm). Lactate dehydrogenase (LDH) (Gay et al., 1968Go), total BALF protein (Watanabe et al., 1986Go), and alkaline phosphatase were measured in the supernatant using a Hitachi 911 (Roche Diagnostics, Basel Switzerland) clinical chemistry analyzer.

Histopathology was assessed in the fixed left lung by light microscopy. Three transverse slices about 3-mm thick (one through the lung cranial to the hilus, one through the main airway just caudal to the hilus, and one near the end of the main axial airway) were embedded in paraffin, sectioned (5 µm), and stained with hematoxylin and eosin. Histologic grading was performed by a board certified veterinary pathologist (A.P.G.) based on the responses of the lung to the instilled material. Briefly, the character, severity, and distribution of pathologic responses such as inflammatory infiltrates (e.g., suppurative bronchiolitis, mononuclear perivascular cuffing), edema, Type II pneumocyte hypertrophy or hyperplasia, fibrosis, and others were graded using a scale from 0 (no pathology; absence of that particular response/diagnosis) to 5 (extreme pathology; severe and widespread presence of that particular response/diagnosis). For analysis, the specific histopathologic diagnoses were then pooled into categories of responses indicating cytotoxicity (e.g., diagnosis of necrosis), inflammation (e.g., bronchiolitis), or parenchymal changes (e.g., fibrosis) as previously described (Seagrave et al., 2002Go). For each animal, the sum of the individual scores for each pathologic response/diagnosis within a category was calculated. The sum of all scores was also analyzed as an overall pathological response.

Statistical Analysis of Data
We plotted the responses of each emission sample as a function of the doses (in m3 of exhaust) of each emission sample. None of the TB sample doses gave responses greater than the control in any of the measured parameters, so the data from all three doses were pooled with the solvent control as the zero dose values. Response functions were fit to the data at three different dose levels per emission sample (plus the baseline control values), and the slope coefficients from these fitted dose-response curves were used to make comparisons among the toxicological potencies of the different emissions. This technique makes optimal use of the entire dose-response curve, providing substantially more statistical power to discriminate among the potencies of the different emissions samples than would be possible from individual dose-to-dose comparisons.

For each endpoint of the toxicity parameters evaluated in rats, a single regression model with emission-specific slope coefficients was utilized:

where Yki = endpoint value in the kth emission group at concentration i, concki = concentration i in kth emission, ck = fitted estimate of control group mean, bk = fitted slope estimate for emissions group k, and log = natural logarithm.

For variables that could exhibit values of zero, log (Yki + 1) was analyzed. Although linear models (i.e.,) fit most of the lavage data equally well, log-linear models were used to obtain potency estimates that were comparable to those of a previous study (Seagrave et al., 2002Go). For the mutagenicity data, which showed strongly linear initial dose-responses up to the level of toxicity to the bacteria, Yki was substituted for log (Yki) in this model. We used generalized least squares regression to estimate the dose-response coefficients for endpoints that exhibited a relationship between sample mean and standard deviation, using squared reciprocals of fitted values from estimated standard deviation functions as weights (Neter et al., 1996Go); ordinary least squares regression was employed for all other endpoints. Statistical significance of individual emission-specific toxicological potency estimates (bk != 0) was assessed using t-tests associated with the regression model. Comparisons between pairs of emission-specific toxicological potencies (bk) were evaluated with F-test contrasts (Searle, 1971Go), calculated by the SAS software system (Cary, NC). To evaluate differences among emission samples for each endpoint, p-values from the pair-wise F-tests were adjusted for multiple comparisons (three paired comparisons among the three samples) using the modified Bonferroni procedure devised by Hochberg (1988)Go. Statistical significance was assessed at p = 0.05.

Exposure Metrics and Interpretation of Potency Factors
In a previous study comparing mutagenicity and in vivo toxicity of gasoline- and diesel-powered vehicle emissions (Seagrave et al., 2002Go), we characterized exposure-related effects relative to the mass of administered emission material (either per plate or per animal). A significant disadvantage to this metric is that it does not incorporate the relative emission rates into the potency factors. Thus, a high emitter and low emitter vehicle with similar mass-based potency factors might not be perceived as different, even though the hazard is greater from the high emitter simply because it produces more mass. In the current study, we have chosen to use equal volumes of the equivalently diluted emissions as the metric, thus incorporating emission rates directly into the potency factors. However, to facilitate comparisons with results from the previous paper, we also present results as mass-based potency factors. These two metrics provide different types of useful information: biological responses as a function of mass dose provide a means to investigate differences in the potencies of the emissions materials themselves, while the volume-based metric provides insight into effects associated with differences in the amounts of materials generated by different emissions sources at equal units of operation and exposure.

It should be noted that conversion between these different metrics is straightforward for the linear dose response curves (mutagenicity): for instance, a potency factor in revertants per microgram can be converted to revertants per m3 by multiplying the mass-based potency factor by the µg/m3. However, because the dose is contained in the exponent of the potency factors derived from exponential fits of the data, the conversion is more complex. More importantly, the response is not a simple linear function of the conversion factors. For more detail, please see the on-line supplementary information.


    RESULTS
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 SUPPLEMENTARY DATA
 REFERENCES
 
Vehicle Emission Rates and Exhaust Composition
PM emission rates, determined as net filter weights, and SVOC emission rates, determined as the mass recovered in the extracts from the PUF/XAD-4/PUF cartridges, are reported in Table 1. The NT vehicle yielded somewhat lower emissions than the NE. These two vehicles produced 4–6 times more SVOC mass than PM. As expected, the HE vehicle had significantly higher PM and SVOC emission rates than the other vehicles, and a much greater increase in PM than SVOC. Despite good (67–100%) PM extraction efficiency using the method described here for gasoline and diesel emission samples (Seagrave et al., 2002Go), the procedure resulted in poor recovery from the CNG emission samples: only 18% for NE, 13.9% for NT, and 14.0% for HE. Because the SVOC emission rates are calculated based on the material that was recovered into solution, the recovery efficiency for the SVOC from the collection medium is not known. PM and SVOC were collected also from a volume of tunnel dilution air similar to that used for the emission samples. The mass in these fractions corresponded to less than 1% of the PM and about 2.5% of the SVOC of the equivalent volumes collected for the HE sample, but as much as 20% of the PM and SVOC of the NT sample. Because this background is presumed to contribute to the mass and the toxicity of the emission samples, the values presented have not been corrected for this additional mass.


View this table:
[in this window]
[in a new window]
 
TABLE 1 PM and SVOC Emission Rates and Recovery

 
The proportion of carbon, inorganic ions (sulfate and nitrate), and elements in the emission samples and TB, reconstructed from the masses of these constituents detected by the analyses, are summarized in Figure 1. Inorganic carbon was similar, approximately 4% of total mass, in all samples. Organic carbon contributed the major fraction of the HE emissions, with smaller relative proportions of other elements and ions. In contrast, the NT and NE had high proportions of elements (and to a lesser extent ions). Figure 2 gives a breakdown of the major non-carbonaceous elements (primarily metals) detected by XRF analysis in the samples. No attempt was made to correct these values for the mass as oxides or other chemical forms. These elements are all typical for engine emissions (as shown in Docekal et al., 2004Go), corresponding with typical lube/fuel additives (calcium, zinc, phosphorus) as well as engine wear (iron). Although lead comprised a very small amount of the emissions, and is therefore not shown on the graph, it is noteworthy that the HE vehicle emissions contained more than 100 times the concentration of lead found in the other two vehicles' emissions, possibly due to wear of the engine bearings. The large proportion of these non-carbonaceous elements in the NT and NE exhausts (relative to total PM) is not typical for gasoline or diesel engine exhaust, but it should be noted that the concentration of these elements in the exhaust is still less than those in the HE. The TB sample was primarily composed of organic carbon and a lesser amount of sulfate, with little to no elemental carbon or other elements (data not shown). The absence of these constituents suggests that background contamination did not arise from mechanical re-entrainment of particulates from the tunnel wall. Organic carbon, which dominated the composition, likely arose from desorption of small amounts of SVOC from the walls of the tunnel. This interpretation is supported by the fact that essentially all the PAHs measured in the TB sample were found in the SVOC volatility range (individual data not shown).



View larger version (40K):
[in this window]
[in a new window]
 
FIG. 1. Constituents of PM fraction of emissions from CNG buses. Data are shown as the percent of total reconstructed mass of PM collected on the appropriate medium for the various assays. Organic carbon has been multiplied by 1.2 to account for total organic mass.

 


View larger version (26K):
[in this window]
[in a new window]
 
FIG. 2. XRF elemental analysis of PM from CNG buses. The major non-carbon elements detected by XRF in the PM fraction are shown as a fraction of the total of these elements. Sulfur has been omitted from these analyses, since it is present as sulfate and is shown in Figure 1.

 
Table 2 shows the concentrations of PAH and nitro PAH compounds as well as hopanes/steranes among the exhaust samples and TB. Although they accounted for a small portion of the total mass, hopanes and steranes were nearly 20-fold higher in the HE samples than in the other emission samples. Hopanes and steranes are associated with lubrication oil emissions and, importantly, were shown in our previous work to trend with the toxicity of vehicle emissions (McDonald et al., 2004Go). PAHs as a fraction of the total mass of the emissions were smaller in the HE emission sample than the properly functioning vehicles, although because of the much higher emission rates, the aerosol concentrations of this class were substantially higher in the HE emissions than the other two vehicles. In each of the vehicle emission samples, the PAHs were primarily present in the SVOC fraction (data not shown). The heavier PAHs (e.g., molecular weight >240) were present in low to nondetectable amounts. These samples contained a relatively high proportion of nitro-PAH compounds, which accounted for over 30% of the total PAH concentration in the HE and over 90% of the PAHs in the NE emission samples. However, the absolute levels (ng/m3) of collected nitro-PAH emissions were highest in the NE vehicle, intermediate (approximately 50% of the NE) in the HE emissions, while nitro-PAH emissions from the NT vehicle were much lower.


View this table:
[in this window]
[in a new window]
 
TABLE 2 Concentrations and Fractions of Selected Emission Components per m3 of Diluted Exhaust

 
Mutagenicity
The TB sample induced only extremely low levels of mutations in the bacterial assay. Emissions from all three CNG vehicles were mutagenic in both TA98 and TA100 strains of Salmonella. The highest tested doses appeared to be toxic to the bacteria as indicated by a reduction in bacterial lawn. The presence of S9 reduced mutagenicity in TA100, but not in TA98. Under all four protocols (TA98 or TA100, with or without S9), the mutations induced by equivalent volumes of the emissions were much higher for the HE than for the NE, with the NT vehicle producing the lowest numbers of revertants (Fig. 3).



View larger version (21K):
[in this window]
[in a new window]
 
FIG. 3. CNG emission samples cause mutations in Salmonella. Samples of PM suspensions combined with SVOC extracts were tested for mutagenicity in Salmonella strains TA98 and TA100, with (+) and without (–) S9 added in a rat liver microsomal preparation. Mutation frequencies were determined by linear regression of the mutations per plate over a range of subtoxic concentrations of the extracts. Mutagenicity is shown as the revertants produced by the emissions recovered from specified volumes of the diluted emissions. Error bars are the estimated error of the regression coefficients. All samples produced significant mutations. The + indicates significantly different from the NE sample and # indicates significantly different from the NT sample.

 
Lung Toxicity
Generally, responses were fairly mild. One animal in the highest concentration group of the NT sample died before the scheduled sacrifice time, but it is unclear whether this was a direct result of the treatment since the only histopathological finding was pulmonary edema, indistinguishable from that commonly occurring post-mortem. There were no significant changes in lung to body weight ratios in any of other exhaust sample-treated animals. The sum of the histopathological indicators was driven almost entirely by inflammatory responses, discussed below.

The lavage fluid from one animal in the highest group of NE (24 m3) contained noticeable red blood cells. This animal also had exceptionally high levels of LDH, protein, and total nucleated cells (more than 10 times the levels in any of the other animals in this group). However, the histopathology from this rat did not reflect these effects. We speculate that despite efforts to prevent influx of blood during the removal of the lungs, the lavage fluid was contaminated with blood from the thoracic cavity. In addition, three other animals (one in the diluent control, one in the high-dose TB, and one in the mid-high NE group) had red blood cells in the lavage fluid and levels of at least three lavage parameters that met statistical criteria as outliers (most of these values were also more than 10 times the values of the other animals in the respective group), and these data have been omitted from the analysis (along with the animal in the highest concentration of NT that died before the scheduled sacrifice).

The two primary acute cytotoxicity indicators, lavage total protein (shown in Fig. 4) and LDH, showed very similar dose-response curves, with the effects of the HE sample being pronounced at low exposures when plotted on the basis of the equivalent volumes (Fig. 4A). The NE and NT samples showed much lower potencies. However, on a per unit mass basis (Fig. 4B), the HE sample was similar in potency to the other two CNG samples for these parameters, indicating that the HE toxicity (volume-based) related to a higher emissions rate, rather than increased toxicological potency of emissions substances. The potency factors for alkaline phosphatase (AP) and the histopathological indications of cytotoxicity followed similar response patterns for these three emission samples (Fig. 5).



View larger version (17K):
[in this window]
[in a new window]
 
FIG. 4. CNG emission samples increase protein in lavage fluid. Data shown are the mean and SEM of the protein levels measured in lavage fluid 24 h after instillation with the emission samples. Panel A shows the results based on the volume of emissions, while the bottom panel shows the results based on mass. Note logarithmic y axes. Responses to 1 mg of NIST SRM2975 DS are shown in Panel B for comparison. Dashed line indicates the mean protein levels in rats instilled with samples from the TB or the solvent control. Note logarithmic y axes.

 


View larger version (14K):
[in this window]
[in a new window]
 
FIG. 5. Potency factors for cytotoxicity. Potency factors were estimated by fitting an exponential function to the data, with dose in m3 of collected emissions as described in the text. Error bars represent the standard errors of the estimated regression coefficients. Panel A: Cytotoxicity and general potency factors. Lung wt: lung to body weight ratio, LDH: lactate dehydrogenase activity; Protein: total lavage protein concentration; AP: alkaline phosphatase; HP cytotoxicity: histopathological indicators of cytotoxicity. The * indicates significantly different from the zero and + indicates significantly different from both NE and NT samples.

 
The patterns of inflammatory responses also distinguished among these samples. On a per unit volume basis, the HE sample produced a dose-dependent increase in total cells (Fig. 6A), including statistically significant increases in both macrophages and neutrophils (Fig. 7). In contrast, only weak increases in macrophages and little or no increase in neutrophils were produced by the NE and NT samples. On a per unit mass basis (Fig. 6B), there was also evidence that the HE sample was more potent in than the other samples in the production of total cells and macrophages, indicating that the volume-based HE effect was not due solely to the higher emission rate (Table 4). Inflammatory indicators in the histopathological examination were not statistically significant due to relatively large uncertainties in the potency estimates, but the response pattern was consistent with that which was exhibited by lavage parameters (Fig. 7).



View larger version (20K):
[in this window]
[in a new window]
 
FIG. 6. CNG emission samples increase inflammatory cells in lavage fluid. Data shown are the mean and SEM of the number of cells per ml of lavage fluid 24 h after instillation with the emission samples. Panel A shows the results based on the volume of emissions, while the bottom panel shows the results based on mass. Note logarithmic y axes. Responses to 1 mg of NIST SRM2975 DS are shown in Panel B for comparison. Dashed line indicates the mean cell concentration in rats instilled with samples from the TB or the solvent control.

 


View larger version (14K):
[in this window]
[in a new window]
 
FIG. 7. Potency factors for inflammation. Potency factors were estimated by fitting an exponential function to the data, with dose in m3 of collected emissions as described in the text. Error bars represent the standard errors of the estimated regression coefficients. Total cells/ml, macrophage/ml, PMNs/ml, and HP inflamm (histopathological indicators of inflammation). The * indicates significantly different from the zero and + indicates significantly different from the NE sample.

 

View this table:
[in this window]
[in a new window]
 
TABLE 4 Lung Toxicity Potency Factors Based on Dose in Massa

 
Comparisons with Gasoline and Diesel Emissions
PM emission rates for the NE (7 mg/mi) and NT (5 mg/mi) vehicles were well below those of the normal emitter gasoline or diesel vehicles (10 and 144 mg/mi, respectively, [Zielinska et al., 2004Go]). The HE emission rate (>400 mg/mi) was intermediate, falling between the normal emitter gasoline and diesel vehicles and the white smoker gasoline (WG) (770 mg/mi) or high emitter diesel (483 mg/mi) vehicles.

The mutagenicity of these samples can be compared with the data previously reported using identical methods for the PM + SVOC fractions of diesel and gasoline vehicles by the conversion process described in Materials and Methods. It is important to note that the previous set of emission samples were from light-duty gasoline and diesel vehicles, while only CNG buses (heavy-duty vehicles) were used in this study. However, all samples from both studies were obtained from vehicles operated under their typical operating conditions, and these comparisons provide a starting point for comparing the toxicity of the emissions produced from vehicles utilizing these three fuel types.

The mutagenicity in revertants per µg, shown in Table 3, indicates that the much greater potency of the HE sample compared with the NE sample on a volumetric basis (Fig. 3) was primarily due to the greater mass of the emissions, since less than two-fold differences in mutagenicity per unit mass were observed between the HE and NE samples in both strains with and without metabolic activation. However, the NT sample contained substantially less mutagenic activity per unit of mass (as well as per emission volume). Table 3 also shows data for selected samples from the previous study. Like the gasoline and diesel emissions samples, the range of mutagenicity caused by the CNG emissions samples was fairly limited (less than a 10-fold difference between the least and most potent conditions). Comparing the normal emitters of the three classes, effects in TA98 with or without metabolic activation were greater for the NE sample than for the gasoline or diesel normal emitter samples. The HE sample had comparable mutagenicity to the high emitter diesel sample and white smoker gasoline sample, while the NE sample was the least mutagenic sample in this strain. Metabolic activation in TA98 had little effect on the CNG or diesel samples, but increased the potency for the gasoline samples. In TA100 without metabolic activation, the HE sample was the most potent of the CNG samples, comparable to previously reported potencies for the white smoke gasoline and high emitter diesel samples. The NE sample resembled the normal emitter diesel and black smoker gasoline samples, and the NT sample was similar in mutagenic potency to the normal emitter gasoline sample. Metabolic activation in this strain decreased mutagenicity by 50–70% for the three CNG samples. Reduced mutagenicity with metabolic activation was also observed in TA100 for the high emitter diesel and white smoker gasoline, but not for the other gasoline and diesel emission samples.


View this table:
[in this window]
[in a new window]
 
TABLE 3 Mutagenicity Potency Factors Based on Dose in Massa

 
Table 4 compares the mass-based potency factors for rat lung lavage parameters after instillation of emission samples from CNG vehicles with those for gasoline and diesel vehicle samples (Seagrave et al., 2002Go). These data indicate that the CNG mass-based potency factors are, in general, lower than those for the previously studied gasoline and diesel vehicles. Specifically, there were no significant effects of any of the CNG samples on lung to body weight ratios, and only the HE sample had a significant effect on any of the histological parameters, while all of the gasoline and diesel samples had significant potencies for all these parameters. Even for the endpoints where effects were observed for the CNG samples (e.g., LDH, protein) the mass-based potency factors were only 10–20% of those derived for the gasoline and diesel samples. One interesting exception is the greater potency of the HE and NT samples for increasing AP activity, compared with the gasoline and diesel vehicle emission samples.


    DISCUSSION
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 SUPPLEMENTARY DATA
 REFERENCES
 
In this study we have compared the composition, mutagenic potential, and acute lung toxicity of collected emissions from heavy-duty CNG-fueled vehicles using methods previously applied to emissions collected from a series of gasoline and diesel-fueled vehicles. However, to take the emission rates into account in evaluating the biological effects in the present study, the potencies were initially derived on the basis of the emissions present in equivalent or overlapping ranges of the volumes of the diluted exhaust.

One important result of this study was that under all assay conditions (either strain of Salmonella and with or without metabolic activation), the mutagenic potencies, as revertants per m3 (thus taking into account the differences in emission rates) of the combined PM + SVOC fractions of the HE vehicle were much greater than for the NT vehicle, with the NE sample having intermediate activity. These observations confirm and extend the previous data for an organic extract of the PM fraction of a CNG-powered heavy duty truck (Lapin et al., 2002Go), showing that emissions from CNG-fueled vehicles contain substantial amounts of direct-acting mutagens. This was confirmed by the presence of nitro-PAH in both the HE and NE samples that have been shown in several previous studies (McDonald et al., 2004Go) to be associated with bacterial mutagenicity. An important consideration for both these samples and the previously reported data (McDonald et al., 2004Go) is that many of these nitro-PAH compounds can be formed by PAH reacting with NOx that pass through the filter and sorbent during sample collection (Arey et al., 1988Go). This artifact was also suggested for diesel and gasoline emission samples reported previously (McDonald et al., 2004Go), but likely is even more important for CNG samples (especially low emitters) because low vehicle mass emissions required the samples to be collected for longer time periods (up to 50 h for the NE sample) to obtain enough sample for the toxicity and mutagenicity studies. An even longer time was required to collect sufficient sample of the NT emissions (63 h), which had a lower proportion of nitro PAHs than the NE vehicle. However, the NOx emission rate of the NT vehicle was only 0.01 g/mi, compared with 0.253 for the NE and 2.05 for the HE.

Intratracheal instillation of collected materials, as used in this study, is widely accepted as a useful method for screening relative toxicity (Henderson et al., 1995Go), thus justifying further investigations. Although some variation of deposition into different lung lobes may occur with the instillation method, in our hands reasonably reproducible responses are achieved, comparable in value to those achieved following inhalation studies (Seagrave et al., in press).

The HE sample produced greater lung toxicity than the NE and NT samples for nearly all parameters when data were adjusted for the difference in emission rates. There was little difference between the NE and NT emission sample potencies on the basis of mass. Interestingly, the HE sample did not have the highest PAH concentrations (relative to total mass), but had somewhat higher proportions of hopanes and steranes, two classes of organic compounds that are unique to oil and were shown in our previous work to be statistically associated with lung toxicity of emission samples (McDonald et al., 2004Go).

Despite this study advancing our knowledge of the lung toxicity and mutagenicity of natural gas emissions, it has several limitations. The use of collected, acetone-extracted, and processed samples (as opposed to a study conducted by inhalation) was a limitation of this study. Although attempts were made to quantitatively remove 100% of the PM from the filters, only ~15–20% of the PM could be removed. This was much less than the expected efficiency, based on results obtained using the same procedures with gasoline and diesel emissions, perhaps suggesting very different composition of the original exhaust. There may be several explanations for the sub-optimal extraction. First, the organic solvent may not have efficiently removed the large amounts of metallic/ionic elements that were associated with the PM from the CNG vehicles. Second, the organic PM appeared to be stickier than previous emission samples, making removal of the material from the filters by mechanical agitation less effective. However, more vigorous physical manipulations were avoided to prevent fragmentation of the filters, that would have resulted in release of fibers into the suspensions. Additional losses of PM during evaporation/concentration may explain some of the differences as well.

Toxicity potency per unit of total mass (including that which could not be extracted) could be either greater or less than the values reported, depending on whether extraction was selective for more or less toxic agents. If the extracted material was representative of the whole mixture, the mass-based potency factors of the extracts would accurately represent the potency of the whole. If most of the toxic materials were extracted, and only relatively inert materials were left on the filter, the mass-based potency of the whole would be less than the potency of the extracts since the denominator (mass) would increase without increasing the numerator (toxicity). In contrast, if the most toxic constituents were not extracted, the potency of the whole would be greater than the potency of the extracts. However, the toxicity and mutagenicity, reported as potency factors per unit volume of emissions, are unlikely to be less than those indicated, since more effective extraction would only add constituents to the mixture without changing the denominator.

In summary, these data demonstrate that collected PM + SVOC emission samples from CNG-fueled vehicles contain both mutagenic and toxic constituents, and serve to place the potential hazard from these emissions in the context of potential health effects from diesel and gasoline-fueled vehicles' emissions.


    SUPPLEMENTARY DATA
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 SUPPLEMENTARY DATA
 REFERENCES
 
This section describes the conversion between potency factors based on different dose metrics (mass, volume, or other) for both linear dose-response functions and exponential dose-response functions. Supplementary data are available online at www.toxsci.oupjournals.org.


    ACKNOWLEDGMENTS
 
This work was supported by the Freedom CAR and Vehicle Technology Program of the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy. The authors gratefully acknowledge the contributions of Brenda Pacheco, Teresa Espindola, and the necropsy staff at Lovelace Respiratory Research Institute (LRRI) for excellent technical assistance. LRRI animal facilities are fully accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care International. This article was prepared as an account of work sponsored by an agency of the U.S. Government. Neither the U.S. Government nor any agency thereof, nor any of their employees, makes any warranty, expressed or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe on privately owned rights. Reference herein to any specific commercial product, process, service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the U.S. Government or any agency thereof. The views and opinions of the authors expressed herein do not necessarily state or reflect those of the U.S. Government or any agency thereof. No copyright is asserted in the works of U.S. Government employees. Conflict of interest: none declared.


    REFERENCES
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 SUPPLEMENTARY DATA
 REFERENCES
 
Ames, B. N., McCann, J., and Yamasaki, E. (1975). Methods for detecting carcinogens and mutagens with the Salmonella/mammalian-microsome mutagenicity test. Mutat. Res. 31, 347–364.[ISI][Medline]

Arey, J., Zielinska, B., Atkinson, R., and Winer, A. M. (1988). Formation of nitroarenes during ambient high-volume sampling. Environ. Sci. Technol. 22, 457–462.[CrossRef][ISI]

Brauer, M., Hoek, G., van Vliet, P., Meliefste, K., Fischer, P. H., Wijga, A., Koopman, L. P., Neijens, H. J., Gerritsen, J., Kerkhof, M., Heinrich, J., Bellander, T., and Brunekreef, B. (2002). Air pollution from traffic and the development of respiratory infections and asthmatic and allergic symptoms in children. Am. J. Respir. Crit. Care Med. 166, 1092–1098.[Abstract/Free Full Text]

Brunekreef, B., Janssen, N. A., de Hartog, J., Harssema, H., Knape, M., and van Vliet, P. (1997). Air pollution from truck traffic and lung function in children living near motorways. Epidemiology 8, 298–303.[CrossRef][ISI][Medline]

Chow, J. C., Watson, J. G., Crow, D., Lowenthal, D. H., and Merrifield, T. (2001). Comparison of IMPROVE and NIOSH carbon measurements. Aerosol Sci. Technol. 34, 23–34.[CrossRef][ISI]

Docekal, B., Krivan, V., Pelz, N., and Fresenius, J. (2004). Anal. Chem. 343, 873–878.

Englert, N. (2004). Fine particles and human health–a review of epidemiological studies. Toxicol. Lett. 149, 235–242.[CrossRef][ISI][Medline]

Finkelstein, M. M. (2003). Mortality and indicators of traffic-related air pollution. Lancet 361, 430.

Gay, R. J., McComb, R. B., and Bowers, Jr., G. N., (1968). Optimum reaction conditions for human lactate dehydrogenase isoenzymes as they affect total lactate dehydrogenase activity. Clin. Chem. 14, 740–753.[Abstract/Free Full Text]

Henderson, R. F., Driscoll, K. E., Harkema, J. R., Lindenschmidt, R. C., Chang, I. Y., Maples, K. R., and Barr, E. B. (1995). A comparison of the inflammatory response of the lung to inhaled versus instilled particles in F344 rats. Fundam. Appl. Toxicol. 24, 183–197.[CrossRef][ISI][Medline]

Hochberg, Y. (1988). A sharper Bonferroni procedure for multiple tests of significance. Biometrika 75, 800–802.[ISI]

Lapin, C. A., Gautam, M., Zielinska, B., Wagner, V. O., and McClellan, R. O. (2002). Mutagenicity of emissions from a natural gas fueled truck. Mutat. Res. 519, 205–209.[ISI][Medline]

Maron, D. M., and Ames, B. N. (1983). Revised methods for the Salmonella mutagenicity test. Mutat. Res. 113, 173–215.[CrossRef][ISI][Medline]

McCormick, R. L., Graboski, M. S., Alleman, T. L., and Yanowitz, J. (2000). Idle emissions from heavy-duty diesel and natural gas vehicles at high altitude. J. Air Waste Manag. Assoc. 50, 1992–1998.[ISI][Medline]

McDonald, J. D., Eide, I., Seagrave, J., Zielinska, B., Whitney, K., Lawson, D. R., and Mauderly, J. L. (2004). Relationship between composition and toxicity of motor vehicle emission samples. Environ. Health Perspect. 112, 1527–1538.[ISI][Medline]

Neter, J., Kutner, M. H., Nactsheim, C. J., and Wasserman, W. (1996). Applied Linear Statistical Methods. Irwin, Chicago.

Nightingale, J. A., Maggs, R., Cullinan, P., Donnelly, L. E., Rogers, D. F., Kinnersley, R., Fan, C. K., Barnes, P. J., Ashmore, M., and Newman-Taylor, A. (2000). Airway inflammation after controlled exposure to diesel exhaust particulates. Am. J. Respir. Crit. Care Med. 162, 161–166.[Abstract/Free Full Text]

Nordenhall, C., Pourazar, J., Blomberg, A., Levin, J. O., Sandstrom, T., and Adelroth, E. (2000). Airway inflammation following exposure to diesel exhaust: A study of time kinetics using induced sputum. Eur. Respir. J. 15, 1046–1051.[Abstract/Free Full Text]

Rudell, B., Blomberg, A., Helleday, R., Ledin, M. C., Lundback, B., Stjernberg, N., Horstedt, P., and Sandstrom, T. (1999). Bronchoalveolar inflammation after exposure to diesel exhaust: Comparison between unfiltered and particle trap filtered exhaust. Occup. Environ. Med. 56, 527–534.[Abstract]

Salvi, S., Blomberg, A., Rudell, B., Kelly, F., Sandstrom, T., Holgate, S. T., and Frew, A. (1999). Acute inflammatory responses in the airways and peripheral blood after short-term exposure to diesel exhaust in healthy human volunteers. Am. J. Respir. Crit. Care Med. 159, 702–709.[Abstract/Free Full Text]

Seagrave, J. C., McDonald, J. D., Gigliotti, A. P., Nikula, K. J., Seilkop, S. K., Gurevich, M., and Mauderly, J. L. (2002). Mutagenicity and in vivo toxicity of combined particulate and semivolatile organic fractions of gasoline and diesel engine emissions. Toxicol. Sci. 70, 212–226.[Abstract/Free Full Text]

Seagrave, J. C., McDonald, J. D., Reed, M. D., Seilkop, S. K., and Mauderly, J. L. (2005). Responses to subchronic inhalation of low concentrations of diesel exhaust and hardwood smoke measured in rat bronchoalveolar lavage fluid. Inhal. Toxicol. 17, in press.

Searle, S. R. (1971). Linear Models. J. Wiley and Sons, New York.

Stenfors, N., Nordenhall, C., Salvi, S. S., Mudway, I., Soderberg, M., Blomberg, A., Helleday, R., Levin, J. O., Holgate, S. T., Kelly, F. J., Frew, A. J., and Sandstrom, T. (2004). Different airway inflammatory responses in asthmatic and healthy humans exposed to diesel. Eur. Respir. J. 23, 82–86.[Abstract/Free Full Text]

Watanabe, N., Kamei, S., Ohkubo, A., Yamanaka, M., Ohsawa, S., Makino, K., and Tokuda, K. (1986). Urinary protein as measured with a pyrogallol red-molybdate complex, manually and in a Hitachi 726 automated analyzer. Clin. Chem. 32, 1551–1554.[Abstract/Free Full Text]

Zielinska, B., Sagebiel, J., Arnott, W. P., Rogers, C. F., Kelly, K. E., Wagner, D. A., Lighty, J. S., Sarofim, A. F., and Palmer, G. (2004). Phase and size distribution of polycyclic aromatic hydrocarbons in diesel and gasoline vehicle emissions. Environ. Sci. Technol. 38, 2557–2567.[CrossRef][ISI][Medline]





This Article
Abstract
Full Text (PDF)
Supplementary Data
All Versions of this Article:
87/1/232    most recent
kfi230v1
Alert me when this article is cited
Alert me if a correction is posted
Services
Email this article to a friend
Similar articles in this journal
Similar articles in ISI Web of Science
Similar articles in PubMed
Alert me to new issues of the journal
Add to My Personal Archive
Download to citation manager
Disclaimer
Request Permissions
Google Scholar
Articles by Seagrave, J.
Articles by Mauderly, J. L.
PubMed
PubMed Citation
Articles by Seagrave, J.
Articles by Mauderly, J. L.