Bacterial mutagenicity of pyrolysis tars produced from chloro-organic fuels.

Droplets of toluene and three chlorinated organics, ortho-dichlorobenzene, 1,2-dichloroethane, and trichloroethylene, were pyrolyzed in pure nitrogen. The composition and bacterial mutagenicity of the product tars were measured. The presence of organic chlorine was found to affect both pyrolysis product tar composition and total tar mutagenicity. Pyrolysis in the absence of chlorine produced tars whose bacterial mutagenicity was found to be largely due to the presence of cyclopenta[cd]pyrene, fluoranthene, and benzo[a]pyrene. Small amounts of chlorine in the fuel (i.e., Cl/H molar ratios of less than 0.3) enhanced the formation of highly condensed polycyclic aromatic hydrocarbons (including cyclopenta[cd]pyrene) and increased tar mutagenicity. Larger amounts of organic chlorine (Cl/H ratios of between 0.3 and 0.6) resulted in significant yields of mono- and dichlorinated aromatics and higher levels of tar mutagenicity, which could not be accounted for by the presence of mutagens produced by pyrolysis in the absence of chlorine. Furthermore, unlike tars containing little or no chlorine, tars containing aryl chlorine were more mutagenic in the absence of added enzymes (intended to mimic in vivo mammalian metabolism) than in their presence. We hypothesize that at least one of the chlorinated aromatic products is strongly mutagenic. Two specific conditions that gave notably different results were a) the low-temperature (i.e., below 1400 K) pyrolysis of ortho-dichlorobenzene, which produced tri- and tetrachlorinated biphenyls almost exclusively; and b) the chlorine-rich pyrolysis of trichloroethylene, during which mostly perchloroaromatics were formed. Neither of these tars was found to mutate bacteria.


Introduction
Many of the polycyclic aromatic compounds (PACs) condensed as tar on the surface of soot are mutagenic (1,2) or carcinogenic (3,4). The types and amounts ofthese toxic combustion by-products that are produced by pyrolysis in diffusion flames are functions ofthe elemental and structural composition ofthe fuel and the conditions governing the rate and extent offuel conversion during pyrolysis. In the case of incineration, the presence oforganically bound chlorine in the waste stream can have a significant effect on pyrolysis chemistry. The pyrolysis of chloro-organics has been found to produce chlorinated aromatics, with yields dependent largely on the molecular structure ofthe fuel at low temperatures and on the elemental content ofthe fuel at high temperatures (5). Pyrolysis in a chlorine-rich environment produces perchlorocompounds (6). The presence ofchlorine in fuels has also been found to promote aromatic condensation and soot formation during pyrolysis, resulting in tars that contain a large fraction ofperi-fused aromatic structures (7).
Few studies have directly linked incinerator emissions with health impact, however. In one epidemiological investigation that did, it was found that the frequencies of dairy cattle and human twinning increased in areas near incinerators burning municipal and chemical wastes containing chlorine (18). It was hypothesized that the presence ofpolychlorinated hydrocarbons, some ofwhich have estrogenic and fertility-related properties, may account for this association; however, causality could not be established. With the limited amount ofdata from field studies, controlled laboratory studies using bioassays as indicators of potential health risk are needed to establish incineration guidelines and regulations.
Bacterial mutation assays represent one method for assessing overall health risk associated with a complex sample; their use has been reviewed by Lewtas (19). Recently, these assays have been used to measure the induced mutation ofbacteria by products ofethylene combustion (20), coal pyrolysis (21), municipal waste incineration (22)(23)(24), and agricultural waste incineration (25,26). Biological activity, like chemical reactivity, is a function of molecular structure. It has been shown that factors governing the biological behavior of PACs include the size and configuration of the parent aromatic molecule (27,28) and the presence or absence of heteroatoms or substituent groups (29)(30)(31)(32)(33)(34). Further complicating matters, the frequency of mutation is also dependent on the set of catalysts (enzymes) that are present in an exposed cell. In the human body, this is best considered an idiosyncratic set for each cell (tissue) type. Bacteria have their own set of metabolizing enzymes, providing one measure of induced mutation when exposed to chemicals. Enzymes can also be added to cell cultures in bioassay experiments in an attempt to mimic specific types of mammalian metabolism.
Although the pyrolytic transformation of fuel to products results in a large increase in the number of effluent chemicals, it is generally believed that the number of mutagens contributing significantly to the total mutagenicity is small. Of PAHs formed during the high-temperature combustion of hydrocarbons, the major mutagens that have been found are cyclopenta[cd]pyrene (CPP), fluoranthene, and benzo[a]pyrene. Little is known, however, about the mutagenicity of pyrolysis products of chlorinated organics. DeMarini et al. (23) found that the mutagenic potency of rotary kiln emissions, as measured by a bacterial mutation assay with metabolic activation via the addition of mammalian enzymes, was reduced by the presence ofpolyvinylchloride and carbon tetrachloride in the feed stock. The chemical composition of the chlorinated organic emission fraction was not measured, however. In analysis oforganic emissions oftwo municipal waste incinerators burning polyethylene plastic, DeMarini et al. (24) found mutagenic activity to be greater in the absence of mammalian enzymes than in their presence. Furthermore, they found that the moderately polar fraction ofthe organic emissions was responsible for the mutagenic activity. As in the rotary kiln study, a detailed chemical analysis of the emissions was not reported.
In this paper, effects ofchlorine on pyrolysis product pathways are reviewed; these results have been reported elsewhere (5-7). We then discuss measurements ofbacterial mutation induced by the pyrolysis tar extracts, comparing them with the total contribution of all identified mutagens. Based on this comparison, we speculate on the presence of new mutagens in the chloro-organic pyrolysis tars. We also consider the effects on mutagenicity ofthe enhanced formation of pen-fused PAH mutagens during pyrolysis ofchlorinated organics and the presence ofchlorinated arenes in the tars.

Methods
Reactor and Sampling Apparatus Streams of monodisperse droplets, generated by a vibrating orifice device, were injected down the center of a laminar flow, isothermal drop-tube reactor in an environment ofpure nitrogen.
A metered stream ofdroplets, 40 Iam in diameter and 2.5 droplet diameters in spacing, was injected coaxially with nitrogen, preheated to furnace wall temperatures ranging from 1100 to 1500 K. Average gas residence times ofbetween 1.1 and 1.5 sec were maintained, sufficient for complete vaporization ofthe fuel droplets. A collection probe with gas quench and wall purge was used to channel the entire product stream through aO.2 um pore, mesh filter at ambient temperature. Tar yield was measured gravimetrically, after extraction from the filter sample by solubility in dichloromethane (DCM). Soot, defined here as the DCM-insoluble fraction, was determined by difference. This equipment and these procedures are described in greater detail elsewhere (5-7).
Toluene, three chlorinated organics, and equal volume mixtures of toluene and the chlorinated organics were pyrolyzed. The chlorinated compounds were ortho-dichlorobenzene (o-C6H4C12), 1,2-dichloroethane (1, 2-C2H4C12), and trichloroethylene (C2HCl3). Selection of these compounds was based on their physical properties (e.g., viscosity, boiling point, and density) being similar to those oftoluene and their composition and structure allowing for the study of effects of chlorine type (aliphatic versus aromatic) and amount (chlorine-to-hydrogen molar ratios ranging from 0 to 3).

Analytical Chemistry Instrumentation
The tar fractions were subjected to the following analyses: gas chromatography (GC) with flame ionization detection (FID), with mass spectroscopy (MS) and with Fourier transform infrared detection (FTIR), and high-performance liquid chromatography (HPLC) with ultraviolet-visible (UV-vis) spectrometric detection. All of the GC work was performed on Hewlett-Packard Model 5890A systems with Quadrex methyl (5 % phenyl) silicone fused-silica open tubular columns, using a helium gas mobile phase with identical flow and temperatureramping protocol. This allowed for easy cross-referencing of spectral data by GC retention time. The HPLC analysis was performed on a Hewlett-Packard Model 1090 system with a 190 to 600 nm diode array detector and a Vydac reverse-phase column.

Bacterial Mutation Assay
Forward-mutation assay to 8-azaguanine resistance in Salmonella typhimurium strain TM677 was used to measure the mutagenicities of total tar samples, as described previously (35,36); a brief summary is given here. The solvent containing the pyrolysis tar product sample was exchanged from dichloromethane (DCM) to dimethyl sulfoxide (DMSO). These test samples, at concentrations ranging from 10 to 300 jig/mL, were then exposed to exponentially growing bacteria in the presence and in the absence of 5 % (v/v) Aroclor induced postmitochondrial supernatant (PMS). After 2 hr, aliquots ofthe cell cultures were plated in the presence and in the absence ofthe selective agent (8-azaguanine, 50 ig/mL). Two independent cultures were used for each treatment point.
Colonies were counted after 48 hr. The mutant fraction was determined as the number of colonies formed under selective conditions divided by the number ofcolonies formed under nonselective conditions. If this ratio was larger than that found for simultaneous untreated control cultures (n = 2) with greater than 99 % confidence, and if the ratio also exceeded the 95 % upper confidence limit of the mutant fraction for the cumulative historical control (n 2 1000), the test was considered positive. The induced mutant fraction was calculated by subtracting the background mutation level. Induced mutant fraction measurements are presented in this paper at a dose of 30 ,g/mL.

Tar Product Composition
Toluene pyrolysis yielded a broad distribution of PAHs, with structures ranging in aromatic carbon number from 10 to 26. These compounds were formed by carbon-fragment addition, aryl-aryl recombination, condensation, and isomerization. At high conversion offuel to tar and soot (i.e., at high temperatures), PAHs consisting of four and five fused rings were found to be most abundant. The presence oforganically-bound chlorine was found to affect pyrolysis product composition in two ways: by the formation of chlorinated aromatics and by the promotion of aromatic condensation and soot formation. At high temperatures, little chlorine was found in the aromatic tars produced by the pyrolysis offuels with Cl/H molar ratios ofless than 0.3; most ofthe aryl chlorine that was detected in these tars was bound to unfused benzene rings. In pyrolysis offuels with higher chlorine contents (0.3 < Cl/H < 0.6), the product tars contained significant quantities ofmono-and dichlorinated aromatics. With regard to aromatic structure, the formation of pen-fused aromatics, including CPP, was favored by increased chlorine content over the formation of ortho-fused structures and biaryls. Only trace amounts (less than 0.5 mole % of the total tar yield) of new structures were detected in the tars from chlorinated hydrocarbon pyrolysis; these consisted mostly of highly condensed structures with ethynyl substituents.
Two sets ofconditions that were studied gave notably different tar product distributions. Low-temperature pyrolysis of o-C6H4C12 (temperatures less than 1400 K) yielded tri-and tetrachlorinated biphenyls almost exclusively (5). In the absence of ring rupture, carbon growth appeared to occur entirely by aryl dimerization. Above 1400 K, fragmentation occurred, and a broad distribution ofproducts, similar to those described in the preceding paragraph, was found. A second condition that produced a unique product distribution was the pyrolysis of a chlorine-rich fuel, C2HC13. Here, mostly perchloroaromatics were found (6). Carbon growth in this system was limited, with no identified structure containing more than 14 carbon atoms.

Mutagenicity with Postmitochondrial Supernatant
The mutagenicities of 30 jtg/mL doses of product tars in the presence of mammalian enzymes (i.e., +PMS) are shown as a function of pyrolysis temperature in Figure 1. As a reference point, the induced mutant fraction ofpure benzo[a]pyrene at 30 jtg/mL is about 90x10 -5. All of the tars that contained a broad distribution of aromatic products (i.e., all tars from pyrolysis of toluene-containing fuels, as well as tars from 1,2-C2H4C12 pyrolysis and high temperature o-C6H4C12 pyrolysis) were mutagenic. The tars containing a narrow distribution of products, namely, those produced by pure C2HC13 pyrolysis (containing predominantly perchlorocompounds) and those produced by low temperature pyrolysis of o-C6H4Cl2 (composed almost exclusively of PCBs), were not mutagenic.
Six PAHs that are known bacterial mutagens were detected in significant abundance (greater than 1 mole %) in these tars. In order of decreasing potency, these PAH mutagens are CPP, fluoranthene, benzo[a]pyrene, acephenanthrylene, aceanthrylene, and cyclopenta[hi]acephenanthrylene. Five of these were detected in the toluene pyrolysis tars; their relative yields are given in Table 1. Also shown are the relative potencies of these compounds as bacterial mutagens. The total mutagenicity ofthe toluene pyrolysis tar increased as temperature increased, as did the yields of the five mutagens. Assuming additivity, mutagenicity is determined by the sum of the product of the potency and yield ofeach tar constituent. Mutagen additivity in each of the toluene pyrolysis tars accounts for the total mutagenicity measurement to within 25 %. CPP, fluoranthene, and benzo[a]pyrene appear to be responsible for about 90% of the biological activity.  (2) fluoranthene; (3) benzo[alpyrene; (4) acephenanthrylene; (5) aceanthrylene; (6) cyclopenta[hilacephenanthrylene.
Yields of the six PAH mutagens identified in the tars from pyrolysis of each fuel, over all of the temperatures studied, are shown in Figure 2. Adding o-C6H4CI2 to toluene had little effect on the product distribution and mutagenicity of the resulting pyrolysis tars (Fig. la). Pyrolysis of the 1,2-C2H4Cl2/toluene mixture yielded tars with slightly increased mutagenicity (10-50%). Adding C2HC13 to toluene produced pyrolysis tars of significantly higher mutagenicity (50-200%). In general, the yields of known mutagens increased with increasing pyrolysis temperature, consistent with the measurements of total tar mutagenicity. This was particularly true of CPP, the most potent mutagen found in these tars. As discussed elsewhere (7), CPP formation is promoted in environments that favor the formation ofperi-fused aromatics, i.e., pyrolysis at high temperatures and in the presence of chlorine. Thus, the measurements ofpotential health impact by the bacterial mutation assay appear to be best correlated with the presence of CPP.
The contributions of known PAH mutagens do not appear to account for all of the measured mutagenicity in several of the pyrolysis tars, however. Pure o-C6H4C12 pyrolysis at 1460 K, pure 1, 2-C2H4C12 pyrolysis at low temperatures and pyrolysis of the C2HCl3/toluene mixture at high temperatures yielded very mutagenic tars in which the summed contributions of known PAHs do not appear to account for the measured levels ( Table 2). Each of these tars was composed of a broad distribution of aromatics, including large amounts of partially chlorinated aromatics. We hypothesize that at least one unidentified mutagen is present in the tars from pyrolysis ofchlorinated organics under the conditions specified above. It is possible that a mutagen not produced during pure hydrocarbon pyrolysis was formed either by a chloro-organic-dependent pathway to an unidentified PAH mutagen or by the addition ofchlorine atom to an aromatic structure. We speculate that it was the latter mechanism because these tars contained no PAHs in more than trace amounts that were not found in the pure toluene pyrolysis tars. Furthermore, each of these tars contained significant amounts of mono-and dichlorinated congeners of the major PAH mutagens.
Data that support this speculation are shown in graph form in Figure 3. Pyrolysis of 1,2-C2H4C12 at low temperatures produced highly mutagenic tars containing very small amounts ofthe major PAH mutagens and large amounts of mono-and dichlorinated aromatics. As pyrolysis temperature was increased, the difference between the measured total tar mutagenicity and the sum of the PAH mutagen contributors decreased. The yields of chlorinated arenes also decreased. Similar correlations between chlorinated aromatic yields and unaccounted-for mutagenicity were obtained for the other tars studied. Pyrolysis ofo-C6H4CI2 at low temperatures yielded tars containing only PCBs and having no significant mutagenicity. Pyrolysis of pure C2HC13 produced nonmutagenic tars containing BACTERIAL

Mutagenicity without Postmitochondrial Supernatant
The bacterial mutagenicities of selected tar samples were measured without addition ofmammalian enzymes (i.e., -PMS). One tar from pyrolysis ofeach of thepure liquids was analyzed; dose-response curves for both the +PMS and -PMS tests are presented in Figure 4. The tar from pure toluene pyrolysis was a more potent +PMS mutagen that a -PMS mutagen. This is consistent with other data indicating that unsubstituted PAHs are generally inactive in -PMS tests. Tars from pyrolysis ofthe three chlorinated organics each contained more potent -PMS mutagens than +PMS mutagens. These tars contained significant amounts ofchlorinated arenes. We speculate that chlorine substitution is responsible for the -PMS mutagenicity, consistent with the finding of DeMarini et al. (24) that moderately polar compounds appear to be more potent mutagens in the absence ofmammalian enzymes.
Even in the two cases in which +PMS mutagenicity was negligible (i.e., the case of C2HCl3 pyrolysis at 1370 K yielding perchloroaromatics and the case ofo-C6H4C12 pyrolysis at 1295 K yielding PCBs), the -PMS mutagenicity was significant. In the   bInduced mutant fraction measured for pyrolysis tar samples. cBrackets denote tars in which additivity of known mutagens does not appear to account for the observed total tar mutagenicity. case of 1,2-C2H4C12 pyrolysis at 1360 K, the +PMS mutagenicity was high, and the -PMS mutagenicity was even higher. More work is needed to isolate and identify the specific compound(s) responsible for this mutagenicity.

Conclusions
The composition and bacterial mutagenicity ofpyrolysis product tars were found to be affected by the presence of organic chlorine. In the absence of organic chlorine, bacterial mutagenicity was found to be largely due to the presence ofCPP, fluoranthene, and benzo[a]pyrene in the product tars. Small amounts of chlorine (Cl/H < 0.3) enhanced the formation of pen-fused PAHs, including CPP, and increased tar mutagenicity. Larger amounts of chlorine (0.3 < Cl/H < 0.6) resulted in significant yields ofmono-and dichlorinated aromatics and increased levels of tar mutagenicity. This mutagenicity could not be accounted for by the presence of mutagens found in the pyrolysis tars ofunchlorinated fuels (e.g., CPP) and appears to be related to the presence ofchlorinated aromatics. Furthermore, unlike the tars containing little or no chlorine, the tars containing chlorine were more mutagenic in the absence of PMS than in the presence ofPMS. Therefore, we hypothesize that these tars contain at least one unidentified mutagen, and we speculate that it is a mono-or dichlorinated aromatic. Two conditions that gave notably different results are the lowtemperature pyrolysis (i.e., at temperatures below 1400 K) of o-C6H4C12, which produced triand tetrachlorinated biphenyls almost exclusively, and the chlorine-rich pyrolysis of C2HC13, during which mostly perchloroaromatics were formed. Neither of these tars was found to mutate bacteria.