Unexpected Mechanism of Biodegradation and Defluorination of 2,2-Difluoro-1,3-Benzodioxole by Pseudomonas putida F1

ABSTRACT Perfluorinated carbon atoms in a diether linkage are common in commercial anesthetics, drugs, fungicides, and insecticides. An important chemical group comprising perfluorodiethers is the 2,2-fluoro-1,3-benzodioxole (DFBD) moiety. The fluorine atoms stabilize the molecule by mitigating against metabolism by humans and microbes, as used in drugs and pesticides, respectively. Pseudomonas putida F1 catalyzed defluorination of DFBD at an initial rate of 2,100 nmol/h per mg cellular protein. This is orders of magnitude higher than previously reported microbial defluorination rates with multiply fluorinated carbon atoms. Defluorination rates declined after several hours, and the medium darkened. Significant defluorination activity was observed with cells grown on toluene but not l-arginine. Defluorination required only toluene dioxygenase. Pseudomonas and recombinant Escherichia coli cells expressing toluene dioxygenase oxidized DFBD to DFBD-4,5-dihydrodiol. The dihydrodiol could be oxidized to 4,5-dihydroxy-DFBD via the dihydrodiol dehydrogenase from P. putida F1. The dihydrodiol dehydrated with acid to yield a mixture of 4-hydroxy-DFBD and 5-hydroxy-DFBD. All those metabolites retained the difluoromethylene group; no fluoride or dark color was observed. The major route of DFBD-4,5-dihydrodiol decomposition produced fluoride and 1,2,3-trihydroxybenzene, or pyrogallol, and that was shown to be the source of the dark colors in the medium. A mechanism for DFBD-4,5-dihydrodiol transformation to two fluoride ions and pyrogallol is proposed. The Pseudomonas genome database and other databases revealed hundreds of bacteria with enzymes sharing high amino acid sequence identity to toluene dioxygenase from P. putida F1, suggesting the mechanism revealed here may apply to the defluorination of DFBD-containing compounds in the environment.

For more insights into DFBD defluorination, we chose to examine model Pseudomonas strains for which genomes are available and enzymes have been purified and characterized. The first defluorinating enzyme studied, fluoroacetate defluorinase, was purified and characterized from a Pseudomonas strain (22). In one recent study, fluoroacetate was shown to undergo defluorination by multiple bacterial strains, but difluoro-and trifluoroacetates were uniformly recalcitrant (23). However, Pseudomonas strains are known to contain organochlorine dehalogenases that react with dichloro-and trichloro-substrates (24,25). Moreover, Pseudomonas genomes are known to encode .25% proteins of unknown function, suggesting novel defluorinating activities may yet be discovered (26,27). After initially screening several Pseudomonas strains against a small library of fluorinated compounds, we observed significant levels of fluoride anion in the medium with strain Pseudomonas putida F1 and DFBD. This model strain has had its genome sequenced and multiple enzymes characterized in detail (28,29). The most well-studied enzymes are involved in the metabolism of aromatic hydrocarbons (30), and DFBD contains an aromatic ring. Moreover, the DFBD moiety is known to be quite stable, underlying its incorporation into numerous commercial products (Fig. 1).
The present study focused on DFBD transformation by P. putida F1 and the mechanism by which free fluoride was released into the medium. Using the combined tools of nuclear magnetic resonance (NMR) spectroscopy, gas chromatography (GC), and mass spectrometry (MS), we propose here a novel mechanism underlying fluoride displacement from the DFBD group. It was demonstrated that the only enzyme required for defluorination of DFBD is toluene dioxygenase, and the mechanism elucidated here has, to our knowledge, never been shown previously. Moreover, many environmental bacteria express toluene dioxygenase homologs, suggesting that the mechanism revealed here may be operative with DFBD compounds found in soil and water.

RESULTS AND DISCUSSION
Metabolic studies with wild-type Pseudomonas putida F1. Wild-type cells of P. putida F1 were grown to mid-exponential phase on toluene vapors to induce toluene dioxygenase and related enzymes. The toluene vapor was removed and replaced with 2,2-difluoro-1,3-benzodioxole (DFBD) vapor, and free fluoride ion was detected immediately. Over time, color change of the medium was also observed (see Fig. S1 in the supplemental material). The release of free fluoride ion was measured using a fluoride electrode, which showed a rapid accumulation of fluoride in the first 4 h ( Fig. 2A). After this, fluoride in the medium plateaued at ;1 mM, which was followed by a darkening of the medium (Fig. 2B). Cells grown on L-arginine as a carbon source that were not induced with toluene produced minimal levels of fluoride (0.1 mM over 24 h). Additionally, DFBD was stable in medium in the absence of cells. No fluoride was released, and there was no darkening of medium. When sodium fluoride was added to medium, no color change was observed. In multiple experiments, color change was always linked to fluoride release but lagged behind temporally as shown in Fig. 2.
The initial rate of fluoride ion release was 2,100 nmol/h per mg of cell protein. By way of comparison, the rates of initial enzymatic attack on toluene by different Pseudomonas strains are reported to range from 720 to 10,800 nmol/h per mg protein (31). Defluorination of difluoromethylene carbon atoms by microorganisms has rarely been reported, and when it has, it is typically orders of magnitude slower and measured over weeks or months (7,8). The rapid plateau reached after 4 h suggests that some factor becomes limiting. This argues against a hydrolytic defluorination since that would require only water that would not become limiting. The observations are more consistent with a redox reaction, since NADH or other sources of electrons could conceivably become depleted. The observation that growth on toluene is required suggests a role for one or more components of the toluene metabolic pathway.
Metabolic studies with mutant P. putida F39/D and Escherichia coli (pDTG601a) expressing toluene dioxygenase. The observations with wild-type P. putida F1 suggested that one or more components of the toluene degradation pathway were involved in DFBD defluorination and the darkening of the medium. Since the pathway is initiated by toluene dioxygenase (TDO), we first aimed to examine that enzyme's reaction with DFBD in isolation. Toluene dioxygenase consists of three components that are highly unstable and separate upon cell lysis and purification. They are a flavoprotein reductase (TodA), a ferredoxin (TodB), and the Rieske dioxygenase protein (TodC 1 C 2 ) (32). So, an in vivo approach was pursued with P. putida F39/D, a derivative of P. putida F1 that lacks a functional second enzyme in the toluene catabolic pathway (33), and the recombinant strain E. coli (pDTG601a), which contains the todC1C2BA genes encoding the complete TDO system but no other enzymes of the toluene pathway (29). It is plausible that these two systems could give different results. For example, the TodE catechol oxygenase in P. putida F39/D could hypothetically catalyze ring cleavage leading to defluorination and E. coli DTG601a does not contain that enzyme.
P. putida F39/D was grown on L-arginine and induced with toluene vapor before incubation with DFBD. E. coli (pDTG601a) was induced with isopropyl-b-D-thiogalactopyranoside (IPTG) and subsequently incubated with DFBD. In both cases, the media were extracted with ethyl acetate, concentrated using a rotary evaporator, and analyzed by GC-MS and NMR as described in Materials and Methods. The GC-MS and 19 F-NMR analyses were both consistent with the formation of a 4,5-dihydro-2,2-difluoro-1,3-benzodioxole-4,5-diol (4,5-DD-DFBD) (Fig. 3). Similar results were obtained with the P. putida F39/D and E. coli (pDTG601a) extracts. Because dihydrodiols are subject to decomposition in high temperatures of the GC inlet, extracted medium was derivatized prior to GC-MS analysis. Both the GC retention time and the mass spectrum of the prominent peak at 12.64 min are consistent with its identification as a derivatized dihydrodiol ( Fig. 3A and B). Further structural information was gained from 19 F-NMR that showed that both fluorine atoms were retained in the product (Fig. 3C). The fluorine resonance splitting demonstrated that the plane of the benzene ring was asymmetric, consistent with a cis-dihydrodiol. The oxidation of bicyclic fused ring compounds, such as indan (34), by toluene dioxygenase has not been observed to produce bridgehead diols, and the enzyme consistently produces cis-dihydrodiols. Therefore, it was predicted that the cis-dihydrodiol being formed was most likely a 4,5-cis-dihydrodiol.
Dehydration of the cis-dihydrodiol-DFBD. It was decided that following the fate of 4,5-DD-DFBD would lead to insights into darkening of medium and the defluorination mechanism. Dihydrodiols are known to readily undergo dehydration to yield phenols. The dehydration reaction with substituted benzenes often produces a mixture of two phenols (33). The initial evidence that the DFBD dihydrodiol is unstable came from its low yield and subsequent NMR analysis, initially described in the previous section. When it was dried and taken up in CDCl 3 , a solvent known to be slightly acidic, the 19 F-NMR spectrum showed a collapse of the fluorine doublet into a single resonance, consistent with a rearomatization of the benzene ring that would occur with dehydration, producing a monohydroxylated aromatic ring (Fig. S6). With rearomatization, the environments of the two fluorine substituents become identical. Unexpected Mechanism for Defluorination of DFBD ® From the NMR spectrum, it could not be discerned whether DFBD-4-ol, DFBD-5-ol, or both are formed.
Direct evidence was obtained that both DFBD-4-ol and DFBD-5-ol form in growth medium. This was determined by GC-MS, following derivatization to make both trimethylsilyl derivatives, and in comparison to authentic standards of each (Fig. 4). Together, they are not major products and account for ,20% of total products. Moreover, addition of DFBD-4-ol and DFBD-5-ol to P. putida F1 growth medium did not lead to formation of fluoride; both compounds are stable in medium. In addition, no evidence for the metabolism of standard DFBD-4-ol and DFBD-5-ol by P. putida F39/D or E. coli (pDTG601a) in minimal salts basal (MSB) medium was observed. Based on these observations, additional metabolic studies were needed to uncover the mechanism of fluoride displacement. E. coli (pDTG602) expressing toluene dioxygenase and diol dehydrogenase. P. putida F1 is known to oxidize cis-dihydrodiols to catechols that can oxidize to form products that darken growth medium (29). In that context, we sought to accumulate the catechol from DFBD and determine if that could help explain medium darkening and fluoride release. The experiment used E. coli (pDTG602), which contains genes encoding toluene dioxygenase and dihydrodiol dehydrogenase and is known to accumulate catechols (29). This study produced a metabolite consistent with 4,5-dihydroxy-DFBD (4,5-DH-DFBD), as demonstrated by 1 19 F-NMR (Fig. S8). Moreover, similar to the presence of DFBD-4-ol and DFBD-5-ol in E. coli (pDTG601a) cultures, 4,5-dihydroxy-DFBD was formed in only low yield. More importantly, we observed that medium from E. coli pDTG602 incubated with DFBD did darken but not as quickly or to the same extent with as with P. putida F39/D and E. coli (pDTG601a). This suggested that the DFBD-4,5-dihydrodiol is undergoing some other reaction to give a colored product(s) and fluoride.
Identification of pyrogallol, a defluorinated product that explained medium darkening. The volatility of DFBD prevented our determining the stoichiometry of DFBD oxidized to the amount of fluoride released. The initial rate of fluoride release (Fig. 2) was high, and the yields of DFBD-4-ol, DFBD-5-ol, and 4,5-DH-DFBD were low, suggesting that the DFBD-cis-4,5-dihydrodiol undergoes another reaction leading to defluorination. In this context, another organic product was sought in cultures of E. coli (pDTG601a) and P. putida F39/D. GC-MS revealed the transient formation of a nonfluorinated product, 1,2,3-benzenetriol, also known as pyrogallol (Fig. 6A). 1,2,3-Benzenetriol is known to undergo spontaneous oxidation in air (35,36), and our cultures are vigorously shaken to provide a continuous supply of oxygen for the dioxygenase reaction. These oxidation products absorb at varying wavelengths, which leads to dark-colored microbiological medium (37,38).
To follow pyrogallol formation and decomposition, we used UV-visible (UV-vis) spectroscopy of biological material in comparison to standard pyrogallol and purpurogallin ( Fig. 6B and C; see also Fig. S9 and S10). Pyrogallol, prior to oxidation, absorbs Unexpected Mechanism for Defluorination of DFBD ® only below 300 nm, which is difficult to analyze because DFBD absorbs in the 250-to 300-nm region. So, we analyzed at wavelengths above 300 nm. When added to sterile growth medium, pyrogallol quickly oxidized to a species that absorbs at 320 nm, purportedly purpurogallin (Fig. S10). Over time, this decayed with the appearance of absorbance at 420 to 440 nm, indicating the formation of purpurogallin-quinone (35). Over time, other absorbing species led to the previously described brown to black coloration of pyrogallol medium that was observed here with DFBD in medium in the presence of toluene dioxygenase (Fig. 6B) and standard pyrogallol in medium (Fig. 6C).
Mechanism of defluorination. A satisfactory explanation for DFBD defluorination had to align several key data. First, fluoride release was immediate and rapid and required toluene-grown cells, and medium darkening lagged temporally. All bacterial strains expressing toluene dioxygenase, even singly, released fluoride and darkened medium. However, known further transformation products of the toluene dioxygenase, which are 4,5-DD-DFBD and the phenols and catechols characterized here, did not undergo defluorination. Indeed, the presence of the enzyme dihydrodiol dehydrogenase, which  transformed at least some 4,5-DD-DFBD to the corresponding catechol, diminished fluoride release and color formation. Those observations, together with the immediate fluoride release, indicated that 4,5-DD-DFBD was undergoing another reaction, and that reaction was fast enough to compete with the dehydrogenase kinetically. This previously undescribed reaction pathway had to explain rapid fluoride release, pyrogallol appearance in the medium, and a lag in darkening of the medium.
It was not possible to isolate and store 4,5-DD-DFBD due to its extreme instability compared to other bicyclic dihydrodiols (34). Its instability was presumably linked to the highly electron-withdrawing -OCF 2 -moiety adjacent to the dihydrodiol group, precisely the atoms that are lost (Fig. 7). The carbon atom linking the two groups referred to above is a bridgehead carbon. This bridgehead carbon would be highly susceptible to nucleophilic attack by a hydroxide anion in solution. The resultant trihydroxy compound (middle left, Fig. 7) is expected to decompose with C-O breakage, rearomatization, and opening of the five-membered 1,3-dioxole ring. We propose here that these are spontaneous reactions, driven by the electrophilicity of the bridgehead carbon atom and the stabilization energy inherent in rearomatization that occurs with C-O bond cleavage. However, we cannot rule out the possibility of a cell component assisting in hydroxide attack on the bridgehead carbon. Once the C-O bond cleaves, the resultant -OCF 2 O-moiety is highly unstable, to undergo very rapid hydrolysis to produce carbon dioxide and the two fluorine atoms as fluoride. Pyrogallol, the other product, is not colored but over time will oxidize to a range of colored products that darken the medium (Fig. 7).
Potential for DFBD defluorination by other aromatic oxygenase-containing bacteria. A question naturally arises as to whether DFBD defluorination by this mechanism is common, since genes encoding the oxygenation of toluene and other aromatic hydrocarbons are widespread in bacterial genomes (39)(40)(41). In addition to dioxygenation, many bacteria oxidize toluene via monooxygenases (39). Given the stability of monohydroxy and catecholic DFBD metabolites, we did not expect strains with monooxygenases to defluorinate DFBD. To test this hypothesis, Burkholderia cepacia G4, Unexpected Mechanism for Defluorination of DFBD ® containing toluene-2-monoxygenase, was grown on glucose and induced with toluene. After 24 h of incubation with DFBD, this strain did not release fluoride or create a dark color in the medium.
Dioxygenases that oxidize multiring substrates were considered to be better candidates to test for DFBD defluorination. Defluorination and darkening of the medium were not observed with Pseudomonas sp. strain NCIB 9816 or Pseudomonas sp. strain 9816-11, each of which expresses a naphthalene dioxygenase. Biphenyl dioxygenases that have different amino acid sequences were also tested with two bacteria, Burkholderia xenovorans LB400 and Sphingobium yanoikuyae B1. After induction with biphenyl and incubation with DFBD for 24 h, no fluoride release or color change was observed with either strain.
Identifying enzymes with greater similarity to the toluene dioxygenase from P. putida F1. Toluene dioxygenase is known to be a Rieske oxygenase, a very common class of proteins in aerobic bacteria (39)(40)(41). The Rieske oxygenases are composed of two protein subunits. The cofactors and the active site are within the larger, or alpha-, subunit. The alpha-subunit contains a Rieske iron sulfur cluster and a mononuclear iron center. Since the alpha-subunit of Rieske dioxygenases determines substrate specificity (42,43), sequence comparisons with this subunit are most germane to the question as to which enzymes and organisms will defluorinate DFBD.
The amino acid sequence of the P. putida F1 toluene dioxygenase alpha-subunit (WP_012052601.1) was used to search the Pseudomonas genome database (22,23) for homologs to the alpha-subunit of toluene dioxygenase on 6 August 2021 using DIAMOND-BLAST. The closer the sequence to the toluene dioxygenase studied here, the more likelihood that the active site would bind and oxygenate DFBD in a manner leading to defluorination (Table 1). This search identified proteins with amino acid sequences that were 100% identical in the following Pseudomonas strains with the locus tags given in the parentheses: Pseudomonas monteilii SB3101 (X970_10585), P. monteilii SB3078 (X969_10930), P. putida YKD221 (TR32_RS03575), Pseudomonas sp. strain NBRC 111125 (APH29_RS14695), and P. putida UV4/95 (CBP06_RS10555). Published substrate specificity studies with P. putida UV4/95 have shown the same substrate spectrum as P. putida F1 (44), consistent with the 100% sequence identity of the dioxygenase alpha-subunits.
Next, we queried the RHObase, or Ring-Hydroxylating Oxygenase database, which focuses specifically on characterized Rieske oxygenases using the term "toluene dioxygenase" on 7 August 2021. This identified several enzymes denoted as toluene dioxygenase or nitrotoluene dioxygenases. One of these, from Pseudomonas putida DOT-T1E (ADI95397), was 100% identical to the alpha-subunit of toluene dioxygenase from P. putida F1.
The large GenBank nonredundant database, with 419,490,021 sequences as of 6 August 2021, was searched to determine the approximate number of homologs and particularly close sequence matches. A BLAST search yielded over 1,200 homologous sequences with an E value lower than e 250 , indicating they matched extremely well. Of those, 293 showed sequence identity of .50% and 9 showed sequence identity of  (49), and the query sequence coverage refers to the extent over which the two sequences aligned.
.90% (Table 1). Closely related sequences are found in Pseudomonas, Comamonas, Sphingomonas, Paraburkholderia, Nocardioides, and Rhodococcus. Next, the Rieske dioxygenases in the bacteria that were tested here and found not to release fluoride were analyzed for amino acid sequence relatedness to the P. putida F1 toluene dioxygenase. The biphenyl dioxygenase from Sphingobium yanoikuyae B1 is 37% identical in amino acid sequence to toluene dioxygenase via pairwise BLAST alignment. The Pseudomonas sp. strain NCIB 9816 naphthalene dioxygenase is 32% identical to the toluene dioxygenase alpha-subunit. The Paraburkholderia xenovorans LB400 biphenyl dioxygenase is 65% identical.
It is not currently known how much a sequence might deviate from 100% identity, and if slightly different, what specific requirements would ensure reactivity with DFBD at the 4,5-position leading to defluorination. Further experimental work is warranted to test other bacteria with closely related Rieske dioxygenases. It is possible that docking and molecular dynamics simulations could also be done to predict enzyme reactivity and regioselectivity with DFBD (45,46), but those detailed computational experiments are beyond the scope of the present study. A search of the Protein Data Bank for homologs of toluene dioxygenase yielded four distinct Rieske dioxygenases with Xray structures and .64% sequence identity. Based on the bioinformatic data presented here and the wide diversity of bacteria known to express Rieske dioxygenases, we suggest that the DFBD moiety in natural environments might be subject to the type of metabolism revealed here with P. putida F1.
Conclusions. Pseudomonas strains and oxygenases have the potential to catalyze defluorination reactions in unexpected ways as demonstrated here. Previously, Pseudomonas sp. strain T-12 expressing toluene dioxygenase was shown to hydroxylate a carbon bearing a fluorine substituent, leading to defluorination by gem elimination (17). In this study, a difluorinated ether was shown to be defluorinated by P. putida F1 via a different mechanism but also demonstrated to be dependent upon toluene dioxygenase. Bioinformatic analyses strongly suggest that DFBD and its derivatives will undergo defluorination by dioxygenases found in other known Pseudomonas strains and in natural environments.

MATERIALS AND METHODS
Bacterial strains, growth, and manipulations. Bacteria, growth substrates, enzymes expressed, and references are compiled in Table 2. Further specific details on growth and manipulations are provided below. Pseudomonas putida F1 was grown on Luria-Bertani (LB) plates and then grown overnight on minimal salts basal medium (MSB) (47) with carbon supplied via a vapor bulb containing toluene, as previously described (48). This 10-ml culture was then added to an additional 25 ml of MSB, and the toluene bulb was replaced with a 2,2-difluoro-1,3-benzodioxole bulb. This flask was incubated at 30°C with shaking at 200 rpm. Aliquots of the culture were taken out at various time points. Samples intended for measuring fluoride release were centrifuged to remove cells, and the supernatant was removed and placed into a clean tube. For photographs of color change, the medium aliquots were taken directly and then stored at 220°C to be photographed simultaneously. Pseudomonas putida F39/D was originally streaked onto an LB plate and grown at 28°C. For liquid cultures, this strain was grown on MSB and 0.2% (wt/vol) L-arginine. A saturated 10-ml culture of P. putida F39/D, grown overnight, was added to 200 ml of MSB and 0.2% (wt/vol) arginine and allowed to grow to log phase at 30°C. Once log phase was attained, the culture was induced with toluene for 1 h. After induction, the toluene bulb was replaced with a DFBD bulb. After 2 h of incubation with DFBD, the cells were pelleted, and the supernatant was removed. E. coli pDTG601a was maintained on LB plates with 100 mg/ml ampicillin at 37°C. For experimental use, it was grown overnight on LB with ampicillin liquid, cells were pelleted, and then the cells were resuspended in MSB, 0.2% (wt/vol) glucose, and 100 mg/ml ampicillin. After growing at 37°C for 2 h, this strain was induced with 1 mM IPTG for 1 h. After induction, a DFBD vapor bulb was added to the flask and the culture was left to incubate for a further 2 h. E. coli pDTG602 cells were maintained on LB-ampicillin plates at 37°C. For experimental use, this strain was grown overnight on MSB medium with glucose and 100 mg/ml ampicillin. This overnight culture was used to inoculate a culture of 50 ml of MSB, glucose, and ampicillin and induced with IPTG for 1 h. After induction, DFBD was added in vapor form. Aliquots were taken at various time points and left to incubate overnight. Pseudomonas sp. NCIB 9816 and 9816-11 were maintained on LB plates at 28°C. NCIB 9816 was grown to log phase on 100 ml of MSB and naphthalene for induction of naphthalene dioxygenase. Subsequently, the culture was filtered through glass wool to remove naphthalene crystals and a vapor bulb of DFBD was added to the culture. NCIB 9816-11 was grown on MSB medium and 15 mM pyruvate. This culture was added to 25 ml of MSB and induced with 0.2% naphthalene. This culture was also filtered through glass wool before adding the DFBD vapor bulb, and then the culture was left to incubate for 24 h. Burkholderia cepacia G4 was maintained on LB plates at 28°C. It was grown in MSB and 0.2% glucose overnight. This culture was added to 25 ml of MSB and induced with toluene for 1 h. Sphingobium yanoikuyae B1 was maintained on LB plates at 28°C. Overnight, this strain was grown on MSB and biphenyl, which induces the biphenyl dioxygenase. Paraburkholderia xenovorans LB400 was maintained on nutrient agar plates at 28°C. Overnight, this strain was grown on MSB and biphenyl, which induces the biphenyl dioxygenase enzyme.
Chemicals. 2,2-Difluoro-1,3-benzodioxole (DFBD) was purchased from Matrix Scientific and 1 H-NMR indicated 97% purity. 2,2-Difluoro-1,3-benzodioxole-4-ol and 2,2-difluoro-1,3-benzodioxole-5-ol were from AstaTech with 95% purity. 1,2,3-Benzenetriol was purchased from Mallinckrodt with a purity of ;95%. Purpurogallin was from Cayman Chemical and had a purity of 95%, as determined by 1 H-NMR. N, O-Bis(trimethylsilyl)trifluoroacetamide was purchased from Fluka Analytical and had a purity of 99%. CDCl 3 and CD 3 CN were obtained from Cambridge Isotope Laboratories, Inc., and had .99% purity. Methyl-tert-butyl ether (MTBE) and ethyl acetate (EtOH), solvents used for GC-MS, were from Sigma-Aldrich, both high-performance liquid chromatography (HPLC) grade, .99% pure. Toluene used for specified cell growth and induction was from Fisher Chemical and is 99% pure. Sodium fluoride (NaF) used for fluoride standards was obtained from Aldrich Chemical company, 99% pure. D-Glucose, used for growth of E. coli recombinant strains, was purchased from Sigma-Aldrich and was 99.5% pure. IPTG was obtained from Gold Bio. L-Arginine, used for growth of Pseudomonas strains, is from Calbiochem. Sodium pyruvate is from Alfa-Aesar with a purity of 99%. Biphenyl was from Spectrum Chemical and had a purity of 99%. Naphthalene was from Alfa-Aesar with a purity of 99%.
Nuclear magnetic resonance (NMR) experiments. Cell culture supernatants from both E. coli pDTG601a and P. putida F39/D after incubation with DFBD were extracted with an equal volume of ethyl acetate (EtOAc). In each case, the aqueous layer was drawn off and the organic layer was dried (MgSO 4 ), transferred to a round-bottom flask, and rotary evaporated. The residue was taken into CD 3 CN for initial 1 H-and 19 F-NMR acquisitions on a Varian INOVA 500-MHz NMR spectrometer. The initial 1 H-NMR spectra were too complex to identify signals due to 4,5-dihydro-DFBD-4,5-diol, but 19  In the proton NMR, the high field doublet at 6.51 ppm is partly buried in a neighboring peak. Additional significant peaks that do not describe the 4,5-DD-DFBD are shown to be 1,2,3benzenetriol ( Fig. S7 and S9).
GC-MS experiments. For gas chromatography and mass spectrometry (GC-MS), cell culture supernatants were extracted with methyl-tert-butyl ether (MTBE). Equal volumes of organic and aqueous phases were used. Extraction was done directly in glass GC vials, which were shaken vigorously, and then the organic layer was drawn off and placed in a clean vial. One microliter of derivatizing agent, N,O-bis(trimethylsilyl)trifluoroacetamide, was added to each extracted sample. Product ion spectra were identified in positive ion mode on an HP6890 gas chromatograph with an HP5973 MS detector. GC conditions were as follows: helium gas, 1 ml/min; HP-1ms column (100% dimethylpolysiloxane capillary; 30 m by 250 mm by 0.25 mm); starting temperature 50°C with a ramp of 10°C/min to 320°C. Standards run on the GC-MS used the same program specified above. 2,2-Difluoro-1,3-benzodioxole-5-ol and 2,2-difluoro-1,3benzodioxole-4-ol were prepared by placing a small volume of each into separate GC vials with 500 ml of MTBE. One microliter of the derivatizing agent was added to each vial. Pyrogallol (1,2,3-benzenetriol) was prepared similarly, by putting a small solid amount of the compound into 500 ml of MTBE and adding 1 ml of derivatizing agent.
Fluoride measurements. Fluoride ion measurements were determined using an ISE Ionplus Sure-Flow Orion fluoride probe from Thermo Fisher and the Orion Star A214 meter. To perform fluoride measurements, 1 ml of cell culture was pelleted. The supernatant was removed and placed into a 5-ml tube with 1 ml of TISAB (58.5 g/liter NaCl, 15 g/liter CH 3 COOH, 66 g/liter CH 3 COONa, and 1 g/liter 1,2-cyclohexane diaminetetraacetic acid). Calibrations of the probe were made with standard concentrations ranging from 10 25 M to 10 22 M. Two standards were required for calibration. Standards were made with sodium fluoride in MSB medium.
UV-vis spectroscopy. UV-vis absorption scans were performed on a Cary 3500 double-beam UV-vis spectrophotometer containing a Xenon flash lamp from Agilent. Measurements of supernatant from an E. coli pDTG601a culture were taken after 72 h of incubation with DFBD. An aliquot was taken from the cell culture and pelleted to remove cells. The supernatant was removed and placed in a 1.5-ml disposable cuvette. The sample was then measured (300 to 700 nm) in increments of 5 nm. Standard 1,2,3-benzenetriol was prepared to a concentration of 0.1 M in water. To scan this compound, 1 ml of MSB was placed in a 1.5-ml disposable cuvette and 1,2,3-benzenetriol stock solution was added to a final concentration of 0.1 mM. This cuvette of MSB and 0.1 mM pyrogallol was left to sit at room temperature for 72 h and measured after that.
Bioinformatic methods. BLAST was used to query the NCBI GenBank nonredundant database (49). DIAMOND-BLAST was used to query the Pseudomonas genome database (50). The RHObase, or Ring-Hydroxylating Oxygenase database, was queried with a word search (51). All hits obtained from the searches were compiled manually. Comparisons with toluene dioxygenase and other single enzymes were carried out using pairwise BLAST (49) on the National Center for Biotechnology Information server.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.