Degradation of Chlorinated Dibenzofurans and Dibenzo-p-Dioxins by Two Types of Bacteria Having Angular Dioxygenases with Different Features

ABSTRACT Two kinds of bacteria having different-structured angular dioxygenases—a dibenzofuran (DF)-utilizing bacterium,Terrabacter sp. strain DBF63, and a carbazole (CAR)-utilizing bacterium, Pseudomonas sp. strain CA10—were investigated for their ability to degrade some chlorinated dibenzofurans (CDFs) and chlorinated dibenzo-p-dioxins (CDDs) (or, together, CDF/Ds) using either wild-type strains or recombinant Escherichia coli strains. First, it was shown that CAR 1,9a-dioxygenase (CARDO) catalyzed angular dioxygenation of all mono- to triCDF/Ds investigated in this study, but DF 4,4a-dioxygenase (DFDO) did not degrade 2,7-diCDD. Secondly, degradation of CDF/Ds by the sets of three enzymes (angular dioxygenase, extradiol dioxygenase, and meta-cleavage compound hydrolase) was examined, showing that these enzymes in both strains were able to convert 2-CDF to 5-chlorosalicylic acid but not other tested substrates to the corresponding chlorosalicylic acid (CSA) or chlorocatechol (CC). Finally, we tested the potential of both wild-type strains for cooxidation of CDF/Ds and demonstrated that both strains degraded 2-CDF, 2-CDD, and 2,3-diCDD to the corresponding CSA and CC. We investigated the sites for the attack of angular dioxygenases in each CDF/D congener, suggesting the possibility that the angular dioxygenation of 2-CDF, 2-CDD, 2,3-diCDD, and 1,2,3-triCDD (10 ppm each) by both DFDO and CARDO occurred mainly on the nonsubstituted aromatic nuclei.

Polychlorinated dibenzofurans and polychlorinated dibenzop-dioxins (PCDF/Ds) are highly toxic compounds but have become widely distributed in the environment. They are formed as unwanted by-products during the synthesis of pesticides and herbicides or during municipal incineration and are not degraded under environmental conditions. Recently, bioprocess treatment methods have been proposed for the destruction of dioxins, and bioremediation of contaminated soil is being used as an alternative to more costly physical and chemical treatments (10,12,20). Over the last decade, a number of bacteria capable of degrading dibenzofuran (DF), dibenzo-pdioxin (DD), and some of their lightly chlorinated derivatives have been isolated, and the cooxidative degradation of chlorinated DFs (CDFs) and chlorinated DDs (CDDs) (or, together, CDF/Ds) have been elucidated (14,17,21,25,30,35,37,38). However, compared with bacterial degradation of polychlorinated biphenyls carrying di-to pentachlorine substituents (6,11,19,31,32,33,34), knowledge concerning bacterial cometabolism of CDF/Ds carrying two or more chlorine substituents is still lacking. One reason is the shortage of information about the initial dioxygenases that can act on the angular position adjacent to the oxygen atom of DF and DD skeletons, though many biphenyl dioxygenases have been well studied. This initial dioxygenase, termed angular dioxygenase, is a very important enzyme for the degradation of dioxin-related compounds (7,13,36). Since angular dioxygenation produces unstable hemiacetal compounds that rearomatize spontaneously, prompting cleavage of the ether bond ( Fig. 1), the reduction of toxicity of dioxins via the ring cleavage is accomplished by this single-step reaction.
Besides those of DFDO from strain DBF63 and CARDO from strain CA10, only one detailed report has been made of the cloning of angular dioxygenase genes. This enzyme, dioxin dioxygenase from Sphingomonas wittichii strain RW1 (38,39), has been studied at the molecular level (1,2,3,5,8). Although substrate specifities of these angular dioxygenases for various aromatic compounds have already been investigated (8,16,22), data concerning the degradation of CDF/Ds by angular dioxygenases have not been reported.
For bioremediation of dioxin-contaminated soils, it is important to know the influence of the chlorine substitution pattern on the degradation of CDF/Ds by these angular dioxygenases with different features as well as the possible fate of CDF/Ds in the environment after degradation by different types of dioxin degraders. In this work, we describe the substrate selectivity of DFDO and CARDO and the potential of strains DBF63 and CA10 to cooxidize various mono-, di-, and triCDF/Ds. From the results obtained, we discuss the sites of angular dioxygenation by DFDO and CARDO and the transformation capabilities of the subsequent enzymes.

MATERIALS AND METHODS
Bacterial strains, plasmids, media, and culture conditions. Bacterial strains and plasmids used in this study are listed in Table 1. Terrabacter sp. strain DBF63 was routinely cultivated in a carbon-free mineral medium (CFMM) (23) supplemented with DF at a concentration of 0.05% (wt/vol) at 30°C. Pseudomonas sp. strain CA10 was similarly cultivated in CFMM with CAR at the same concentration. DF and CAR were added to CFMM from stock solutions (0.1 g/ml) dissolved in dimethyl sulfoxide and sterilized by filtration. Ten milliliters of precultures of strains DBF63 and CA10 grown on CFMM supplemented with 0.05% (wt/vol) DF and CAR, respectively, was inoculated into 1 liter of the same fresh medium in 5-liter flasks and cultivated at 30°C for another 3 and 2 days, respectively. When the optical density at 600 nm reached 1.5, cells were harvested by centrifugation, washed twice with 50 mM phosphate buffer (pH 7.   (27) at 37°C, which was supplemented with ampicillin and/or chloramphenicol at a final concentration of 50 g/ml and 30 g/ml, respectively. The resting cells of E. coli harboring pDF32, pUCARA, pUCA1, and pLM1RM were prepared as described previously (15,22). Biotransformation analysis of CDDs and CDFs. Biotransformation reactions proceeded at 30°C in glass tubes sealed with silicon caps. The tubes containing 5 ml of the washed-cell suspension and the substrate at 10 ppm (10 g/ml) were incubated on a reciprocal shaker (300 strokes/min) at 30°C for 18 h. Stock solutions of CDF/Ds (1 mg/ml) were made up in dimethyl formamide. The reaction mixtures were extracted twice with 5 ml of ethyl acetate under acidic or neutral conditions. Extracts were dried over anhydrous Na 2 SO 4 , and concentrated in vacuo. Metabolites formed from CDF/Ds were analyzed by gas chromatography-mass spectrometry (GC-MS) after derivatization with N-methyl-Ntrimethylsilyltrifluoroacetamide (MSTFA) and verified by coinjection of authentic samples. When authentic compounds were not available, structures of metabolites were proposed by MS spectra. We determined the rate of substrate depletion or metabolite formation by comparing the peak areas for molecular ions of each compound extracted from the reaction mixture containing E. coli cells harboring pDF32, pUCARA, pUCA1, and pDF32 plus pLM1RM with those of each compound extracted from the reaction mixture containing E. coli cells harboring pUC119 or pUC119 plus pSTV28. We repeated the above experiments two or three times.
Analytical methods. GC-MS analyses were performed with a model JMS-Automass 150 GC-MS system (JEOL, Ltd., Tokyo, Japan) fitted with a fusedsilica chemically bonded capillary column (DB-5; 0.25 mm [inside diameter] by 15 m; 0.25 m film thickness; J&W Scientific, Inc., Folsom, Calif.). After trimethylsilylation with MSTFA, each sample was injected into the column at 80°C in the splitless mode. After 2 min at 80°C, the column temperature was increased at 16°C/min to 280°C. The head pressure of the helium carrier gas was 65 kPa.

Conversion of CDFs and CDDs by DFDO and CARDO.
To investigate whether two kinds of angular dioxygenases, DFDO and CARDO, transform CDF/Ds, these compounds were incubated with recombinant E. coli strains carrying respective genes, and the products were characterized by GC-MS (Table  2).
Although 2-CDF has two possible sites for angular dioxygenation, the GC-MS analytical data showed the production of one compound. In contrast to 2-CDF, 2-CDD has four possible sites for the initial angular dioxygenation. However, the sample of 2-CDD incubated with either DFDO or CARDO appeared to contain two metabolites showing very similar fragmentation patterns. In addition, CARDO degraded 2,7-DCDD, producing a metabolite, whereas DFDO did not metabolize 2,7-DCDD. In the sample of 1,2,3-TCDD with DFDO and CARDO, we detected one metabolite (retention time [RT], 13.03 min) and two metabolites, respectively.
The following data obtained by GC-MS strongly suggested that these metabolites are the chlorinated THB and THDE originating from the parent CDFs and CDDs. (i) All products had the possible molecular ions whose mass unit corresponded to the calculated molecular weight of the trimethylsilyl (TMS)derivatives of chlorinated THB or chlorinated THDE as shown in Table 2. (ii) In our previous reports (15,22), the mass spectra of the TMS derivatives of 2,2Ј,3-THB (M ϩ , 418), 2,2Ј,3-THDE (M ϩ , 434), 2,2Ј,3-trihydroxydiphenylmethane Comparison of the isotope ratio between the TMS-derivatized metabolites and the parent substrate on mass spectra showed that both contained the same number of chlorine substituents (data not shown). From these findings, it was suggested that CARDO from strain CA10 could catalyze the angular dioxygenation of all the tested mono-to triCDF/Ds. On the other hand, DFDO from strain DBF63 could not degrade 2,7-DCDD.
Conversion of CDFs and CDDs by DbfA1A2BC or Car-AaAcAdBaBbC. The formation of corresponding chlorosalicylic acid (CSA) and CC from CDFs and CDDs by DbfA1A2BC, which are thought to be involved in the metabolism of DF to salicylic acid in strain DBF63, were investigated (Table 3). E. coli cells harboring both pDF32 containing dbfA1A2 genes and pLM1RM containing dbfBC genes were used for synthesizing the three enzymes, i.e., angular dioxygenase, extradiol dioxygenase, and meta-cleavage compound hydrolase. As a result, 2-CDF was converted to 5-CSA but not to salicylic acid. On the other hand, 5-CSA and CCs were not produced from 2,8-DCDF, 2-CDD, 2,3-DCDD, 2,7-DCDD, and 1,2,3-TCDD. However, approximately 96, 100, 98, 0, and 95% of the substrates were degraded, respectively, and also, the corresponding chlorinated THB and chlorinated THDE were not detected. These results suggest that CDF/D congeners except 2,7-DCDD may undergo meta-cleavage by DbfB after angular dioxygenation, although direct evidence for the presence of the meta-cleavage product is not provided. The fact that approximately 100% of 2,7-DCDD remained was consistent with the result of the transformation of 2,7-DCDD by DFDO.
We also investigated whether CAR-degrading enzymes from strain CA10, CarAaAcAdBaBbC, involved in the metabolism of CAR to anthranilic acid, could catalyze the degradation of CDFs and CDDs to CSAs and CCs, respectively (Table 3). Like DbfA1A2BC, GC-MS analysis revealed that the resting cells of E. coli harboring pUCA1 containing the carAaAcAd BaBbC genes produced a small amount of 5-CSA from 2-CDF, but salicylic acid was not detected. From the other tested CDF/Ds, corresponding chlorinated THB, chlorinated THDE, CSA, and CCs were not detected. Based on the facts that approximately 70% of 2,8-DCDF, 100% of 2-CDD, 80% of 2,3-DCDD, 30% of 2,7-DCDD, and 60% of 1,2,3-TCDD were degraded after the biotransformation, it was suggested that CarBaBb might catalyze the degradation of chlorinated THB and chlorinated THDE, but that CarC could not degrade effectively the resultant metabolites.
When 2-CDD was incubated with the resting cells of strain DBF63, we detected the single product (RT, 6.32 min) whose mass spectrum exhibited fragment ions at m/z 290 (M ϩ , 37   The starting concentration of substrates in the biotransformations used to estimate the yields of substrate depletion and the product formation was 10 ppm. b Values represent the percentages of the respective CDF/D degradation rates by resting cells of each strain. After 18-h biotransformation experiments, we compared the peak area for the molecular ion of each substrate extracted from each reaction mixture containing DBF63 cells, CA10 cells, and E. coli cells harboring either dbfABC or carABC with that for each substrate extracted from the reaction mixture containing E. coli cells harboring only pUC119 or pUC119 plus pSTV28 (control). We repeated the experiments two or three times and estimated the substrate depletion.
c Values represent the percentages of the corresponding CSA or CC formation rates by resting cells of each strain. After 18-h biotransformation experiments, we compared the peak area for the molecular ion of TMS derivatives of each product extracted from each reaction mixture containing DBF63 cells, CA10 cells, and E. coli cells harboring either dbfABC or carABC with that of 10 ppm of authentic CSA or CC. We repeated the experiments two or three times and estimated the yields of products. d Identification of products formed was made by GC-MS after derivatization of MSTFA. e ND, the corresponding CSA and CC were not detected. In all these cases, both salicylic acid and catechol were not detected. to the experiment of 2-CDD transformation, no catechol was detected from the extract of the reaction mixture. In addition, it was revealed that 2,7-DCDD and 1,2,3-TCDD were not converted to either corresponding CC or catechol. However, 2,7-DCDD was also degraded up to 10 to 25%, though DFDO did not attack this substrate. Since the predicted monohydroxylated derivative of 2,7-DCDD was detected by GC-MS (data not shown), 2,7-DCDD was thought to be degraded by other oxygenases or hydroxylases in strain DBF63. Cooxidation of CDFs and CDDs by Pseudomonas sp. strain CA10. CAR-grown CA10 cells converted 2-CDF to 5-CSA but not to salicylic acid. Based on the GC-MS analysis, it was also revealed that 4-CC and 4,5-DCC were formed from 2-CDD and 2,3-DCDD, respectively, but no catechol was produced from both substrates by strain CA10 (Table 3). These results are the same with the conversion by strain DBF63, although the yields of 5-CSA, 4-CC, and 4,5-DCC by strain CA10 were much lower than those by strain DBF63 (Table 3). These results showed that strain CA10 had a weaker ability to degrade CDF/Ds than did strain DBF63.

DISCUSSION
We investigated the concentration of the total PCDF/Ds in the actual dioxin-contaminated soil at an incinerator site located in Japan. Since it amounted to 725 ng/g of dry soil (10), we first tried to use a low concentration (0.5 ppm) of CDF/Ds as target substrates. However, since the analysis of metabolites at this low concentration was difficult, probably due to low transformation rates of CDF/Ds, 10 ppm of each substrate was used.
In this study, the positions of the hydroxy groups in the chlorinated THB and THDE produced by DFDO and CARDO could not be determined exactly, because biotransformation experiments at low concentrations of substrates did not produce enough amounts of the products for additional analyses such as 1 H nuclear magnetic resonance spectrometry. Therefore, assignment of the positions of angular dioxygenation from GC-MS data can be made only for 2,8-DCDF, the attack at only a single angular position of which will yield a 5,5Ј-dichloro-2,2Ј,3-THB (Fig. 2, panel 2, reaction A). However, the combination of the results obtained from the transformation of various CDF/D congeners by angular dioxygenase frequently permitted assignment of the position of initial dioxygenation. Also, it was quite useful for position assignment of dioxygenation to determine what type of salicylic acid or catechol derivatives was produced by biotransformation experiments using wild-type strains (35). Possible assignment of the positions of angular dioxygenation is shown in Fig. 2.
In experiments, neither nonsubstituted salicylic acid nor cat-echol was detected after biotransformation using either strain DBF63 or CA10. The nonsubstituted aromatic nucleus of 2-CDF seems to be attacked because 5-CSA was produced from 2-CDF by wild-type strains DBF63 and CA10 (Table 3) as well as E. coli cells harboring DbfA1A2BC and Car AaAcAdBaBbC. In addition, as shown in Table 2, only one angular dioxygenation product was detected by DFDO and CARDO. These results suggest that DFDO and CARDO mainly act on the nonsubstituted aromatic nucleus of 2-CDF to create a 5Ј-chloro-2,2Ј,3-THB (Fig. 2, panel 1, reaction A).
2-CDD may also be attacked at the site of the nonsubstituted aromatic nucleus, because 4-CC was produced by wildtype strains DBF63 and CA10 in the biotransformation experiments (Table 3). In addition, two angular dioxygenation products were produced by DFDO and CARDO. In contrast to 2-CDF, which contains only two sites for the initial enzymatic attack, 2-CDD has four possible sites for the initial angular dioxygenation. Considering that DFDO and CARDO act on the nonsubstituted aromatic nucleus of 2-CDF, these two metabolites are thought to be 5Ј-chloro-2,2Ј,3-THDE (Fig.  2, panel 3, reaction A) and 4Ј-chloro-2,2Ј,3-THDE (Fig. 2, panel 3, reaction AЈ).
Previously, Harms et al. revealed that a mutant strain from the DF-degrading bacterium Pseudomonas sp. strain HH69, which is defective in 2,3-dihydroxybiphenyl-1,2-dioxygenase, led to the formation of an equal amount of 4Ј-chloro-2,2Ј,3-THB and 4-chloro-2,2Ј,3-THB from 3-CDF (14). Wilkes et al. demonstrated that strain RW1 converted 2-CDF to both 5-CSA and salicylic acid and converted 2-CDD to both 4-CC and catechol (35). Wittich et al. also reported that the conversion of 3-CDF in the presence of an inhibitor of meta-cleavage of chlorinated THB by Sphingomonas sp. strain RW16 led to accumulation of two catabolites that are likely to be 4-chloro-2,2Ј,3-THB and 4Ј-chloro-2,2Ј,3-THB (37). These results indicated that angular dioxygenases from these bacteria attacked both the substituted and the nonsubstituted aromatic nuclei of monochlorinated DFs and DDs nonselectively. In these biotransformation experiments, the concentrations of target substrates were from 100 to 1,000 ppm. On the other hand, Parsons et al. proposed that initial dioxygenation apparently took place on the nonchlorinated ring during the degradation of 2-CDF and 2-CDD by the biphenyl-utilizing Burkholderia strain JB1, since 5-CSA and 4-CC as well as only one isomer each of chlorinated THB and chlorinated THDE were detected (25). In their experiment, the concentration of substrate was low (0.3 ppm). Our results with 10 ppm of target substrates suggested that both DFDO and CARDO might attack the nonchlorinated ring of 2-CDF and 2-CDD as does the initial dioxygenase from strain JB1.
On the other hand, none of the meta-cleavage compounds of CDDs were hydrolyzed effectively by E. coli cells harboring the dbfC gene or carC gene, whereas the wild-type strains DBF63 and CA10 apparently could hydrolyze meta-cleavage compounds of both 2-CDD and 2,3-DCDD. Previously, Kasuga et al. showed that DbfBC was capable of converting 2,2Ј,3-THB to salicylic acid and 2,2Ј,3-THDE to the corresponding metacleavage product but not to catechol (15). Therefore, it seems reasonable that DbfB catalyzed meta-cleavage of chlorinated THDE but DbfC did not hydrolyze the corresponding metacleavage compound. In the DD-degrading bacterium strain RW1, the meta-cleavage compound hydrolase, DxnB, was supposed to be involved in the DD degradation system (2, 4), but direct proof of the hydrolysis of meta-cleavage compounds of THDE has not been reported (9). On the other hand, Pfeifer et al. reported that 2,3-dihydroxydiphenyl ether was converted to phenol and 2-pyrone-6-carboxylic acid by the action of purified extradiol dioxygenase from Pseudomonas cepacia strain Et4 (26). If a similar reaction occurred in the case of chlorinated THDE by the action of DbfB or CarBaBb, formation of the corresponding CCs instead of phenol would be expected. However, we could not detect CCs in the biotransformation experiments using DbfA1A2BC and CarAaAcAdBaBbC. In strains DBF63 and CA10, it also remains unclear what type of compounds are formed after meta-cleavage and which hydrolase is involved in these reactions. There exists a possibility that an esterase is involved in this hydrolysis reaction.
Considering the above results, the influence of the chlorine substitution pattern of mono-to triCDF/Ds on the formation of corresponding CSAs and CCs may not be dependent on the substrate specificity of the initial angular dioxygenases but on that of the subsequently acting enzyme, especially hydrolases. This result is consistent with the report that BphD is a key determinant in the aerobic microbial degradation of polychlorinated biphenyls (31,32).
In conclusion, our study suggests the sites for the attack of angular dioxygenases and the transformation capabilities of subsequently acting enzymes toward some CDF/Ds using two different types of dioxin degraders. Although it remains to be investigated which enzyme catalyzes hydrolysis of the metacleavage compounds of CDF/Ds in these bacteria, the results of this study add useful knowledge to the bacterial degradation of dioxins. For a complete understanding of the influence of chlorine substituents on the ability of angular dioxygenases to oxidize chlorinated dioxins, transformation of more highly CDF/Ds and determination of the structure of these enzymes will be required. Accumulation of such information will make it possible to perform efficient bioremediation of dioxins.