Subtle Difference between Benzene and Toluene Dioxygenases of Pseudomonas putida

Abstract

Benzene dioxygenase and toluene dioxygenase from Pseudomonas putida have similar catalytic properties, structures, and gene organizations, but they differ in substrate specificity, with toluene dioxygenase having higher activity toward alkylbenzenes. The catalytic iron-sulfur proteins of these enzymes consist of two dissimilar subunits, α and β; the α subunit contains a [2Fe-2S] cluster involved in electron transfer, the catalytic nonheme iron center, and is also responsible for substrate specificity. The amino acid sequences of the α subunits of benzene and toluene dioxygenases differ at only 33 of 450 amino acids. Chimeric proteins and mutants of the benzene dioxygenase α subunit were constructed to determine which of these residues were primarily responsible for the change in specificity. The protein containing toluene dioxygenase C-terminal region residues 281 to 363 showed greater substrate preference for alkyl benzenes. In addition, we identified four amino acid substitutions in this region, I301V, T305S, I307L, and L309V, that particularly enhanced the preference for ethylbenzene. The positions of these amino acids in the α subunit structure were modeled by comparison with the crystal structure of naphthalene dioxygenase. They were not in the substrate-binding pocket but were adjacent to residues that lined the channel through which substrates were predicted to enter the active site. However, the quadruple mutant also showed a high uncoupled rate of electron transfer without product formation. Finally, the modified proteins showed altered patterns of products formed from toluene and ethylbenzene, including monohydroxylated side chains. We propose that these properties can be explained by a more facile diffusion of the substrate in and out of the substrate cavity.

The ring-hydroxylating dioxygenases of prokaryotes are nonheme iron enzymes that convert aromatic compounds to dihydrodiols in the presence of dioxygen and NADH (6). They act upon a wide range of compounds, which the organisms use as carbon and energy sources. The enzymes are important in the bioremediation of environmental pollutants. Although none of them is completely specific, the dioxygenases display selectivity for the compounds that they metabolize. Moreover, they convert these substrates into stereo- and regiospecific products. These properties make them valuable in the enantio-controlled synthesis of compounds (21).

The benzene 1,2-dioxygenase from Pseudomonas putida ML2 (EC 1.14.12.3) is an extensively studied enzyme system (6, 33, 35, 44) that catalyzes the dihydroxylation of benzene to (1R,2S)-cis-cyclohexa-3,5-diene-1,2-diol (benzene cis-dihydrodiol), the first step in biodegradation of this xenobiotic compound. The enzyme system has three components: a flavoprotein reductase and a ferredoxin, which transfer electrons from NADH, and a catalytic iron-sulfur protein (ISP_BED). ISP_BED contains a Rieske-type [2Fe-2S] cluster, a mononuclear nonheme iron oxygen activation center, and a substrate-binding site (6, 22, 25). The ISP_BED component comprises two dissimilar subunits, ISP_BEDα and ISP_BEDβ (55). A closely related enzyme that preferentially metabolized ethylbenzene and toluene, toluene dioxygenase ISP_TOD (EC 1.14.12.11), was isolated from P. putida F1 (13). Much evidence has accumulated that the α subunit of the ISP contains the region that confers substrate specificity (2, 3, 5, 8, 9, 17, 26, 34, 36-38, 47). The genes encoding benzene and toluene dioxygenase systems have been cloned, sequenced, and expressed separately and in combination in Escherichia coli (35, 49, 50, 57, 58), but the molecular structures of the enzymes have not been determined. However several structures are now available for the naphthalene dioxygenase ISP (25), which has a homologous amino acid sequence (35.2% identity with ISP_BED) (24). The enzyme comprises a trimeric arrangement of two types of subunits, α₃β₃.

The manipulation of enzyme specificity, both for the substrates used and for the products formed, is important for biotechnology. It is necessary to minimize undesirable side reactions, such as the generation of products that cannot be metabolized or the direct oxidation of NADH without substrate oxidation (the uncoupling reaction). Enzymes that are isolated from the natural environment have been optimized by evolution to avoid these difficulties. To better design new enzymes, it is worthwhile to examine naturally occurring enzymes with different specificities in order to determine which features of the sequence are necessary for efficient and specific catalysis.

Alignment of the nucleotide and amino acid sequences of ISP_BEDα (50) and ISP_TODα (57) revealed that they differ by 33 amino acids, distributed through the sequence. The first objective of this study was to investigate which of these amino acid differences is necessary for conversion of the substrate preference of benzene dioxygenase from benzene to alkyl-substituted aromatic hydrocarbons, such as toluene and ethylbenzene. The N- and C-terminal domains of toluene and benzene dioxygenases were interchanged, and then individual amino acids were mutated. The recombinant proteins were subcloned and expressed in an isogenic background in E. coli. The properties of the variant enzymes were investigated by examining the activities with benzene, toluene, and ethylbenzene by an oxygen electrode assay and by performing product formation assays with radiolabeled substrates, TLC, and GC-MS.

MATERIALS AND METHODS

Abbreviations.

GC-MS, gas-chromatography-mass spectrometry; TLC, thin-layer chromatography; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis; ISP, iron-sulfur protein of terminal dioxygenase; ISP_BEDα and ISP_TODα, iron-sulfur proteins of benzene and toluene dioxygenases, respectively; ISP_BODα and ISP_TEDα, chimeric α subunits with N-terminal regions from benzene and toluene dioxygenases and C-terminal regions from toluene and benzene dioxygenases, respectively; ISP_BOD2α, ISP_BEDα with residues 364 to 450 of ISP_TODα; ISP_BOD3α, ISP_BEDα with residues 281 to 363 of ISP_TODα.

Bacterial strains and plasmids.

Bacterial strains and plasmids used in this study are listed in Table 1. E. coli JM109 was used for standard cloning and gene expression. E. coli CJ236 was used in the mutagenesis experiment with uracil-containing DNA. Transformations were carried out by the method of Hanahan (16).

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Relevant characteristics
Strains
E. coli JM109a	endA1 recA1 gyrA96 thi hsdR17 relA1 supE44 Δ(lac-proAB) [F′ traD36 proAB lacIqZΔM15]
E. coli CJ236b	dut-1 ung-1 thi-1 rela1 F′ pCJ105 (Cmr)
Plasmids
pGEM 3Zf(+)	Plac lacZ Ampr
pKK223-3	Plac Ampr
pCB1c	1.6-kb DraI/EcoRI bedC1 insert in pGEM 3Zf(+)
pJRM502c	2-kb EcoRI/Pst1 bedC1C2 insert in pKK223-3
pJRM502-1	2-kb EcoRI/PstI todC1-bedC2 insert in pKK223-3
pJRM502-2	2-kb EcoRI/PstI bodC1-bedC2 insert in pKK223-3
pJRM503d	1.7-kb EcoRI bedBA insert in pKK223-3
pJRM504c	1.6-kb EcoRI/HindIII bedC1 insert in pKK223-3
pJRM505d	0.7-kb EcoRI/PstI bedC2 insert in pKK223-3
pJRM711	0.66-kb EcoRI todC2 insert in pKK223-3
pSA109d	1.6-kb EcoRI/PstI todC1 insert in pKK223-3
pJRM5042	1.6-kb EcoRI/PstI todC1 modified insert in pKK223-3
pJRM5041	1.6-kb EcoRI/PstI bodC1 insert in pKK223-3
pJRM5041-1	1.6-kb EcoRI/HindIII bod2C1 insert in pKK223-3
pJRM5041-2	1.6-kb EcoRI/HindIII bod3C1 insert in pKK223-3
pCMB101	1.6-kb EcoRI/PstI tedC1 insert in pKK223-3
pJRM5043	1.6-kb EcoRI/HindIII tedC1 modified insert in pKK223-3
pCB1M1	pCB1 (L285W)
pCB1M2	pCB1 (A291S)
pCB1M3	pCB1 (L285W/A291S)
pCB1M5	pCB1 (V324I/I327V)
pCB1M7	pCB1 (G404D)
pCB1M9	pCB1 (K436R)
pJRM504M1	pJRM504(L285W)
pJRM504M2	pJRM504(A291S)
pJRM504M3	pJRM504(L285W/A291S)
pJRM504M4e	pJRM504(I301V/T305S/I307L/L309V)
pJRM504M5	pJRM504(V324I/I327V)
pJRM504M6e	pJRM504(L285W/A291S/G404D)
pJRM504M7	pJRM504(G404D)
pJRM504M8e	pJRM504(I412V)
pJRM504M9	pJRM504(K436R)
pJRM504M10e	pJRM504(E444D)

^aSee reference 54.

^bObtained from Bio-Rad.

^cGift of C. S. Butler.

^dGift of S. Anguravirutt.

^eMutant ISP_BEDα generated by PCR-based mutagenesis.

Cloning strategies.

To construct ISP_BODα encoded by plasmid pJRM5041 (Fig. 1A), the EcoRI-BstEII fragment from pJRM504 encoding ISP_BEDα (residues 1 to 280) was ligated to the BstEII-HindIII fragment (residues 281 to 450) from pSA109 encoding ISP_TODα. ISP_TEDα encoded by plasmid pCMB101 was constructed by ligating the EcoRI-BstEII fragment from pSA109 with the BstEII-HindIII fragment from pJRM504. The two constructs that had the 5′ end from plasmid pSA109 had a lower expression level due to a difference of nine nucleotides with a high adenosine content in the ribosome-binding domain (45). To increase the level of expression, this sequence (CCAAAAAGT) was introduced into the Shine-Dalgarno sequence by carrying out PCR with pSA109 by using forward primer KPIf, which contained the 9-nucleotide sequence and a KpnI restriction site, and reverse primer BEIIr, which contained a BstEII restriction site. The positions of the primers are shown in Table 2. PCRs were carried out for 25 cycles of denaturation at 94°C for 20 s, primer annealing at 55°C for 20 s, and primer extension at 74°C for 60 s. The amplified DNA was digested with KpnI and BstEII and ligated into the BstEII-KpnI fragment of pJRM504 to construct pJRM5043 (ISP_TEDα modified). Plasmid pJRM5042, containing ISP_TODα modified in the Shine-Dalgarno region, was constructed by ligation of the pJRM5043 HindIII-BstEII fragment with the BstEII-HindIII fragment of pSA109 (Fig. 1A).

FIG. 1.

(A) Diagram of the genes encoding ISP_BEDα, ISP_TODα, ISP_BODα, ISP_TEDα, ISP_BOD2α, and ISP_BOD3α and relative activities of the corresponding cell extracts. (B) Diagram of the single and multiple mutations created in the ISP_BEDα C-terminal region and the effects of the mutations on the relative activity. The bars indicate the open reading frames encoding ISP α subunit proteins. The nonconserved amino acids targeted for mutation are indicated by solid boxes. The letters above the bars indicate translated amino acids in ISP_BEDα, and the letters below the bars indicate amino acids in ISP_TODα. The numbering corresponds to the numbering for the amino acids in the ISP_BEDα sequence. Solid diamonds indicate the conserved histidine (H98 and H119) and cysteine (C96 and C116) residues responsible for coordination of the [2Fe-2S] cluster. Solid circles indicate the conserved amino acids (two histidines [H222 and H228] and one aspartate [D376]) involved in coordination of the nonheme iron. The EcoRI, KpnI, BstEII, SnaBI, MluI, and HindIII sites are the restriction sites used for cloning the recombinant α subunit genes. The relative activities with toluene (tol) and ethylbenzene (ethylben) are compared to the activity with benzene (ben), and the data are based on the rate of O₂ uptake measured polarographically by using ISP_BED or ISP_TOD β subunits, as described in Materials and Methods (means of three determinations).

TABLE 2.

Oligonucleotides used in this study

Oligonucleotidea	Oligonucleotide sequenceb	Presence, loss, or gain of restriction site
Oligonucleotides for modification of the Shine-Dalgarno region
KPIfc	5′ CTCGGTAC(CCAAAAAGT)GAGAAGACAATG 3′	Gain of KpnI
BEIIrc	5′ CAAGTAGCTGGTGACCTTCGGCCC 3′	BstEII
Mutant oligonucleotidesd
L285W	5′ CCAGCTACTGGACCGAAGGCCCGGCG 3′	Loss of DraII
A291S	5′ GAAGGCCCCGCGTCGGAAAAGGCGG 3′	Loss of SacII
G404D	5′ GCTGAGATGTCCATGGACCAAACCGTTG 3′	Gain of NcoI
I301V/T305S/I307L/L309V	5′ GCTCGACCATGAGTTTGGAGCCGCGCTCCACACTAC 3′	Gain of BstXI
V324I/I327V	5′ CCCAGGTATCAATACGGTCCGAACATGG 3′	Gain of AvaII
I412V	5′ CGACCCGGTATACCCTGGTCG 3′	Gain of Bst1107I
K436R	5′ GCACATTGGCTCCGGATGATGACC 3′	Gain of BspEI
E444D	5′ GACTGGGACGCATTAAAGGCGACG 3′	Gain of BsmFI
Nonmutagenic oligonucleotides for PCR-based site-directed mutagenesis and/or cloning of bod2C1, bod3C1, todC1-bedC2, and bodC1-bedC2 genes
SBIfc	5′ CGTCAGACACTACGTACCTTCTCTGCC 3′	SnaBI
MUIrc	5′ TTAGGTTTAACGCGTCGCCTTTAATGC 3′	MluI
BEIIfc	5′ GGGCCGAAGGTCACCAGCTACTTG 3′	BstEII
SBIrc	5′ GGCAGAGAAGGTACGTAGTGTCTGACG 3′	SnaBI
HDIIIrc	5′ GGCTGAAAATCTTCTCTCATCCGC 3′	HindIII
PKKc	5′ GCGGATAACAATTTCACACAGG 3′	EcoRI
AGIr	5′ CGGGCGCTACCGGTGCCGGC 3′	AgeIe

^aMost oligonucleotides were synthesized by Phil Marsh (King's College London); the exceptions were KPIf and BEIIr, which were synthesized by Pharmacia (UK) and Perkin-Elmer Ltd., respectively.

^bThe base changes are indicated by boldface type. The underlined bases indicate loss or addition of the restriction site. Additional bases are in parentheses.

^cFor the locations of nonmutagenic oligonucleotides see the positions of restriction sites in Fig. 1, sequences are within or close to the primer sequence.

^dIn the mutant oligonucleotide designations the first letter is the amino acid in the ISP_BEDα sequence, the number corresponds to the position in the amino acid sequence, and the second letter is the replacing amino acid in the ISP_TODα sequence.

^eThe AgeI restriction site is located between the bedC1 and bedC2 genes in the pJRM502 plasmid.

Two chimeric ISP α subunits in the C-terminal region, ISP_BOD2α and ISP_BOD3α, were generated by using an approach similar to the approach described above (Fig. 1A). ISP_BOD2α contained the N-terminal region of ISP_BEDα (residues 1 to 363) linked to the SnaBI-HindIII C-terminal region of ISP_TODα (residues 364 to 450). ISP_BOD3α was ISP_BEDα with the BstEII-SnaBI region (residues 281 to 363) replaced with the BstEII-SnaBI region of ISP_TODα.

To construct the coexpression vectors encoding ISP_TODα-ISP_BEDβ and ISP_BODα-ISP_BEDβ, PCRs were carried out with plasmids pJRM5042 (encoding ISP_TODα) and pJRM5041 (encoding ISP_BODα), respectively, by using primers PKKf and AGIr (Table 2). The PCR products were digested with EcoRI and AgeI and ligated into pJRM502 (encoding ISP_BEDα-ISP_BEDβ) previously digested with the same enzymes to construct plasmids pJRM502-1 and pJRM502-2, respectively. The correct clones were selected by restriction analysis and DNA sequencing.

Mutagenesis.

Site-directed mutagenesis was performed by using both the uracil-containing DNA-based mutagenesis method developed by Kunkel (27) and a PCR-based mutagenesis method adapted from the method of Kammann et al. (23). For uracil-containing DNA-based mutagenesis, plasmid pCB1 containing an F1 origin and encoding ISP_BEDα was transformed into E. coli strain CJ236 (ung dut). After production of single-stranded DNA with the VCM13 helper phage (Stratagene), mutagenesis was carried out with a Muta-gene kit (Bio-Rad). The mutated DNA was further cloned into expression vector pJRM504 (ISP_BEDα). For PCR-based mutagenesis, the first PCR was carried out with the wild-type ISP_BEDα gene (plasmid pJRM504) by using the mutant oligonucleotide and a nonmutagenic oligonucleotide facing in the opposite direction at a distance of about 100 bp. The product was then used as a megaoligonucleotide in a second PCR with another nonmutagenic oligonucleotide facing in the opposite direction at a distance of about 250 bp. The final product was digested and ligated into the wild-type ISP_BEDα gene. The mutagenic oligonucleotides were designed with a silent mutation that generated a new restriction site (Table 2) to facilitate identification of clones carrying the desired mutation. Successful mutations were identified by restriction analysis and DNA sequencing.

Expression of the genes in E. coli JM109 and preparation of cell extracts.

Cultures of E. coli JM109 cells carrying the different plasmids were grown at 37°C in L broth containing ampicillin (100 μg ml⁻¹) to an optical density at 660 nm of 0.7 and then were grown at 30°C after induction with 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG). The cells were harvested in the late exponential phase, washed once in 50 mM potassium phosphate (pH 7.2), and resuspended in 25 mM potassium phosphate (pH 7.2)-1 mM dithiothreitol (2 ml/g [wet weight]). The cells were disrupted by three passages through a chilled French pressure cell (Aminco, Silver Spring, Md.) at 27.5 MPa. After centrifugation at 17,400 × g for 1 h, the supernatant was used for enzyme activity measurements.

Protein determination, SDS-PAGE, and Western blot analysis.

Protein concentrations were determined by the modified Lowry method of Hess et al. (19). Proteins were resolved by SDS-PAGE (28). Western blot analysis with antibodies raised to purified ISPα was performed as described by Zamanian and Mason (55). To determine the ISPα concentration expressed per milligram of cell extract protein, samples were resolved by SDS-PAGE, stained with Coomassie brilliant blue R250, and subjected to quantitative image analysis with the EASYWin program (Herolab, Wiesloch, Germany). Duplicate samples (5 and 10 μg of cell extract protein) were taken, and a calibration curve was constructed by using a range of concentrations of pure ISP_BEDα (1 to 15 μg).

Oxygen uptake measurement.

Dioxygenase activities of cell extracts were determined polarographically by measuring O₂ consumption with a Clark-type oxygen electrode as previously described (11). The assay mixture contained 0.42 mM NADH, 0.5% (vol/vol) Triton X-100, 0.1 mM ferrous ammonium sulfate, and 1.5 mg of cell extract protein from the strain expressing reductase_BED and ferredoxin_BED (plasmid pJRM503). The following enzymes were also added to the assay mixture: for the coexpressed enzymes, 1.5 mg of cell extract protein from the strains expressing ISP_BEDα-ISP_BEDβ (plasmid pJRM502), ISP_TODα-ISP_BEDβ (plasmid pJRM502-1), or ISP_BODα-ISP_BEDβ (plasmid pJRM502-2); and when the α and β subunits were expressed separately, 0.25 mg of ISP α subunits present in the cell extracts from the different strains and 0.5 mg of either ISP_BEDβ present in the cell extract from the strain containing plasmid pJRM505 or ISP_TODβ present in the cell extract from the strain containing plasmid pJRM711. The reaction was initiated by addition of 300 μM benzene, toluene, or ethylbenzene dissolved in buffer. Enzyme activity was expressed in nanomoles of O₂ per minute per milligram of ISPα.

Enzyme activity.

Benzene dioxygenase activity was determined by using ¹⁴C-labeled substrates by measuring the accumulation of nonvolatile products on TLC squares by using a modification of the method described by Jiang et al. (22). Each reaction mixture (0.5 ml) was similar to the reaction mixture described for the polarographic assay, except that after 5 min of preincubation at room temperature, 1 mM NADH and either 20 μl of l-[U-¹⁴C]benzene, 20 μl of l-[U-¹⁴C]toluene (Sigma) in methanol (22.2 × 10⁴ dpm μl⁻¹; final concentration, 0.8 mM), or 20 μl of l-[U-¹⁴C]ethylbenzene (Sigma) in methanol (4.8 × 10⁴ dpm μl⁻¹; final concentration, 0.8 mM) were added. The enzyme specific activity was expressed in nanomoles of total product formed per minute per milligram of ISPα.

Detection and quantification of reaction products.

The oxidation products of the recombinant proteins from the different substrates were determined by using a modified method described by Ensley et al. (7). The assay with a radiolabeled substrate was allowed to proceed for 45 min. Aliquots (8 μl for experiments in which [¹⁴C]benzene and [¹⁴C]toluene were used and 16 μl for experiments in which [¹⁴C]ethylbenzene was used) were applied to the origin of a TLC plate (Silica Gel 60 F254; thickness, 0.2 mm). Standard samples of cis-3,5-cyclohexadiene-1,2-diol, catechol, 3-methylcatechol, o-cresol, m-cresol, p-cresol, and benzyl alcohol were also applied to the chromatograms, which were air dried for 30 min and developed with chloroform-acetone (80:20, vol/vol). Radioactive spots were located by autoradiography with X-ray film exposed at −70°C for 6 days before development. The area containing the radioactive product was scraped from the TLC plate, and the amount of each nonvolatile ¹⁴C-labeled product was determined by scintillation counting. Nonradioactive compounds were detected by spraying the plate with 2% (wt/vol) 2,6-dichloroquinone-4-chloroimide (13).

The identities of the reaction products were further established by GC-MS. Assays with unlabeled substrates were performed as described above, and the reaction products were extracted with 2 volumes of ethyl acetate. The organic phase was dried over anhydrous sodium sulfate, and the samples were concentrated to 100 μl under a flow of nitrogen. Samples were injected into a Hewlett-Packard 6890 GC coupled to a 5970 mass selective detector operating in splitless mode (injector temperature, 300°C) with a fused silica capillary column (inside diameter, 0.25 mm; length, 25 m) coated with a 0.1-μm phase of DB-5 (5% phenyl, 95% dimethylpolysiloxane; J& W Scientific, Folsom, Calif.). The column was kept at 50°C for 3 min, and then the temperature was increased to 300°C at a rate of 15°C/min and finally kept at 300°C for 10 min. The detector was operated in the scan mode with electron impact ionization at 70 eV and with scanning from 550 to 50 atomic mass units with a scan time of 0.86 s. The transfer line temperature was 280°C. All products were identified either by comparison of their GC-MS properties (retention times and mass fragments) with those of reference compounds used in the TLC experiments or by interpretation of the mass spectra. Relative yields of products were determined by integration of total ion current peak areas.

RESULTS

Chimeric proteins were constructed with the sequences containing residues 1 to 280 and 281 to 450 of ISP_BEDα and ISP_TODα, which yielded proteins designated ISP_BODα and ISP_TEDα (Fig. 1A). Two additional chimeric α subunits in the C-terminal region were constructed. ISP_BOD2α contained the N-terminal region of ISP_BEDα (residues 1 to 363) linked to the C-terminal region of ISP_TODα (residues 364 to 450). ISP_BOD3α was ISP_BEDα with residues 281 to 363 replaced with the residues of ISP_TODα. The cell extracts from each of the recombinant strains contained a protein that immunocross-reacted with a band whose relative mobility was identical to that of purified ISP_BEDα (data not shown). The level of protein production was determined by quantitative SDS-PAGE (Fig. 2A). Despite the presence of identical promoter and ribosome-binding sites, the level of expression for the recombinant plasmids generally was not as high as that for the ISP_BEDα protein (Fig. 2A, lane 2).

FIG. 2.

SDS-PAGE-resolved E. coli cell extracts expressing recombinant ISP α subunits. The numbers in parentheses indicate the levels of expression of the soluble ISP α subunit in the cell extract (in milligrams per milligram of total protein; means of two determinations). (A) Wild-type and chimeric ISP α-subunit cell extracts. Lane 1, purified ISP_BEDα (2 μg); lanes 2 to 7, cell extracts (5 μg of protein per lane) from E. coli strains containing plasmids pJRM504 (ISP_BEDα), pJRM5041 (ISP_BODα), pJRM5042 (ISP_TODα), pJRM5043 (ISP_TEDα), pJRM5041-2 (ISP_BOD3α), and pJRM5041-1 (ISP_BOD2α), respectively. (B) Wild-type and mutant ISP α subunit cell extracts. Lanes 1 to 12, cell extracts (5 μg of protein per lane) from E. coli strains containing plasmids pKK223-3, pJRM504 (ISP_BEDα), pJRM504M1 [ISP_BEDα(L285W)], pJRM504M2 [ISP_BEDα(A291S)], pJRM504M7 [ISP_BEDα(G404D)], pJRM504M3 [ISP_BEDα(L285W/A291S)], pJRM504M6 [ISP_BEDα(L285W/A291S/G404D)], pJRM504M4 [ISP_BEDα(I301V/T305S/I307L/L309V)], pJRM504M5 [ISP_BEDα(V324I/I327V)], pJRM504M8 [ISP_BEDα(I412V)], pJRM504M9 [ISP_BEDα(K436R)], and pJRM504M10 [ISP_BEDα(E444D)], respectively.

When the O₂ uptake assay was used to measure activity, all the recombinant ISP α subunits reconstituted with ISP_BEDβ or ISP_TODβ were active with the three substrates (Fig. 1 and Table 3). The ratios of the oxygen uptake rates measured with toluene or ethylbenzene to the oxygen uptake rates measured with benzene (Fig. 1) gave a first measure of the substrate preferences of the enzymes. The chimeric ISP_BODα and ISP_BOD3α proteins combined with ISP_BEDβ had a pattern of substrate specificity more like that of ISP_TODα; the toluene/benzene activity ratios of the three proteins were 1.07, 0.88, and 1.26, respectively, and the ethylbenzene/benzene activity ratios were 1.10, 0.94, and 1.20, respectively. Furthermore, ISP_TEDα and ISP_BOD2α were more similar to ISP_BEDα, and the ethylbenzene/benzene activity ratios of the three proteins were 0.35, 0.30, and 0.21, respectively (Fig. 1). The nature of the β subunit did not have a consistent effect on the substrate specificity. It should be noted that the specific activities of ISP_TODα, ISP_BODα, ISP_TEDα, and ISP_BOD3α combined with ISP_BEDβ were lower than the specific activity of ISP_BEDα (Table 3). However, this had no effect on the substrate preference since it was the relative activity with all three substrates that was important.

TABLE 3.

Comparison of the rates of oxygen uptake and nonvolatile product formation by different ISP α subunits combined with ISP_BED β subunit^a

Source of ISP α subunit in assay (plasmid)	Type of assay	Sp act (nmol of O2 min−1 mg−1 or nmol of product min−1 mg−1)			% of uncoupling
		Benzene	Toluene	Ethyl-benzene	Benzene	Toluene	Ethylbenzene
ISPBED (pJRM504)	Polarographic	115 ± 9.6b	70.5 ± 5.5	25 ± 1.5	0	22	0
	Radiolabeled	120 ± 2.9	54.9 ± 1.9	30.7 ± 1.7
ISPTOD (pJRM5042)	Polarographic	44.9 ± 2.2	56.4 ± 3.9	54.1 ± 3.4	0	0	21
	Radiolabeled	45.5 ± 0.7	61.3 ± 2.5	42.5 ± 3.4
ISPBOD (pJRM5041)	Polarographic	44.5 ± 3.1	47.5 ± 4.1	49.1 ± 5.6	0	0	45
	Radiolabeled	47.6 ± 2.6	46.2 ± 0.7	26.8 ± 2.3
ISPTED (pJRM5043)	Polarographic	39.1 ± 2.3	25.3 ± 3.0	13.8 ± 0.5	0	0	48
	Radiolabeled	42.1 ± 2.9	25.5 ± 1.7	8.8 ± 0.7
ISPBOD2 (pJRM5041-1)	Polarographic	112 ± 8.5	79.7 ± 8.1	33.7 ± 4.1	0	37	24
	Radiolabeled	115 ± 2.9	49.6 ± 2.2	25.7 ± 2.6
ISPBOD3 (pJRM5041-2)	Polarographic	36.8 ± 4.6	32.2 ± 6.1	34.5 ± 4.0	0	0	31
	Radiolabeled	39.1 ± 1.5	32.2 ± 0.6	23.6 ± 2.5
I301V/T305S/I307L/L309V (pJRM504M4)	Polarographic	84.5 ± 4.4	62.1 ± 8.9	70.7 ± 9.9	0	0	44
	Radiolabeled	85.1 ± 1.9	70.8 ± 6.1	39.9 ± 1.4

^aActivities were determined with ISP_BED β subunit, ferredoxin_BED, and reductase_BED cell extracts. The specific activity is expressed as the rate of O₂ uptake measured polarographically (three determinations) or as the amount of nonvolatile ¹⁴C-labeled product formed from oxidation of [¹⁴C]benzene, [¹⁴C]toluene, and [¹⁴C]ethylbenzene (nine determinations).

^bMean ± standard deviation.

The results for the chimeric α subunits (Fig. 1A) revealed that a region of the protein responsible for specificity was between residues 281 and 450. In this region, benzene and toluene dioxygenases differ at 12 amino acid positions. Single and multiple amino acid mutations were generated, in which the nonconserved residues of ISP_BEDα were replaced by the corresponding residues of ISP_TODα (Fig. 1B). The ISPα concentration was estimated by quantitative SDS-PAGE (Fig. 2B) to be between 0.17 and 0.42 mg mg of protein⁻¹. When the activity was measured polarographically, the majority of the mutant proteins had a substrate preference similar to that of the wild type. However, ISP_BEDα(I301V/T305S/I307V/L309V) showed an increased preference for ethylbenzene, with an ethylbenzene/benzene activity ratio of 0.84 (Fig. 1B).

Table 3 shows the rates of product formation measured by ¹⁴C assays for a representative range of recombinant proteins combined with ISP_BEDβ, and the data were compared with the rates of oxygen uptake under the same conditions, as shown in Fig. 1. The difference between the rate of product formation and the rate of oxygen consumption represented the rate of uncoupled oxidation of NADH. With benzene as the substrate, the rate of O₂ consumption for ISP_BEDα was equal to the amount of products formed, which was indicative of tight coupling, but for the chimeric proteins uncoupling was observed with toluene and, more particularly, with ethylbenzene. For example, the percentage of uncoupling (percentage of O₂ consumed in excess of the amount of product formed) was 45% for ISP_BODα and 44% for ISP_BEDα(I301V/T305S/I307L/L309V), both with ethylbenzene (Table 3). Similar uncoupling reactions have been observed with naphthalene dioxygenase, in which the oxidation of benzene, toluene, and ethylbenzene was partially uncoupled from O₂ consumption (29, 31). However, the ethylbenzene/benzene activity ratios as determined by the radiolabel assay were 0.67 for ISP_BODα and 0.71 for ISP_BOD3α, compared to 0.23 for ISP_BEDα (Table 3), indicating that these enzymes had a higher substrate preference for ethylbenzene than the wild type. This phenomenon was less pronounced for ISP_BEDα(I301V/T305S/I307L/L309V) which had a ratio of 0.47.

The products formed from the three substrates were investigated. Samples from ¹⁴C assays of the recombinant proteins, either expressed separately or coexpressed (ISP_BEDα-ISP_BEDβ, ISP_TODα-ISP_BEDβ, and ISP_BODα-ISP_BEDβ), were subjected to TLC (Fig. 3). With benzene as the substrate, a single compound with an R_f of 0.25, corresponding to the reference compound benzene cis-dihydrodiol, was detected in every case (Fig. 3A). This result is consistent with previous studies (12). With toluene as the substrate, two compounds (R_f, 0.31 and 0.67) were detected (Fig. 3B). By analogy to the benzene dioxygenase product, the compound with an R_f of 0.31 was provisionally identified as the expected cis-2,3-dihydroxy-1-methyl-cyclohexa-4,6-diene (toluene cis-dihydrodiol) (12). The R_f of 0.67 corresponded to the R_f of benzyl alcohol. Similarly, with all the dioxygenases that oxidized ethylbenzene, two compounds with R_f values of 0.35 and 0.71 were detected (Fig. 3C), and these compounds were provisionally identified as cis-2,3-dihydroxy-1-ethyl-cyclohexa-4,6-diene (ethylbenzene cis-dihydrodiol) and 1-phenethyl alcohol, respectively.

FIG. 3.

Autoradiograms of the products formed by using [¹⁴C]benzene (A), [¹⁴C]toluene (B), and [¹⁴C]ethylbenzene (C) by wild-type, chimeric, and mutant ISP_BEDα subunits combined with the ISP_BEDβ subunit and coexpressed ISP_BEDα-ISP_BEDβ, ISP_TODα-ISP_BEDβ, and ISP_BODα-ISP_BEDβ in the dioxygenase assay. The R_f values of the radiolabeled products are indicated on the left. The numbers on the autoradiograms indicate the percentages of the chain monohydroxylated and cis-diol products. The values are averages of two determinations. A repeat experiment gave similar results. Details of the experiment are described in Materials and Methods.

The products of the reactions described above, but without the use of radiolabeled substrates, were further analyzed by GC-MS (Tables 4 and 5). Since some compounds are known to undergo thermal decomposition during analysis, the phenol and catechols that were detected by GC-MC and not by other methods were interpreted as being the products of thermal dehydration of the cis-diols. Thus, with benzene as the substrate, peaks with masses corresponding to those of phenol, benzene cis-dihydrodiol, and catechol were detected, and the sum of the masses of these peaks corresponded to the mass of the single product, presumed to be benzene cis-dihydrodiol, observed in the autoradiogram (Fig. 3A).

TABLE 4.

Gas chromatography-mass spectrometric results for reaction products

Iona	Retention time (min)	Product [ion mass (relative abundance)]
Phenol	5.0	94 (100), 66 (38), 65 (28), 63 (9), 55 (9), 51 (7)
Benzene cis-dihydrodiol	5.3	12 (15), 94 (23), 83 (36), 68 (26), 66 (100), 65 (32), 55 (40)
Catechol	7.8	110 (100), 92 (12), 81 (19), 64 (47), 63 (23), 53 (18)
Benzaldehyde	4.5	106 (82), 105 (82), 77 (100), 51 (53)
Benzyl alcohol	5.7	108 (73), 107 (49), 79 (100), 77 (59), 51 (31)
o-Cresol	6.0	108 (100), 107 (90), 90 (25), 79 (54), 77 (51), 51 (25)
m-Cresol	6.3	108 (100), 107 (94), 79 (47), 77 (46), 53 (16), 51 (19)
Toluene cis-dihydrodiol	6.4	126 (7), 108 (60), 97 (21), 80 (82), 79 (100), 77 (36), 65 (23), 55 (22) 53 (22) 51 (20)
Phenethyl alcohol	6.0	122 (20), 107 (77), 79 (100), 77 (57), 51 (29)
2-Hydroxyethylbenzene	7.0	122 (39), 107 (100), 79 (17), 77 (34), 51 (10)
3-Hydroxyethylbenzene	7.4	122 (41), 107 (100), 79 (10), 77 (32), 51 (8)
Ethylbenzene cis-dihydrodiol	7.5	140 (8), 122 (8), 111 (19), 94 (40), 84 (49), 83 (86), 81 (30), 79 (38), 68 (49), 67 (31), 57 (46), 55 (100), 53 (28), 51 (17)

^aAll the products formed by dioxygenase-catalyzed reactions with ISP_BEDα, ISP_TODα, or ISP_BEDα(I301V/T305S/I307L/L309V) combined with the ISP_BED β subunit by using benzene, toluene, or ethylbenzene as the substrate.

TABLE 5.

Percentages of products formed by dioxygenase-catalyzed reactions with ISP_BEDα, ISP_TODα, or ISP_BEDα(I301V/T305S/I307L/L309V) combined with the ISP_BED β subunit by using benzene, toluene, or ethylbenzene as the substrate

Substrate	Product	% of product formeda
		ISPBEDα	ISPTODα	ISPBEDα mutant
Benzene	Benzene cis-dihydrodiol	46	37	42
	Phenolb	35	30	37
	Catecholb	19	33	21
Toluene	Toluene cis-dihydrodiol	3	45	12
	Benzaldehyde	18	1	8
	Benzyl alcohol	53	6	31
	o-Cresolb	22	38	42
	m-Cresolb	4	10	7
Ethylbenzene	Ethylbenzene cis-dihydrodiol	5	24	16
	1-Phenethyl alcohol	24	39	5
	2-Hydroxyethylbenzeneb	58	30	65
	3-Hydroxyethylbenzeneb	13	7	14

^aThe values are averages of two determinations. A repeat experiment gave similar results.

^bProduct expected from thermal decomposition. See text for details.

When toluene was used, toluene cis-dihydrodiol and o- and m-cresols, the expected dehydration products, were detected (Table 5); again, the sum of the levels of these products was taken to represent the level of cis-diols produced. In addition, benzyl alcohol and benzaldehyde were observed only in trace amounts with ISP_TOD, but the amounts were larger with ISP_BED. In principle, the benzyl alcohol could be formed either by direct monohydroxylation of toluene by benzene dioxygenase or by production of toluene cis-dihydrodiol, followed by dehydration and isomerization to benzyl alcohol. This possibility was eliminated by mixing ISP_TODα and ISP_BEDβ in a dioxygenase assay mixture and allowing them to react with [¹⁴C]toluene to produce the cis-dihydrodiol. The preparation was then further incubated with a reaction mixture containing ISP_BEDα and ISP_BEDβ, and the results were analyzed by TLC. No further formation of benzyl alcohol was observed (data not shown). Therefore, the benzyl alcohol was the result of chain monohydroxylation of the substrate by the enzyme system; benzaldehyde was most probably the direct product of further oxidation of benzyl alcohol by the dioxygenase. Analogous reactions have been observed during oxidation of toluene by naphthalene dioxygenase (30) and xylene monooxygenase (18). However, the possibility that a nonspecific host alcohol dehydrogenase caused the conversion of benzyl alcohol to benzaldehyde cannot be ruled out.

Similarly, for ethylbenzene, in addition to the cis-dihydrodiol detected in the autoradiogram analysis, the dehydration products 2- and 3-hydroxyethylbenzene were observed by GC-MS (Table 5). As observed for the oxidation of toluene, ethylbenzene was monohydroxylated on its side chain to form 1-phenethyl alcohol. Lee and Gibson (31) reported that naphthalene dioxygenase oxidized ethylbenzene to 1-phenethyl alcohol, followed by slower oxidation of the latter compound to acetophenone. This rate-determining step (31) may explain why acetophenone was not observed.

Table 5 shows that all of the enzymes tested catalyzed the side chain monohydroxylation of toluene and ethylbenzene. The products were quantified by TLC (Fig. 3), and in general, toluene dioxygenase was more efficient than benzene dioxygenase in the dihydroxylation of toluene and produced less of the monohydroxylation products benzyl alcohol and benzaldehyde. ISP_BEDα(I301V/T305S/I307L/L309V) was intermediate between these two proteins. As expected for benzene, which has no side chains to be oxidized, no products other than the dihydrodiol were detected.

DISCUSSION

In this work, sections and individual amino acids of the toluene dioxygenase sequence were replaced in the benzene dioxygenase α subunit in order to probe the regions responsible for substrate preference for toluene and ethylbenzene. All of the ISP α subunits were active when they were combined with β subunits in vitro, in contrast to some other dioxygenases, in which the α and β subunits must be coexpressed to form a functional enzyme (10, 20, 46, 48). The ISP_BOD3α chimera narrows the region for substrate preference for alkylbenzenes between residues 281 and 363. In this region, the amino acid substitutions I301V, T305S, I307L, and L309V resulted in a recombinant enzyme with a rate of O₂ uptake with ethylbenzene that was substantially higher than that of the wild type (Table 3). The dioxygen data combined with the amount of nonvolatile products showed that for some of the ISP α subunits tested, partial uncoupling of O₂ consumption was observed with toluene and, more particularly, with ethylbenzene. The first product of uncoupled reduction of O₂ is expected to be H₂O₂ from the protonation of iron-bound peroxide species, as proposed for cytochrome P450 (14). However, formation of H₂O₂ is not observed as a result of uncoupling in assays of benzene dioxygenase (1). As suggested by Lee (29), this could be due to iron-catalyzed Fenton chemistry with H₂O₂ and, ultimately, water as the final product. ISP_BODα and ISP_BOD3α in which partial uncoupling reactions were observed did have higher substrate specificity for ethylbenzene than the wild-type ISP_BEDα. However, the amino acids I301, T305, I307, and L309 are not the only amino acids involved in substrate specificity because although they increased the substrate preference for ethylbenzene, the high uncoupling reaction (44% with ethylbenzene) made them less efficient than the wild-type ISP_TOD.

There was no discernible correlation between the degree of uncoupling and the type of ISP α subunit in the assay; for example, ISP_BEDα showed partial uncoupling of oxygen consumption with toluene (22%), while ISP_TEDα, which contained the same C-terminal region, showed no significant uncoupling with the same substrate. However, ethylbenzene, which had the largest substituent, induced uncoupling of O₂ uptake in most cases, indicating that substrate shape or size is a factor. The fact that the specific activities of some of the chimeric α subunits combined with ISP_BEDβ were lower with all three substrates (Table 3) than the specific activity of wild-type ISP_BEDα could be explained by some mismatch at the subunit interface in the heterologous α-β pair, although, as indicated above, this would not have any effect on the relative activity with the substrates. The source of the β subunit had no systematic effect on substrate specificity (Fig. 1).

Analysis of product formation indicated that both dihydroxylation and chain monohydroxylation occurred in all the recombinant ISP α subunits combined with ISP β subunits. Several studies have reported that dioxygenases, such as toluene dioxygenase (4, 32, 40-42, 51), catalyze monohydroxylation reactions. However, to our knowledge, this study is the first study to report the chain monohydroxylation of toluene and ethylbenzene by benzene dioxygenase to form benzyl alcohol and 1-phenethyl alcohol, respectively. The amounts of cis-diols and chain monohydroxylated products were similar whether the ISP α and β subunits were expressed separately or not (Fig. 3), confirming that the two subunits can combine with each other to form a functional enzyme. For example, around 50% of toluene cis-dihydrodiol and benzyl alcohol was produced with ISP_BEDα and coexpressed ISP_BEDα-ISP_BEDβ when toluene was the substrate. This also showed that the amount of cis-diols and chain monohydroxylated compounds depended upon the nature of the α subunit in the assay. Furthermore, ISP_TODα, ISP_BODα, and ISP_BOD3α, all of which had the C-terminal region of ISP_TOD, had different amounts of chain monohydroxylated products (for example, 4, 16, and 48% benzyl alcohol, respectively), indicating that no region of the ISP α subunit seemed to correlate with the amount of the benzyl and 1-phenethyl alcohols.

We modeled the sequence of the ISP_BEDα(I301V/T305S/I307L/L309V) by homology with the known structure of naphthalene dioxygenase (25) (Fig. 4). As determined by sequence comparisons, I301V, T305S, I307L, and L309V correspond to V287, I291, R293, and H295 in naphthalene dioxygenase. Surprisingly, these residues are not in the active site cavity, nor do they line the channel leading from this site to the surface. H295 is located behind F352, and modification of the latter amino acid has been shown to alter the regiospecificity of naphthalene dioxygenase (39). L307 in toluene dioxygenase corresponds to the residue which was shown by Zhang et al. (56) (by using substitution in a proline) to change the proportion of 1-indenol, 1-indanone, and cis-indandiol produced. The other amino acids are aligned adjacent to residues lining the substrate channel. Replacement of I301 and L309 in benzene dioxygenase by the less bulky amino acid valine should facilitate access to the larger substrates, such as toluene and ethylbenzene, either by decreasing steric hindrance or by increasing flexibility in the channel lining. Although the amino acids surrounding the binding site for the aromatic ring are unchanged, the effects of the mutations on the conformational flexibility of the substrate cavity are difficult to predict. The consequences of the combination are a higher rate of oxygen uptake for toluene and ethylbenzene but a high rate of uncoupling, especially with ethylbenzene. Some monohydroxylated products were observed, but the amount was intermediate between the amounts for all the novel enzymes. The uncoupling is not directly associated with the formation of the chain monohydroxylated product because ISP_BEDα(I301V/T305S/I307L/L309V) had a high level of uncoupling with ethylbenzene yet produced the largest proportion of cis-diols.

FIG. 4.

Structural representation of the putative substrate entry channel in ISP_BEDα(I301V/T305S/I307L/L309V). The protein is represented in spacefill. Iron is indicated by red, and the four amino acid substitutions (I301V, T305S, I307L, and L309V) are indicated by blue spheres. The view is a section through the substrate channel, and the arrow indicates the putative route that the substrate takes to get to the iron catalytic site. The model was generated with the SWISS-MODEL program (15; http://expasy.hcuge.ch/swissmod/SWISS-MODEL.html) and was displayed by using RasWin Molecular Graphics Window, version 2.6 (43).

One must assume that the native enzymes evolved to maximize the dihydroxylation of the preferred substrates to the required stereospecific products and to avoid the side reactions. From the crystal structures of naphthalene dioxygenase it is clear that the position of the substrate relative to the Fe-bound O₂ is favorable for the dioxygenase reaction (24). The three different reactions, dihydroxylation, monohydroxylation, and uncoupled reduction of O₂ to H₂O₂, may be considered to be the result of competing pathways in the oxidoreduction reactions of iron with oxygen (Fig. 5). Wolfe et al. (52, 53) have shown that there is strict control of the redox events within the enzyme. To avoid uncontrolled reduction of O₂, electron transfer from the Rieske cluster to Fe(III) does not occur until substrate is present in the active site, and the product is not released until the iron is reduced to Fe(II) for the next cycle. The dihydroxylation of the substrate by the adjacent side-on peroxo complex is the dominant reaction for native benzene and toluene dioxygenases. However, in the chimeric enzymes the shape of the substrate channel is distorted, and it is possible that the substrate is less tightly bound than it is in the native enzyme. Movement of the substrate away from the binding pocket should allow the side chain hydroxylation, possibly by repositioning of the substrate in the active site, and the uncoupling reaction with reduction to water. Thus, the different product ratios can be understood in terms of the tightness of binding of the substrate to the active site. Following the course of directed evolution seems to be a promising strategy for the engineering of enzymes to metabolize novel substrates.

FIG. 5.

Alternative pathways for dioxygenase and monooxygenase activities and uncoupling. Dihydroxylation results from the addition of O₂ and two-electron reduction to the peroxo derivative. In the structure of naphthalene dioxygenase with O₂ and substrate in the active site (24), the oxygen is bound side-on to the iron and next to the aromatic ring. Addition of both oxygen atoms, with two protons, would lead to addition of two —OH groups. The monooxygenation reactions are concerted reactions in which one oxygen atom is protonated and reduced to H₂O and the other oxygen atom is inserted into the C—H bond of the side chain. The uncoupling reaction is shown as a protonation of the Fe-peroxo species, releasing H₂O₂.