Bioassay-directed fractionation and salmonella mutagenicity of automobile and forklift diesel exhaust particles.

Many pulmonary toxicity studies of diesel exhaust particles (DEPs) have used an automobile-generated sample (A-DEPs) whose mutagenicity has not been reported. In contrast, many mutagenicity studies of DEPs have used a forklift-generated sample (SRM 2975) that has been evaluated in only a few pulmonary toxicity studies. Therefore, we evaluated the mutagenicity of both DEPs in Salmonella coupled to a bioassay-directed fractionation. The percentage of extractable organic material (EOM) was 26.3% for A-DEPs and 2% for SRM 2975. Most of the A-EOM (~55%) eluted in the hexane fraction, reflecting the presence of alkanes and alkenes, typical of uncombusted fuel. In contrast, most of the SRM 2975 EOM (~58%) eluted in the polar methanol fraction, indicative of oxygenated and/or nitrated organics derived from combustion. Most of the direct-acting, base-substitution activity of the A-EOM eluted in the hexane/dichloromethane (DCM) fraction, but this activity eluted in the polar methanol fraction for the SRM 2975 EOM. The direct-acting frameshift mutagenicity eluted across fractions of A-EOM, whereas > 80% eluted only in the DCM fraction of SRM 2975 EOM. The A-DEPs were more mutagenic than SRM 2975 per mass of particle, having 227 times more polycyclic aromatic hydrocarbon-type and 8-45 more nitroarene-type mutagenic activity. These differences were associated with the different conditions under which the two DEP samples were generated and collected. A comprehensive understanding of the mechanisms responsible for the health effects of DEPs requires the evaluation of DEP standards for a variety of end points, and our results highlight the need for multidisciplinary studies on a variety of representative samples of DEPs.

Long-term exposure to particulate air pollution, including diesel exhaust, is associated with an increasing incidence of respiratory allergy, cardiopulmonary mortality, and risk of lung cancer (Pope et al. 2002;Sydbom et al. 2001). Although the health effects of diesel exhaust particles (DEPs) have been studied for many years (Lewtas 1982), a comprehensive identification of the chemical components responsible for the biologic effects and a full understanding of the underlying mechanisms remain incomplete (Mauderly 2001;Rosenkranz 1996).
The two most-studied health effects of DEPs are pulmonary toxicity and mutagenicity, and different chemical and physical features of DEPs have been associated with the induction of these two end points. For mutagenicity, early studies suggested that nitroarenes were a primary class of mutagens in organic extracts of DEPs (Austin et al. 1985;Claxton 1981;Claxton and Huisingh 1980); analytical studies confirmed this (Paputa-Peck et al. 1983;Salmeen et al. 1982) and also identified a role for polycyclic aromatic hydrocarbons (PAHs) (Rosenkranz 1996). For pulmonary effects, some PAHs have been shown to enhance pro-inflammatory and allergic responses induced by DEPs in the airways (Diaz-Sanchez 1997;Kawasaki et al. 2001;Tsien et al. 1997). However, the size and surface reactivity of particles also may play a role in the induction of pulmonary effects (Dick et al. 2003;Donaldson et al. 1996).
Many studies of the mutagenicity of DEPs have been conducted using a standard reference material (SRM), such as SRM 2975, which was derived from a forklift truck and developed by the National Institute of Standards and Technology (NIST) (Claxton et al. 1992;Hughes et al. 1997); however, only a few studies have characterized the pulmonary effects of this sample (Lovik et al. 1997;Madden et al. 2000). In contrast, many studies on the pulmonary toxicity of DEPs have used an automobile-derived DEP sample (A-DEP) (Kobayashi and Ito 1995;Sagai et al. 1993), but the mutagenicity of this DEP sample has not been reported. Although Seagrave et al. (2002) have examined the same DEPs for pulmonary effects as well as mutagenicity, no one has done so for the extensively studied A-DEP sample.
The chemical composition of DEPs is influenced by the age and type of engine, fuel composition, load characteristics, lube oil components, presence and efficiency of control devices, and sampling procedures (Claxton 1983;Mauderly 2001;Rosenkranz 1996;Schuetzle 1983). Given that these factors vary for the SRM and A-DEP samples, we reasoned that the biologic activities of these DEPs were likely to be different. To assess the potential impact of these differences on mutagenicity, we evaluated the two DEP samples using a bioassay-directed fractionation (Schuetzle and Lewtas 1986) coupled with the Salmonella mutagenicity assay.
Various DEP samples have been subjected to such analyses since the first report (Huisingh et al. 1979), including an earlier SRM of DEPs, SRM 1650 (Savard et al. 1992). In this study, we sequentially extracted an organic extract of each DEP with solvents of increasing polarity on a silica-gel column and then evaluated the fractions for mutagenicity in various strains of Salmonella. The results were expressed as the distribution of mass and mutagenicity among the fractions. Combined with physical and chemical features of these DEPs, as well as their pulmonary toxicities (Singh et al. 2004), we propose that DEP samples should be characterized chemically, physically, and biologically at multiple end points to understand the mechanisms associated with the health effects of DEPs.

Generation and collection of DEPs.
A-DEPs were provided by one of the authors (T.K.), and the generation and collection conditions of these DEPs have been described previously (Kobayashi and Ito 1995;Sagai et al. 1993). Briefly, DEPs were collected "cold" at a sampling temperature of 50°C onto glass-fiber filters (GD-100R, 203 mm × 254 mm) in a constant-volume sampling system fitted at the end of a dilution tunnel. The particles were generated using a light-duty (2,740 cc), 4-cylinder, 4JB1-type Isuzu diesel engine. The engine had a torque load of 6 kg/m generated by an Many pulmonary toxicity studies of diesel exhaust particles (DEPs) have used an automobilegenerated sample (A-DEPs) whose mutagenicity has not been reported. In contrast, many mutagenicity studies of DEPs have used a forklift-generated sample (SRM 2975) that has been evaluated in only a few pulmonary toxicity studies. Therefore, we evaluated the mutagenicity of both DEPs in Salmonella coupled to a bioassay-directed fractionation. The percentage of extractable organic material (EOM) was 26.3% for A-DEPs and 2% for SRM 2975. Most of the A-EOM (~55%) eluted in the hexane fraction, reflecting the presence of alkanes and alkenes, typical of uncombusted fuel. In contrast, most of the SRM 2975 EOM (~58%) eluted in the polar methanol fraction, indicative of oxygenated and/or nitrated organics derived from combustion. Most of the directacting, base-substitution activity of the A-EOM eluted in the hexane/dichloromethane (DCM) fraction, but this activity eluted in the polar methanol fraction for the SRM 2975 EOM. The directacting frameshift mutagenicity eluted across fractions of A-EOM, whereas > 80% eluted only in the DCM fraction of SRM 2975 EOM. The A-DEPs were more mutagenic than SRM 2975 per mass of particle, having 227× more polycyclic aromatic hydrocarbon-type and 8-45× more nitroarenetype mutagenic activity. These differences were associated with the different conditions under which the two DEP samples were generated and collected. A comprehensive understanding of the mechanisms responsible for the health effects of DEPs requires the evaluation of DEP standards for a variety of end points, and our results highlight the need for multidisciplinary studies on a variety of representative samples of DEPs. ECDY dynamometer (Meiden-Sya, Tokyo, Japan) and was run at 2,000 rpm. SRM 2975 was generated by a forklift truck and was purchased from NIST (Gaithersburg, MD, USA). The DEPs were generated by a heavy-duty diesel engine and collected using a filtering system designed for diesel forklifts under "hot" conditions without a dilution tunnel at the Donaldson Company, Inc. (Minneapolis, MN, USA; personal communication). The certified analyses of these particles are available (NIST 2000).
Organic extractions and fractionation. DEPs were sonicated for 20 min in dichloromethane (DCM) at 2× the estimated volume of the particles, and the tube was centrifuged at approximately 2,000 rpm for 10 min. The solvent was transferred to another glass tube, and the extraction was repeated two more times. The pooled solvent extract was concentrated, and the percentage of extractable organic material (EOM) was determined by gravimetric measurement. The remaining extract was concentrated to 1 mL and readjusted to 5 mL with hexane.
Silica gel (10 g of grade 62, 60-200 mesh) was added to a 40 × 300 mm open column with a medium-porosity ground-glass frit. The silica was washed with DCM followed by hexane. The extract was added to the column, and the EOM was eluted serially with hexane, 50:50 hexane:DCM, DCM, and methanol. Each fraction was then concentrated under nitrogen, and the mass of EOM for each fraction was determined as described above for the whole extract. Stock solutions at 20 µg of EOM/mL in dimethyl sulfoxide (DMSO) of the whole extract and each fraction were prepared for bioassay by solvent exchange.
Aroclor-induced Sprague-Dawley rat liver S9 was obtained from Moltox (Boone, NC, USA) and used at 1 mg S9 protein/plate. Plates were incubated for 3 days, the colonies were counted, and linear regressions were calculated over the linear portion of the dose-response curves to determine the mutagenic potencies expressed as revertants (rev) per microgram.

Results
Mutagenic potencies of EOM. The basis for the interpretation of the mutagenicity data is summarized in Table 1, and the mutagenicity dose-response data for the EOM samples are shown in Table 2. The mutagenic potencies derived from the dose-response data for both the EOM and particles are shown in Table 3.
The A-EOM exhibited 18× more PAH-type mutagenic activity (TA100 +S9) than did the SRM 2975 EOM (Table 3). In the absence of S9, the base-substitution mutagenic potency (TA100) of the SRM 2975 EOM was 3× greater than that of the A-EOM. The two EOM samples had rather similar frameshift mutagenic potencies (TA98); however, the SRM 2975 EOM had 2.3× more frameshift activity in the absence of S9 than in the presence of S9 (Table 3). Also, the SRM 2975 EOM had 5× more frameshift potency than base-substitution potency in the presence of S9. In the absence of S9, the rankings of the mutagenic potencies among the strains, with the exception of TA100, were similar for the two EOM samples (Table 3). did the A-EOM. However, based on data from strain YG1021, the opposite result was found, with the A-EOM having approximately 174× more nitroarene-type mutagenic activity than did the SRM 2975 EOM. Mutagenic potencies of particles. To convert the mutagenic potencies of the EOM (rev per microgram of EOM) to mutagenic potencies of the particles (rev per microgram of particle), the potencies of the EOM were multiplied by the percent EOM of the particle. The percent EOM was 26.3% for A-DEPs and 2% for SRM 2975. Based on these calculations, the A-DEPs had 227× more PAH-type mutagenic activity (TA100 +S9) than did the SRM 2975 particles. In addition, the A-DEPs had approximately 8× more nitroarene-type activity than did the SRM 2975 particles based on data from TA98NR, TA98/1,8-DNP6, or YG1024 (Table 4). Based on data from strain YG1021, the A-DEPs had even more (45×) nitroarene-type mutagenic activity than did the SRM 2975 particles. The A-DEPs had greater mutagenic potency in both TA98 and TA100 than did the SRM 2975 particles (Table 3), perhaps due to higher amounts of nitroareneand aromatic amine-type mutagenic activity in the A-DEP sample relative to the SRM 2975 sample. Except for the juxtaposition of TA98 and TA100, the mutagenic potency rankings of the two DEPs were similar among the strains (Table 3).
Mutagenic potencies of fractions of EOM. The dose-response data for the fractions are shown in Table 5, and the mutagenic potencies of the fractions of the two EOM are shown in Table 6. For example, in TA100 +S9, the most potent A-EOM eluted in the hexane/DCM and DCM fractions, whereas the most potent SRM 2975 EOM eluted in the DCM fraction. Thus, the classes of compounds accounting for S9-dependent, basesubstitution mutagenicity in the A-EOM were less polar than were those in the SRM 2975 EOM. When comparing the reduction in mutagenic potencies in TA98NR relative to TA98, which is an indication of the presence of nitroarenes, the greatest reduction for the SRM 2975 EOM occurred in the hexane/DCM fraction, whereas the greatest reduction for the A-EOM occurred in the DCM and methanol fractions, which are much more polar than hexane/DCM. A variety of other differences of this sort can be noted by comparing the results in Table 6.
Distribution of recovered mass and mutagenicity of EOM. Most (84%) of the mass of the A-EOM and all (103%) of the mass of the SRM 2975 EOM were recovered from the silica-gel column after the fractionation (Tables 7  and 8). However, the distribution of the mass across the fractions was the opposite for the two EOM (Tables 7 and 8, Figure 1). Thus, approximately 55% of the A-EOM eluted in the hexane fraction and approximately 33% in the highly polar methanol fraction, whereas approximately 29% of the SRM 2975 EOM eluted in the hexane fraction and approximately 58% in the methanol fraction.
Likewise, the distribution of mutagenicity across the fractions was different for the two EOMs. For example, 3× more PAH-type mutagenic activity (TA100 +S9) eluted in the hexane/DCM fraction from the A-EOM than from the SRM 2975 EOM (Tables 7 and 8, Figure 2), confirming our other data that the A-DEP sample had more PAH-type mutagenic activity than did the SRM 2975. The distribution of direct-acting, base-substitution mutagenic activity (TA100 -S9) was completely the opposite for the two EOMs, with most of this activity eluting in the hexane/DCM fraction for the A-EOM but in the polar methanol fraction for the SRM 2975 EOM (Tables 7 and 8). The direct-acting frameshift mutagenic activity requiring minimal nitroreductase (TA98NR -S9) was due to compounds having a range of polarity for the A-EOM, because this activity  A-DEP ~174× > SRM A-DEP 45× > SRM a Comparisons were made by determining the difference in mutagenic potency for each EOM or particle in Table 3 between TA98 -S9 and the strains listed above -S9. For example, for EOM in TA98NR, 138.5 -15.5 (rev/µg in TA98 -TA98NR) = 123 rev/µg; for SRM 2975 this was 218.1 -27.3 = 190.8 rev/µg. Then, 190.8 ÷ 123 = 1.6; thus, based on data in TA98NR, SRM 2975 EOM had 1.6× more nitroarene-type mutagenic activity than did A-EOM.  was dispersed across three fractions, whereas for the SRM 2975 EOM, > 80% of this activity eluted in only the DCM fraction. Despite the many differences, one feature was identical for both EOM: neither had detectable mutagenic activity in the hexane fraction. This was true even for the A-EOM, despite approximately 55% of its mass eluting in the hexane fraction.

Mutagenicity of EOM and DEPs.
DCM is the most effective solvent for the extraction of mutagenic organics from diesel exhaust (Montreuil et al. 1992;Petersen and Chuang 1982); therefore, it was used here to prepare the initial organic extract of both DEP samples. Although the mutagenicity of SRM 2975 EOM in several strains of Salmonella had been reported previously (Hughes et al. 1997), this sample had not been subjected previously to a bioassay-directed fractionation of the type described here, and no mutagenicity data had been reported previously for the A-EOM. With regard to the whole, unfractionated SRM 2975 EOM, the mutagenic potency ranking in the absence of S9 among the strains obtained here was similar to that obtained previously by Hughes et al. (1997) except that our sample ranked as more potent in TA100 than that of Hughes et al. (1997). One possible reason is that Hughes et al. (1997) evaluated the soxhlet extract of SRM 2975 purchased from NIST, whereas we prepared our own extract by sonication from SRM 2975. In the absence of S9, the mutagenic potency of the SRM 2975 EOM was generally greater than that of the A-EOM; the opposite was the case in the presence of S9 (Table 3). However, when the mutagenic potencies of the EOM were combined with the percent EOM of the particles, the mutagenic potency of the A-DEPs per mass of particle was greater in all strains than that of the SRM 2975 DEP regardless of S9 (Table 3). This occurred partly because the A-DEPs had > 10× the percent EOM than did SRM 2975. The A-DEPs had 227× more PAH-type activity and approximately 8-45× more nitroarene-type activity than did the SRM 2975 particles. The A-DEPs also had 3-21× more frameshift mutagenic activity than did the SRM 2975 particles, which was most likely due to the excess amount of nitroarene and possibly aromatic amine activity in the A-DEPs. Considering the various strains used to infer the proportion of nitroarene-type mutagenic activity in the samples, all the strains but YG1021 led to similar conclusions (Table 4). Perhaps the additional nitroreductase present in YG1021 activated many compounds present at low concentrations that were not activated by the normal levels of nitroreductase present in TA98. Thus, continued caution must be exercised regarding inferences about the role of nitroarenes in the mutagenicity of DEPs using these and other such strains (Rosenkranz 1981).
Mutagenicity of fractions of EOM. Various methods have been used to fractionate organic extracts of diesel particles, including acid/base/neutral fractionation procedures (Crebelli et al. 1991) and solid-phase extractions using Sephadex (Bechtold et al. 1985) or silica gel (Hayakawa et al. 1997;Strandell et al. 1994). Hexane is a neutral solvent in which chemical classes such as alkanes and alkenes elute (Hayakawa et al. 1997), whereas methanol is a highly polar solvent in which Article | Mutagenicity of diesel exhaust particles Environmental Health Perspectives • VOLUME 112 | NUMBER 8 | June 2004 817  Table 6 by the number of micrograms of EOM recovered for each fraction as noted in the second column of this table. These values, rev/fraction, were then expressed as a percentage relative to the sum (Σ) of the recovered mass of the fractions noted in the second column of this table.  Table 6 by the number of micrograms of EOM recovered for each fraction as noted in the second column of this table. These values, rev/fraction, were then expressed as a percentage relative to the sum (Σ) of the recovered mass of the fractions noted in the second column of this table.  compounds such as oxygenated aromatics elute (Strandell et al. 1994). PAHs, aromatic amines, and nitroarenes can elute in the hexane/DCM and DCM fractions (Hayakawa et al. 1997;Schuetzle 1983). Hayakawa et al. (1997) showed that as much as 53% of the mutagenicity of the DCM fraction of DEPs was due to nitroarenes. As summarized by Singh et al. (2004), the two DEP samples had different physical properties, chemical compositions, and pulmonary toxicities. The mass and mutagenicity distributions of the two samples described here reflected clear differences in the chemical composition of these DEPs. The mass distributions of the fractionated extracts of the two DEP samples were the opposite of each other, with most of the A-EOM eluting in the hexane fraction but most of the SRM 2975 EOM eluting in the highly polar methanol fraction. The lack of mutagenic activity in the hexane fraction (Tables 7 and 8) was consistent with the presence of unsubstituted alkanes and alkenes, which are not mutagenic. These results are also supported by the photomicrographs and other chemical and combustion data demonstrating that much of the A-EOM is uncombusted fuel, possibly neutral alkanes and alkenes (Singh et al. 2004), which would have eluted in the hexane fraction.
As reported previously (Hayakawa et al. 1997), the sum of the mutagenic potencies of the EOM across the fractions was generally much higher than that of the whole, unfractionated material, and this was true for all the strains used here. For example, in TA98 -S9, the mutagenic potency of the A-EOM was 138.5 rev/µg (Table 3), but the sum of the potencies of the fractions in this strain was 526.8 rev/µg (Table 7). Likewise, the mutagenic potency of the SRM 2975 EOM was 218.1 rev/µg, but that of the sum of its fractions was 3018.5 rev/µg, an approximately 14× increase. This showed that the fractionation unmasked the mutagenic potency of the compounds by separating some of the nonmutagenic cytotoxic compounds from the mutagenic compounds in the mixture. The ability of the fractionation to separate a large amount of nonmutagenic mass from mutagenic mass (Tables 7 and 8, Figure 1) is one of the goals of a successful bioassay-directed fractionation (Schuetzle and Lewtas 1986).
As shown previously, the distribution of mutagenic activity of the EOM across chemical fractions can vary depending on the type of engine (Clark et al. 1981;Sjogren et al. 1996), type of fuel (Clark et al. 1982;Crebelli et al. 1995;Sjogren et al. 1996;Westerholm et al. 2001), the running conditions (Bechtold et al. 1984;Courtois et al. 1993), and the collection conditions (Claxton and Barnes 1981;Lies et al. 1986). As demonstrated by extensive studies, much of the mutagenic activity of diesel exhaust is due to PAHs and nitroaromatics, especially nitropyrenes (Austin et al. 1985;Claxton 1981Claxton , 1983Lies et al. 1986;Nakagawa et al. 1983;Rosenkranz 1996;Tokiwa and Ohnishi 1986). An important role for direct-acting acid/neutral compounds that are not nitropyrenes has also been noted (Crebelli et al. 1991). In this regard, the directacting, base-substitution mutagenicity (TA100 -S9) of the two EOM eluted in opposite fractions, indicating that different chemicals accounted for this activity for the two samples.
The mutagenic potencies of most of the fractions of the A-EOM were enhanced by S9 (Table 6), whereas the opposite result was found for the SRM 2975 EOM. This indicated that, across chemical classes, A-EOM contained more S9-dependent mutagenicity than did the SRM 2975 EOM. Consistent with this is the evidence presented here and by Singh et al. (2004) for greater amounts of PAH-type mutagenicity and PAHs in the A-DEPs compared with SRM 2975. There is ample evidence that the genotoxic activity associated with the EOM of DEPs inhaled into the lung may be bioavailable by virtue of the solubilization and dispersion properties of pulmonary surfactant components (Belisario et al. 1984;Keane et al. 1991;King et al. 1981).

Conclusions
A summary of some of the relative differences between the two DEPs (Table 9) shows that they have disparate mutagenic activities and chemical compositions due to the different conditions under which they were generated and collected. The fact that one sample (A-DEPs) has been used extensively in pulmonary toxicity studies but never studied previously for mutagenicity, and virtually the opposite situation pertains for the other sample (SRM 2975), illustrates the need for scientists to engage in collaborative, multidisciplinary research efforts in this area. Similar to the mutagenic activities of these particles, the pulmonary toxicities of these two DEPs were also strikingly different (Singh et al. 2004). These biologic data, combined with the physical and chemical features of these two DEPs (Singh et al. 2004), provide a basis for comparing these two DEPs that was not available previously.
A screening battery for a variety of DEPs involving pulmonary toxicity and mutagenicity has been proposed by Seagrave et al. (2002), and this could be extended to include a bioassay-directed fractionation as shown here. Until comparative data among a variety of DEP samples for various end points are available, a comprehensive understanding of the mechanisms associated with the health effects of any DEPs will be hindered (HEI 2002).